article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in the present work a method for solving the problem of time - dependent electromagnetic wave propagation through an isotropic inhomogeneous medium is developed . several ideas concerning such mathematical notions as bicomplex and biquaternionic reformulations of electromagnetic models , hyperbolic vekua - type equations and transmutation operators from the theory of ordinary linear differential equations
are combined in our approach which results in a simple and practical representation for solutions .
we observe that the 1 + 1 maxwell system for inhomogeneous media can be transformed into a hyperbolic vekua equation .
this gives us the possibility to obtain the exact solution of the problem of a normally incident electromagnetic time - dependent plane wave propagated through an inhomogeneous layer in terms of a couple of darboux - associated transmutation operators .
this is a new representation for a solution of the classic problem .
application of the recent results on the analytic approximation of such operators allows us to write down the electromagnetic wave in an approximate analytic form which is then used for numerical computation .
the numerical implementation of the proposed approach reduces to a certain recursive integration and solution of an approximation problem , and can be based on the usage of corresponding standard routines of such packages as matlab . in section [ sect2 ]
we recall a biquaternionic reformulation of the maxwell system and use it to relate the 1 + 1 maxwell system for inhomogeneous media with a hyperbolic vekua equation .
we establish the equivalence between the electromagnetic transmission problem and an initial - value problem for the vekua equation . in section [ sect3 ]
we obtain the exact solution of the problem in terms of a couple of the transmutation operators . in section [ sect4 ] the analytic approximation of the exact solution is obtained in the case when the incident wave is a partial sum of a trigonometric series .
other types of initial data ( and hence modulations ) are discussed in section [ sect5 ] .
section [ sect6 ] contains several exactly solved examples used as test problems for the resulting numerical method . in section [ sect7 ]
we formulate some additional conclusions .
the algebraic formalism in studying electromagnetic phenomena plays an important role since maxwell s original treatise in which hamilton s quaternions were present . in his phd thesis of 1919
lanczos wrote the maxwell system for a vacuum in the form of a single biquaternionic equation .
this elegant form of maxwell system was rediscovered in several posterior publications , see , e.g. , @xcite . in @xcite ( see also @xcite ) the maxwell system describing electromagnetic phenomena in isotropic inhomogeneous media was written as a single biquaternionic equation .
the maxwell system has the form@xmath0@xmath1@xmath2@xmath3 here @xmath4 and @xmath5 are real - valued functions of spatial coordinates , @xmath6 and @xmath7 are real - valued vector fields depending on @xmath8 and spatial variables , the real - valued scalar function @xmath9 and vector function @xmath10 characterize the distribution of sources of the electromagnetic field .
the following biquaternionic equation obtained in @xcite is equivalent to this system @xmath11 here @xmath12 where @xmath13 is the wave propagation velocity and @xmath14 is the the intrinsic impedance of the medium .
all magnitudes in bold are understood as purely vectorial biquaternions , and the asterisk denotes the complex conjugation ( with respect to the complex imaginary unit @xmath15 ) .
the operator @xmath16 is the main quaternionic differential operator introduced by hamilton himself and sometimes called the moisil - theodoresco operator .
it is defined on continuously differentiable biquaternion - valued functions of the real variables @xmath17 , @xmath18 and @xmath19 according to the rule@xmath20 where @xmath21 and @xmath22 are basic quaternionic units . in @xcite with the aid of the representation of the maxwell system in the form ( [ maxwell quat ] ) it was observed that in the sourceless situation ( i.e. , @xmath9 and @xmath10 are identically zeros ) and when all the magnitudes involved are independent of two spatial coordinates , say , @xmath18 and @xmath19 , @xmath23 and @xmath24 , the maxwell system is equivalent to the following vekua - type equation @xmath25 where @xmath26 , @xmath27 is a hyperbolic imaginary unit , @xmath28 commuting with @xmath15 , @xmath29 is a bicomplex - valued function of the real variables @xmath30 and @xmath8 , @xmath31 and @xmath32 , @xmath33 are complex valued ( containing the imaginary unit @xmath15 ) .
the function @xmath34 is real valued and depends on @xmath30 only .
the conjugation with respect to @xmath27 is denoted by the bar , @xmath35 .
the maxwell system in this case can be written in the form@xmath36 where @xmath37 , @xmath38 , @xmath39 .
the relation between ( [ maxwell1 + 1 ] ) and ( [ bicompw ] ) involves the change of the independent variable @xmath40 .
the function @xmath34 in ( [ bicompw ] ) is related to @xmath4 and @xmath5 by the equality @xmath41 where and below the tilde means that a function of @xmath42 is written as a function of @xmath30 , @xmath43 .
the function @xmath29 is written in terms of @xmath44 and @xmath45 as follows@xmath46 let us consider the problem of a normally incident plane wave transmission through an inhomogeneous medium ( see , e.g. , ( * ? ? ?
* chapter 8) ) .
the electromagnetic field @xmath47 and @xmath48 is supposed to be known at @xmath49,@xmath50 .
\label{init cond}\ ] ] we assume @xmath51 and @xmath52 to be continuously differentiable functions . the problem ( [ maxwell1 + 1 ] ) , ( [ init cond ] ) can be reformulated in terms of the function ( [ relation w ] ) .
find a solution of ( [ bicompw ] ) satisfying the condition@xmath53 where @xmath54 is a given continuously differentiable function .
first , let us consider the elementary problem@xmath55 the hyperbolic cauchy - riemann system ( [ cr hyper ] ) was studied in several publications ( see , e.g. , @xcite , @xcite , @xcite and more recent @xcite ) .
its general solution can be written in the form@xmath56 where @xmath57 and @xmath58 are arbitrary continuously differentiable scalar functions , @xmath59 .
for the scalar components of @xmath60 we introduce the notations @xmath61 when @xmath62 we obtain @xmath63 .
hence the unique solution of the cauchy problem ( [ cr hyper ] ) , ( [ initial condition w ] ) has the form@xmath64 where @xmath65 in @xcite there was established a relation between solutions of ( [ cr hyper ] ) and solutions of ( [ bicompw ] ) . any solution of ( [ bicompw ] )
can be represented in the form @xmath66 + jt_{1/f}\left [ \mathcal{i}\left ( w(\xi , t)\right ) \right ] \label{w transmut w}\ ] ] where @xmath60 is a solution of ( [ cr hyper ] ) , @xmath67 and @xmath68 are darboux - associated transmutation operators defined in @xcite , see also @xcite and @xcite , and applied with respect to the variable @xmath30 .
both operators have the form of volterra integral operators , @xmath69 with continuously differentiable kernels .
moreover , the operators @xmath67 and @xmath68 preserve the value at @xmath62 giving additionally to ( [ w transmut w ] ) the relation @xmath70 .
this together with ( [ elementary sol ] ) allows us to write down the unique solution of the problem ( [ bicompw ] ) , ( [ initial condition ] ) in the form @xmath71 + \frac{j}{2}t_{1/f}\left [ w_{0}^{+}(t+\xi)-w_{0}^{-}(t-\xi)\right ] .
\label{solution initial problem}\ ] ] in what follows we use the convenience of this representation and the recent results @xcite , @xcite on the construction of the operators @xmath67 and @xmath68 .
the single - wave approximation of the solution of ( [ bicompw ] ) , ( [ initial condition ] ) ( see ( * ? ? ?
* subsection 8.5.2 ) ) can be obtained from the representation ( [ solution initial problem ] ) by removing integrals from the definition of @xmath67 and @xmath68 or in other words , replacing @xmath67 and @xmath68 by the identity operator .
initial data interesting in practical problems correspond to modulated electromagnetic waves which are represented as partial sums of trigonometric series ( other types of initial data are discussed in the next section ) .
in other words , consider initial data of the form @xmath72 this leads to a similar form for the initial data @xmath73 in ( [ initial condition ] ) , @xmath74 where the bicomplex numbers @xmath75 are related to @xmath76 , @xmath77 as follows@xmath78 due to ( [ solution initial problem ] ) , the unique solution of the problem ( [ bicompw ] ) , ( [ initial condition ] ) with @xmath73 given by ( [ w0exp ] ) can be written in the form@xmath79 + \gamma _ { m}^{-}t_{f}\left [ e^{-i(\omega_{0}+m\omega)\xi}\right ] \right)\right .
\\ & + j\left.\sum_{m =- m}^{m}e^{i(\omega_{0}+m\omega)t}\left ( \gamma_{m}^{+}t_{1/f}\left [ e^{i(\omega_{0}+m\omega)\xi}\right ] -\gamma_{m}^{-}t_{1/f}\left [ e^{-i(\omega_{0}+m\omega)\xi}\right ] \right )
\right ) \end{split}\label{wxit}\ ] ] where @xmath80 .
although the explicit form of the operators @xmath67 and @xmath68 is usually unknown , in @xcite it was shown how their kernels can be approximated by means of generalized wave polynomials .
in particular , for the images of the functions @xmath81 the following approximate representations are valid@xmath82 & \cong e^{\pm i(\omega_{0}+m\omega)\xi}+2\sum_{n=0}^{n}a_{n}\sum_{\text{even } k=0}^{n}\binom{n}{k}\varphi_{n - k}(\xi)\int_{0}^{\xi}\tau^{k}\cos(\omega_{0}+m\omega)\tau\,d\tau\\ & \quad\pm2i\sum_{n=1}^{n}b_{n}\sum_{\text{odd } k=1}^{n}\binom{n}{k}\varphi_{n - k}(\xi)\int_{0}^{\xi}\tau^{k}\sin(\omega_{0}+m\omega)\tau\,d\tau \end{split } \label{tfexp}\ ] ] and@xmath83 & \cong e^{\pm i(\omega_{0}+m\omega)\xi}-2\sum_{n=0}^{n}b_{n}\sum_{\text{even } k=0}^{n}\binom{n}{k}\psi_{n - k}(\xi)\int_{0}^{\xi}\tau^{k}\cos(\omega_{0}+m\omega ) \tau\,d\tau\\ & \quad\mp2i\sum_{n=1}^{n}a_{n}\sum_{\text{odd } k=1}^{n}\binom{n}{k}\psi _ { n - k}(\xi)\int_{0}^{\xi}\tau^{k}\sin(\omega_{0}+m\omega)\tau\,d\tau . \end{split } \label{t1/fexp}\ ] ] here all integrals are easily calculated in a closed form , the functions @xmath84 and @xmath85 are defined as follows .
consider two sequences of recursive integrals@xmath86 and @xmath87 the two families of functions @xmath88 and @xmath89 are constructed according to the rules @xmath90 and @xmath91 finally , the coefficients @xmath92 and @xmath93 are obtained by solving an approximation problem described in @xcite . as an important feature of the representations
( [ tfexp ] ) and ( [ t1/fexp ] ) in @xcite there were obtained estimates of their accuracy uniform with respect to the parameter @xmath94 .
note that the direct evaluation of expressions and requires @xmath95 algebraic operations for each pair of @xmath96 and @xmath30 leading to the computation complexity @xmath97 for the evaluation of .
the change of summation order in and allows one to evaluate with the computation complexity of @xmath98 .
for example , consider @xmath99 the coefficients @xmath100 can be precomputed once for every @xmath30 in @xmath95 operations and the outer sum requires @xmath101 operations .
the proposed method is not restricted exclusively to initial data which can be represented or closely approximated by .
when one can efficiently calculate ( at least numerically ) the indefinite integrals @xmath102 the approximations of the transmutation operators can be used .
one of the examples of such initial data arises in digital signal transmission . for widely used modulations like phase - shift keying or qam
, the transmitted signal can be represented as @xmath103\mathbf{1}_{[nf_s , ( n+1)f_s)}(t),\ ] ] where @xmath104 is the characteristic function of the interval @xmath105 , @xmath106 is the carrier frequency , @xmath107 is the symbol rate and the coefficients @xmath108 and @xmath109 encode transmitted information .
other examples include gaussian rf pulses and linear frequency modulation ( also known as chirp modulation ) used for radars .
we refer the reader to @xcite for further details . returning to , consider @xmath110 $ ] . in @xcite
the following approximation was constructed @xmath111(\xi ) & \cong w_0^+(t+\xi)+\sum_{n=0}^n a_n \sum_{\text{even } k=0}^n \binom nk \varphi_{n - k}(\xi)\int_{-\xi}^{\xi } \tau^k w_0^+(t+\tau)\ , d\tau \\ & \quad + \sum_{n=1}^n b_n \sum_{\text{odd } k=1}^n \binom nk \varphi_{n - k}(\xi)\int_{-\xi}^{\xi } \tau^k w_0^+(t+\tau)\ , d\tau .
\end{split}\label{tfw}\ ] ] we have @xmath112 hence reordering terms in one obtains @xmath113 & \cong w_0^+(t+\xi)+\sum_{\ell=0}^n \left(\int_{t-\xi}^{t+\xi } z^\ell w_0^+ ( z)\ , dz\right)\\ & \quad \times \sum_{n=\ell}^n \bigg\{a_n\sum_{\text{even } k\ge \ell}^n + b_n\sum_{\text{odd } k\ge \ell}^n\bigg\ } ( -1)^{k-\ell}\binom nk \binom k\ell \varphi_{n - k}(\xi ) t^{k-\ell } , \end{split}\ ] ] an expression which can be easily evaluated once the indefinite integrals of @xmath114 are known .
similarly , @xmath115 & \cong w_0 ^ -(t-\xi)+\sum_{\ell=0}^n \left(\int_{t-\xi}^{t+\xi } z^\ell w_0 ^ - ( z)\ , dz\right)\\ & \quad \times \sum_{n=\ell}^n \bigg\{a_n\sum_{\text{even } k\ge \ell}^n + b_n\sum_{\text{odd } k\ge \ell}^n\bigg\ } ( -1)^{\ell}\binom nk \binom k\ell \varphi_{n - k}(\xi ) t^{k-\ell}. \end{split}\ ] ] to obtain approximations for @xmath116 $ ] on can change in and the coefficients @xmath117 and @xmath118 by @xmath119 and @xmath120 respectively and change functions @xmath121 by @xmath122 .
[ ex1 ] let us consider the system ( [ maxwell1 + 1 ] ) with the permittivity of the form @xmath123 where @xmath124 and @xmath125 are some real numbers , such that @xmath126 does not vanish on the interval of interest and @xmath127
. then @xmath128 .
hence @xmath129 in this case the vekua equation ( [ bicompw ] ) has the form @xmath130 where the coefficient @xmath131 is constant , @xmath132 .
the vekua equation ( [ bicompw ] ) is a special case of the main vekua equation @xmath133 ( see @xcite , @xcite ) . for its solutions
one has that @xmath134 satisfies the equation @xmath135 where @xmath136 and @xmath137 , and @xmath138 satisfies the equation @xmath139 where @xmath140 .
the relation between @xmath134 and @xmath138 is akin to the relation between harmonic conjugate functions and can be found in @xcite , @xcite . in the case of equation ( [ vekuaconst ] )
consider a solution @xmath32 of the equation @xmath142 in the form @xmath143 where @xmath144 and @xmath145 are arbitrary real constants .
then a solution @xmath29 of ( [ vekuaconst ] ) such that @xmath146 can be chosen in the form ( see @xcite , @xcite ) @xmath147 this leads to the following solution of the maxwell system @xmath148 and@xmath149 or in terms of the variables @xmath42 and @xmath8 , @xmath150{\mu}\sqrt{\alpha x+\beta}e^{\frac{i\alpha t}{2\sqrt{\mu}}}\left ( \frac{\sqrt{\mu}}{\alpha}a\log\frac{\alpha x+\beta } { \beta}+b\right)\ ] ] and@xmath151{\mu}\sqrt{\alpha x+\beta}}\left ( \frac{\sqrt{\mu}}{\alpha}a\log\frac{\alpha x+\beta}{\beta}+b+\frac{2a\sqrt{\mu}}{\alpha}\right ) .\ ] ] thus the functions and are the exact solutions of the maxwell system with @xmath152 given by and with the initial conditions @xmath153{\mu}\sqrt{\beta}e^{\frac{i\alpha t}{2\sqrt{\mu}}}b\quad \text{and}\quad
\mathcal{h}(0,t)= -\frac{e^{\frac{i\alpha t}{2\sqrt{\mu}}}}{\sqrt[4]{\mu } \sqrt{\beta}}\left ( b+\frac{2a\sqrt{\mu}}{\alpha}\right).\ ] ] below we present the results of numerical solution of the same system by the proposed method . for the numerical experiment we considered an interval @xmath154
$ ] for both @xmath42 and @xmath8 and took @xmath155 , @xmath156 , @xmath157 and @xmath158 , @xmath159 .
the initial condition in this example corresponds to the case @xmath160 and @xmath161 in ( [ w0exp ] ) .
the permittivity @xmath152 was approximated on uniform mesh of @xmath162 points .
the new variable @xmath30 was obtained by the modified 6 point newton - cottes integration formula , see @xcite for details .
the same integration formula was used for calculation of all other integrals , i.e. , for computation of and and for the integration of the potential @xmath163 required to obtain approximations and . note that the integration with respect to the variable @xmath30 requires integration over a non - uniform mesh ,
however such inconvenience can be easily avoided observing that @xmath164 for any function @xmath165 .
all calculations were performed using matlab 2012 in the machine precision .
the exact expressions were used only for the function @xmath152 and its derivatives , all other functions involved were computed numerically .
the developed program found the optimal value of @xmath166 for the approximations and to be @xmath167 .
the computation time required was 0.4 seconds . on figure [ fig : ex1 ]
we show the graphs of the absolute errors of the computed @xmath168 and @xmath169 .
[ ex2 ] let us consider the same parameters of the medium as in example [ ex1 ] but instead of choosing a solution of ( [ vekuaconst ] ) in the form ( [ solvekuaconst ] ) we take the solution@xmath170 here @xmath144 and @xmath171 are arbitrary constants , @xmath172 , @xmath173 .
it is obtained similarly to ( [ solvekuaconst ] ) starting with a solution @xmath32 of the equation @xmath174 in the form @xmath175 .
we have then that @xmath176{\mu}\sqrt{\alpha x+\beta}e^{i\omega t}\left ( \left ( \frac{\alpha x+\beta}{\beta}\right ) ^{\frac{d\sqrt{\mu}}{\alpha}}+\frac{d+c}{d - c}\left ( \frac{\alpha x+\beta}{\beta}\right ) ^{-\frac { d\sqrt{\mu}}{\alpha}}\right)\ ] ] and@xmath177{\mu}\sqrt{\alpha x+\beta}}\left ( \left ( \frac{\alpha x+\beta}{\beta}\right ) ^{\frac{d\sqrt{\mu}}{\alpha}}-\left ( \frac{\alpha x+\beta}{\beta}\right ) ^{-\frac{d\sqrt{\mu}}{\alpha}}\right)\ ] ] satisfy the maxwell system with the permittivity and the initial conditions @xmath178{\mu}\sqrt{\beta}e^{i\omega t}\quad\text{and}\quad \mathcal{h}(0,t)=0.\ ] ] for the numerical calculation we considered the same values of the parameters @xmath124 , @xmath125 , @xmath5 as in example [ ex1 ] and took the interval @xmath179 $ ] for both @xmath42 and @xmath8 . for the initial condition we took the sum of four terms , each of the form having @xmath180 , @xmath181 .
since the expression for @xmath62 reduces to @xmath182 , we took @xmath183 , @xmath184 and obtained initial conditions @xmath185 .
for this example the optimal @xmath166 was equal to 13 and the whole computation time was @xmath186 seconds . on figure
[ fig : ex2 ] we present the graphs of the initial condition and the computed @xmath168 .
the absolute errors of the computed @xmath168 and @xmath169 were less than @xmath187 and @xmath188 respectively .
[ ex3 ] for this example let us consider the system with the permittivity of the form @xmath189 where @xmath190 and @xmath124 , @xmath191 are such that @xmath192 on the interval of interest .
then @xmath193 .
hence @xmath194 in @xcite we show how one can construct the transmutation operators @xmath195 and @xmath68 when @xmath196 , @xmath197 .
the procedure can be easily generalized for functions of the form @xmath198 , @xmath197 .
hence for values of @xmath199 of the form @xmath200 , @xmath197 one can explicitly construct the pair of transmutation operators @xmath195 and @xmath68 and obtain the solution of . for the numerical experiment we took @xmath201 , @xmath202 , @xmath203 and @xmath204 . for such parameters
the function @xmath205 is equal to @xmath206 and the integral kernels of the transmutation operators @xmath195 and @xmath68 are given by ( see @xcite ) @xmath207 we considered the interval @xmath208 $ ] for @xmath42 and the interval @xmath209 $ ] for @xmath8 and took the gaussian pulse @xmath210 , @xmath211 as the initial condition . for such initial condition
the expression can be evaluated in the terms of @xmath212 function .
the approximate solution was computed using and .
we used 2001 points to represent the permittivity @xmath152 .
the formal powers were computed as it was explained in example [ ex1 ] while to evaluate the indefinite integrals @xmath213 , @xmath214 we approximated the integrands as splines and used the function @xmath215 from matlab .
the main reason for such choice is that despite the uniform mesh was taken for both @xmath42 and @xmath8 , the resulting mesh for @xmath30 may not be uniform leading to rather large set of values which can be taken by @xmath216 and @xmath217 .
all computations were performed in machine precision in matlab 2012 . on figure
[ fig : ex3 ] we show the obtained graphs of @xmath168 and @xmath169 .
the absolute errors of the computed solutions were less than @xmath218 and @xmath219 , respectively .
a method for solving the problem of electromagnetic wave propagation through an inhomogeneous medium is developed .
it is based on a simple transformation of the maxwell system into a hyperbolic vekua equation and on the solution of this equation by means of approximate transmutation operators . in spite of elaborate mathematical results which are behind of the proposed method ,
the final representations for approximate solutions of the electromagnetic problem have a sufficiently simple form , their numerical implementation is straightforward and can use standard routines of such packages as matlab .
h. campos , v. v. kravchenko , l. m. mndez , _ complete families of solutions for the dirac equation : an application of bicomplex pseudoanalytic function theory and transmutation operators _ , adv .
clifford algebr . 22
( 2012 ) , issue 3 , 577594 .
v. v. kravchenko , _ quaternionic equation for electromagnetic fields in inhomogeneous media_. in : progress in analysis , v. 1 , eds .
h. begehr , r. gilbert and m. wah wong , 361366 , world scientific , 2003 .
l. a. ostrovsky , a. i. potapov , _ introduction into the theory of modulated waves _ , fizmatlit , moscow , 2003 ( in russian ) ,
the revised and extended translation of _ modulated waves : theory and applications _ , johns hopkins university press , 2002 . | the time - dependent maxwell system describing electromagnetic wave propagation in inhomogeneous isotropic media in the one - dimensional case reduces to a vekua - type equation for bicomplex - valued functions of a hyperbolic variable , see @xcite . using this relation we solve the problem of the transmission through an inhomogeneous layer of a normally incident electromagnetic time - dependent plane wave .
the solution is written in terms of a pair of darboux - associated transmutation operators @xcite , and combined with the recent results on their construction @xcite , @xcite can be used for efficient computation of the transmitted modulated signals .
we develop the corresponding numerical method and illustrate its performance with examples . |
in this work we seek a mathematical understanding of phase transitions in the steady state of stochastic many - body systems .
systems at equilibrium with their environment provide examples of such steady states , and the mechanisms underlying equilibrium phase transitions are long known and understood .
experimentally , one can distinguish between two types of transition : the first - order transition at which there is phase coexistence , e.g. between a high - density solid and a low - density fluid , and the continuous transition at which fluctuations and correlations grow to such an extent as to be macroscopically observable . from a thermodynamical perspective
, one can understand first - order transitions by associating with each phase a free energy .
for a given set of external parameters , the phase ` chosen ' by the system is that with the lowest free energy , and so a phase transition occurs when the free energies of two ( or more ) phases are equal .
the sudden changes in macroscopically measurable quantities that take place at first - order transitions are described mathematically as discontinuities in the first derivative of the free energy .
discontinuities in higher derivatives relate to continuous ( higher - order ) phase transitions .
the tools of equilibrium statistical physics allow one in principle at least to express the free energy solely in terms of microscopic interactions .
more specifically , the free energy is given by the logarithm of a _ partition function _
, a quantity that normalises the steady - state probability distribution of microscopic configurations .
initially it was not universally accepted that this approach could faithfully describe phase transitions , in particular the first - order solid - fluid transition @xcite . in order to show that the statistical mechanical approach _ can _ reproduce the correct discontinuities in the free energy at a first - order transition , lee and yang @xcite introduced a description of phase transitions concerning zeros of the partition function when generalised to the complex plane of an intensive thermodynamic quantity .
initially , the theory was couched in terms of zeros in the complex fugacity plane which is appropriate for fluids in the grand canonical ( fluctuating particle number ) ensemble . however , by mapping a lattice gas onto the ising model @xcite , the theory was found to hold equally well for a magnet in a fixed magnetic field . later @xcite it became clear that the distribution of zeros in the complex temperature plane can reveal information about phase transitions in the canonical ensemble . in section [ eqm ]
we present a brief , self - contained discussion of the lee - yang theory of equilibrium phase transitions which relates nonanalytic behaviour in the free energy to zeros of the partition function . in the remainder of this article ,
we move our attention to phase transitions in _ nonequilibrium _ steady states , that is , those that carry currents of mass , energy or some other quantity .
we give therefore a very brief introduction to the established practice of modelling nonequilibrium physics through stochastic dynamics in section [ nss ] .
in particular , we will identify a candidate quantity for use in place of an equilibrium partition function when performing an lee - yang analysis of phase transitions .
finally , in section [ neqm ] we review recent work in which the lee - yang theory has been successfully applied to specific nonequilibrium phase transitions .
in this section we recount the relationship of zeros of a partition function to phase transitions in the system it describes .
we present here a brief but self - contained tour of the key points . for more detail ( and mathematical rigour )
the reader is referred to textbooks , such as @xcite or , for a more advanced presentation , @xcite , as well as early treatments of the subject @xcite .
for concreteness we shall describe the theory in terms of a simple model system of @xmath0 spins in thermal contact with a heat reservoir . in this model system
the energy of the system can take values @xmath1 with @xmath2 and the number of microstates corresponding to the @xmath3 energy level is @xmath4 .
the canonical partition function is given by @xmath5 where @xmath6 is the inverse temperature .
to investigate the zeros of this partition function , it is useful to make the change of variable @xmath7
. then , the partition function is explicitly a polynomial in @xmath8 of degree @xmath9 and can be factorised as @xmath10 clearly the quantities @xmath11 are the @xmath9 zeros of the partition function @xmath12 ; meanwhile , @xmath13 is a constant which we can safely ignore in the following .
since all the @xmath4 are positive , no zero of @xmath12 can be real and positive : that is , the zeros @xmath11 will generally lie in the complex plane away from the physical values of @xmath8 which lie on the positive real axis .
we now define for all complex @xmath8 _ except _ the points @xmath14 the ( complex ) free energy @xmath15 using ( [ zfactor ] ) we rewrite this as @xmath16 and note that a taylor series expansion of @xmath17 around a point @xmath18 has a finite radius of convergence given by the distance to the nearest zero from @xmath8 .
this then implies that that @xmath17 can be differentiated infinitely many times over any region of the complex plane that is devoid of partition function zeros .
since we identify a phase transition through a discontinuity in a derivative of the free energy , we see that such a transition can only occur at a point @xmath19 in the complex plane if there is at least one zero of the partition function @xmath12 within any arbitrarily small region around the point @xmath19 .
clearly this scenario is impossible if the number of zeros @xmath9 is finite , except at the isolated points @xmath11 where the free energy exhibits a logarithmic singularity .
since such a point can not lie on the positive real @xmath8 axis , there is no scope for a phase transition in a finite spin system , such as the simple example ( [ zspinsys ] ) . on the other hand ,
if the partition function zeros accumulate towards a point @xmath19 on the real axis as we increase the number of spins @xmath0 to infinity there is the possibility of a phase transition . in order to deal with the thermodynamic limit ( see @xcite for rigorous considerations ) we shall assume that the limit @xmath20 exists and we may write @xmath21 where @xmath22 is the local density of zeros in the complex-@xmath8 plane in the thermodynamic limit . since the imaginary part of this complex free energy @xmath23 is multi - valued it will at times be more convenient to work with the _ potential _ @xmath24 defined as @xmath25 an expression for the density @xmath22 in terms of the potential @xmath24 is easily obtained once one recognises that @xmath26 is the green function for the two - dimensional laplacian .
specifically , if @xmath27 , we have @xmath28 using this expression , we can take the laplacian of both sides of ( [ potential ] ) to find @xmath29 such an equation is familiar from electrostatics and so @xmath24 is analogous to an electrostatic potential . in analogy to electrostatics ,
as long as we can integrate @xmath30 over bounded regions containing any singularities of @xmath31 , the potential will be a continuous function @xcite .
the significance of this statement is that we can derive a rule for locating phase boundaries given a partition function .
let us suppose that around points @xmath32 and @xmath33 in the complex plane , one has ( analytic ) expressions for the potential @xmath34 and @xmath35 such that @xmath36 ( i.e. not the same function over the entirety of the complex plane ) . in order for the potential to be continuous at all points on the complex plane , we must have a phase boundary at those values of @xmath8 for which the condition @xmath37 holds .
this is basically the definition of a transition mentioned above in the introduction . since @xmath38 and @xmath39 are different functions , some derivatives of @xmath24 will not exist at these values of @xmath8 and we expect the density of zeros at these points to be non - zero .
( it is also possible for zeros to be present at other points in the complex plane , but we do not need to consider this possibility here . ) typically , a solution of ( [ boundary - rule ] ) describes a curve @xmath40 that intersects the positive real @xmath8 axis at a point @xmath19 .
having already established that we require zeros to accumulate at the point @xmath19 in the thermodynamic limit for a physical phase transition to take place , we are interested in the line density of zeros per unit length of this curve .
we introduce the arclength @xmath41 which measures the distance along @xmath40 from the transition point @xmath19 .
small area @xmath42 covering a length @xmath43 of the curve of zeros @xmath40 required to derive a relationship for the line density @xmath44 of zeros along the curve . ] to obtain an expression for the line density of zeros @xmath44 along the curve @xmath40 , consider a short section of @xmath40 of length @xmath45 enclosed by an area @xmath42 that has two sides parallel to @xmath40 and the other two sides perpendicular to @xmath40see figure [ line - density ] . integrating the density @xmath22 over this area we have @xmath46 meanwhile , we have from ( [ subharmonic ] ) and an application of the divergence theorem that @xmath47 = \frac{\delta s}{2\pi } [ \nabla { \ensuremath{\varphi}}_2(z ) - \nabla { \ensuremath{\varphi}}_1(z ) ] \cdot \hat{n}\ ] ] in which the functions @xmath34 and @xmath35 relate to the limiting values of @xmath24 as the point @xmath8 on the curve @xmath40 is approached from either side , and the vector @xmath48 is the unit vector normal to @xmath40 at that point .
recall that @xmath24 is the real part of the complex free energy @xmath23 and therefore , away from any zeros of the partition function , satisfies the cauchy - riemann equations @xmath49 in which @xmath50 is the imaginary part of @xmath23 and @xmath27 .
then we have that @xmath51 where @xmath52 is the unit vector tangent to the curve @xmath40 at the point @xmath8 .
thus we recognise the scalar product in ( [ density - integral2 ] ) as the directional derivative of the imaginary part of the free energy @xmath23 along the curve @xmath40 . putting this together with ( [ density - integral1 ] ) we find that @xmath53\ ] ] in which @xmath54 $ ] and the subscripts indicate opposite sides of a phase boundary .
let us now assume that in the thermodynamic limit , there is a phase transition at the point @xmath19 .
then , either side of the transition the free energy can be written as @xmath55 where @xmath56 and @xmath57 is valid for @xmath58 and @xmath59 valid for @xmath60 .
note that for the free energy to be real along the real @xmath8 axis , the coefficients @xmath61 and @xmath62 must also be real . from the criterion that the real part of @xmath23 is continuous across a phase boundary , we find that the boundary lies along the curve @xmath63 where @xmath64 and @xmath65 are the real and imaginary parts of @xmath66 respectively .
we consider the conditions under which a discontinuous ( first order ) or continuous ( second or higher order ) appear .
* first - order transition * + for the case where @xmath67 , and the free energy has a discontinuity in its first derivative , the curve of zeros is a hyperbola that passes smoothly through the transition point @xmath19 .
hence the tangent to the curve of zeros is parallel to the imaginary axis at @xmath19 and using the rule ( [ density - rule ] ) we find that @xmath68 \right|_{\tilde{x}=\tilde{y}=0 } = \frac{b_2-b_1}{2\pi } \;.\ ] ] hence we see that the density of zeros at the transition point @xmath19 is nonzero at a first - order phase transition ( i.e. one at which the first derivative of the free energy is discontinuous ) .
* second - order transition * + if @xmath69 but @xmath70 we have that the curve @xmath40 obeys the equation @xmath71 .
since the zeros of the partition function come in complex - conjugate pairs , we find that the zeros approach the point @xmath19 along straight lines that meet at a right - angle .
if we consider the line @xmath72 , we find from ( [ density - rule ] ) that @xmath73 this reveals that at a second - order phase transition , the density of partition function zeros decreases linearly to zero at the phase transition point .
* higher - order transition * + more generally , one can consider a difference in the free energies either side of the transition point to have the leading behaviour @xmath74 .
then the condition that @xmath75=0 $ ] suggests that the curve of zeros approaches the real axis at an angle @xmath76 . unless @xmath77 , the curve does not pass smoothly through the real axis . in any case , the imaginary part of the free energy difference grows as @xmath78 giving rise to a density of zeros that behaves as @xmath79 for small arclength @xmath41 .
this means that for the density of zeros to be finite at the transition point we must have @xmath80 . in this section
we have summarised what we refer to as the lee - yang theory of phase transitions .
this describes how partition function zeros are related to phase transitions : the accumulation of zeros at a point along the physical ( real , positive ) axis of the control parameter gives the critical value and the density of zeros near to the accumulation point determines the order of the phase transition .
we derived a rule ( [ boundary - rule ] ) for locating phase boundaries and a further rule ( [ density - rule ] ) for finding the density of zeros along such boundaries . at a first - order transition
there is a nonzero density of zeros at the transition point whereas the density decays as a power - law to zero at the transition point when the associated phase transition is continuous .
although we have discussed the theory of partition function zeros with reference to the specific system described by the partition function ( [ zspinsys ] ) we should note that the ideas hold much more generally .
firstly , one is not restricted to the canonical ensemble : indeed in the original exposition of the theory @xcite , lee and yang worked in the grand canonical ensemble in which the quantity generalised to the complex plane was not a function of the temperature but rather the chemical potential .
of course , in the equilibrium theory , these intensive field - like quantities play similar mathematical roles and so there is no reason why the lee - yang theory should nt apply to all of them .
for historical reasons , zeros in the complex fugacity ( or chemical potential ) plane are often referred to as lee - yang zeros @xcite and those in the complex temperature plane fisher zeros @xcite .
also it appears that one can just as easily generalise physical ` field - like ' variables ( such as temperature ) or ` fugacity - like ' variables ( such as the quantity @xmath8 considered above ) to the complex plane without altering the properties of the partition - function zero densities at first - order and continuous phase transitions described above .
despite the apparently wide generality of the lee - yang theory of equilibrium phase transitions , proving its validity in the general case is a difficult task ( although we note recent work in this direction @xcite ) .
therefore , most rigorous results , such as those discussed in @xcite , tend to rely on specific properties of a particular partition function .
perhaps the most spectacular of these is that obtained by lee and yang in their original work .
specifically , they found that the zeros of the partition function for an ising ferromagnet in an external field @xmath81 all lie on the circle @xmath82 .
the significance of this result , which does not depend on the number of spatial dimensions , lattice structure and details of the spin - spin interactions , is that if there is a phase transition induced by varying the magnetic field @xmath81 , it can only occur at the point @xmath83 , and then only if the partition function zeros accumulate there in the thermodynamic limit .
a nonequilibrium steady state differs from its equilibrium counterpart in that it may admit a flow of , say , energy or mass . more generally , these states have a circulation of probability in the space of microscopic configurations . in the past few years , it has become customary to model nonequilibrium systems as stochastic processes , i.e. simple models defined by local dynamical rules
. extensive study of these has revealed a range of phenomena including nonequilibrium phase transitions @xcite .
we present below the key elements of this approach to nonequilibrium physics with a view to understanding nonequilibrium phase transitions in the framework of the lee - yang theory described above .
in particular , we will need to propose a quantity to use in place of the equilibrium partition function ( [ zspinsys ] ) .
let us begin by discussing how one might realise the dynamics of the _ equilibrium _ spin system of the previous section , for example in a computer simulation .
the aim is to generate a sequence of spin configurations such that the frequency with which a particular configuration @xmath84 is generated is proportional to the boltzmann weight @xmath85 $ ] where @xmath86 is the energy of configuration @xmath84 .
usually this is achieved by choosing the next configuration @xmath87 from the current configuration @xmath84 with a probability proportional to the transition rate @xmath88 that satisfies the detailed balance condition @xmath89 apart from the fact that this choice of transition rates guarantees convergence to the desired equilibrium distribution of microstates @xcite , it also has the feature that the mean current of any quantity in the steady state is zero ( as it should be at equilibrium ) .
since we are interested in nonequilibrium systems that support steady - state currents we must work with transition rates that do not satisfy the condition ( [ db ] ) or equivalent criteria ( given in e.g. @xcite ) .
clearly one has a lot of freedom in the choice of transition rates , and so in practice one often devises dynamics that seem physically reasonable given the phenomena one is trying to describe .
later , in section [ dds ] we will give concrete examples in the form of driven diffusive systems , in which particle moves are biased in a particular direction to model the effect of an external field . in order to find the set of steady - state weights associated with a particular set of prescribed transition rates , we impose the condition that the total flow of probability into a configuration @xmath84 is balanced by the corresponding outflow .
that is , we must have @xmath90 = 0\ ] ] for every configuration @xmath84 . in general
there might be more than one solution to this set of equations for the rates @xmath91 , each corresponding to a steady state that is reached with some probability that depends on the starting configuration .
we shall assume for simplicity that the steady state is unique and therefore reached with certainty from every initial state .
a sufficient condition for a unique steady state is that it is possible to reach each configuration from every other via a sequence of microscopic transitions @xcite .
once the steady state weights @xmath91 are known from ( [ steady - state ] ) one can obtain the corresponding probability distribution of microscopic configurations @xmath92 through a normalisation @xmath93 such that @xmath94 .
then , one can compute averages of physical quantities and look for nonanalyticities to locate nonequilibrium phase transitions as one varies the transition rates . at equilibrium
we saw that the zeros of the function ( [ zspinsys ] ) that normalises the boltzmann weights encodes the phase behaviour of the system .
thus we might hope that more generally , the zeros of the normalisation ( [ zneqm ] ) will provide information about nonequilibrium phase behaviour .
in section [ neqm ] we review recent work that suggests that the steady - state phase behaviour of certain nonequilibrium driven diffusive and reaction - diffusion systems _ is _ correctly described by the zeros of the normalisation @xmath95 . although we are unaware of any rigorous argument for this to be the case , some suggestive evidence is provided by an observation made in @xcite which we now outline .
first note that the equation ( [ steady - state ] ) is linear in the weights @xmath91 .
this implies that these weights , and hence the normalisation , can always be written as sums of products of the transition rates @xmath96 .
now , by numbering the microscopic configurations , one can construct the transition rate matrix @xmath97 whose off - diagonal elements @xmath98 are equal to the transition rates from configuration @xmath99 to @xmath100 and the diagonal elements @xmath101 are negative and express the total rate of departure from configuration @xmath100 . using elementary results from matrix theory ( see @xcite for the details ) , one can show that an expression for the normalisation @xmath95 that is polynomial in the transition rates @xmath102 is @xmath103 in which @xmath104 are the eigenvalues of the matrix @xmath97 . with each of the eigenvalues @xmath104
is associated an eigenvector of @xmath97 describing a ` mode ' of the stochastic process that decays exponentially with a timescale @xmath105 @xcite .
we have assumed that the process described by @xmath97 has only one steady state , and so only one of the eigenvalues is equal to zero ( since clearly the relaxation time of a steady state is infinite ) .
equation ( [ zevprod ] ) states that the normalisation @xmath95 can be written as a product of the remaining eigenvalues .
now , at a phase transition we expect diverging timescales : physically one encounters metastable ( long - lived ) states near first - order transition points and long correlation lengths and times at continuous transitions .
the presence of long timescales @xmath106 implies small eigenvalues @xmath104 of @xmath97 and , from ( [ zevprod ] ) , that the normalisation @xmath95 approaches zero . hence it appears that it is appropriate to consider the zeros of @xmath95 as given by ( [ zneqm ] ) to locate nonequilibrium phase transitions .
we shall now review recent progress in applying the lee - yang theory to nonequilibrium phase transitions .
we consider first in section [ dds ] driven diffusive systems , where most work has so far been focussed @xcite .
an appealing feature , discussed in more detail below , is that some one - dimensional cases have been solved exactly , the normalisation ( [ zevprod ] ) calculated , and nonequilibrium phase transitions analysed .
( the existence of one - dimensional phase transitions contrasts with the case of one - dimensional equilibrium models which do not admit phase transitions if the interactions are short - ranged . )
then in section [ rdanddp ] we move onto reaction - diffusion systems and directed percolation . in their original papers on partition function zeros , lee and
yang @xcite made use of the mapping between the ising model of a magnet and the lattice gas .
essentially one associates up - spins with particles and down - spins with vacancies so that a positive interaction strength @xmath107 in the ising model results in a particle - particle attraction in the lattice gas , and a negative interaction strength gives rise to repulsive gas particles .
note that there is implicitly at most one particle per lattice site , so there is a hard - core exclusion in the lattice gas .
as discussed in section [ nss ] , one can realise the dynamics of the lattice gas through a set of transition rates that satisfy the detailed balance condition ( [ db ] ) with respect to the boltzmann distribution .
thirty years after lee and yang s work , katz , lebowitz and spohn ( kls ) @xcite introduced a _ driven _ lattice gas model in which the rate at which particles hop in the direction of an external field is enhanced and the hop - rate against the field is suppressed .
this model is well - studied and many results are discussed in @xcite .
the principal effect of the driving field is to introduce anisotropy in the phase - separation that occurs below the critical temperature associated with the spontaneous magnetisation of the ising ferromagnet ( @xmath108 ) in two or more spatial dimensions and in the critical exponents that characterise this transition .
as yet , the kls model remains unsolved for general interaction strength , although in one dimension the steady state is known for some parameters @xcite .
the particular case of one dimension and zero interaction strength @xmath109 , known as the asymmetric simple exclusion process ( asep ) , had already been studied at least at a mean - field level by biophysicists interested in the kinetics of biopolymerisation @xcite .
the mean - field approach predicts phase transitions in the steady state as parameters controlling the rate of insertion and extraction of particles at the boundaries are varied @xcite .
the existence of these phase transitions is confirmed through an exact solution of the asep @xcite , achieved using a powerful matrix product approach @xcite which has subsequently been used to solve many generalisations of the asep .
the details of the matrix product method are not necessary for the following suffice to say that one ends up calculating a normalisation proportional to ( [ zevprod ] ) through a product of matrices , often of infinite dimension . in this way one obtains some explicit formulas for the normalisation ( [ zneqm ] ) which we shall use below .
the asymmetric exclusion process with open boundaries is perhaps the simplest exactly solved nonequilibrium model that exhibits both a first - order and continuous phase transition in its steady state .
therefore it is an ideal candidate for testing the hypothesis outlined in section [ nss ] that zeros of the normalisation should accumulate towards the positive real axis in the complex plane of transition rates . before outlining the results of this analysis ( the details of which
are presented in @xcite ) we recall the definition of the asep with open boundaries .
the dynamics defining the asep .
the lattice is open and particles may be inserted or extracted at the boundaries as shown . ]
the phase diagram of the asep in the space spanned by the boundary rates @xmath110 and @xmath6 .
the thick line represents a first - order phase boundary , the thin line a continuous phase boundary . ] in this system , a particle on an @xmath0-site lattice can hop one site to the right at unit rate , as long as the receiving site is empty .
meanwhile , particles are inserted onto the leftmost lattice site ( if empty ) at a rate @xmath110 and removed from the rightmost site ( if occupied ) at a rate @xmath6see figure [ asep ] . along the line @xmath111 there is a coexistence between a high- and a low - density phase , with a shock separating the two .
this is indicative of a first - order transtion .
meanwhile , along the lines @xmath112 and @xmath113 there is a continuous transition ( i.e. one accompanied by diverging lengthscales ) to a phase in which the particle current is independent of @xmath110 and @xmath6 . the phase diagram for the model is shown in figure [ asep - phase ] .
the steady - state normalisation ( [ zneqm ] ) for the asep with @xmath0 sites has been calculated @xcite as @xmath114 it is a simple matter to use a computer algebra package to solve this equation for its zeros in the complex-@xmath110 plane at fixed @xmath0 and @xmath6 .
( equivalently , one could look at the complex-@xmath6 zeros at fixed @xmath110 since @xmath115 . ) in figure [ asep - alphazeros ] plots of the zeros are shown for @xmath116 , where a continuous transition occurs at @xmath117 , and for @xmath118 , where a first order transition occurs at @xmath119 .
we immediately notice that the curve of zeros seem to intersect the real positive @xmath110 axis at the correct transition point .
furthermore , the density of zeros near the first - order transition point ( @xmath120 ) seems to be uniform and nonzero , whereas the density of zeros near the second - order transition point ( @xmath121 ) seems to decrease to zero .
both of these observations are in accord with the results known for equilibrium partition function zeros discussed in section [ eqm ] .
zeros of the normalisation in the asep in the complex-@xmath110 plane and @xmath122 and lattice size @xmath123 . ]
as shown in @xcite , the distribution of zeros in the thermodynamic limit can be calculated once one knows that , for large @xmath0 , the normalisation behaves as @xmath124 . in this expression , @xmath125 and @xmath126
depend on @xmath110 and @xmath6 and the quantity @xmath107 is the current of particles across the lattice .
thus , the complex free energy ( [ hdef ] ) in the limit @xmath127 is @xmath128 furthermore for @xmath110 smaller than the transition value @xmath129 , @xmath130 whereas for @xmath131 , @xmath132 .
although an electostatic analogy was used in @xcite to find the zero distribution , the mathematical content is the same as that used to derive the two rules ( [ boundary - rule ] ) and ( [ density - rule ] ) in section [ eqm ] .
applying the first rule , which demands that the real part of the free energy be continuous across a phase boundary , we find that the zeros of @xmath133 should lie along the curve @xmath134 .
it is therefore convenient to change variable to @xmath135 .
then , in the complex-@xmath136 plane , the zeros lie on the circle @xmath137 .
the density of zeros on this circle can be found by setting @xmath138 and parametrising points on the circle as @xmath139 .
then , the second rule ( [ density - rule ] ) gives the density of zeros @xmath44 on the circle as @xmath140 = \frac{1}{2\pi\xi_c } \frac{{\ensuremath{\mathrm{d}}}}{{\ensuremath{\mathrm{d}}}\theta } \theta = \frac{1}{2\pi\xi_c } \;.\ ] ] that is , in the complex-@xmath136 plane the zeros should become evenly distributed on a circle in the thermodynamic limit . transforming the zeros of ( [ zasep ] ) obtained at different system sizes to the complex-@xmath136 plane reveals this to be the case @xcite . finally , one can show that near the intersection point between the curve of zeros in the complex-@xmath110 plane and the positive real @xmath110 axis , the zeros of ( [ zasep ] ) sit on the curve @xmath141 where @xmath142 and @xmath143 are the real and imaginary parts of @xmath110 .
for the case @xmath144 , @xmath145 and the transition is first order .
one finds that the curve of zeros is smooth at the transition point @xmath146 , and that the density of zeros is @xmath147 $ ] which is nonzero .
these are precisely the properties of the equilibrium partition function zeros at a first - order transition point ( see section [ eqm ] ) . at the continuous transition point
( @xmath148 and @xmath149 , the zeros pass through the real @xmath110 axis along the line @xmath150 .
recall from section [ eqm ] that a transition of @xmath3 order has the curve of zeros meeting the positive real axis at an angle of @xmath151 . here
the zeros clearly approach at an angle @xmath152 suggesting a second - order transition .
in fact , one finds the density of zeros is @xmath153 at a distance @xmath41 along this line , confirming that the transition is second - order . in summary
, we have found that the lee - yang theory of first - order and continuous phase transitions applies to the normalisation of the nonequilibrium asymmetric exclusion process just as it does to the partition function of equilibrium systems . of course
, this does not prove that the theory is generally applicable , and so there is some value in investigating other nonequilibrium steady states that exhibit phase transitions .
one such state is that initially studied by arndt , heinzel and rittenberg @xcite .
this model comprises @xmath154 ( @xmath155 ) positively ( negatively ) ` charged ' particles on a closed ring of @xmath0 sites .
when next to vacant sites , the positive particles hop to the right and negative particles to the left at unit rate , with hops in the opposite directions disallowed .
meanwhile , should two oppositely charged particles be next to one another , the following transitions can occur @xmath156 where the label above the arrow indicates the rate at which the transitions occur .
these dynamics are illustrated in figure [ ahr ] .
the dynamics defining the ahr model : labels indicate the rates at which the hops denoted by arrows can occur . note that the @xmath0-site lattice is periodic . ] although a matrix product solution for this model is known to exist @xcite , it is technically difficult to work out the solution in its full generality @xcite .
however in the case of either a single vacancy @xcite or a single negative particle , i.e.@xmath157 @xcite the steady - state normalisation is known explcitly and phase transitions have been identified . in the case @xmath157 the zeros of the partition function in the complex-@xmath158 plane have been studied @xcite and , once again the continuous phase transition present in this model is correctly described by the lee - yang theory . a slightly different lee - yang analysis , that precedes the work described above , has been used to study the nonequilbrium phase transition that occurs at a finite density of vacancies @xcite .
the motivation behind this study arose from computer simulations of the model @xcite for the case where the numbers of positive and negative particles were equal . in these simulations , three phases were initially identified : a ` solid ' phase for small @xmath158 , a ` fluid ' phase for large @xmath158 and a mixed phase comprising the background ` fluid ' and a large mobile droplet in an intermediate regime @xmath159 where @xmath160 is some density - dependent quantity .
the steady - state normalisation @xmath161 for the case of @xmath162 positive and negative particles on the @xmath0-site ring is not known exactly ( at least in the canonical ensemble ) , therefore arndt @xcite chose to study the generating function @xmath163 in the complex-@xmath8 plane .
physically this approach is equivalent to placing the ring with nonequilibrium interactions in chemical equilibrium with a particle reservoir at fixed fugacity @xmath8 .
one can check that as one takes the thermodynamic limit , the relative size of the fluctuations in the density @xmath164 vanish , and so working at fixed fugacity @xmath8 is equivalent to working at fixed density @xmath165 .
( in fact , this is a standard ` trick ' for dealing with closed systems in which the particle number is conserved , see e.g. @xcite . ) the key point here , however , is that @xmath8 is an equilibrium fugacity , and not a microscopic transition rate , and so the lee - yang theory of phase transitions described in section [ eqm ] ought to apply directly here , without reference to the discussion of section [ nss ] . it was found @xcite that when @xmath166 the zeros of ( [ ahrgf ] ) in the complex-@xmath8 plane appear to lie along ellipses , intersecting the positive real axis .
conversely , for larger @xmath158 the zeros appear to describe hyperbol that avoid the positive real axis . after investigating numerically the density of zeros in the complex-@xmath8 plane , arndt concluded that there is a fluid - fluid phase transition at some density - dependent point @xmath167 .
unfortunately , an exact asymptotic ( i.e. large @xmath0 ) analysis of the generating function ( [ ahrgf ] ) shows that the only nonanalyticity occurs at the solid - fluid transition point @xmath168 @xcite .
however it was noted that physical quantities vary rapidly , but continuously , at some a point @xmath169 .
this phenomenom has been explained as an abrupt increase in a corrleation length to an anomalously large , but finite value @xcite .
we believe that resolution with the study of arndt lies in the fact that the system sizes considered in @xcite were quite small ( @xmath170 ) , whereas it was suggested @xcite that the distinction between this crossover behaviour and a genuine phase transition might become apparant only once @xmath0 is increased above about @xmath171 ( an unfeasibly huge number ! ) .
it would be interesting , therefore , to extend the numerical computation of zeros performed in @xcite to much larger systems , to see how the ellipses noted above develop .
for example it might well happen that instead of approaching the positive - real fugacity axis , the zeros of ( [ ahrgf ] ) would terminate a short distance away from it . a large and important class of models with stochastic dynamics is provided by reaction - diffusion systems .
in contrast to driven diffusive systems , where the particle - particle interactions imply conservation of particles , reaction - diffusion systems are characterised by dynamics that result in a change in particle number .
moreover there are a number of such systems that have absorbing states i.e. special configurations generally devoid of particles that once entered can not be left .
phase transitions associated with whether the system has a finite probability of not being absorbed into such a state fall within the _ directed percolation _ and related universality classes @xcite .
we shall shortly discuss the second order phase transition associated with the directed percolation university class in a little detail .
meanwhile as a simple example of a reaction - diffusion system , we review a model for which the steady state that can be solved using the matrix product approach .
the approach again provides us with an explicit expression for the normalisation ( [ zneqm ] ) by virtue of which we can analyse its zeros in the plane of complex reaction rates .
the system in question @xcite has for the dynamics at neighbouring bulk sites on a one - dimensional lattice the processes @xmath172 occurring at the rates indicated , and with @xmath173 representing a particle and @xmath174 an empty lattice site .
it was demonstrated @xcite that the matrix product scheme used for the asep could be generalised to cater for the steady state of the present reaction - diffusion system on a lattice of @xmath0 sites with reflecting boundary conditions ( i.e. which is neither periodic nor has particle input or removal at the boundaries ) . in the matrix product approach , the normalisation @xmath175 is given by a scalar derived from @xmath176 where @xmath40 is a square matrix .
if @xmath40 is a finite dimensional @xmath177 matrix and can be diagonalised , the resulting expression for @xmath175 has the form @xmath178 in which @xmath179 is the @xmath3 eigenvalue of @xmath40 and the coefficients @xmath180 arise from the details of the way in which one obtains a scalar from the matrix @xmath176 .
a normalisation of this form leads to a complex free energy @xmath181\ ] ] in which the maximum means the eigenvalue with the largest
_ absolute _ value .
then , there is the possibility of a phase boundary when the magnitude of the two largest eigenvalues of @xmath40 are equal .
it turns out that for the closed reaction - diffusion system introduced above , @xmath40 is a @xmath182 matrix that can be diagonalised when @xmath183 @xcite .
when @xmath184 , @xmath40 can not be diagonalised on account of the largest eigenvalue being degenerate , and a phase transition is found to occur at this point . note that this scenario contrasts with the transfer - matrix approach to one - dimensional equilibrium systems where the partition function is also written as a product of matrices .
since all elements of the transfer matrix are positive the largest eigenvalue can not become degenerate therefore there can be no phase transition @xcite .
however there is no such restriction on the elements of @xmath40 , and so eigenvalue crossing is permitted and nonequilibrium one - dimensional phase transitions can occur .
the structure of the normalisation @xmath185 is even simpler in the case where particles are inserted onto and removed from the left boundary at rates @xmath110 and @xmath6 respectively ( with the right boundary remaining reflecting ) , and these rates satisfy the relation @xmath186 @xcite . in this case
@xmath40 is a @xmath187 matrix and , for @xmath188 , is diagonalisable with eigenvalues @xmath189 we see then immediately from ( [ hmaxev ] ) and an application of the rule ( [ boundary - rule ] ) that there is a phase boundary in the complex @xmath158 plane along the circle @xmath190 . also ,
since one of the eigenvalues does not depend on @xmath158 , the free energy @xmath191 is a constant in one of the phases which , as the analysis of the normalisation zeros for the asep demonstrated , implies that the density of zeros on this circle is constant in the thermodynamic limit . in turn
, this implies that the phase transition is first order , as confirmed by explicitly calculating the density profile in the two phases @xcite .
we finally turn our attention to a process that comprises symmetric decoagulation ( that is , @xmath192 and @xmath193 taking place with equal probability in each direction ) and spontaneous decay ( @xmath194 ) occurring at independent rates . introduced as a crude model of an epidemic @xcite , this _ contact process _
is known to exhibit a transition from a phase in which the absorbing state ( empty lattice ) is reached with certainty to a phase in which there is some probability that the epidemic remains active forever in the thermodynamic ( infinite system size ) limit @xcite .
this transition occurs as the ratio between the decoagulation and decay rates is increased beyond a critical value .
directed percolation .
bonds are open ( solid lines ) with probability @xmath195 and closed ( broken lines ) with probability @xmath196
. a fluid starts at layer @xmath197 and may flow only downwards through open bonds .
bonds carrying the fluid are shown with thick lines , so that the point @xmath42 on layer @xmath198 is connected to the origin whereas point @xmath199 is not . ]
although not proven , it is widely accepted on the basis of simulation and approximate methods ( such as series expansions ) that the transition just described is continuous and characterised by _ directed percolation _ ( dp ) exponents ( see @xcite for an in - depth discussion of these issues ) .
directed percolation itself @xcite is a geometric construction designed to model fluid flow through a random porous medium under the influence of gravity . consider the rhombic lattice shown in figure [ dp ] and imagine ` pouring ' fluid in from the uppermost site .
the fluid is allowed to flow along a bond pointing diagonally downwards and then only if it is open .
each bond has a probability @xmath195 of being open ( and hence a probability @xmath196 of being blocked ) , and the state of each bond is independent of any other and may not change with time .
the main quantity of interest in this system is the percolation probability @xmath200 that the fluid can penetrate to a depth @xmath100 in the medium ( see figure [ dp ] ) , averaged over all possible bond configurations .
the order parameter for the model is the probability @xmath201 that the fluid can penetrate infinitely far given a bond probability @xmath195 .
if @xmath195 is below some critical percolation threshold @xmath202 , the fluid only ever penetrates a finite distance i.e .
a layer becomes dry with certainty .
however , for larger @xmath195 , the order parameter becomes nonzero , growing as @xmath203 with @xmath204 near the critical point .
the importance of the dp transition is that a wide range of models that have a transition into an absorbing state are expected to have that transition characterised by the dp exponents , one of which is @xmath6 @xcite . despite a huge amount of interest in directed percolation
, none of the models expected to belong to its universality class has been solved exactly .
recently an attempt has been made to shed further light on the dp transition by studying the zeros of the percolation probability @xmath200 @xcite . at first glance ,
a connection between this probability and a partition function is not obvious .
however , associating with each site a ` spin ' state @xmath205 , that is up if site @xmath206 is connected to a point on the @xmath3 layer and down otherwise , yields a form for @xmath200 that has the structure of a transfer - matrix representation of a partition function for an equilibrium system with three - spin interactions @xcite .
it is not clear that being able to give @xmath200 the appearance of a partition function necessarily implies that the lee - yang theory should hold . in particular
, @xmath200 has some features that set it apart from ` standard ' equilibrium partition functions , in that it does not grow exponentially in the number of bonds @xmath0 , at least in the range @xmath207 .
it is also uncertain whether a free energy can meaningfully be defined for this system , especially in the region @xmath208 in which @xmath209 . given these observations , it is perhaps not surprising that the zeros of @xmath200 calculated numerically for @xmath210 ( leading to polynomials of degree up to @xmath211 ) have a more complicated distribution than for the ( exactly solved ) models considered so far . as is evident from figure [ ddh - zeros ] , which was presented in @xcite , the zeros do not lie along a single curve but along a sequence of curves meeting at the critical point ( as well as inside some region that encloses part of the real axis for @xmath212 ) . by considering the sequences of zeros approaching the positive real axis
, one can estimate the critical point @xmath202 and the density of zeros as it is reached @xcite .
the latter yields a prediction for one of the dp critical exponents and one finds good agreement with the most precise estimates of both the transition point and the associated exponent obtained through other means .
hence we have further evidence for the applicability of lee and yang s ideas concerning partition function zeros to a much wider range of statistical distributions than equilibrium steady states .
in this work we have revisited the lee - yang description of equilbrium phase transitions with a view to seeing whether the ideas apply to more general nonequilibrium transitions .
recently there have been a number of studies of zeros of partition - function - like quantities that arise in systems with nonequilibrium dynamics , and we have seen in our review of these works that the lee - yang theory , as described in section [ eqm ] , seems to hold quite generally . we have argued that for dynamic models with a unique steady state , the normalisation defined as a sum over the steady - state configurational weights ( [ zneqm ] ) serves as a suitable ` partition function ' in the sense that its zeros , in the complex plane of any model parameter , should accumulate towards physical transition points in the thermodynamic limit .
furthermore , the density of zeros and angle of approach to the real axis indicate whether the transition is first - order ( manifested physically through phase - coexistence ) or continuous ( i.e. characterised by divergent correlation lengths and times ) .
thus studying the zeros of the normalisation ( [ zneqm ] ) provides an unambiguous classification of nonequilibrium phase transitions as do the lee - yang zeros in the equilibrium case .
the observation that backs up this scenario is embodied by equation ( [ zevprod ] ) which reveals that the reciprocal of the steady - state normalisation is equal to the product of the characteristic relaxation times in the dynamics . since near a phase transition
one expects timescales to diverge , one also expects the normalisation to approach zero .
however , a rigorous argument for this to be the case is still lacking .
moreover ( [ zevprod ] ) implies a possible link between systems for which the steady state normalisation can be calculated and those for which eigenvalues of the transition matrix can in principle be calculated .
a different class of systems encompasses those whose steady state is not unique .
the contact process is , in fact , an example of such a model , in which the absorbing state is reached with certainty below the critical decoagulation rate , whereas above it , and on an infinite system , a second steady state can also be reached with some nonzero probability .
since this additional steady state exists only when the lattice size becomes infinite , one must take that limit first , before taking time to infinity .
otherwise , on a finite system the steady state is simply the absorbing state and the steady - state normalisation is trivially equal to a constant .
nevertheless the work of @xcite , which we reviewed in section [ rdanddp ] , indicates that the lee - yang theory can be relevant when one considers other properties of the nonequilibrium system such as the percolation probability .
although we have not discussed this in great detail here , it should be noted that the lee - yang approach gives a method for extrapolating to the thermodynamic limit from solutions for small system sizes . in the work of @xcite ,
numerical solutions for small systems were used successfully to estimate the transition point and density of zeros as it is approached . from this information
one learns about the nature of the phase transition and , for example , can estimate the values of critical exponents .
this is a common technique in equilibrium statistical physics ( see e.g. @xcite ) and it may be the case that stochastic processes unyielding to analytical treatment could be understood this way .
d. mukamel ( 2000 ) phase transitions in nonequilibrium systems . in _ soft and fragile matter _ , edited by m. e. cates and m. r. evans , volume 53 of _ scottish universities summer school in physics _
, page 237 ( institute of physics publishing , bristol ) .
g. m. schtz ( 2001 ) exactly solvable models for many - body systems far from equilibrium . in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz , volume 19 , page 1 ( academic press , london ) .
p. f. arndt , t. heinzel and v. rittenberg ( 1999 ) spontaneous breaking of translational invariance and spatial condensation in stationary states on a ring : i. the neutral system .
_ j. stat .
phys . _ * 97 * 1 . s. redner ( 1997 ) scaling theories of diffusion - controlled and ballistically - controlled bimolecular reactions . in _
nonequilibrium statistical mechanics in one dimension _
, edited by v. privman , chapter 1 ( cambridge university press , cambridge ) . | we present a pedagogical account of the lee - yang theory of equilibrium phase transitions and review recent advances in applying this theory to nonequilibrium systems . through both general considerations and explicit studies of specific models ,
we show that the lee - yang approach can be used to locate and classify phase transitions in nonequilibrium steady states . |
quantum machine learning aims at merging the methods from quantum information processing and pattern recognition to provide new solutions for problems in the areas of pattern recognition and image understanding @xcite . in the first aspect the research in this area
is focused on the application of the methods of quantum information processing @xcite for solving problems related to classification and clustering @xcite .
one of the possible directions in this field is to provide a representation of computational models using quantum mechanical concepts . from the other perspective
the methods for classification developed in computer engineering are used to find solutions for problems like quantum state discrimination @xcite , which ares tightly connected with the recent developments in quantum cryptography .
+ using quantum states for the purpose of representing patterns is naturally motivated by the possibility to exploit quantum algorithms to boost the computational intensive parts of the classification process .
in particular , it has been demonstrated that quantum algorithms can be used to improve the time complexity of the @xmath2__nearest neighbor _ _ ( @xmath3nn ) method . using the algorithms presented in @xcite it is possible to obtain polynomial reductions in query complexity in comparison to the corresponding classical algorithm .
another motivation comes from the possibility of using quantum - inspired algorithms for the purpose of solving classical problems .
such an approach has been exploited by various authors . in @xcite authors
propose an extension of gaussian mixture models by using the statistical mechanics point of view . in their approach
the probability density functions of conventional gaussian mixture models are expressed by using density matrix representations . on the other hand , in @xcite authors
utilize the quantum representation of images to construct measurements used for classification
. such approach might be particularly useful for the physical implementation of the classification procedure on quantum machines . in the last few years , many efforts to apply the quantum formalism to non - microscopic contexts @xcite and to signal processing @xcite have been made .
moreover , some attempts to connect quantum information to pattern recognition can be found in @xcite .
exhaustive survey and bibliography of the developments concerning applications of quantum computing in computational intelligence are provided in @xcite . even if these results seem to suggest some possible computational advantages of an approach of this sort , an extensive and universally recognized treatment of the topic is still missing @xcite .
the main contribution of our work is the introduction of a new framework to encode the classification process by means of the mathematical language of density matrices @xcite .
we show that this representation leads to two different developments : @xmath0 it enables us to provide a representation of the _ nearest mean classifier _
( nmc ) in terms of quantum objects ; @xmath4 it can be used to introduce a _ quantum classifier _ ( qc ) that can provide a significative improvement of the performances on a classical computer with respect to the nmc .
the paper is organized as follows . in section [ sec : pre ] basic notions of quantum information and pattern recognition are introduced . in section [ sec : pdp ] we formalize the correspondence between arbitrary two - feature patterns and pure density operators and we define the notion of _ density pattern_. in section [ sec : cp ] we provide a representation of nmc by using density patterns and by the introduction of an _ ad hoc _ definition of distance between quantum states . in section [ qcp ]
is devoted to describe a new quantum classifier qc that has not a classical counterpart in the standard classification process .
numerical simulations for both qc and nmc are presented . in section [ sec : general ] a geometrical idea to generalize the model to arbitrary @xmath5-feature patterns is proposed .
finally , in section [ sec : concl ] concluding remarks and further developments are discussed .
in the standard quantum information theory @xcite , the states of physical systems are described by unit vectors and their evolution is expressed in term of unitary matrices ( _ i.e. _ quantum gates ) . however
, this representation can be applied for an ideal case only , because it does not take into account some unavoidable physical phenomena , such as interactions with the environment and irreversible transformations . in modern quantum information theory
@xcite , another approach is adopted .
the states of physical systems are described by density operators also called _ mixed states _
@xcite and their evolution is described by quantum operations .
the space @xmath6 of density operators for @xmath5-dimensional system consists of positive semidefinite matrices with unit trace .
a quantum state can be _ pure _ or _
mixed_. we say that a state of a physical system is pure if it represents `` maximal '' information about the system , _
i.e. _ an information that can not be improved by further observations . a probabilistic mixture of pure states is said to be a _
mixed _ state .
generally , both pure and mixed states are represented by density operators , that are positive and hermitian operators ( with unitary trace ) living in a @xmath5-dimensional complex hilbert space @xmath7 .
formally , a density operator @xmath8 is pure iff @xmath9 and it is mixed iff @xmath10 .
+ if we confine ourselves in the @xmath11-dimensional hilbert space @xmath7 , a suitable representation of an arbitrary density operator @xmath12 is provided by @xmath13 where @xmath14 are the pauli matrices .
this expression comes to be useful in order to provide a geometrical representation of @xmath8 .
indeed , each density operator @xmath12 can be geometrically represented as a point of a radius - one sphere centered in the origin ( the so called _ bloch sphere _ ) , whose coordinates ( _ i.e. _ _ pauli components _ ) are @xmath15 ( with @xmath16 ) . by using the generalized pauli matrices
@xcite it is also possible to provide a geometrical representation for an arbitrary @xmath5-dimensional density operator , as it will be showed in section [ sec : general ] .
again , by restricting to a @xmath11-dimensional hilbert space , the points on the surface of the bloch sphere represent pure states , while the inner points represent mixed states .
quantum formalism turns out to be very useful not only in the microscopic scenario but also to encode classical data .
this has naturally suggested several attemps to represent the standard framework of machine learning through the quantum formalism @xcite . in particular ,
pattern recognition @xcite is the scientific discipline which deals with theories and methodologies for designing algorithms and machines capable of automatically recognizing `` objects '' ( i.e. patterns ) in noisy environments .
some typical applications are multimedia document classification , remote - sensing image classification , people identification using biometrics traits as fingerprints .
a pattern is a representation of an object
. the object could be concrete ( i.e. , an animal , and the pattern recognition task could be to identify the kind of animal ) or an abstract one ( i.e. a facial expression , and the task could be to identify the emotion expressed by the facial expression ) .
the pattern is characterized via a set of measurements called _ _
features__. features can assume the forms of categories , structures , names , graphs , or , most commonly , a vector of real number ( feature vector ) @xmath17 . intuitively , a class is the set of all similar patterns . for the sake of simplicity , and without loss of generality , we assume that each object belongs to one and only one class , and we will limit our attention to 2-class problems .
for example , in the domain of ` cats and dogs ' we can consider the classes @xmath18 ( the class of all cats ) and @xmath19 ( the class of all dogs ) . the pattern at hand is either a cat or a dog , and a possible representation of the pattern could consist in the height of the pet and the length of its tail . in this way ,
the feature vector @xmath20 is the pattern representing a pet whose height and length of the tail are @xmath21 and @xmath22 respectively .
+ now , let us consider an object @xmath23 whose membership class is unknown .
the basic aim of the classification process is to establish which class @xmath23 belongs to . to reach this goal ,
standard pattern recognition designs a _ classifier _ that , given the feature vector @xmath23 , has to determine the true class of the pattern .
the classifier should take into account all the available information about the task at hand ( i.e. , information about the statistical distributions of the patterns and information obtained from a set of patterns whose true class is known ) .
this set of patterns is called ` training set ' , and it will be used to define the behavior of the classifier . + if no information about the statistical distributions of the patterns is available , an easy classification algorithm that could be used is the _
nearest mean classifier _ ( nmc ) @xcite , or minimum distance classifier .
the nmc * computes the centroids of each class , using the patterns on the training set @xmath24 where @xmath25 is the number of patterns of the training set belonging to the class @xmath26 ; * assigns the unknown pattern @xmath23 to the class with the closest centroid . in the next section we provide a representation of arbitrary 2d patterns by means of density matrices , while in section [ sec : cp ] we introduce a representation of nmc in terms of quantum objects .
let @xmath27 be a generic pattern , _
i.e. _ a point in @xmath28 . by means of this representation ,
we consider all the @xmath3 features of @xmath29 as perfectly known .
therefore , @xmath29 represents a maximal kind of information , and its natural quantum counterpart is provided by a pure state . for the sake of simplicity
, we will confine ourselves to an arbitary two - feature pattern indicated by @xmath30 in this section , a one - to - one correspondence between each pattern and its corresponding pure density operator is provided .
+ the pattern @xmath31 can be represented as a point in @xmath32 .
the stereographic projection @xcite allows to unequivocally map any point @xmath33 of the surface of a radius - one sphere @xmath34 ( except for the north pole ) onto a point @xmath35 of @xmath32 as @xmath36 the inverse of the stereographic projection is given by @xmath37 therefore , by using the bloch representation given by eq .
( [ bl ] ) and placing @xmath38 we obtain the following definition .
[ dp ] given an arbitrary pattern @xmath35 , the density pattern ( dp ) @xmath39 associated to @xmath31 is the following pure density operator @xmath40 it is easy to check that @xmath41 .
hence , @xmath39 always represents a pure state for any value of the features @xmath42 and @xmath43 .
+ following the standard definition of the bloch sphere , it can be verified that @xmath44 with @xmath45 and @xmath14 are pauli matrices .
let us consider the pattern @xmath46 .
the corresponding @xmath39 reads @xmath47 the introduction of the density pattern leads to two different developments .
the first is showed in the next section and consists in the representation of the nmc in quantum terms . moreover , in section [ qcp ] , starting from the framework of density patterns , it will be possible to introduce a quantum classifier that exhibits better performances than the nmc .
as introduced in section [ sec : pre ] , the nmc is based on the computation of the minimum euclidean distance between the pattern to be classified and the centroids of each class . in the previous section , a quantum counterpart of an arbitrary classical " pattern was provided . in order to obtain a quantum counterpart of the standard classification process
, we need to provide a suitable definition of distance @xmath48 between dps . in addition to satisfy the standard conditions of metric , the distance @xmath48 also needs to satisfy the _ preservation of the order _ : given three arbitrary patterns @xmath49 such that @xmath50 , if @xmath51 are the dps related to @xmath49 respectively , then @xmath52 in order to fulfill all the previous conditions , we obtain the following definition . [ ntr ] the normalized trace distance @xmath53 between two arbitrary density patterns @xmath54 and @xmath55 is given by formula @xmath56 where @xmath57 is the standard trace distance , @xmath58 , with @xmath59 representing the eigenvalues of @xmath60 @xcite , and @xmath61 is a normalization factor given by @xmath62 , with @xmath63 and @xmath64 representing the third pauli components of @xmath54 and @xmath55 , respectively .
[ ntd ] given two arbitrary patterns @xmath65 and @xmath66 and their respective density patterns , @xmath54 and @xmath55 , we have that @xmath67 it can be verified that the eigenvalues of the matrix @xmath60 are given by @xmath68 using the definition of trace distance , we have @xmath69 by applying formula ( [ r123 ] ) to both @xmath63 and @xmath64 , we obtain that @xmath70 using proposition [ ntd ] , one can see that the normalized trace distance @xmath53 satisfies the standard metric conditions and the preservation of the order . due to the computational advantage of a quantum algorithm able to faster calculate the euclidean distance @xcite , the equivalence between the normalized trace distance and the euclidean distance turns out to be potentially beneficial for the classification process we are going to introduce .
let us now consider two classes , @xmath71 and @xmath72 , and the respective centroids and @xmath73 do not represent true centroids , but centroids estimated on the training set . ]
@xmath74 and @xmath75 the classification process based on nmc consists of finding the space regions given by the points closest to the first centroid @xmath76 or to the second centroid @xmath73 . the patterns belonging to the first region
are assigned to the class @xmath71 , while patterns belonging to the second region are assigned to the class @xmath72 .
the points equidistant from both the centroids represent the _ discriminant function _
( df ) , given by @xmath77 thus , an arbitrary pattern @xmath78 is assigned to the class @xmath71 ( or @xmath72 ) if @xmath79 ( or @xmath80 ) .
+ let us notice that the eq .
( [ df ] ) is obtained by imposing the equality between the euclidean distances @xmath81 and @xmath82 .
similarly , we obtain the quantum counterpart of the classical discriminant function .
let @xmath83 and @xmath84 be the dps related to the centroids @xmath76 and @xmath73 , respectively .
then , the _ quantum discriminant function ( qdf ) _ is defined as @xmath85 where * @xmath86 , * @xmath87 , @xmath88 are pauli components of @xmath83 and @xmath84 respectively , * @xmath89 * @xmath90 in order to find the @xmath91 , we use the equality between the normalized trace distances @xmath92 and @xmath93 , where @xmath94 is a generic dp with pauli components @xmath95 , @xmath96 , @xmath97 .
we have @xmath98 the equality @xmath99 reads @xmath100 in view of the fact that @xmath83 , @xmath84 and @xmath101 are pure states , we use the conditions @xmath102 and we get @xmath103 this completes the proof .
+ similarly to the classical case , we assign the dp @xmath94 to the class @xmath71 ( or @xmath72 ) if @xmath104 ( or @xmath105 ) .
geometrically , eq . ( [ qdf ] ) represents the surface equidistant from the dps @xmath83 and @xmath84 .
+ let us remark that , if we express the pauli components @xmath87 , @xmath88 and @xmath106 in terms of classical features by eq .
( [ r123 ] ) , then eq . ( [ qdf ] ) exactly corresponds to eq .
( [ df ] ) . as a consequence ,
given an arbitrary pattern @xmath78 , if @xmath107 ( or @xmath108 ) then its relative dp @xmath94 will satisfy @xmath109 ( or @xmath110 , respectively ) .
+ the comparison between the classical and quantum discrimination functions for the _ moon _ dataset is presented in fig .
[ fig : discr - compare ] . plots in figs .
[ fig : class - moon ] and [ fig : quant - moon ] present the classical and quantum discrimination , respectively .
it is worth noting that the correspondence between pattern expressed as a feature vector ( according to the standard pattern recognition approach ) and pattern expressed as a density operator is quite general .
indeed , it is not related to a particular classification algorithm ( nmc , in the previous case ) nor to the specific metric at hand ( the euclidean one ) .
therefore , it is possible to develop a similar correspondence by using other kinds of metrics and/or classification algorithms , different from nmc , adopting exactly the same approach .
this result suggests potential developements which consist in finding a quantum algorithm able to implement the normalized trace distance between density patterns .
so , it would correspond to implement the nmc on a quantum computer with the consequent well known advantages .
the next section is devoted to explore another developement , that consists in using the framework of density patterns in order to introduce a purely quantum classification process ( without any classical counterpart ) more convenient than the nmc on a classical computer .
in section [ sec : cp ] we have shown that the nmc can be expressed by means of quantum formalism , where each pattern is replaced by a corresponding density pattern and the euclidean distance is replaced by the normalized trace distance .
representing classical data in terms of quantum objects seems to be particularly promising in quantum machine learning . quoting lloyd et al.@xcite `` estimating distances between vectors in @xmath111-dimensional vector spaces takes time @xmath112 on a quantum computer .
sampling and estimating distances between vectors on a classical computer is apparently exponentially hard '' .
this convenience was already exploited in machine learning for similar tasks @xcite .
hence , finding a quantum algorithm for pattern classification using our proposed encoding could be particularly beneficial to speed up the classification process and it can suggest interesting developments , that , however , are beyond the scopes of this paper . what we propose in this section is to exhibit some explicative examples to show how , on a classical computer , our formulation can lead to meaningful improvements with respect to the standard nmc .
we also show that these improvements could be further enhanced by combining classical and quantum procedures . in order to get a real advantage in the classification process we need to be not confined in a pure representation of the classical procedure in quantum terms . for this reason
, we introduce a purely quantum representation where we consider a new definition of centroid .
the basic idea is to define a _
quantum centroid _ not as the stereographic projection of the classical centroid , but as a convex combination of density patterns . *
( quantum centroid ) * given a dataset @xmath113 with + @xmath114 let us consider the respective set of density patterns @xmath115 the quantum centroid is defined as : @xmath116 generally , @xmath117 is a mixed state that has not an intuitive counterpart in the standard representation of pattern recognition , but it turns out to be convenient in the classification process .
indeed , the quantum centroid includes some further information that the classical centroid generally descards .
in fact , the classical centroid does not involve all the information about the distribution of a given dataset , _
i.e. _ the classical centroid is invariant under uniform scaling transformations of the data .
consequently , the classical centroid does not take into account any dispersion phenomena .
standard pattern recognition conpensates for this lack by involving the covariance matrix @xcite .
+ on the other hand the quantum centroid is not invariant under uniform scaling . let us consider the set of @xmath5 points @xmath113 where @xmath114 and let @xmath118 be the respective classical centroid .
a uniform rescaling of the @xmath5 points of the dataset corresponds to move each point @xmath119 along the line joining itself with @xmath120 , whose generic expression is given by : @xmath121 let @xmath122 be a generic point on this line .
obviously , a uniform rescaling of @xmath119 by a real factor @xmath123 is represented by the map : @xmath124 even if the classical centroid is not dependent on the rescaling factor @xmath123 , on the other hand the expression of the quantum centroid is : @xmath125 that , clearly , is dependent on @xmath126 according to the same framework used in section [ sec : cp ] , given two classes @xmath71 and @xmath72 of real data , let @xmath127 and @xmath128 the respective quantum centroids .
given a pattern @xmath129 and its respective density pattern @xmath130 , @xmath129 is assigned to the class @xmath71 ( or @xmath72 ) if @xmath131 ( or @xmath132 , respectively ) .
let us remark that we do not need any normalization parameter to be added to the trace distance @xmath133 , because the exact correspondence with the euclidean distance is no more a necessary requirement in this framework .
from now on we refer to the classification process based on density patterns , quantum centroids and trace distances as the _ quantum classifier _ ( qc ) .
we have shown that the quantum centroid is not independent on the dispersion of the patterns and , intuitively , it could contain some additional information with respect to the classical centroid .
consequently , it is reasonable to expect that qc could provide some better performances than the nmc
. the next subsection will be devoted to exploit this convenience by means of numerical simulations on different datasets . + before presenting the experimental results , let briefly remark in what consists the `` convenience '' of a classification process with respect to another . in order to evaluate the performances of a supervised learning algorithm , for each class
it is tipical to refers to the respective confusion matrix @xcite .
it is based on four possible kinds of outcome after the classification of a certain pattern : * true positive ( tp ) : pattern correctly assigned to its class ; * true negative ( tn ) : pattern correctly assigned to another class ; * false positive ( fp ) : pattern uncorrectly assigned to its class ; * false negative ( fn ) : pattern uncorrectly assigned to another class . according to above ,
it is possible to recall the following definitions able to evaluate the performance of an algorithm with tp .
similarly for tn , fp and fn . ] .
+ true positive rate ( tpr ) , or sensitivity or recall : @xmath134 ; false positive rate ( fpr ) : @xmath135 ; true negative rate ( tnr ) : @xmath136 ; false negative rate ( fnr ) : @xmath137 .
+ let us consider a dataset of @xmath120 elements allocated in @xmath138 different classes .
we also recall the following basic statistical notions : * error : @xmath139 * accuracy : @xmath140 * precision : @xmath141 further , another statistical index that is very useful to indicate the reliability of a classification process is given by the cohen s @xmath3 , that is @xmath142 , where @xmath143 and @xmath144 .
the value of @xmath3 is such that @xmath145 , where the case @xmath146 corresponds to a perfect classification procedure . in this subsection
we implement the qc on different datasets and we show the difference between qc and nmc in terms of the values of error , accuracy , precision and other probabilistic indexes summarized above .
+ we will show how our quantum classification procedure exhibits a convenience with respect to the nmc on a classical computer by using different datasets .
+ we refer to the following very popular two - features datasets , extracted from common machine learning repositories : the _ gaussian _ and the _ moon _
datasets , composed of @xmath147 patterns allocated in two different classes , the _ banana _ dataset , composed of @xmath148 patterns allocated in two classes and the _ 3classgaussian _ , composed of @xmath149 patterns allocated in three classes .
this dataset consists of @xmath147 patterns allocated in two classes ( with equal size ) , following gaussian distributions whose means are @xmath150 , @xmath151 and covariance matrices are @xmath152 , @xmath153 , respectively . + as depicted in figure 2 , the classes appear particularly mixed and the qc is able to classify a number of true positive patterns that is significantly larger than the nmc . hence , the error of the qc is ( about @xmath154 ) smaller than the error of the nmc .
in particular , the qc turns out to be strongly beneficial in the classification of the patterns of the second class .
further , also the values related to accuracy , precision and the other statistical indexes exhibit relevant inprovements with respect to the nmc .
+ on the other hand , there are some patterns correctly classified by the nmc which are neglected by the qc . on this basis , exploiting their complementarity , it makes sense to consider a combination of both classifiers .
the so - called _ oracle _ is an hypothetical selection rule that , for each pattern , is able to select the most appropriate classifier .
its aim is to show the potentiality of an ensemble of classifiers ( in this case , qc and nmc ) if we were able to select the most appropriate classifier depending on the test pattern .
fig 2(d ) shows the effect of the oracle whose performances are summarized in table 1 .
these performances represent the theoretical upper bound of the ensemble composed by qc and nmc . + we denote the variables listed in the tables as follows : e= error ; ei= error on the class i ; ac= accuracy ; pr= precision ; k = cohen s k ; tpr= true positive rate ; fpr = false positive rate ; tnr= true negative rate ; fnr= false negative rate .
let us remark that : _ i _ ) the values listed in the table are referred to the mean values over the classes ; _ ii _ ) in case the number of classes is equal to @xmath11 , is @xmath155 , @xmath156 and @xmath157 . lll c c c c c c
c&e&e1&e2&pr&k&tpr&fpr + + nmc&*0.445&0.41&0.48&*0.555&*0.11&0.555&0.445 + qc&*0.24&0.28&0.2&*0.762&*0.52&0.76&0.24 + nmc - qc&*0.13&0.14&0.12&*0.87&*0.74&0.87&0.13 + * * * * * * * * * this dataset consists of @xmath147 patterns equally distributed in two classes . in this case , the correctly classified patterns of the first class are exactly the same for both classifiers but the qc turns out to be beneficial in the classification of the second class .
+ differently from the gaussian dataset , for this dataset the patterns correctly classified by the nmc are a proper subset of the ones correctly classified by the qc . on this basis , the qc is fully convenient with respect to the nmc and a combination of the two classifiers is useless . .moon
dataset [ gd ] [ cols="<,<,^,^,^,^,^,^,^ " , ] even if the previous examples have shown how the qc can be particularly beneficial with respect to the nmc , according to the well known _ no free lunch theorem _
@xcite , there is no a classifier whose performance is better than the others for any dataset @xcite .
this paper is focused on the comparison between the nmc and the qc because these methods are exclusively based on the pattern - centroid distance .
anyway , a widely comparison among the qc and other commonly used classifiers ( such as the lda - linear discriminant analysis - and the qda - quadratic discriminant analysis - ) will be proposed for future works , where also other quantum metrics ( such as the fidelity , the bures distance etc ) instead of the trace distance will be considered to provide an adaptive version of the quantum classifier .
in section [ sec : pdp ] we provided a representation of an arbitrary two - feature pattern @xmath31 in the terms of a point on the surface of the bloch sphere @xmath34 , _
i.e. _ a density operator @xmath39 . a geometrical extension of this model to the case of @xmath5-feature patterns inspired by quantum framework is possible .
+ in this section we introduce a method for representing an arbitrary @xmath5-dimensional real pattern as a point in the radius - one hypersphere @xmath158 , centered in the origin . a quantum system described by a density operator @xmath8 in an @xmath5-dimensional hilbert space @xmath7 , can be represented by a linear combination of the @xmath5-dimensional identity i and @xmath159 @xmath160-square matrices @xmath161 ( _ i.e. _ _ generalized pauli matrices _
@xcite ) : @xmath162 where the real numbers @xmath106 are the pauli components of @xmath8 . hence , by eq . ( [ rn ] )
, a density operator @xmath8 acting on an @xmath5-dimensional hilbert space can be geometrically represented as a @xmath163-dimensional point @xmath164 in the bloch hypersphere @xmath165 , with @xmath166 .
therefore , by using the generalization of the stereographic projection @xcite we obtain the vector @xmath167 , that is the correspondent of @xmath129 in @xmath168 in fact , the generalization of eqs .
( [ sp])([sp1 ] ) are given by @xmath169 @xmath170 hence , by eq .
( [ gsp ] ) , a @xmath11-dimensional density matrix is determined by three pauli components and it can be mapped onto a @xmath11-dimensional real vector .
analogously , a @xmath171-dimensional density matrix is determined by eight pauli components and it can be mapped onto a @xmath172dimensional real vector . generally , an @xmath5-dimensional density matrix is determined by @xmath173 pauli components and it can be mapped onto an @xmath174 dimensional real vector . now , let consider an arbitrary vector @xmath175 with @xmath176 . in this case eq .
( [ gsp1 ] ) can not be applied because @xmath177 in order to represent @xmath178 in an @xmath5-dimensional hilbert space , it is sufficient to involve only @xmath179 pauli components ( instead of all the @xmath173 pauli components of the @xmath5-dimensional space ) .
hence , we need to project the bloch hypersphere @xmath180 onto the hypersphere @xmath181 .
we perform this projection by using eq .
( [ gsp1 ] ) and by assigning some fixed values to a number of pauli components equal to @xmath182 . in this way
, we obtain a representation in @xmath181 that involves @xmath179 pauli components and it finally allows the representation of an @xmath138-dimensional real vector . let us consider a vector @xmath183 by eq .
( [ gsp1 ] ) we can map @xmath31 onto a vector @xmath184 hence , we need to consider a @xmath171-dimensional hilbert space @xmath7 .
then , an arbitrary density operator @xmath185 can be written as @xmath186 with @xmath106 pauli components such that @xmath187 and @xmath161 generalized pauli matrices . in this case
@xmath161 is the set of eight @xmath188 matrices also known as _ gell - mann matrices _ , namely @xmath189 consequently , the generic form of a density operator @xmath8 in the @xmath171-dimensional hilbert space is given by @xmath190 then , for any @xmath8 it is possible to associate an @xmath191-dimensional bloch vector @xmath192 .
however , by taking @xmath193 for @xmath194 we obtain @xmath195 that , by eq .
( [ gsp1 ] ) , can be seen as point projected in @xmath196 where @xmath197 the generalization introduced above , allows the representation of arbitrary patterns @xmath198 as points @xmath199 also the classification procedure introduced in section [ sec : cp ] can be naturally extended for an arbitrary @xmath5-feature pattern where the normalized trace distance between two dps @xmath54 and @xmath55 can be expressed using eq .
( [ gsp ] ) in terms of the respective pauli components as @xmath200 ^ 2}}{(1-r_{a_{n+1}})(1-r_{b_{n+1}})}.\ ] ] analogously , also the qc could be naturally extended to a @xmath5-dimesional problem ( without lost of generality ) by introducing a @xmath5-dimensional quantum centroid .
in this work a quantum representation of the standard objects used in pattern recognition has been provided . in particular , we have introduced a one - to - one correspondence between two - feature patterns and pure density operators by using the concept of _ density patterns_. starting from this representation , firstly we have described the nmc in terms of quantum objects by introducing an _ ad hoc _ definition of _ normalized trace distance_. we have found a quantum version of the discrimination function by means of pauli components .
the equation of this surface was obtained by using the normalized trace distance between density patterns and geometrically it corresponds to a surface that intersects the bloch sphere .
this result could be considered potentially useful because suggests to find an appropriate quantum algorithm able to implement the normalized trace distance between density patterns . in this way
, we could reach a replacement of the nmc in a quantum computer , with a consequent significative reduction of the computational complexity of the process . secondly , the definition of a _ quantum centroid _ that has not any kind of classical counterpart permits to introduce a purely quantum classifier .
the convenience of using this new quantum centroid lies in the fact that it seems to contain some additional information with respect to the classical one because the first takes into account also the distribution of the patterns .
the main implementative result of the paper consists in showing how the quantum classifier performs a meaningful reduction of the error and improvement of the accuracy , precision and other statistical parameters of the algorithm with respect to the nmc .
further developments will be devoted to compare our quantum classifier with other kinds of commonly used classical classifiers .
finally , we have presented a generalization of our model that allows to express arbitrary @xmath5-feature patterns as points on the hypersphere @xmath201 , obtained by using the generalized stereographic projection .
however , even if it is possible to associate points of a @xmath5-hypersphere to @xmath5-feature patterns , those points do not generally represent density operators . in @xcite
the authors found some conditions that guarantee the one - to - one correspondence between points on particular regions of the hypersphere and density matrices .
a full development of our work is therefore intimately connected to the study on the geometrical properties of the generalized bloch sphere .
this work has been partly supported by the project `` computational quantum structures at the service of pattern recognition : modeling uncertainty '' [ crp-59872 ] funded by regione autonoma della sardegna , l.r . 7/2007 ( 2012 ) .
d. aerts , b. dhooghe .
classical logical versus quantum conceptual thought : examples in economics , decision theory and concept theory .
, _ lecture notes in comput .
_ , * 5494 * : 128142 .
springer , berlin ( 2009 ) . | we introduce a framework suitable for describing pattern recognition task using the mathematical language of density matrices .
in particular , we provide a one - to - one correspondence between patterns and pure density operators .
this correspondence enables us to : @xmath0 represent the nearest mean classifier ( nmc ) in terms of quantum objects , @xmath1 ) introduce a quantum classifier ( qc ) . by comparing the qc with the nmc on different 2d datasets ,
we show the first classifier can provide additional information that are particularly beneficial on a classical computer with respect to the second classifier . |
in astrophysics it is of high importance to construct a realistic physical model of stellar compact objects such as neutron stars and white dwarfs ( wds ) which fits with observations .
consequently , all physical phenomena and quantities must be duly taken into account in the equation of state ( eos ) .
up to now there exist three eos to describe the degenerate matter of wds : the chandrasekhar eos,@xcite the salpeter eos@xcite and the relativistic feynman - metropolis - teller ( rfmt ) eos.@xcite the main differences , advantages and drawbacks among theses eos are discussed by rotondo et al@xcite . in detail .
moreover , according to shapiro et al.@xcite it is necessary to investigate wds in general relativity ( gr ) in order to analyze their stability though the corrections of gr can be neglected for low mass wds . in this work
, we investigate hot white dwarfs by employing the chandrasekhar eos@xcite including the finite - temperature effects .
similar approach to include the effects of finite temperatures in case of the rfmt eos has been used by s. m. de carvalho et al.@xcite here we perform similar analyses following s. m. de carvalho et al.@xcite in order to construct the mass - radius relations of white dwarfs at finite temperatures in gr for the sake of completeness .
we compare our results with the estimated masses and radii from the sloan digital sky survey data release 4 ( sdss - e06 catalog).@xcite
for the sake of clarity we make use of the chandrasekhar eos since it is the simplest eos for wd matter , is well known and widely used in the description of physical properties of wds.@xcite in general , the eos is determined through the total pressure and the total energy density of matter . within the chandrasekhar s approximation ,
the total pressure is due to the pressure of electrons @xmath1 since the pressure of positive ions @xmath2 ( naked nuclei ) is insignificant whereas the energy density is due to the energy density of nuclei @xmath3 as the energy density of the degenerate electrons @xmath4 is negligibly small .
thus , the chandrasekhar eos is given by @xmath5 @xmath6 where @xmath7 is the average atomic weight , @xmath8 is the number of protons , @xmath9 g is the unified atomic mass , @xmath10 is the speed of light and @xmath11 is the electron number density . in general
, the electron number density follows from the fermi - dirac statistics and is determined by@xcite @xmath12 + 1 } , \ ] ] where @xmath13 is the boltzmann constant , @xmath14 is the electron chemical potential without the rest - mass , and @xmath15 , with @xmath16 and @xmath17 the electron momentum and rest - mass , respectively .
it is possible to show that ( [ eq : ne ] ) can be written in an alternative form as @xmath18 , \label{eq : ne2}\ ] ] where @xmath19 is the relativistic fermi - dirac integral , @xmath20 , @xmath21 and @xmath22 are degeneracy parameters.@xcite consequently , the total electron pressure for @xmath23 k is given by @xmath24.\end{aligned}\ ] ] white dwarf for selected temperatures in the range @xmath25 k. ] in fig .
[ fig : f1 ] we plot the total pressure ( [ eq : pe2 ] ) as a function of the total density ( [ eq : ech0 ] ) of the system for selected temperatures @xmath26 $ ] k. as one can see the effects of temperature is essential in the range of small densities .
when @xmath27 one can write for the number density of the degenerate electron gas the following expression from the eq .
( [ eq : ne ] ) @xmath28 and the total electron pressure @xmath29 \label{eq : eos2}\ , , \end{aligned}\ ] ] where @xmath30 is the dimensionless fermi momentum .
from spherically symmetric metric @xmath31 the equations of equilibrium can be written in the tolman - oppenheimer - volkoff ( tov ) form , @xmath32}\,\label{eq : nu},\\ \frac{d m(r)}{dr } & = & 4 \pi r^2 \frac{{\cal e}(r)}{c^2}\,\label{eq : mr},\\ \frac{d p(r)}{dr } & = & - \frac{1}{2 } \frac{d \nu(r)}{dr } [ { \cal e}(r)+p(r)]\,\label{eq : pr},\end{aligned}\ ] ] where the mass enclosed at the distance @xmath33 is determined through @xmath34 , @xmath35 is the total energy density , and @xmath36 is the total pressure , given by eqs .
( [ eq : ech0 ] ) and ( [ eq : pch0 ] ) . from the eqs .
( [ eq : nu ] ) and ( [ eq : pr ] ) the total pressure can be rewritten in the following form @xmath37\left[1+\frac{4 \pi r^2 p(r)}{m(r ) c^2}\right]\left[1-\frac{2
g m(r)}{r c^2}\right]^{-1}.\ ] ] the tov equation completely determines the structure of a spherically symmetric body of isotropic material in equilibrium .
the first two factors in square brackets represent special relativistic corrections of order @xmath38 that arise from the mass - energy relation so that the denominators , @xmath39 and @xmath40 , vary relativistically in connection with einstein s famous equation @xmath41 .
the last term in brackets is a general relativistic correction due to non - negligible strength of the gravitational potential ( in units of @xmath42 ) and the meaning of @xmath43 as the total integrated mass out to a radial distance @xmath33 .
these corrections each act to strengthen the gravitational interaction .
if terms of order @xmath38 are neglected , the tov equation becomes the newtonian hydrostatic equation , @xmath44 @xmath45 and @xmath46 used to find the equilibrium structure of a spherically symmetric body of isotropic material when general - relativistic corrections are not important . here
@xmath47 is the newtonian potential of the gravitational field inside the star .
by solving the tov equations eqs .
( [ eq : nu])-([eq : pr ] ) numerically , we obtained mass - central density @xmath48 , and mass - radius @xmath49 relations for static hot wds . in fig .
[ fig : f4 ] we plot the mass - central density for @xmath50 white dwarfs using the chandrasekhar eos at different temperatures in the range of densities where the finite - temperature effects are more important .
white dwarfs for selected temperatures from @xmath51 k to @xmath52 k. ] white dwarfs at temperatures @xmath53\ $ ] k in the range @xmath54 km . ] in addition , in fig .
[ fig : f3 ] we show the mass - radius relations .
one can see that the finite - temperature effects are essential for small mass wds .
k to @xmath52 k and their comparison with the masses and radii of white dwarfs taken from the sloan digital sky survey data release 4 ( black dots ) . ] in fig .
[ fig : f2 ] we plot @xmath49 relation for hot static white dwarfs with @xmath50 ( see fig .
[ fig : f4 ] ) and compare them with the estimated masses and radii of white dwarfs from the sloan digital sky survey data release 4 ( see ref . ) .
note , the temperature we use here is the temperature of the wd isothermal core @xmath55 . in order to find the surface effective temperature @xmath56 , one needs to know the temperature gradient ( so the heat flux ) between the isothermal core and the surface of the star from which photons finally escape , i.e. a @xmath55-@xmath56 relation .
we adopt here the koester formula @xmath57 , where @xmath58 is the surface gravity.@xcite
in this work we studied the properties of static wds by using the chandrasekhar eos at finite - temperatures .
to investigate wds in gr we solved the tov equation numerically in order to construct @xmath48 and @xmath49 relations at finite temperatures .
furthermore we superposed our results with the estimated values of masses and radii obtained by tremblay et al.@xcite as a result we showed that all observational data , at least in the range of low masses , can be described by the chandrasekhar eos at finite - temperatures . in our computations
we used the values of temperature of the core of the wd . to compare with the real surface temperature of wds we exploited the koester formula which establishes the connection between the effective surface temperature and the temperature of the core of wds .
we found that most of the observed wds have core temperatures lower than @xmath59 k ( see fig .
[ fig : f2 ] ) . a precise analysis using empirical mass - radius relations obtained from the spectroscopic or photometric measurements of masses and radii
is still needed to confirm and extend our results .
this work was supported by the ministry of education and science of the republic of kazakhstan , grant no .
3101/gf4 ipc-11/2015 . | [ sec:1]recently , it has been shown by s. m. de carvalho et al .
( 2014 ) that the deviations between the degenerate case and observations were already evident for 0.7 - 0.8 m@xmath0 white dwarfs . such deviations were related to the neglected effects of finite temperatures on the structure of a white dwarf .
therefore , in this work by employing the chandrasekhar equation of state taking into account the effects of temperature we show how the total pressure of the white dwarf matter depends on the mass density at different temperatures .
afterwards we construct equilibrium configurations of white dwarfs at finite temperatures .
we obtain the mass - radius relations of white dwarfs for different temperatures by solving the tolman - oppenheimer - volkoff equation , and compare them with the estimated masses and radii inferred from the sloan digital sky survey data release 4 . |
denote by @xmath3 $ ] the model @xmath4-space with curvature @xmath0 ; i.e. , @xmath3 $ ] is the simply connected @xmath4-dimensional riemannian manifold with constant curvature @xmath0 .
we will also use shortcuts @xmath5 $ ] for the unit @xmath4-sphere , and @xmath6 $ ] for the euclidean @xmath4-space .
in this paper we play with applications of the following lemma . the proof is given in section [ sec : key - lenmma ] .
this lemma was first discovered by buyalo in the case of nonpositive curvature ; see ( * ? ? ?
* lemma 5.8 ) .
buyalo s lemma[lem : key ]
let @xmath7 be a complete riemannian manifold with sectional curvature either @xmath8 or @xmath2 .
let @xmath9 be a tetrahedron in @xmath10 $ ] and @xmath11 be a union of three out of four faces of @xmath9 .
then any immersion @xmath12 which is isometric and geodesic on each face can be extended to an isometric geodesic immersion @xmath13 .
moreover , @xmath14 is uniquely determined by @xmath15 .
here is an immediate corollary : corollary let @xmath16 be a complete riemannian metric on @xmath17 with curvature @xmath18 ( or @xmath19 ) such that all three coordinate planes of @xmath17 are flat geodesic hypersurfaces in @xmath20
. then @xmath20 is isometric to euclidean space .
we would suggest that reader checks that the last statement does not follow from the standard theorems ; in particular the splitting theorems can not help here directly .
let us now introduce some terminology to state further applications . * a riemannian manifold ( possibly not complete ) of constant curvature @xmath0 will be called _ @xmath0-flat_. * a @xmath0-flat riemannian manifold ( possibly not complete ) which satisfies the conclusion of buyalo s lemma will be called _ two - convex_. this definition is discussed in more details in section [ sec : k - convex ] . *
given a riemannian manifold @xmath7 , its maximal open subset of constant curvature @xmath0 will be called _
@xmath0-flat domain of @xmath7 _ and it will be denoted as @xmath21 . from buyalo s lemma one easily gets the following ; a formal proof is given in section [ sec : key - lenmma ] .
observation[obs : flat - domain ] let @xmath7 be a complete riemannian manifold either with curvature @xmath8 or @xmath2 . then @xmath21 is two - convex . here is an application .
theorem[thm : rigid - sphere ] let @xmath22 and @xmath7 be a complete connected @xmath4-dimensional manifold with curvature @xmath23 or @xmath24 which admits a totally geodesic immersion of the closed unit hemisphere @xmath25 and an open neighborhood of @xmath26 in @xmath7 has constant curvature @xmath27 .
then @xmath7 has constant curvature @xmath27 . *
note that diameter - sphere rigidity does not help here directly ; in principle , the diameter of @xmath7 might be @xmath28 . * note that @xmath29 equipped with the canonical
metric is an example of a space with curvature @xmath23 and @xmath30 , which admits totally geodesic immersions of 2-spheres of constant curvature @xmath27 and @xmath31 .
i.e. , the condition in theorem [ thm : rigid - sphere ] that the curvature is constant in a neighborhood of @xmath26 is necessary . * in the case of curvature @xmath23 , theorem [ thm : rigid - sphere ] also holds in dimension @xmath32 ; this is proved by zalgaller in @xcite ; see theorem [ thm : rigid - sphere-2d ] and the discussion around . to prove the theorem
, one needs to show that if a neighborhood @xmath33 of @xmath34 in @xmath35 admits an immersion in a two - convex manifold @xmath36 then @xmath36 has to be complete
. then observation [ obs : flat - domain ] implies that @xmath37 ; i.e. , @xmath7 is a spherical space form . in other words ,
any neighborhood @xmath33 of @xmath26 in @xmath35 is _ exhaustive _ in the sense of the following definition .
definition[def : exhaustive ] let @xmath33 be a @xmath0-flat manifold .
assume that any connected two - convex manifold @xmath36 that appears as the target of an open isometric immersion @xmath38 is complete , and at least one such @xmath36 exists .
then we say that @xmath33 is _
exhaustive_. using this definition , we can formulate the following generalization of theorem [ thm : rigid - sphere ] : theorem[thm : exhaustive ] let @xmath7 be a complete connected riemannian manifold with curvature @xmath8 or @xmath2 .
assume there is an open isometric immersion @xmath39 from an exhaustive @xmath0-flat manifold @xmath33 .
then @xmath7 has constant curvature @xmath0 . in order to apply
this theorem one only has to find a source of exhaustive manifolds .
in section [ sec : k - convex ] , we introduce the notion of the _ two - hull _ of a @xmath0-flat simply connected manifold @xmath33 ; in some sense this is the minimal simply connected two - convex manifold which contains an _ immersed copy _ of @xmath33 .
it is easy to see that if the two - hull of a manifold @xmath33 is isometric to @xmath3 $ ] then @xmath33 is exhaustive .
this permits one to present a number of examples of exhaustive manifolds .
this is done in section [ sec : exhaustive ] , here is a list of examples : ( proposition [ prop : comp - conv ] . )
: : for @xmath22 , any non - empty open subset of @xmath3 $ ] with convex complement .
( proposition [ prop : comp - conv+ ] . ) : : more generally : any open simply connected subset @xmath40 $ ] which satisfies the following property . for any @xmath41 $ ]
there is a 3-dimensional subspace @xmath42 of @xmath3 $ ] containing @xmath43 ( @xmath42 is an isometric copy of @xmath10 $ ] ) such that @xmath44 and each connected component of @xmath45 is a convex set .
+ in particular , @xmath46 is exhaustive .
( proposition [ prop : s - m ] . ) : : any open subset of @xmath35 which contains the standard 2-dimensional hemisphere .
this type of manifolds is used in theorem [ thm : rigid - sphere ] .
( this list can be continued . )
one outcome of theorem [ thm : exhaustive ] is a sufficient condition on the piece ] of the model space @xmath3 $ ] , which _ can not _ be exchanged to another piece that has sectional curvature not smaller or not bigger .
this condition is nontrivial only for @xmath22 .
the similar conditions for scalar and ricci curvature were studied .
the case of deformation with nondecreasing curvature turned out to be very different from the one with nonincreasing curvature .
after rescaling one can only consider three cases @xmath47 or @xmath27 .
if @xmath48 , the case of nondecreasing scalar curvature leads to so called _ positive mass conjecture _ which is proved by schoen yau and witten in @xcite and @xcite .
this implies in particular that _ the metric of euclidean space can not be perturbed in a bounded region so that the scalar curvature does not decrease_. an analogous statement holds for @xmath49 ; i.e. , _ the metric of lobachevsky space can not be perturbed in a bounded region so that the scalar curvature does not decrease_. the later was proved by min - oo in @xcite .
the case @xmath50 was considered in @xcite , where min - oo makes an attempt to show that the standard metric on the @xmath4-sphere can not be perturbed inside of hemisphere so that the scalar curvature does not decrease .
but in @xcite , brendle , marques and neves find a counterexample .
one can not perturb the metric in a sufficiently small domain of sphere , but optimal bounds on such domain seem to be not known .
on the other hand as it was shown by hang and wang in @xcite , one can not perturb the metric of the standard sphere inside its hemisphere with nondecreasing ricci curvature .
the two - dimensional case of the above statements for @xmath48 and @xmath51 follows from gauss bonnet formula and the case @xmath50 was done by zalgaller ( see the appendix ) . in @xcite ,
lohkamp proves that for all @xmath52 , one can perturb the metric of @xmath3 $ ] in any open region in such a way that its ricci curvature does not increase . moreover , this can be done without changing the topology and with arbitrary small change of the geometry of the space . in two - dimensional case ,
attaching a handle can be done in arbitrary small region with decreasing its curvature . on the other hand ,
if we fix the topology , for @xmath48 and @xmath51 , gauss bonnet formula prevents any change of metric in bounded regions with nonincreasing curvature . for @xmath50 ,
even if topology is fixed , the metric can be changed ( by inserting a bubble ) in arbitrary small open subset so that the curvature in the region decreases .
however , it seems that for proper subsets of hemisphere , there is no continuous deformation of this type .
definition[def:2-conv-2 ] let @xmath33 be a @xmath4-dimensional @xmath0-flat manifold .
we say that @xmath33 is _ two - convex _ if the following condition holds : given a tetrahedron @xmath9 in @xmath10 $ ] with a choice of a subset @xmath53 formed by 3 out of 4 faces , any immersion @xmath12 which is isometric and geodesic on each face of @xmath11 can be extended to an isometric geodesic immersion @xmath13 .
definition[def:2-hull ] let @xmath33 be a simply connected @xmath4-dimensional @xmath0-flat manifold .
a two - convex manifold @xmath36 is called the _ two - hull _ of @xmath33 ( briefly @xmath54 ) if there is an open immersion @xmath55 such that for any open isometric immersion @xmath56 into a two - convex manifold @xmath57 there is a isometric immersion @xmath58 which makes the following diagram commutative : @xmath59 \obj(3,5)[aa]{$\omega$ } \obj(0,0)[bb]{$\phi$ } \obj(6,0)[cc]{$\psi$ } \mor{aa}{bb}{$\phi$}[-1,0 ] \mor{aa}{cc}{$\psi$ } \mor{bb}{cc}{$\theta$ } \enddc$}\ ] ] further the immersion @xmath55 will be called _ two - hull immersion_. let us notice that even though for some manifolds @xmath33 the two - hull immersion @xmath55 is in fact an embedding , in general one should not expect this .
our next goal is to prove existence of the two - hull .
proposition[prop:2-hull ] for any simply connected @xmath0-flat manifold @xmath33 , its two - hull @xmath36 is uniquely defined up to isometry .
moreover , a. if @xmath55 and @xmath60 are two - hull immersions then there is an isometry @xmath61 such that @xmath62 . b. @xmath36 is simply connected . to prove the above proposition , we mimic the proof of existence of ordinary convex hull as the intersection of all convex sets containing the given set .
fix a simply connected @xmath4-dimensional @xmath0-flat manifold @xmath33 . note that @xmath33 admits an open isometric immersion @xmath63 $ ] .
let us construct a category @xmath64 .
the class of objects in @xmath64 is formed by all open isometric immersion @xmath65 where target @xmath57 is a two - convex @xmath0-flat manifold , and the morphisms are commutative triangles of isometric immersions @xmath59 \obj(3,5)[aa]{$\omega$ } \obj(0,0)[bb]{$\psi_1 $ } \obj(6,0)[cc]{$\psi_2 $ } \mor{aa}{bb}{$\psi_1$}[-1,0 ] \mor{aa}{cc}{$\psi_2 $ } \mor{bb}{cc}{$\theta$ } \enddc$ } \eqno{({*})}\ ] ] the category @xmath64 contains at least one object , the immersion @xmath66 $ ] mentioned above ( this is the terminal object of @xmath64 ) .
the existence of the two - hull of @xmath33 is equivalent to the existence of an initial object in @xmath64 .
a choice of point @xmath67 for each object @xmath65 in @xmath64 is called the _ inverse point system _ , if for any morphism as in @xmath68 we have @xmath69 . note that for any point @xmath70 , the choice of points @xmath71 forms an inverse point system .
set @xmath36 to be the set of all inverse point systems .
note that @xmath36 comes with natural maps @xmath72 and @xmath73 for any object @xmath65 in @xmath64 such that the following diagram commutes .
\obj(3,5)[aa]{$\omega$ } \obj(0,0)[bb]{$\phi$ } \obj(6,0)[cc]{$\psi$ } \mor{aa}{bb}{$\phi$}[-1,0 ] \mor{aa}{cc}{$\psi$ } \mor{bb}{cc}{$\theta_\psi$ } \enddc\ ] ] let us equip @xmath36 with the weakest topology which makes all maps @xmath75 continuous .
clearly , with this topology all @xmath75 and @xmath76 become immersions .
let @xmath77 be the maximal open set in @xmath36 which is homeomorphic to @xmath4-manifold .
note that @xmath77 comes with a natural @xmath0-flat metric so that each @xmath75 is an open isometric immersion of @xmath77 .
it is easy to see that @xmath77 is two - convex and it contains @xmath78 .
therefore , the isometric immersion @xmath79 is an object of @xmath64 .
hence there is an immersion @xmath80 which commutes with the natural embedding @xmath81 ; i.e. , @xmath82 . in other words
@xmath36 is isometric to the two - hull of @xmath33 .
the last two statements of the proposition follow easily from above .
in particular , @xmath36 coincides with its universal cover @xmath83 because @xmath84 lifts to @xmath85 ( @xmath86 ) , and there is a morphism from @xmath87 to @xmath76 in @xmath64 , proving that @xmath87 and @xmath76 are isomorphic because @xmath76 is initial in @xmath64 by construction .
in this section we prove buyalo s lemma [ lem : key ] and observation [ obs : flat - domain ] .
the proof of the following proposition is left to the reader .
proposition[lem : triv ] let @xmath88 and @xmath89 be ( possibly noncomplete ) riemannian manifolds and @xmath90 be an open set of unit speed geodesics in @xmath88 , covering all points of @xmath88 . then @xmath91 is an isometric geodesic immersion if and only if for any @xmath92 , the curve @xmath93 is a unit speed geodesic in @xmath89 . set @xmath94 .
note that the statement of buyalo s lemma trivially holds if @xmath95 .
further we assume @xmath22 . by choosing an isometric geodesic embedding @xmath96 $ ]
, we can consider @xmath9 as a subset of @xmath3 $ ] .
let us denote by @xmath97 the common vertex of the faces in @xmath11 and let @xmath98 be the remaining vertexes of @xmath9 . denote by @xmath99 the corresponding points in @xmath7 ; i.e. @xmath100 set @xmath101
. assume first that the injectivity radius at any point in @xmath102 is at least @xmath103 . in this case
@xmath15 is distance preserving on each face .
claim[clm : dist - preserv ] @xmath104 is a distance preserving map . on the geodesic
@xmath105 $ ] consider two unit normal fields that go in the directions of the images of the faces adjacent to @xmath105 $ ] .
note that both fields are parallel .
thus the angle between the images of the faces in @xmath11 is constant along the common side .
taking the point on geodesic @xmath105 $ ] close to @xmath43 , one can see that angles between faces of @xmath106 in @xmath7 coincide with the corresponding angles in @xmath107 $ ] .
consider points @xmath108 , & \tilde y'&\in [ \tilde p\tilde y ] , & \tilde z'&\in [ \tilde p\tilde z ] , \\ x'&=f(\tilde x ' ) , & y'&=f(\tilde y ' ) , & z'&=f(\tilde z').\end{aligned}\ ] ] from above , we have that corresponding angles in the triangles @xmath109 $ ] and @xmath110 $ ] are equal ; i.e. , the angles in triangle @xmath109 $ ] coincide with its comparison angles .
let @xmath111 and @xmath112 be arbitrary points on the sides of triangle @xmath110 $ ] and @xmath113 and @xmath114 . in both cases ( curvature @xmath8 or @xmath2 ) the above angle equality implies that @xmath115}.\ ] ] ( here @xmath116 denotes distance function in a metric space @xmath88 . ) note that for any @xmath117 there is a triangle @xmath110 $ ] as above which contains @xmath111 and @xmath112 on its sides .
hence the claim follows .
note that there is a map @xmath118 satisfying the following properties : 1 .
@xmath119 ; 2 .
@xmath120 , and the differential of @xmath14 at @xmath43 is an isometry @xmath121 ; 3 . @xmath14 sends all unit speed geodesics through @xmath97 to unit speed geodesics through @xmath43 .
claim the restriction of any such @xmath14 to @xmath9 satisfies buyalo s lemma ; this claim is proved separately in the two cases : by toponogov comparison theorem , the diffeomorphism @xmath118 is non - expanding .
this fact together with claim [ clm : dist - preserv ] imply that the restriction of @xmath14 to @xmath9 is distance preserving on any geodesic in @xmath9 with ends in @xmath11 .
applying proposition [ lem : triv ] we get that the restriction of @xmath14 to @xmath9 is isometric and geodesic in the interior of @xmath9 and hence the same holds on whole @xmath9 .
set @xmath122 to be the set of all minimizing geodesics with ends in @xmath106 and let @xmath123 be the subset of @xmath7 covered by geodesics in @xmath122 . by toponogov comparison ,
the diffeomorphism @xmath118 is distance nondecreasing , while its inverse @xmath124 is a distance non - increasing diffeomorphism . since @xmath15 is distance preserving
, it follows that @xmath124 is isometric on each of the geodesic in @xmath122 ; moreover , any minimizing geodesic between points in @xmath11 can be presented as @xmath125 for some @xmath126 .
it follows that @xmath127 , or equivalently @xmath128 . in particular , @xmath14 is distance preserving on each minimizing geodesic with ends in @xmath11 .
applying proposition [ lem : triv ] the same way as above , we conclude that the restriction of @xmath14 to @xmath9 is distance preserving and geodesic . to treat the general case , choose @xmath129 so that the injectivity radius at any point in @xmath130 is at least @xmath131 .
note that one can cover the interior of @xmath9 by an infinite sequence of tetrahedra @xmath132 with a choice of three faces @xmath133 in each @xmath134 such that @xmath135 and @xmath136 for each @xmath137 .
then it remains to apply the above argument sequentially to @xmath132 and pass to the closure .
set @xmath94 .
choose any point @xmath138 and @xmath139 $ ] .
choose a map @xmath140\to \flat^\kappa m$ ] such that 1 .
@xmath120 , and the differential of @xmath14 at @xmath43 is an isometry @xmath121 ; 2 .
@xmath14 sends all unit speed geodesics through @xmath97 to unit speed geodesics through @xmath43 .
let @xmath141 $ ] be the maximal open star - shaped w.r.t .
@xmath97 set such that the map @xmath14 induces an open isometric immersion of @xmath142 .
let @xmath143 be the set of all tetrahedra with one vertex at @xmath97 and three adjacent faces in @xmath144 and let @xmath145 be the union of all tetrahedra in @xmath146 .
clearly @xmath145 is open and @xmath147 .
according to buyalo s lemma , the map @xmath14 is isometric on each geodesic lying in a tetrahedron from @xmath143 . applying proposition [ lem : triv ]
, we get that @xmath14 is an open isometric immersion @xmath148 .
thus , @xmath149 for any @xmath150 , hence the result .
let @xmath33 be a simply connected @xmath0-flat manifold . denote by @xmath151 the two - hull of @xmath33 ( see definition [ def:2-hull ] ) . from the definition of the two - hull , we have that if @xmath151 is isometric to the model space @xmath3 $ ] then @xmath33 is exhaustive ( see definition [ def : exhaustive ] ) . in this section
we use the above observation to construct examples of exhaustive manifolds .
the following two propositions follow directly from the discussion above .
( in other words , the proof is left to the reader . )
proposition[prop : comp - conv ] assume @xmath22 and suppose @xmath152 $ ] is an nonempty open set with convex complement . then @xmath151 is isometric to @xmath3 $ ] . in particular , @xmath33 is exhaustive . here is a generalization of the above proposition : proposition[prop : comp - conv+ ] suppose @xmath22 and suppose @xmath152 $ ] is an nonempty open set such that through any point @xmath41 $ ] passes a @xmath153-dimensional subspace @xmath42 ( i.e. , an isometric copy of @xmath10 $ ] ) such that each connected component of @xmath45 is a convex set
. then @xmath151 is isometric to @xmath3 $ ] .
in particular , @xmath33 is exhaustive .
the proof of the following proposition requires some work .
proposition[prop : s - m ] assume @xmath22 and suppose @xmath154 $ ] admits a geodesic isometric immersion @xmath155 .
then @xmath151 is isometric to @xmath35 .
fix two embeddings @xmath156 , denote their composition by @xmath157 .
note that for any point @xmath158 there is unique embedding @xmath159 such that @xmath160 and @xmath161 for any @xmath162 .
it is easy to see that one can choose a tetrahedron @xmath9 in @xmath35 , such that one face of @xmath9 belongs to @xmath163 and contains all points in the set @xmath164 , while the rest of the faces is arbitrary close to @xmath26 , in particular these faces belong to @xmath33 .
applying to @xmath9 the definition of two - convexity , we get an isometric geodesic immersion @xmath165 .
it is easy to see that the map @xmath166 is independent on the choice of @xmath9 ; moreover , the obtained map @xmath167 is an open isometric immersion .
since @xmath151 is simply connected ( see proposition [ prop:2-hull ] ) we have that @xmath151 is isometric to @xmath35 .
the definition of two - convexity ( [ def:2-conv-2 ] ) can be generalized to `` @xmath168-convexity '' ; one has to change tetrahedron @xmath9 to a @xmath169-dimensional simplex and @xmath11 to the set formed by @xmath168 faces out of @xmath169 in @xmath9 . in this case , @xmath27-convexity is equivalent to the usual convexity of each connected component of @xmath33 . in ( * ? ? ?
* section @xmath170 ) , gromov introduced the following closely related notion which we will call further as _ _
lefschetz-@xmath168-convexity _ _ ; in gromov s notations lefschetz-@xmath168-convexity in @xmath171 is called @xmath172-convexity . ] .
definition[def : lefschetz ] an open set @xmath33 in @xmath171 is _ lefschetz-@xmath168-convex _ if for any @xmath168-dimensional affine subspace @xmath173 the natural homology homomorphism @xmath174 is injective .
this definition can be generalized to @xmath0-flat manifolds , one only has to replace @xmath175 by @xmath168-dimensional manifolds @xmath176 that admit proper of riemannian manifolds @xmath176 and @xmath33 is called _ proper _ if for any point @xmath70 there is @xmath129 such that each connected component of @xmath177 is compact . ]
isometric geodesic immersion @xmath178 .
it is easy to show that lefschetz-@xmath168-convexity in @xmath171 implies our @xmath168-convexity .
we know that the converse holds in two trivial cases : @xmath179 and @xmath180 , but in all other cases we do not know the answer to the following question .
open problem is it true that any @xmath168-convex open subset of @xmath171 is lefschetz-@xmath168-convex ? to get a feeling of definition of @xmath168-convexity , it is useful to observe the following .
proposition[prop : smooth ] if @xmath33 is an open subset of @xmath171 with smooth boundary @xmath181 , then it is @xmath168-convex if and only if the hypersurface @xmath181 has at most @xmath182 negative principle curvatures at any point .
it is well known that any convex set in @xmath171 can be approximated by a convex set with smooth boundary .
it turns out that for @xmath168-convex sets ( as well as for lefschetz-@xmath168-convex sets ) this is no longer true .
one of the reasons comes from the fact that for @xmath168-convex sets with smooth boundary the homeomorphism in @xmath183 is injective for subspaces @xmath173 of arbitrary dimension ; the proof is an exercise in morse theory , see ( * ? ? ?
* section @xmath170 ) .
thus , any @xmath168-convex set which does not satisfy this condition can not be approximated . to give an explicit example ,
let @xmath184 be the complement to the union of the following two 3-dimensional halfspaces : @xmath185 clearly @xmath33 is 2-convex and simply connected , but @xmath186 for @xmath187 the following example shows a problem with extension of the two - hull construction to non simply connected case .
consider an isometric action @xmath188 with two fixed points ; then take @xmath33 to be the orbit space @xmath189 with singular orbits removed .
note that @xmath33 admits no open isometric immersion into a two - convex @xmath27-flat manifold .
hence the two - hull of @xmath33 can not be defined in the class of manifolds . on the other hand
it can be defined in the class of `` riemannian megafolds '' ; these creatures were introduced in @xcite and under a different name in @xcite ; they look a lot like riemannian manifolds , but fail to be topological spaces .
( in the above example , the two - hull of @xmath33 is the riemannian orbifold @xmath190 . ) here is a possible generalization of proposition [ prop : s - m ] : question is it true that the two - hull of any open simply connected set @xmath191 which contains a closed geodesic is isometric to @xmath35 ? the following question of d. burago and b. kleiner is open for long time .
it is not directly relevant to all above , but it was one of the initial motivations for our work .
question is it possible to construct a riemannian metric @xmath16 on the product of a torus and an open disc @xmath192 such that the torus @xmath193 is flat and the curvature is strictly positive outside of @xmath194 ?
an answer to this question might lead to a better understanding of manifolds with _ almost positive curvature _
( see @xcite ) .
let us yet mention two related questions from http://mathoverflow.net/[mathoverflow ] : * http://mathoverflow.net/questions/55788[question 55788 ] about two - convexity and lefschetz property .
* http://mathoverflow.net/questions/50889[question 50889 ] about possible generalization of buyalo s lemma .
# 1 _ _
here we briefly repeat the proof of a theorem from @xcite .
we do this since the result which interests us ( theorem [ thm : rigid - sphere-2d ] ) was not formulated as a separate statement ; it appeared as an intermediate statement in the proof .
theorem[thm : zalgaller ] let @xmath195 and @xmath196 be two simple spherical polygons ( not necessary convex ) with equal corresponding sides .
assume @xmath173 lies in an open hemisphere and @xmath197 for each @xmath198 .
then @xmath173 is congruent to @xmath199 . at first
this result might look unrelated to the content of this article .
but the proof relies on the following 2-dimensional analog of theorem [ thm : rigid - sphere ] . recall that _ spherical polyhedron _ is a simplicial complex equipped with a metric such that each simplex is isometric to a simplex in a standard sphere .
theorem[thm : rigid - sphere-2d ] let @xmath200 be a spherical polyhedron which is homeomorphic to @xmath201 and has curvature @xmath23 in the sense of alexandrov .
assume that an open neighborhood of @xmath34 in @xmath201 admits a locally isometric immersion in @xmath200 .
then @xmath200 is isometric to the standard sphere . to deduce theorem [ thm : zalgaller ] from theorem [ thm : rigid - sphere-2d ]
, zalgaller cuts the polygon @xmath173 from the sphere and glues instead polygon @xmath199 . as a result he gets the spherical polyhedron @xmath200 as in theorem [ thm : rigid - sphere-2d ] .
( in fact , if we drop the condition that @xmath173 lies in a hemisphere , we can obtain this way any spherical polyhedral metric on @xmath201 with curvature @xmath23 . )
theorem [ thm : rigid - sphere-2d ] is proved by induction on the number @xmath137 of singular points in @xmath200 .
the base case @xmath202 is trivial . to do the induction step ,
choose two singular points @xmath203 , cut @xmath200 along a geodesic @xmath204 $ ] and patch the hole so that the obtained new polyhedron @xmath205 has curvature @xmath23 .
the patch is obtained by gluing two copies of a spherical triangle along two sides . for the right choice of the triangle ,
the points @xmath43 and @xmath206 become regular in @xmath205 and exactly one new singular point appears in the patch .
this way , the case with @xmath137 singular points is reduced to the case with @xmath207 singular points ( if @xmath208 ) . applying polyhedral approximation
, one can extend theorem [ thm : rigid - sphere-2d ] to any surface with curvature @xmath23 in the sense of alexandrov ; in particular , this shows that theorem [ thm : rigid - sphere ] holds in addition for @xmath209 and curvature @xmath23 . | we observe that the maximal open set of constant curvature @xmath0 in a riemannian manifold of curvature @xmath1 or @xmath2 has a convexity type property , which we call _ two - convexity_. this statement is used to prove a number of rigidity statements in comparison geometry . |
there are several advantages in studying a heavy - light system on a lattice .
our meson is much more simple than in true qcd : one of the quarks is static with the light quark `` orbiting '' it .
this makes it very beneficial for modelling . on the lattice
an abundance of data can be produced , and we know which state we are measuring the physical states can be a mixture of two or more configurations , but on the lattice this complication is avoided . however , our results on the heavy - light system can still be compared to the @xmath2 meson experimental results .
we have measured both angular and radial excitations of heavy - light mesons , and not just their energies but also some radial distributions . since the heavy quark spin decouples from the game we may label the states as @xmath3 , where l is the angular momentum and @xmath4 is the spin of the light quark .
the measurements were done on a @xmath5 lattice .
we have two degenerate quark flavours with a mass that is close to the strange quark mass .
the lattice configurations were generated by the ukqcd collaboration . more details of the different lattices used in this study can be found in refs .
@xcite , and references therein .
two different levels of fuzzing ( 2 and 8 iterations of conventional fuzzing ) were used in the spatial directions to permit the extraction of the excited states .
the 2-point correlation function ( see figure [ fig : corr ] ) is defined as @xmath6 where @xmath7 is the heavy ( infinite mass)-quark propagator and @xmath8 the light anti - quark propagator .
@xmath9 is a linear combination of products of gauge links at time @xmath10 along paths @xmath11 and @xmath12 defines the spin structure of the operator .
the @xmath13 means the average over the whole lattice .
the energies ( @xmath14 ) and amplitudes ( @xmath15 ) are extracted by fitting the @xmath16 with a sum of exponentials , @xmath17 fuzzing indices have been omitted for clarity .
.heavy - light meson energy differences in gev . lattice 1 uses a static heavy quark , whereas lattices 2 and 3 are slightly smeared in the time direction .
lattice 2 uses ape type smearing and lattice 3 uses hypercubic blocking . see @xcite for details .
@xmath18 fm was used to convert the energies to physical units .
( note : in the first version of this article @xmath19 fm was used . ) [ cols="^,^,^,^,^,^,^,^",options="header " , ] the energy spectrum obtained is shown in fig .
[ fig : espectr ] and in table [ energytable ] .
there seems to be some spread in the extracted energy for the p@xmath20 state between the different lattices , but the reason for this is not understood yet .
the energy of the d@xmath21 state was expected to be near the spin average of the d@xmath22 and d@xmath20 energies , but it turned out to be a poor estimate of this average .
thus it is not clear whether or not the f@xmath21 energy is near the spin average of the two f - wave states , as was hoped .
our earlier results can be found in ref .
@xcite .
one interesting point to note here is that the spin - orbit splitting of the p - wave states is small , almost zero .
we extracted the energy difference of the p@xmath20 and p@xmath22 states in two different ways : 1 .
_ indirectly _ by simply calculating the difference using the energies given by the fits in eq
. [ c2fit ] , when the p@xmath20 and p@xmath22 data are fitted separately .
2 . combining the p@xmath20 and p@xmath22 data and fitting the ratio @xmath23 , which enables us to go _ directly _ for the spin - orbit splitting , @xmath24 .
the results of the fits are @xmath25 mev and @xmath26 mev , respectively .
d - wave spin - orbit splitting was also extracted in a similar manner , giving the results @xmath27 mev and @xmath28 mev , respectively . to obtain predictions of the @xmath2 meson excited state energies , we can now interpolate between the static heavy quark lattice calculations and @xmath29 meson experimental results , i.e. interpolate between the static quark and the charm quark .
linear extrapolation seems to work well , as can be seen in fig .
[ fig : bs_interpolation ] . the p - wave states seem to lie near the @xmath30 thresholds .
for evaluating the radial distributions of the light quark a 3-point correlation function shown in fig .
[ fig : corr ] is needed .
it is defined as @xmath31 this is rather similar to the 2-point correlation function in eq .
[ 2point ] .
we have now two light quark propagators , @xmath32 and @xmath33 , and a probe @xmath34 at distance @xmath35 from the static quark ( @xmath36 for the vector ( charge ) and @xmath37 for the scalar ( matter ) distribution ) . knowing the energies @xmath14 and the amplitudes @xmath15 from the earlier @xmath16 fit , the radial distributions , @xmath38 s , are then extracted by fitting the @xmath39 with @xmath40 the results are plotted in figs .
[ s11 ] and [ petd ] .
the error bars in these figures show statistical errors only .
see @xcite and @xcite for earlier calculations .
a simple model based on the dirac equation is used to try to describe the lattice data .
since the mass of the heavy quark is infinite we have essentially a one - body problem .
the potential in the dirac equation has a linearly rising scalar part , @xmath41 , as well as a vector part @xmath42 .
the one gluon exchange potential , @xmath43 , is modified to @xmath44 where @xmath45 mev and the dynamical gluon mass @xmath46 mev ( see @xcite for details ) .
the potential also has a scalar term @xmath47 , which is needed to increase the energy of higher angular momentum states .
however , this is only a small contribution ( about 30 mev for the f - wave ) .
the solid lines in the radial distribution plots are predictions from the dirac model fit with @xmath48 gev , @xmath49 , @xmath50 gev / fm , @xmath51 and @xmath52 .
these are treated as free parameters with the values obtained by fitting the ground state energies of p- , d- and f - wave states and the energy of the first radially excited s - wave state ( 2s ) .
note that the excited state energies in fig .
[ fig : espectr ] were not fitted .
* there is an abundance of lattice data , energies and radial distributions , available for this @xmath2-like system . *
the p - wave spin - orbit splitting is small ( essentially zero ) , which supports the symmetry @xmath53 as proposed in @xcite .
the d - wave spin - orbit splitting is clearly non - zero and positive , whereas another lattice group seems to find it to be slightly negative ( see @xcite ) . * the energies and radial distributions of s- ,
p- and d - wave states can be qualitatively understood by using a one - body dirac equation .
i am grateful to my supervisor a. m. green and to our collaborator , professor c. michael .
i wish to thank the ukqcd collaboration for providing the lattice configurations .
i also wish to thank the center for scientific computing in espoo , finland , for making available the computer resources .
the eu grant hprn - ct-2002 - 00311 euridice is also gratefully acknowledged .
ukqcd collaboration , a. m. green , j. koponen , c. michael , c. mcneile and g. thompson , phys .
d 69 , 094505 ( 2004 ) ; + ukqcd collaboration , j. koponen , ` hep - lat/0411015 ` ukqcd collaboration , j. koponen , pos lat2006 ( 2006 ) 112 ukqcd collaboration , a. m. green , j. koponen , c. michael and p. pennanen , phys .
d 65 , 014512 ( 2002 ) and eur .
j. c 28 , 79 ( 2003 ) ; ukqcd collaboration , a. m. green , j. ignatius , m. jahma , j. koponen , c. mcneile and c. michael , pos lat2005 ( 2005 ) 205 | this is a follow - up to our earlier work for the energies and the charge ( vector ) and matter ( scalar ) distributions for s - wave states in a heavy - light meson , where the heavy quark is static and the light quark has a mass about that of the strange quark .
we now study excited states of these mesons with higher angular momentum and with radial nodes .
the calculation is carried out with dynamical fermions on a @xmath0 lattice with a lattice spacing @xmath1 fm .
the lattice configurations were generated by the ukqcd collaboration .
attempts are now being made to understand these results in terms of the dirac equation . in nature
the closest equivalent of this heavy - light system is the @xmath2 meson , which allows us to compare our lattice calculations to experimental results ( where available ) or give a prediction where the p - wave states should lie .
we pay particular attention to the spin - orbit splitting , to see which one of the states ( for a given angular momentum l ) has the lower energy . |
the recent euro - zone crisis remind us that ` risk analysis ' is always an essential part of the theory of portfolio management .
markowitz(1952 ) @xcite , first proposed volatility ( or standard deviation ) as the risk measure .
as volatility provides an idea about average loss ( or gain ) of portfolio ; more new risk measures like ` _ _ value at risk _ _ ' ( var ) and ` _ _ expected short fall _ _ ' ( esf ) for extreme losses are also popular .
although basel ii regulatory framework requires inclusion of var and esf , till date volatility plays an important role in finance .
for example , volatility can be traded directly on the open market through _ exchange traded fund _ of vix index and indirectly through derivatives .
it is essential to measure the volatility and identify the main sources of volatility
. often the number of financial instruments of well diversified mutual funds or pension funds are more than thousands .
such funds invest in foreign countries on a regular basis ; sometimes in more than fifty to sixty different countries . however , portfolio managers are concerned about the stationarity in long time - series data and interested about only recent volatility which consider daily returns of a month and sometimes even less . as the number of sectors , or countries , or components , of a portfolio is greater than the number of days of return , the rank of the portfolio covariance matrix is less than full ; such cases yield non - unique solutions .
generally it is known as the ill - posed " problem . in this paper
, we mainly focus on estimating and analysing the sources of portfolio volatility under ill - posed " condition .
we address the mean - variance optimization problem which admits an analytical solution for the optimal weights of a given portfolio .
this optimization procedure requires an estimate of the portfolio covariance matrix .
but due to the ill - posed " structure of the problem , regular sample covariance would not work here .
therefore , we propose a regularized plug - in bayes estimator for the portfolio covariance matrix and use the optimized weights for the given portfolio . using this setup
we evaluate its out - sample performance using a monte carlo algorithm . in a 3-series paper , ledoit and wolf ( 2003 , 2004a
, 2004b)@xcite showed the use of shrinkage estimator on actual stock market data keeping all other steps of optimization process the same . by doing
so they reduced the tracking error relative to a ad - hoc index . as a result
it substantially increased the realized information ratio of active portfolio managers .
ledoit and wolf ( 2004a ) suggested a distribution - free approach to regularizing the covariance matrix , in this paper however we impose a probability structure on the covariance matrix for obvious reasons .
the proposed regularized plug - in bayes estimator in this paper bears a direct relationship with the empirical bayes estimator suggested by ledoit and wolf ( 2004a ) @xcite , for the covariance matrix . also , golsnoy and okhrin ( 2007)@xcite showed the improvement of portfolio selection by using the multivariate shrinkage estimator for the optimal portfolio weights .
recently , das and dey ( 2010 ) @xcite introduced some bayesian properties of multivariate gamma distribution for covariance matrix . in this paper
we use this bayesian approach to regularize the estimation problem . under certain conditions ,
the posterior distribution of the portfolio covariance matrix is proper and has a closed form inverted multivariate gamma distribution .
consequently , the solution of covariance matrix is unique .
the rest of the article is organized as follows . in section [ postcov ]
, we discuss the posterior distribution of portfolio covariance matrix and the condition under which it is proper . in section [ riskana ]
, we present a parallelizable monte carlo algorithm to obtain posterior inference on the risk . in section [ empstudy ] ,
we demonstrate the method with two empirical data sets .
the methodology of inference is applied initially to a small dataset , consisting of different asset classes .
next we consider a portfolio consisting of the stocks from national stock exchange of india " ( nsei ) .
section [ concl ] concludes the paper .
suppose , @xmath0 is a real symmetric sample portfolio covariance matrix of order @xmath1 with @xmath2 variables @xmath3 . @xmath4
is the corresponding population portfolio covariance matrix , such that for a diagonal matrix @xmath5 with diagonal elements @xmath6 and @xmath7 , @xmath8 has non - positive off - diagonal elements . hence due to bapat s condition , bapat ( 1989 ) @xcite , @xmath0 with characteristic function as @xmath9=|i_p - i\beta \sigma t|^{-\alpha},\ ] ] has the density function @xmath10 has an infinitely divisible multivariate gamma distribution with parameters @xmath11 , @xmath12 and @xmath13 a positive definite matrix .
note that if @xmath14 , @xmath0 has a degenerate distribution .
if we choose @xmath15 and @xmath16 then @xmath0 follows a wishart distribution , i.e. , @xmath17 ( see anderson , ( 1984 ) @xcite , pp 252 ) . if @xmath18 , then @xmath0 is less than full rank , and the sampling distribution of @xmath0 is degenerate and that is no valid statistical inference can be implemented for such cases .
das and dey ( 2010 ) @xcite showed that if @xmath13 has prior as inverted multivariate gamma distribution , i.e. , if @xmath19 then the posterior distribution of @xmath13 is @xmath20 note that as long as @xmath21 , the posterior distribution is proper .
suppose @xmath22 , i.e. , @xmath23 , where @xmath15 ; then the sampling distribution of @xmath0 is degenerate .
however , if we choose the prior degrees of freedom parameter @xmath24 to be such that @xmath25 that is @xmath26 then the posterior distribution of @xmath13 is proper .
hence , we will be able to carry out bayesian inference .
if we choose @xmath27 and @xmath16 then the prior of @xmath13 is inverse wishart distribution , @xmath28 and posterior distribution of @xmath13 is @xmath29 for details see anderson ( 1984 ) @xcite .
the 3-series paper of ledoit and wolf ( 2003 , 2004a , 2004b ) , established the reasons for the sample covariance matrix failing to provide a good estimate for the portfolio covariance structure , and showed the need for regularization even when the problem is not ill - posed " . however , there was a lack of scope for implementing inferential procedures on their structural framework , regarding quantities like _ conditional contribution to total risk_. the main reason behind this , being the problem of assigning a suitable model for the anticipated return distribution . in this paper
an assumption for the covariance matrix @xmath30 , has an underlying assumption of normal / normal - component - mixture distribution on the anticipated returns .
the justification for the assumption @xmath30 is provided by the bayesian formulation as in gelman et al . @xcite .
the marginal posterior distribution for anticipated return would be @xmath31 .
this would provide a superior apprach to modelling the anticipated return in comparision to using a normal distribution . if @xmath32 , then we choose the prior degrees of freedom as @xmath33 for @xmath34 .
this ensures posterior distribution to be proper .
the posterior mode of @xmath13 is @xmath35 where @xmath36 .
clearly posterior mode of @xmath13 is a shrinkage estimator which is a weighted average of prior distribution s mode and sample covariance estimator .
das and dey ( 2010 ) @xcite showed posterior mode is also a bayes estimator under a kullback - leibler type loss function .
therefore under properly chosen prior degrees of freedom ( @xmath37 or @xmath38 ) and positive definite @xmath39 , posterior mode of @xmath13 is also a bayes shrinkage estimator which regularizes the solution .
the posterior distribution of @xmath13 is proper , which helps us to regularize the portfolio optimization procedure while conducting our empirical study in section [ empstudy ] . under the assumption of @xmath40=\sigma$ ] , where @xmath0 is the usual sample portfolio covariance matrix and @xmath41 ; ledoit and wolf ( 2004a ) @xcite showed that the folowing estimator , @xmath42 is asymptotically optimal under the frobenius norm . in this work
we assume @xmath17 , which implies @xmath43 . hence , the corresponding estimator is @xmath44 now we consider the definition of functional order from knuth(1976 ) @xcite . *
definition : * for functions @xmath45 and @xmath46 , let @xmath47 , @xmath48 and @xmath49 be defined as * @xmath50 , @xmath51 is the set of all such functions whose magnitude is upper - bounded by constant times @xmath52 . * @xmath53 , @xmath54 is the set of all such functions whose magnitude is lower - bounded by constant times @xmath52 .
* @xmath55 , @xmath56 is the set of all such functions whose magnitude is lower as well as upper - bounded by constant / s time / s @xmath52 .
* note 1 : * now we will check if the co - efficient of the first term in ( [ smode ] ) and @xmath57 in ( [ dhdest1 ] ) are of the same functional order .
this facilitates a measure for the amount of shrinkage towards the target . in ( [ smode ] ) ,
the shrinkage target is the prior , while in ( [ ledoitest ] ) and ( [ dhdest1 ] ) , it is @xmath58 in the equation ( [ dhdest1 ] ) @xmath59 . here ,
@xmath60 , @xmath61 $ ] and @xmath62 $ ] .
due to the imposed probability structure on the covariance matrix @xmath0 , we have the following results .
@xmath63 ^ 2}{p}\big]=\ ; \frac{1}{p^2}\big[(p-1)\sum_{i=1}^{p}\sigma_{ii}^2+\mathop{\sum \sum}_{i\neq j}^{p } \sigma_{ii}\sigma_{jj}(p\;\rho_{ij}^2 - 1)\big]$ ] @xmath61 = \frac{2}{n-1}\big[\frac{1}{p}\sum_{i=1}^{p}\sigma_{ii}^2\big]$ ] @xmath64 ^ 2}{p}\big\rbrace+\frac{2}{n-1}\big [ \frac{1}{p}\sum_{i=1}^{p}\sigma_{ii}^2\big]$ ] @xmath65 .
result 4 follows from the definition of fr@xmath66benius norm and the proofs of result 1,2 and 3 are presented in section [ app ] .
now we examine the denominator of @xmath57 in ( [ dhdest1 ] ) , @xmath67 ^ 2}{p}\big\rbrace+\frac{2}{n-1}\big [ \frac{1}{p}\sum_{i=1}^{p}\sigma_{ii}^2\big ] , \nonumber \\ \delta^2 & = & \frac{1}{p^2}\big[(p-1)\sum_{i=1}^{p}\sigma_{ii}^2+\mathop{\sum \sum}_{i\neq j}^{p}\sigma_{ii}\sigma_{jj}(p\;\rho_{ij}^2 - 1)\big]+\frac{2}{n-1}\big[\frac{1}{p}\sum_{i=1}^{p}\sigma_{ii}^2\big ] .
\label{deltaeqn}\end{aligned}\ ] ] multiplying both sides of equation [ deltaeqn ] by @xmath68 we get , @xmath69\nonumber\\ & = & a_1+a_2 , \nonumber\end{aligned}\ ] ] where , @xmath70 is a function of the variances , and @xmath71\end{aligned}\ ] ] is a function of the variances and the covariances .
we consider @xmath72 from @xmath73 , and if @xmath74 which reduces to @xmath75 , as @xmath76 , which implies @xmath77 . now , we consider @xmath78 from ( [ numer ] ) @xmath79 therefore , ( [ denomo ] ) provides a lower - bound for the denominator of @xmath57 .
now we consider the form of @xmath80 , from results 2 and 4 , @xmath81\big[\sum_{i=1}^{p}\sigma_{ii}\big]\label{varst}\\ & \leq & 2p^2 \big\lbrace \mathop{\mathrm{max}}_{i } \sigma_{ii}^2\big\rbrace \big\lbrace \mathop{\mathrm{max}}_{i } \sigma_{ii}\big\rbrace\nonumber\\ & \rightarrow & ( n-1)p^2\beta^2\mu = o(p^2).\nonumber\\\end{aligned}\ ] ] it provides an upper - bound for the numerator of @xmath57 .
hence we combine these findings in the following theorem .
if s , is the sample portfolio covariance matrix , such that @xmath17 , then for ( [ smode ] ) and ( [ dhdest1 ] ) we have , @xmath82 provided @xmath83 .
theorem 1 establishes the functional equivalence of the distribution - free approach and the bayesian approach .
note that from ( [ smode ] ) @xmath84 * note 2 : * theorem 1 implies that the ` degree of shrinkage ' towards the target for ( [ ledoitest ] ) , is less than a constant times @xmath85 ; whereas in ( [ smode ] ) , the degree of shrinkage towards the target , is exactly @xmath85 .
equation ( [ varst ] ) is a function of the variances , therefore we propose the following modification to the regularization , @xmath86 where , @xmath87 .
we have two choices of weights for the regularization . using ( [ orderbayes ] )
, we can choose @xmath88 as the bayesian weights and compare the performance with the asymptotic weights for @xmath88 presented in ledoit and wolf ( 2003 ) @xcite .
the asymptotic weight @xmath88 is obtained by minimizing @xmath89 $ ] , under the squared error loss , that is @xmath90 on solving the optimisation problem as in section [ app ] , under the probability structure imposed on @xmath0 , we have @xmath91\sum_{i=1}^{p}\sigma_{ii}^2-(n-2)\sum_{i=1}^{p}\sigma_{ii}-(n-1)\sum_{i=1}^{p}\sum_{j=1}^{p}\sigma_{ij}^2}{(n-2)^2[2\sum_{i=1}^{p}\sigma_{ii}^2+\sum_{i=1}^{p}\sigma_{ii}]}\big].\ ] ] we compare the performance of the resulting optimal estimators under the different weights , the first being @xmath92 with @xmath88 as in result 5 , and the bayesian shrinkage estimator @xmath93
ledoit and wolf ( 2004a ) @xcite presented asymptotic properties of the regularized covariance estimator , under a distribution free approach .
the lack of any distributional assumptions , prevents us from applying any inferential procedures for estimating quantities of interest , for example ` marginal contribution to total risk ' ( mctr ) and ` cctr ' .
the ` mctr ' and ` cctr ' of individual securities provide justification for the risk - behaviour of the portfolio . however , the distribution - free approach bypassed the need to model stock returns explicitly , and provided a justification for the method of regularization . in an attempt to approach a better regularization procedure , ledoit and wolf ( 2003 )
@xcite exploits the dependence structure of individual securities traded in a market , with the market index .
this approach resulted in a market specific regularization , by altering the shrinkage target .
in bayesian terminology this indicates that the prior information in comparision to ledoit and wolf ( 2004a ) is altered . from , the applicability point of view , the downside of this approach , was the computationally intensive calculation of the weights , for the regularization procedure of the covariance matrix .
as we have seen in section [ postcov ] , the theorem , provides a simple set of weights , showing functional equivalence with the posterior mode , of a shrinkage estimator under the minimal assumptions of a wishart probability structure on the covariance matrix .
moreover the new regularization procedure suggested for the covariance matrix , does not utilise any additional correlation structure with the market index .
this provides a more generalized and implementable approach to regularize the covariance matrix . under the assumption of no ` short - selling '
, the performance of markowitz s @xcite expectation - variance optimisation procedure with @xmath94 , has been known to falter due to singularity conditions already stated in the section [ intro ] .
regularizing , the covariance matrix provides a bypass to the problem .
@xmath13 , being ill - posed implies @xmath95 . for a strictly positive definite prior - information matrix @xmath96 @xmath97 @xmath98 , since @xmath99 .
the problem of maximizing the expectation and minimizing the variance as in markowitz ( 1952 ) , is transformed into the problem of utility maximization of the investor as in fabozzi et.al(2008 ) @xcite . under the general setup for optimization , we have the expected utility function e as a function of the returns and the variance .
@xmath100 where @xmath101 is the increase in risk , the investor is willing to tolerate per unit increase in return .
it serves as a relative risk aversion measure .
we assume that the expected utility function is @xmath102 , while proceeding with the optimization .
therefore , @xmath101 is a matrix of the relative risk aversion towards each security @xmath103 which is a function of the sharpe ratio , for indvidual securities .
the regularized covariance matrix is used to implement a quadratic - optimizer sub - routine to determine the weights for the optimal portfolio .
consequently , we can calculate the portfolio risk for the selected number of stocks in the portfolio .
once portfolio risk has been calcuated using the weights , the next step is to evaluate the sources of risk and how they interrelate .
there could be many different sources of risk , like individual stocks , sectors , asset classes , industries , currencies or style factors .
therefore , we keep the notion for _ sources of risk _ generic .
we consider an investment period where @xmath104 denote the return of source @xmath105 for the same period , where @xmath106 .
the anticipated portfolio return over the period is @xmath107 where @xmath108 is the portfolio s exposure to the source @xmath105 , i.e. , the portfolio weight , such that @xmath109 and @xmath110 , see ruppert ( 2004 ) @xcite . portfolio volatility is defined as @xmath111 where @xmath112 . the portfolio manager determines the size of @xmath108 at the beginning of the investment period , typically using markowitz - type optimization .
clearly , the weights ( @xmath108 ) play an important role as regulators of the portfolio s total volatility , along with the covariance structure of the portfolio .
we have already dealt with the issue of regularizing the portfolio s covariance structure .
however , it is also important for a manager to quantify , how sensitive the portfolio volatility is with respect to a small change in @xmath113 .
this can be achieved by differentiating the volatility with respect to weight , i.e. , @xmath114 where @xmath115 , is known as the ` marginal contribution to total risk ' ( mctr ) , see menchero _ etal . _
( 2011 ) @xcite and baigent ( 2014 ) @xcite .
note that the mctr for source @xmath116 is given by the following expression @xmath117 the cctr is the amount that a source contributes to the total portfolio volatility .
in other words , if @xmath118 is the cctr of source @xmath105 then @xmath119 therefore the total volatility can be viewed as weighted average of the mctr .
now in order to estimate the mctr and cctr , a regularized estimate of @xmath13 is required , because the portfolio weights @xmath120 are pre - determined by the manager . however , for all practical purposes a manager is more interested in estimating the @xmath121 or @xmath122 .
the reason being , @xmath123 or @xmath124 implies that the source @xmath105 reduces the total risk . in section [ postcov ]
, we presented that the posterior distribution of @xmath13 follows @xmath125 . also , in order to estimate the @xmath126 we present a monte carlo ( mc ) method based on independent replications as follows : * step 1 : for iteration @xmath116 , generate a sample @xmath127 from @xmath125 * step 2 : compute @xmath128 * step 3 : compute @xmath129 * step 4 : compute @xmath130 for all @xmath131 . * step 5 : set @xmath132 and go to step 1 .
note that these are independent replications , that is all the steps of iteration @xmath116 , does not depend on previous step @xmath133 .
if two parallel processors are available then iteration @xmath116 and @xmath133 can be implemented in parallel at the same time .
in fact one can consider the algorithm to be an ` embarrassingly parallel ' algorithm ( see matloff ( 2011 ) @xcite ) .
implementing the algorithm in parallel might not be required if @xmath1 is small , for example , if we consider four or five different asset classes as the risk sources .
but if @xmath1 is very large , like for any mutual fund portfolio , number of stocks might be more than thousands . for a large p , generating @xmath134 will be slow and thus , consequent calculations in all the other steps will be slow as well . in such cases ,
parallelization of the algorithm is required to improve the time complexity for the algorithm .
once the samples are generated , required mc statistics can be estimated easily , like @xmath135 where @xmath136 is the simulation size , @xmath137 if @xmath101 is true and @xmath138 if @xmath101 is false .
in this section we illustrate the methodology with two different sets of empirical data .
each exposes the inference procedure to a different practical situation .
the first dataset , displays the performance of the algoritm on a small dataset , consisting of 5 asset classes and 3 months of monthly return for each asset class .
the second dataset displays the performance of the algorithm on a large dataset , consisting daily log - returns of @xmath139 to @xmath140 stocks , over a period of 10 years .
firstly , we consider a portfolio with five asset classes , viz .
( i ) hybrid bond , ( ii ) emerging market , ( iii ) commodity , ( iv ) bond and ( v ) stocks , taken from the r - package ` ghyp ' ( @xcite ) .
the investment timeline consists of two time periods of study .
the first consisting three months of monthly return data ( may , june and july of 2008 ) , and the second time period ( for august , september and october of 2008 ) . as the second month of period 2 is september 2008 , when events like ` bankruptcy of lehman brothers ' took place , the total volatility during this period is very high .
however , for a portfolio manager it is important to identify the different sources of risk contribution from highest to least .
we consider the same portfolio weights ( see table [ portweighttable ] ) during both periods for fair comparison . here
@xmath141 and @xmath142 , which implies that regular sample covariance matrix estimator is not stable , as it provides non - unique solution for both periods .
we use the bayesian approach as discussed in section [ postcov ] .
we select the degrees of freedom for wishart prior to be @xmath143 , as @xmath144 , we maintain a balanced prior information by our choice of @xmath145 and choose @xmath146 , and @xmath147 is the identity matrix of order @xmath1 , as described in equation ( [ wishartprior ] ) , ( [ wishartposterior ] ) and ( [ eqchoosedf ] ) .
we considered simulation size as @xmath148 .
we present posterior estimates of total volatility in table [ totalvoltable ] and posterior density plot in figure [ post - total - vol ] .
monthly level total portfolio volatility is estimated by posterior mean and it goes up from @xmath149 to @xmath150 during first to second period .
the portfolio density becomes more positively skewed during the period 2 as presented in figure [ post - total - vol ] .
we present the posterior density of cctr ( @xmath151 ) of five asset classes , viz .
( i ) hybrid bond , ( ii ) emerging market , ( iii ) commodity , ( iv ) bond and ( iv ) stock for the three sets of weights in figures [ ad - hoc - cctr ] , [ blw - cctr ] , [ dhd - cctr ] and [ bond - cctr ] .
clearly we can say from these figures that posterior density of cctr of four asset classes ( viz .
hybrid bond , emerging market , commodity and stock ) have shifted towards right and have become more positively skewed during the second period .
it means that these four asset classes have contributed heavily to the total volatility - risk during the second period .
posterior estimates of all five asset classes are presented in tables [ poststatcctrm ] , [ poststatcctrsd ] and [ poststatcctrci ] . except ` bond ' , for four of the other asset classes , the posterior mean of cctr per unit increase in standard deviation went up significantly during the second period .
therefore we can conclude these four assets classes contributed statistically significantly large amounts towards the total volatility risk of the portfolio .
one point to be noted here is that the ad - hoc indicates that among these four asset classes , the posterior mean cctr of ` bond ' is the least compared to cctr s of ` emerging market ' , ` commodity ' and ` stock ' . in case of the other two weights namely , blw ( ledoit wolf with bayesian weights ) and dhd ( proposed weights , in section ( [ riskana ] ) ) , the least posterior mean cctr is presented by ` emerging market ' and ` commodity ' respectively .
this is explained by noting that the two methods focus mainly on decreasing volatility ( [ portweighttable ] ) for the portfolio by optimising the weights accordingly .
the allocation of funds is mainly towards ` bond ' .
this results in a comparitively superior portfolio , with the proposed weight structure ( dhd ) dominating over the others ( [ port - inf ] ) , showing lower portfolio volatility - risk , maximized return , and consequently a higher sharpe ratio in comparision to the others , during the ` crisis ' period . however , ` bond ' is the only asset class whose posterior density , as presented in figure [ bond - cctr ] , did not shift towards right .
rather it became more concentrated around zero under the proposed weight structure .
note that ` hybrid bond ' shows the maximum shift to the positive disrection in the second period , contributing greatly to increasing portfolio risk .
also we can see from table [ poststatcctrci ] that bayesian @xmath152 for bond shows a considerable spread to the negative side compared to the other asset classes . finally , using the formula as presented in equation ( [ eqprobcctr ] ) , we calculated the probability of positive cctr for both periods and presented the result in table [ tabprobcctr ] .
probability of positive cctr for all the asset classes , except bond , goes up during period 2 . during the same period , the probability of positive cctr for bond decreases
this showed us that during period 2 , which is commonly termed now as a ` crisis ' period , bond was the only asset class whose volatility - risk showed a decreasing behaviour .
this analysis clearly aligns with the intuitive understanding of the bonds market .
the advantage of this analysis is that it can provide reasoning to such intuitive understanding in terms of bayesian probability .
here we consider a market portfolio consisting of equities taken from the nsei ( national stock exchange of india ) market .
the investment period is fixed at half yearly intervals .
period 1 consists of the ( january - june ) , and period 2 ( july - december ) , for the years 2005 - 2014 .
the data for the respective years is downloaded from the site < https://www.quandl.com / data / nse?keyword= > .
the number of securities in the 10 year time period selected varies from @xmath153 to @xmath154 .
we consider the daily log - returns for each of the available securities , providing us with @xmath155 days of data , on an average on p securities for both periods .
the portfolio weights are obtained using the quadratic - optimisation technique on the data , using regularization procedures listed in section [ postcov ] .
we compare portfolios obtained under the two broad methodologies , that is ledoit and wolf ( 2004a ) and the proposed methodolgy as in section [ riskana ] , using @xmath156 , as the regularization matrix .
another comparision is made with the regularization @xmath157 ( [ lw - weight ] ) , for the sample covariance matrix .
portfolios obtained under the two methods are compared using following basis for comparision ( i ) half - yearly portfolio return , ( ii ) half - yearly portfolio risk , ( iii ) sharpe ratio and ( iv ) portfolio size .
we use the bayesian weights while conducting the regularization of the covariance matrix .
the wishart prior has the degree of freedom as @xmath158 , for the respective year .
the shrinkage target or prior @xmath159 , where @xmath160 , as defined in @xmath156 ( [ bayes - wt ] ) in section [ riskana ] .
the simulation size is @xmath148 .
the figures presented in ( [ return - risk],[esf - var],[sr ] ) provide a comparative summary for the two methodologies , on the basis of more recent methods of measuring risk , namely ` var'(value at risk ) and ` esf'(expected shortfall ) .
we present the out - sample performance , including the anticipated return and the volatility of the portfolios over the years .
the out - sample performance summary , for the portfolio constructed under the proposed weights is provided in table ( [ dhd - infer ] ) . in the case of out - sample performances ,
the period 2007 - 09 shows a clear dominance of the proposed method , while in other out - sample half - yearly periods it is dominant over the ledoit and wolf method , on an average .
we consider the period 2013 ( jul - dec ) in table ( [ infer-13 - 14 ] ) to demonstrate the procedure of inference on a portfolio constructed solely on equity , considering this as the investment period .
the same procedure of inference is carried out as in the case of ` indices ' dataset , the only difference is in terms of the size of the market portfolio is much larger , making the problem considerably more " ill - posed .
the shrinkage towards target is more sensitive to @xmath161 , while @xmath162 remains constant , implying @xmath163 .
note that from table ( [ dhd - infer ] ) , the size of the portfolio constructed under the proposed weights utilizes 10@xmath164 of the market.we now proceed to examine the stocks common to both portfolios .
for a stock with higher cctr that is , @xmath165 $ ] included in the portfolio constructed using the ledoit and wolf weights , presents an even higher @xmath165 $ ] in the portfolio constructed using proposed weights as shown in table ( [ infer-13 - 14 ] ) which provides a major point of difference between the two portfolios , apart from the significant difference in portfolio size .
the posterior summary statistics for cctr of 6 stocks over the two periods are presented in tables ( [ post - infer - dhd-1h ] , [ post - infer - dhd-2h ] ) .
the stocks x20microns and abbotindia in the first period ( jul - dec13 ) have a negative cctr , and contribute to decreasing the volatility - risk of the portfolio in the same period . comparing this to the second period
we see that abbotindia presents a positive cctr , showing that the @xmath165 $ ] has increased , which implies increase in portfolio volatility - risk .
x20microns preserves a negative cctr and shows more concentration near the posterior mean . in the second period ( jan - jun14 ) abgship has a negative cctr and shows a considerable decrease in cctr from the first period . the stocks astral and bajajhldng show decreased cctr , their distributions being more concentrated about the posterior mean , in comparision to the first period .
the stock bhushanstl , however shows no considerable change over the two periods .
finally , using the asymptotic weights , instead of the bayesian weights for the ledoit wolf method with the same shrinkage target as indicated in @xmath157 as in ( [ lw - weight ] ) , we can compute the value for @xmath88 , providing us with a comparable , but inferior result in comparision with the proposed portfolio weights .
the proposed weights also provide a portfolio of smaller size that lowers transaction cost for the investor .
in this paper we discussed about bayesian approach to regularize the ` ill - posed ' covariance estimation problem , establishing equivalence of the bayesian technique with the existing non - parametric techniques .
the method also analyzes the sources of risk by estimating the probability of cctr being positive for any particular security .
as cctr sums up to total volatility , it provides each source s contribution to total volatility .
regular method only estimates cctr of a source , but it does not estimate the probability of cctr to be significantly greater ( or less ) than zero .
this paper discussed the methodology to do so .
the existing and relatively new measures of risk , like esf and var , are used to analyze the performance of the proposed portfolio weights in comparison to the traditional methods .
we presented a parallelizable and easy to implement monte carlo method to carry out inference regarding the individual contribution of securities towards total risk .
we further presented two empirical studies , the first showed that during the crisis of 2008 , a portfolio consisting of five asset classes experienced large volatility risk due to significant increase in the contribution of stock and hybrid bond . during the same period
` bond ' was the asset class which contributed least on an average to the risk exposure of the portfolio .
secondly , the performance of the regularization technique under relatively higher dimensions , under computationally efficient bayesian analysis , to produce effective construction of a portfolio and inference regarding its risk exposure .
* proof of result 1 : * @xmath166 ^ 2}{p } & = & \sum_{i=1}^{p } \sum_{j=1}^{p } \sigma_{ij}^2 - \frac{1}{p}\big [ \sum_{i=1}^{p } \sigma_{ii}\big]^2\\ & = & \frac{1}{p}\big [ p\sum_{i=1}^{p } \sigma_{ii}^2 + p\mathop{\sum \sum}_{i\neq j}^{p } \sigma_{ij}^2- \big\lbrace \sum_{i=1}^{p } \sigma_{ii}^2 + \mathop{\sum \sum}_{i\neq j}^{p } \sigma_{ii}\sigma_{jj}\big\rbrace\big]\\ & = & \frac{1}{p}\big [ ( p-1)\sum_{i=1}^{p } \sigma_{ii}^2+\mathop{\sum \sum}_{i\neq j}^{p } ( p\ ; \sigma_{ij}^2-\sigma_{ii}\sigma_{jj})\big]\\ & = & \frac{1}{p}\big[(p-1)\sum_{i=1}^{p } \sigma_{ii}^2+\mathop{\sum \sum}_{i\neq j}^{p } \sigma_{ii } \sigma_{jj}(p\ ; \rho_{ij}^2 - 1)\big].\end{aligned}\ ] ] consequently we have , @xmath167.\ ] ] * * proof of result 2:**note that ,
the following result is true for the covariance matrix , s , @xmath168&=&\mathrm{tr}[\mathrm{e}(s^2)]\\ & = & \mathrm{tr}\big[\mathrm{var}(s)+[\mathrm{e}(s)]^2 \big]\\ & = & \mathrm{tr}\big[(n-1)((\sigma_{ij}^2+\sigma_{ii}\sigma_{jj}))_{i=1\ldots p , j=1\ldots p}+(n-1)^2\sigma^2 \big]\\ & = & ( n-1)\big[2\sum_{i=1}^{p}\sigma_{ii}^2+(n-1)\ ; \mathrm{tr}(\sigma^2)\big].\end{aligned}\ ] ] using the result obtained above we have , @xmath169 \\ & = & \frac{1}{p } \mathrm{e}\big[\mathrm{tr}\big\lbrace \big(\frac{s}{n-1 } -\sigma \big)\big(\frac{s}{n-1 } -\sigma \big)^{\prime}\big\rbrace \big]\\ & = & \frac{1}{p } \mathrm{e}\big[\mathrm{tr } \big(\frac{s^2}{(n-1)^2}\big ) -2 \mathrm{tr}\big(\frac{s}{n-1}\sigma \big ) + \mathrm{tr } \big(\sigma^{2}\big)\big\rbrace \big].\\\end{aligned}\ ] ] now using the fact that @xmath170=\ ; \mathrm{tr}[\mathrm{e}(.)]$ ] , @xmath171=\ ; ( n-1)\sigma$ ] , and the result above , we have @xmath172\\ & = & \frac{1}{p } \big [ \frac{(n-1)\lbrace 2\sum_{i=1}^{p } \sigma_{ii}^2 + ( n-1 ) \mathrm{tr}(\sigma^2)\rbrace}{(n-1)^2}-\frac{2}{n-1}(n-1)\mathrm{tr}(\sigma^2)+\mathrm{tr}(\sigma^2)\big]\\ & = & \frac{2}{n-1}\big [ \frac{1}{p}\sum_{i=1}^{p}\sigma_{ii}^2\big].\end{aligned}\ ] ] * proof of result 3 : * @xmath173\\ & = & \frac{1}{p}\mathrm{e}\big [ \mathrm{tr } \big\lbrace \big(\frac{s}{n-1}-\mu\mathrm{i}\big)\big(\frac{s}{n-1}-\mu\mathrm{i}\big)^{\prime}\big\rbrace\big]\\ & = & \frac{1}{p}\mathrm{e}\big[\mathrm{tr } \big(\frac{s^2}{(n-1)^2}\big)-\frac{2\mu}{n-1}\mathrm{tr}(s)+\mu^2\mathrm{tr}\mathrm{i}\big]\\ & = & \frac{1}{p}\mathrm{tr}\big [ \mathrm{e}\big(\frac{s^2}{(n-1)^2}\big)-\frac{2\mu}{n-1}(n-1)\sigma+p\mu^2\big]\\ & = & \frac{1}{p}\big [ \frac{(n-1)\lbrace 2\sum_{i=1}^{p } \sigma_{ii}^2 + ( n-1 ) \mathrm{tr}(\sigma^2)\rbrace}{(n-1)^2}-2\frac{[\mathrm{tr}(\sigma)]^2}{p}+\frac{[\mathrm{tr}(\sigma)]^2}{p } \big]\\ & = & \frac{2}{n-1}\big [ \frac{1}{p}\sum_{i=1}^{p}\sigma_{ii}^2\big ] + \frac{1}{p}\big\lbrace\mathrm{tr}(\sigma^2 ) -\frac{[\mathrm{tr}(\sigma)]^2}{p}\big\rbrace=\ ; \alpha^2+\beta^2.\end{aligned}\ ] ] * proof of result 5 : * @xmath174 ~=~ \mathop{\mathrm{argmin}}_{\rho \geq 0 } \big|\big|\sigma_3-\sigma\big|\big|^2\\ & = & \mathop{\mathrm{argmin}}_{\rho \geq 0 } \big|\big|\rho \lambda^{\prime}\mathrm{i_{p}}+(1-\rho)\frac{s}{n-1}-\sigma\big|\big|^2~=~\mathop{\mathrm{argmin}}_{\rho \geq 0 } \big|\big|\rho \mathop{\mathrm{diag}}_{i}~\lbrace s_{11},s_{22},\ldots , s_{pp}\rbrace+(1-\rho)\frac{s}{n-1}-\sigma\big|\big|^2.\\\end{aligned}\ ] ] now considering @xmath175 we have , @xmath176&= & \frac{1}{p}\mathrm{e}\big[\mathrm{tr}\big\lbrace\big(\rho \mathop{\mathrm{diag}}_{i}~\lbrace s_{11},s_{22},\ldots , s_{pp}\rbrace+(1-\rho)\frac{s}{n-1}-\sigma\big)\big(\rho \mathop{\mathrm{diag}}_{i}~\lbrace s_{11},s_{22},\ldots , s_{pp}\rbrace+(1-\rho)\frac{s}{n-1}-\sigma\big)^{\prime}\big\rbrace\big]\\ & = & \frac{1}{p } \mathrm{tr } \big(\mathrm{e}\big[\rho^2\mathop{\mathrm{diag}}_{i}~\lbrace s_{ii}\rbrace^2 + 2\rho(1-\rho)\mathop{\mathrm{diag}}_{i}~\lbrace s_{ii}\rbrace\frac{s}{n-1}-2\rho\mathop{\mathrm{diag}}_{i}~\lbrace s_{ii}\rbrace\sigma-2(1-\rho)\sigma\frac{s}{n-1}+\sigma^2\big]\big).\\\end{aligned}\ ] ] since , @xmath177 implies , @xmath40=(n-1)\sigma$ ] , we have @xmath176 & = & \frac{1}{p}\big[\sum_{i=1}^{p}(n-1)\rho^2(2\sigma_{ii}^2+\sigma_{ii})+2\sum_{i=1}^{p}\rho(1-\rho)(2\sigma_{ii}^2+\sigma_{ii})-2\sum_{i=1}^{p}\rho(n-1)\sigma_{ii}^2+\sum_{i=1}^{p}\frac{(1-\rho)^2}{n-1}(2\sigma_{ii}^2+\sigma_{ii})\big].\end{aligned}\ ] ] re - arranging the terms in the above equation we have @xmath176 & = & \frac{1}{p}\big[2\big\lbrace\big(\sqrt{n-1}\rho+\frac{1-\rho}{\sqrt{n-1}}\big)^2-\rho(n-1)\big\rbrace\sum_{i=1}^{p}\sigma_{ii}^2+\big(\sqrt{n-1}\rho+\frac{1-\rho}{\sqrt{n-1}}\big)^2\sum_{i=1}^{p}\sigma_{ii}+(2\rho-1)\sum_{i=1}^{p}\sum_{j=1}^{p}\sigma_{ij}^2\big].\end{aligned}\ ] ] if we denote @xmath178 $ ] , then to minimize @xmath179 we have the normal equation , by differentating w.r.t @xmath88 , and rearranging terms , for co - efficient of @xmath88 @xmath180\rho & = & [ ( n-2)^2 + 1]\sum_{i=1}^{p}\sigma_{ii}^2-(n-2)\sum_{i=1}^{p}\sum_{i=1}^{p}\sigma_{ii}-(n-1)\sum_{i=1}^{p}\sum_{j=1}^{p}\sigma_{ij}^2,\\\end{aligned}\ ] ] which is the required result . 99 golosnoy v. , okhrin y. multivariate shrinkage for optimal portfolio weights . the european journal of finance , 2007 ; 13:441 - 458 .
ledoit o. , wolf m. honey , i shrunk the sample covariance matrix .
journal of portfolio management , 2004 ; 30:4:110 - 119 .
ledoit o. , wolf , m. a well - conditioned estimator for large - dimensional covariance matrices .
journal of multivariate analysis , 2004 ; 88:365 - 411 .
ledoit o. , wolf m. improved estimation of the covariance matrix of stock returns with an application to portfolio selection .
journal of empirical finance , 2003 ; 10:5:603 - 621 .
markowitz harry m. portfolio selection .
journal of finance , 1952 ; 77 - 91 .
das s. , dey d. k. on bayesian inference for generalized multivariate gamma distribution .
statistics and probability letters , 2010 ; 80 : 14921499 .
bapat , r. b. infinite dividibility of multivariate gamma distributions and m - matrices .
sankhya : the indian journal of statistics , 1989 ; 51:1:73 - 78 .
anderson , t. w. an introduction to multivariate statistical analysis 2 ed .
wiley , 1984 .
ruppert d. statistics and finance an introduction .
springer , 2004 .
menchero j. , davis .
b. risk contribution is exposure time s volatility time s correlation : decomposing risk using the x - sigma - rho formula .
the journal of portfolio management , 1989 ; 37:2:97 - 106 .
baigent g. g. x - sigma - rho and market efficiency
. international journal of finance and banking , 2014 ; 1:1:39 - 43 .
matloff n. the art of r programming : a tour of statistical software design .
the journal of portfolio management , 2011 ; isbn : 9781593274108 : 347 e. graham , knuth .
big o micron and big omega and big theta .
sigact news , 1976 ; : 18 - 24 rachev svetlozar t. , john s. , hsu j. , biliana s. bayesian methods in finance ( frank j. fabozzi series ) .
wiley , 2008 .
andrew gelman , john b. carlin , hal s. stern , david b. dunson , aki vehtari , donald b. rubin .
bayesian data analysis .
chapman and hall / crc , 2004 .
david l. , wolfgang b. ghyp : a package on the generalized hyperbolic distribution and its special cases .
r package v 1.5.6 , 2013 .
< http://cran.r - project.org / package = ghyp>.
.portfolio weights for five asset classes [ portweighttable ] [ cols=">,>,>,>,>,>",options="header " , ] | it is important for a portfolio manager to estimate and analyse recent portfolio volatility to keep the portfolio s risk within limit .
though the number of financial instruments in the portfolio can be very large , sometimes more than thousands , daily returns considered for analysis are only for a month or even less . in this case
rank of portfolio covariance matrix is less than full , hence solution is not unique .
it is typically known as the ill - posed " problem . in this paper
we discuss a bayesian approach to regularize the problem .
one of the additional advantages of this approach is to analyze the source of risk by estimating the probability of positive ` conditional contribution to total risk ' ( cctr ) .
each source s cctr would sum up to the portfolio s total volatility risk .
existing methods only estimate cctr of a source , and does not estimate the probability of cctr to be significantly greater ( or less ) than zero .
this paper presents bayesian methodology to do so .
we use a parallelizable and easy to use monte carlo ( mc ) approach to achieve our objective .
estimation of various risk measures , such as value at risk and expected shortfall , becomes a by - product of this monte - carlo approach .
* keyword * : monte carlo , parallel computation , risk analysis , shrinkage method , volatility |
variable stars in globular clusters ( particularly binaries ) play an important role in understanding cluster dynamical evolution . despite this ,
such clusters have seldom been the targets of detailed study for the presence of variables , mainly due to the difficulty in obtaining accurate photometry from the ground of faint stars in very crowded fields . due to technological advances in recent years , however , the number of clusters studied and the list of variable stars discovered has increased dramatically . as a general overview of the field ,
those major clusters investigated recently include omega centauri @xcite , m5 @xcite , m71 @xcite , m4 @xcite , ngc6397 @xcite , m22 @xcite , ngc6946 @xcite , m69 @xcite , m15 @xcite , m13 @xcite and m3 @xcite .
@xcite discovered six spectroscopic binaries in ngc5053 . @xcite and @xcite review recent and more historical studies into globular variables .
a few previous surveys have searched for variable stars in 47 tuc .
@xcite discovered two variables .
@xcite used a hst dataset to detect 75 variable stars , including eclipsing binaries and variability among k - giants .
@xcite uncovered 107 variable stars , the largest number to date , and derived an overall binary frequency of 14@xmath54@xmath6 using the same hst dataset as @xcite to search for hot jupiter planets in the cluster .
@xcite performed a wider field survey on the cluster , uncovering 42 variables .
we present a new extensive variable star catalogue , a natural byproduct of a photometric project to search for transiting hot jupiter planets in 47 tuc .
the results of the planetary search will be the subject of a separate paper .
the total catalogue comprises 28 eclipsing binary systems , 20 long period variables and 41 small magellanic cloud ( smc ) rr lyraes , four halo rr lyraes , two cepheids and four apparent @xmath2 scuti stars , and one anomalous short period small magellanic cloud ( smc ) star .
discrimination between cluster / smc memberships is achieved using the location of the variable on the cluster colour magnitude diagram .
our large number of new discoveries is due primarily to the very large field of view ( 52@xmath052@xmath1 ) of our survey and also its photometric depth . in this paper
, we present the phase - wrapped v+r lightcurves , preliminary investigations into the detected variable stars , and a description of our photometric , astrometric and lightcurve database .
our survey covers a larger area than any previous search , and extends to deeper photometry than presented by @xcite .
we recover 31 of kaluzny s stars and discover a further 69 variables .
the unrecovered variables either lie between ccds , or within regions of no data caused by telescope offsets .
the cluster core can not be easily imaged by ground - based telescopes due to the extreme crowding . on our 300s exposures ( v+r ) , the inner 6@xmath1 of 47 tuc
is saturated .
the cluster field is located at a high galactic latitude ( l=305.9 deg , [email protected] deg ) , providing low foreground contamination by the milky way and low reddening .
the field is significantly contaminated by background stars from the small magellanic cloud .
our field of view extends to @xmath860@xmath6 of the 47 tuc tidal radius .
study of the smc rrlyrae stars , as standard candles , presents an opportunity to investigate the distance to the smc from a location some seven degrees nw of the centre .
contact eclipsing binaries ( ecb ) are very useful as distance indicators . observing any double line spectroscopic binaries offers an opportunity to measure directly the primary stellar parameters such as mass , luminosity , radius and hence distance .
@xcite presents a method of distance determination for contact binaries by including the period , unreddened v - i colour and system metallicity .
we present the results of an application of rucinski s calibration for our binary sample with periods less than 1 day .
section 2 describes the observations and data reduction , along with a description of the method used to obtain the lightcurves .
here we also discuss the astrometry and variable detection methodology , along with notes on cluster membership of the variables .
our survey completeness and the quality of the photometry is described .
section 3 deals with the cluster colour magnitude diagram ( cmd ) , the corresponding photometric database , and describes the photometric calibrations .
section 4 contains a discussion and preliminary analysis of the different types of variables in the catalogue and presents their lightcurves .
we summarise and present conclusions in section 5 .
finder charts of these stars are provided on a webpage link to allow easy identification on 47 tuc wide field images .
this project was undertaken using the australian national university ( anu ) 40-inch ( 1 m ) telescope at siding spring observatory .
we used the wide field imager ( wfi ) which consists of a 4@xmath02 array of 2048@xmath04096 pixel back - illuminated ccds .
the ccds are arranged to give a total format of 8k@xmath08k pixels .
the detector scale is 0.38@xmath9/pix at the 1 m telescope cassegrain focus , with a field of view of 52 arcmin on a side .
this very large field allows us to image much of 47 tuc , approaching 60@xmath6 of the cluster tidal radius of 45.9 arcminutes @xcite in one exposure .
the field layout is presented in fig.[astromplot ] .
our main aim of detecting planetary transits requires an effective sn ratio of 200 or more at v=18 for a 3@xmath10 detection and , to maximise the in - transit sampling , exposures no longer than 300s were employed .
our broadband filter covers the combined wavelength range of cousins v and r , giving a significant increase in signal - to - noise while maintaining the image degradation due to atmospheric dispersion at an undetectable level . from previous experience with this filter , for a star of v=18.5 in 2@xmath11 seeing ,
the photon noise s / n decreases from 220 at 7-day moon to 165 at bright moon . the globular cluster
47 tucanae was observed for a total of 33 nights from 2002 august 22 to 2002 september 24 with a field centre of ra=00h24m05.2s dec=@xmath772@[email protected]@xmath11 .
approximately 80@xmath6 of the observing time was useful for the main planetary transit project , with mean seeing of 2.2 arcsec .
the temporal coverage of the cluster was maximised as much as possible ; and averaged an image every 6 minutes for around 10 hours per night .
each image was checked for quality independently after readout .
unsuitable images caused by satellite trails , bad seeing periods , etc , were discarded .
in total we have 1220 images of the same field centered on 47 tuc , which have been used to produce time - series lightcurves for 109,866 unsaturated stars , with apparent magnitude 14.5@[email protected] .
this covers a large range of stellar mass and type , encompassing most of the red giant branch ( rgb ) the subgiant branch , the cluster turn - off , and the cluster main sequence down to a magnitude of v=22.5 .
this dataset therefore covers a large range of variable types .
initial reduction of the raw images was undertaken using the mscred package of iraf .
a significant number of calibration images were obtained over the 33-night run , and allowed for correction of time - dependent variations , eg .
, differences in flat fields from one observing period to the next , across the wfi array .
initial reductions included region trimming and overscan correction , bias correction , flat - fielding and dark current subtraction .
a dark - sky flat was obtained , but was not used because it degraded the large scale image quality .
after checking the final reduced images for flatness ( using imstat ) and pixel - to - pixel variations , the resultant 1220 object images were then ready for the main photometric pipeline and analysis .
the primary photometric method to generate lightcurves was carried out using a difference image analysis ( dia ) method originally described as a optimal psf matching algorithm by @xcite , and modified by @xcite .
detailed discussion of each program making up the pipeline can be found in wozniak s paper .
we summarize the method here .
dia consists of the following steps : firstly , all frames are transformed onto a common coordinate system . then a template image is produced by combining the best quality images with small offsets into a master image .
the point spread function ( psf ) of the stars on each image is determined using a combination of two gaussians , one for the core and one a factor of 1.83 wider for the wings , each multiplied by a third order polynomial . the best psf - matching kernel is found and each image is then subtracted from the template ( reference ) frame .
the coordinates of all stars on the template images are found , and finally the profile photometry is extracted from those star positions .
central to this method is the assumption that the pixels containing psf information in the analysis image vary only slightly due to seeing variations .
@xmath13 knowledge of the psf and the sky background is not required , and the method works better as the crowding increases , as in denser fields a larger number of pixels contain information about any psf differences .
heavy blending can cause trouble , however , as the pipeline photometry does not model surrounding stars and thus faint and/or heavily blended objects can be contaminated by neighbouring stars . for this reason ,
a crowding flag is introduced : a flag of 1 indicates that a pixel brighter than 0.15 @xmath0 distance in pixels @xmath0 pixel flux lies with in the immediate four pixel neighbourhood . for the production of the psf
, a star is rejected if another local maximum at least 2@xmath10 above the background level is present . for ease of data handling ,
each of the eight wfi ccds were cut into 32 subsections of 512@xmath0512 pixels each .
once the images were reduced , the 45 exposures with the best mean seeing , measured at 1.1@xmath11 were then coadded to create a template frame , a very high signal - to - noise version of each subsection .
this allowed initial flux measurements of the stars to be made , which defines the zero point of the output lightcurves .
the coordinates of all the stars detected by daofind on these 512@xmath0512 template images were fed into the main photometric pipeline .
those stars within 22 pixels of the subframe edge were discarded ; due to telescope pointing offsets those stars do not have complete temporal coverage . due to these offsets , @xmath821,000 stars ( @xmath815@xmath6 of total ) could not be produced for the final lightcurve database . after the subtraction process is complete
, the derived centroid of any variable object is unbiased by surrounding unvarying stars .
the dia method measures the flux differences between the frames , rather than the total flux .
the lightcurve of the variable object can be converted to total flux units using the stellar flux from the template .
as originally implemented , the method automatically detects objects that are classified as variable from frame to frame using the following conditions : at least three consecutive points depart by at least 3@xmath10 from the baseline in the same sense , and there are at least 10 points departing by 4@xmath10 from the baseline in either sense .
the method then produces a lightcurve at the pixel coordinates of the detected varying object . as the main aim of the project was to find the periodic @xmath81 - 2@xmath6 drop in brightness caused by a planetary transit , which is practically undetectable by this automatic process , we altered the method slightly .
the program was forced to produce a lightcurve for every visible star on our best - seeing image , which yielded the largest number of stars of any of our dataset images .
using daofind within iraf , the pixel locations of all stars were determined . the main photometric pipeline within dia
was then instructed to produce a lightcurve for each of these locations . the variable stars presented in this paper
were detected using this modified method .
this implementation of the method gave us a total 1200-point lightcurve database across the 47 tuc field for 109,866 unsaturated stars .
this is a smaller number than that of our total astrometric database , due to telescope offsets and the gaps between images .
the dia method produces lightcurves that are not on a standard magnitude system .
the lightcurves are produced with linear flux units , from which a constant reference flux taken from the template images has been subtracted . to convert this unit into a magnitude scale , the flux of each star on the reference image must be determined .
an estimate of the flux was made by running daophot on the reference template images and measuring the flux ( in counts ) for each star .
the change in light output in magnitudes can then be calculated via : @xmath14 $ ] where @xmath15 is the flux of star @xmath16 on the template image , and @xmath17 is the difference flux on a given frame in the time series .
we then compared our range of @xmath18 for those variables found by @xcite for verification .
these @xmath18 are thus used as the magnitude total variability level of our detected variables .
fig.[ampmags ] shows the amplitude comparison between the two datasets for those variables which can be crossidentified , and shows a scatter around the zero - point with a mean of -0.0016 and a standard deviation of 0.13 magnitudes .
the slope of the lpv variables towards fainter magnitudes is attributed to the different passbands of @xcite ( v ) and ourselves ( v+r ) .
those ecb which are not close to the zeropoint are members of the binary main sequence , and hence are more likely to be composed of different colour components .
the rr lyrae scatter is attributed to the relative faintness of this sample .
our photometric errors are presented in fig.[rmstot ] , which shows the v+r rms photometric uncertainties of our lightcurves as a function of v magnitude .
as we have a significant change in crowding between the inner four ccds and the outer four ( see fig.[astromplot ] ) , two rms curves are presented .
it can be seen that the outer ccds produce comparable lightcurve quality for stars up to 0.5mag fainter for any given rms level compared to the inner ccds .
we therefore have sufficient photometric precision to detect a 1.5 - 2.5@xmath6 dip ( typical of the general range of a hot jupiter planet @xcite in the lightcurves of v=18.5 stars in a crowded inner ccd and v=19.0 in an uncrowded outer ccd .
this limit allows us to probe the brightest 1.5 - 2 magnitudes of the cluster main sequence for any orbiting giant planets ; these results will be presented in a subsequent paper .
our rms precision also allows us to detect variable stars to v@xmath821 with amplitudes of ( 0.14,0.20 ) magnitudes for the two crowding levels , giving us an excellent chance to detect previously unknown variable stars .
fig.[starcount ] shows the magnitude depth of our photometry for all eight ccds , with recovered stars counted in 0.25 magnitude bins .
our photometry is limited to the range 13.0@[email protected] , with different chips having slightly different sensitivities .
the apparent peaks in the stellar distributions are due to incomplete sampling at faint magnitudes .
the gradual decrease in star numbers towards ccd8 can be clearly seen , indicative of increasing distance from the small magellanic cloud ( smc ) , and hence decreasing background contamination .
astrometry was obtained for all 143,814 stars detected in our best seeing image , across the eight ccds of the 47 tuc field allowing a determination of the position of all stars in both our photometric and lightcurve database .
we used a program to search the usno ccd astrograph catalogue ( ucac1 ) for astrometric standard stars within the field ( b.p.schmidt , 2003 private communication ) .
several hundred such stars were cross - identified , allowing for an accurate determination of the astrometric solution of our star lists .
the rms residual of the astrometry was @xmath80.15 arcsecs .
our astrometry is presented in fig.[astromplot ] , with the ccd number overplotted .
the cluster core and our extensive coverage of the cluster is apparent .
the v , v - i colour magnitude diagram ( cmd ) used in the production of the variable colour data is presented in fig.[47tuccmd_total ] .
these data were originally taken at the mssso 40@xmath11 telescope by ken freeman and michelle doherty to allow for a placement of any candidate transiting system on the v , v - i system .
the data cover the same 52@xmath052@xmath1 fov as the lightcurve dataset presented in this paper , and totalling 43,067 stars .
the magnitude range is 13.0@[email protected] in v and i. the cmd was calibrated against that presented by @xcite .
the authors warn in that paper of systematic errors caused by non - linearity of the ogle ccd chip . for faint stars these errors are likely to be more significant .
the ogle dataset as available on their website ( calibrated ) was overplotted on top of our uncalibrated cmd data .
our data was then shifted until the two datasets overlay eachother .
a histogram of the distribution of the stellar magnitudes was produced , and our data was shifted further until a more accurate match was found by comparing the v magnitude of the horizontal branch .
this was repeated for v - i calibration .
this simple calibration method produced photometry accurate to @xmath40.03 mag , across the full magnitude range .
fig.[47tuccmd_total ] shows our cmd .
variable stars in the v+r dataset were cross - identified by comparing the template image containing the variable and the corresponding ccd image of the cmd dataset .
the v and v - i information in tables 3,4 and 5 are associated with uncertainties of 0.03 and 0.04 respectively , and are the non - random errors in the zero - point determination and incorporate the errors in the ogle calibration . a lomb - scargle periodogram ( lsp ) method was chosen to search for periodic variables in the final lightcurve database @xcite .
this method was produced specifically to overcome the problem of unevenly spaced data , which in our case is caused by daylight and cloudy nights .
a fourier power spectrum is produced , with the same statistical properties as a standard power spectrum .
if a periodicity ( p ) is detected in the data , a spike is produced in the spectrum at a frequency of 2@xmath19/p .
as implemented , any spike above 2@xmath0rms of the power spectrum is identified as a candidate . using this condition , 29,314 lightcurves which contained such an apparent periodicity ( @xmath823@xmath6 total database
, @xmath8300 times the final variable number ) were found , which were then examined in detail .
it was noticed that the vast majority of this list was composed of common systematic effects inherent in the data , lightcurves from stars close to saturation , stars very close to our magnitude limits and the @xmath202@xmath10 variables .
all those candidates which exhibited a clear periodicity by eye were catalogued , leaving a sample of 100 for the variable list .
fig.[detlim ] shows our detection limits .
the log of the total amplitude ( in magnitudes ) of the detected variables have been plotted against v. it is clear that there is a sharp cutoff to the detections , marked with a dotted line . any amplitudes which are @xmath203@xmath6 are detectable to v=17.0 , and only those with an amplitude @xmath201 mag can be found for v@xmath2022 . hence those variables which lie underneath this limit are missed in our dataset . by phase - wrapping
the candidate lightcurves at the detected period , the nature of the periodicity is seen .
the lsp - derived period is very close @xmath80.03d to the chosen true periods presented for our variables in table1 .
once phase - wrapped , the period was altered ( at the third significant figure ) until a minimum in the scatter of the points was seen ; this value was recorded as the period of the variable .
our cross - identified variables had their derived periods compared to those in @xcite , and are in excellent agreement .
phase - wrapped lightcurves of all 100 variable stars detected in our dataset are presented in figs.[varplot1]-[varplot10 ] .
we arrange the lightcurves by type , ecb ( figs.[varplot1]-[varplot3 ] ) , rr lyraes ( figs.[varplot4]-[varplot7 ] ) , long period variables ( lpvs , figs.[varplot8]-[varplot9 ] ) and miscellaneous variables ( fig.[varplot10 ] ) . for the ecb ,
it is clear that examples of contact ( ie v6 ) , semi - contact ( ie v7 ) and detached configuration ( ie v69 ) binaries are present , classified due to the appearance ( or lack thereof ) of features in the lightcurves indicating tidal distortion of the stellar components , the period , and the length of time between eclipses .
the ecb sample consists mostly of w ursae majoris - type lightcurves , with short periods .
most of these stars seem to be candidate blue stragglers with cluster membership ( see fig.[linecmd ] ) .
the lightcurves of some of these stars ( ie , v6 , v7 , v12 ) display secondary variations outside the main variability , which could be indicative of star - spots , flares or roche - lobe overflow between the components ( e.c sung , 2003 private communication ) .
one of the main explanations for the phenomenon of blue straggler stars ( bss ) is that the two binary components are tranferring mass , and hence remain on the main sequence when they otherwise would have evolved off .
@xcite showed that many bss are members of short - period binaries , and as such are likely to be transferring material from one to the other as one component evolves . hence it is not surprising that some of our sample of blue stragglers ( ie v6 ) seem to show evidence of such mass transfer activity in their lightcurves .
a number of ecb appear to lie on the cluster binary main sequence ( bms ) and as such all are very likely members of 47 tuc .
detached systems are also seen in the sample ( v39 , v41 , v69 , v78 , v84 , v89 and v93 ) . of these
, four do not show detectable secondary eclipses to our rms level ( v69 , v78 , v89 , v93 ) , and thus could conceivably be orbited by low luminosity companions , most likely m - dwarfs .
our data for v78 and v93 is very limited , with only one eclipse visible across our temporal range .
our period estimates for these two variables are therefore not well determined .
we have estimated a period for these two stars which would hold if all other eclipses occured during cloud or daylight .
further observations are required to derive an accurate period .
if these stars are indeed orbited by m - dwarfs , these variables would be important to determine the survivability and long - term stability of such low - mass companions inside globular clusters @xcite .
we present phase wrapped lightcurves of these two special cases in fig.[mdwarf ] , with the eclipses plotted more clearly than in the general lightcurve database figures .
v78 also shows an apparent variation at 1.015d , as well as the longer period for which only one eclipse is seen .
this is shown in fig.[varplot2 ] ( two plots for v78 ) , and as the period is so close to 1 day , it is almost certainly due to terrestrial effects . it is included for completeness .
the lightcurve data suggest that the companion sizes are approximately 0.25 - 0.3 solar radius , assuming a mid - to late k - type primary at the distance of 47 tuc .
the rest of our ecb sample are detached binaries on the cluster bms .
the apparent frequency of the occurence of detectable 47 tuc contact binaries in the field is 21/124073 @xmath21 [email protected]@xmath010@xmath22 , which is slightly higher , but consistent with , the estimate of 1@xmath010@xmath22 presented in @xcite , and is more than an order of magnitude lower than that observed in the core of 47tuc @xcite , and in fields containing galactic open clusters @xcite and ogle fields located in baade s window , close to the galactic centre @xcite .
the reasons for this difference are unclear , although mass segregation may play an important role in determining the radial distribution of such stars . for comparison ,
some other globular clusters studied with variability surveys give eclipsing binary frequencies of 7.5@xmath010@xmath22 for m71 @xcite and 1.4@xmath010@xmath23 for m5 @xcite .
our dataset shows an apparent segregation in binary period , with shorter period binaries located closer ( in projection ) to the core of 47 tuc .
this can be seen clearly in fig.[ecbdist ] where we have plotted the number of contact and detached ecb as a function of radius from the core in bins of 6 arcminutes .
a ( normalised ) histogram of the total star population has been overplotted for comparison .
the two ecb distributions are clearly different , with a significance of around 4@xmath10 , as determined by a kolmogorov smirnov test .
contact systems are preferentially located closer to the core ( in projection ) than detached systems of the same magnitude .
this segregation is not a classical indication of mass segregation , for which we would expect both detached and contact binary systems to be preferentially located closer to the core , but may be an indication of dynamical effects , with binaries closer to the core having lost much of their orbital energy .
the binaries further out retain their energy , and hence remain in longer period orbits . due to dynamical effects
, it is expected that most of the cluster contact binaries would be located in the core @xcite , although some would have been ejected from the core by star - star encounters .
a theoretical king profile has also been overplotted to show the total stellar distribution , without the crowding our data suffers towards the core .
the inner 6@xmath1 is not sampled in our dataset ; the dropoff on binary number at the inner 6 arcminutes is entirely due to this effect .
fig.[ecbpcol ] shows the sample of our contact ecb ( periods @xmath40.5d ) indicating a clear relationship between period and v - i colour .
this period - colour relationship was first noted by @xcite , and further investigated by @xcite .
the explanation as offered by these authors is that this relation is indicative of the differing system masses , the higher mass systems are bluer and have longer orbital periods than the lower mass systems .
as one moves further down the cluster cmd ( in our case ) , the mass of the system in total decreases , and the overall system colour gets redder . since massive stars are physically larger than low - mass stars ,
those binaries which have high mass components must have a greater orbital separation to accommodate this larger stellar volume .
therefore the lower the likely masses of the components the redder they appear , and the shorter the orbital period will be .
system metallicity has a bearing on the positions of the contact binaries in the period colour relationship .
@xcite showed that the reduced line blanketing of metal - poor stellar atmospheres accounted for some of the differences between a globular cluster period colour relationship and that of field binaries .
the lower metallicity also makes stellar radii smaller , affecting the resultant orbital period for such systems .
we do not have definitive proof that our variable stars are members of 47 tuc , the background smc , or the galactic halo .
spectroscopic observations are required to confirm membership ; the heliocentric radial velocity of 47 tuc ( -19.4km / s ; @xcite ) is very different to that of the galactic halo and the smc ( + 158km / s ; @xcite ) . in this paper , we have estimated membership for variables from their location on the cluster colour magnitude diagram ( cmd ) , which is shown schematically in fig.[linecmd ] . as a globular cluster ,
47 tucanae has a very small spread of stellar metallicity , so that the stars lie on tight loci on the cluster cmd , without the spread in colour observed in other objects such as the smc .
as such , we can preliminarily assign memberships which are presented in tables 1 - 5 . for the contact eclipsing binary ( ecb ) systems , we have adopted rucinski s absolute brightness calibration @xcite to calculate m@xmath24 , distances and thus memberships .
the method estimates m@xmath24 from the period , unreddened colour , and system metallicity via : @xmath25}.$ ] for all our ecb systems , we have adopted [ fe / h]=-0.76 and e(v - i)=0.05 @xcite .
fig.[ruccal ] shows the derived distance modulus of all ecb for which we have complete colour information and have periods p@xmath261d ; such systems can be regarded as contact systems .
the derived distance modulus ( dm ) of 47 tuc from this plot is [email protected] , which is in agreement with the estimate of 13.21 presented by @xcite , and that derived by @xcite , using main sequence fitting , of @xmath27 . from fig.[ruccal ] , it is clear that we have detected 10 eclipsing binaries that are likely members of 47 tuc .
v95 and v26 appear to be foreground members of the galactic halo , whereas v20 is a likely member of the small magellanic cloud @xcite .
interestingly , v11 and v75 both have distance moduli which lie inbetween 47 tuc and the smc .
it is interesting to note the very small amplitude of variation associated with the foreground ecb v26 , perhaps indicating low mass components . to estimate the total number of variable stars present in the wfi field we consider how many of @xcite s sample we missed due to telescope offsets and gaps between our ccds .
we recovered 31 of the 42 variables presented in that paper . using this result
, we estimate that we missed @xmath826@xmath6 of the variables we are capable of detecting .
we therefore expect that in total there are 126@xmath311 detectable variable stars present in the field . a significant number of small magellanic cloud ( smc ) rr lyrae stars were found in our data .
their phase - wrapped lightcurves are presented in figs.[varplot4]-[varplot7 ] , along with a preliminary examination .
all but four rr lyrae stars are clustered around v=19.68 , indicating their membership in the smc .
the other four stars are very likely to be located in the galactic halo , one of which ( oglegc223=v10 ) was presented in @xcite .
three populations of rr lyrae stars are found in our data : type ab , type c , and two examples of rr lyrae stars with periods of about 1 day .
these latter stars ( type ahb1 ) are described by @xcite , and constitute very low metallicity post - horizontal branch ( hb ) stars passing rapidly through the instability strip in the vicinity of the horizontal branch but on bright evolutionary tracks @xcite .
both of our ahb1 stars are galactic halo stars , and hence indeed likely to be of a low metallicity .
the identification of these ahb1 stars is presented in the rr lyrae period distribution ( fig.[rrlyrp ] ) , and are also identified on the rr lyrae p - v diagram ( fig.[rrlyrpv ] ) . by comparing the density of rr lyraes in our sample with those previously published for the field of 47 tuc , we can draw conclusions about the completeness of our rr lyrae sample .
@xcite searched for variables in an area covering 4680 arcmins@xmath28 north of 47 tuc , which included a small part of the cluster .
the rr lyrae density presented in that paper was 0.016 variables per square arcminute , and is the generally accepted value for the rr lyrae density in this region of the sky .
the surface area covered by our search is 2704 arcmin@xmath28 , and yields a derived rr lyrae density value @xmath21 [email protected] rrlyr / arcmin@xmath28 , consistent with the @xcite result .
the limiting magnitude of graham s search was [email protected] , whereas ours is [email protected] .
we did not identify any 47 tuc rr lyraes in our dataset , due to their relative brightness compared to our target stars . with our 300s exposures , at v=14.06 @xcite such cluster stars would be very close to saturation .
fig.[rrlyrpv ] shows the period - luminosity diagram for our sample of rr lyraes .
those that lie in the smc are easily distinguishable from those that lie in the galactic halo ( marked with a h ) .
the long - period ahb1 rr lyraes are also identified .
the average v magnitude of the smc stars lies at [email protected] .
the mean v absolute magnitude of rr lyraes in the smc is m@xmath29(rr)=0.75 , as assumed by @xcite , and also used by @xcite and includes correction for metallicity effects .
it therefore follows that the smc distance modulus ( m - m)@xmath24 from our sample of rr lyraes is [email protected] or [email protected] kpc .
the large errorbar is due to the relatively small sample size .
the average magnitude of each of the rr lyrae stars was found , by integrating across the lightcurve , and then the phase of the variability amplitude at the time of the cmd dataset was used to determine the actual apparent magnitude at this same time . the difference between this and the mean magnitude
was then measured . the error in this measurement as plotted on fig.[rrlyrpv ]
was taken as the residual scatter in the phase - wrapped lightcurve points at this time .
this method allows us to account for the variability amplitude when finding the v magnitudes of our stars , and hence a more accurate measure of the smc distance .
our result compares favourably with that of [email protected] presented by @xcite
. a larger sample size would help to indicate more conclusively if the smc is extended in the radial dimension towards the direction of 47 tuc .
the blazhko effect is a little understood feature of some rr lyrae stars , in which the amplitude of variation itself varies with a certain periodicity .
one popular theory to explain
this erratic behaviour is that it is related to the presence of a strong photospheric magnetic field @xcite , yet recently @xcite has ruled out a magnetic field for the brightest blazhko ( bl ) star - rr lyrae itself , the prototype of the class .
neither this , nor the rotating resonant pulsator model @xcite explain observed bl star properties .
the frequency of such bl stars seems to be dependent on the metallicity of the environment in which they occur , @xcite has found a different incident rate of bl stars in the lmc ( @xmath811@xmath6 ) when compared to @xcite , who found a 23@xmath6 incidence rate for the galactic bulge .
both papers agree that metallicity is the most probable reason .
the reason for bl behaviour in general therefore remains unknown .
it is seen on a few of our rr lyrae sample ( v2 , v3 , v18 , v86 ) .
fig.[v86b ] shows the remarkable regularity of the blahzko effect for v86 , for which we plot the phase - wrapped lightcurve over four periods .
the amplitude of the variation decreases significantly over three primary periods .
this behaviour was constant over the whole 33-night run .
the other three blahzko examples do not show such regularity .
a few of the rr lyrae lightcurves show significant scatter , primarily due to crowding , and variable seeing .
our sample of variables also includes 20 long period variables ( lpvs ) with locations on the cmd that are consistent with two main populations : those present in the red giant branch ( rgb ) of 47 tuc itself , and those which lie on the agb / rgb of the smc .
we define a long period as being significantly longer than that of rr lyrae and ecb stars , from about 5 - 6 days .
stars with very long periods can not have their periods accurately determined from our dataset , as our temporal coverage is insufficient to show even a single period ; we present lower limits on the periods of the more extreme lpvs .
some of these stars were found by @xcite .
these extreme lpvs are likely to be examples of agb stars ( miras ) in the smc .
v27 ( oglegc230 ) is affected by bad seeing and extreme crowding , and hence has significantly more scatter on its lightcurve .
our derived period for this star is half that derived by @xcite , who indicated that their period might require revision .
only about six of our lpv sample can be given tentative 47 tuc membership , a number too small to give meaningful results to study their radial distributions .
it is very likely that the majority of cluster rgb variables are brighter than our magnitude limit .
a shorter exposure search for variability among 47 tuc rgb stars has recently been started by kiss et al ( 2003 , private communication , also with wfi on the mssso 40-inch ) , and our lpv sample will overlap somewhat with their results .
this should allow a more accurate study into the cluster lpv radial distribution .
a small number of other variables were also discovered in our dataset , including two cepheids , four @xmath2 scuti stars , and an anomalous short - period red variable , which is a likely smc star .
the two cepheids ( v24 and v37 ) are identified from their position in the schematic cluster cmd ( fig.[linecmd ] ) .
they are significantly brighter and redder than the rr lyrae stars , but are of short period ( 0.387d and 2.572d respectively ) for cepheids , and as such could be classified as anomalous .
v24 has been tentatively identified as a typeii cepheid based on the secondary variation seen on the lightcurve at phase @xmath80.5 .
we detected four @xmath2 scuti stars in our search ( v35 , v54 , v67 and v80 ) .
all are certainly members of the smc , as they have v@xmath821 .
this is at the limit of our detectability , but they were found due to their large amplitude of variation . v35,v54 and
v67 all have very short periods @xmath260.1d , typical of @xmath2 scuti stars .
v80 has a period that is longer , at 0.2144d , and is among the faintest variables in our catalogue at v@xmath822 .
it has been classified as a @xmath2 scuti star due to the shape of the lightcurve , and the amplitude of variation .
the @xmath2 scuti lightcurves presented here show a significant amount of scatter , which is attributed to photometric scatter caused by the faintness of the sample . to investigate the possibility of multi - periodicity ,
the periodograms for these four stars were compared to those of non - variables of the same magnitude .
the lightcurves were phase - wrapped to all significant periods found with the lsp ( see earlier ) , the majority of the extra periodicities were found to be common , and attributed to systematic effects . no obvious secondary period was seen for these stars down to our sensitivity level .
the amount of photometric scatter at v@xmath821 is @xmath80.3 mag , and is consistent with the scatter seen on these four lightcurves .
finally , from its position on the schematic cmd v60 is an apparent smc agb star , with v - i=2.99 , yet a periodicity of only 0.2544d .
the amplitude of variation is moderate ( @[email protected] ) , and as such indicates an apparent giant red star with a very short period pulsation .
it follows that such a star would likely be unstable at such short periods . at the distance of the smc
derived above , the absolute magnitude would be -1.8 , typical of an a - type star .
it therefore seems likely that this star is actually much bluer , but could possibly be a post - agb star surrounded by a dusty shell .
future monitoring and imaging of this star would be useful to unravel the mystery .
we have presented data for 100 variable stars detected across a wide ( 52@xmath052@xmath1 ) field centered on the globular cluster 47 tucanae . of these 100 , 69 are new discoveries .
the sample consists of 41 apparent small magellanic cloud rr lyrae stars , four halo rr lyrae stars , 28 eclipsing binaries , 20 long period variables , four @xmath2 scuti stars and two cepheids .
we also detected one anomalous short period red giant , perhaps surrounded by a dusty region .
four of our rr lyrae sample display blahzko effect variations , one with remarkable regularity .
this catalogue more than doubles the number of known variables in the 47 tuc field . of the ecb sample
, four variables are perhaps orbited by faint companions , most likely m - dwarf stars .
such stars are important in determining the long - term stability and survivability of low - mass objects in close orbits inside globular clusters .
future spectroscopic observations of these candidates are planned . as well as presenting this new variable catalogue
, this paper presents a new complete database of v and i photometry , astrometry and 33-night v+r lightcurves for 109,866 stars across the field .
the distance modulus of both 47 tucanae and the smc have been determined from our sample .
the values are consistent with those already published for these two objects .
it is clear from the eclipsing binary results , as well as those previously published in the literature , that the relative frequency of contact binaries in the field of 47 tuc is very low compared to other studied regions .
the reasons for this difference are unclear , although mass - segregation and dynamical effects seem to play an important role .
our sample of ecb shows a distinct period / radial - distance segregation , perhaps indicative of dynamical relaxation .
quite possibly the more massive shorter period contact binaries are located preferentially within the cluster core , an unsampled area in our experiment .
the authors wish to thank the following people for their help in the producing this piece of work : matthew coleman for helping the first author relieve the insanity of observing for 33 nights straigh ; eduard westra for help in debugging the photometric pipeline , and for answering many emails to that effect ; brian schmidt for allowing use of his astrometric package , as well as instructions on its operation ; laszlo kiss for looking over some of the lightcurves and clearing up the identifications of the more noisy variables , and doug welch for helpful comments while acting as referee . finding charts for all variables are available on the electronic edition of this paper .
the orientation of these charts has north to the top and east to the left , produced from the corresponding template image which contains the variable , and they are each 3.24@xmath1 on a side .
adams , f. c. & laughlin , g. 2003 , icarus , 163 , 290 alard , c. & lupton , r. h. 1998 , , 503 , 325 albrow , m. d. , gilliland , r. l. , brown , t. m. , edmonds , p. d. , guhathakurta , p. , & sarajedini , a. 2001 , , 559 , 1060 alcock , c. , et al .
2003 , , 598 , 597 bretthorst , g. l. 2001 , aip conf .
proc . 568 : bayesian inference and maximum entropy methods in science and engineering , 568 , 241 chadid , m. , wade , g. a. , shorlin , s. l. s. , & landstreet , j. d. 2004 , , 413 , 1087 charbonneau , d. , brown , t. m. , latham , d. w. , & mayor , m. 2000 , , 529 , l45 clement , c. m. et al . 2001 , , 122 , 2587 cousens , a. 1983 , , 203 , 1171 dziembowski , w. a. & cassisi , s. 1999 , acta astronomica , 49 , 371 edmonds , p. d. , gilliland , r. l. , guhathakurta , p. , petro , l. d. , saha , a. , & shara , m. m. 1996 , , 468 , 241 edmonds , p. d. & gilliland , r. l. 1996 , , 464 , l157 eggen , o. j. 1967 , , 70 , 111 gilliland , r. l. et al .
2000 , , 545 , l47 graham , j. a. 1975 , , 87 , 641 gregorsok , j. d. , layden , a. c. , welch , d. l. , & webb , t. m. a. 2003 , american astronomical society meeting , 203 , guldenschuh , k. a. , layden , a. c. , welch , d. l. , & webb , t. m. a. 2003 , american astronomical society meeting , 203 , haggard , d. , dorfman , j. l. , cool , a. m. , anderson , j. , bailyn , c. d. , edmonds , p. d. , & grindlay , j. e. 2003 , american astronomical society meeting , 203 , harries , t. j. , hilditch , r. w. , & howarth , i. d. 2003 , , 339 , 157 harris , w. e. 1996 , , 112 , 1487 sawyer hogg , h. 1973 , , 67 , 8 hut , p. et al .
1992 , , 104 , 981 hut , p. 1996 , iau symp . 174 : dynamical evolution of star clusters : confrontation of theory and observations , 174 , 121 kaluzny , j. , kubiak , m. , szymanski , m. , udalski , a. , krzeminski , w. , & mateo , m. 1996 , , 120 , 139 kaluzny , j. 1997 , , 122 , 1 kaluzny , j. & rucinski , s. m. 1993 , asp conf . ser . 53 : blue stragglers , 164 kaluzny , j. , kubiak , m. , szymanski , m. , udalski , a. , krzeminski , w. , & mateo , m. 1997 , , 125 , 343 kaluzny , j. , thompson , i. b. , & krzeminski , w. 1997 , , 113 , 2219 kaluzny , j. , kubiak , m. , szymanski , m. , udalski , a. , krzeminski , w. , mateo , m. , & stanek , k. z. 1998 , , 128 , 19 kopacki , g. , koaczkowski , z. , & pigulski , a. 2003 , , 398 , 541 leon , s. , meylan , g. , & combes , f. 2000 , , 359 , 907 leonard , p. j. t. 1996 , , 470 , 521 mayor , m. et al .
1984 , , 134 , 118 moskalik , p. & poretti , e. 2003 , , 398 , 213 percival , s. m. , salaris , m. , van wyk , f. , & kilkenny , d. 2002 , , 573 , 174 pietrukowicz , p. & kaluzny , j. 2003 , acta astronomica , 53 , 371 richter , o .- g . ,
tammann , g. a. , & huchtmeier , w. k. 1987 , , 171 , 33 rubenstein , e. p. 2001
, , 121 , 3219 rucinski , s. m. 1993 , , 105 , 1433 rucinski , s. m. 1997 , , 113 , 1112 sandage , a. , diethelm , r. , & tammann , g. a. 1994 , , 283 , 111 strader , j. , everitt , h. o. , & danford , s. 2002 , , 335 , 621 strom , k. m. , strom , s. e. , rood , r. t. , & iben , i. j. 1970 , , 2 , 347 yan , l. & reid , i. n. 1996 , , 279 , 751 yan , l. & mateo , m. 1994 , , 108 , 1810 yan , l. & cohen , j. g. 1996 , , 112 , 1489 westra.e.a.m 2003 , masters thesis at rijksuniversiteit groningen wozniak , p. r. 2000 , acta astronomica , 50 , 421 zheleznyak , a. p. & kravtsov , v. v. 2003 , astronomy letters , 29 , 599 lll @xmath31 & @xmath32 & @xmath33 + & h m s & @xmath12 @xmath1 @xmath11 + 1 & 00:26:40 & @xmath771:44:14 + 2 & 00:26:40 & @xmath771:57:42 + 3 & 00:26:40 & @xmath772:10:46 + 4 & 00:26:40 & @xmath772:23:42 + 5 & 00:21:02 & @xmath772:24:20 + 6 & 00:21:02 & @xmath772:11:12 + 7 & 00:21:02 & @xmath771:58:20 + 8 & 00:21:02 & @xmath771:45:15 + llllll @xmath34 & @xmath35 & @xmath36 & @xmath37 & @xmath38 & @xmath31 + & & h m s & @xmath12 @xmath1 @xmath11 & & + v1 & rr lyr & 00:28:52.205 & @xmath772:09:31.00 & - & 3 + v2 & rr lyr & 00:28:56.555 & @xmath772:08:18.75 & - & 3 + v3 & rr lyr & 00:28:15.293 & @xmath772 06:11.23 & - & 3 + v4 & lpv & 00:26:19.372 & @xmath772:14:02.10 & oglegc254 & 3 + v5 & rr lyr & 00:26:09.379 & @xmath772:12:28.64 & oglegc255 & 3 + v6 & ecb & 00:26:10.514 & @xmath772:11:07.75 & oglegc227 & 3 + v7 & ecb & 00:26:08.836 & @xmath772:07:01.45 & oglegc228 & 3 + v8 & lpv & 00:25:38.176 & @xmath772:16:37.62 & oglegc252 & 3 + v9 & lpv & 00:25:34.105 & @xmath772:09:58.17 & oglegc222 & 3 + v10 & rr lyr & 00:25:42.014 & @xmath772:06:01.43 & oglegc223 & 3 + v11 & ecb & 00:27:33.659 & @xmath772:04:38.26 & - & 3 + v12 & ecb & 00:26:42.960 & @xmath772:15:20.71 & oglegc253 & 3 + v13 & rr lyr & 00:26:56.834 & @xmath772:10:13.19 & oglegc232 & 3 + v14 & ecb & 00:25:11.146 & @xmath772:12:14.42 & oglegc250 & 3 + v15 & lpv & 00:25:06.187 & @xmath772:10:23.48 & oglegc219 & 3 + v16 & lpv & 00:24:31.778 & @xmath772:09:29.45 & oglegc251 & 3 + v17 & ecb & 00:24:25.499 & @xmath772:08:31.60 & - & 3 + v18 & rr lyr & 00:29:20.024 & @xmath772:00:07.21 & - & 2 + v19 & rr lyr & 00:29:02.392 & @xmath771:58:57.71 & - & 2 + v20 & ecb & 00:28:55.205 & @xmath771:59:56.93 & - & 2 + v21 & ecb & 00:29:09.429 & @xmath771:51:26.42 & - & 2 + v22 & rr lyr & 00:28:31.478 & @xmath772:02:07.5 & - & 2 + v23 & rr lyr & 00:28:27.374 & @xmath771:58:32.19 & - & 2 + v24 & tyii ceph & 00:28:15.314 & @xmath771:58:21.58 & - & 2 + v25 & lpv & 00:27:56.580 & @xmath771:57:26.71 & - & 2 + v26 & ecb & 00:26:59.860 & @xmath771:55:09.77 & - & 2 + v27 & lpv & 00:26:12.380 & @xmath772:02:13.24 & oglegc230 & 2 + v28 & lpv & 00:26:02.920 & @xmath772:03:02.94 & oglegc229 & 2 + v29 & lpv & 00:25:37.513 & @xmath771:56:03.71 & - & 2 + v30 & ecb & 00:25:15.920 & @xmath771:56:06.76 & oglegc238 & 2 + v31 & lpv & 00:24:42.540 & @xmath771:59:23.92 & oglegc220 & 2 + v32 & ecb & 00:25:00.451 & @xmath772:00:02.83 & oglegc221 & 2 + v33 & lpv & 00:24:36.217 & @xmath771:57:19.26 & - & 2 + v34 & ecb & 00:24:49.296 & @xmath771:56:19.84 & - & 2 + llllll @xmath34 & @xmath35 & @xmath36 & @xmath37 & @xmath38 & @xmath31 + & & h m s & @xmath12 @xmath1 @xmath11 & & + v35 & @xmath2 scuti & 00:29:06.307 & @xmath772:28:48.03 & - & 4 + v36 & lpv & 00:28:55.595 & @xmath772:19:50.77 & - & 4 + v37 & cepheid & 00:28:24.641 & @xmath772:28:37.04 & - & 4 + v38 & rr lyr & 00:28:40.161 & @xmath772:24:03.96 & - & 4 + v39 & det.ecb & 00:28:21.873 & @xmath772:21:28.41 & - & 4 + v40 & rr lyr & 00:27:56.936 & @xmath772:21:56.20 & - & 4 + v41 & det.ecb & 00:27:45.610 & @xmath772:23:00.27 & - & 4 + v42 & lpv & 00:27:38.760 & @xmath772:21:46.11 & - & 4 + v43 & ecb & 00:27:58.730 & @xmath772:18:47.36 & - & 4 + v44 & rr lyr & 00:27:24.760 & @xmath772:23:38.47 & - & 4 + v45 & rr lyr & 00:26:56.533 & @xmath772:22:08.39 & - & 4 + v46 & lpv & 00:26:55.296 & @xmath772:21:31.04 & - & 4 + v47 & rr lyr & 00:26:07.430 & @xmath772:24:39.51 & - & 4 + v48 & rr lyr & 00:24:48.950 & @xmath772:26:42.95 & - & 4 + v49 & rr lyr & 00:24:45.564 & @xmath772:24:33.29 & - & 4 + v50 & lpv & 00:25:05.083 & @xmath772:26:58.64 & - & 4 + v51 & ecb & 00:24:58.186 & @xmath772:22:11.08 & oglegc249 & 4 + v52 & rr lyr & 00:24:31.318 & @xmath772:28:13.25 & - & 4 + v53 & lpv & 00:24:11.112 & @xmath772:27:15.40 & - & 4 + v54 & @xmath2 scuti & 00:24:01.445 & @xmath772:26:57.49 & - & 4 + v55 & lpv & 00:23:28.058 & @xmath772:26:36.56 & oglegc248 & 5 + v56 & ecb & 00:23:15.809 & @xmath772:18:53.93 & oglegc244 & 5 + v57 & rr lyr & 00:21:48.627 & @xmath772:25:21.48 & - & 5 + v58 & rr lyr & 00:21:53.866 & @xmath772:22:22.83 & - & 5 + v59 & ecb & 00:19:26.139 & @xmath772:28:52.87 & - & 5 + v60 & agb puls & 00:18:21.268 & @xmath772:26:24.44 & - & 5 + v61 & ecb & 00:22:00.581 & @xmath772:02:04.03 & oglegc214 & 7 + v62 & lpv & 00:22:11.470 & @xmath771:59:17.88 & - & 7 + v63 & lpv & 00:21:11.387 & @xmath772:00:57.68 & - & 7 + v64 & rr lyr & 00:21:06.552 & @xmath771:56:37.49 & - & 7 + v65 & rr lyr & 00:20:49.302 & @xmath772:03:09.62 & oglegc213 & 7 + v66 & rr lyr & 00:20:17.035 & @xmath771:57:26.36 & - & 7 + v67 & @xmath2 scuti & 00:19:09.879 & @xmath772:04:27.80 & - & 7 + llllll @xmath34 & @xmath35 & @xmath36 & @xmath37 & @xmath38 & @xmath31 + & & h m s & @xmath12 @xmath1 @xmath11 & & + v68 & rr lyr & 00:18:39.969 & @xmath772:03:58.19 & - & 7 + v69 & det.ecb & 00:22:52.951 & @xmath772:03:40.68 & - & 7 + v70 & rr lyr & 00:22:32.785 & @xmath771:59:30.94 & - & 7 + v71 & rr lyr & 00:22:43.840 & @xmath771:57:20.66 & oglegc234 & 7 + v72 & rr lyr & 00:23:36.984 & @xmath771:45:14.81 & - & 8 + v73 & rr lyr & 00:21:30.496 & @xmath771:46:31.43 & - & 8 + v74 & rr lyr & 00:20:55.545 & @xmath771:41:30.92 & - & 8 + v75 & ecb & 00:29:05.153 & @xmath771:43:25.90 & - & 1 + v76 & rr lyr & 00:28:50.459 & @xmath771:43:27.06 & - & 1 + v77 & rr lyr & 00:28:15.717 & @xmath771:46:11.79 & - & 1 + v78 & det.ecb & 00:28:21.065 & @xmath771:43:40.72 & - & 1 + v79 & rr lyr & 00:28:21.180 & @xmath771:40:31.50 & - & 1 + v80 & @xmath2 scuti & 00:26:42.279 & @xmath771:44:07.98 & - & 1 + v81 & rr lyr & 00:26:42.083 & @xmath771:40:24.53 & - & 1 + v82 & rr lyr & 00:25:58.078 & @xmath771:46:57.09 & - & 1 + v83 & rr lyr & 00:28:22.977 & @xmath771:28:55.44 & - & 1 + v84 & rr lyr & 00:25:11.427 & @xmath771:43:35.59 & - & 1 + v85 & det.ecb & 00:24:43.872 & @xmath771:47:38.26 & oglegc240 & 1 + v86 & rr lyr & 00:24:58.198 & @xmath771:42:13.35 & - & 1 + v87 & rr lyr & 00:28:23.004 & @xmath771:28:22.70 & - & 1 + v88 & rr lyr & 00:18:49.569 & @xmath772:09:38.80 & - & 6 + v89 & det.ecb & 00:18:54.254 & @xmath772:06:32.35 & - & 6 + v90 & rr lyr & 00:20:57.408 & @xmath772:12:04.35 & - & 6 + v91 & rr lyr & 00:21:51.437 & @xmath772:10:49.96 & - & 6 + v92 & rr lyr & 00:21:23.958 & @xmath772:15:33.79 & oglegc243 & 6 + v93 & det.ecb & 00:21:07.603 & @xmath772:08:55.80 & - & 6 + v94 & rr lyr & 00:22:33.373 & @xmath772:13:49.70 & oglegc246 & 6 + v95 & ecb & 00:22:47.469 & @xmath772:13:17.20 & oglegc245 & 6 + v96 & lpv & 00:22:40.876 & @xmath772:09:22.51 & oglegc216 & 6 + v97 & ecb & 00:23:06.015 & @xmath772:09:30.52 & - & 6 + v98 & rr lyr & 00:23:25.283 & @xmath772:19:35.85 & oglegc247 & 5 + v99 & rr lyr & 00:25:49.582 & @xmath771:58:27.35 & oglegc226 & 2 + v100 & ecb & 00:25:32.213 & @xmath772:01:51.37 & oglegc225 & 2 + llllll @xmath34 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath43 + v6 & bs & 0.3788 & 16.73 & 0.52 & 0.45 + v7 & bs & 1.1506 & 16.25 & 0.34 & 0.4 + v11 & bms & 0.4294 & 18.96 & 0.50 & 0.2 + v12 & bs & 0.4465 & 16.66 & 0.57 & 0.45 + v14 & bs & 0.3514 & 16.47 & 0.44 & 0.28 + v17 & - & 0.3005 & - & - & 0.25 + v20 & smc.bs ? & 0.8561 & 20.30 & 0.13 & 0.7 + v21 & bms & 0.2812 & 18.88 & 0.99 & 0.75 + v26 & fgrnd ? & 0.3476 & 15.99 & 0.73 & 0.09 + v30 & bms & 0.2506 & 18.56 & 0.89 & 0.15 + v32 & bms & 0.3136 & 18.14 & 0.72 & 0.6 + v34 & - & 0.2155 & - & - & 0.2 + v39 & bms & 4.6015 & 18.17 & 1.06 & 0.18 + v41 & - & 5.3648 & @xmath821 & - & 1.0 + v43 & - & 0.2602 & - & - & @xmath81 + v51 & msto & 0.3226 & 17.43 & 0.71 & 0.35 + v56 & bs & 0.3837 & 16.27 & 0.38 & 0.25 + v59 & - & 0.4618 & @xmath821 & - & @xmath81 + v61 & bms & 0.2737 & 18.02 & 0.77 & 0.4 + v69 & msto & 5.2239 & 17.08 & 0.58 & 0.65 + v75 & fgrnd ?
& 0.4922 & 18.04 & 0.04 & 0.18 + v78 & bms & @xmath44 & 19.31 & 0.88 & 0.2 + v85 & - & 2.1559 & @xmath820 & - & 0.55 + v89 & - & 1.6215 & @xmath821 & - & 0.8 + v93 & bms & @xmath44 & 19.71 & 0.93 & 0.2 + v95 & fgrnd ? &
0.2789 & 15.49 & 0.69 & 0.4 + v97 & - & 0.3973 & @xmath819 & - & 0.3 + v100 & bms & 0.2347 & 19.47 & 1.04 & 0.3 + lllll @xmath34 & @xmath40 & @xmath41 & @xmath42 & @xmath43 + v1 & 0.5052 & - & - & 0.8 + v2 & 0.3812 & 19.36 & 0.39 & 0.6 + v3 & 0.3647 & 19.36 & 0.35 & 0.7 + v5 & 0.5251 & 19.80 & 0.50 & 0.9 + v10 & 0.2971 & 17.60 & 0.33 & 0.25 + v13 & 0.3633 & 19.66 & 0.49 & 0.6 + v18 & 0.5794 & - & - & 0.6 + v19 & 0.5766 & 19.63 & 0.53 & 0.9 + v22 & 0.4244 & 19.52 & 0.37 & 0.6 + v23 & 0.7008 & 19.68 & 0.66 & 0.9 + v38 & 0.5495 & 19.60 & 0.17 & 0.9 + v40 & 0.5627 & 19.82 & 0.37 & 0.6 + v44 & 0.6190 & 19.91 & 0.64 & 0.8 + v45 & 0.6524 & - & - & 0.6 + v47 & 0.6129 & 19.88 & 0.63 & 0.7 + v48 & 0.6210 & 19.58 & 0.37 & 0.6 + v49 & 0.3727 & 19.60 & -0.13 & 0.4 + v52 & 0.5820 & 19.64 & 0.46 & 0.6 + v57 & 0.6084 & 19.66 & 0.58 & 0.7 + v58 & 1.2875 & 17.49 & 0.53 & 0.8 + v64 & 0.5963 & 19.30 & 0.84 & 0.35 + v65 & 0.6327 & - & - & 0.5 + v66 & 0.6503 & 19.74 & 0.73 & 0.4 + v68 & 0.6131 & 19.73 & 0.67 & 0.55 + v70 & 0.2653 & - & - & 0.6 + v71 & 0.6151 & 19.30 & 0.36 & 0.5 + v72 & 0.6227 & - & - & 0.55 + v73 & 0.6283 & 16.33 & 0.41 & 0.04 + v74 & 0.6291 & 19.73 & 0.49 & 0.7 + v76 & 0.5954 & 19.68 & 0.42 & 0.8 + v77 & 0.2645 & 20.00 & 0.49 & 0.7 + v79 & 0.6053 & 19.59 & 0.55 & 0.8 + v81 & 1.4301 & 17.82 & 0.49 & 0.75 + v82 & 0.6252 & 19.33 & 0.47 & 0.6 + lllll @xmath34 & @xmath40 & @xmath41 & @xmath42 & @xmath43 + v83 & 0.5465 & 19.97 & 0.51 & 1.0 + v84 & 0.5882 & 19.69 & 0.52 & 1.0 + v86 & 0.5247 & 19.05 & 0.54 & 0.5 + v87 & 0.3951 & 19.51 & 0.38 & 0.7 + v88 & 0.6313 & 19.52 & 0.46 & 0.7 + v90 & 0.6952 & 19.70 & 0.70 & 0.9 + v91 & 0.6533 & - & - & 0.3 + v92 & 0.6256 & 20.10 & 0.59 & 0.7 + v94 & 0.5721 & 19.65 & 0.80 & 0.9 + v98 & 0.5114 & 19.90 & 0.51 & 0.9 + v99 & 0.6475 & - & - & 0.25 + lllll @xmath34 & @xmath40 & @xmath41 & @xmath45 & @xmath46 + v4 & @xmath2040 & 16.51 & 1.75 & 0.2 + v8 & @xmath2050 & 17.32 & 1.90 & @xmath200.3 + v9 & @xmath2020 & 16.13 & 0.92 & @xmath200.1 + v15 & @xmath2030 & 15.36 & 1.00 & 0.19 + v16 & 3.4629 & 16.72 & 1.05 & 0.18 + v24 & 0.3871 & 18.13 & 1.82 & 0.2 + v25 & 5.8524 & 16.72 & 0.97 & 0.1 + v27 & 2.416 & 17.57 & 1.12 & 0.08 + v28 & 4.159 & 15.00 & 1.03 & 0.13 + v29 & 4.639 & 18.79 & 1.64 & 0.1 + v31 & 10.21 & 16.23 & 0.90 & 0.18 + v33 & @xmath2050 & 16.57 & 1.51 & @xmath200.2 + v35 & 0.0932 & @xmath822 & - & 1.0 + v36 & 10.7 & 19.66 & 0.91 & 0.3 + v37 & 2.5724 & 16.97 & 0.71 & 0.9 + v42 & 18.3 & 16.66 & 1.49 & 0.04 + v46 & @xmath2040 & 16.39 & 1.89 & @xmath200.15 + v50 & @xmath20100 & 16.90 & 2.34 & @xmath200.2 + v53 & @xmath20100 & 16.57 & 1.73 & @xmath200.4 + v54 & 0.0834 & @xmath821 & - & 1.0 + v55 & @xmath2060 & 16.46 & 1.76 & @xmath200.2 + v60 & 0.2544 & 17.20 & 2.99 & 0.07 + v62 & @xmath2050 & 17.56 & 2.03 & @xmath200.45 + v63 & @xmath2030 & 17.09 & 1.46 & @xmath200.08 + v67 & 0.0829 & @xmath820 & - & 0.7 + v80 & 0.2144 & @xmath822 & - & @xmath81 + v96 & 8.235 & 16.56 & 1.14 & 0.07 + | we present the results of a comprehensive search for stellar variability in the globular cluster 47 tucanae . using the mount stromlo 40-inch ( 1 m ) telescope at siding spring observatory and a combined v+r filter , we have detected 100 variable stars across a 52@xmath052@xmath1 field centered on the cluster .
the main aim of this project is to search for transiting hot jupiter planets in this cluster , the results of which shall be discussed in a separate paper . here
we present the v+r lightcurves and preliminary investigations of the detected variable stars , which comprise 28 eclipsing binaries ( 21 contact binaries and 7 detached systems ) , 45 rr lyrae stars ( 41 of which belong to the small magellanic cloud and four seemingly to the galactic halo ) , and 20 k - giant long period variables ( lpvs ) .
we also detected four @xmath2 scuti stars , one typei cepheid , and one typeii cepheid .
one variable appears to be a possible dust - enshrouded smc star with a short period pulsation .
of these 100 variables , 69 are new discoveries .
our eclipsing binary sample indicates a clear radial segregation in period , and includes two binaries that are seemingly orbited by low - luminosity stellar companions .
one rr lyrae star shows a blahzko effect with remarkable regularity .
those variables previously known are cross - identified with kaluzny and coworkers . in agreement with previous studies
, this work shows that the relative frequency of detectable variable stars ( particularly contact binaries ) in 47 tuc is very low compared to other studied regions .
a distance modulus of [email protected] for the small magellanic cloud and [email protected] for 47 tucanae has been estimated from our sample , and are in agreement with values previously published .
the total database presented here contains v and i photometry for 43,067 47 tuc field stars ( 13.0@[email protected] ) , along with 33-night v+r lightcurves and astrometry for 109,866 stars ( 14.5@[email protected] ) . |
the fermi gamma ray space telescope , hereafter _ fermi _ , represents the next generation of satellite based high - energy gamma - ray observatory .
the _ fermi _ satellite hosts two instruments : the large area telescope ( lat ) @xcite and the gamma - ray burst monitor ( gbm ) @xcite .
the former employs a pair - conversion technique to measure photons from 20 mev to energies greater than 300 gev , while the latter uses nai and bgo scintillation counters to record transient phenomena in the sky in the energy range from 8 kev to 40 mev .
the lat has no consumables , and a very stable response unlike its predecessor , the energetic gamma ray emission telescope ( egret ) @xcite .
for the energy range above 10 gev the sensitivity of the lat is at least one order of magnitude greater than that of egret , allowing the sky to be explored at these energies essentially for the first time @xcite .
the lat consists of a tracker / converter ( tkr ) for direction measurements @xcite , followed by a calorimeter ( cal ) for energy measurements @xcite .
sixteen tkr and cal modules are combined to form sixteen towers , which are assembled in a 4@xmath544 mechanical support structure .
an anticoincidence detector ( acd ) , enclosed by a micrometeoroid shield , surrounds the tkrs and rejects charged cosmic - ray background @xcite .
the lat has about one million detector readout channels . on - orbit
calibrations relate to all aspects of lat measurements and data analysis results , from absolute timing to energy and direction measurements for individual events , to fluxes and positions of gamma - ray sources .
the accurate timestamps of the lat are obtained using the global positioning system ( gps ) of the _ fermi _ spacecraft , which provides timing and position information . those are needed for phase folding pulsars and correlating gamma - ray observations with those at other wavelengths .
as the number of photons in an observation increases , the centroid of their spatial distribution becomes better measured , and eventually the error is dominated by uncertainties in the alignment of the lat , both internal and with respect to the _ fermi _ spacecraft .
source localization at gev energies enables the lat to resolve bright , adjacent sources previously labeled as unidentified @xcite and will help elucidate the origin of gamma - ray emissions from galactic cosmic rays accelerated in supernova remnants .
the energy calibrations at higher energies are of utmost importance for detection of dark matter particle signals .
some extensions to the standard model predict narrow spectral lines due to the annihilations of as - yet unknown massive particles . for detecting and characterizing these features ,
accurate energy determination is vital .
even though the lat was designed to measure gamma - rays it can also study , though not separately , the cosmic - ray electron and positron spectra . cosmic - ray electron and positron spectra and intensities may also contain signatures for new physics . in this case , energy calibrations and position determination using extrapolated tracks into the cal play an important role . at lower energies , broad features in the photon energy spectra of active galactic nuclei , or supernovae remnants originating from pion
decays and bremsstrahlung may help unravel outstanding questions concerning particle acceleration in these sources .
the purpose of this paper is to document the on - orbit calibration procedures used by the lat ; it begins with an overview of calibrations in section [ sec : overview ] .
details on trigger , acd , cal and tkr calibrations are described in sections [ sec : daq ] , [ sec : acd ] , [ sec : cal ] and [ sec : tkr ] , respectively .
the evaluation and updates to the perimeter of the south atlantic anomaly ( saa ) are presented in section [ sec : saa ] and measurements of live time are discussed in section [ sec : livetime ] .
the results from absolute timing follow in section [ sec : absolute ] .
finally , internal and spacecraft boresight alignments are explained in section [ sec : ta ] .
we conclude with a table that summarizes the calibration results in section [ sec : conclusion ] .
assessment of the current lat performance is described in a separate publication @xcite , while tests performed at particle accelerators are presented elsewhere @xcite .
the section describing each calibration in shown in the last column of table [ table : calib ] .
for this paper , the word calibration represents synchronization of trigger signals , optimization of delays for latching data , determination of detector thresholds , gains and responses , evaluation of the perimeter of the south atlantic anomaly ( saa ) , measurements of live time and of absolute time and internal and spacecraft boresight alignments .
table [ table : calib ] summarizes all calibration types , classified by category .
configuration refers to operational settings that define the state of the hardware , which are fixed before data are acquired .
calibrations are related to quantities that can change after data are processed and analyzed .
as shown in table [ table : calib ] , for some types of calibrations , updates are not as frequent as the acquisition of the relevant data .
.list of lat calibrations ( calib ) and configurations ( config ) that can impact lat scientific results .
note that pedestals are used both as a configuration on - board and as a calibration on the ground .
detailed descriptions are given in the sections listed in the last column .
frequency of updates correspond to current best estimates for the period beyond the first year of operations . [ cols="^,^,^,^,^,^,^ " , ]
the _ fermi _ lat collaboration acknowledges the generous ongoing support of a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy and the istituto nazionale di astrofisica , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation and the swedish national space board in sweden . additional support for science analysis during the operations phase from the following agencies is also gratefully acknowledged : the istituto nazionale di astrofisica in italy and the k. a. wallenberg foundation in sweden for providing a grant in support of a royal swedish academy of sciences research fellowship for jc . 99 , in press _
astrophysical journal_. c. meegan _ et al _ , in preparation .
et al . _ ,
vol * 86 * ( 1993 ) 629 .
w. b. atwood _ et al .
_ , astropart .
phys . * 28 * , ( 2007 ) 422 .
t. ohsugi _
et al _ , nucl .
inst . and meth . * a 541 * ( 2005 ) 29 .
r. p. johnson , p. poplevin , h. f .- w .
sadrozinski , and e. n. spencer , ieee trans .
sci . * 45 * , ( 1998 ) 927 .
l. baldini _ et al .
_ , ieee trans .
sci . * 53 * , ( 2006 ) 466 .
j. e. grove _ et al .
_ , in preparation .
a. moiseev _
_ , astropart . phys .
* 27 * , ( 2007 ) 339 .
a. moiseev _
et al . _ ,
inst . and meth . * a 583 * , ( 2007 ) 372
. a. abdo _
et al . _ , in preparation .
w. atwood _
et al . _ ,
inst . and meth . * a 446 * ( 2000 ) 444 .
e. do couto e silva _ et al .
inst . and meth .
* a 474 * , ( 2001 ) 19 .
et al . _ , in preparation .
et al . _ , in preparation .
a. moiseev _
et al . _ ,
* 22 * , ( 2004 ) 275 .
et al . _ ,
inst . and meth .
* 560 * , ( 2006 ) 395 .
d. m. sawyer and j. i. vette , _
ap-8 trapped proton environment for solar maximum and solar minimum _ , nssdc / wdc - a - r&s 76 - 06 , 1976 .
d. heynderickx , m. kruglanski , v. pierrard , j. lemaire , m. d. looper , and j. b. blake , ieee trans .
, * 46 * , ( 1999 ) 1475 .
igrf-10 , iaga division v - mod geomagnetic field modeling , http://www.ngdc.noaa.gov/iaga/vmod/igrf.html m. casolino , et al .
_ observations of primary , trapped and quasi - trapped particles with the pamela experiment_. to appear in the proceedings of the 30@xmath55 international cosmic ray conference ( icrc 2007 ) 3 - 11 jul 2007 , merida , yucatan , mexico .
et al _ submitted to _ science _
d. a. smith _ et al .
_ a&a * 492 * ( 2008 ) 923 .
m. de naurois , j. holder _ et al .
_ apj * 566 * , ( 2002 ) 343 .
et al . _ , in press _ astrophysical journal _ ( 2008 ) . global postioning system standard positioning service performance standard , 4th edition ( september 2008 ) , http://pnt.gov/public/docs/2008-spsps.pdf , p. a-16 , section a.4.8 a. abdo _
et al . _ ,
submitted to apj .
et al . _ , in preparation .
et al . _ , in preparation . | the large area telescope ( lat ) on board the _ fermi _ gamma ray space telescope began its on orbit operations on june 23 , 2008 .
calibrations , defined in a generic sense , correspond to synchronization of trigger signals , optimization of delays for latching data , determination of detector thresholds , gains and responses , evaluation of the perimeter of the south atlantic anomaly ( saa ) , measurements of live time , of absolute time , and internal and spacecraft boresight alignments . here
we describe on orbit calibration results obtained using known astrophysical sources , galactic cosmic rays , and charge injection into the front - end electronics of each detector .
instrument response functions will be described in a separate publication .
this paper demonstrates the stability of calibrations and describes minor changes observed since launch .
these results have been used to calibrate the lat datasets to be publicly released in august 2009 . *
the on orbit calibration of the _ fermi _ large area telescope * _ the fermi lat collaboration _ a. a. abdo@xmath0 , m. ackermann@xmath1 , m. ajello@xmath1 , j. ampe@xmath2 , b. anderson@xmath3 , w. b. atwood@xmath3 , m. axelsson@xmath4 , r. bagagli@xmath5 , l. baldini@xmath5 , j. ballet@xmath6 , g. barbiellini@xmath7 , j. bartelt@xmath1 , d. bastieri@xmath8 , b. m. baughman@xmath9 , k. bechtol@xmath1 , d. bdrde@xmath10 , f. bellardi@xmath5 , r. bellazzini@xmath5 , f. belli@xmath11 , b. berenji@xmath1 , d. bisello@xmath8 , e. bissaldi@xmath12 , e. d. bloom@xmath1 , g. bogaert@xmath13 , j. r. bogart@xmath1 , e. bonamente@xmath14 , a. w. borgland@xmath1 , p. bourgeois@xmath10 , a. bouvier@xmath1 , j. bregeon@xmath5 , a. brez@xmath5 , m. brigida@xmath15 , p. bruel@xmath13 , t. h. burnett@xmath16 , g. busetto@xmath8 , g. a. caliandro@xmath15 , r. a. cameron@xmath1 , m. campell@xmath1 , p. a. caraveo@xmath17 , s. carius@xmath18 , p. carlson@xmath19 , j. m. casandjian@xmath6 , e. cavazzuti@xmath20 , m. ceccanti@xmath5 , c. cecchi@xmath14 , e. charles@xmath1 , a. chekhtman@xmath21 , c. c. cheung@xmath22 , j. chiang@xmath1 , r. chipaux@xmath23 , a. n. cillis@xmath22 , s. ciprini@xmath14 , r. claus@xmath1 , j. cohen - tanugi@xmath24 , s. condamoor@xmath1 , j. conrad@xmath25 , r. corbet@xmath22 , s. cutini@xmath20 , d. s. davis@xmath26 , m. deklotz@xmath27 , c. d. dermer@xmath2 , a. de angelis@xmath28 , f. de palma@xmath15 , s. w. digel@xmath1 , p. dizon@xmath29 , m. dormody@xmath3 , e. do couto e silva@xmath1 , p. s. drell@xmath1 , r. dubois@xmath1 , d. dumora@xmath30 , y. edmonds@xmath1 , d. fabiani@xmath5 , c. farnier@xmath24 , c. favuzzi@xmath15 , e. c. ferrara@xmath22 , o. ferreira@xmath13 , z. fewtrell@xmath2 , d. l. flath@xmath1 , p. fleury@xmath13 , w. b. focke@xmath1 , k. fouts@xmath1 , m. frailis@xmath28 , d. freytag@xmath1 , y. fukazawa@xmath31 , s. funk@xmath1 , p. fusco@xmath15 , f. gargano@xmath32 , d. gasparrini@xmath20 , n. gehrels@xmath33 , s. germani@xmath14 , b. giebels@xmath13 , n. giglietto@xmath15 , f. giordano@xmath15 , t. glanzman@xmath1 , g. godfrey@xmath1 , j. goodman@xmath1 , i. a. grenier@xmath6 , m .- h . grondin@xmath30 , j. e. grove@xmath2 , l. guillemot@xmath30 , s. guiriec@xmath24 , m. hakimi@xmath1 , g. haller@xmath1 , y. hanabata@xmath31 , p. a. hart@xmath1 , p. hascall@xmath34 , e. hays@xmath22 , m. huffer@xmath1 , r. e. hughes@xmath9 , g. jhannesson@xmath1 , a. s. johnson@xmath1 , r. p. johnson@xmath3 , t. j. johnson@xmath33 , w. n. johnson@xmath2 , t. kamae@xmath1 , h. katagiri@xmath31 , j. kataoka@xmath35 , a. kavelaars@xmath1 , h. kelly@xmath1 , m. kerr@xmath16 , w. klamra@xmath19 , j. kndlseder@xmath36 , m. l. kocian@xmath1 , f. kuehn@xmath9 , m. kuss@xmath5 , l. latronico@xmath5 , c. lavalley@xmath24 , b. leas@xmath2 , b. lee@xmath34 , s .- h .
lee@xmath1 , m. lemoine - goumard@xmath30 , f. longo@xmath7 , f. loparco@xmath15 , b. lott@xmath30 , m. n. lovellette@xmath2 , p. lubrano@xmath14 , d. k. lung@xmath34 , g. m. madejski@xmath1 , a. makeev@xmath21 , b. marangelli@xmath15 , m. marchetti@xmath11 , m. m. massai@xmath5 , d. may@xmath2 , g. mazzenga@xmath11 , m. n. mazziotta@xmath32 , j. e. mcenery@xmath22 , s. mcglynn@xmath19 , c. meurer@xmath37 , p. f. michelson@xmath1 , m. minuti@xmath5 , n. mirizzi@xmath15 , p. mitra@xmath1 , w. mitthumsiri@xmath1 , t. mizuno@xmath31 , a. a. moiseev@xmath38 , m. mongelli@xmath32 , c. monte@xmath15 , m. e. monzani@xmath1 , e. moretti@xmath7 , a. morselli@xmath39 , i. v. moskalenko@xmath1 , s. murgia@xmath1 , d. nelson@xmath1 , l. nilsson@xmath40 , s. nishino@xmath31 , p. l. nolan@xmath1 , e. nuss@xmath24 , m. ohno@xmath41 , t. ohsugi@xmath31 , n. omodei@xmath5 , e. orlando@xmath12 , j. f. ormes@xmath42 , m. ozaki@xmath41 , a. paccagnella@xmath43 , d. paneque@xmath1 , j. h. panetta@xmath1 , d. parent@xmath30 , v. pelassa@xmath24 , m. pepe@xmath14 , m. pesce - rollins@xmath5 , p. picozza@xmath11 , m. pinchera@xmath5 , f. piron@xmath24 , t. a. porter@xmath3 , s. rain@xmath15 , r. rando@xmath8 , e. rapposelli@xmath5 , w. raynor@xmath2 , m. razzano@xmath5 , a. reimer@xmath1 , o. reimer@xmath1 , t. reposeur@xmath30 , l. c. reyes@xmath44 , s. ritz@xmath33 , s. robinson@xmath45 , l. s. rochester@xmath1 , a. y. rodriguez@xmath46 , r. w. romani@xmath1 , m. roth@xmath16 , f. ryde@xmath19 , a. sacchetti@xmath32 , h. f .- w .
sadrozinski@xmath3 , n. saggini@xmath5 , d. sanchez@xmath13 , l. sapozhnikov@xmath1 , o. h. saxton@xmath1 , p. m. saz parkinson@xmath3 , a. sellerholm@xmath37 , c. sgr@xmath5 , e. j. siskind@xmath47 , d. a. smith@xmath30 , p. d. smith@xmath9 , g. spandre@xmath5 , p. spinelli@xmath15 , j .- l .
starck@xmath6 , t. e. stephens@xmath22 , m. s. strickman@xmath2 , a. w. strong@xmath12 , m. sugizaki@xmath1 , d. j. suson@xmath48 , h. tajima@xmath1 , h. takahashi@xmath31 , t. takahashi@xmath41 , t. tanaka@xmath1 , a. tenze@xmath5 , j. b. thayer@xmath1 , j. g. thayer@xmath1 , d. j. thompson@xmath22 , l. tibaldo@xmath8 , o. tibolla@xmath49 , d. f. torres@xmath50 , g. tosti@xmath14 , a. tramacere@xmath51 , m. turri@xmath1 , t. l. usher@xmath1 , n. vilchez@xmath36 , n. virmani@xmath29 , v. vitale@xmath11 , l. l. wai@xmath52 , a. p. waite@xmath1 , p. wang@xmath1 , b. l. winer@xmath9 , d. l. wood@xmath2 , k. s. wood@xmath2 , h. yasuda@xmath31 , t. ylinen@xmath53 , m. ziegler@xmath3 1 .
national research council research associate 2 .
space science division , naval research laboratory , washington , dc 20375 3 .
w. w. hansen experimental physics laboratory , kavli institute for particle astrophysics and cosmology , department of physics and slac national laboratory , stanford university , stanford , ca 94305 4 .
santa cruz institute for particle physics , department of physics and department of astronomy and astrophysics , university of california at santa cruz , santa cruz , ca 95064 5 .
the oskar klein centre for cosmo particle physics , albanova , se-106 91 stockholm , sweden 6 .
department of astronomy , stockholm university , se-106 91 stockholm , sweden 7 .
istituto nazionale di fisica nucleare , sezione di pisa , i-56127 pisa , italy 8 .
laboratoire aim , cea - irfu / cnrs / universit paris diderot , service dastrophysique , cea saclay , 91191 gif sur yvette , france 9 .
istituto nazionale di fisica nucleare , sezione di trieste , i-34127 trieste , italy 10 .
dipartimento di fisica , universit di trieste , i-34127 trieste , italy 11 .
istituto nazionale di fisica nucleare , sezione di padova , i-35131 padova , italy 12 .
dipartimento di fisica g. galilei " , universit di padova , i-35131 padova , italy 13 .
department of physics , center for cosmology and astro - particle physics , the ohio state university , columbus , oh 43210 14 .
irfu / dir , cea saclay , 91191 gif sur yvette , france 15 .
istituto nazionale di fisica nucleare , sezione di roma tor vergata " , i-00133 roma , italy 16 .
dipartimento di fisica , universit di roma tor vergata " , i-00133 roma , italy 17 .
max - planck institut fr extraterrestrische physik , 85748 garching , germany 18 .
laboratoire leprince - ringuet , cole polytechnique , cnrs / in2p3 , palaiseau , france 19 .
istituto nazionale di fisica nucleare , sezione di perugia , i-06123 perugia , italy 20 .
dipartimento di fisica , universit degli studi di perugia , i-06123 perugia , italy 21 .
dipartimento di fisica m. merlin " delluniversit e del politecnico di bari , i-70126 bari , italy 22 .
istituto nazionale di fisica nucleare , sezione di bari , 70126 bari , italy 23 .
department of physics , university of washington , seattle , wa 98195 - 1560 24 .
inaf - istituto di astrofisica spaziale e fisica cosmica , i-20133 milano , italy 25 .
school of pure and applied natural sciences , university of kalmar , se-391 82 kalmar , sweden 26 .
department of physics , royal institute of technology ( kth ) , albanova , se-106 91 stockholm , sweden 27 .
agenzia spaziale italiana ( asi ) science data center , i-00044 frascati ( roma ) , italy 28 .
george mason university , fairfax , va 22030 29 .
nasa goddard space flight center , greenbelt , md 20771 30 .
irfu / sedi , cea saclay , 91191 gif sur yvette , france 31 .
laboratoire de physique thorique et astroparticules , universit montpellier 2 , cnrs / in2p3 , montpellier , france 32 .
department of physics , stockholm university , albanova , se-106 91 stockholm , sweden 33 .
university of maryland , baltimore county , baltimore , md 21250 34 .
stellar solutions inc .
, 250 cambridge avenue , suite 204 , palo alto , ca 94306 35 . dipartimento di fisica , universit di udine and istituto nazionale di fisica nucleare , sezione di trieste , gruppo collegato di udine , i-33100 udine , italy 36 .
atk space products , beltsville , md 20705 37 .
cnrs / in2p3 , centre dtudes nuclaires bordeaux gradignan , umr 5797 , gradignan , 33175 , france 38 .
universit de bordeaux , centre dtudes nuclaires bordeaux gradignan , umr 5797 , gradignan , 33175 , france 39 .
department of physical science and hiroshima astrophysical science center , hiroshima university , higashi - hiroshima 739 - 8526 , japan 40 .
university of maryland , college park , md 20742 41 .
orbital network engineering , 10670 north tantau avenue , cupertino , ca 95014 42 .
department of physics , tokyo institute of technology , meguro city , tokyo 152 - 8551 , japan 43 .
centre dtude spatiale des rayonnements , cnrs / ups , bp 44346 , f-30128 toulouse cedex 4 , france 44 . center for research and exploration in space science and technology ( cresst ) , nasa goddard space flight center , greenbelt , md 20771 45 .
mtfakta i kalmar ab , 30477 kalmar , sweden 46 .
institute of space and astronautical science , jaxa , 3 - 1 - 1 yoshinodai , sagamihara , kanagawa 229 - 8510 , japan 47 .
department of physics and astronomy , university of denver , denver , co 80208 48 .
dipartimento di ingegneria dellinformazione , universit di padova , i-35131 padova , italy 49 .
kavli institute for cosmological physics , university of chicago , chicago , il 60637 50 .
current address : pacific northwest national laboratory , richland , wa 99352 51 .
institut de ciencies de lespai ( ieec - csic ) , campus uab , 08193 barcelona , spain 52 .
nycb real - time computing inc . ,
lattingtown , ny 11560 - 1025 53 .
department of chemistry and physics , purdue university calumet , hammond , in 46323 - 2094 54 .
max - planck - institut fr kernphysik , d-69029 heidelberg , germany 55 .
instituci catalana de recerca
i estudis avanats ( icrea ) , barcelona , spain 56 .
consorzio interuniversitario per la fisica spaziale ( cifs ) , i-10133 torino , italy 57 .
current address : yahoo ! inc . ,
sunnyvale , ca 94089 * keywords : * glast , fermi , lat , gamma - ray , calibrations * pacs classification codes : * 07.87.+v ; 95.55.ka |
in their famous paper @xcite einstein , podolsky and rosen ( epr ) suggested a gedankenexperiment which , as they believed , must prove the incompleteness of quantum mechanics .
an interesting analysis of this problem was given by bohr @xcite .
some progress was achieved by bell @xcite .
he showed that under assumption of the epr arguments some inequalities must be fulfilled .
if bohr s arguments are correct then these inequalities can be violated .
it was only in @xmath0 when the bohr s arguments were experimentally verified @xcite .
now the arguments by epr are considered to be incorrect , but nevertheless it is the work @xcite that initiated the discussion of basics of quantum mechanics . in this work
i analyze the connection between the boolean functions theory and bell inequalities for multi - qubit systems ( which were obtained in @xcite ) .
surprisingly enough , in many aspects bell inequalities theory is analogous to the combinatorial problems of computer logic circuits design developed a year earlier then the bell s work @xcite appeared .
there is also a relation between bell inequalities and applications of boolean functions theory to classical cryptography .
for example , classification of bell inequalities discussed in @xcite is closely connected to the jevons group , studied in @xcite , and the maximal quantum violation of a given bell inequality is connected to the nonlinearity of the corresponding boolean function .
i made an attempt to clarify these connections .
but there are still many open questions , both combinatorial ( like classification of bell inequalities with respect to the group @xmath1 ) and analytical ( like calculation of the maximal quantum violation @xmath2 ) .
another interesting problem is the connection between the maximal quantum violation and the uncertainty relation for boolean functions . in this work
it is shown that the bell inequalities whose maximal quantum violation is the largest ( mermin inequalities ) , minimize this uncertainty relation .
all mathematical constructions discussed in the text are illustrated with _
mathematica_. i choose _ mathematica_since it has an extremely flexible and unified programming language and a very rich set of built - in mathematical functions . due to this
it is possible to present the algorithms illustrating the discussed quantities in a very compact form . using _
, it is possible to code all the illustrating examples using high - level constructions , avoiding worrying about low - level programming details which have nothing to do with the problem under study .
i do not pretend to give the most effective _
mathematica_code for calculating different features of boolean functions , my goal is to present a compact and ready - to - use working code .
all the examples can be typed in and run provided that they are entered to _
mathematica_in the given order .
i also presented the _ c _ source code for the fast walsh - hadamard transform and a way to turn it into an executable program which can be used in _
the bell inequalities for a multi - qubit system , which were obtained in @xcite , are closely connected with boolean functions theory .
there is a natural one - to - one correspondence between the set of all @xmath3 boolean functions of @xmath4 boolean variables and the set of bell inequalities for @xmath4-qubits . in this section
i give a short overview of the notions and results of the boolean functions theory which are needed for applications to bell inequalities .
the most important notion discussed in this section is the walsh - hadamar transform , which turns out to be the coefficients of the bell inequality corresponding to a given boolean function . here
i also introduce a visualization technique of boolean functions which is useful to graphically represent different classes of boolean functions .
this approach is based on the fact that for all @xmath4 the number of boolean functions of @xmath4 variables is a square ( in fact , @xmath5 ) , so that one can associate the boolean functions with the cells of a @xmath6 array . briefly speaking ,
this is done numerating boolean functions with integers , interpreting the vectors of their values as binary decompositions and then using division modulo @xmath7 . in this section
i visualize boolean functions with respect to their degree and uncertainty . in both cases
the pictures show some kind of fractal behavior .
let @xmath8 be the finite field with two elements @xmath9 . the sum and the product of elements @xmath10 are denoted as @xmath11 and @xmath12 respectively .
the product of @xmath4 copies of @xmath13 we denote as @xmath14 and refer to its elements as ( @xmath4-dimensional ) boolean vectors .
it is clear that @xmath15 .
the notation @xmath16 is used for the scalar product of two boolean vectors @xmath17 and @xmath18 : @xmath19 there is a natural one - to - one correspondence @xmath20 between the set @xmath21 and the set @xmath22 : @xmath23 in other words , the boolean vector @xmath24 corresponds to the integer @xmath25 whose binary representation is given by @xmath24 .
the most significant bit in the binary decomposition of @xmath25 is the first component of @xmath24 and the least significant bit is the last component .
the correspondence @xmath20 is natural in the following sense . for an integer
@xmath26 the set @xmath27 can be identified with a subset of @xmath21 by padding @xmath28-dimensional boolean vectors on the left to extend them to @xmath4 dimensions .
then the diagram @xmath29 is commutative .
this means that one can apply @xmath20 to @xmath28-dimensional boolean vectors with @xmath26 .
the correspondence @xmath20 is illustrated by table [ tbl : b ] .
.the correspondence @xmath20 between boolean vectors @xmath30 and integer numbers @xmath31 . [ cols="^,^,^,^,^",options="header " , ] the cross - product is defined as follows . for powers of variables
we have @xmath32 where @xmath33 and @xmath34 are the greatest common divisor and the least common multiple of integers @xmath35 and @xmath36 respectively . for two monomials
the cross - product is defined via @xmath37 and then extended for arbitrary polynomials by bilinearity .
for example , let us calculate the cross - product @xmath38 : @xmath39 the cycle index @xmath40 for small @xmath4 is shown in the second column of the table [ tbl:1 ] .
now let us add an action on the set @xmath41 .
let @xmath42 be the cyclic group of the second order .
the action @xmath43 we define as : @xmath44 and @xmath45 being the logical _ not _ , @xmath46 and @xmath47 . for cycle lengthes
we have @xmath48 according to for the number of equivalence classes we have the following expression ( see also @xcite ) : @xmath49 these numbers are shown in the third column of the table [ tbl:1 ] .
the last column of the table shows the number @xmath50 of equivalence classes with respect to the equivalence under study . the number @xmath51 was taken from _
larsed / boolean/_. the numbers @xmath50 for @xmath52 are unknown .
in conclusion , the relation between the boolean functions theory and the general bell inequalities for @xmath4-qubits is established .
the classification of bell inequalities with respect to the jevons group is obtained , which is a weaker result then the problem posed in @xcite .
nevertheless , to my knowledge it is the only approach to the more general classification .
this approach is based on the works @xcite done for computer logic circuits theory , which shows the connection between quite different problems qubit system description and computer logic circuit design .
there are still many unsolved problems .
two the most important ones are : 1 .
classification of bell inequalities ( or boolean functions ) with respect to the group @xmath1 , 2 .
characterization of the maximal quantum violation @xmath2 of a given boolean function @xmath53 in terms of properties of @xmath53 , in particular , finding the relation between @xmath2 and the nonlinearity @xmath54 . and , of course , a very interesting question how the ideas from different applications of the boolean functions theory ( not only to cryptography or computer logic circuits design ) can be used in quantum information theory .
in this appendix i present the _ mathematica_code figs . [
fig : o ] , [ fig : u ] , [ fig : v ] , [ fig : bell ] and table [ tbl:1 ] were obtained with .
first of all , the set @xmath21 can be coded as @xmath55:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{v[n\ _ ] : = tuples[\{0 , 1\ } , n]}\ ] ] the _
mathematica_function @xmath56}$ ] gives the list of the base-@xmath57 digits of @xmath4 , padding it on the left if necessary to give a list of length @xmath58 . using this function
, one can code @xmath21 in another way as @xmath59 : = integerdigits[\ # , 2 , n]\ & /@
range[0 , 2^n-1 ] } \nonumber\ ] ] this code works a few times more slowly , but this approach can be useful if it necessary to construct only some part of @xmath21 , not the whole set @xmath21 . we know that @xmath60 , but since @xmath61 , the simple definition @xmath62 : = \ v[2^n]}$ ] will not work for @xmath63 .
a general boolean function can be coded as follows : @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{f[c\ _ , s\ _ ] : = function[x , mod[c.(apply[times , \#]\ & /@\
( part[x , \#]\ & /@\
s ) ) , 2]]}\ ] ] where @xmath65 is the list of all the multi - indices @xmath66 for which the coefficients @xmath67 and @xmath68 the list of these coefficients . note that the @xmath69-es of @xmath68 and @xmath65 must be the same . for example , the expression @xmath70 , \{m\}]]}\ ] ] gives a homogeneous polynomial of the degree @xmath28 ( except the case when all elements of @xmath68 are zero ; when all elements of the list @xmath68 are @xmath71 then is the @xmath28-th symmetric polynomial : @xmath72\ } ] , subsets[range[n ] , \{m\}]][\{x_1 , \ldots , x_n\ } ] = s_m(x_1 , \ldots , x_n)}\ ] ] similarly , the expression @xmath73 , m]]}\ ] ] is a polynomial of the degree not greater then @xmath28 .
the walsch - hadamard transform can be coded as @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{wht : = \ function[u , plus\ @@\
( function[x , ( -1)^{\#[x]+(x.u ) } ] /@\
v[length[u]])]\&}\ ] ] this method of calculation the walsch - hadamard transform is very simple and quite ineffective . in appendix
b a much better method is presented .
the only disadvantage of that method is the fact that it works much faster only when it is necessary to calculate all the numbers @xmath74 , @xmath75 simultaneously , and it can not be applied to calculate only one number @xmath74 for a given @xmath76 .
@xmath77 is a functional : @xmath78}$ ] is a function and @xmath78[\{u_1 , \ldots , u_n\}]}$ ] is its value @xmath74 at @xmath79 .
the inverse walsch - hadamard transform can be coded as @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{whit : = \ function[x , ( 1 - 2^{-length[x ] } plus\ @@\
( function[u , ( -1)^{x.u}\ \#[u ] ] /@\
v[length[x]]))/2 ] \&}\ ] ] like @xmath77 , it is also a functional .
the autocorrelation of two boolean functions can be coded as @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{\delta : = \ function[u , plus\ @@\ ( function[x , ( -1)^{\#1[x ] + \#2[mod[\ # , 2]\ & /@\ ( x+u ) ] } ] /@\
v[length[u ] ] ) ] \&}\ ] ] it is a functional of two arguments : the expression @xmath80[\{u_1 , \ldots , u_n\}]}$ ] is the value @xmath81 at @xmath79 . the uncertainty can be coded as @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{u[n\ _ ] : = \ function[f , 2^{-n } count[wht[f ] /@\
v[n ] , \_?(\ # ! = 0 \ & ) ] count[\delta[f , f ] /@\ v[n ] , \_?(\ # ! = 0 \&)]]}\ ] ] for a function @xmath82 the expression @xmath83[f]}$ ] gives the uncertainty @xmath84 of @xmath53 .
to calculate @xmath84 for all @xmath82 we need a way to define @xmath53 given its number @xmath85 , @xmath86 .
the following code solves this problem : @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{itof[n\ _ , b\ _ ] : = \ part[integerdigits[b,2 , 2^n ] , 2^n - fromdigits[\ # , 2 ] ] \&}\ ] ] given @xmath87 , the expression @xmath88}$ ] is the corresponding boolean function , to which one can apply the syntaxis @xmath88[\{x_1 , \ldots , x_n\}]}$ ] .
we can visualize boolean functions with respect to their uncertainty using the code @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{ut[n\ _ ] : = \ partition[table[u[n][itof[n , b ] ] , \{b , 0 , 2^{2^n}-1\ } ] , 2^{2^{n-1}}]}\ ] ] in _
mathematica_this table can be immediately plotted with the @xmath89 function , what was doe for the case of @xmath90 , for the other two cases i used a simple script to generate _ pstricks _ code from this table and then compiled it with latex(_pstricks _ code produced huge pictures in the case of @xmath90 ) . in this way fig .
[ fig : u ] was obtained . the code @xmath91 //flatten //sort //split
//length}\ ] ] gives the number of different values of the uncertainty .
for @xmath92 it is @xmath71 ( all @xmath93 boolean functions have the same uncertainty @xmath71 ) , for @xmath94 it is @xmath95 ( the values are @xmath71 and @xmath96 ) and for @xmath90 it is @xmath97 ( the values are @xmath71 , @xmath98 , @xmath96 and @xmath93 ) .
unfortunately , these are all the values of @xmath4 for which this simple code works , for larger @xmath4 another technique is needed .
now i will show how the fig . [
fig : o ] was obtained .
the key point is the function @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{d[n\ _ , m\ _ ] : = } \ & \mathtt{function[c , fromdigits[f[c , subsets[range[n ] , m ] ] /@\
v[n ] , 2 ] ] } \\ & \mathtt{/@\ v[plus\ @@\ ( binomial[n , \ # ] \ & /@\ range[0 , m ] ) ] } \end{split}\ ] ] which returns the numbers @xmath85 of the boolean functions @xmath82 of degree @xmath100 . to visualize boolean functions with respect to their degree
let us create a list @xmath101};$ ]
. then we can fill it in as @xmath102 + 1 ] ] = n - i]}\ ] ] the table @xmath103}$ ] gives the desired visualization .
now let us discuss the equivalence of bell inequalities .
the map @xmath104 can be coded as @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{p[pi\_]:=\ function[x , \#[permute[x , inversepermutation[pi]]]]\&}\ ] ] the expression @xmath105}$ ] is a functional : @xmath105[f]}$ ] gives the function @xmath106 , and @xmath105[f][\{x_1 , \ldots , x_n\}]}$ ] gives its value @xmath107 at @xmath108 .
the maps @xmath109 , @xmath110 and @xmath111 can be coded as @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{\delta : = \ function[x , 1-\#[x]]\ & } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{s[y\ _ ] : = \ function[x , \#[function[z , mod[z , 2 ] ] /@\
( x + y)]]\ & } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{t[z\ _ ] : = \ function[x , mod[\#[x ] + x.z , 2]]\ & } \end{split}\ ] ]
let us illustrate the relations between the maps under study .
as an example , consider the first relation .
i show that this relation is valid by applying the maps from both sides to all boolean functions @xmath82 and comparing the results .
to do it , we need operations which produce the lists of values @xmath112 and @xmath113 , @xmath30 given the list @xmath114 , @xmath30 of values of @xmath82 . the code @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{ps[n\ _ , pi\ _ , y\_]}\ & \mathtt{:= composition[p[pi ] , s[y]][function[x , part[\ # , fromdigits[x,2]+1 ] ] ] /@\
v[n ] \ & } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{sp[n\ _ , pi\ _ , y\_]}\ & \mathtt{:= composition[s[y ] , p[pi]][function[x , part[\ # , fromdigits[x,2]+1 ] ] ] /@\ v[n ] \ & } \end{split}\ ] ] is a solution to this problem .
the expression @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{ps[3 , \{2,3,1\ } , \{1,1,0\ } ] /@\
v[2 ^ 3 ] - sp[3 , \{2,3,1\ } , \{1,0,1\ } ] /@\
v[2 ^ 3 ] //short}\ ] ] produces a list of zero - lists , as expected .
this example clearly demonstrates that the first relation is valid ( in the case of @xmath94 ) .
@xmath115 given a number @xmath87 the function @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{e[n\ _ , b\_]}&\ \mathtt{:=\ replace[fromdigits[map[\ # , v[n ] ] , 2 ] \ & } \\ & \mathtt{/@\ function[f , apply[function[\{\varepsilon , y , pi , z\ } , composition[\varepsilon , t[z ] , p[pi ] , s[y ] ] } \\ & \mathtt{[function[x , part[integerdigits[f , 2 , 2^n ] , fromdigits[x , 2 ] + 1 ] ] ] ] , \#]\ & } \\ & \mathtt{/@\ tuples[\{\{identity , \delta\ } , v[n ] , permutations[range[n ] ] , v[n]\}]][b ] } \\ & \mathtt{//sort //split , \{(i\ _ ) .. \ } \to i , 1 ] } \end{split}\ ] ] produces the list of numbers which correspond to the functions , equivalent to the one corresponding to @xmath116 . with this functions it is easy to get the equivalence classes .
let us illustrate the general idea by the case of @xmath94 .
we start with @xmath117 : @xmath118 //short}$ ] produces @xmath119 it is the first class of equivalence ( which contains @xmath93 elements ) .
the smallest number which is not in this class is @xmath71 ; @xmath120 //short}$ ] produces @xmath121 it is the second class of equivalence ( which contains @xmath122 elements ) .
the smallest number which is not in either class found so far is @xmath123 ; @xmath124 //short}$ ] produces @xmath125 it is the third class of equivalence ( which contains @xmath126 elements ) .
the ext number to try is @xmath127 : @xmath128 //short}$ ] produces @xmath129 it is the fourth class of equivalence ( which also contains @xmath126 elements ) .
the next number is @xmath130 ; @xmath131 //short}$ ] produces @xmath132 it is the fifth class of equivalence ( which contains @xmath93 elements ) . since @xmath133 ,
we exhausted all numbers ( boolean functions ) which means that in the case of @xmath94 there are @xmath134 classes of equivalence , they are shown in fig .
[ sfig : bell - b ] .
the same approach allows one to get fig .
[ sfig : bell - b2 ] .
the maximal quantum violations can be found as follows .
let us introduce the functions @xmath99:=
\quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{s[\varphi\_]\ : = \ \{cos[\varphi ] , i\ sin[\varphi]\ } } \\
{ { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{s[\varphi\_list]\ : = \ apply[times , \#]\ & } \\ &
\mathtt{/@\ tuples[s /@\ \varphi ] } \end{split}\ ] ] the maximal violation can now be coded as @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{v[f\ _ , \varphi\_]\ : = \ abs[((-1)^{f[\#]}\ & } \\ & \mathtt{/@\ v[length[\varphi]]).s[\varphi ] ] } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{v2[b\ _ , \varphi\_]\ : = \ v[itof[length[\varphi ] , b ] , \varphi ] } \end{split}\ ] ] the quantity @xmath135}$ ] give the maximal quantum violation @xmath2 of the function @xmath53 corresponding to the number @xmath116 .
let us illustrate the calculation of maximal quantum violation for mermin inequalities .
the mermin inequalities ( whose coefficients are given by and ) can be coded as ( using the standard package _ algebrasymmetricpolynomials _ ) @xmath136\ : = \ mod[symmetricpolynomial[x , 2 ] , 2]}\ ] ] the maximal quantum violation of mermin inequalities can be found as @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{findmaximum[v[m , \{\varphi_1 , \ldots , \varphi_n\ } ] , } \\ & \mathtt{\{\varphi_1 , \varphi^0_1\ } , \ldots , \{\varphi_n , \varphi^0_n\ } ] } \end{split}\ ] ] for example , the code @xmath137 , \{\varphi_1 , 1\ } , \{\varphi_2 , 1\}]}\ ] ] produces @xmath138 the code @xmath139 , } \\ & \mathtt{\{\varphi_1 , 1\ } , \{\varphi_2 , 1\ } , \{\varphi_3 , 1\ } ] } \end{split}\ ] ] produces @xmath140 in full agreement with the relation . since @xmath2 is the same for equivalent functions , using the classes of equivalence calculated before , it is easy to get fig .
[ fig : v ] .
now i will show how the table [ tbl:1 ] was obtained . to calculate the cycle index
@xmath141 we need to calculate the sum .
the vectors @xmath142 for a given @xmath4 can be obtained using the standard package @xmath143 , containing the function @xmath144}$ ] , which returns the list of partitions of @xmath4 .
any partition @xmath35 of this list has the following form @xmath145 with @xmath146 and @xmath147 .
the relation @xmath148 gives a one - to - one correspondence between the partitions of @xmath4 and the vectors @xmath142 .
here @xmath149 is on @xmath150-th place and the other components are zero .
this correspondence can be realized in _
@xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{ct[p\ _ ] : = } \ & \mathtt{module[\{c = table[0 , \{plus\ @@\ p\}]\ } , } \\ &
\mspace{20mu } \mathtt{set[part[c , \#[[1 ] ] ] , length[\#]]\ & /@\ split[p ] ; } \\ & \mspace{20mu } \mathtt{return[c ] } \\ & \mathtt { ] } \end{split}\ ] ] then we need to define the cross - product .
we introduce the object @xmath151}$ ] which represents the @xmath4-th power of the @xmath35-th independent variable .
the code @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{cp[var[p\ _ , n\ _ ] , var[q\ _ , m\_]]\ : = \ var[lcm[p , q ] , n\ m\ gcd[p , q]]}\ ] ] reproduces the definition . to extend the cross - product for arbitrary polynomials
we need the following definitions ( the order in which they are given is important ) :
@xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[v\ _ , c\ _ ] : = c\ v /;\ numericq[c ] } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[c\ _ , v\ _ ] : = c\ v /;\ numericq[c ] } \\
{ { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[v1\ _
+ v2\ _ , v3\ _ ] : = cp[v1 , v3 ] + cp[v2 , v3 ] } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[v3\ _ , v1\ _ + v2\ _ ] : = cp[v3 , v1 ] + cp[v3 , v2 ] } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[v3\ _ , v1\ _ v2\ _ ] : = cp[v3 , v1]\ cp[v3 , v2 ] } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[v1\ _ v2\ _ , v3\ _ ] : = cp[v1 , v3]\ cp[v2 , v3 ] } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[c\ _ v1\ _ , v2\ _ ] : = c\ cp[v1 , v2 ] /;\ numericq[c ] } \\
{ { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{cp[v1\ _ , c\ _ v2\ _ ] : = c\ cp[v1 , v2 ] /;\ numericq[c ] } \end{split}\ ] ] the following code reproduces the identities @xmath152 and @xmath153 respectively : @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{var /:\
power[var[p\ _ , n\
_ ] , m\_]\ : = \ var[p , n\ m ] } \\ { { \scriptstyle in[\arabic{inpromtcount}]:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{var /:\ var[p\ _ , n\_]\ var[p\ _ , m\_]\ : = \
var[p , n + m ] } \end{split}\ ] ] the code for the functions @xmath154 and @xmath57 is obvious : @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{a[k\_]\ : = \ \frac{1}{k } plus\ @@\
moebiusmu\left[\frac{k}{\#}\right]\ & /@\
divisors[k]\right ) } \\ { { \scriptstyle in[\arabic{inpromtcount}]:=
\quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{b[k\_]\ : = \ \frac{1}{k } plus\ @@\ \left(2^{\#/2 } moebiusmu\left[\frac{k}{\#}\right]\ &
/@\ complement[divisors[k ] , \
divisors[k/2]]\right ) } \end{split}\ ] ] a factor of of the sum can be coded as @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{factor[i\_]\ : = } \ & \mathtt{times\ @@\ ( var[\ # , a[\ # ] ] \ & /@\
divisors[i ] ) + times\ @@\ ( var[\ # , b[\ # ] ] \ & } \\ & \mathtt{/@\ complement[divisors[2i ] , divisors[i ] ] ) } \end{split}\ ] ] a term of the sum is coded as @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}&\mathtt{term[c\_]\ : = \ module[\{t\ } , } \\ & \mspace{20mu}\mathtt{t = mapindexed[if[\#1 = = 0 , 1 , if[\#1 = = 1 , factor[\#2[[1 ] ] ] , } \\
& \mspace{40mu}\mathtt{fold[cp , factor[\#2[[1 ] ] ] , table[factor[\#2[[1]]],\{\#1 - 1\}]]]]\ & , c ] ; } \\ & \mspace{20mu}\mathtt{fold[cp , first[t ] , rest[t ] ] } \\ & \mathtt { ] /(times\ @@\ ( factorial /@\
c ) times\ @@\ mapindexed[(2 \#2[[1]])^{\#1 } \ & , c ] ) } \end{split}\ ] ] the whole sum is given by @xmath64:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{cycleindex[n\_]\ : = \
( ct /@\ partitions[n]))}\ ] ] the number @xmath155 can be obtained as follows : @xmath99:= \quad \setcounter{outpromtcount}{\value{inpromtcount}}\stepcounter{inpromtcount}}}\mathtt{ub[n\_]\ : = } \ & \mathtt{\frac{1}{2 } ( ( cycleindex[n]\ /.\ \{var[p\ _ , k\_]\ \to\ 2^k\ } ) } \\ & \mathtt{+ ( cycleindex[n]\ /.\ \{var[p\ _ , k\_]\ : \to\ if[evenq[p ] , 2^k , 0]\ } ) ) } \end{split}\ ] ] then one can get the numbers form the table [ tbl:1 ] : the code @xmath156 , \{n , 1 , 5\}]}$ ] produces @xmath157 .
a much faster way of calculating the walsch - hadamar transform of a boolean function @xmath82 ( i.e. calculation of all @xmath158 numbers @xmath74 , @xmath75 ) is based on the following decomposition of the hadamard matrix : @xmath159 which can be proved by induction .
the relation @xmath160 can be written as @xmath161 for @xmath162 and @xmath163 .
the walsh - hadamard transform can be calculated according to as @xmath164 starting with the vector @xmath165 we calculate vectors @xmath166 for @xmath167 .
then @xmath168 is @xmath169 .
the algorithm in _
c _ is presented in the function _
whtl _ in the file _
whtl.c _ below . to turn _ whtl.c _ into a program usable from inside
_ mathematica_another file , _
_ is necessary .
\textcolor{gray}{// \textit{xlen is of the form $ 2^n$ , returns $ n$ } } \textcolor{blue}{long } _ _ log(\textcolor{blue}{long } xlen ) \ {
i = 1 , m = xlen ; \textcolor{blue}{while}((m > > = \textcolor{red}{1 } ) ! = \textcolor{red}{1 } ) i++ ; \textcolor{blue}{return } i ; \ } \textcolor{blue}{void } whtl(\textcolor{blue}{int * } x , \textcolor{blue}{long } xlen ) \ { \textcolor{blue}{int * } y = ( \textcolor{blue}{int*})calloc(xlen , \textcolor{blue}{sizeof}(\textcolor{blue}{int } ) ) ; \textcolor{blue}{long } i , j , k , t , pow1 , pow2 , ind1 , ind2 ; \textcolor{blue}{long } n = _ _ log(xlen ) ; \textcolor{blue}{for}(k = \textcolor{red}{1 } ; k
< = n ; k++ ) \ { pow1 = \textcolor{red}{1}<<(k-\textcolor{red}{1 } ) ; pow2 = \textcolor{red}{1}<<(n - k ) ; \textcolor{blue}{for}(j = \textcolor{red}{0 } ; j < pow2 ; j++ ) \
{ t = j<<k ; \textcolor{gray}{// \textit{in c arrays are numbered from 0 } } \textcolor{blue}{for}(i = \textcolor{red}{0 } ; i < pow1 ; i++ ) \
{ ind1 = t + i ; ind2 = ind1 + pow1 ; y[ind1 ] = x[ind1 ] + x[ind2 ] ; y[ind2 ] = x[ind1 ] - x[ind2 ] ; \ } \ } \textcolor{blue}{if}(k <
n ) memcpy(x , y , \textcolor{blue}{sizeof}(\textcolor{blue}{int})*xlen ) ; \ } on a unix system the files are compiled with the following command @xmath170 this command produces an executable file _
whtl_. to use it in _
mathematica_it must be installed as @xmath171;}\ ] ] then it can be used as @xmath172}\ ] ] where all @xmath173 , @xmath174 are @xmath175 . | the relation between the boolean functions and bell inequalities for qubits is analyzed .
the connection between the maximal quantum violation of a bell inequality and the nonlinearity of the corresponding boolean function is discussed .
a visualization scheme of boolean functions is proposed .
an attempt to classify bell inequalities for qubits is made , a weaker result ( classification with respect to jevons group ) is obtained .
the fractal structure of the classification is shown .
all constructs are illustrated by _
mathematica_code . |
qcd sum rules @xcite represent one approach towards understanding the masses of hadrons .
hadronic parameters , among them the masses , are linked to expectation values of qcd operators , the condensates .
the nonzero values of condensates point to a complicated structure of the qcd vacuum , i.e. , of the ground state of strong interaction .
there are a few particularly important condensates : ( i ) the chiral condensate , @xmath2 , is connected with the spontaneous breaking of chiral symmetry .
together with the explicit chiral symmetry breaking by nonzero quark masses @xmath3 , the pion mass @xmath4 is related via the gell mann - oakes - renner relation @xmath5 @xcite to the pion decay constant @xmath6 and the chiral condensate .
( ii ) the gluon condensate @xmath7 ( with @xmath8 as chromodynamic field strength tensor ; @xmath9 is the strong coupling ) is anchored in the qcd trace anomaly and the breaking of the dilatation symmetry .
( iii ) the quark - gluon condensate @xmath10 may also be considered as order parameter of chiral symmetry @xcite .
there are ( infinitely ) many other condensates which may be ordered according to their mass dimension .
these quantities occur in the evaluation of the current - current correlator by means of an operator product expansion ( ope ) .
since organized as corrections in increasing powers of inverse hadron momentum , the ( poorly known ) higher - order condensates are expected to be of minor importance , so that the few low - dimension condensates suffice . however , there are special circumstances , for instance in the borel transformed qcd sum rules for light vector mesons , in which the low - order condensates are numerically suppressed .
specifically , the chiral condensate enters in the renormalization invariant combination @xmath11 being a tiny contribution due to the smallness of @xmath3 . instead
, the four - quark condensates become important . in a flavor - symmetric vacuum
, there are 10 independent four - quark condensates @xcite , all with highly unsettled values .
the situation becomes even more dramatic when extending the qcd sum rule approach to a strongly interacting medium @xcite . at finite baryon density @xmath12 and temperature @xmath13 the condensates
are modified , as exhibited in fig .
[ fig.1 ] for the chiral and the gluon condensate , and further condensate structures arise .
this modification is expected to be reflected in hadron properties @xcite . in particular , the approach to chiral symmetry restoration was thought to be measurable by investigating hadrons embedded in a strongly interacting medium @xcite .
the qcd sum rule approach provides , however , a more involved relation .
the in - medium modifications of light vector mesons depend crucially on the density dependence of a certain combination of four - quark condensates @xcite . in a strongly interacting medium ,
there are 44 independent four - quark condensates in the @xmath14 sector @xcite ; if flavor symmetry is satisfied this number reduces to 20 .
these enter the qcd sum rules for different hadrons in different combinations .
therefore , one can hope that a systematic investigation of various hadrons may shed light on the important four - quark condensate combinations .
density and temperature dependence of chiral condensate ( left panel ) and gluon condensate ( right panel ) .
the employed approximation is reliable only for small values of @xmath13 and @xmath12 @xcite .
note the different slopes in @xmath12 direction.,title="fig:",width=226 ] density and temperature dependence of chiral condensate ( left panel ) and gluon condensate ( right panel ) .
the employed approximation is reliable only for small values of @xmath13 and @xmath12 @xcite .
note the different slopes in @xmath12 direction.,title="fig:",width=226 ] formally , four - quark condensates are qcd ground state expectation values of hermitian products of four quark operators which are to be dirac and lorentz scalars , color singlets and are to be invariant under parity and time reversal ( the latter for equilibrated systems ) .
physically , the four - quark condensates quantify the correlated production of two quark - antiquark pairs in the physical vacuum .
in contrast to the square of the chiral ( two - quark ) condensate , which accounts for uncorrelated production of two of these pairs , the four - quark condensates are a measure of the correlation and thus expose the complexity of the qcd ground state .
especially , deviations from factorization , i.e.the approximation of unknown four - quark condensates in terms of the squared chiral condensate , justified in the large @xmath15 limit @xcite , represent effects of these more involved correlations . as mentioned above , the chiral condensate @xmath16 is often considered an order parameter for the @xmath17 chiral symmetry of qcd and , therefore , its density and temperature dependence is investigated extensively . the change of @xmath16 ,
however , might partially originate from virtual low - momentum pions and thus could not clearly signal partial restoration of chiral symmetry in matter @xcite .
the interpretation of four - quark condensates as order parameters for spontaneous breaking of chiral symmetry is still an open issue . in @xcite a specific combination of four - quark condensates arising from the difference between vector and axial - vector correlators
is proposed as such an alternative parameter .
this combination ( shown to agree with vacuum factorization in the analysis of @xmath18 decay data @xcite ) is distinct from the four - quark condensates in the nucleon channel as well as from the combination in the @xmath0 sum rule @xcite .
for instance , in vacuum qcd sum rules for the nucleon @xcite the four - quark condensate combination contains a chirally invariant part when assuming flavor symmetry ( @xmath19 is a flavor vector ) @xmath20 - \frac34 \left [ \text{color structures with } \ , \lambda^a \right ] , \nonumber\ ] ] since the condensates @xmath21 and @xmath22 are invariants w.r.t .
the @xmath17 transformation @xmath23 and contains another part @xmath24 - \frac34 \left [ \text{color structures with } \ , \lambda^a \right ] \nonumber\ ] ] including condensates which break the @xmath17 symmetry .
the vanishing of the four - quark condensates in this second part would restore the broken @xmath17 symmetry .
the combinations are weighted by polynomials of the mixing parameter @xmath25 allowing general forms of the nucleon interpolating field . in the preferred case @xmath26 ( the ioffe current for the nucleon ) only the chirally invariant part survives and thus this remainder can not be used as an order parameter .
additional insight into the change of four - quark condensates and their role as order parameters of spontaneous chiral symmetry breaking could be acquired from other hadronic channels . nevertheless , when identifying chiral symmetry restoration with degenerating vector ( @xmath27 ) and axial - vector ( @xmath28 ) spectral functions the four - quark condensates occur as order parameter @xcite @xmath29 with @xmath30 showing ( i ) that four - quark condensates are important for chiral symmetry restoration in the sense of degeneracy of vector and axial - vector currents , and ( ii ) that such formulations refer to moments , i.e. , weighted integrals .
our contribution is organized as follows . in section [ sec : disprel ] , we consider a dispersion relation being the starting point of the qcd sum rule approach .
section [ sec : probes ] is devoted to specific cases : the status of the @xmath0 meson is recapitulated ( [ sec : probes].1 ) and the nucleon is discussed ( [ sec : probes].2 ) .
the case of the @xmath1 meson as an example for the light - heavy flavor sector is dealt with ( [ sec : probes].3 , where a brief remark is made on the @xmath31 as example for the heavy - heavy sector ) .
the summary can be found in section [ sec : summary ] .
let be given the analytical properties of the correlation function @xmath32 of a selected hadron ( actually , all hadronic excitations with a given set of the same quantum numbers ) , such that it can be expressed through its spectral function @xmath33 via the dispersion relation @xmath34 this leads to the celebrated form employed in qcd sum rules .
the l.h.s . contains the hadronic spectral function , while the r.h.s .
is evaluated , for large euclidian energies @xmath35 and fixed momenta @xmath36 , by means of the ope .
we assume for the moment being that the r.h.s.is settled . in order to attempt the isolation of the contribution of the lowest hadronic excitation one can perform the following chain of reformulations .
define even ( e ) and odd ( o ) contributions @xmath37 and split the integrals with respect to the low - lying hadron under consideration in the intervals @xmath38 for positive energy and @xmath39 for negative energy @xmath40 the integrals on the r.h.s . can be converted into expressions similar to the ope by exploiting the semi - local duality hypothesis .
defining further the moments @xmath41 the even and odd sum rules can be rephrased to formulate relations which rather depend on ratios than purely on absolute spectral integrals . in a combined resulting sum rule @xmath42 the integral over
the negative energy contribution seems to be eliminated .
( this contribution , however , is still included in the ratio @xmath43 . ) in the special case of an anti - symmetric spectral function , @xmath44 , e.g. for @xmath45 and @xmath0 , one obtains with @xmath46 that @xmath47 , @xmath48 and ( using @xmath49 ) @xmath50 .
therefrom one can define a normalized moment for the borel transform @xmath51 where @xmath52 is the borel mass .
determined by the ratio of @xmath53 and its logarithmic derivative , this quantity is model independent ( but suffers from the ad hoc introduced continuum threshold @xmath54 ) .
its meaning becomes obvious for a pole ansatz , @xmath55 , where @xmath56 follows . in the generic case , however , the excitation spectrum of particles and anti - particles in a medium is asymmetric , and the dispersion relation ( [ eq : dispersionrelationansatz ] ) alone seems not to allow separate model - independent statements about the low - lying excitation at positive energy .
this appears as a severe limitation of this form of the qcd sum rule approach .
the @xmath0 and @xmath45 mesons , including their mixing , have been dealt with in @xcite in detail . in @xcite
the combined four - quark condensates contributing to the ope terms @xmath57 in the qcd sum rule for the @xmath0 meson have been assigned a strong density dependence , at least @xmath58 .
otherwise the numerically large landau damping term would push up the weighted strength @xcite .
in contrast , the experiment @xcite finds a lowering of the @xmath0 strength .
[ fig.2 ] exhibits the density dependence of the mass parameter ( [ mass_parameter ] ) .
note that this parameter coincides only in zero - width approximation with the @xmath0 pole mass squared ; in general it is a normalized moment of @xmath59 to be calculated from data or models .
-3 mm the nucleon qcd sum rule is more involved since several dirac structures need to be considered @xcite : @xmath60 i.e. , besides the even and odd parts one has @xmath61 with @xmath62 , where @xmath63 refers to the four - velocity of the medium .
we evaluate the qcd sum rule by utilizing the ansatz @xmath64 , \ ] ] with the implication @xmath65 the quantities @xmath66 are related , in the ansatz , to the scalar and vector contributions to the self - energy , @xmath67 and @xmath68 .
figure [ fig.3 ] exhibits @xmath69 for special values of the density dependence parameters @xmath70 @xmath71 @xmath72 entering the three independent four - quark condensate combinations @xmath73 = { \ensuremath{\kappa^{\rm med } _ { \rm s } } } { \ensuremath{\left \langle \bar{q } q \right \rangle_{\rm vac } } } \dfrac{3}{2 } n , \ ] ] -6 mm @xmath74 = \dfrac32 \left ( { \ensuremath{\kappa^{\rm vac } _ { \rm q } } } { \ensuremath{\left \langle \bar{q } q \right \rangle_{\rm vac}}}^2 + { \ensuremath{\kappa^{\rm med } _ { \rm q } } } { \ensuremath{\left \langle \bar{q } q \right \rangle_{\rm vac } } } \dfrac{\sigma_{\rm n } } { m_{\rm q } } n \right),\end{aligned}\ ] ] -6 mm @xmath75 = \dfrac32 { \ensuremath{\tilde{\kappa}^{\rm med } _ { \rm v } } } { \ensuremath{\left \langle \bar{q } q \right \rangle_{\rm vac } } } \dfrac{\sigma_{\rm n } } { m_{\rm q } }
n .\end{aligned}\ ] ] we use further @xmath76 and assume equal continuum thresholds for the three components of the sum rule for contributions @xmath62 .
changes of the self - energies under individual variations of @xmath77 , @xmath78 , @xmath79 are reported in @xcite .
the variation of the genuine chiral condensate @xmath2 causes only tiny changes .
the @xmath1 meson qcd sum rule is entered @xcite by the combination @xmath80 , i.e. , the charm mass @xmath81 acts as magnifier of the chiral condensate .
the evaluation of the sum rule for the @xmath82 , @xmath83 , @xmath84 multiplet involves the renormalization of non - perturbative condensates in order to compensate the parts of perturbative contributions , which are dominated by the non - vanishing mass scales .
these parts are of non - perturbative origin and thus should be covered by the appropriately defined non - perturbative condensates ( usually called non - normal ordered
condensates @xcite ) @xmath85 \right \rangle,\ ] ] which quantify the non - perturbative regime @xcite .
the integral denotes the perturbatively calculated contribution @xcite .
the operator @xmath86 is a function of the covariant derivative and contains general dirac structures , @xmath28 is the gluon gauge field in fixed - point gauge and @xmath87 its fourier transform ; @xmath88 denotes normal ordering , and @xmath89 is the quark propagator in the gluonic background field
. application of this condensate renormalization cancels mass logarithms in ope calculations @xcite and causes a mixing between different types of condensates , e.g. , the chiral condensate mixes with the gluon condensate ( cf .
also @xcite ) @xmath90 which relates in the heavy quark sector the heavy quark condensate ( e.g. @xmath91 for the charm condensate ) to gluon condensates @xcite @xmath92 this is utilized in the qcd sum rule for the @xmath31 meson entered by the combination @xmath93 , which one expects to exhibit the weak density dependence of the gluon condensate , as shown in fig . 1 .
in fact , the evaluation of the qcd sum rule for @xmath31 shows only a tiny change of the in - medium mass @xcite . in finite density qcd sum rules for the @xmath1 meson ,
medium - specific condensates appear and introduce , via their mixing , further gluon condensates into the ope .
an example is @xmath94 where @xmath95 is the renormalization scale , already introduced in ( [ eq.17 ] ) . the complete reevaluation of the in - medium @xmath1 meson qcd sum rules , in particular the complete ope side up to mass dimension 6 for products of quark masses and condensates , is under investigation .
in summary we survey the qcd sum rules for the @xmath0 meson and the nucleon .
the sum rules allow for a direct relation of hadron properties to qcd condensates which change at nonzero baryon density .
the genuine chiral condensate has numerically a small influence in the borel transformed sum rules . instead , the four - quark condensates determine to a large extent the density dependence of these hadrons composed of light quarks .
in contrast , in the heavy - light quark sector , e.g. , represented by @xmath1 mesons , the impact of the chiral condensate is expected to become noticeable . a systematic investigation of various hadron species is needed to investigate the role of the many four - quark condensates entering the qcd sum rules in different combinations . 99 -2pt m.a .
shifman , a.i .
vainshtein , v.i .
zakharov , _ nucl .
_ b 147 ( 1979 ) 385 g. colangelo , j. gasser , h. leutwyler , _ phys .
86 ( 2001 ) 5008 t. doi , n. ishii , m. oka , h. suganuma , _ phys .
_ d 70 ( 2004 ) 034510 r. thomas , t. hilger , b. kmpfer , _ nucl .
phys . _ a 795 ( 2007 ) 19 a. i. bochkarev , m. e. shaposhnikov , _ nucl .
_ b 268 ( 1986 ) 220 t. hatsuda , s.h .
lee , _ phys .
_ c 46 ( 1992 ) r34 g.e .
brown , m. rho , _ phys .
_ 363 ( 2002 ) 85 and further references therein r. thomas , s. zschocke , b. kmpfer , _ phys .
_ 95 ( 2005 ) 232301 r. thomas , s. zschocke , t. hilger , b. kmpfer , invited talk at 44th international winter meeting on nuclear physics , bormio , italy , 29 jan - 5 feb 2006 , hep - ph/0602163 s. leupold , _ phys .
_ b 616 ( 2005 ) 203 m. birse , _ phys .
_ c 53 ( 1996 ) 2048 s. leupold , _ j. phys .
_ g 32 ( 2006 ) 2199 j. bordes , c. a. dominguez , j. penarrocha , k. schilcher , _ jhep _ 02 ( 2006 ) 037 j.i .
kapusta , e.v .
shuryak , _ phys .
_ d 49 ( 1994 ) 4694 s. zschocke , b. kmpfer , _ phys .
_ c 70 ( 2004 ) 035207 s. zschocke , b. kmpfer , o.p .
pavlenko , _ eur .
_ j. a 15 ( 2002 ) 529 ; _ phys .
_ b 562 ( 2003 ) 57 ; b. kmpfer , s. zschocke , _ prog .
_ 53 ( 2004 ) 317 d. trnka et al .
( cb - taps ) , _ phys .
_ 94 ( 2005 ) 192303 t.d .
cohen , r.j .
furnstahl , d.k .
kriegel , x.m .
jin , _ prog .
_ 35 ( 1995 ) 221 o. plohl , c. fuchs , _ phys .
_ c 74 ( 2006 ) 034325 a. hayashigaki , _ phys .
_ b 487 ( 2000 ) 96 p. morath , w. weise , s.h .
lee , 17th autumn school ; qcd : perturbative or nonperturbative ?
lisbon , portugal , 29 sep - 4 oct 1999 , p. 425
s. narison , _ qcd as a theory of hadrons _ , cambridge univ
. press , 2004 v.p .
spiridonov , k.g .
chetyrkin , _ sov .
_ 47 ( 1988 ) 522 a.g .
grozin , _ int . j. mod
. phys . _ a 10 ( 1995 ) 3497 m. jamin , m. mnz , _ z. phys .
_ c 60 ( 1993 ) 569 d.j .
broadhurst , s.c .
generalis , _ phys .
_ b 165 ( 1985 ) 175 f. klingl , s .- s .
kim , s. h. lee , p. morath , w. weise , _ phys .
lett . _ 82 ( 1999 ) 3396 | the qcd sum rule approach to the in - medium behavior of hadrons is discussed for @xmath0 meson , nucleon and @xmath1 meson .
emphasis is devoted to the impact of four - quark condensates and to order parameters of spontaneous symmetry breaking . |
in recent decades researchers have focused their interest on materials in the nano - scale regime . the blue shift in band - gap of materials as their grain size reduces
is well established .
@xcite also , when the particle size of noble metals such as gold , silver etc are reduced to the nano - scale , one observes interesting changes in their physical and chemical properties .
for example , when the metal grain size is of the order of the incident light s wavelength , the free electrons of the metal are set into oscillation .
the collective / quantized oscillation of these electrons are called plasmons .
@xcite at a certain wavelength of the incident light , they resonate .
this phenomenon is called surface plasmon resonance ( spr ) .
@xcite the main attraction of spr is the tunability of the absorbance in the uv - visible range based on the metal nano - particle s size and shape along with their and their surrounding medium s dielectric constant .
@xcite spr finds application in sensors and scintillation screens .
@xcite recent works by the authors have shown that alkali halides , such as cesium bromide and cesium iodide , may give spr peaks in their absorption spectra either by inclusion of silver metal nano - particles @xcite or due to segragation of cesium from the cesium halide molecule on the formation of color centers with aging .
@xcite fig [ summary2 ] is a pictorial representation of cesium nano - rods formation in cesium iodide background . within the grains of csi films ( shown in the cross - sectional view in fig [ summary2 ] ) ,
color centers are formed with the halide centers jumping out of the lattice .
these lattices of only cesium metal atoms take - up tetragonal lattice structure and apply strain on neighbouring csi unit cells .
@xcite the strain results in the csi unit cells to take up a tetragonal structure .
meanwhile , cesium metal structures migrate towards the grain boundary where grain division results in cesium nano - rods .
@xcite thus explaining the segragation of cesium from cesium iodide with time ( ageing ) .
it is these metal nano - rods in csi background ( not shown in fig [ summary2 ] ) that results in spr peaks in the uv - visible absorption spectra .
, width=264 ] as stated previously , the spr peak position is strongly influenced by the dielectric constant ( refractive index ) of the background material .
in fact , the matrix background s dielectric constant ( @xmath3 ) is an important input parameters in standard models which are used for estimating the spr peak positions , like the mie @xcite or gans model .
@xcite however , due to the transparent nature of the cesium iodide films , the experimental determination of its refractive index , an in term dielectric constant in the visible range , is very difficult .
hence , the dielectric constant of csi reported in literature is usually obtained by theoretical considerations . * ? ? ?
* however , the emphasis in all the reports have been on the insulator to metal transition that occurs when the unit cell is under pressure , with pressure applied being isotropic .
this results in the variation in lattice parameters abid with the simple cubic structure maintained . hence
, using data from literature would at best give results that are indicative in nature .
since , dielectric constant of cesium iodide with tetragonal unit cell structure is not reported , we have evaluated the same theoretically using abinit ( version 8.0.8 ) .
abinit is a standard package/ software used to compute the energy band structure diagram for evaluation of band - gap ( @xmath4 ) , the density of states ( dos ) and dielectric constant ( @xmath5 ) .
the following section discusses the theoretical background of the software and the results generated by our calculations .
there are two methods by which the quantum mechanical behaviour of electrons in a crystal can be evaluated . the first is called the pseudo - potential ( pp ) method . @xcite the pp theory excludes the core - shell electrons from consideration while determining the oscillating wave functions
. the results are hence smooth ( pseudo ) wave functions of valence electrons that can be decomposed in a series for about @xmath6 plane waves , as compared to the true all - electron ( including inner shell electrons ) wave functions which would be a series of @xmath7 plane waves .
the second method involves including all the electrons and is called the full potential linearized augmented plane wave method ( flapw ) . @xcite in this approach , the volume of the crystal is divided into two regions
. sites of the atoms are represented by spheres called the muffin - tin spheres ( mt ) , while the remainning is empty space/ volume of the crystal surrounding the mt spheres is termed the interstitial region " .
the flapw basis functions are represented within these two regions as @xmath8 & \text{muffin - tin sphere}\\ \end{cases}\ ] ] the vector terms are shown in bold .
the charge density is given as ( the potential is also defined in similar way ) @xmath9 for this article , we have used the projector augmented waves ( paw ) method .
@xcite this method uses features of both above mentioned approaches , namely , pp and flapw . as in flapw , the crystal is mapped into two parts , the so - called spheres of augmentation centered at atom sites and the interstitial region .
the representation of the wave function in these two regions is also different . within the augmentation sphere , near the center
, the wave function of the core electron is represented by atomic partial waves , @xmath10 but in the middle and peripheral part of the sphere , the pseudo wave function is expressed in terms of auxiliary partial waves @xmath11 in the interstitial region the smooth pseudo wave function is expanded as plane waves @xmath12 the crystal s all electron wave function of an electron is hence composed of these three components and is given as @xmath13 where @xmath14 is the projector function .
the all electron function is derived from the pseudo wave function by the transformation operator @xmath15 , i.e. @xmath16 where @xmath17 and ` i ' is the identity operator.one can now substitute this expression into the schrodinger wave equation and then solve the eigenvalue problem written in the form @xmath18 results and discussion ---------------------- in its natural state , cesium iodide crystal takes a simple cubic structure , with the center of the cube being occupied by an iodine atom and cesium atoms positioning themselves at the corners of the cube ( see fig [ summary ] ) .
this arrangements of the atoms implies that there are two atoms per unit cell with p4/mmm space symmetry . for the unstrained lattice ,
the lattice paramters ( a=4.57 ) taken were those reported in the astm card no .
06 - 0311 .
the electronic properties of csi crystal were evaluated by means of the abinit code . @xcite the exchange - correlation function of predew- burke- emzerhof ( pbe ) was used for calculations .
@xcite the projector augmented waves ( paw ) method was used for solving the system of equations given in the previous section with respect to crystal eigenenergies @xmath19 and pseudo wave functions @xmath20 .
the paw basis states have been generated by means of the atompaw program @xcite for the valence basis states of @xmath21 for cesium and @xmath22 for iodine .
the system of equations was solved self - consistently and the full convergence of the binding energy of the unit cell was achieved with accuracy of @xmath23 ha .
the calculations for the unstrained lattice allowed us to verify our theoretical calculations with those determined and reported experimentally ( 6 ev ) .
@xcite , width=188 ] .[tablei ] listing of lattice paramters which were used for calculations in this study .
the lattice parameters were experimentally observed in a previous study by the authors.@xcite [ cols="^,^,^ " , ] 0.1 cm while the variation of the band - gaps ( both direct and indirect ) showed a linear increase with lattice parameter ` c ' , it linearly decreased with increasing ` a ' ( fig [ linear ] ) . hence , both lattice parameters , ` a ' and ` c ' play an important role in the formation of bands and in turn band - gap in csi . however , considering the contribution of the valence band came from the iodine atoms and the conduction bands were contributed by the cesium s d - orbital , it would be interesting to establish whose contribution influences csi s band - gap under stress .
for this , we did a chi - square fit to determine the functional dependence of @xmath4 on the two independent variables .
the calculations were done using standard software ( origin 8.0 ) .
a very good fit , with low residual ( @xmath24 ) was obtained @xmath25 fig [ 3dbandgap ] shows the 3d plot with family of curves along with the theoretically determined band - gaps for tetragonal lattices .
the family of curve essentially helps us to visualize the variation in @xmath4 when one of the lattice parameters is held constant .
the equation ( eqn [ chi ] ) and the graph clearly shows that the band - gap strongly depends on the change in lattice parameter ` c ' compared to ` a ' .
clearly , the interaction of cesium - iodine atoms plays a more prominent role as compared to the cesium - cesium atoms interaction .
0.2 cm the systematic variation in band - gap of csi under stress would also reflect in other optical properties such as refractive index , or dielectric constant .
theoretical calculations with abinit were extended for determining the dielectric constant of csi for the lattice constants listed in table i. fig [ epr1](a ) shows the variation of the real and imaginary part of the dielectric constant with incident photon energy . in the uv - visible - nir region ( region of our interest )
we find the imaginary component to be non - existing , i.e. @xmath26 .
this is expected since csi films/ crystals are transparent in nature and hence non - absorbing in visible and large uv wavelengths .
the real component of the dielectric constant increases as the wavelength decreases ( as we move from nir to uv wavelengths ) .
fig [ epr1](b and c ) shows the variation of @xmath27 with lattice parameters .
the trend is inverse to the trend seen in band - gaps ( fig [ linear ] ) , e.g. while the band - gap increased with increasing ` c ' , @xmath28 decreases with increasing ` c ' .
shukla et al @xcite did theoretical calculations for varying lattice constants of cubic " csi and reported decreasing dielectric constant with increasing lattice paramter .
however , the work does not relate to experimental results since csi under stress takeup tetragonal structure . in the following section ,
the dielectric constants obtained from here are used as input parameters for gans model simulation .
the simulation would theoretically predict the peak in extinction peak position ( @xmath2 ) indicating surface plasmon resonance ( spr ) caused due to the formation in nano - clusters of metal particles .
the absorption spectra of metallic nanoparticles can be explained by mie theory , which essentially is maxwell s equations solved with appropriate boundary conditions corresponding to light interacting with spheical metal particles . as the electric field of the light wave oscillates , electrons of the metal move resulting in a polarization given as @xmath29e_o\nonumber\end{aligned}\ ] ] where ` @xmath5 ' is the complex dielectric constant ( @xmath30 ) of the metal nano - particles of radius ` a ' and ` @xmath31 ' is the dielectric constant of the surrounding media . a maximum , corresponding to resonance
is achieved when the denominator ( @xmath32 ) is a minimum .
the extinction coefficient of spherical particles is given by @xmath33 shows the dependency of the dielectric constants of metal and the surrounding medium .
gan @xcite extended the mie theory for non - spherical particles and showed that the above equation needs modification for nano - particle that are deformed from spherical shapes and have taken an ellipsoid ( prolate/ oblate ) shape . according to theory
, spr oscillations occur in two directions of the ellipsoid .
the theory predicts splitting of spr into longitudinal and transverseal modes when light is incident along or perpendicular to the major axis , respectively .
@xcite the extinction coeffiecient ( @xmath34 ) for the ellipsoid shaped nano - particle is given as @xmath35 ^ 2}\ ] ] where ` v ' is the volume occupied by the particle , @xmath36 the angular frequenecy of the incident light ,
c " is the speed of light and @xmath37 is the polarization factor along the three axes a , b and c of the ellipsoid , with @xmath38 .
@xmath37 is defined as ( j subscript identifies the axis a , b or c ) @xcite @xmath39\ ] ] and ` e ' is the aspect ratio ( which defines the shape of the non - spherical metal particle ) given by @xmath40^{1/2}\ ] ] again , the denominator of the scattering extinction coefficient should be minimum for the resonance .
mathematically , the resonance condition is given by @xmath41 ^ 2+\epsilon^2_2=min\ ] ] thus , the nature of spr peaks depends on the shape and the dielectric constant of the metal as well as the nano - particle s surrounding media ( via it s dielectric constant @xmath31 ) . for gan model calculations
, we have assumed the cesium metal structure to be nano - rod or ellipsoidal shaped with an aspect ratio of 0.86 ( same as that observed in our earlier experimental studies ) .
for this aspect ratio , the peak due to longitutional and transverse modes are very near and unresolvable , giving an appearance to a very broad peak .
also , instead of a constant @xmath31 for csi background , we have used the theoretically evaluated @xmath42 .
0.2 cm since the stressed csi lattice showed anisotropic behavior , we have done theoretical calculations ( gan model ) with light incident along ` a ' and ` c ' axis respectively .
fig [ gan ] shows gan model calculations for cubic and tetragonal lattice structured csi when light is incident along ` a ' and ` c'-axis respectively .
a red - shift in @xmath2 is clearly evident as we repeat our calculations for lattice parameters listed from s.no .
1 to 5 of table i. that is , a red - shift in the spr peak position is evident with increasing stress in the lattice .
it might appear strange that a red - shift in spr peak position is seen in both graphs of fig [ gan ] considering the inverted slopes of fig [ epr1 ] ( b and c ) .
however , because of the inverted slope , as lattice parameter ` a ' increases ( and ` c ' decreases ) from s.no . 1 to 5 of table i , @xmath43 increases .
this increase in the background s dielectric constant , @xmath43 results in a red - shift in @xmath2 .
also , since the slopes of the curves in fig [ epr1 ] ( b and c ) show that the variation in @xmath43 along ` a ' direction ( @xmath0 ) is stronger than that along the ` c ' direction ( @xmath1 ) , a more prenounced red - shift in @xmath2 is seen along the ` a'-axis of the deformed cesium iodide s lattice .
the theoretical results are in agreement with the results presented by kuldeep et al .
@xcite however , the red shift reported experimentally was more than what we have evaluated here .
this may be due to the commulative effect of cesium grain size and surrounding dielectric coming into play in the experimental study , where it was impossible to resolve the contribution from each factor .
in conlcusion , we were able to calculate the electronic energy bands and dielectric constant of cesium iodide crystals taking into account the tetragonal lattice .
tetragonal lattices of csi experimentally occur when color centers form within the csi matrix .
the color centers give rise to non - uniform deformation of csi unit cell along the ` a ' and ` c ' directions .
the energy band gaps calculated here are in good agreement with experimental results .
finally , the dielectric constants have been calculated for all the deformed lattices ( table i ) .
the theoretical results give important input parameters needed for gan model calculation of extinction coefficients of csi .
the novel treatment gives interesting insight to the behavior of cesium iodide .
99 a.l.efros and a.l.efros , interband absorption of light in a semiconductor sphere , sov .
semicond . ,
* 16 * , 772 ( 1982 ) . a.i.ekimov and a.a.onushchenko , quantum size effect in the optical - spectra of semiconductor micro - crystals . ,
semicond . ,
* 16 * , 775 ( 1982 ) .
d.bohm and d.pine , a collective description of electron interactions : iii .
coulomb interactions in a degenerate electron gas , phys .
, * 92 * , 609 ( 1953 ) .
c.f.bohren and d.r.huffman , absorption and scattering of light by small particles ,
( john wiley & sons , inc .
, ny , 1983 ) .
c.noguez , surface plasmons on metal nanoparticles : the influence of shape and physical environment , j. phys .
c * 111 * , 3806 ( 2007 ) .
b.d.gupta and r.k.verma , surface plasmon resonance - based fiber optic sensors : principle , probe designs , and some applications , journal of sensors , * 2009 * , 1 ( 2009 ) .
lovkush , chhaya ravi kant , p.arun and kuldeep kumar , spr sensitivity of silver nanorods in csbr - ag nanocomposite thin films , mater .
, * 3 * , 076403 ( 2016 ) .
kuldeep kumar , p.arun , chhaya ravi kant , and bala krishna juluri , metal cluster s effect on the optical properties of cesium bromide thin films , appl .
lett . , * 100 * , 243106 ( 2012 ) .
kuldeep kumar , p.arun , chhaya ravi kant , n.c.mehra , vincent mathew , the effect of cesium metal clusters on the optical properties of cesium iodide thin films , appl .
a * 99 * , 305 ( 2010 ) .
g.mie , beitrage zur optik truber medien : speziell kolloidaler metallosungen , annalen der physik * 330 * , 377 ( 1908 ) .
r.gans , uber die form ultramikroskopischer silberteilchen , ann .
phys . , * 47 * , 270 ( 1915 ) .
d. vanderbilt , soft self - consistent pseudopotentials in a generalized eigenvalue formalism , phys .
b * 41 * , 7892 ( 1990 ) .
e.wimmer , h.krakauer , m.weinert , and a.j.freeman , full potential self consistent linearized augmented plane wave method for calculating the electronic structure of molecules and surfaces : @xmath44 molecule , phys .
b , * 24 * , 864 ( 1981 ) .
p.e.blochl , projector augmented - wave method , phys .
b * 50 * , 17953 ( 1994 ) .
x.gonze , b.amadon , p.m.anglade , j.m.beuken , f.bottin , p.boulanger , f.bruneval , d.caliste , r.caracas , m.cote , t.deutsch , l.genovese , ph.ghosez , m.giantomassi , s.goedecker , d.r.hamann , p.hermet , f.jollet , g.jomard , s.leroux , m.mancini , s.mazevet , m.j.t.oliveira , g.onidab , y.pouillon , t.rangel , g.m.rignanese , d.sangalli , r.shaltaf , m.torrent , m.j.verstraete , g.zerah , j.w.zwanziger , abinit : first - principles approach to material and nanosystem properties , comput . phys .
commun . , * 180 * , 2582 ( 2009 ) .
j.p.perdew , k.burke , m.ernzerhof , generalized gradient approximation made simple , phys .
letters , * 77 * 3865 ( 1996 ) .
n.a.w.holzwarth , a.r.tackett , g.e.matthews , a projector augmented wave ( paw ) code for electronic structure calculations , part i : atompaw for generating atom - centered functions , comput . phys . commun . ,
* 135 * 329 ( 2001 ) . j.j.hopfield and j.m.worlock , two - quantum spectrum of ki and csi , phys rev * 137 * , a1455 ( 1965 ) .
a.buzulutskov , a.breskin and r.,chechik , field enhancement of the photoelectric and secondary electron emission from csi , j. appl . phys . * 77 * 2138 ( 1995 ) .
triloki , r.rai , nikita gupta , nabeel f.a .
jammal , b.k.singh , effect of humid air exposure on photoemissive and structural properties of kbr thin film photocathode , nuclear instruments and methods in physics research a , * 787 * , 161 ( 2015 ) .
h.k.mao , y.wu , r.j.hemley , l.c.chen , j.f.shu , l.w.finger and d.e.cox , high - pressure phase transition and equation of state of csi , phys . rev . lett . * 64 * , 1749 ( 1990 ) . t.huang and a.l.ruoff , equation of state and high - pressure phase transition of csi , phys .
b * 29 * , 1112(r ) ( 1984 ) .
i.v.aleksandrov , a.f.goncharov , i.n.makarenko and s.m.stishov , new data on the equation of state for the phase transitions of cesium iodide ( the role of hydrostatics ) , sov
jetp , * 70 * , 745 ( 1990 ) .
s.lebegue , b.arnaud , m.alouani , and p.e.blochl , implementation of an all - electron gw approximation based on the projector augmented wave method without plasmon pole approximation : application to si , sic , alas , inas , nah , and kh , phys .
b , * 67 * , 155208 ( 2003 ) . f.aryasetiawan and o.gunnarsson , the gw method , rep . prog .
phys . , * 61 * , 237 ( 1998 ) .
l.hedin , new method for calculating the one - particle green s function with application to the electron - gas problem , phys .
, * 139 * , a796 ( 1965 ) .
s.shukla , d.kumar , n.n.shukla , r.prasad , dielectric critical behaviour at a metal - insulator transition under lattice compression , j. phys . chem .
solids * 66 * , 1150 ( 2005 ) a.burchanti , a.bogi , c.marinelli , c.maibohm , e.mariotti , s.sanguinetti and l.moi , optical characterization and manipulation of alkali metal nanoparticles in porous silica , j. eur .
d , * 49 * 201 ( 2008 ) .
s.link , m.b.mohamed , m.a.el-sayed , simulation of the optical absorption spectra of gold nanorods as a function of their aspect ratio and the effect of the medium dielectric constant , j. phys .
b , * 103 * , 3073 ( 1999 ) . | alkali halides like cesium iodide have received renewed attention as a material with potential application as detector coating due to the appearance of surface plasmon resonance absorption peak in their uv - visible absorption spectrum .
the formation of surface plasmon resonance peak has been traced to the formation of cesium metal clusters due to the aglomeration of color centers embedded in cesium iodide background .
this paper , based on experimental observation that cubic cesium iodide experiences stress due to color centers forming in the neighborhood and take up tetragonal lattice structure , investigates the tetragonal cesium iodide s band structure and dielectric constant using ab initio calculations .
both , the band - gap and dielectric constant of cesium iodide , show a systematic variation with changing lattice constant .
while the band - gap is seen to strongly depend on the lattice parameter ` c ' , the dielectric constant s variation along ` a ' direction ( @xmath0 ) was found to be stronger than that along the ` c ' direction ( @xmath1 ) .
the calculated dielectric constants were used as input parameter for gan model calculations of the extinction coefficient and evaluation of surface plasmon resonance peak position ( @xmath2 ) .
a red - shift in @xmath2 was seen with increasing dielectric constant that appears with stress acting on cesium iodide s lattice . |
in high-@xmath6 cuprates , the electronic density of states ( dos ) near the fermi energy has been demonstrated to decrease gradually with decreasing temperature , resulting in the pseudogap ( pg ) formation @xcite .
this pg , which has been found to progressively `` destroy '' the fermi surface ( fs ) @xcite , is on one hand a challenge to the conventional view of the fs itself , while on the other hand it allows one to reconcile the small number of carriers that participate in the charge transport in underdoped cuprates with the large fs observed by photoemission @xcite .
although the existence of the pg has been documented by many experiments @xcite , its nature and , particularly , its relation to the superconductivity ( sc ) remain far from being clear .
for example , according to photoemission and surface - tunneling studies @xcite , the pg evolves smoothly into the sc gap below @xmath6 , which implies a precursor - pairing origin of the pg ; in contrast , recent observations of a distinct sc gap that coexists with the pg below @xmath6 and tends to close at @xmath6 @xcite suggest that the pg might have nothing to do with superconductivity @xcite .
apparently , a crucial test for the origin of the pg would be its sensitivity to the magnetic field ; however , studies of the magnetic - field dependence of the pg have reported surprisingly controversial results @xcite . while the electronic dos in cuprates has mostly been studied by photoemission @xcite or surface - tunneling @xcite spectroscopies ,
the extremely anisotropic nature of the bi - based cuprates , where the crystal structure itself forms a stack of tunnel junctions , offers a possibility of _ intrinsic _ tunneling spectroscopy @xcite .
in fact , it has been shown that the @xmath0-axis transport in the bi - based cuprates is governed by the tunneling between cuo@xmath7 planes and , accordingly , the @xmath0-axis voltage - current characteristics directly reflect the dos structure of the cuo@xmath7 planes @xcite . it was reported that the @xmath0-axis differential resistivity @xmath8 at high bias shows virtually no temperature dependence , indicating that the tunneling mechanism as well as the dos away from the fermi level are temperature independent @xcite .
however , the low - bias @xmath8 reveals a pronounced upturn upon decreasing temperature below @xmath5 , demonstrating the opening of the pg near the fermi energy @xcite .
this tunneling behavior of @xmath8 is in sharp contrast to the metallic in - plane resistivity that is governed by scattering and is believed to decrease with the pg opening @xcite . since the conventional @xmath0-axis resistivity @xmath9 corresponds to @xmath8 at zero bias , @xmath9 of the bi - based cuprates measured above @xmath6 reflects the dos at the fermi energy @xcite .
this means that @xmath9 provides a convenient probe to trace the behavior of the pg , whatever its nature is .
indeed , the characteristic temperature @xmath5 below which @xmath10 shows a `` semiconducting '' upturn coincides well with the pseudogap temperatures determined by other methods @xcite .
correspondingly , the @xmath0-axis magnetoresistance ( mr ) , @xmath11 , is expected to bear information on the magnetic - field dependence of the pg , and the negative @xmath0-axis mr @xcite has been proposed to reflect a partial recovery of the dos with magnetic field .
so far , however , most of the mr studies were done on cuprates at nearly optimum doping , where @xmath5 closely approached @xmath6 , and thus it was difficult to distinguish whether the observed effect was related to the sc fluctuations @xcite or to the normal - state pg @xcite .
therefore , it is desirable to extend the mr study to the heavily - underdoped region where @xmath5 and @xmath6 are far apart . in this letter
, we report the study of @xmath9 and the @xmath0-axis mr in bi@xmath7sr@xmath12la@xmath3cuo@xmath4 ( bslco ) and bi@xmath1sr@xmath13cuo@xmath4 ( bsco ) single crystals for a wide doping range , from superconducting compositions with @xmath6 of @xmath14 to heavily - underdoped non - superconducting ones .
the evolution of @xmath10 indicates that @xmath5 increases with decreasing hole doping without being interrupted by the disappearance of sc . in heavily - underdoped samples
, the mr appears to be extremely small in a wide temperature range below @xmath5 , suggesting a magnetic - field - insensitive nature of the normal - state pg .
however , it is found at the same time that a noticeable negative mr , indicating a recovery of the pseudogapped dos , starts to show up upon approaching @xmath6 .
we discuss that the appearance of the negative mr well below @xmath5 is likely to be associated with a secondary pseudogap , which can be suppressed with magnetic fields .
the data therefore point to a possibility of two distinct mechanisms for the pg in cuprates : the first one is insensitive to the magnetic - field , gains strength with decreasing hole doping , and ultimately causes a transition into an insulating state ; the second one is sensitive to the magnetic field and is clearly related to the superconductivity .
single crystals with nominal compositions of bi@xmath7sr@xmath12la@xmath3cuo@xmath4 ( @xmath15 ) and bi@xmath16sr@xmath17cuo@xmath4 are grown by the floating - zone method @xcite .
substituting sr with la reduces the hole density in bi@xmath7sr@xmath12la@xmath3cuo@xmath4 ; optimum doping corresponds to @xmath18 ( @xmath19 ) and the system becomes non - sc for @xmath20 ( heavily underdoped ) .
the bi@xmath16sr@xmath17cuo@xmath4 crystal is nearly optimally doped ( its hole density can be estimated @xcite to be @xmath2117% per cu ) , and yet its @xmath6 ( onset at @xmath22 ) is much smaller than in slightly underdoped bslco with @xmath23 ( @xmath24 ; @xmath25 ) , whose hole density is @xmath2115% per cu @xcite .
the @xmath9 measurements are done on samples with typical sizes of @xmath26 using the ac four - probe technique . to provide a homogeneous current flow along the @xmath0-axis , two current contacts are painted to
almost completely cover the opposing @xmath27-faces of the crystal ; two voltage contacts are placed in the small windows reserved in the center of the current contacts @xcite .
the mr measurements are carried out by sweeping the magnetic field to @xmath28 at fixed temperatures stabilized by a capacitance sensor with an accuracy of @xmath29 ; this method allows reliable measurements of @xmath30 as small as @xmath31 at @xmath32 .
all the bi-2201 samples demonstrate a steep increase in @xmath10 at low temperatures ( fig .
[ fig1 ] ) , which has been shown to originate from the pg formation and corresponding decrease in the dos at the fermi energy @xcite .
one can infer that the onset of the @xmath33 upturn shifts to higher temperatures with decreasing hole doping , as has been reported for other cuprates @xcite , and that the evolution of the @xmath10 behavior is apparently uninterrupted by the disappearance of sc [ fig .
[ fig1](b ) ] . as is mentioned in the introduction , whether the pg is affected by the magnetic field or not depends crucially on its origin , and the mr data are useful in clarifying this point . for all superconducting samples
we have observed that an application of the magnetic field causes a considerable decrease in the normal - state @xmath9 in a rather wide temperature range above @xmath6 ; as an example , precise mr data for @xmath23 are shown in fig .
note that @xmath30 is strictly proportional to @xmath34 at all temperatures down to @xmath6 when @xmath35 is parallel to @xmath27 [ fig .
[ fig2](a ) ] , while the proportionality to @xmath34 is observed only at high enough temperatures when @xmath35 is parallel to @xmath0 [ fig .
[ fig2](b ) ] . in the non - sc ( @xmath36 ) sample ,
the mr turns out to be much weaker and positive at all temperatures ( fig .
[ fig3 ] ) , which gives evidence for an intimate connection between the negative @xmath0-axis mr and superconductivity .
the temperature dependences of the mr for underdoped and optimally - doped crystals are depicted in fig .
[ fig4 ] ( all the plotted data are from the region where deviation from the @xmath37 behavior is insignificant ) .
it is notable that a rapid growth of the negative mr takes place with approaching @xmath6 that decreases with decreasing hole density , rather than correlates with the upturn in @xmath10 which shifts to higher temperatures .
intriguingly , the system seems to `` know '' whether it will eventually become superconducting already at @xmath38 , where the sign of the mr appears to be determined by the low - temperature ground state .
this indicates that the mechanism which causes the negative mr sets in already at such high temperatures as @xmath38 and that this mechanism only exists when the sample becomes superconducting at low temperatures .
based on the behaviors of @xmath10 and @xmath30 , a schematic phase diagram of bslco is sketched in fig .
[ fig5 ] , where @xmath5 is defined as a temperature where @xmath10 deviates from its linear high - temperature behavior by 10% [ fig .
[ fig5](a ) ] .
the evolution of @xmath5 with increasing hole density turns out to be notably non - monotonic : a rapid change of @xmath5 in the underdoped region is separated by a kink from a much slower decrease in the overdoped region . on the other hand , the hatched area in fig .
[ fig5](b ) , which marks the region where considerable negative mr is observed , follows @xmath6 throughout the phase diagram , and apparently ignores the increase of @xmath5 at low hole densities . for example , @xmath39 of the @xmath40 sample ( for which @xmath5 is above @xmath41 ) stays less than 0.1% at @xmath42 in a wide temperature range below @xmath5 and starts to grow rapidly only below @xmath43 , while in the @xmath23 sample ( for which @xmath5 is about @xmath44 ) a rapid growth of the negative mr is observed below @xmath45 ; this trend is irrespective of the field direction ( fig .
[ fig4 ] ) .
a simple estimate shows that away from @xmath6 the magnetic - field scale for suppression of the upturn in @xmath10 ( and thus of the normal - state pg ) is extremely high : in the @xmath40 sample , for instance , @xmath9 increases by @xmath213 times upon cooling from @xmath5 to @xmath44 , indicating that by @xmath44 , @xmath212/3 of the @xmath0-axis conductivity is wiped out by the pg . at the same time , an application of @xmath46 at @xmath44 reduces @xmath9 by only @xmath210.04% ; therefore , even if the mr kept the steep @xmath37 behavior up to the highest field , it would require @xmath47 to completely suppress the pg .
it is worth noting that in yba@xmath7cu@xmath48o@xmath4 the negative @xmath0-axis mr has also been found to disappear when sc is killed by underdoping @xcite .
the above result therefore distinguishes two essentially different pseudogapped regions on the underdoped side of the phase diagram : one is marked by the semiconducting @xmath10 behavior and negligibly weak magnetic - field dependence ; the other is marked by the sc - related large negative mr .
this distinction can be understood if one assumes that there are two distinct mechanisms that separately contribute to the pg behavior .
the predominant one that sets in around @xmath5 in underdoped samples is almost insensitive to the magnetic field , gains strength with decreasing hole concentration , and smoothly extends into the non - sc region .
the second one exists only in samples that show superconductivity , causes further pseudogapping , and can be suppressed by the magnetic field . on the overdoped side of the phase diagram ,
the distinction between these two is not clear any more . the first mechanism of the pg characterized by @xmath5
can probably be associated with antiferromagnetic ( af ) @xcite or stripe @xcite correlations that develop in the cuo@xmath7 planes ; these correlations are fundamentally governed by the af interactions and gain strength as the hole density is reduced towards the af region .
note that the large exchange coupling @xmath49 ( @xmath50 ) makes af interactions ( and the pg caused thereby ) robust against the magnetic field . on the other hand ,
the second pg mechanism can most easily be understood by attributing it to the sc fluctuations , which are magnetic - field sensitive .
one may naturally expect that the pg features associated with sc fluctuations @xcite would occur in the vicinity of the sc region , exhibit a singularity at @xmath6 , and disappear in non - sc compositions
exactly the behavior that has been observed for the secondary pg .
in fact , useful information on the pg mechanisms can be obtained from the mr anisotropy shown in fig .
[ fig6 ] . at @xmath51 ,
the mr is almost isotropic : difference between the longitudinal mr ( @xmath52 for @xmath53 ) and transverse mr ( @xmath54 for @xmath55 ) is virtually zero [ fig .
[ fig6](a ) ] and thus the ratio @xmath56 is nearly one [ fig .
[ fig6](b ) ] , which suggests that the spin terms dominate at high temperatures . at lower temperatures , however , the longitudinal mr significantly exceeds the transverse mr in the sc samples ( but not in the non - sc one )
; this means that the magnetic field @xmath53 , which affects the _ in - plane _ motion of carriers , becomes more effective in suppressing the pg than @xmath57 .
this anisotropy may be readily understood if the mr originates from the suppression of two - dimensional sc fluctuations in cuo@xmath7 planes @xcite , where the `` orbital '' effects of the magnetic field are relevant only for @xmath53 .
it is intriguing that the temperature range where the large negative mr is observed appears to be notably broader than is usually expected for sc fluctuations : in bi@xmath16sr@xmath17cuo@xmath4 , it extends up to @xmath58 , though @xmath6 is only @xmath22 .
this corresponds nicely to the recent scanning tunneling spectroscopy of the pg in overdoped bsco @xcite , which also finds that the pg temperature is extremely high compared to @xmath6 .
the existence of two distinct mechanisms , which both cause the depression of the dos , might be the answer to the long - standing debates on whether the pg has a sc @xcite or non - sc @xcite origin : there appears to be both a non - sc pseudogap ( that gains strength at low doping and tends to vanish in the overdoped region @xcite ) and a sc - induced suppression in the dos ( that persists in a certain range above @xmath6 at all dopings @xcite ) .
the two mechanisms with contrasting doping and magnetic - field dependences seem to be the main source of confusion in studies of the pg in cuprates ; measurements that probe the pg at high temperatures and low doping apparently see the magnetic - field insensitive part of the pg @xcite , while those performed at low temperatures and higher doping see the sc - related part of the pg that is sensitive to the magnetic field @xcite . in summary
, we present @xmath0-axis magnetotransport data of bi-2201 that strongly suggest existence of two distinct mechanisms for the pseudogap in cuprates .
the first one is almost insensitive to the magnetic field and causes the primary pseudogap opening at @xmath5 in the underdoped region ; we discuss this may be related to the development of the magnetic correlations or stripes .
the second one is very magnetic - field sensitive , which is the source of the large negative mr , and causes a further pseudogapping as temperature is decreased towards @xmath6 .
this second mechanism is found to be present only in superconducting samples and is most strongly suppressed when the magnetic field is applied along the @xmath0-axis , which points to its relation to the superconductivity . | measurements of the @xmath0-axis resistivity and magnetoresistance have been used to investigate the pseudogap ( pg ) behavior in bi@xmath1sr@xmath2la@xmath3cuo@xmath4 ( bi-2201 ) crystals at various hole densities .
while the pg opening temperature @xmath5 increases with decreasing hole doping , the magnetic - field sensitivity of the pg is found to have a very different trend : it appears at _ lower _ temperatures in more underdoped samples and vanishes in non - superconducting samples .
these data suggest that besides the field - insensitive pseudogap emerging at @xmath5 , a distinct one is formed above @xmath6 as a precursor to superconductivity . |
it is common in astrophysics to find systems composed by an external material halo surrounding an inner region which is either void or has a massive black hole in the center .
examples of systems that are realistically described by such a model are elliptical galaxies and globular clusters@xcite . the gravitational field in the inner region due to the halo satisfies the vacuum field equations , and in newtonian gravitation it can be expanded in a sum of multipole terms of various orders .
due to the intrinsic nonlinearity of general relativity , in general this is not true for general relativistic fields .
however , in many situations the halo can be supposed to be axisymmetric ; in this case , the general solution of einstein s equations is known , and the inner relativistic field can also be expanded in a series of multipolar terms . in this article
, we study the dynamics of test particles in pure multipolar halo fields ( without any matter in the inner region ) , both in the newtonian and relativistic theories , for several multipole orders , for material particles and ( in the relativistic cases ) for light .
we find that in each case the systems have distinct well - defined _ escapes .
_ different escapes are chosen by the test particles depending on its initial conditions ; if the escape route is a fractal function of the initial conditions , the outcome of a particle starting from a given initial condition shows a sensitive dependence on these initial conditions . in this sense
, we say that the system is chaotic@xcite .
associated with this , we have a conveniently defined fractal dimension , which quantifies this sensitiveness .
fractal escapes have been found in a number of physical systems , including the henon - heiles system@xcite , astrophysical potentials@xcite , the motion of test particles in gravitational waves@xcite , classical ionization@xcite , motion of particles in a fluid@xcite , inflationary cosmology@xcite , and others .
we study the fractal dimension for different multipole orders and for different energies , in the case of material test particles . for simplicity ,
we confine ourselves to the case of zero angular momentum , when the motion is planar .
this article is organized as follows : in section 2 , we investigate the classical multipole gravitational field , showing the fractality of the escape basins and calculating the corresponding dimension for several multipole orders . in section 3
, we turn to the relativistic case , and investigate the dynamics of both material particles and massless particles ; we compare the results with those obtained in the classical case . in section 4 ,
we summarize our results and draw some conclusions .
the newtonian gravitational field in the inner region due to the mass in the halo satisfies the vacuum ( laplace ) field equation @xmath2 , and therefore can be uniquely expanded in a sum of multipole terms . if the field is axisymmetric , as we admit from now on , its multipole expansion can be written as : @xmath3 where @xmath4 , @xmath5 , and @xmath6 and @xmath7 are the usual cylindrical coordinates ; the @xmath8 are legendre polynomials .
if we consider the inner region to be empty of matter , @xmath9 can not have singularities there .
this means that @xmath10 for all @xmath0 in the inner region .
thus : @xmath11 where @xmath12 is a constant which can without lack of generality be chosen as zero .
the multipole expansion of @xmath9 contains only terms that become singular as @xmath13 ( with the exception of the constant term @xmath12 ) .
here we are interested in studying the dynamics of the field due to each multipole term separately .
we therefore define the @xmath0-th order multipole field @xmath14 as : @xmath15 we now consider a test particle moving in the field @xmath14 with no angular momentum in the direction of the symmetry axis @xmath7 ; in this case , the particle s motion is restricted to a plane that contains the @xmath7 axis
. a convenient set of coordinates in this plane is given by the @xmath7 axis and the extended @xmath6 axis , which reaches negative values . in order to avoid confusion
, we shall denote the extended @xmath6 axis by @xmath16 , with @xmath17 .
@xmath9 is written in terms of @xmath16 simply by replacing it for @xmath6 .
the hamiltonian of a test particle of unit mass acted upon by the field @xmath14 is : @xmath18 the corresponding equations of motion are @xmath19 , @xmath20 .
since @xmath14 is a homogeneous potential , satisfying @xmath21 , these equations are invariant with respect to the similarity transformations @xmath22 , @xmath23 , @xmath24 , and @xmath25 , where @xmath26 is an arbitrary scale factor .
thus , to each orbit in the @xmath27 potential there corresponds a similar orbit in the transformed potential @xmath28 , as long as @xmath26 is positive .
this means that the dynamics for all positive values of @xmath29 is the same except for scaling ; we have to consider only one value of @xmath29 for each multipole potential @xmath14 when we study its dynamical properties .
we could use this scaling property of the hamiltonian to make @xmath30 in eq .
( [ psinclass ] ) , but we prefer to keep the constant arbitrary because of numerical convenience . the region in the @xmath31-plane accessible by a particle moving in the potential @xmath14 with energy @xmath29 is bounded by the equipotential curves @xmath32 ; the particle must satisfy @xmath33 . this inequality can only be satisfied for a positive energy if @xmath14 is positive . from equation ( [ psinclass ] )
, it follows that the directions @xmath34 satisfying @xmath35 are accessible for all positive energies , whereas directions satisfying @xmath36 , if followed far enough , reach values of @xmath37 which violate the inequality @xmath33 .
therefore , the directions with negative values of @xmath38 define distinct _ escapes _ of the potential @xmath14 ; these escapes are separated by inaccessible regions lying at angles with @xmath36 . from the properties of legendre polynomials , it follows that the potential @xmath14 has @xmath0 distinct escapes . as an example , fig .
1 shows equipotential curves for some energies for the octopole field @xmath39 the curves for different energies have the same shapes , which is a direct consequence of the potential s homogeneity .
the three openings in the equipotential curves are seen clearly . to each escape
we define the corresponding _ basin _ , which is the set of initial conditions in the phase space that lead to that escape .
if the escape chosen by a particle is a fractal function of its initial condition , then the basins of the various escapes are mixed in a complex fashion , down to arbitrarily small scales , and the final state of the particle ( i.e. which escape it chooses ) can be extremely sensitive to its initial state .
this happens when the basin boundary is fractal .
we then say that the system presents chaos .
this is the appropriate definition of chaos to open systems , whose motions are not confined to a limited volume of the phase space , and this is the definition used throughout this article .
the directions @xmath40 which are minima of @xmath38 correspond to valleys in the potential @xmath14 ; the escape regions are centered around these directions .
the walls on either side of these valleys became increasingly steeper as we go to @xmath13 .
this makes the orbits of escaping particles become focused in beams around these directions as they move towards @xmath13 .
this phenomenon can be seen in fig .
2a , which shows some numerically integrated orbits all beginning with initial conditions on @xmath41 , and varying initial velocity direction angle @xmath42 , defined by : @xmath43 where @xmath44 and @xmath45 are the initial velocities in the @xmath16 and @xmath7 directions , and the velocity modulus @xmath46 is fixed by the conservation of energy : @xmath47 since @xmath48 .
the fractal nature of the basin boundary can be observed in a plot of the escape angle @xmath40 as a function on the initial velocity direction angle .
this is shown in fig .
we had to choose a `` cutoff distance '' @xmath49 such that the integration stops when @xmath37 reaches @xmath49 . in the case of fig.3
, we chose @xmath50 .
the discreteness of the values assumed by the escape angle is clearly seen in fig .
3 ; this is an illustration of the focusing effect .
the magnification of a tiny region of fig .
3a shown in fig . 3b gives us a convincing proof that the basin boundary is fractal .
an example of the sensitivity to initial conditions implied by the fractality of the basin boundary is given in fig .
2b , where three orbits with @xmath42 differing by less that one thousandth of a radian choose different escapes .
we see that initially the orbits are indistinguishable , and after a time they separate rather abruptly and go to their distinct escapes .
figures 2 and 3 were plotted with @xmath51 and @xmath52 , but as we explained above , the result is exactly the same regardless of the parameters values .
we note that we have found no bounded orbits for any of the @xmath14 potentials , which might have been expected from the shapes of the equipotentials .
the reason why the escape basins are fractally mixed is that the particle can bounce many times off the potential `` walls '' before escaping .
this means that associated with the fractality in the basin boundary we have also fractality in the escape time . to see that this is so
, we plot the time it takes for a particle to escape ( without discriminating among the various escapes ) as a function of its initial velocity angle @xmath42 , for a fixed initial position @xmath53 , @xmath54 .
figure 4 shows the results for the octopole potential , with @xmath55 .
the magnification shown in fig .
4b shows clearly the fractal nature of the escape time .
notice that fractal and non - fractal areas are mixed for all scales .
the fractal nature of the escape time is a result of the existence of a fractal set of singularities in the escape time function , where the escape time diverges .
this set of singularities results from the intersection between the set of initial conditions ( a topological circle in this case ) and the set in the phase space composed of orbits that never escape for @xmath56 , which in its turn is the stable manifold of the so - called _ invariant set _ , which is also fractal in our case and is composed of orbits that never escape for both @xmath56 and @xmath57 .
the sensitivity to initial conditions implied by the fractal nature of the basin boundary is made precise with the introduction of the _ box - counting dimension @xmath58 _ , calculated in the following way . for a given initial position ( which is @xmath59 in our case ) , we pick a large number of random initial velocity angles @xmath42 , with @xmath60 . for each of these
we integrate the equations of motion and see to which basin it belongs .
we then take two neighboring angles @xmath61 and @xmath62 ( with the same initial position ) , with @xmath63 being a small number , and find to which basin they belong .
if all these three initial conditions do not belong to the same basin ( if they lead to different escapes ) , then @xmath42 is said to be an `` uncertain '' angle .
the number @xmath0 of uncertain angles found in a sample of @xmath64 randomly chosen points is a function of the displacement @xmath63 , and for large @xmath64 the fraction @xmath65 of uncertain points scales as @xmath66 where @xmath58 is by definition the box - counting dimension of the intersection of the basin boundary with the one - dimensional manifold in which we chose the initial conditions .
we numerically calculate @xmath67 for several values of @xmath63 , and @xmath58 is directly obtained from the plot of @xmath68 versus @xmath69 , which should be a straight line , and whose inclination @xmath70 is related to @xmath58 by @xmath71 .
we have calculated in this way the box - counting dimension for multipole fields with orders from 2 to 9 .
since the dynamics is the same for every value of the energy , @xmath58 does not depend on @xmath29 , and is unique for each multipole order .
figure 5 shows the plot of @xmath68 versus @xmath69 we have obtained for the octopole field ( @xmath72 ) .
the data fit quite well into a straight line , whose inclination ( @xmath73 ) gives a dimension @xmath74 , confirming the fractal nature of the basin boundary .
figure 6 shows @xmath58 as a function of the multipole order @xmath0 .
we see that @xmath58 increases monotonically with @xmath0 , which means that the fractality of the basin boundary increases with @xmath0 , and thus the dynamics becomes `` more chaotic '' with multipolar fields of higher orders .
the dipole and quadrupole fields ( @xmath75 and @xmath76 , respectively ) have regular basin boundaries , and show no chaos in their dynamics .
in fact , the potentials @xmath77 and @xmath78 are separated using the @xmath79 coordinates , as can be immediately seen from ( [ psinclass ] ) , which implies that the hamiltonian is integrable and the basin boundaries are regular .
a field of the general form ( [ classexp ] ) , with the contribution of more than one multipole order , is not homogeneous , and its dynamics changes with the energy .
we have studied how the box - counting dimension changes with @xmath29 for the field @xmath80 , with equal contributions of the 3rd and 4th multipole orders . for high values of @xmath29 , the particle s motion is dominated by the 16-pole component of the field , and the equipotentials are practically the same as the pure 16-pole field .
accordingly , for these energies the box - counting dimension was found to be numerically indistinguishable from the value we obtained for the pure 16-pole field @xmath81 .
as the energy is lowered , however , @xmath58 does _ not _ go to its pure octopole field value , as might be imagined .
in fact , for @xmath82 , we found @xmath83 ; this value of @xmath58 is higher then that of both the pure octopole field and the quadrupole field .
thus , the fractal dimension from a field made by the mixture of two multipole orders is not necessarily an intermediate value between the dimensions of each pure multipole order separately .
in this section we investigate the dynamics with a general - relativistic generalization of the newtonian multipole fields we studied in the previous section . in general relativity , the dynamics results from the space - time metric @xmath84 , where @xmath85 is the metric tensor and @xmath86 are the space - time coordinates .
the equations of motion for a test particle in the field @xmath85 are given by the geodesic equation @xmath87 where @xmath88 is the christoffel symbol , and @xmath89 , with @xmath26 being a convenient affine parameter .
for an axially symmetric and static field , the metric is most conveniently written in the weyl form@xcite : @xmath90 , \ ] ] where @xmath6 , @xmath7 and @xmath42 are cylindrical - like coordinates , and @xmath91 is the time .
we use units such that @xmath92 throughout .
the metric functions @xmath9 and @xmath93 depend on @xmath6 and @xmath7 only , and satisfy the vacuum einstein field equations : @xmath94 @xmath95 dr+2r\frac{\partial \psi } { \partial r}\frac{\partial \psi } { \partial z}dz.\ ] ] eq .
( [ eq1 ] ) is just laplace s equation in cylindrical coordinates , and eq .
( [ eq2 ] ) is a quadrature whose integrability is automatically guaranteed by eq .
( [ eq1 ] ) .
thus , each axisymmetric solution of newton s field equation corresponds to a general - relativistic metric ( [ weyl ] ) , with @xmath93 being found by integration of eq .
( [ eq2 ] ) with the requirement that the metric be regular in the symmetry axis . based on this ,
we define the relativistic equivalent to the multipolar fields @xmath14 in eq .
( [ psinclass ] ) to be the metric ( [ weyl ] ) with @xmath96 and @xmath97 obtained through eq .
( [ eq2 ] ) .
the integration can be performed analytically , and the result is : @xmath98 , \ ] ] with @xmath99 .
since there are many possible relativistic space - times which in the weak field limit correspond to a given newtonian field , the choice of the metric is not unique . we have chosen the metric defined by @xmath14 and @xmath100 because it is the simplest choice , and because the field obtained in this way shares important features with the newtonian field , as we shall see shortly .
as in the newtonian case , the invariance of the system with respect to time translations and axial rotations implies the existence of two constants of motion , which can be identified as the energy @xmath29 and the @xmath7-component of the angular momentum @xmath101 .
they can be found by using the lagrangian associated with the metric ( [ weyl ] ) : @xmath102 , \ ] ] where the dot denotes differentiation with respect to the affine parameter . since both @xmath9 and @xmath93
depend only on @xmath6 and @xmath7 , the lagrangian @xmath103 does not depend on @xmath91 and @xmath42 ; they are `` cyclical coordinates '' in the language of hamiltonian dynamics , and they are associated with the conserved quantities : @xmath104 @xmath105 since the lagrangian function ( [ lagr ] ) has only kinetic terms , and since @xmath103 does not depend explicitly on the affine parameter , the hamiltonian of this system is simply @xmath103 itself , and its conservation gives us yet another constant of motion : @xmath106 where @xmath107 is the test particle s rest mass . in the case of light , we have @xmath108 ; for massive particles , we can without loss of generality make @xmath109 . rewriting @xmath103 with the help of eqs .
( [ rele ] ) and ( [ rell ] ) , we have : @xmath110 with @xmath111 for particles with zero rest mass , and @xmath112 for massive particles . as in the newtonian case in the previous section
, we will restrict ourselves to motion with zero angular momentum ( @xmath113 ) , when the motion is planar . in complete analogy to what we did for the newtonian potential ,
we define a new coordinate @xmath16 , which is no more than @xmath6 extended to negative values ; the motion of our zero angular momentum test particle occurs on the @xmath31-plane .
all the expressions above are written in terms of @xmath16 simply by replacing @xmath6 by @xmath16 everywhere .
the accessible region of the @xmath31-plane is found by setting @xmath114 in ( [ hamrel ] ) ; taking into account that @xmath113 , we have : @xmath115 the accessible region for a particle with energy @xmath29 is given by @xmath116 .
for massless particles ( @xmath111 ) , @xmath117 , and this condition is always satisfied ; thus , the whole @xmath31-plane is accessible to null geodesics .
this is not so for massive particles ( @xmath112 ) ; the regions they are able to reach on the @xmath118 plane depend on their energy and are determined by @xmath9 .
we now consider the pure multipole potentials @xmath119 , with @xmath100 given by ( [ gamman ] ) .
we first treat the case of massive particles ; the motion of light will be studied later .
the effective potential is @xmath120 , where @xmath121 is the time - time component of the metric .
it is clear that the dynamics in the relativistic field @xmath122 is more complex than its newtonian counterpart .
in particular , since neither the effective potential ( [ potefrel ] ) nor @xmath100 are homogeneous functions , and furthermore the relativistic equations of motion depend on the velocities @xmath123 , the orbits in the relativistic field do not satisfy the homogeneity property that guarantees for the orbits in the newtonian field that the dynamics is the same for all energies , up to a global scaling .
we thus expect the dynamics to change with the energy in the relativistic case ; this is a fundamental difference between the two theories . in spite of this , we see from eq .
( [ potefrel ] ) that the curves of constant effective potential are still given by @xmath124 ( although the constant is no longer just the energy ) .
this means that the escapes in the potentials @xmath14 are the same as those in the newtonian fields .
as @xmath13 along the asymptotically allowed directions , given by @xmath35 ( exactly as in the newtonian case ) , we have @xmath125 , that is , we have an infinite red shift ( as in section 2 , we assume @xmath126 ) : particles are attracted to these regions . on the other hand , for asymptotically forbidden regions , with @xmath35 , we have @xmath127 , an infinite blue shift : particles are attracted _ away _ from these regions .
we now study the dynamics in this field using the same techniques we used in the newtonian case .
we integrate numerically orbits with initial position @xmath55 and variable velocity angle @xmath42 , with the magnitude @xmath46 of the velocity being fixed by the conservation of energy . since @xmath128 for all @xmath0 , from eq .
( [ hamrel ] ) we see that : @xmath129 with @xmath130 the equations of motion for the independent variables @xmath16 and @xmath7 are obtained directly from the geodesci equations : @xmath131 ; \ ] ] @xmath132 , \ ] ] with @xmath133 and @xmath134 ; these equations are valid for @xmath113 .
together with the quadrature ( [ rele ] ) , the above equations give the complete dynamics of the system .
similar to what we found in the previous section , the orbits are focused along directions corresponding to the minima of @xmath38 .
plots of the escape angle for @xmath1 are very similar to fig .
3 , showing clearly that the escape has a fractal dependency on the initial conditions . as in the newtonian case ,
the dipole and quadrupole fields ( @xmath75 and 2 , respectively ) show a regular basin boundary .
however , the equations of motion are not separated in @xmath79 coordinates , as is the case for the newtonian fields . as we explained above ,
we expect the fractal basin boundary present in the @xmath1 fields to change with the energy in the relativistic case , as opposed to the newtonian one . as in the newtonian case
, we have found no bounded orbits , neither for massive particles nor for massless ones .
we calculate the fractal dimension of this set in the same way we did previously for the motion in the newtonian field . in that case
the dimension did not depend on the energy , because the dynamics is the same for all ( positive ) energies .
this is not true for the relativistic dynamics , and we expect the dimension to change with the energy . this is indeed what we have found .
as an example , we take the field @xmath135 .
we consider particles with energies higher than 1 , corresponding to positive energies in the newtonian case . for @xmath29 close to 1 ( corresponding to energies close to 0 in the newtonian field ) , we find that the box - counting dimension is equal to its newtonian value .
for instance , for @xmath136 , we find @xmath137 , which is ( within the error ) the same value we found for the newtonian field @xmath81 . for higher energies
, however , @xmath58 changes with @xmath29 : for @xmath138 , we have @xmath139 , and for @xmath140 , @xmath141 .
we have calculated @xmath58 for more values of @xmath29 than we show here , and it seems that @xmath58 decreases monotonically with @xmath29 , and the system becomes more chaotic for large energies ; this is the contrary to what is usually the case in bounded dynamics , where commonly the chaos grows with the energy .
the same behavior is observed in the octopole field ( @xmath72 ) , but in this case the variations in the dimension are smaller . because of the complexity of the equations of motion for high multipole orders , we have not been able to calculate @xmath58 with enough accuracy for @xmath0 larger than 4 , but we think this behavior of @xmath58 might be a general feature of the dynamics for the @xmath14 fields . in the case of null geodesics ,
the equations of motion are again given by eqs .
( [ eqmot1 ] ) and ( [ eqmot2 ] ) , but the initial conditions are required to satisfy the constraint ( [ hamrel ] ) with @xmath111 ( and @xmath113 in our case ) .
this allows us to scale the affine parameter so as to eliminate @xmath29 from the equations .
this means that the planar motion of light depends only on the chosen field @xmath14 , and not on any other parameter , unlike the motion of massive particles . even though the whole @xmath31-plane is accessible to massless particles , as we have seen
, null geodesics also show focusing effects similar to that shown by massive particles .
an example is given in fig .
7 , where the escape angle is plotted as a function of the initial velocity angle for null geodesics , for the octopole field @xmath72 .
we see clearly that the values assumed by the escape angle are indeed discrete .
we do not see , however , any region where the escape angle shows rapid variations , that is typical of fractal functions .
magnifications of several regions of fig .
7 show no small - scale structures , indicating that the escape angle is not fractal .
this is confirmed by direct calculation of the box - counting dimension , which gives @xmath142 to within statistical accuracy .
this result also holds for @xmath143 and @xmath144 , which is the highest multipole order for which we have been able to obtain reliable results .
we speculate that this result might be true for all @xmath0 .
we have studied the dynamics of test particles in pure multipole fields generated by external halos of matter , when the particles have zero angular momentum and the motion happens on a plane .
such a system could be used to approximate observed astronomical fields , such as the interior of galaxies .
we have considered both the newtonian and the general - relativistic fields , and we have compared the results of the two theories . for the newtonian fields
, the homogeneity of the pure potentials @xmath14 implies that the dynamics is independent on the energy .
we have found that the escape chosen by the test particles is a fractal function of the initial conditions for all pure multipole fields of order higher than 3 . for @xmath145 ,
the hamiltonian is separable and the basin boundaries are regular .
we have calculated numerically the box - counting dimension @xmath58 associated with the escape function . since the dynamics is the same for all energies , @xmath58 depends only on the multipole order @xmath0 .
we have found that @xmath58 is an increasing function of @xmath0 for @xmath0 up to 9 ; higher values of @xmath0 could not be studied because of numerical limitations .
the increasing in the intensity of chaos with @xmath0 can be understood by the fact that fields of higher order have a higher number of escapes .
we have also studied the dynamics of a field made by the superposition of two different multipole orders , specifically , the field @xmath80 . for this field ,
the dynamics depends on the energy , and we found that , although for high energies the box - counting dimension tends to the value of the @xmath81 field , for low energies @xmath58 assumes values higher than the dimensions of both @xmath146 and @xmath81 , maybe somewhat surprisingly .
we now discuss the results found for the relativistic fields , starting by the dynamics of massive particles .
the relativistic pure multipole fields have not the homogeneous property shown by their newtonian counterparts , and their dynamics does depend on the energy , and so does the escape function .
we found that the escape chosen by the particle is a fractal function of the initial conditions for fields of order higher than 2 also in the relativistic case .
the dipole and quadrupole fields have regular basin boundaries , but unlike the newtonian fields , the relativistic equations of motion are not directly separable . for @xmath1
, we expect the box - counting dimension @xmath58 to change with the energy .
we have calculated @xmath58 as a function of the energy , and we have found that for low energies , @xmath58 goes to the newtonian values , as might have been expected . as @xmath29 increases , @xmath58 was found to decrease monotonically , for all the energies for which we have been able to obtain reliable results .
we have obtained these results for @xmath147 and 5 , and this seems to be a general feature of the dynamics of these fields .
for massless test particles , the escape function was found to be regular ( non - fractal ) for @xmath0 up to 5 ; at least as far as the fifth multipole order , null geodesics are not chaotic .
we would like to thank fapesp and cnpq for their financial support .
j. binney and s. tremaine , _ galactic dynamics _ , princeton university press , princeton , new jersey ( 1994 ) c. grebogi , s. w. mcdonald , e. ott and j. a. yorke , phys .
lett . * 99a * , 415 ( 1983 ) s. bleher , c. grebogi , e. ott and r. brown , phys .
a * 38 * , 930 ( 1988 ) e. ott , _ chaos in dynamical systems _ , cambridge university press , cambridge ( 1993 ) a. p. s. de moura and p. letelier , phys .
a * 256 * , 362 ( 1999 ) g. contopoulos , d. kaufmann , _ astron .
astrophys . _ * 253 * , 379 ( 1992 ) ; g. contopoulos , c. polymilis , _ phys .
e _ * 46 * , 1546 ( 1993 ) ; g. contopoulos , h. e. kandrup , d. kaufmann , _ physica d _ * 64 * , 310 ( 1993 ) j. podolsky and k. vasely , phys . rev .
d * 58 * , 081501 ( 1998 ) f. sattin and l. salasnich , phys .
e * 59 * , 1246 ( 1999 ) i. m. jnosi , t. tl , d. e. wolf and j. a. c. gallas , phys .
e * 56 * , 2858 ( 1997 ) n. j. cornish and j. j. levin , phys .
d * 53 * , 3022 ( 1996 ) d. kramer , h. stephani and e. herlt , _ exact solutions of einstein s field equations _
, cambridge university press , cambridge ( 1980 )
fig . 1 : : equipotential curves for the newtonian octopole field @xmath146 .
from the inside out , the energies are @xmath148 , @xmath149 and @xmath150 .
fig . 2 : : orbits in the newtonian field @xmath146 all starting from @xmath41 , with different initial directions and @xmath149 , @xmath52 .
2b shows three orbits with initial velocity angles differing by 1/5000 radian ; the circle marks the starting point @xmath41 .
fig . 3 : : escape angle @xmath40 as a function of the initial velocity angle @xmath42 for the octopole field .
3b is a magnification of a small region of fig .
fig . 4 : : escape time as a function of the initial velocity angle @xmath42 for the octopole field .
4b is a magnification of a small region of fig .
4a . fig . 5 : : @xmath68 vs. @xmath69 for the newtonian octopole field .
the straight line is the result of the fitting .
fig . 6 : : box - counting dimension @xmath58 for newtonian multipole field of order @xmath0 .
the error bars come from the statistical errors in the calculation of @xmath58 .
fig . 7 : : escape angle @xmath40 as a function of the initial velocity angle @xmath42 for the null geodesics in the relativistic octopole field . | we study the planar motion of test particles in gravitational fields produced by an external material halo , of the type found in many astrophysical systems , such as elliptical galaxies and globular clusters .
both the newtonian and the general - relativistic dynamics are examined , and in the relativistic case the dynamics of both massive and massless particles are investigated .
the halo field is given in general by a multipole expansion ; we restrict ourselves to multipole fields of pure order , whose newtonian potentials are homogeneous polynomials in cartesian coordinates .
a pure @xmath0-pole field has @xmath0 different escapes , one of which is chosen by the particle according to its initial conditions .
we find that the escape has a fractal dependency on the initial conditions for @xmath1 both in the newtonian and the relativistic cases for massive test particles , but with important differences between them .
the relativistic motion of massless particles , however , was found to be regular for all the fields we could study .
the box - counting dimension was used in each case to quantify the sensitivity to initial conditions which arises from the fractality of the escape route . |
massive multiple - input multiple - output ( mimo ) systems employing simple linear precoding and combining schemes offer significant performance gains in terms of bandwidth , power , and energy efficiency compared to conventional multiuser mimo systems as impairments such as fading , noise , and interference are averaged out for very large numbers of base station ( bs ) antennas @xcite .
furthermore , in time - division duplex ( tdd ) systems , channel reciprocity can be exploited to estimate the downlink channels via uplink training so that the training overhead scales only linearly with the number of users and is independent of the number of bs antennas @xcite .
however , if the pilot sequences employed in different cells are not orthogonal , so - called pilot contamination occurs and impairs the channel estimates , which ultimately limits the achievable performance of massive mimo systems @xcite . since secrecy and privacy are critical concerns for the design of future communication systems @xcite , it is of interest to investigate how the large number of spatial degrees of freedom in massive mimo systems can be exploited for secrecy enhancement @xcite .
if the eavesdropper ( eve ) remains passive to hide its existence , neither the transmitter ( alice ) nor the legitimate receiver ( bob ) will be able to learn eve s channel state information ( csi ) .
in this situation , it is advantageous to inject artificial noise ( an ) at the transmitter to degrade eve s channel and to use linear precoding to avoid impairment to bob s channel as was shown in @xcite-@xcite and @xcite , @xcite for single user and single - cell multiuser systems , respectively .
however , in multi - cell massive mimo systems , multi - cell interference and pilot contamination will hamper alice s ability to degrade eve s channel and to protect bob s channel .
this problem was studied first in @xcite for simple matched - filter ( mf ) data precoding and null - space ( ns ) and random an precoding .
however , it is well known that mf data precoding suffers from a large loss in the achievable information rate compared to other linear data precoders such as zero - forcing ( zf ) and regularized channel inversion ( rci ) precoders as the number of mobile terminals ( mts ) increases @xcite . since it is expected that this loss in information rate also translates into a loss in secrecy rate , studying the secrecy performance of zf and rci data precoders in massive mimo systems
is of interest .
furthermore , while ns an precoding was shown to achieve a better performance compared to random an precoding @xcite , it also entails a much higher complexity .
similarly , the improved performance of zf and rci data precoding compared to mf data precoding comes at the expense of a higher complexity .
hence , the design of novel data and an precoders which allow a flexible tradeoff between complexity and secrecy performance is desirable .
related work on physical layer security in massive mimo systems includes @xcite where the authors use the channel between alice and bob as secrete key and show that the complexity required by eve to decode alice s message is at least of the same order as a worst - case lattice problem .
physical layer security in a downlink multi - cell mimo system was considered in @xcite-@xcite . however , unlike our work , perfect knowledge of eve s channel was assumed , an injection was not considered , and pilot contamination was not taken into account . furthermore , zf and rci data precoding were analyzed in the large system limit in @xcite . however , neither pilot contamination nor an were taken into account and the secrecy rate was not analyzed . using a concept that was originally conceived for code division multiple access ( cdma ) uplink systems in @xcite and later extended to mimo systems in @xcite , reduced complexity
linear data precoders that are based on matrix polynomials were investigated for use in massive mimo systems in @xcite-@xcite .
however , @xcite-@xcite did not take into account the effect of an leakage for precoder design and did not study the secrecy performance .
hence , the results presented in @xcite-@xcite are not directly applicable to the system studied in this paper . in this paper
, we consider secure downlink transmission in a multi - cell massive mimo system employing linear data and an precoding in the presence of a passive multi - antenna eavesdropper .
we study the achievable ergodic secrecy rate of such systems for different linear precoding schemes taking into account the effects of uplink channel estimation , pilot contamination , multi - cell interference , and path - loss .
the main contributions of this paper are summarized as follows : * we study the performance - complexity tradeoff of selfish and collaborative data and an precoders .
selfish precoders require only the csi of the mts in the local cell but cause inter - cell interference and inter - cell an leakage .
in contrast , collaborative precoders require the csi between the local bs and the mts in all cells , but reduce inter - cell interference and inter - cell an leakage
. however , since the additional csi required for the collaborative precoders can be estimated directly by the local bs , the additional overhead and complexity incurred compared to selfish precoders is limited .
* we derive novel closed - form expressions for the asymptotic ergodic secrecy rate which facilitate the performance comparison of different combinations of linear data precoders ( i.e. , mf , selfish and collaborative zf / rci ) and an precoders ( i.e. , random , selfish and collaborative ns ) , and provide significant insight for system design and optimization . * in order to avoid the computational complexity and potential stability issues in fixed point implementations entailed by the large - scale matrix inversions required for zf and rci data precoding and ns an precoding , we propose polynomial ( poly ) data and an precoders and optimize their coefficients . unlike @xcite and @xcite , which considered polynomial data precoders for massive mimo systems without an generation , we use free probability theory @xcite to obtain the poly coefficients .
this allows us to express the poly coefficients as simple functions of the channel and system parameters .
simulation results reveal that these precoders are able to closely approach the performance of selfish rci data and ns an precoders , respectively .
the remainder of this paper is organized as follows . in section [ s2 ] , we outline the considered system model and review some basic results from @xcite . in sections [ s3 ] and [ s4 ] , the considered linear data and an precoders are investigated , respectively . in section [ s5 ] , the ergodic secrecy rates of different linear precoders are compared analytically for a simple path - loss model . simulation and numerical results are presented in section [ s6 ] , and some conclusions are drawn in section [ s7 ] .
_ notation : _ superscripts
@xmath0 and @xmath1 stand for the transpose and conjugate transpose , respectively .
@xmath2 is the @xmath3-dimensional identity matrix .
the expectation operation and the variance of a random variable are denoted by @xmath4 $ ] and @xmath5 $ ] , respectively .
@xmath6 denotes a diagonal matrix with the elements of vector @xmath7 on the main diagonal . @xmath8 and @xmath9 denote trace and rank of a matrix , respectively .
@xmath10 represents the space of all @xmath11 matrices with complex - valued elements .
@xmath12 denotes a circularly symmetric complex gaussian vector @xmath13 with zero mean and covariance matrix @xmath14 .
@xmath15_{kl}$ ] denotes the element in the @xmath16 row and @xmath17 column of matrix @xmath18 , and @xmath19^+=\max\{x,0\}$ ] .
in this section , we introduce the considered system model as well as the adopted channel estimation scheme , and review some ergodic secrecy rate results .
we consider the downlink of a multi - cell massive mimo system with @xmath20 cells and a frequency reuse factor of one , i.e. , all bss use the same spectrum .
each cell includes one @xmath21-antenna bs , @xmath22 single - antenna mts , and potentially an @xmath23-antenna eavesdropper .
the eavesdroppers try to hide their existence and hence remain passive . as a result
, the bss can not estimate the eavesdroppers csi . to overcome this limitation
, each bs generates an to mask its information - carrying signal and to prevent eavesdropping @xcite . in the following ,
the @xmath16 mt , @xmath24 , in the @xmath25 cell , @xmath26 , is the mt of interest and we assume that an eavesdropper tries to decode the signal intended for this mt . we note that neither the bss nor the mts are assumed to know which mt is targeted by the eavesdropper .
the signal vector , @xmath27 , transmitted by the bs in the @xmath25 cell ( also referred to as the @xmath25 bs in the following ) is given by @xmath28 where @xmath29 and @xmath30 denote the data and an vectors for the @xmath31 mts in the @xmath25 cell , respectively .
@xmath32 \in \mathbb{c}^{n_t \times k}$ ] and @xmath33\in \mathbb{c}^{n_t \times n_t}$ ] are the data and an precoding matrices , respectively , and the efficient design of these matrices is the main scope of this paper .
thereby , the structure of both types of precoding matrices does not depend on which mt is targeted by the eavesdropper .
the an precoding matrix @xmath34 has rank @xmath35 , i.e. , @xmath36 dimensions of the @xmath21-dimensional signal space spanned by the @xmath21 bs antennas are exploited for jamming of the eavesdropper .
the data and an precoding matrices are normalized as @xmath37 and @xmath38 , i.e. , their average power per dimension is one .
the average powers @xmath39 and @xmath40 allocated to the information - carrying signal for each mt and each an signal , respectively , can be written as @xmath41 and @xmath42 , where @xmath43 is the total transmit power and @xmath44 $ ] is a power allocation factor which can be optimized .
for the sake of clarity , in this paper , we assume that all cells utilize the same value of @xmath45 .
the vectors collecting the received signals at the @xmath31 mts and the @xmath23 antennas of the eavesdropper in the @xmath25 cell are given by @xmath46 respectively , with gaussian noise vectors @xmath47 and @xmath48 , where @xmath49 and @xmath50 denote the noise variances at one mt and one eavesdropper receive antenna , respectively .
furthermore , @xmath51 and @xmath52 are the matrices modeling the channels from the @xmath53 bs to the @xmath31 mts and the eavesdropper in the @xmath25 cell , respectively .
thereby , @xmath54 and @xmath55 represent the path - losses from the @xmath53 bs to the @xmath31 mts and the eavesdropper in the @xmath25 cell , respectively .
matrix @xmath56 , with row vector @xmath57 in the @xmath16 row , and matrix @xmath58 represent the corresponding small - scale fading components .
their elements are modeled as mutually independent and identically distributed ( i.i.d . )
complex gaussian random variables ( rvs ) with zero mean and unit variance . for the design of the data and noise precoders , we consider two different approaches : _ selfish _ designs and _ collaborative _ designs . for the selfish designs ,
each bs designs its precoders only based on the estimate of the csi in its own cell , @xmath59 , and without regard for the interference and the an it causes to other cells . in contrast , for the collaborative designs , each bs designs its precoders based on the estimates of the csi to the mts in all cells , @xmath60 , @xmath61 , in an effort to avoid excessive interference and an to other cells .
although collaborative designs introduce more channel estimation overhead at the bs , they may not always outperform selfish designs because of the imperfection of the csi and the limited number of spatial degrees of freedom available for precoder design . as is customary for massive mimo systems , we assume that the downlink and uplink channels are reciprocal and the csi is estimated in an uplink training phase @xcite-@xcite . to this end , all mts emit pilot sequences of length @xmath62 and with pilot symbol power @xmath63 .
we assume that the pilot sequences of the @xmath31 mts in a given cell are mutually orthogonal but the same pilot sequences are used in all cells .
this gives rise to so - called pilot contamination @xcite-@xcite .
furthermore , we assume that the path - loss information changes on a much slower time scale than the small - scale fading .
hence , the path - loss matrices @xmath64 , @xmath65 , can be estimated perfectly and are assumed to be known at the bs for minimum mean - square error ( mmse ) estimation of the small - scale fading gains @xcite . at the @xmath25 bs
, the small - scale fading vector to the @xmath16 mt in the @xmath53 cell , @xmath66 , can be expressed as @xmath67 where the estimate @xmath68 and the estimation error @xmath69 are mutually independent and can be statistically characterized as @xmath70 and @xmath71 , respectively , cf .
@xcite . for future reference
, we collect the estimates and the estimation errors at the @xmath25 bs corresponding to all @xmath31 mts in the @xmath53 cell in matrices
@xmath72^t \in \mathbb{c}^{k \times n_t}$ ] and @xmath73^t \in \mathbb{c}^{k \times n_t}$ ] , respectively .
the performance metric adopted in this paper is the ergodic secrecy rate @xcite . in this section
, we review some results for the ergodic secrecy rate in multi - cell massive mimo systems employing linear data and an precoding from @xcite , as these results will be needed throughout this paper . combining ( [ eq1 ] ) and ( [ eq2 ] ) we observe that the downlink channel comprising the bs , the @xmath16 mt , and the eavesdropper in the @xmath25 cell is an instance of a multiple - input , single - output , multi - eavesdropper ( misome ) wiretap channel @xcite .
hence , the achievable secrecy rate of the @xmath16 mt in the @xmath25 cell is bounded by the difference of the capacities of the channel between the bs and the mt and the channel between the bs and the eavesdropper , see ( * ? ? ?
* lemma 1 ) , ( * ? ? ?
* lemma 2 ) .
thus , a lower bound on the ergodic secrecy rate of the @xmath16 mt in the @xmath25 cell is given by @xcite @xmath74^+ , { \color{blue } k=1,\ldots , k,}\ ] ] where @xmath75 denotes an achievable rate of the @xmath16 mt in the @xmath25 cell and @xmath76 denotes the ergodic capacity of the channel between the bs and the eavesdropper . in order to obtain a tractable lower bound on the ergodic secrecy rate ,
we lower bound the achievable rate of the mt as @xmath77 with signal - to - interference - and - noise ratio ( sinr ) ( * ? ? ?
* eq . ( 10 ) ) @xmath78|^2}{{{\rm var}[\sqrt{\beta^k_{nn } p } { \bf h}^k_{nn } { \bf f}_{nk}]}+\sum\limits^m_{m=1 } \sum\limits_{i=1}^{n_t } \mathbb{e}[|\sqrt{\beta^k_{mn } q}{\bf h}^k_{mn } { \bf a}_{mi}|^2]+ \sum\limits_{\{m , l\ } \neq \{n , k\ } } \mathbb{e}[|\sqrt { \beta^k_{mn } p } { \bf h}^k_{mn } { \bf f}_{ml}|^2]+1}.\ ] ] furthermore , we make the pessimistic assumption that the eavesdropper is able to cancel the received signals of all in - cell and out - of - cell mts except the signal intended for the mt of interest .
this leads to an upper bound for the eavesdropper s capacity , and consequently , to a lower bound for the ergodic secrecy rate .
hence , the ergodic capacity of the eavesdropper is given by ( * ? ? ?
* eq . ( 7 ) )
@xmath79,\ ] ] where @xmath80 denotes the noise correlation matrix at the eavesdropper under the worst - case assumption that the receiver noise at the eavesdropper is negligible , i.e. , @xmath81 .
denoting the normalized number of eavesdropper antennas by @xmath82 , a necessary condition for the invertibility of matrix @xmath83 is @xmath84 .
hence , a non - zero secrecy rate can only be achieved if this condition is met .
consequently , a larger @xmath36 implies that the bs is able to tolerate more eavesdropper antennas .
if @xmath85 and matrix @xmath83 are statistically independent , which in turn means for the data and an precoders that vector @xmath86 and the subspace spanned by the columns of @xmath34 are mutually orthogonal , a simple and tight upper bound on ( [ ceve ] ) can be obtained . since any efficient data / an precoder pair has to keep the an self - interference at the desired mt small
, this orthogonality condition holds at least approximately in practice . in this case , for @xmath87 and @xmath88 , where @xmath89 and @xmath90 , a simple and tight upper bound for @xmath76 is given by ( * ? ? ?
* theorem 1 ) @xmath91 for @xmath92 , we have @xmath93 , i.e. , the bound in ( [ cup ] ) is applicable in the entire range of @xmath94 where @xmath76 in ( [ ceve ] ) is finite . for @xmath95 , we have @xmath96 ,
i.e. , the bound is not applicable for @xmath97 .
however , for strong inter - cell interference , we have @xmath98 and @xmath99 , i.e. , the bound is applicable for all @xmath94 for which @xmath76 in ( [ ceve ] ) is finite .
on the other hand , for weak inter - cell interference , we have @xmath100 , and matrix @xmath83 will be ill - conditioned for @xmath101 and @xmath76 will become very large . hence , the bound is again applicable for the values of @xmath94 ( i.e. , @xmath102 ) , for which @xmath76 in ( [ ceve ] ) assumes practically relevant values . more generally , ( * ? ? ? * figs
. 2 - 4 ) and section [ s6 ] suggest that , for @xmath103 , ( [ cup ] ) is applicable and tight for all values of @xmath94 which permit a non - vanishing secrecy rate . combining ( [ secnk ] ) , ( [ rnk ] ) , and ( [ cup ] ) ,
we obtain a tight and tractable lower bound on the secrecy rate @xcite .
it is noteworthy that the upper bound on the capacity of the eavesdropper in ( [ cup ] ) is only affected by the dimensionality of the an precoder , @xmath36 , but not by the exact structures of @xmath34 and @xmath104 , as long as @xmath86 and the subspace spanned by the columns of @xmath34 are orthogonal . on the other hand ,
the achievable rate of the mt in ( [ rnk ] ) is affected by both the data and the an precoders . in the following two sections ,
we analyze the impact of the most important existing data and an precoder designs on the achievable rate @xmath75 as @xmath103 , respectively , and propose novel low - complexity data and an precoders that are based on a polynomial matrix expansion .
in this section , we analyze the achievable rate of selfish and collaborative zf / rci data precoding , respectively , and develop a novel poly data precoder . in contrast to existing analyses and designs of data precoders for massive mimo , e.g. @xcite , @xcite-@xcite , the results presented in this section account for the effect of an leakage , which is only present if an is injected at the bs for secrecy enhancement .
we are interested in the asymptotic regime where @xmath105 but @xmath106 and @xmath82 are finite
. for @xmath88 , analyzing the achievable rate is equivalent to analyzing the sinr in ( [ rnk ] ) .
thereby , the effect of the an precoder can be captured by the term @xmath107= \sum^m_{m=1 } \beta^k_{mn } \mathbb{e}[{\bf h}^k_{mn } { \bf a}_{m } { \bf a}^h_{m } ( { \bf h}^k_{mn})^h ] \label{eq8}\ ] ] in the denominator of ( [ rnk ] ) , which represents the inter - cell and intra - cell an leakage .
this term is assumed to be given in this section and will be analyzed in detail for different an precoders in section [ s4 ] .
the selfish rci ( srci ) data precoder for the @xmath25 cell is given by @xmath108 where @xmath109 , @xmath110 is a scalar normalization constant , and @xmath111 is a regularization constant . in the following proposition ,
we provide the resulting sinr of the @xmath16 mt in the @xmath25 cell . _
proposition 1 _ : for srci data precoding , the received sinr at the @xmath16 mt in the @xmath25 cell is given by @xmath112 where @xmath113,\ ] ] and @xmath114 with @xmath115 , @xmath116 , @xmath117 @xmath118 , and @xmath119 .
please refer to appendix a. regularization constant @xmath111 can be optimized for maximization of the lower bound on the secrecy rate in ( [ secnk ] ) , which is equivalent to maximizing the sinr in ( [ gammalmmsenpc ] ) . setting the derivative of @xmath120 with respect to @xmath111 to zero
, the optimal regularization parameter is found as @xmath121 , and the corresponding maximum sinr is given by @xmath122 on the other hand , for @xmath123 , the srci data precoder in ( [ srci ] ) reduces to the selfish zf ( szf ) data precoder .
the corresponding received sinr is provided in the following corollary .
_ corollary 1 _ : assuming @xmath124 , for szf data precoding , the received sinr at the @xmath16 mt in the @xmath25 cell is given by @xmath125 @xmath126 in ( [ szf ] ) can be obtained from ( [ gammalmmsenpc ] ) as @xmath127 .
the collaborative rci ( crci ) precoder for the @xmath25 cell is given by @xmath128 where @xmath129 with @xmath130^t \in \mathbb{c}^{mk \times n_t}$ ] , @xmath131 is a normalization constant , and @xmath132 is a regularization constant .
the corresponding sinr of the @xmath16 mt in the @xmath25 cell is provided in the following proposition .
_ proposition 2 _ : for crci data precoding , the received sinr at the @xmath16 mt in the @xmath25 cell is given by @xmath133 where @xmath134 with @xmath135 .
the proof is similar to that for the sinr for the srci data precoder given in appendix a and omitted here for brevity .
furthermore , the optimal regularization constant maximizing the sinr ( and thus the secrecy rate ) in ( [ gammacrcinpc ] ) is obtained as @xmath136 , and the corresponding maximum sinr is given by @xmath137 on the other hand , for @xmath138 , the crci precoder in ( [ crci ] ) reduces to the collaborative zf ( czf ) precoder .
the corresponding received sinr is provided in the following corollary .
_ corollary 2 _ : assuming @xmath139 , for czf data precoding , the received sinr at the @xmath16 mt in the @xmath25 cell is given by @xmath140 @xmath141 in ( [ czf ] ) is obtained by letting @xmath142 in ( [ gammacrcinpc ] ) .
_ remark 1 : _ selfish data precoders require estimation of in - cell csi , i.e. , @xmath143 , only .
in contrast , collaborative data precoders require estimation of both in - cell and inter - cell csi at the bs , i.e. , @xmath144 . furthermore , since collaborative data precoders attempt to avoid interference not only to in - cell users but also to out - of - cell users , more bs antennas are needed to achieve high performance .
this is evident from corollaries 1 and 2 , which reveal that @xmath145 and @xmath146 are necessary for szf and czf data precoding , respectively . on the other hand ,
if successful , trying to avoid out - of - cell interference is beneficial for the overall performance .
hence , whether selfish or collaborative precoders are preferable depends on the parameters of the considered system , cf .
sections [ s5 ] and [ s6 ] .
the rci and zf data precoders introduced in the previous section achieve a higher performance than simple mf data precoding @xcite .
however , they require a matrix inversion which entails a high computational complexity for the large values of @xmath31 and @xmath21 desired in massive mimo . hence , in this section , we propose a low - complexity poly data precoder which avoids the matrix inversion . as the goal is a low - complexity design ,
we focus on selfish poly precoders , although the extension to collaborative designs is possible .
the proposed poly precoder , @xmath104 , for the @xmath25 bs can be expressed as @xmath147 where @xmath148 , and @xmath149^t$ ] are the real - valued coefficients of the precoder matrix polynomial , which have to be optimized . in the following ,
we show that , for @xmath150 , the optimum coefficients @xmath151 do not depend on the instantaneous channel estimates but are constant and can be determined by exploiting results from free probability @xcite and random matrix theory @xcite . to this end , we define the asymptotic average mean - square error ( mse ) of the users in the @xmath25 cell as @xmath152 $ ] with error vector @xmath153 where @xmath154 includes gaussian noise , inter - cell interference , and inter - cell an leakage .
furthermore , @xmath155 is a normalization constant at the receiver , which does not impact detection performance .
the optimal coefficient vector @xmath151 minimizes @xmath156 for a given power budget @xmath157 for the information - carrying signal , i.e. , @xmath158 where we use the notation @xmath159 .
the optimal coefficient vector , @xmath160 , is provided in the following theorem .
_ theorem 1 _ : for @xmath161 , the optimal coefficient vector minimizing the asymptotic average mse of the users in the @xmath25 cell for the poly precoder in ( [ w ] ) is given by @xmath162 where @xmath163^t$ ] , @xmath164_{i , j}={\rm tr}\left\{{\bf d}_{nn}\right\}\zeta^{i+j}+\left({\rm tr}\left\{{\bf d}_{nn } \boldsymbol{\delta}_n\right\ } + \frac{{\rm tr}\left\{\boldsymbol{\sigma}_n\right\}+p_{\rm an}}{n_t p}\right ) \zeta^{i+j-1}$ ] , @xmath165 $ ] , @xmath166 , and @xmath167 @xmath168 $ ] .
furthermore , @xmath169 denotes the @xmath17-order moment of the sum of the eigenvalues of @xmath170 , i.e. , @xmath171 , which converges to @xmath172 for @xmath173 ( * ? ? ?
* theorem 2 ) .
finally , @xmath174 is chosen such that @xmath175 holds .
please refer to appendix b. we note that @xmath160 does not depend on instantaneous channel estimates , and hence , can be computed offline .
we compare the computational complexity of the considered data precoders in terms of the number of floating point operations ( flops ) @xcite .
each flop represents one scalar complex addition or multiplication .
we assume that the coherence time of the channel is @xmath0 symbol intervals of which @xmath176 are used for training and @xmath177 are used for data transmission .
hence , the complexity required for precoding in one coherence interval is comprised of the complexity required for generating one precoding matrix and @xmath177 precoded vectors .
a similar complexity analysis was conducted in ( * ? ? ?
* section iv ) for selfish data precoders without an injection at the bs .
since the an injection does not affect the structure of the data precoders , we can directly adapt the results from ( * ? ? ?
* section iv ) to the case at hand .
in particular , the selfish mf , the szf / srci , and the czf / crci precoders require @xmath178 , @xmath179 , and @xmath180 flops per coherence interval , see ( * ? ? ? * section iv ) .
in contrast , for the poly data precoder , we obtain for the overall computational complexity @xmath181 flops , which assumes implementation of the precoding operation by horner s rule ( * ? ? ? * section iv ) .
the above complexity expressions reveal that the additional complexity introduced by collaborative data precoders compared to selfish data precoders is at most a factor of @xmath182 .
in addition , the complexity savings achieved with the poly data precoder compared to the szf / srci data precoders increase with increasing @xmath31 for a given @xmath0 .
we note however that , regardless of their complexity , poly data precoders are attractive as they avoid the stability issues that may arise in fixed point implementation of large matrix inverses .
in this section , we investigate the performance of selfish and collaborative ns ( s / cns ) and random an precoders .
in addition , a novel poly an precoder is derived . to the best of the authors knowledge ,
poly an precoding has not been considered in the literature before . for a given dimensionality of the an precoder , @xmath36 ,
the secrecy rate depends on the an precoder only via the an leakage , @xmath183 , given in ( [ eq8 ] ) , which affects the sinr of the mt .
furthermore , the optimal poly data precoder coefficients in ( [ muopt ] ) are affected by the an precoder via the leakage term @xmath184 . in this subsection , for @xmath103
, we will provide closed - form expressions for @xmath183 and @xmath184 for the sns , cns , and random an precoders .
the sns an precoder of the @xmath25 bs is given by @xcite @xmath185 which has rank @xmath186 and exists only if @xmath187 .
we divide the corresponding an leakage @xmath188 into an inter - cell an leakage @xmath189 and an intra - cell an leakage @xmath190 , where @xmath191 . for the sns an precoder ,
@xmath189 is obtained as @xmath192=\mathbb{e}\bigg[{\rm tr}\left\{{\bf a}_{m } { \bf a}^h_m\right\}\bigg]\sum_{m \neq n}^m \beta^k_{mn}= ( n_t - k)\sum_{m \neq n}^m \beta^k_{mn},\ ] ] where we exploited ( * ? ? ? * lemma 11 ) and the independence of @xmath193 and @xmath194 .
in contrast , the intra - cell an leakage power is given by @xmath195 @xmath196=\beta^k_{nn}\mathbb{e}\bigg[\tilde{\bf h}^k_{nn } { \bf a}_{n } { \bf a}^h_n ( \tilde{\bf h}^k_{nn})^h\bigg]=(n_t - k)\beta^k_{nn}\frac{1 + p_\tau \tau \sum_{m \neq n}^m \beta^k_{nm}}{1+p_\tau \tau \sum_{m=1}^m \beta^k_{nm}},\ ] ] as the sns an precoder matrix lies in the null space of the estimated channels of all @xmath31 mts in the @xmath25 cell . similarly , the an leakage relevant for computation of the poly data precoder is obtained as @xmath197 for the cns an precoder at the @xmath25 bs , the an is designed to lie in the null space of the estimated channels between all @xmath198 mts and the bs , i.e. , @xmath199 which has rank @xmath200 and exists only if @xmath201 . the corresponding an leakage to the @xmath16 mt in the @xmath25 cell is given by @xmath202= ( n_t - mk ) \sum_{m=1}^m \beta^k_{mn}\frac{1+p_\tau \tau \sum_{l \neq m}^m \beta^k_{ml}}{1+p_\tau \tau \sum_{l=1}^m \beta^k_{ml}}.\ ] ] furthermore , the cns an precoder results in the same @xmath184 as the sns an precoder , cf .
( [ aaa ] ) . for the random precoder ,
all elements of @xmath34 are i.i.d .
random variables independent of the channel @xcite , i.e. , @xmath34 has rank @xmath203 .
hence , @xmath194 and @xmath193 , @xmath204 , are mutually independent , and we obtain @xmath205=n_t\sum_{m=1}^m \beta^k_{mn}.\ ] ] furthermore , we obtain @xmath206 .
_ remark 2 : _ if the power and time allocated to channel estimation are very small , i.e. , @xmath207 , the s / cns an precoders yield the same @xmath208 and @xmath184 as the random an precoder .
this suggests that in this regime all considered an precoders achieve a similar sinr performance for a given mt .
however , for @xmath209 , the s / cns an precoders cause less an leakage resulting in an improved sinr performance compared to the random precoder at the expense of a higher complexity .
to mitigate the high computational complexity imposed by the matrix inversion required for the s / cns an precoders , while achieving an improved performance compared to the random an precoder , we propose a poly an precoder .
similar to the poly data precoder , we concentrate on the selfish design because of the desired low complexity , and hence , set @xmath186 .
the proposed poly an precoder is given by @xmath210 where @xmath211^t$ ] contains the real - valued coefficients of the an precoder polynomial , which have to be optimized .
in particular , @xmath212 is optimized for minimization of the asymptotic average an leakage caused to all mts in the @xmath25 cell @xmath184 .
the corresponding optimization problem is formulated as @xmath213\quad { \rm s.t . : } { \rm tr}\{{\bf a}_n^h { \bf a}_n\}= 1/\beta-1.\ ] ] the solution of ( [ opt2 ] ) is provided in the following theorem . _ theorem 2 _ : for @xmath161 , the optimal coefficient vector minimizing the asymptotic average an leakage caused to the users in the @xmath25 cell for the an precoder structure in ( [ v ] ) is given by @xmath214 where @xmath215_{i , j}=\zeta^{i+j+1}+\epsilon \zeta^{i+j}$ ] and @xmath216 $ ] . here
, @xmath169 denotes again the @xmath17 order moment of the sum of the eigenvalues of matrix @xmath170 , cf .
theorem 1 .
@xmath217 is chosen such that @xmath218
. please refer to appendix c. similarly to the data precoders , the complexity of the an precoders is evaluated in terms of the number of flops required per coherence interval @xmath0 . for the sns an precoder ,
the computation of @xmath34 in ( [ sn ] ) requires the computation and inversion of a @xmath219 positive definite matrix , which entails @xmath220 flops @xcite , and the multiplication of an @xmath221 , an @xmath219 , and an @xmath222 matrix , which entails @xmath223 flops @xcite .
furthermore , the @xmath177 vector - matrix multiplications required for an precoding entail a complexity of @xmath224 flops @xcite , respectively . hence , the overall complexity is @xmath225 flops .
similarly , for the cns an precoder , we obtain a complexity of @xmath226 flops , whereas the random an precoder entails a complexity of @xmath227 flops as only the an vector - matrix multiplications are required .
similar to the precoded data vector ( * ? ? ?
* section iv ) , the poly precoded an vector can be generated using horner s rule .
hence , based on ( [ v ] ) , the transmitted an vector in the @xmath25 cell can be obtained as @xmath228 hence , @xmath229 can be computed efficiently by first multiplying @xmath230 with @xmath231 , which requires @xmath232 flops , then multiplying @xmath233 with the resulting vector , which requires @xmath234 flops , adding @xmath231 to the newly resulting vector , and repeating similar operations @xmath235 times , see @xcite for details of horner s rule .
overall , this leads to a complexity of @xmath236 flops .
in this subsection , we compare the secrecy performances of the considered data and an precoders . thereby , in order to get tractable results , we focus on the relative performances of szf , czf , and mf @xcite data precoders and sns , cns , and random an precoders . the performances of srci , crci , and poly data precoders and the poly an precoder will be investigated via numerical and simulation results in section [ s6 ] . in order to gain some insight for system design and analysis ,
we adopt a simplified path - loss model .
in particular , we assume the path losses are given by @xmath237 where @xmath238 $ ] denotes the inter - cell interference factor . for this simplified model , @xmath239 and @xmath240 in ( [ cup ] )
simplify to @xmath241 and @xmath242 .
furthermore , the sinr expressions of the linear data precoders considered in section [ s3a ] and the mf precoder considered in @xcite can be simplified considerably and are provided in table [ tablea ] , where we use the normalized an leakage @xmath243 .
the expressions for the normalized an leakage @xmath244 , the asymptotic average an leakage @xmath184 , and the dimensionality @xmath36 of the considered linear an precoders are given in table [ tableb ] .
[ cols="^,^",options="header " , ] [ tableb ] in this subsection , we compare the performances achieved with szf , czf , and mf data precoders for a given an precoder , i.e. , @xmath36 and @xmath244 are fixed . since the upper bound on the capacity of the eavesdropper channel is independent of the adopted data precoder , cf .
section [ s2c ] , we compare the considered data precoders based on their sinrs . exploiting the results in table [ tablea ] , we obtain the following relations between @xmath245 , @xmath246 , and @xmath247 : @xmath248 hence , for @xmath249 , we require @xmath250 , and for @xmath251 , we need @xmath252 $ ] .
as expected , ( [ eq25 ] ) suggests that for a lightly loaded system , i.e. , @xmath253 , all three precoders have a similar performance , i.e. , @xmath254 . in the following ,
we investigate the impact of the number of mts and the pilot power on the relative performances of the considered data precoders .
\1 ) _ number of mts _ : from ( [ eq25 ] ) , we find that for @xmath249 and @xmath251 to hold , the number of mts has to meet @xmath255 and @xmath256 , where @xmath257 @xmath258\phi+1/p_t},\ ] ] respectively .
interestingly , both the maximum numbers of mts for which the szf data precoder is advantageous compared to the mf data precoder , @xmath259 , and the maximum number of mts for which the czf data precoder is advantageous compared to the szf data precoder , @xmath260 , decrease with increasing an leakage , @xmath244 , and increasing number of cells , @xmath20 , but increase with the amount of resources dedicated to channel estimation , @xmath261 ( via @xmath262 ) , and consequently with the channel estimation quality .
however , while @xmath259 decreases with increasing inter - cell interference factor , @xmath263 ( via @xmath239 ) , @xmath260 increases .
\2 ) _ pilot energy _ : from ( [ eq25 ] ) , we find that for @xmath249 and @xmath251 to hold , pilot energy @xmath264 has to fulfill @xmath265 where we have assumed that sns an precoding is adopted , i.e. , @xmath266 , to arrive at insightful expressions .
similar results can be obtained for other an precoders . from ( [ przf ] )
, we observe that mf , szf , and czf data precoding are preferable if @xmath267 , @xmath268 , and @xmath269 , respectively . in general
, the more mts are in the system ( i.e. , the larger @xmath270 ) , the larger the pilot energy has to be to make szf and czf data precoding beneficial .
in fact , from ( [ przf ] ) we observe that if @xmath270 exceeds @xmath271 $ ] , mf data precoding is always preferable regardless of the value of @xmath261 .
similarly , if @xmath270 exceeds @xmath272 $ ] , szf data precoding is always preferable compared to czf data precoding regardless of the value of @xmath261 . in this subsection , we analyze the impact of the an precoders on the secrecy rate .
an precoders affect the ergodic capacity of the eavesdropper via @xmath36 and the achievable rate of the mt via the leakage , @xmath244 .
since the upper bound on the ergodic secrecy rate of the eavesdropper in ( [ cup ] ) is a decreasing function in @xmath36 , we have @xmath273 on the other hand , from table [ tableb ] , we observe @xmath274 . since according to table [ tablea ] the sinrs for all data precoders are decreasing functions of @xmath275 , for a given data precoder , we obtain for the lower bound on the ergodic rate of the @xmath16 mt in the @xmath25 cell @xmath276 considering ( [ a-1 ] ) , ( [ a-2 ] ) , and the expression for the ergodic secrecy rate , @xmath277^+$ ] , it is not a priori clear which an precoder has the best performance . in fact , our numerical results in section [ s6 ] confirm that it depends on the system parameters ( e.g. @xmath94 , @xmath270 , @xmath20 , @xmath261 , and @xmath263 ) which an precoder is preferable . in this subsection , we provide closed - form results for the ergodic secrecy rate for szf , czf , and mf data precoding for the simplified path - loss model in ( [ eq21 ] ) .
thereby , the simplified path - loss model is extended also to the eavesdropper , i.e. , @xmath278 and @xmath279 , @xmath280 , is assumed . combining ( [ secnk ] ) , ( [ cup ] ) , and the results in table [ tablea ] , we obtain the following lower bounds for the ergodic secrecy rate of the @xmath16 mt in the @xmath25 cell : @xmath281^+ & \quad { \rm for~mf},\\\bigg[\log_2 \left(\frac{(\tilde{q}+1/p_t)\beta+(a-\theta-\tilde{q})\beta \phi+c\theta(1-\beta)\phi}{(\tilde{q}+1/p_t)\beta+(a-\theta-\tilde{q})\beta \phi+(c-1)\theta(1-\beta)\phi}\cdot \frac{-\chi \phi+\chi}{(1-\chi)\phi+\chi}\right)\bigg]^+ & \quad { \rm for~szf},\\\bigg[\log_2 \left(\frac{(\tilde{q}+1/p_t)\beta+(a - a\theta-\tilde{q})\beta \phi+c\theta(1-m\beta)\phi}{(\tilde{q}+1/p_t)\beta+(a - a\theta-\tilde{q})\beta \phi+(c-1)\theta(1-m\beta)\phi}\cdot \frac{-\chi \phi+\chi}{(1-\chi)\phi+\chi}\right)\bigg]^+ & \quad { \rm for~czf},\end{cases}\ ] ] where @xmath282 , and @xmath275 and @xmath36 are given in table [ tableb ] for the considered an precoders .
( [ rmfzf ] ) is easy to evaluate and reveals how the ergodic secrecy rate of the three considered data precoders depends on the various system parameters . to gain more insight , we determine the maximum value of @xmath94 which admits a non - zero secrecy rate .
this value is denoted by @xmath283 in the following , and can be shown to be a decreasing function of @xmath45 for all conidered data precoders .
hence , we find @xmath283 by setting @xmath284 in ( [ rmfzf ] ) and letting @xmath285 .
this leads to @xmath286 eq .
( [ alphas ] ) reveals that for a given an precoder , independent of the system parameters , the mf data precoder can always tolerate a larger number of eavesdropper antennas than the szf data precoder , which in turn can always tolerate a larger number of eavesdropper antennas than the czf data precoder .
this can be explained by the fact that the high an transmit power required to combat a large number of eavesdropper antennas drives the receiver of the desired mt into the noise - limited regime , where the mf data precoder has a superior performance compared to the s / czf data precoders . on the other hand ,
since @xmath283 depends on both @xmath275 and @xmath36 , it is not a priori clear which an precoder can tolerate the largest number of eavesdropper antennas . for a lightly loaded network with small @xmath270 and small @xmath20 , according to table [ tableb ] , we have @xmath287 for all three an precoders .
hence , in this case , we expect the cns an precoder to outperform the sns and random an precoders as it achieves a smaller @xmath244 . on the other hand , for a heavily loaded network with large @xmath270 and @xmath20 ,
the value of @xmath283 of the cns an precoder is compromised by its small value of @xmath36 and sns and even random an precoders are expected to achieve a larger @xmath283 .
in this section , we evaluate the performance of the considered secure multi - cell massive mimo system .
we consider cellular systems with @xmath288 and @xmath289 hexagonal cells , respectively , and to gain insight for system design , we adopt the simplified path - loss model introduced in section [ s5 ] , i.e. , the severeness of the inter - cell interference is only characterized by the parameter @xmath290 $ ] .
the pilot sequence length is @xmath291 .
the simulation results for the ergodic secrecy rate of the @xmath16 mt in the @xmath25 cell are based on ( [ secnk ] ) , ( [ ceve ] ) , and the expression for the ergodic rate of the mt ( * ? ? ?
* eq . ( 8) ) and are averaged over @xmath292 random channel realizations .
note that , in this paper , we consider the ergodic secrecy rate of a certain mt , i.e. , the @xmath16 mt in the @xmath25 cell .
the cell sum secrecy rate can be obtained by multiplying the secrecy rate of the @xmath16 mt by the number of mts , @xmath31 , as for the considered channel model , all mts in the @xmath25 cell achieve the same secrecy rate .
the values of all relevant system parameters are provided in the captions of the figures . to enable a fair comparison , throughout this section
, we adopted the selfish sns an precoder when we compare different data precoders and the selfish zf data precoder when we compare different an precoders . in fig .
[ fig0 ] , we show the ergodic capacity of the eavesdropper for the considered conventional an precoders . first , we note that the upper bound in ( [ cup ] ) is very tight since the number of bs antennas is large ( @xmath293 ) and @xmath294 holds for all considered an precoders and all consider values of @xmath94 and @xmath270 .
furthermore , as @xmath270 increases , the ergodic capacity of all an precoders decreases since the power allocated to the information - carrying signal of the user that the eavesdropper tries to intercept decreases with increasing @xmath270 as the total power allocated to the information - carrying signals of all users is fixed . as expected , the eavesdropper s capacity benefits from larger values of @xmath94 .
furthermore , as predicted in ( [ a-1 ] ) , because of their different values of @xmath36 , the cns an precoder yields the largest eavesdropper capacity , while the random an precoder yields the lowest .
the performance differences between the different an precoders diminish for small values of @xmath94 and @xmath270 as the dependence of the eavesdropper capacity on @xmath36 becomes negligible for small @xmath94 , cf .
( [ cup ] ) , and @xmath287 holds for all precoders for small @xmath270 , cf .
table [ tableb ] . , for a system with @xmath293 , @xmath295 , @xmath296 db , @xmath297 , and @xmath288.,title="fig:",width=355 ] + in figs .
[ fig1 ] and [ fig2 ] , we show the ergodic secrecy rates of the @xmath16 mt in the @xmath25 cell vs. the number of bs antennas for the mf , szf , czf , srci , and crci data precoders for a lightly loaded and a dense network , respectively , and a fixed power allocation factor of @xmath295 . in both figures
, the analytical results were obtained from ( [ secnk ] ) , ( [ ceve ] ) , and ( [ srci - opt ] ) for the srci data precoder , ( [ crci - opt ] ) for the crci data precoder , and ( [ rmfzf ] ) for the mf , szf , and czf data precoders . for all considered precoders , the analytical results provide a tight lower bound for the ergodic secrecy rates obtained by simulations .
furthermore , as expected , the rci data precoders outperform the zf data precoders for both the selfish and the collaborative strategies , but the performance gap diminishes with increasing number of bs antennas . for the lightly loaded network in fig .
[ fig1 ] , we assume @xmath288 cells , @xmath298 users , and a small inter - cell interference factor of @xmath299 . for this scenario , the collaborative designs outperform the selfish designs and c / szf precoding yields a large performance gain compared to mf precoding .
this is expected from our analysis in section [ s5a ] as for the parameters valid for fig .
[ fig1 ] , we obtain from ( [ betazf ] ) , @xmath300 and @xmath301 for @xmath302 .
intuitively , as the network is only lightly loaded , the collaborative data precoder can efficiently reduce interference to the other cell despite the pilot contamination .
, for a dense network with @xmath295 , @xmath296 db , @xmath303 , @xmath304 , @xmath305 , @xmath297 , and @xmath289.,title="fig:",width=355 ] + , for a dense network with @xmath295 , @xmath296 db , @xmath303 , @xmath304 , @xmath305 , @xmath297 , and @xmath289.,title="fig:",width=355 ] + for the dense network in fig .
[ fig2 ] , we assume @xmath289 cells , @xmath305 users , and a larger inter - cell interference factor of @xmath297 . in this case , for the considered range of @xmath21 , the collaborative precoder designs are not able to suppress inter - cell interference and an leakage to other cells sufficiently well to outperform the selfish precoder designs .
in fact , for @xmath302 , we obtain from ( [ betazf ] ) @xmath306 , i.e. , our analytical results suggest that the szf precoder outperforms the czf precoder for @xmath305 which is confirmed by fig .
[ fig2 ] . nevertheless , for @xmath307 , the ergodic secrecy rate for the czf data precoder will eventually surpass that for the szf data precoder .
in this subsection , we investigate the dependence of the ergodic secrecy rate on the power allocation factor @xmath45 and study the impact of system parameters such as @xmath270 , @xmath20 , and @xmath263 on the optimal @xmath45 that maximizes the ergodic secrecy rate .
the results in this subsection were generated based on the analytical expressions in ( [ secnk ] ) , ( [ ceve ] ) , and ( [ srci - opt ] ) for the srci data precoder , ( [ crci - opt ] ) for the crci data precoder , and ( [ rmfzf ] ) for the mf , szf , and czf data precoders . for different data precoders for a network with @xmath296 db , @xmath308 , @xmath303 , @xmath304 , @xmath309 , and @xmath299 . , title="fig:",width=355 ] + for different data precoders for a network with @xmath296 db , @xmath308 , @xmath303 , @xmath304 , @xmath309 , and @xmath299 . , title="fig:",width=355 ] + fig . [ fig3 ]
depicts the ergodic secrecy rate of the @xmath16 mt in the @xmath25 cell for the selfish data precoders srci , szf , and mf as a function of the power allocation factor @xmath45 .
all curves are concave and have a single maximum . for @xmath310
only an is transmitted , hence @xmath311 results since no data can be transmitted . for @xmath312 ,
no an is transmitted , hence @xmath311 results since the capacity of the eavesdropper becomes unbounded ( recall that we make the worst - case assumption that the eavesdropper can receive noise - free ) . for @xmath313
, a positive secrecy rate may result depending on the system parameters and the precoding schemes . since we keep the total transmit power fixed
, the transmit power per mt decreases with increasing @xmath270 . to compensate for this effect
, the portion of the total transmit power allocated to data transmission should increase .
this is confirmed by fig .
[ fig3 ] where the optimal value of @xmath45 for @xmath314 is larger than that for @xmath309 .
furthermore , for a given @xmath270 , the optimal @xmath45 is the larger , the better the performance of the adopted data precoder is , i.e. , for a more effective data precoder , transmitting the data signal with higher power is more beneficial , whereas for a less effective data precoder impairing the eavesdropper with a higher an power is more beneficial . in fig .
[ fig4 ] , we show the ergodic secrecy rate vs. @xmath45 for the crci , czf , and szf precoders . similar to our observations in fig .
[ fig3 ] , for given system parameters , the optimal @xmath45 tends to be larger for more effective precoders that achieve a better performance .
for the system with @xmath289 , this can be observed by comparing the optimal @xmath45 for the szf and czf precoders .
furthermore , while for the smaller system with @xmath288 cells collaborative precoding is always preferable , for @xmath289 , szf precoding outperforms czf and crci precoding for all considered values of @xmath45 , as the collaborative designs are not able to effectively suppress the interference and an leakage to the @xmath315 users in the other cells with the available @xmath308 antennas . in particular , from ( [ betazf ] ) , we obtain @xmath316 for @xmath288 and @xmath317 for @xmath289 , which confirms the results shown in fig .
[ fig4 ] . vs. @xmath270 for different data and an precoders for a network with @xmath296 db , @xmath308 , @xmath303 , @xmath297 , and @xmath288 .
, title="fig:",width=355 ] + vs. @xmath270 for different data and an precoders for a network with @xmath296 db , @xmath308 , @xmath303 , @xmath297 , and @xmath288 . ,
title="fig:",width=355 ] + fig .
[ fig45 ] depicts the ergodic secrecy rate vs. @xmath45 for the considered conventional an precoder structures .
we consider a lightly loaded network with @xmath318 and a moderately loaded network with @xmath319 .
for @xmath318 , the cns an precoder outperforms the sns an precoder since , in this case , for the cns an precoder , the negative impact of having ( slightly ) fewer dimensions available for degrading the eavesdropper s channel ( smaller value of @xmath36 ) is outweighed by the positive impact of causing less an leakage ( smaller value of @xmath244 ) .
on the other hand , for @xmath319 , the cns an precoder has a substantially smaller @xmath36 than the sns precoder which can not be compensated by its larger @xmath244 . despite having the largest value of @xmath36
, the random an precoder has the worst performance for both considered cases because of its large an leakage . in section [ s5c ]
, we showed that a positive ergodic secrecy rate is possible only if @xmath320 . in fig .
[ fig5 ] , using ( [ alphas ] ) , we plot @xmath283 as a function of @xmath270 . in the left hand side subfigure , we compare mf , szf , and czf data precoding for sns an precoding , and in the right hand side subfigure , we compare random , sns , and cns an precoding for szf data precoding .
the comparison of the data precoders reveals that although szf and czf entail a much higher complexity , mf precoding achieves a larger @xmath283 . therefore ,
if the eavesdropper has a large number of antennas and small ergodic secrecy rates are targeted , simple mf precoding is always preferable .
on the other hand , whether sns or cns an precoder is preferable depends on the system load . for small values of @xmath270 , cns an precoding can tolerate more eavesdropper antennas , whereas for large values of @xmath270 , sns an precoding is preferable .
random an precoding is outperformed by sns an precoding for any value of @xmath270 .
a closer examination of ( [ alphas ] ) reveals that this is always true if s / czf data precoders are employed .
however , for the mf data precoder , there are parameter combination for which random an precoding outperforms sns and cns an precoding . in this subsection
, we evaluate the ergodic secrecy rates of the proposed low - complexity poly data and an precoders . to this end , we consider again a lightly loaded network with little inter - cell interference ( @xmath288 , @xmath309 , @xmath299 ) and a dense network with more inter - cell interference ( @xmath289 , @xmath321 , @xmath297 ) .
all results shown in this section were obtained by simulation . for each simulation point , the optimal value of @xmath45 was found numerically and applied . in figs .
[ fig6 ] and [ fig7 ] , we show the ergodic secrecy rate of the @xmath16 mt in the @xmath25 cell as a function of the pilot energy , @xmath322 .
as expected , for all considered schemes , the ergodic secrecy rate is monotonically increasing in the pilot energy since more accurate channel estimates improve the performance . , @xmath296 db , @xmath293 , and @xmath304 . ,
title="fig:",width=355 ] + , @xmath296 db , @xmath293 , and @xmath304 .
, title="fig:",width=355 ] + in fig .
[ fig6 ] , we depict the ergodic secrecy rates for the proposed poly data precoder for different values of @xmath323 and compare them to those of conventional selfish data precoders . for the sake of comparison ,
all data precoders are combined with the sns an precoder . as the number of terms of the polynomial @xmath323 increase , the performance of the poly data precoder quickly improves and approaches that of the srci data precoder .
the convergence is faster for the dense network considered in the right hand side subfigure , where the performance difference between all precoders is smaller in general since interference can not be as efficiently avoided as for the lightly loaded network . in fig .
[ fig7 ] , we show the ergodic secrecy rates for the proposed poly an precoder for different values of @xmath324 and compare them to those of the random and sns an precoders . for the sake of comparison ,
all an precoders are combined with szf data precoding .
the poly an precoder quickly approaches the performance of the sns an precoder as the polynomial order @xmath324 increases .
similar to the poly data precoders , the convergence is faster for the dense network where the performance differences between different an precoders are also smaller . for the denser network ,
even the random an precoder is a viable option and suffers only from a small loss in performance compared to the sns an precoder .
db , @xmath325 , @xmath326 , @xmath288 , @xmath299 , @xmath327 , and an szf data precoder.,title="fig:",width=355 ] + db , @xmath325 , @xmath326 , @xmath288 , @xmath299 , @xmath327 , and an szf data precoder.,title="fig:",width=355 ] + in this subsection , we investigate the tradeoff between the ergodic secrecy rate performance and the computational complexity of the proposed data and an precoders in figs .
[ fig8 ] and [ fig9 ] , respectively .
in particular , figs . [ fig8 ] and [ fig9 ] depict the ergodic secrecy rate on the left hand side and the computational complexity ( in giga flop ) on the right hand side , both as a function of the numbers of users in a cell . for the considered setting , the performance gains of collaborative data and an precoding compared to selfish strategies
are moderate , but the associated increase in complexity is substantial , especially for large @xmath31 .
[ fig8 ] illustrates that for the considered setting a poly data precoder with @xmath328 achieves a better performance than the mf precoder but has substantially lower complexity than the srci precoder . for large @xmath323
, the poly data precoder has a lower complexity than the srci precoder for large @xmath31 .
however , even for small @xmath31 , the poly precoder may be preferable as it does not incur the stability issues that may arise in the implementation of the large - scale matrix inversions required for the srci precoder . fig .
[ fig9 ] shows that for the considered setting the proposed poly an precoder with @xmath329 outperforms the random an precoder .
the poly an precoder with @xmath330 achieves almost the same performance as the sns an precoder but with a substantially lower complexity .
we further observe that for small @xmath31 , because of its efficient implementation via horner s scheme , cf .
( [ ahorner ] ) , the proposed poly an precoder requires an even lower complexity than the random an precoder .
in this paper , we considered downlink multi - cell massive mimo systems employing linear data and an precoding for physical layer security provisioning .
we analyzed and compared the achievable ergodic secrecy rate of various conventional data and an precoders in the presence of pilot contamination . to this end , we also optimized the regularization constants of the selfish and collaborative rci precoders in the presence of an and multi - cell interference .
in addition , we derived linear poly data and an precoders which offer a good compromise between complexity and performance in massive mimo systems .
interesting findings of this paper include : 1 ) collaborative data precoders outperform selfish designs only in lightly loaded systems where a sufficient number of degrees of freedom for suppressing inter - cell interference and sufficient resources for training are available .
2 ) similarly , cns an precoding is preferable over sns an precoding in lightly loaded systems as it causes less an leakage to the information - carrying signal , whereas in more heavily loaded systems , cns an precoding does not have sufficient degrees of freedom for effectively degrading the eavesdropper channel and sns an precoding is preferable .
3 ) for a large number of eavesdropper antennas , where only small positive secrecy rates are achievable , mf data precoding is always preferable compared to szf and czf data precoding .
4 ) the proposed poly data and an precoders approach the performances of the srci data and sns an precoders with only a few terms in the respective matrix polynomials and are attractive options for practical implementation .
considering ( [ eq3 ] ) and ( [ srci ] ) , the effective signal power , i.e. , the numerator in ( [ rnk ] ) , can be expressed as @xcite @xmath331 = \gamma^2_1 \mathbb{e}^2[{\bf h}^k_{nn } { \bf l}_{nn } ( \hat{\bf h}^k_{nn})^h]=\gamma^2_1 \mathbb{e}^2\bigg[\frac{{\bf h}^k_{nn } { \bf l}_{n , k } ( \hat{\bf h}^k_{nn})^h}{1+\hat{\bf h}^k_{nn } { \bf l}_{n , k } ( \hat{\bf h}^k_{nn})^h}\bigg ] = \frac{\gamma^2_1 ( x_{nk}+a_{nk})^2}{(1+x_{nk})^2},\ ] ] where @xmath332 , @xmath333 $ ] , and @xmath334 $ ] . on the other hand , the intra - cell interference term in the denominator of ( [ rnk ] )
can be expressed as @xmath335 = \gamma^2_1 \mathbb{e}\bigg[\frac{{\bf h}^k_{nn } { \bf l}_{n , k } \hat{\bf h}^h_{n , k } \hat{\bf h}_{n , k } { \bf l}_{n , k } ( { \bf h}^k_{nn})^h}{\left(1+\hat{\bf h}^k_{nn } { \bf l}_{n , k } ( \hat{\bf h}^k_{nn})^h\right)^2}\bigg]=\frac{\gamma^2_1 ( y_{nk}+b_{nk})}{(1+x_{nk})^2},\ ] ] where @xmath336 is equal to @xmath337 with the @xmath16 row removed , and @xmath338 $ ] and @xmath339 $ ] . due to pilot contamination , the data precoding matrix of the @xmath53 bs is a function of the channel vectors between the @xmath53 bs and the @xmath16 mts in all cells .
hence , the inter - cell interference from the bss in adjacent cells is obtained as @xmath340=\frac{\gamma^2_1 ( x_{nk}+a_{nk})^2}{(1+x_{nk})^2}+\frac{1+p_\tau \tau \sum_{l \neq m}^m \beta^k_{ml}}{1+p_\tau \tau \sum_{l=1}^m \beta^k_{ml}}.\ ] ] meanwhile , by exploiting ( [ sp ] ) , ( [ eq13a ] ) , and the definition of the variance , i.e. , @xmath341=\mathbb{e}[x^2]-\mathbb{e}^2[x]$ ] , we obtain for the first term of the denominator of ( [ rnk ] ) , @xmath342=\frac{1+p_\tau \tau \sum_{m \neq n}^m \beta^k_{nm}}{1+p_\tau \tau \sum_{m=1}^m \beta^k_{nm}}$ ] . according to ( * ? ? ?
( 16 ) ) and ( * ? ? ?
* theorem 7 ) , for @xmath103 and constant @xmath270 , @xmath343 converges to @xmath344 defined in ( [ g ] ) and @xmath345 . similarly , @xmath346 and @xmath347 approach @xmath348 and @xmath349 respectively , where @xmath350 .
moreover , the inter - cell interference from other mts ( i.e. , not the @xmath16 mts ) is calculated as @xmath351=\mathbb{e}\bigg[{\rm tr}\left\{{\bf f}_{m , k}{\bf f}^h_{m , k}\right\}\bigg]=k-1,\ ] ] where @xmath352 is equal to @xmath353 with the @xmath16 column removed . the first equality in ( [ io ] )
is due to the fact that the precoding matrix for the other mts ( i.e. , not the @xmath16 mts ) in adjacent cells are independent of @xmath194 and ( * ? ? ?
* lemma 11 ) , while the second equality holds for @xmath354 . on the other hand , the constant scaling factor @xmath110 for srci precoding is given by ( * ? ? ?
* eq . ( 22 ) ) @xmath355 hence , employing ( [ sp])-([gamma12 ] ) in ( [ rnk ] ) , the received sinr in ( [ gammalmmsenpc ] ) is obtained , which completes the proof of _ proposition 1_. the objective function in ( [ opt1 ] ) can be rewritten as @xmath356 \nonumber\\ \nonumber & & + \varsigma^2 p \mathbb{e}\bigg[{\rm tr}\bigg\{\sum_{i=0}^{\cal i } \mu_i \bigg(\hat{\overline{\bf h}}_{nn } \hat{\overline{\bf h}}_{nn}^h\bigg)^i \hat{\overline{\bf h}}_{nn}\tilde{\bf \overline{h}}_{nn}^h { \bf d}_{nn } \tilde{\bf \overline{h}}_{nn } \hat{\overline{\bf
h}}_{nn}^h \sum_{i=0}^{\cal i } \mu_i \bigg(\hat{\overline{\bf h}}_{nn } \hat{\overline{\bf h}}_{nn}^h\bigg)^i\bigg\}\bigg]\nonumber\\ & & -2 \varsigma \sqrt{p}\mathbb{e}\bigg[{\rm tr}\bigg\{{\bf
d}^{1/2}_{nn}\sum_{i=0}^{\cal i } \mu_i \bigg(\hat{\overline{\bf h}}_{nn } \hat{\overline{\bf h}}_{nn}^h\bigg)^{i+1}\bigg\}\bigg]+1 + \varsigma^2 p_{\rm an}+\varsigma^2 { \rm tr}\left\{\boldsymbol{\sigma}_n\right\},\end{aligned}\ ] ] where we exploited @xmath357={\bf i}_k$ ] , the definition of @xmath184 given in theorem 1 , the definition of @xmath104 in ( [ w ] ) , the definition @xmath358 , and @xmath359 . in the following ,
we simplify the right hand side ( rhs ) of ( [ x1 ] ) term by term . to this end
, we denote the first three terms on the rhs of ( [ x1 ] ) by @xmath360 , @xmath361 , and @xmath362 , respectively . using a result from free probability theory @xcite
, the first term converges to ( * ? ? ?
* theorem 1 ) @xmath363 , \label{eq49}\ ] ] as matrix @xmath364 is free from @xmath365 .
similarly , the third term converges to @xmath366.\ ] ] furthermore , the second term can be rewritten as @xmath367\nonumber\\ & \stackrel{{\rm ( { b})}}{=}&\varsigma^2 p n_t { \rm tr}\left\{{\bf d}_{nn } \boldsymbol{\delta}_n\right\ } , \label{eq51}\end{aligned}\ ] ] where ( a ) follows again from ( * ? ? ?
* theorem 1 ) and ( b ) results from @xmath368={\rm tr}\left\{{\bf d}_{nn } \boldsymbol{\delta}_n\right\}$ ] , where @xmath369 is defined in theorem 1 , ( [ w ] ) , and the constraint in ( [ opt1 ] ) . exploiting ( [ eq49])-([eq51 ] ) and the eigen - decomposition of matrix @xmath370 , where diagonal matrix @xmath371 contains all eigenvalues and unitary matrix @xmath372 contains the corresponding eigenvectors , the asymptotic average mse becomes @xmath373 \nonumber\\ & & + 1 + \varsigma^2 p_{\rm an}+\varsigma^2 { \rm tr}\left\{\boldsymbol{\sigma}_n\right\}+\varsigma^2pn_t { \rm tr}\left\{{\bf d}_{nn } \boldsymbol{\delta}_n\right\}.\end{aligned}\ ] ] next , we introduce the vandermonde matrix @xmath374 , where @xmath375_{i , j}=\lambda^{j-1}_i$ ] , and @xmath376^t$ ] , which allows us to rewrite ( [ eq53 ] ) in compact form as @xmath377\nonumber\\ & & \qquad+1 + \varsigma^2 p_{\rm an}+\varsigma^2 { \rm tr}\left\{\boldsymbol{\sigma}_n\right\ } + \varsigma^2p n_t { \rm tr}\left\{{\bf d}_{nn } \boldsymbol{\delta}_n\right\}.\end{aligned}\ ] ] similarly , the constraint in ( [ opt1 ] ) can be expressed as @xmath378=n_t . \label{eq55}\ ] ] thus , the lagrangian function of primal problem ( [ opt1 ] ) can be expressed as @xmath379-n_t)$ ] , where @xmath380 is the lagrangian multiplier . taking the gradient of the lagrangian function with respect to @xmath151 , and setting the result to zero , we obtain for the optimal coefficient vector @xmath160 : @xmath381 \boldsymbol{\mu } = \frac{{\rm tr}\left\{{\bf d}^{1/2}_{nn}\right\ } } { \varsigma \sqrt{p } { \rm tr}\left\{{\bf d}_{nn}\right\ } } \lim_{k \to \infty } \frac{1}{k } \mathbb{e}\left[{\bf c}^t_1 \boldsymbol{\lambda}\right ] .
\label{eq56}\ ] ] furthermore , taking the derivative of @xmath382 with respect to @xmath155 and equating it to zero , and multiplying both sides of ( [ eq56 ] ) by @xmath383 and applying ( [ eq55 ] ) , we obtain @xmath384 the expressions involving @xmath385 , @xmath386 , and @xmath387 in ( [ eq56 ] ) can be further simplified .
for example , we obtain @xmath388_{m , n}\bigg]= \lim_{k \to \infty } \mathbb{e}\bigg[\frac{1}{k } \sum_{k=1}^k \lambda^{m+n-1}_k\bigg]$ ] . simplifying the other terms in ( [ eq56 ] ) in a similar manner and inserting ( [ eq57 ] ) into ( [ eq56 ] )
we obtain the result in theorem 1 . exploiting @xmath389={\bf i}_{n_t}$ ] , the constraint in ( [ opt2 ] ) , and a similar approach as was used to arrive at ( [ aaa ] ) , the objective function in ( [ opt2 ] ) can be simplified as @xmath167 @xmath390=q\mathbb{e}\bigg[{\rm tr}\left\ { { \bf d}_{nn } \hat{{\bf h}}_{nn } { \bf a}_n { \bf a}^h_n \hat{\bf h}^h_{nn}\right\}\bigg ] + ( 1-\phi)p_t{\rm tr}\{{\bf d}_{nn } \boldsymbol{\delta}_n\}.\ ] ] defining vandermode matrix @xmath392 , where @xmath393_{i , j}=\lambda_i^{j-1}$ ] , we can rewrite ( [ y22 ] ) in compact form as @xmath167 @xmath394+(1-\phi)p_t{\rm tr}\{{\bf d}_{nn } \boldsymbol{\delta}_n\},\ ] ] where @xmath395 denotes the all - ones column vector . taking into account the constraint in ( [ opt2 ] )
, we can formulate the lagrangian as @xmath396 + 1)$ ] with lagrangian multiplier @xmath397 .
the optimal coefficient vector @xmath398 is then obtained by taking the gradient of the lagrangian function with respect to @xmath212 and setting it to zero : @xmath399\boldsymbol{\nu}=\lim_{k \to \infty } \mathbb{e}\bigg[{\bf c}^t_2 \left(\boldsymbol{\lambda}+\epsilon { \bf i}_k\right ) \boldsymbol{\lambda}\bigg],\ ] ] where we used @xmath400 . simplifying the terms in ( [ nuopt ] ) by exploiting a similar approach as in appendix b
, we obtain the result in theorem 2 .
20 f. rusek , d. persson , b. k. lau , e. g. larsson , t. l. marzetta , o. edfors , and f. tufvesson , `` scaling up mimo : opportunities and challenges with very large arrays , '' _ ieee sig .
proc . mag .
_ , vol . 30 , no .
40 - 46 , jan . 2013 .
h. q. ngo , e. g. larsson , and t. l. marzetta , `` energy and spectral efficiency of very large multiuser mimo systems , '' _ ieee trans .
commun . _ , vol .
1436 - 1449 , apr . 2013 .
j. jose , a. ashikhmin , t. l. marzetta , and s. vishwanath , `` pilot contamination and precoding in multi - cell tdd systems , '' _ ieee trans .
wireless commun .
10 , no . 8 , pp .
2640 - 2651 , aug .
a. mukherjee , s. a. a. fakoorian , j. huang , and a. l. swindlehurst , `` principles of physical layer security in multiuser wireless networks : a survey , '' _ ieee commun .
surveys & tutorials _ , vol .
3 , third quarter , 2014 .
f. oggier and b. hassibi , `` the secrecy capacity of the mimo wiretap channel , '' _ ieee trans .
inform . theory _
57 , no . 8 , pp.4961 - 4972 , aug .
s. goel and r. negi , `` guaranteeing secrecy using artificial noise , '' _ ieee trans .
wireless commun .
_ , vol . 7 , no . 6 , pp
. 2180 - 2189 , june 2008 . x. zhou and m. r. mckay , `` secure transmission with artificial noise over fading channels : achievable rate and optimal power allocation , '' _ ieee trans . veh .
3831 - 3842 , july 2010 .
w. liao , t. chang , w. ma , and c. chi , `` qos - based transmit beamforming in the presence of eavesdroppers : an optimized artificial - noise - aided approach , '' _ ieee trans .
1202 - 1216 , march 2011 .
g. geraci , h. s. dhillon , j. g. andrews , j. yuan , i. b. collings , `` physical layer security in downlink multi - antenna cellular networks , '' _ ieee trans .
62 , no . 6 , pp .
2006 - 2021 , jun . 2014 .
g. geraci , m. egan , j. yuan , a. razi , and i.b .
collings , `` secrecy sum - rates for multi - user mimo regularized channel inversion precoding , '' _ ieee trans .
3472 - 3482 , nov .
g. geraci , j. yuan , and i. b. collings , `` large system analysis of linear precoding in miso broadcast channels with confidential messages,''__ieee journal on sel .
areas in commun . _
9 , pp . 1660 - 1671 , sept .
2013 .
v. k. nguyen and j. s. evans , `` multiuser transmit beamforming via regularized channel inversion : a large system analysis , '' in _ proc . ieee global communications conference _ , new orleans , lo , us , pp . 1 - 4 , dec . 2008
. s. zarei , w. gerstacker , r. r. muller , and r. schober , `` low - complexity linear precoding for downlink large - scale mimo systems , '' in _ proc .
personal , indoor and mobile radio commun .
( pimrc ) _ , sept .
2013 .
a. kammoun , a. mller , e. bjrnson , and m. debbah , `` linear precoding based on polynomial expansion : large - scale multi - cell mimo systems , '' _ ieee journal of sel .
topics in sig .
_ , vol . 8 , no .
861 - 875 , oct . 2014 .
j. evans and d. n. c. tse , `` large system performance of linear multiuser receivers in multipath fading channels , '' _ ieee trans .
inform . theory _
46 , no . 6 , pp . 2059 - 2018 , sept .
r. hunger , `` floating point operations in matrix - vector calculus , '' technische universitt mnchen , associate institute for signal processing , tech . rep . , 2007 .
a. m. tulino and s. verdu , `` random matrix theory and wireless communications , '' _ foundations and trends in communications and information theory _ , vol . 1 , no
. 1 , pp . 1 - 182 , jun . 2004 . | in this paper , we consider secure downlink transmission in a multi - cell massive multiple - input multiple - output ( mimo ) system where the numbers of base station ( bs ) antennas , mobile terminals , and eavesdropper antennas are asymptotically large .
the channel state information of the eavesdropper is assumed to be unavailable at the bs and hence , linear precoding of data and artificial noise ( an ) are employed for secrecy enhancement .
four different data precoders ( i.e. , selfish zero - forcing ( zf)/regularized channel inversion ( rci ) and collaborative zf / rci precoders ) and three different an precoders ( i.e. , random , selfish / collaborative null - space based precoders ) are investigated and the corresponding achievable ergodic secrecy rates are analyzed .
our analysis includes the effects of uplink channel estimation , pilot contamination , multi - cell interference , and path - loss .
furthermore , to strike a balance between complexity and performance , linear precoders that are based on matrix polynomials are proposed for both data and an precoding .
the polynomial coefficients of the data and an precoders are optimized respectively for minimization of the sum mean squared error of and the an leakage to the mobile terminals in the cell of interest using tools from free probability and random matrix theory .
our analytical and simulation results provide interesting insights for the design of secure multi - cell massive mimo systems and reveal that the proposed polynomial data and an precoders closely approach the performance of selfish rci data and null - space based an precoders , respectively . |
the 6df galaxy survey ( 6dfgs ; www.aao.gov.au/local/www/6df ; jones et al . 2004 , 2005 ) is a redshift and peculiar velocity survey of galaxies in the local universe .
the observations for the survey were obtained during 20012006 using the uk schmidt telescope and the 6df spectrograph ( watson et al .
the 6dfgs covers 92% of the southern sky with @xmath2 .
its primary sample is drawn from the 2mass extended source catalog ( xsc ; jarrett et al .
2000 ) and consists of galaxies with @xmath3 ; for this sample the redshift completeness is 88% .
it also includes secondary samples complete down to @xmath4 , @xmath5 ( from the 2mass xsc ) and @xmath6 , @xmath7 ( from the supercosmos sky survey ; hambly et al .
the 6dfgs peculiar velocity survey uses the fundamental plane relation to derive distances and velocities for about 15,000 bright early - type galaxies .
the 6dfgs database comprises 137k spectra and 124k galaxy redshifts , plus photometry and images .
the final data release will be made public in august 2007 ( jones et al . in prep . ; see www-wfau.roe.ac.uk/6dfgs ) . as well as its intended purpose as a survey of the structure and motions in the local universe
, the 6dfgs also provides a benchmark sample for studying the properties of the low - redshift galaxy population .
here we present some preliminary results from an analysis of the stellar populations in a sample of about 6000 galaxies with high - quality spectra from the 6dfgs second data release ( dr2 ; jones et al .
velocity dispersions ( @xmath8 ) have been measured for these galaxies using the cross - correlation technique of tonry & davis ( 1979 ) .
comparisons with other high - quality samples show good agreement and imply the 6dfgs dispersions have a median error of 10.9% ( campbell et al .
, in prep . )
stellar population parameters for these galaxies have been derived from lick index measurements for approximately 15 indices covering the spectral range 40005500 ( proctor et al .
, in prep . ) .
ages , metallicities and @xmath0-element over - abundances were determined by comparing these index measurements to the simple stellar population models of korn , maraston & thomas ( 2005 ) using the @xmath9-fitting method introduced by proctor & sansom ( 2002 ) .
the method rejects those indices that give poor fits to the models , and this excludes h@xmath10 for galaxies both with and without h@xmath0 emission . for passive galaxies without h@xmath0 emission ,
the uncertainties in the derived ages are typically about 0.17 dex in age , 0.15 dex in [ z / h ] and 0.10 dex in [ @xmath0/fe ] . for galaxies that do show significant h@xmath0 emission , the fits also reject h@xmath11 and h@xmath12 , which results in somewhat larger uncertainties .
the distribution of age and metallicity is shown in figure [ fig : agez ] both for passive galaxies ( those with no detected h@xmath0 emission ) and for star - forming galaxies ( those in which h@xmath0 emission is detected ) .
in general , the passive galaxies have metallicities in the range [ z / h]=@xmath13@xmath14 and luminosity - weighted ages of 314gyr , while the star - forming galaxies have solar or higher metallicities and ages less than 3gyr .
both samples show a trend of increasing metallicity for younger galaxies that is substantially stronger than the residual correlation between the errors in age and metallicity .
if we just consider the passive galaxies , then as well as the correlation between age and metallicity , there are also correlations between metallicity and velocity dispersion and between age and [ @xmath0/fe ] , as shown in the left and middle panels of figure [ fig : trends ] : more massive galaxies tend to have higher metallicities , while older galaxies tend to have larger @xmath0-element over - abundances .
there is also a downsizing relation between age and velocity dispersion , in which the age of the _ youngest _ galaxies decreases with decreasing velocity dispersion , as indicated in the right panel of figure [ fig : trends ] .
this latter relation would appear to be the projection in the age@xmath15 plane of the [ z / h]@xmath15 and age[@xmath0/fe ] relations .
/fe ] ( middle panel ) ; there is also a downsizing relation between age and velocity dispersion ( right panel ) . ]
it is interesting to examine the effects of age and metallicity on the mass - to - light ratios for galaxies in different passbands .
if we consider only old galaxies ( greater than 10gyr ) , then the left panel of figure [ fig : mtol ] shows the trends of @xmath1 in the @xmath16 , @xmath17 and @xmath18 bands against both dynamical mass and @xmath18-band luminosity .
two points are worth noting : ( i ) the trends with dynamical mass are steeper than those with luminosity in all passbands , and ( ii ) the trends with both mass and luminosity are steeper in bluer passbands . as well as eliminating age effects by selecting only old galaxies , we can also attempt to remove the dependence on metallicity in these relations
. the right panel of figure [ fig : mtol ] shows as thick solid lines the same binned relations as in the left panel ; it also shows these relations after correcting the mass - to - light ratio for differences in metallicity ( dashed lines ) .
these corrections are determined from the luminosities predicted by bruzual & charlot ( 2003 ) models for stellar populations having the metallicity ( based on the mass metallicity relation ) and age of each mass or luminosity bin , and converting to the predicted luminosity for a stellar population with the same metallicity as the highest mass or luminosity bin .
the corrected mass - to - light ratios resulting from this procedure have almost identical dependences on mass or luminosity , regardless of passband , as can be seen by comparing the dashed lines in the right panel of figure [ fig : mtol ] with the thin solid lines ( which are just the corrected relations for the @xmath18-band ) .
thus the variations of @xmath1 with passband for this sample of old galaxies can be fully accounted for by the mass metallicity relation .
it is worth noting that the correction has almost no effect in the @xmath18-band , underlining the minimal impact of metallicity on @xmath18-band luminosities and the usefulness of this passband for studies of @xmath1 and the fundamental plane .
) ratios of old galaxies ( age @xmath1910gyr ) in @xmath16 , @xmath17 and @xmath18 bands is plotted against dynamical mass and @xmath18-band luminosity .
solid lines are means in five bins along the @xmath20-axis ; the rms scatter is indicated by dashed lines . the average error on individual points is shown at the bottom right of each panel . at right ,
the mean @xmath1 ratios are shown before ( solid line ) and after ( dashed line ) correction for the effect of the mass metallicity relation on the luminosity .
the thin solid lines in each panel show the @xmath18-band relation for the purposes of comparison.,title="fig:",scaledwidth=48.0% ] ) ratios of old galaxies ( age @xmath1910gyr ) in @xmath16 , @xmath17 and @xmath18 bands is plotted against dynamical mass and @xmath18-band luminosity .
solid lines are means in five bins along the @xmath20-axis ; the rms scatter is indicated by dashed lines .
the average error on individual points is shown at the bottom right of each panel . at right ,
the mean @xmath1 ratios are shown before ( solid line ) and after ( dashed line ) correction for the effect of the mass metallicity relation on the luminosity .
the thin solid lines in each panel show the @xmath18-band relation for the purposes of comparison.,title="fig:",scaledwidth=48.0% ]
the 6dfgs provides a large sample of galaxies with spectra of sufficiently high quality and spectral range to allow reliable measurements of both line indices and velocity dispersions .
this sample can be used to explore the correlations between mass , age , metallicity and [ @xmath0/fe ] .
some preliminary results from such an analysis are presented here ; a more detailed study will appear in proctor et al.(2007 , in prep ) .
we find that age and metallicity are significantly anti - correlated ( younger galaxies are more metal - rich ) for both passive and star - forming samples .
passive galaxies show the well - known strong correlation between mass and metallicity ( more massive galaxies are more metal - rich ) , but also a strong correlation between age and @xmath0-element over - abundance ( older galaxies have higher over - abundances ) .
these two effects combine to produce a downsizing relation between age and mass , such that the age of the _ youngest _ galaxies decreases with decreasing mass . for old ( @xmath1910gyr ) passive galaxies ,
the different trends of @xmath1 with mass and luminosity in different passbands result from the differential effect of the mass metallicity relation on the luminosities in each passband . | the 6df galaxy survey provides a very large sample of galaxies with reliable measurements of lick line indices and velocity dispersions .
this sample can be used to explore the correlations between mass and stellar population parameters such as age , metallicity and [ @xmath0/fe ] .
preliminary results from such an analysis are presented here , and show that age and metallicity are significantly anti - correlated for both passive and star - forming galaxies .
passive galaxies have strong correlations between mass and metallicity and between age and @xmath0-element over - abundance , which combine to produce a downsizing relation between age and mass . for old passive galaxies ,
the different trends of @xmath1 with mass and luminosity in different passbands result from the differential effect of the mass metallicity relation on the luminosities in each passband .
future work with this sample will examine the fundamental plane of bulge - dominated galaxies and the influence of environment on relations between stellar population parameters and mass . |
the presence of metals in the intergalactic medium ( igm ) at all red - shifts @xcite , the present - day galaxy mass - metallicity relationship @xcite , and @xmath11-kpc - scale holes in the igm at red - shift @xmath12 @xcite have all been interpreted as the consequence of energetic metal - bearing outflows from galaxies .
there is no doubt that at least one form of galactic wind exists : starburst galaxies are commonly observed to host supernova - driven galactic - scale winds @xcite in both the local and high redshift universe @xcite .
these superwinds are phenomenologically complex and are comprised of multiple different gas phases moving at different velocities @xcite .
however , the very hot plasma long hypothesized to be the prime mover in superwinds @xcite , often called the wind fluid , has only recently been directly observed in the archetypal superwind associated with the starburst galaxy m82 @xcite . the most fundamental long - standing uncertainty in our understanding of the nature of superwinds and their larger scale role is the value of the energy per particle of the metal - enriched wind - fluid that drives an individual superwind . both supernovae ( sne ) and
stellar winds from massive stars are responsible for supplying the kinetic energy that drive superwinds ( stellar winds only contribute @xmath13% of the total mechanical power of a starburst , but they may supply up to half of the gas mass returned to the ism , see @xcite ) .
what fraction of the @xmath14 of kinetic energy released per star that ultimately goes supernova is actually available to drive bulk motions of the ism ?
as it is understood that sne dominate the energy return this is often termed the sn thermalization efficiency or the sn feedback efficiency , even though stellar winds do play some small part energetically .
is the metal - enriched mix of sn and stellar wind ejecta diluted and cooled by mixing with the ambient ism , i.e. is the wind - fluid mass - loaded
? if this ejecta is to escape the galaxy and enter the igm then its specific energy must exceed its initial gravitational potential energy , and furthermore the total energy in the wind - fluid must be sufficient to perform the mechanical work necessary to move the interstellar and halo gas that lies between the starburst region and the igm @xcite . for moderate to high thermalization efficiency ( efficiencies @xmath15 10% ) , the wind - fluid is expected to have an initial temperature within the starburst region in the range 10 100 million kelvin @xcite , even if it has been lightly mass - loaded with cold ambient gas .
such temperatures are far hotter than the neutral , warm ionized or soft x - ray emitting plasmas that are the most easily observed phases within superwinds @xcite .
however it has only recently become technically possible to detect the diffuse hard x - ray emission that would be associated with a plasma in the temperature range @xmath16 ( k ) @xmath17 .
the starburst regions of even the nearest starburst galaxies only subtend @xmath18
@xmath19 on the sky and are heavily populated with x - ray - luminous compact objects . it was only with the advent of @xmath20-spatial resolution x - ray astronomy in 1999 following the launch of the _ chandra _ x - ray observatory that made it possible for @xcite to resolve out the point source emission in the archetypal nearby starburst galaxy m82 and reveal residual diffuse hard x - ray emission , although of uncertain origin .
it is important to distinguish between the material that actually drives a superwind , and the other more easily observed phases of swept - up and entrained gas that are part of the wind .
we define the wind - fluid as the material whose high thermal and/or ram pressure drives the superwind .
the wind fluid certainly comprises a significant fraction of the merged sn ejecta and massive star stellar wind material , probably admixed ( mass - loaded ) with some ambient gas from with starburst region .
although other gas phases may dominate the total gas mass budget of a superwind , they are less important with respect to the issue of energy and metal transfer into the igm .
furthermore the ejection of these phases is less than certain , as the measured warm neutral and warm ionized medium ( wnm , wim ) gas velocities are typically comparable to the galactic escape velocities @xcite . unfortunately even the most recent attempts to include superwinds in cosmological models of galaxy formation use only simplified single - phase winds , often with energetics and masses based on the observed wnm / wim properties of superwinds ( see e.g. * ? ? ?
* ; * ? ? ?
these models do not properly address the physics of hot - phase metal transport in addition to having overly low values of the energy per particle , which casts the reliability of their conclusions into doubt .
attempts have been made to use measurements of the temperature of the extended soft x - ray emitting plasmas ( @xmath21 @xmath22 k ) in starburst galaxies to constrain whether the hot gas will escape the host galaxy ( see e.g. * ? ? ?
* ) , but such an approach neglects both the significant kinetic energy expected to be associated with this material and that the soft x - ray emitting plasma is an indirect probe of the properties of the true wind - fluid @xcite .
the detection of apparently thermal diffuse hard x - ray emission of temperature @xmath23 k in the starburst region of m82 in the 33 ks - long _ chandra _ acis - i observations of @xcite suggested it was possible to directly observe and constrain the properties of the wind fluid in a starburst superwind .
we recently obtained and analyzed a new 18 ks _ chandra _ acis - s observation of m82 , and also reprocessed archival _
acis - i and _ xmm - newton _ observations ( total exposure of 48 ks with acis - i , 75 ks with _ xmm - newton _ pn and 102 ks with each mos instrument ) . we confirmed the presence of both diffuse continuum and diffuse @xmath24 fe he@xmath0 line emission within the central starburst region ( * ? ? ?
* hence forth referred to as paper i ) .
having assumed that the line - emitting plasma is metal - enriched to @xmath25 , we found that the observed iron line luminosity of @xmath26
@xmath27 was consistent with the properties of the wind fluid expected given m82 s star formation rate , provided that the sn thermalization efficiency was high .
however the diffuse hard x - ray continuum luminosity of @xmath28 was too luminous to be the thermal bremsstrahlung associated with the wind fluid given the parameters assumed when interpreting the iron line emission .
in this paper we consider in greater detail what constraints the observed hard x - ray properties of m82 can place on the properties on the superwind and the wind - fluid , in particular the efficiency of sn heating ( @xmath29 ) and the degree of central mass - loading ( @xmath30 ) .
ultimately these determine the energy per particle of the wind fluid .
we restrict analysis to the starburst region within @xmath31 pc of the center of m82 .
we calculate the x - ray emission using both multi - dimensional hydrodynamical simulations and the 1-dimensional analytical @xcite model ( hence forth referred to as the cc model ) . by using the simple but relatively accurate cc model we can explore the predicted x - ray properties of the wind fluid in models covering a wider range of parameter space than if we used multi - dimensional hydrodynamical simulations alone .
a variety of recent papers have presented theoretical calculations of the broad - band x - ray emission from individual star - forming clusters , using either the adiabatic cc model , or the non - adiabatic solution of silich and collaborators @xcite , with the aim of seeing if the particular model used can explain the observed data .
our motivation differs from these papers , in that our aim is to use theoretical models in combination with observational data on m82 to constrain fundamental properties of the superwind . in addition
our approach differs from these previous papers in its use of x - ray line luminosities rather than purely broad - band x - ray luminosities . in reality
the broad - band emission might be dominated by non - thermal continuum sources .
furthermore line emission is more sensitive than the thermal continuum luminosity to interesting parameters such as the plasma temperature .
the observational constraints derived in paper i on the diffuse hard x - ray emission within @xmath32 pc of the nucleus of m82 from the _ chandra _ and _ xmm - newton _ observations are summarized in table [ tab : paper1 ] .
these comprise a robust 6.69 kev iron line detection ( @xmath33 in significance ) , an upper limit on the 6.96 kev iron line , and from the _ chandra _ observations the luminosity of the diffuse continuum .
for the _ chandra_-based observations the values refer to the diffuse emission alone , after the exclusion of detected point - like x - ray sources and corrections for undetected point sources based on the observed point source luminosity function . these luminosities also include 32% and 20% upward corrections to the luminosities derived from the acis - s and acis - i data to account for regions excluded from the spectra due to remove the emission from detected point sources .
for the _ xmm - newton _ observation the line luminosities include all emission within @xmath32 pc , but the contribution to the line luminosities from point - like sources is expected to be small in this particular observation .
lrrrrr epic & 0206080101 & & @xmath34 & @xmath35 & @xmath36 + _ chandra _ acis - s & 2933 & @xmath37 & @xmath38 & @xmath39 & @xmath40 + _ chandra _ acis - i & 361 + 1302 & @xmath41 & @xmath42 & @xmath43 & @xmath44 + note that the region of bright diffuse hard x - ray emission studied in this paper is associated with the nuclear starburst region . as can be seen from fig .
[ fig : m82_3color ] this region is much smaller than the full 12-kpc - scale superwind that is most often studied in optical @xcite or soft x - ray emission @xcite .
as discussed in paper i the nature of the diffuse broad - band continuum is not clear .
if the continuum is thermal then the relative strength of the iron line emission to the observed continuum is much lower than expected for a plasma of solar elemental abundances , let alone the iron - enriched plasma expected from recent core - collapse sne .
we demonstrated that a wind mass - loaded with additional ambient gas could produce the total broad - band x - ray luminosity that we observe , but the required degree of mass loading could not sufficiently reduce the relative strength of the iron line to the continuum .
thus the best fit bremsstrahlung models for the continuum do not necessarily provide meaningful estimates of the temperature of the iron - emitting plasma , and should be interpreted with caution .
power law spectral models provide a marginally better fit to the continuum than do thermal bremsstrahlung models .
thus a non - thermal process such as inverse compton x - ray emission might be responsible for the majority of the diffuse hard x - ray continuum .
the x - ray luminosity associated with inverse compton emission depends sensitively on the unknown magnetic field strength within the starburst region . in paper
i we estimated it might account for @xmath45% of the observed diffuse x - ray luminosity in the @xmath1 8 kev energy band , although other authors have in the past given higher luminosity estimates for inverse compton emission from m82 @xcite . the upper limits on the iron 6.96 kev / 6.69 kev line flux ratio of @xmath46 place upper limits on the temperature of @xmath47 if the plasma is in collisional equilibrium ( see table [ tab : temp_estimates ] ) .
the coulomb relaxation and ionization equilibrium timescales presented in paper i are much shorter than the timescale for plasma to flow out of the starburst region , so the assumption of collisional ionization equilibrium is reasonable .
lll fe ly@xmath0 / fe he@xmath0 & xmm - newton & @xmath48 + fe ly@xmath0 / fe he@xmath0 & acis - s & @xmath49 + fe ly@xmath0 / fe he@xmath0 & acis - i & @xmath50 + ca he@xmath0 / fe he@xmath0 & xmm - newton & @xmath51 + ca he@xmath0 / fe he@xmath0 & acis - s & @xmath52 + ar he@xmath0 / fe he@xmath0 & xmm - newton & @xmath53 + ar he@xmath0 / fe he@xmath0 & acis - s & @xmath54 + s he@xmath0 / fe he@xmath0 & xmm - newton & @xmath55 + s he@xmath0 / fe he@xmath0 & acis - s & @xmath56 + ( ca ly@xmath0 + ca he@xmath0 ) / fe he@xmath0 & xmm - newton & @xmath57 + ( ca ly@xmath0 + ca he@xmath0 ) / fe he@xmath0 & acis - s & @xmath58 + ( ar ly@xmath0 + ar he@xmath0 ) / fe he@xmath0 & xmm - newton & @xmath59 + ( ar ly@xmath0 + ar he@xmath0 ) / fe he@xmath0 & acis - s & @xmath60 + ( s ly@xmath0 + s he@xmath0 ) / fe he@xmath0 & xmm - newton & @xmath61 + ( s ly@xmath0 + s he@xmath0 ) / fe he@xmath0 & acis - s & @xmath59 + s he@xmath0 / ca he@xmath0 & xmm - newton & @xmath62 + s he@xmath0 / ca he@xmath0 & acis - s & @xmath63 + s he@xmath0 / ar he@xmath0 & xmm - newton & @xmath64 + s he@xmath0 / ar he@xmath0 & acis - s & @xmath65 + ar he@xmath0 / ca he@xmath0 & xmm - newton & @xmath66 + ca ly@xmath0 / ca he@xmath0 & xmm - newton & @xmath67 + ca ly@xmath0 / ca he@xmath0 & acis - s & @xmath68 + ar ly@xmath0 / ar he@xmath0 & xmm - newton & @xmath69 + ar ly@xmath0 / ar he@xmath0 & acis - s & @xmath70 + s ly@xmath0 / s he@xmath0 & xmm - newton & @xmath71 + s ly@xmath0 / s he@xmath0 & acis - s & @xmath72 + bremsstrahlung fit to diffuse continuum & acis - s & @xmath73 + bremsstrahlung fit to diffuse continuum & acis - i & @xmath74 + other lines in the @xmath75 ( kev ) @xmath76 hard x - ray band may in principle be used to constrain the plasma temperature in a starburst region . in particular the helium - like and hydrogen - like ions of sulphur ( at @xmath77 kev and @xmath78 kev respectively ) , argon ( @xmath79 kev and @xmath80 kev ) and calcium ( @xmath81 kev and @xmath82 kev ) can be detected in ccd spectra from the _ chandra _ acis and _ xmm - newton _ epic instruments .
these lines are most intense in plasmas with temperatures in the range @xmath83 7.5 , somewhat lower than the optimal temperature for the iron lines , but this might be advantageous in diagnosing starbursts where mass loading is significant and/or the efficiency of sn thermalization is low . in paper i we specifically excluded data at the expected energies of the s , ar and ca lines from the fitting of the m82 nuclear region spectra . repeating the analysis described paper i but now fitting all the data in the energy range @xmath84 9.0 kev in both the _ xmm - newton _ and _ chandra _ acis - s spectra yields the line detection significances and line luminosities given in table [ tab : s_ar_ca ] .
we fit for the intensity of the helium - like and hydrogen - like s , ar and ca lines , having fixed the line energies at the expected values .
these lines are faint , but in some cases likely to be real . as an example we display the best - fit to the _ xmm - newton _
epic pn + mos spectra in fig .
[ fig : xmm_spectrum ] .
detection significance was assessed using monte carlo methods with several thousand simulated spectra . where the line detection significance was @xmath85 we give the 99% upper limit on the line luminosity ( and not the best - fit value ) .
acis - i spectra gave similar results to those from acis - s , but as the line detections are less significant than in the _ xmm - newton _ data there is little reason to present the values .
llrlll he@xmath0 &
s xv & 2.45 & xmm - newton & @xmath86 & @xmath87 + s ly@xmath0 & s xvi & 2.62 & xmm - newton & @xmath86 & @xmath88 + ar he@xmath0 & ar xvii & 3.13 & xmm - newton & @xmath86 & @xmath89 + ar ly@xmath0 & ar xviii & 3.32 & xmm - newton & @xmath90 & @xmath91 + ca he@xmath0 & ca xix & 3.90 & xmm - newton & @xmath92 & @xmath93 + ca ly@xmath0 & ca xx & 4.11 & xmm - newton & @xmath94 & @xmath95 + s he@xmath0 & s xv & 2.45 & acis - s & @xmath96 & @xmath97 + s ly@xmath0 & s xvi & 2.62 & acis - s & @xmath98 & @xmath99 + ar he@xmath0 & ar xvii & 3.13 & acis - s & @xmath100 & @xmath101 + ar ly@xmath0 & ar xviii & 3.32 & acis - s & @xmath102 & @xmath103 + ca he@xmath0 & ca xix & 3.90 & acis - s & @xmath104 & @xmath105 + ca ly@xmath0 & ca xx & 4.11 & acis - s & @xmath102 & @xmath106 + in the _ xmm - newton _ epic pn + mos spectra of m82 s nucleus the equivalent of the he@xmath0 line from highly ionized s , ar , ca and fe are all detected with significances @xmath107 , while in the shorter ( and hence lower s / n ) _ chandra _ acis - s spectrum of the nuclear diffuse emission the s he@xmath0 and fe he@xmath0 are detected at @xmath107 while the ar he@xmath0 and ca he@xmath0 lines are only detected at @xmath108 . in both instruments si ly@xmath30 is also detected at high significance , but we shall ignore this lower ionization state line in the analysis to follow . the only ly@xmath0-like line detected at @xmath107 is s ly@xmath0 in the _ xmm - newton _ epic spectra . using the robust line detections ( i.e. those with significance @xmath109 ) we can estimate the plasma temperature from the ratio of s ly@xmath0 to s he@xmath0 , and from between different elements from the ratio of the various he@xmath0 lines ( assuming their relative elemental abundances are solar ) .
these temperature estimates are presented in table [ tab : temp_estimates ] .
the s ions tightly constrain the ion temperature to @xmath110 , while the ratio of the s , ar or ca he@xmath0 line to the fe he@xmath0 line yields @xmath111 .
the ratio of s he@xmath0 to either ar he@xmath0 or ca he@xmath0 yields a temperature estimate of @xmath112 .
this pattern of the inferred temperature increasing as higher ionization state ions are used ( which are of course more sensitive to higher temperatures ) indicates that either the plasma is multi - phase , and/or the the plasma is not in collisional ionization equilibrium ( cie ) .
of these explanations we favor the former , as multi - dimensional numerical hydrodynamical simulations predict a broad range of gas temperatures below @xmath113 ( see e.g. * ? ? ?
furthermore we discussed in paper i reasons to expect the fe - emitting plasma to be in or close to cie .
unfortunately these results do not strongly constrain the temperature of the fe - emitting plasma .
one possibility is that the iron line emission arises in the very hottest phases in the superwind , i.e. at @xmath114 , and that most of the s , ar and ca emission comes from somewhat cooler components associated with the wind / ism interaction and the larger scale soft x - ray emission in the m82 superwind . in this case ,
given the possible soft x - ray contribution , the s , ar and ca observed luminosities must be treated as upper limits to the s , ar and ca line emission coming from the volume filling wind - fluid . alternatively the wind fluid is relatively cool , with a temperature corresponding to that measured from the ar and ca lines , i.e. @xmath111 , and the fe lines are produced by some unexplained non - thermal mechanism . in either case a successful model of m82 s nuclear emission must reproduce the observed hard x - ray iron line emission without over - producing the broad - band continuum and s , ar and ca line emission .
hard x - ray emission in multi - dimensional hydrodynamical models of superwinds is dominated by the thermal emission from starburst region itself @xcite .
if the starburst region is spherical then the plasma properties of this region obtained from hydrodynamical simulations agree very well with the simple one - dimensional radially - symmetric analytical cc model , in which the mechanical energy and returned gaseous ejecta from core - collapse supernovae are injected at a uniform rate only within a radius @xmath115 , and radiative energy losses are assumed to be negligible .
gravity and any surrounding ambient gaseous medium are ignored . as the feedback parameters we wish to constrain are related to the observable x - ray emission via the plasma properties in the starburst region it is worthwhile for us to explicitly reproduce the analytical solution obtained by @xcite .
we consider the conditions under which the adiabatic assumption inherent in the cc model may be invalid in appendix [ app : radiative_case ] , but we will show that for most reasonable parameters associated with the m82 starburst ( described below ) the radiative energy losses are not significant .
in real galaxies starburst regions are better described as disks or ring of star - formation rather than spherical regions . to explore the predicted x - ray emission for our default model of the m82 s disk - like starburst we begin by using 2-dimensional hydrodynamical simulations in cylindrical symmetry to predict the plasma properties within the a radius of 500 pc of the center of the galaxy . however , with appropriate scaling ( see appendix [ app : non_spherical ] ) , it is possible to use the 1-dimensional cc model with a spherical starburst to predict the plasma properties of the wind fluid within a non - spherical starburst region to high accuracy , and to predict the hard x - ray emission from even larger regions with an accuracy no worse than the uncertainties in the observed x - ray luminosities .
these scaled cc models are considerably faster to calculate than a full hydrodynamical model , so we will use them to explore how the constraints on the thermalization efficiency and mass loading change if we adopt a different range of parameters for the m82 starburst .
this approach , in particular the use of the scaled cc model , is applicable only to the study of the wind fluid in the vicinity of the starburst region .
it is not intended for use as a model of an entire superwind as it neglects the interaction between the wind fluid and the multiphase ambient ism .
the simplification of neglecting the ambient ism works because the only significant source of thermal hard x - ray emission in a superwind is the wind fluid itself , in particular the wind fluid within the starburst region .
the interactions between the wind fluid and the multiphase ism are important in the large scale wind and for its emission at soft x - ray and longer wavelengths ( e.g. * ? ? ?
* ; * ? ? ?
* ) , but with the exception of altering the net efficiency of sn+sw heating and any mass - loading of the wind fluid these interactions are not expected to alter the properties of the wind fluid within the starburst region to any significant degree given its energetic dominance .
the hydrodynamical simulations were performed using the multi - dimensional numerical hydrodynamics code vh-1 @xcite .
simulations were performed in 2-dimensional cylindrical symmetry with a cell size of 2 pc over 500 by 500 cell computational grid . reflecting boundary conditions were imposed along the @xmath116 and @xmath117 axes , with
inflow / outflow boundary conditions are the other edges of the computational grid .
energy and mass are injected within the starburst region in the manner described in @xcite . in general the starburst region
was modeled as disk - like region of the size described in [ sec : theory : applying ] .
no gravitational forces or radiative cooling are considered for the purposes of consistency with the analytical cc model described below .
the computational grid was initially populated with a tenuous ism of density @xmath118 and temperature @xmath119 k. the simulations were run until the wind had blown the initial ism off the grid and reached a steady - state solution , which typically occurred @xmath120 2 myr after the onset of the simulation .
gas properties and x - ray luminosities are measured at a time of 3 myr within a spherical region of radius of 500 pc to closely match the observed region discussed in [ sec : observations ] .
for completeness we present the analytical flow solution derived by cc , and how we use it to derive the fluid properties as a function of radial position .
more complicated , but also more generalized , methods of deriving the solution are presented in @xcite and @xcite .
the basic model is completely specified by only three parameters : the energy and mass injection rates ( @xmath121 ) ; and the radius of the starburst region ( @xmath115 ) within which the energy and mass are injected .
cc present the stationary solution to the hydrodynamical equations as a function of the mach number @xmath122 , where @xmath123 is the local speed of the flow and the sound speed is @xmath124 .
the pressure @xmath125 , density @xmath126 and adiabatic index @xmath127 have their normal meanings . at radii @xmath128 mass and energy
are injected at a uniform constant rate per unit volume @xmath129 and @xmath130 , where the starburst volume @xmath131 , and the solution for the mach number is @xmath132 this approach to the energy and mass injection implicitly assumes that the exact manner in which stars return mechanical energy to the ism can be ignored due to rapid local thermalization and mixing , i.e. individual stellar winds and sn blast waves collide and merge on spatial scales small compared to the size of the starburst region . if radiative energy losses are to be small then these collisions must occur with a time shorter than the cooling time scale for stellar wind bubbles or supernova remnants ( snrs ) .
theoretical investigations of some of these issues may be found in @xcite , @xcite , @xcite , @xcite and @xcite . at larger radii @xmath133
then @xmath134 , and the solution is @xmath135 for any given fractional radius @xmath136 the mach number is found using a numerical root - finding technique ( in our case the newton - raphson method ) on the appropriate equation . to obtain the density at any radius @xmath128 we integrate equation 1 of cc to obtain @xmath137 within the starburst region the gas density , pressure and temperature vary only slowly with radius , with maxima at @xmath138 .
the central density is effectively determined by the rate at which the injected material can flow out of the starburst region , @xmath139 where @xmath140 is the terminal velocity of the wind at large radius , and the central temperature is set by the energy injected per particle .
more specifically , the central temperature and density are @xmath141 and @xmath142 for @xmath143 . at radii @xmath144 the mass flow rate through the spherical surface of radius @xmath116
is equal to the total mass injection rate within the starburst region , so the density is @xmath145 to derive the local velocity of the flow @xmath123 we must first determine the speed of sound .
if we make the assumption that the flow is adiabatic then we can obtain the speed of sound at any point ( @xmath128 or @xmath144 ) , as we know the mach number and the initial energy per particle is conserved .
specifically we solve equation 3 of cc and substitute for @xmath126 using equation [ equ : cc_rho_smallr ] , to obtain @xmath146 the speed of the flow at any point is then simply @xmath147 . for @xmath128
the flow is subsonic , with @xmath148 .
the sonic point is reached at @xmath149 , beyond which the flow velocity rapidly tends toward the terminal velocity @xmath150 . with the density @xmath126 and
the speed of sound @xmath151 determined at every point within the flow the thermal pressure @xmath125 is also now known .
the kinetic temperature @xmath152 , where @xmath153 is the mean mass per particle for a fully ionized plasma of solar composition .
numerically we derive the gas density , temperature and velocity as a function of radius , iterating outward in thin radial shells .
the numerical implementation we use is based on code originally created by ian stevens @xcite .
this semi - analytical code reproduces the density and pressure to within 0.1% of the values predicted by the full 2-d hydrodynamical simulation using the vh-1 code for a spherical starburst region .
the scaling used to approximate a non - spherical starburst region using the spherical cc model is described in appendix [ app : non_spherical ] .
figure [ fig : radial_solution ] presents the radial variation in several fluid variables and associated derived quantities for one particular set of input parameters , as predicted using our 1-dimensional cc model code .
the flow times for the region of interest ( @xmath154 pc ) are short ( a few myr ) compared to the lifetime of the massive stars powering the flow ( @xmath155 myr ) .
the properties of the starburst region will very rapidly approach those of the steady - state cc solution after any change in the energy and mass injection rate .
nevertheless , in order to apply the cc model to m82 we must determine the history ( i.e. age and star formation rate as a function of time ) of the starburst event in order to derive the current mechanical energy and mass injection rates and the elemental abundances in the injected gas .
furthermore , determining the degree to which the m82 starburst event can be treated as one the simple limiting cases of either a single instantaneous burst ( sib ) or as continuous star formation ( csf ) is important , in particular for comparison to starburst99 models @xcite .
recent observational studies have greatly expanded our knowledge of m82 s global star formation history since its close encounter with m81 approximately 0.6 1 gyr ago ( see e.g. * ? ? ?
nevertheless significant gaps still remain in our understanding of the history of the nuclear starburst .
exterior to the central kiloparsec star formation appears to have peaked @xmath156 myr ago in a burst of comparable intensity to the current nuclear starburst event , but star formation effectively ceased in the disk @xmath157 30 myr ago @xcite . within the central kiloparsec ( projected separations from the nucleus of @xmath154 pc )
the majority of recent studies have concentrated on the @xmath158 200 compact star clusters found there @xcite .
the brightest @xmath157 star clusters typically appear to have ages @xmath159 @xmath160 myr @xcite , although two massive clusters ( m82-f and m82-l ) at the edge of this region have ages of @xmath161 myr @xcite . @xcite
present a detailed analysis of the nir emission ( 1 45@xmath162 ) from a variety of regions within @xmath163 pc , and find that two bursts ( of ages @xmath159 and @xmath13 myr respectively ) are required in each region to simultaneously explain the strength of both nebular emission features and co absorption from red super giant ( rsg ) stars .
this work , and the estimated ages of the brightest star clusters , indicates that the nuclear star formation is not well represented as a single instantaneous starburst event .
note that the cluster - based studies do not provide a complete view of the starburst , as they are mainly observed in regions of lower - than - average extinction 6 magnitudes , see @xcite . ] toward the nucleus @xcite . nor does this work strongly constrain the star formation rate between 10 60 myr ago . old clusters in this age range , of similar mass to the bright clusters , would be considerably fainter and thus harder to detect than the @xmath164 myr - old clusters myr after formation , and that the cluster can simultaneously produce nebular emission and have a substantial number of rsgs . if this is indeed the case then the clusters identified as having ages of @xmath13 myr may be as old as @xmath155 myr . ] , due to the rapid decrease of the ionizing and bolometric output at ages @xmath165 myr .
uv spectroscopic studies of nearby low - obscuration starbursts have shown that the massive star population within the clusters is typically younger the the field massive star population ( see e.g. * ? ? ?
the probable cause for this effect is the gradual dissolution of the massive star clusters on times scales of @xmath13 30 myr ( for a recent review see * ? ? ?
* and references therein ) .
we are not aware of any similar uv study of the field stellar population in m82 s nucleus , so there is no direct evidence for the presence of star formation intermediate in age between the @xmath166 10 myr - old clusters and the @xmath167 myr - old clusters .
nevertheless it remains very plausible that significant star formation continued in between the two periods .
another reason to suspect that the nuclear burst has been active for longer than @xmath13 myr is the dynamical age of the h@xmath0 cap , a large nebular feature associated with the superwind at a height @xmath168 kpc above the mid - plane of the m82 @xcite .
its true space velocity is unknown , but assuming a mean velocity of @xmath169 equivalent to that of the h@xmath0 filaments in the lower wind @xcite it would have taken @xmath157 myr to reach its current location if its origin was within the disk of the galaxy .
we thus choose to treat m82 s nuclear starburst as having experienced a uniform rate of star formation since the onset of the burst @xmath170 years ago , with negligible star formation before that time .
we will treat @xmath170 as a model parameter in the range @xmath160 50 myr . in table
[ tab : burst ] we present the bolometric luminosity , and total sn plus stellar wind energy and mass injection rates ( all per unit star formation rate ) for continuous star formation models calculated using version 5.1 of starburst99 population synthesis code @xcite .
the star formation rate is @xmath171 for a salpeter imf between the mass limits of 1 @xmath172 . for a @xcite imf between the mass limits 0.01
@xmath172 producing the same total mass of massive stars ( mass range 8 @xmath172 ) as in the salpeter imf model , the star formation rate is a factor 1.76 times larger .
the geneva stellar evolutionary tracks for an initial metallicity equal to solar , and with high stellar wind mass loss rates , were used .
lrrrrrr 10 & @xmath173 & @xmath174 & 0.19 & 6.1 & @xmath175 & 1.1 + 20 & @xmath176 & @xmath177 & 0.26 & 5.1 & @xmath178 & 1.3 + 30 & @xmath179 & @xmath180 & 0.29 & 4.7 & @xmath181 & 1.4 + 40 & @xmath182 & @xmath183 & 0.31 & 4.5 & @xmath184 & 1.4 + 50 & @xmath182 & @xmath185 & 0.31 & 4.4 & @xmath186 & 1.3 + the total 8 @xmath187 ir luminosity of m82 is @xmath188 , based on the latest re - calibrated 12
@xmath189 iras fluxes @xcite .
ir emission dominates m82 s spectral energy distribution , so by equating @xmath190 to the starburst bolometric luminosity for each assumed age yields star formation rates @xmath191 between 4 @xmath192 ( salpeter imf ) .
these values and the associated sn plus stellar wind energy and mass injection rates are tabulated in table [ tab : burst ] . by way of comparison the formulae in @xcite , which assume csf with a large age ,
give a star formation rate of @xmath193 for m82 based on @xmath190 , when corrected to the 1 @xmath172 mass range salpeter imf we use .
these star formation rates are not unreasonable . over the time scale of interest
the net mass of stars formed is at maximum @xmath194% of the dynamical mass of @xmath195 within @xmath196 pc @xcite , even if we assume a kroupa imf extending down to a lower mass limit of @xmath197 .
the brightest region of diffuse non - thermal 5 ghz radio emission in m82 subtends an angular extent of @xmath198 ( @xmath199 ) , the same region occupied by the compact radio sources presumed to be young sn remnants @xcite .
this is similar to the size of the region occupied by the brightest star clusters seen in the nir , @xmath200 ( @xmath201 , @xcite ) .
these major - axis angular diameters closely match the base of the superwind as measured by the region of split optical emission lines , which has a diameter of @xmath202 @xcite . if we approximate the starburst region as a flat cylinder of diameter @xmath203 and total height @xmath204 then the starburst volume @xmath205 and surface area @xmath206 in each case is @xmath207 pc@xmath208 , @xmath209 pc@xmath210 ( 5 ghz ) and @xmath211 pc@xmath208 , @xmath212 pc@xmath210 ( nir ) .
the equivalent spherical cc model should then have a starburst region radius of @xmath213 pc or @xmath214 pc ( appendix [ app : non_spherical ] ) .
the starburst in m82 is possibly concentrated along the stellar bar ( see e.g. * ? ? ?
* ; * ? ? ?
* ) , and hence is more geometrically complex than a uniform disk . furthermore star formation may not have occurred simultaneously throughout the full starburst region , but evidence for inwardly or outwardly propagating star formation within @xmath215 pc of the nucleus remains inconclusive @xcite .
exploring the full significance of these effects requires 3-dimensional hydrodynamical simulations that are outside the scope of this paper .
however we can investigate the significance of concentrating the supernova activity within a smaller disk - like region than the two considered above .
we adopt the starburst disk geometry used in @xcite which has @xmath216 pc and @xmath217 pc .
this produces significantly lower starburst volumes and surface areas : @xmath218 pc@xmath208 and @xmath219 pc@xmath210 . the equivalent spherical starburst radius is @xmath220 pc . for the majority of the models we consider we adopt the first starburst region geometry described above : a disk of diameter @xmath221 pc and total height @xmath222 pc ( or @xmath213 pc for the scaled cc models ) .
the effect of the size of the starburst region is explored using the scaled cc model with radii @xmath214 pc and @xmath220 pc .
the most significant contributors to the energy and mass return rates in a starburst event are core - collapse sne and the stellar winds from the massive stars .
stellar winds are thought to contribute an order of magnitude less mechanical energy than the sne ( assuming equal thermalization efficiencies ) , but they may contribute up to @xmath194% of the total mass returned to the ism @xcite .
as described above , we used the starburst99 population synthesis code to find the combined sne plus stellar wind energy and mass injections rates for a continuous burst of star formation as a function of age .
note that the values of @xmath223 and @xmath224 given in table [ tab : burst ] have not been corrected for any initial mechanical energy losses due to radiation before thermalization occurs ( the thermalization efficiency @xmath29 ) or for mass loading by the addition of cold ambient ism ( with a rate @xmath225 ) .
the thermalization efficiency and degree of mass loading are expected to be a function of the local gaseous environment the massive stars find themselves in , as well as the local supernova rate per unit volume and the density of massive stars .
low ambient gas density and/or high sn rate per unit volume should lead to a higher hot gas filling factor , which in turns increases the thermalization efficiency by reducing the initial radiative energy losses @xcite .
however , at very high stellar densities the sn and stellar wind ejecta may itself become dense enough to cause drastic energy losses , e.g. within the central regions of super star clusters @xcite . to account for these factors
we assume that only a fraction @xmath226 of all the sne and stellar winds occur in regions where conditions lead to non - negligible thermalization , and the mean thermalization efficiency in these regions is what we define as @xmath29 .
essentially @xmath226 is a modification to the total star formation rate of the starburst to account for localized spatial variations within the starburst of the sn and stellar wind feedback .
we will refer to @xmath226 as the participation fraction , as it is effectively the fraction of the total star formation that is actually involved in driving the superwind . for the remaining fraction 1-@xmath226 of the sne and stellar winds
we assume that radiative energy losses are so significant that no mechanical energy is supplied to the starburst region , and that the ejecta never becomes part of the outflow .
we treat @xmath226 as a model variable , and refer the reader to appendix [ app : zeta ] for a discussion of plausible values for @xmath226 given the gas content of the starburst and the star cluster mass function .
the net energy injection rate into the hot volume - filling gas in the starburst region is @xmath227 the net mass injection rate is @xmath228 where the mass loading factor @xmath30 is a parameter commonly used to indicate the relative amount of mass loading . if @xmath229 then the wind is not mass loaded . in this paper
we consider only mass loading that operates within the starburst region , i.e. central mass loading in the terminology adopted by @xcite and @xcite
. written as a function of the thermalization efficiency , the mass loading factor and participation factor , the central temperature , pressure and density are @xmath230 @xmath231 and @xmath232 the central temperature is thus linearly proportional to the ratio of the mechanical energy thermalization efficiency to the mass loading factor , and independent of the size of the starburst region @xmath115 or participation factor @xmath226 .
thus it is an excellent diagnostic tool for quantifying the effects of feedback , provided that this temperature can be accurately measured .
the terminal velocity of the wind fluid @xmath233 is also a function of @xmath234 .
an observational determination of @xmath140 is beyond the capabilities of current x - ray observatories , but the x - ray microcalorimeter spectrometer ( xms ) instrument on the future international x - ray observatory ( ixo ) will provide the high resolution non - dispersive x - ray spectroscopy necessary to directly measure the velocity of the hot phases in superwinds .
the mixed sn ejecta and stellar wind material will be enriched in heavy elements compared to the cooler ambient ism within the starburst region .
as the emissivity of the thermal x - ray emission from a hot plasma depends strongly on the metal abundance of the hot plasma it is necessary to consider both sn - related enrichment and dilution via mass loading .
if we assume mass - loading is associated with efficient mixing , then the elemental abundance @xmath235 of an element @xmath236 in the wind fluid is @xmath237 where @xmath238 is the abundance of the element in the sn plus stellar wind ejecta and @xmath239 is the abundance in the ambient ism , and we have made use of equation [ equ : mdot_tot ] which defines @xmath30 .
the primary elemental coolants for plasmas with temperature @xmath240 k are oxygen , neon and iron ( e.g. see * ? ? ?
iron is of particular importance for this work given the @xmath241 kev and @xmath242 kev lines from helium - like and hydrogen - like iron ions .
elements such as s , ar and ca also produce weaker x - ray lines at energies @xmath243 kev that may also serve as a diagnostic of the wind - fluid .
@xcite present the imf - averaged sn ejecta ( both sn ii and sn ib / c ) element abundances with respect to solar .
their stellar evolution models include mass loss due to stellar winds , but the quoted yields do not include the hydrogen and lighter elements returned to the ism by the winds .
they find @xmath244 , @xmath245 , @xmath246 , @xmath247 , @xmath248 and @xmath249 .
note the low o / fe abundance ratio compared to the imf - averaged yields presented in @xcite , which is most probably a consequence of wind mass loss in the stellar evolution models used by @xcite .
the time - averaged nature of an imf - averaged yield is appropriate for continuous star formation with an age of @xmath250 myr , but the abundances in the ejecta for a younger or instantaneous burst may differ from that in @xcite . to investigate this , and the effect of using a different set of sn yields , we computed the abundances of the combined sn ejecta and stellar wind material from the yields given by starburst99 .
starburst99 includes wind yields by combining the stellar evolutionary wind mass loss rates with observed stellar surface abundances .
sn yields are based on @xcite for sn type ii , but note that the the stellar evolution models that @xcite used did not in general include wind mass loss .
starburst99 also assumes that stars that become wolf - rayet stars ( initial mass @xmath251 ) are assumed to explode as sn type ib / c and contribute no metals to the final yields . for a continuous star formation model the sn plus stellar wind ejecta abundances predicted by starburst99
are relative constant for ages @xmath252 myr after the onset of the burst , at levels of @xmath253 , @xmath254 , and @xmath255 ( starburst99 does not calculate the yields of ne , ar and ca ) .
given the uncertainties we will adopt @xmath256 for these models . for computational convenience we will further assume the same abundance relative to solar for the other elements .
the stellar and region o and ne abundances in m82 are solar or higher @xcite .
the are very few measurements of iron abundances , the only one we are aware of is that of @xcite who derive a stellar iron abundance of @xmath257 from nir spectroscopy .
the total baryonic mass of m82 is @xmath258 based on its peak rotational velocity ( see * ? ? ?
galaxies of this mass have a mean nebular oxygen abundance of @xmath259 ( on the @xcite scale ) , based on the galaxy mass - metallicity relationship @xcite .
thus the measured metallicity of the stellar population and the warm ism in m82 are not unusual given its mass , nor do they suggest that the ambient ism has been significantly enriched with nucleosynthetic products from the current starburst .
we adopt @xmath260 for our models , again assuming the same value for all other elements .
the luminosity per unit volume @xmath261 ( @xmath262 ) due to the thermal emission from an element of gas of density @xmath126 , temperature @xmath263 and metal abundance @xmath264 is @xmath265 where @xmath266 is the mass fraction of hydrogen for a solar abundance plasma , @xmath267 is the mass of a hydrogen atom and @xmath268 is the cooling function for the specific energy band or line transition being considered . for broad - band x - ray emission , or the total cooling rate ,
the cooling function for a hot optically thin plasma may be split into a metallicity - independent component @xmath269 and metallicity - dependent component @xmath270 ( evaluated for solar abundance ) , such that @xmath271 , where @xmath264 is the metal abundance with respect to the solar .
this assumes that the relative metal abundances are solar , as otherwise the contribution from each element must be considered separately . to assess the total cooling rate we use the zero - field non - equilibrium cooling functions from @xcite .
we calculated the metallicity - dependent cooling @xmath272 from their published solar and zero metallicity cooling functions . at temperatures above @xmath273
k ( typical of the vast majority of the models we will consider ) the difference between equilibrium and non - equilibrium cooling functions is negligible . to calculate the cooling functions for the @xmath1 8 kev hard x - ray band and the he@xmath0 and ly@xmath0 like lines of s , ar , ca and fe we used version 1.3.1 of apec hot plasma code ( which assumes collision ionization equilibrium ) , accessed using the xspec spectral fitting program @xcite . for the line emission from any specific element @xmath236 the cooling function is an exactly linear function of the metallicity @xmath274 , i.e. @xmath275 . in fig .
[ fig : emis ] we plot the total and 2 8 kev cooling functions ( split into their metallicity - independent and metallicity - dependent components ) that we use in our calculations .
the line - specific cooling functions for hydrogen - like and helium - like sulphur , argon , calcium and iron for unit metal abundance are also shown .
the x - ray line emissivities we use apply to hot gas in collisional ionization equilibrium ( cie ) . in our the calculations using the scaled cc model we evaluate the following expression in each radial shell as we numerically integrate outwards : @xmath276
if the ionization parameter @xmath277 then the plasma is likely to be in cie @xcite .
if @xmath278 s then the outflowing gas is over - ionized and in recombining non - equilibrium .
@xcite show that the broad band x - ray emissivity of such a recombining plasma is less than that of a cie plasma .
the ionization parameter @xmath279 scales with thermalization efficiency and mass loading proportionally as @xmath280 , as increasing the mass loading or decreasing the thermalization efficiency acts both to increase the plasma density and decrease the outflow velocity . for parameters appropriate to m82
we find that the plasma is typically in cie at radii @xmath281 even when @xmath282 .
even if we integrate the emission out to radii @xmath283 the total luminosity is dominated by emission from within @xmath115 , justifying our use of cie emissivities . the radial variation in @xmath279 and
the cumulative emission integral @xmath284 shown in fig [ fig : radial_solution]c demonstrate this for one set of possible model parameters .
this result is consistent with the alternative ionization equilibrium and coulomb relaxation timescale calculations presented in paper i , which also indicated that the hot plasma within the starburst region should be in ionization equilibrium . for a single temperature plasma
the luminosity is the product of the emission integral @xmath285 with the cooling function @xmath286 .
the cumulative volume emission integral is defined as @xmath287 . for a cc - style radial wind this
is dominated by material within the relatively dense starburst region , as shown in fig [ fig : radial_solution]c .
thus the total emission integral is proportional to @xmath288 which again increases as either the mass loading factor @xmath30 is increased , or as the thermalization efficiency @xmath29 is decreased . however such variations in @xmath30 and/or @xmath29 are also associated with a decrease in the central temperature , and for increased mass loading a decrease in the mean metal abundance , so that the net luminosity can ultimately decrease even as increasing @xmath30 or decreasing @xmath29 monotonically increase the net emission integral .
the primary aim of this project is to assess the range of starburst properties that are consistent with the hard x - ray fe - line - emitting plasma in m82 , and which models are deemed acceptable depends on which constraints are used this is unavoidable .
these are not an arbitrary set of observational and conceptual constraints but have been chosen based on realistic physical expectations .
based on the results and discussion presented in
[ sec : observations ] we adopt the following as the full observational constraints on the properties of the wind fluid within a radius of 500 pc of the center of m82 .
* the total x - ray luminosity of the wind fluid in the @xmath1 8 kev energy band must be less than or equal to the broad band luminosity estimated from the acis - s observation for the diffuse hard x - ray emission ( table [ tab : paper1 ] ) , i.e. @xmath289 . *
the fe he@xmath0 line luminosity must be consistent with the observed value given in table [ tab : paper1 ] . in order to be conservative in our estimation of the constraints the observations place on the thermalization efficiency and mass loading , i.e. to accept the largest number of models matching the observational constraints
, we adopt the _ weakest _ constraints on the allowed value of @xmath290 , i.e. using the acis - s observation rather than the better constrained values from the _ acis - i _ or _ xmm - newton _ data sets .
models producing fe line luminosities within 68.3% confidence region from the acis - s observation , @xmath291 , will be judged as matching the constraint on the @xmath292 kev line emission .
this measurement of @xmath290 is slightly larger than , but still consistent with the line luminosity estimated from the acis - i and xmm - newton observations . *
the fe ly@xmath0 luminosity must be less than or equal to the observationally - determined upper limits , of which the best determined value is from the _ xmm - newton _ observation .
we specify that @xmath293 should be @xmath294 for a model to be judged to be consistent with the observational constraints ( table [ tab : paper1 ] ) . * the luminosity of the predicted he@xmath0 and ly@xmath0-like lines of s , ar and ca must be less than or equal to the observed values and/or 99% upper limits . here
the best observational constraints are from the _ xmm - newton _ data ( table [ tab : s_ar_ca ] ) .
acceptable models must then satisfy all of the following : @xmath295 , @xmath296 , @xmath297 , @xmath298 , @xmath299 , and @xmath300 .
upper limits are used as the s , ar and ca - line - emitting plasma is not necessarily the same phase as the fe - line - emitting plasma .
models that ( a ) have total radiative losses less than 30% of the net energy injection rates ( see appendix [ app : radiative_case ] ) and ( b ) predict x - ray luminosities consistent with the full set of constraints given above are considered as having successfully matching the full set of constraints .
we will consider the effect of using a different set of observational constraints .
these alternatives deliberately ignore what we know about the iron line emission ( the iron lines are not used as constraints ) and instead attempt to fully account for all of either the broad - band hard x - ray luminosity or the detected s , ca and ar line luminosities : * the total x - ray luminosity of the wind fluid in the @xmath1 8 kev energy band must be within the 3@xmath301 envelope given in table [ tab : paper1 ] for the model to considered successful , i.e. @xmath302 .
in addition the the total radiative losses must be less than 30% of the net energy injection rate .
we shall refer to these constraints as the broad - band - only constraints . *
the predicted line luminosities for the s he@xmath0 , s ly@xmath0 , ar he@xmath0 and ca he@xmath0 must be within @xmath303 of the values given in table [ tab : s_ar_ca ] for the best fit to the _ xmm - newton _ spectrum .
the allowed limits on the ar ly@xmath0 and ca ly@xmath0 line luminosities are the same as given for the full set of constraints . in addition
the model must predict broad band and total radiative energy losses less than the upper limits also used in the full set of constraints : @xmath289 and total radiative losses less than 30% of the net energy injection rate .
we shall refer to these constraints as the sarca constraints .
we allow larger ranges in the broad - band or line luminosity ( @xmath303 ) as we found that in practice that it is relatively hard to find regions of parameter space where models can satisfy the broad - band - only or sarca constraints .
mass loading factors of @xmath304 or thermalization efficiencies of @xmath305 mark the practical upper and lower limits on these parameters for a m82-like starburst .
each of these values on its own will reduce the central temperature to @xmath306 k , a temperature at or below which the emissivity of the fe he@xmath0 emission drops dramatically ( fig .
[ fig : emis ] ) . to explore this region of the mass loading and thermalization efficiency parameter space
we adopt the following values for the other important model parameters : @xmath307 , a disk - like starburst with @xmath221 pc and @xmath222 pc ( @xmath308 pc ) , @xmath309 and @xmath310 .
we treat the starburst as continuous star formation at a rate of @xmath311 beginning 30 myr ago , which leads to @xmath312 and @xmath313 .
we shall refer to this combination of input model parameters as parameter set a ( see table [ tab : parameter_sets ] ) .
lcrcrcr a & csf / 30 myr & @xmath181 & 1.4 & 300.0 & 5.0 & 1.00 + b & csf / 10 myr & @xmath175 & 1.1 & 300.0 & 5.0 & 1.00 + c & sib / 10 myr & @xmath314 & 2.2 & 300.0 & 5.0 & 1.00 + d & csf / 30 myr & @xmath181 & 1.4 & 125.5 & 5.0 & 1.00 + e & csf / 30 myr & @xmath181 & 1.4 & 404.9 & 5.0 & 1.00 + f & csf / 30 myr & @xmath181 & 1.4 & 300.0 & 2.5 & 1.00 + g & csf / 30 myr & @xmath181 & 1.4 & 300.0 & 7.5 & 1.00 + h & csf / 30 myr & @xmath181 & 1.4 & 300.0 & 5.0 & 0.71 + i & csf / 30 myr & @xmath181 & 1.4 & 300.0 & 5.0 & 0.50 + we initially computed models covering the full range of @xmath315 parameter space . to prevent the total number of simulations from becoming excessive we considered only a limited number of distinct values for @xmath29 and @xmath30 , specifically @xmath316 and @xmath317 , where n ranges from 0 to 3 in increments of 0.25 , resulting in the computation of 169 models using the 2-dimensional hydrodynamical code described in [ sec : hydromodel ] .
we find that only limited regions of parameter space yield models that satisfy any one of the sets of observational constraints we chose to apply ( fig .
[ fig : new_medium_vh1 ] ) .
furthermore , the regions in @xmath29:@xmath30 parameter space satisfying the different sets of constraints are well separated from one another .
models that satisfy the full set of observational constraints ( labeled f in fig .
[ fig : new_medium_vh1 ] ) have moderately high thermalization efficiencies ( @xmath318 ) and moderately low but non - negligible mass loading ( @xmath319 ) .
these models reproduce the observed fe line fluxes and limits without overproducing s , ar or ca line emission , but in doing so they can only account for a small fraction ( @xmath13 @xmath320% ) of the total observed diffuse luminosity of @xmath321 in the @xmath1 8 kev energy band ( fig . [
fig : new_medium_vh1]b ) .
the range of central thermal pressures found in these matching models , @xmath322 @xmath323 k @xmath324 ( given in table [ tab : allowed_regions ] ) is consistent with estimates of the pressure within the starburst region obtained from optical spectroscopy : @xmath325 @xmath326 k @xmath324 @xcite . to match the observed s , ar and ca line fluxes using a wind model ( the region labeled sarca ) , models require a lower thermalization efficiency and a comparable or slightly higher mass loading factor ( @xmath327 , @xmath328 ) . to reproduce the observed diffuse x - ray luminosity in the @xmath1
8 kev energy band with thermal emission alone requires both high thermalization efficiency and high mass - loading factors : @xmath318 and @xmath329 ( the region labeled bb in fig . [
fig : new_medium_vh1 ] ) .
models in this region of parameter space fail to match the full set of observational constraints because they typically predict more s , ar or ca emission ( the s ly@xmath0 luminosity is shown in fig .
[ fig : new_medium_vh1]c ) than is observed when the fe line luminosity is correctly reproduced , or too little fe he@xmath0 emission when they predict less than or equal to the amount of s , ar or ca line emission observed .
thus we can only account for @xmath330% of the observed @xmath1 8 kev diffuse x - ray luminosity with @xmath331 pc with the sum of the predicted thermal emission from the starburst region and the inverse compton x - ray emission ( paper i ) .
we conclude that either our estimate of the inverse compton luminosity is in error , or there must be an additional non - thermal or quasi - thermal source of hard x - ray continuum emission within the starburst region ( e.g. * ? ? ?
* ; * ? ? ?
solving this enigma will require higher resolution spectroscopy to establish the shape of the continuum and the ionization state of the line emission .
models where the total radiative losses are higher than 30% of the energy injection rate , and the adiabatic assumption is no longer valid , are labeled na in fig .
[ fig : new_medium_vh1 ] .
note that for these models the temperatures and x - ray luminosities we predict are not valid .
the location of this region in @xmath315 space in consistent with the discussion in appendix [ app : radiative_case ] . for m82 s star formation rate and starburst region size
we find that as long as @xmath332 then the adiabatic assumption is applicable to either a disk - like or spherical starburst .
parameters that give rise to highly radiative winds can be rejected even though we can not predict their exact x - ray luminosities .
cooling is so efficient under such conditions that the central temperature @xmath333 k and there is no possibility of fe he@xmath0 emission ( e.g. * ? ? ?
to further refine our knowledge of the exact region of the allowed @xmath315 parameter space for m82 s starburst we ran a further set of full hydrodynamical simulations a subset of the full @xmath315 grid described above .
we used a linearly - spaced grid , varying @xmath29 between 0.1 and 1.0 in steps of @xmath334 and @xmath30 between 1.0 and 3.0 in steps of @xmath335 ( a total of of 494 models ) .
the predicted regions of allowed parameter space for parameter set a are shown in fig .
[ fig : new_fine_vh1 ] .
this figure also demonstrates that the exact shape of the allowed region using the full set of constraints is most - strongly influenced by the combination of the constraints on the fe he@xmath0 line and the upper limit on the s ly@xmath0 line , ar or ca line luminosities as a function of @xmath29 and @xmath30 . ] . for a given value of the thermalization efficiency @xmath29 the lower boundary of acceptable mass loading factors @xmath30
is set by requirement to achieve sufficient fe he@xmath0 emission ( the necessary condition is @xmath336 ) , while the maximum @xmath30 is set by the requirement not to exceed the observed s ly@xmath0 luminosity ( @xmath296 ) .
for a given value of @xmath29 the highest allowed @xmath30 is a less likely solution than a lower allowed value of @xmath30 .
the highest allowed values of @xmath30 produce the entire observed s ly@xmath0 flux in the wind fluid , without leaving leeway for s ly@xmath0 emission from the soft x - ray - emitting plasma within the starburst region . were we to change the s ly@xmath0 constraint from an upper limit to a specific range in values around the observed line luminosity the resulting allowed region would be a narrow strip along what is currently the upper @xmath30 margin of the full allowed region .
adopting such a constraint would be somewhat ad - hoc , given that the s he@xmath0 and ar he@xmath0 lines are equally well detected as the s ly@xmath0 line ( table [ tab : s_ar_ca ] ) .
furthermore , an attempt to match the s he@xmath0 , ly @xmath0 and ar he@xmath0 line fluxes , rather than treating them as upper limits , simultaneously with the fe he@xmath0 line flux leads to no acceptable models being found . ultimately this is because the s , ar and ca line fluxes and upper limits are indicative of a plasma temperature @xmath337 ( see table [ tab : temp_estimates ] or the sarca region on fig .
[ fig : new_fine_vh1]d ) , yet such a low temperature plasma produces too little fe he@xmath0 emission to match the observations . as shown in fig .
[ fig : vh1_vs_scaledcc ] , these same general regions of acceptable parameter space are reproduced in simulations making use of the scaled cc model , although in detail there are minor difference caused by the @xmath338% errors in the luminosity predicted by the scaled cc model compared to the multi - dimensional hydrodynamical simulations ( appendix [ app : non_spherical ] ) .
given the computational cost of the full hydrodynamical simulations it is preferable to use the scaled cc model to explore the effect of varying other model parameters such as starburst region size , metal abundance and participation factor .
our experience with parameter set a indicates that the scaled cc model reproduces the allowed values of @xmath29 and @xmath30 to an accuracy of @xmath339 for the full set of observational constraints that include the fe lines .
parameter sets b through i are shown in table [ tab : parameter_sets ] .
these alternate parameter sets explore the influence of the chosen star formation mode and age , starburst region size , supernova and stellar wind metal abundance , and participation factor on the region of allowed @xmath315 parameter space .
all of these calculations were performed using the scaled cc model alone .
predicted central density , pressure and temperature , along with the terminal velocity , total cooling , broad - band hard x - ray ( @xmath1 8 kev band ) luminosity and fe he@xmath0 and s ly@xmath0 line luminosities for @xmath340 and @xmath229 are shown in table [ tab : general_properties ] .
lrcccccc a & 52 & 0.55 1.0 & 1.4 2.2 & 7.56 7.82 & 7.15 7.38 & 34.31 34.54 & 1555 2097 + a & 75 & 0.40 1.0 & 1.2 2.3 & 7.51 7.84 & 7.08 7.40 & 34.23 34.55 & 1470 2164 + b & 26 & 0.75 1.0 & 1.8 2.3 & 7.49 7.66 & 7.13 7.24 & 34.29 34.39 & 1434 1745 + c & 32 & 0.45 1.0 & 1.0 1.4 & 7.64 7.87 & 7.22 7.47 & 34.37 34.62 & 1709 2239 + d & 21 & 0.45 1.0 & 1.0 1.4 & 7.67 7.87 & 7.80 8.05 & 34.20 34.44 & 1778
2240 + e & 60 & 0.55 1.0 & 1.7 2.6 & 7.48 7.76 & 6.94 7.17 & 34.35 34.58 & 1426 1976 + f & 47 & 0.60 1.0 & 1.9 2.6 & 7.50 7.72 & 7.26 7.43 & 34.41 34.58 & 1452
1874 + g & 76 & 0.30 1.0 & 1.0 2.0 & 7.50 7.87 & 6.96 7.37 & 34.11 34.52 & 1452 2240 + h & 43 & 0.65 1.0 & 2.1 2.8 & 7.47 7.70 & 7.15 7.29 & 34.30 34.44 & 1409 1829 + i & 0 & & & & & & + parameter set
b differs from the previously discussed parameter set a only in that is assumes that ongoing star formation began 10 myr ago , rather than 30 myr ago .
as the younger stellar population is intrinsically more luminous , less star formation is required to match the ir luminosity of m82 and consequently the mechanical energy and mass injection rates in parameter set b are lower than in parameter set a. furthermore the ratio of energy to mass injection is lower , resulting in a lower central temperature of @xmath341 k for unit @xmath29 and @xmath30 compared to @xmath342 k in parameter set a , and a lower temperature than parameter set a for any given value of @xmath29 and @xmath30
. cumulatively these differences result in a significant shift in the location of the the models in @xmath315 parameter space that match either the full or sarca constraints ( see fig .
[ fig : matches_modb_modc]a ) .
generation of plots equivalent to fig .
[ fig : new_fine_vh1 ] ( not shown ) reveals that the region matching the full set of constraints is again determined by the intersection of the fe he@xmath0 and s ly@xmath0 observational constraints .
furthermore , at a given value of @xmath29 and @xmath30 the predicted broad band and ( in particular ) line luminosities are lower for parameter set b than for parameter set a , with the least difference in luminosity occurring at @xmath282 .
application of equation [ equ : ei ] indicates that parameter set b yields emission integrals consistently @xmath343 dex lower than parameter set a for all values of @xmath29 and @xmath30 .
for the fe and s lines the lower temperature of parameter set b when @xmath344 results in a higher cooling function that offsets the reduced emission integral because the lower temperature is closer to the peak of the line - specific cooling functions , such that the predicted values of @xmath290 and @xmath345 are comparable to those of parameter set a ( table [ tab : general_properties ] ) .
but over much of the @xmath315 parameter space the lower temperatures in parameter set b result in lower line emissivities that in combination with the lower emission integral result in lower line luminosities than in parameter set a. ultimately the regions in @xmath315 parameter space that match the full and sarca constraints in parameter set b require higher mass loading and higher thermalization efficiency than in parameter set a in order to increase the emission integral without further decreasing the temperature .
increasing @xmath30 strongly increases the net emission integral as @xmath346 , although it slightly decreases the plasma temperature ( which must then be offset by increasing the thermalization efficiency ) .
in contrast , reducing the thermalization efficiency is counter productive as the increase in @xmath285 is counteracted by the decrease in temperature that consequently reduces the line emissivities .
parameter set c represents the star formation in the nucleus m82 as a single instantaneous burst occurring 10 myr ago .
compared to parameter set a this results in a higher mechanical energy and mass injection rate , although the ratio of energy to mass injection is much closer to that in parameter set a than the lower values of parameter set b. the higher mass injection for parameter set c leads to higher predicted x - ray luminosities for a given value of @xmath29 and @xmath30 than for parameter set a , which has a significant effect on the regions of @xmath315 parameter space that match the various constraints ( fig .
[ fig : matches_modb_modc]b ) .
the region matching the @xmath1 8 kev broad band luminosity appears at much lower @xmath30 than in parameter set a , although in terms of absolute mass injection @xmath347 the location of this region is consistent between parameter set a and parameter set c. similarly , lower values of @xmath30 are required to match the full set of constraints , the allowed region extending between @xmath348 and @xmath349 .
models with @xmath282 are again excluded as possible matches to the observed properties of m82 , but for different reasons than in the case of parameter sets a and b. for parameter set c models with @xmath350 and @xmath351 generate sufficient fe he@xmath0 luminosity to match the required constraint , but these models predict a fe ly@xmath0/he@xmath0 line ratio greater than the upper limit of 0.34 ( i.e. @xmath352 is too high ) , and hence do not match the full set of constraints .
a combination of constraints shapes upper edge of the allowed region .
in addition to the requirement not to overproduce the s ly@xmath0 emission that we encountered with parameter sets a and b , the upper edge of the allowed region in parameter set c is also set where the fe he@xmath0 luminosity exceeds the observational limits .
parameter set d adopts the compact starburst region discussed in
[ sec : theory : applying : sbsize ] . in terms of the effective radius of the starburst region parameter set
d uses @xmath353 pc instead of @xmath308 pc , but is otherwise identical to parameter set a. consequently the gas density within the starburst region ( @xmath354 ) is considerably higher than in parameter set a ( @xmath355 ) or even parameter set c ( @xmath356 ) , and results in a significant increase in the x - ray emission despite the reduction in starburst region volume ( see equ . [
equ : ei ] ) .
the regions of @xmath315 parameter space over which parameter set d matches the various sets of constraints is very similar to that parameter set c ( fig . [ fig : matches_modd_mode]a ) for the same reasons we discussed above .
the region of @xmath315 parameter space that matches the full set of constraints is smaller than in parameter set c because of the greater increase in luminosity caused by the higher gas density , and no region of @xmath315 matches the sarca constraints .
it is possible that with tighter observational constraints on @xmath290 and @xmath345 there would be no region of @xmath315 parameter space that would match the full set of observational constraints .
the lower volume of the starburst region adopted in parameter set d also leads to higher central pressures .
models that match the full set of observational constraints have @xmath357 , values that are 2 4 times higher than the highest independent observational estimates of the thermal pressure in the core of m82 .
this leads us to suspect that parameter set d is not a valid physical representation on the m82 starburst .
parameter set e adopts the largest of the possible starburst region geometries discussed in
[ sec : theory : applying : sbsize ] , for which the effective radius is @xmath358 pc and the resulting central number density is @xmath359 .
the region that matches the full set of observational constraints now moves toward higher @xmath29 and higher @xmath30 with respect to the same region in parameter set a ( fig .
[ fig : matches_modd_mode]b ) , for similar reasons to those we discussed regarding parameter set b. increased mass loading is required to make up for the reduction in gas density due to the larger starburst , but the thermalization efficiency must also increase so as to maintain a central temperature that maximizes the fe he@xmath0 emissivity ( @xmath360 , for parameter sets a , b and e. see table [ tab : allowed_regions ] ) . in [ sec
: theory : applying : z ] we discussed the expected metal abundance of the merged sn ejecta and stellar wind ejecta .
the calculations of @xcite predict a higher iron abundance for the sn ejecta ( @xmath361 ) than the starburst99 calculations that combine @xcite sn yields with stellar wind ejecta ( @xmath362 ) .
furthermore the @xcite calculations predict that sulphur , argon and calcium are less enriched than iron , while starburst99 predicts that sulphur is more strongly enriched than iron . in all of the calculations we have discussed until now we adopted a uniform relative enrichment of all elements , @xmath363 .
parameter sets f and g explore the influence of the adopted metal abundance on the region of @xmath315 parameter space that match the observational constraints : parameter set f adopts @xmath364 while parameter set g adopts @xmath365 .
given the already large uncertainties in the yields we feel that exploring the effect of adopting different yields for different elements is beyond the scope of this paper .
the lower metal abundance in parameter set f leads to lower lower s , ar , ca and fe line luminosities than for parameter set a. for @xmath229 the line luminosities are exactly half those in parameter set a , but the fractional difference becomes less as @xmath30 increases and the influence of the mass - loaded material on the net metal abundance of the wind fluid @xmath366 becomes larger ( equ .
[ equ : zwf ] ) .
the result in terms of the region of @xmath315 parameter space that matches the observational constraints is shown in fig .
[ fig : matches_modf_modg]a , and is very similar to that in parameter sets b and e that also generate lower line luminosities for different reasons .
conversely the higher metal abundance in parameter set g leads to higher line luminosities than parameter set a , although they are generally less than those in parameter sets c and d ( see e.g. table [ tab : general_properties ] ) .
the @xmath315 region of matches is shown in fig .
[ fig : matches_modf_modg]b .
the upper @xmath30 edge of the region that matches the full set of observational constraints is set by the upper limit on the s ly@xmath0 luminosity , while the lower @xmath30 edge of this region is set by a combination of the fe ly@xmath0/he@xmath0 line ratio constraint and the minimum acceptable fe he@xmath0 line luminosity .
parameter sets h and i explore the effect of varying the participation fraction @xmath226 ( see [ sec : theory : applying : edotmdot ] ) . reducing @xmath226 has the effect of reducing the effective star formation rate without altering the energy per particle , and the emission integral and luminosities for a given value of @xmath29 and @xmath30 are equal to those from parameter set a except multiplied by a factor @xmath367 . for parameter
set h , where @xmath368 , the factor 2 reduction in all luminosities moves the regions that match both the sarca and full set of constraints to higher @xmath30 and higher @xmath29 ( fig .
[ fig : matches_modh_modi]a ) .
parameter set i has @xmath369 and the reduction in luminosity is so significant that the region matching the sarca constraints has moved off of the region shown in fig .
[ fig : matches_modh_modi]b to @xmath370 , @xmath371 .
more significantly , there are no values of @xmath29 or @xmath30 that can match the full set of observational constraints when using parameter set i. lllcccrrrr a & -0.80 & 7.22 & 8.02 & 2651 & 278 & 38.67 & 38.27 & 37.46 & 36.22 + b & -0.85 & 7.06 & 7.91 & 2341 & 278 & 38.56 & 38.17 & 37.45 & 36.24 + c & -0.59 & 7.40 & 7.98 & 2548 & 285 & 39.09 & 38.70 & 37.92 & 36.69 + d & -0.04 & 7.98 & 8.02 & 2651 & 121 & 39.17 & 38.75 & 37.93 & 36.74 + e & -1.06 & 6.96 & 8.02 & 2651 & 369 & 38.49 & 38.09 & 37.27 & 35.98 + f & -0.80 & 7.22 & 8.02 & 2651 & 278 & 38.51 & 38.11 & 37.16 & 35.92 + g & -0.80 & 7.22 & 8.02 & 2651 & 278 & 38.80 & 38.39 & 37.64 & 36.40 + h & -0.95 & 7.07 & 8.02 & 2651 & 272 & 38.37 & 37.97 & 37.16 & 35.92 + i & -1.10 & 6.92 & 8.02 & 2651 & 266 & 38.07 & 37.67 & 36.86 & 35.62 +
in @xcite we concluded that the diffuse fe he@xmath0 emission from the central @xmath372 pc of m82 could have been produced by thermalized sn ejecta and stellar wind material from the starburst based on relatively simple estimates of the plasma properties of the emitting material .
the more rigorous approach to the interpretation of the diffuse hard x - ray emission taken in this paper strengthens that conclusion : the plasma responsible for the hard x - ray iron - line emission is the same gas that drives the larger - scale superwind in this starburst galaxy .
furthermore our method allows us to make quantitative statements regarding supernova feedback and the properties of m82 s starburst - driven superwind .
the primary aim of this paper was to establish what constraints could be made on the efficiency of supernova heating and the net rate of mass injection within a starburst region .
it is generally accepted that in more normal , or quiescent , star forming galaxies such as the milky way on average only @xmath13% of the @xmath373 erg of initial kinetic energy per supernova ends up heating or moving the ism , with the remainder being lost as radiation ( see e.g. * ? ? ?
* ; * ? ? ?
although it had long been suspected that the efficiency of supernova heating is a function of the local interstellar and star forming conditions there was no direct method by which to measure this thermalization efficiency in a starburst galaxy until , as described in [ sec : introduction ] , the launch of the _ chandra _ x - ray observatory .
theoretical arguments have been made for either lower or higher sn heating efficiencies in more actively star - forming galaxies .
@xcite argued that radiative energy losses must always be severe because massive stars form in regions of high gas density .
in contrast @xcite argue that thermalization must be efficient in starbursts because a high supernova rate per unit volume leads to an increased filling factor of the hot low density ism , which in turn reduces the radiative energy losses experienced by subsequent sne .
thus the sn thermalization efficiency depends on the sn - modulated porosity of the ism ( e.g. * ? ? ? * ) .
recent numerical simulations by @xcite appear to qualitatively demonstrate such a time - dependent non - linear thermalization efficiency .
intriguingly they find that the thermalization efficiency ( which they term the heating efficiency ) can only be low ( @xmath374% ) or very high ( @xmath11% ) for any significant period of time , transitioning between to two states very rapidly .
let us briefly review how the hard x - ray line emission from a starburst region is related the average thermalization efficiency @xmath29 and the mass loading factor @xmath30 . both of these parameters strongly affect the net volume emission integral @xmath375 ( equation [ equ : ei ] ) and the line emissivities @xmath268 through the plasma temperature @xmath376 ( equation [ equ : final_tc ] , fig .
[ fig : emis ] ) and the metal abundance of the wind fluid @xmath366 ( equation [ equ : zwf ] ) . in the specific case of m82
it is the fe he@xmath0 and s ly@xmath0 lines that drive the allowed values of @xmath29 and @xmath30 we derive . for many of the sets of model parameters
we have explored it should be noted that @xmath282 will not reproduce the observed constraints because too little fe he@xmath0 emission is produced as the plasma density within the starburst region is too low .
thus successful models must have a higher volume emission integral which implies @xmath377 and/or @xmath378 . as we described in
[ sec : model : epsilon_and_beta ] , the lower boundary of acceptable mass loading factors @xmath30 for a given value of the thermalization efficiency @xmath29 is set by requirement to achieve sufficient fe he@xmath0 emission ( @xmath336 ) , while the maximum @xmath30 is set by the requirement not to exceed the observed s ly@xmath0 luminosity ( @xmath296 ) .
furthermore the central temperature ( and hence @xmath234 ) must be high enough that the fe he@xmath0 emissivity is significant with respect to the emissivity of the lower ionization lines , in particular the sulphur lines .
this constraint ultimately controls the minimum allowed value of the thermalization efficiency @xmath29 , and is the reason why the different parameter sets tend to have similar minimum values of @xmath379 , specifically @xmath380 ( see table .
[ tab : allowed_regions ] ) .
the central temperature can not exceed @xmath381 k ( @xmath382 ) as otherwise the upper limit on the fe ly@xmath0/he@xmath0 line ratio would be violated . for a given value of @xmath29
this constraint determines the lowest allowed values of @xmath30 in cases where the fe he@xmath0 luminosity is sufficiently high enough to match the observational constraints .
this upper temperature constraint is only significant when the model parameters generate higher central densities or luminosities ( more mass injection , a more compact starburst , or a higher metal abundance in the ejecta ) , and thus there is a broader range in the maximum allowed @xmath379 than in the minimum allowed @xmath379 .
returning to the consideration of our results , we find that for all parameter sets that produce at least one valid solution that can match the full set of observational constraints for m82 , those valid solutions always include thermalization efficiencies of 100% .
furthermore @xmath29 is never less that 30% , and over the range of parameter sets we have explored the minimum valid thermalization efficiency @xmath383 ranges from 30% to 75% ( see table [ tab : allowed_regions ] ) . within this range
there is a small degree of variation associated with different starburst model parameters we have explored .
we find that the starburst history and ejecta metal abundance @xmath384 have some effect on the allowed thermalization efficiency , while the starburst region size appears to have little influence on the minimum allowed thermalization efficiency ( table [ tab : allowed_regions ] ) .
for example in a younger continuous star formation ( csf ) model for the burst ( parameter set b ) the minimum allowed thermalization efficiency is @xmath385 in comparison to the older csf we adopt as the default starburst history ( @xmath386 ) or the single instantaneous burst model ( parameter set c , @xmath387 ) . that @xmath29 must be @xmath388% is a striking result .
if the diffuse @xmath241 kev iron line in m82 arises in a collisional plasma then the average thermalization efficiency in the m82 starburst region can be as high as 100% , while it can not be as low as 10% .
our results give quantitative support to the arguments presented by @xcite and @xcite for a higher thermalization efficiency in starbursts than in normal star - forming galaxies .
the issue of whether the thermalization efficiency must be 100% , or whether high but intermediate values are allowed , remains an interesting but open question . with respect to the mass - loading of the superwind in m82
we find that the majority of the parameter sets we have considered require some centralized mass loading of the wind fluid , although at low - to - moderate levels .
over all parameter sets @xmath389 , but only parameter sets c , d and g allow @xmath229 and more typically @xmath390 .
parameter sets c , d and g allow zero additional mass - loading ( @xmath229 ) as a valid solution when coupled with lower values of the thermalization efficiency ( @xmath391 70% ) . in parameter set c
this is because the single instantaneous starburst injects more mass than the constant star formation model used in the other parameter sets , and hence does not require additional sources of mass to increase the plasma density within the starburst region . in parameter sets d and g the non - mass - loaded plasma within the starburst region is sufficiently luminous without additional mass - loading because of either the compactness of the starburst ( d ) , or the higher metal abundance ( g ) , increases the net luminosity .
high values of @xmath29 are excluded as valid solutions when @xmath392 for these three parameter sets because the high central plasma temperature violates the fe ly@xmath0/he@xmath0 line ratio constraints .
thus these results suggest that it is plausible that the raw sn and stellar wind ejecta from the massive stars in m82 s starburst have been efficiently mixed with a roughly equivalent mass of initially cooler ambient gas , although no mass loading is also a possible solution .
the net effect is an average mass outflow rate from the starburst region in the very hot phase of the wind fluid of @xmath393 , approximately half of the gas consumption due to the star formation rate alone ( [ sec : theory : applying : sbmag ] ) .
the allowed range of wind fluid outflow rate covering all parameter sets except parameter set d lies in the range @xmath394 ( a fraction @xmath395 0.8 of the star formation rate for the salpeter imf we adopt ) .
this level of mass - loading is significantly lower than the values others have derived from soft x - ray observations of superwinds .
@xcite used fairly sophisticated 1-dimensional hydrodynamical modeling to estimate that a mass - loading factor of @xmath396 6 was required in to match the soft x - ray properties of m82 known at that time from _ einstein _ and _ rosat _ hri observations . for the dwarf starburst galaxy ngc 1569 @xcite estimated @xmath397
@xmath398 ( where @xmath279 is the volume filling factor of the x - ray - emitting plasma ) based on the estimated mass of soft x - ray - emitting gas seen in _
chandra _ observations .
@xcite derived a value of @xmath399 for the soft x - ray - emitting plasma in the local lyman break galaxy analog haro 11 .
both @xcite and @xcite obtain similar estimates of @xmath30 based on the best - fit metal abundances in the soft x - ray - emitting plasma compared to the expected abundances of a sn - ejecta - enriched plasma ( see equation [ equ : zwf ] ) .
however a direct comparison between our findings and the existing results in the literature is somewhat misleading .
the results from the literature are all based on soft x - ray observations of starbursts that probe gas in the temperature range 2 10 million k , not the hotter 30 70 million gas probed by the observations of m82 s diffuse hard x - ray emission .
the high mass loading factors simply reflect the fact that there is more mass in the soft x - ray emitting plasma than can be accounted for by sn and stellar wind ejecta alone .
soft x - ray mass loading factors are further based on measurements that cover the entire soft x - ray nebula associated with the target starburst galaxy ( over @xmath159 kpc - scales for m82 and haro 11 , and over @xmath400 kpc in ngc 1569 ) , whereas our measurement constrains the mass loading of the wind fluid within the @xmath372 pc starburst region itself ( in the terminology introduced by @xcite this is central mass loading ) .
it is plausible that the wind fluid experiences additional mass - loading as it flows outward beyond the starburst region ( distributed mass loading ) .
the older measurements referred to above can not differentiate between a localized or a global mass - loading process ( by global we mean a process that is essentially volume filling ) , although we have previously argued that a significant fraction of the soft x - ray emission must arise from a low - volume filling factor component of the wind @xcite . in contrast the mass loading factor derived in this paper is for a volume - filling mass loading .
it is worth noting that it is not the case that the hottest plasma in a superwind is necessarily also the least mass - loaded phase .
the technique we have presented could detect a mass loading factor of @xmath159 10 for the very hot gas in the starburst region .
the line luminosities are strong functions of the mass loading factor ( fig .
[ fig : new_medium_vh1 ] ) , so that both the fe and s line emission would be more intense than observed in the case of a higher mass - loading factor .
both low thermalization efficiency and high mass loading are more readily observable . in a sense it is the relative faintness of these lines from m82 that force us toward the allowed values of the thermalization efficiency and mass loading factor given in table [ tab : allowed_regions ] .
even if part of the starburst region experienced higher mass loading than another part our final estimate of the allowed mass loading factor would be skewed toward the higher value as @xmath30 is essentially luminosity - weighted .
thus we conclude that the @xmath401 k component of m82 s superwind ( the wind fluid ) genuinely has a relatively low degree of mass - loading within the starburst region . despite the broad range of input model parameters we have explored in the various parameter sets we find that the physical properties of the base of the m82 superwind in the allowed models are all relatively similar ( table [ tab : allowed_regions ] ) .
we have already described the relatively narrow allowed range for the central temperature , @xmath402 , and the physical constraints that drive this result .
the result is all the more notable given the relatively poor constraints provided by the traditional line - ratio diagnostics shown in table [ tab : temp_estimates ] .
similarly the central pressure and wind momentum injection rate are constrained to lie within a range of @xmath403 dex ( @xmath404 , @xmath405 ) , if we exclude parameter set d from consideration because of its unrealistically high pressure .
these pressures are in good agreement with independent estimates of the thermal pressure within the ism in the starburst region , as previously discussed .
the momentum injection rate is an important parameter with regard to the acceleration of the wnm and wim by the wind fluid ( see e.g. * ? ? ?
* ) and @xmath406 , initially exposed to a radially - expanding sn - driven wind at a radius @xmath407 from the origin , is proportional to @xmath408 .
notably the terminal velocity of the cloud can be lower than the terminal velocity of the wind .
for a radiation - pressure driven cloud the final result is essentially the same , with the substitution of @xmath409 for @xmath410 .
clouds of entrained material are effectively momentum - driven irrespective of whether they are in an energy - driven hot wind or a radiation - driven wind . ] . by way of comparison the momentum injection rate assuming 100% thermalization efficiency and
no mass loading is @xmath411 for the star formation models adopted in parameter sets a and b , and @xmath412 for parameter set c. thus taking account of imperfect thermalization and additional ( central ) mass loading does not significantly alter the m82 starburst momentum injection rate from the values one would estimate from sne and stellar winds alone .
the moderately tight limits on the central temperature we have derived also make for tight limits on the terminal velocity of the wind fluid , as @xmath413 ( equation [ equ : vinf ] ) .
over all the parameter sets , models that match the full set of the observational constraints predict @xmath140 in the range 1410 2240 @xmath414 .
this estimate of @xmath140 is significantly higher than previous estimates of the terminal velocities in local starbursts ( including m82 ) based on the temperature of the soft x - ray emitting gas , @xmath415 ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , it has been long understood that these estimates were effectively lower limits on @xmath140 given the technical difficulties associated with robustly detecting a hotter and more tenuous ( and thus intrinsically fainter ) component of the wind . strictly speaking the terminal velocity is the characteristic velocity the wind fluid would asymptote to once outside the starburst region ,
if it expanded into a vacuum ( see fig .
[ fig : radial_solution ] ) . for a superwind expanding into a gaseous medium
there are additional characteristic velocities , some of which are phase - dependent ( * ? ? ? * strickland & dinge , in preparation ) , but all are less than or equivalent to @xmath140 .
nevertheless @xmath140 remains an accurate measure of both the effective energy per particle of the most metal - enriched gas and the maximum velocity attainable by gas in a pressure - driven superwind . for a wind that has broken - out of both the gaseous disk and halo of its host galaxy
a comparison between @xmath140 and the local escape velocity from the gravitational potential @xmath416 should give a qualitative indication of whether the material in the wind can escape .
estimates of m82 s escape velocity ( and mass ) have surprisingly large uncertainties for such a well - studied galaxy , because of its very unusual rotation curve and the possibility that its encounter with m81 has stripped it of its dark matter halo @xcite .
the extreme options are a potential dominated by either a central mass within a radius of @xmath417 kpc or a dark matter halo extending out to 100 kpc , the two of which crudely limit the escape velocity to lie in the range @xmath418 @xmath419 . for a peak rotational velocity of @xmath420 at a radius of @xmath372 pc @xcite , the escape velocity is @xmath421 @xmath422 , much lower than the allowed range of @xmath7 @xmath8 .
our estimated terminal velocity for m82 s wind is also much larger than estimates of the velocity of the h@xmath0 emitting warm ionized medium , which achieves a maximum inclination - corrected velocity of @xmath423
@xmath424 at heights between @xmath425 to 2000 pc from the nucleus @xcite .
this is not surprising , as the @xcite model predicts a large difference between the velocities of the wind fluid and the velocity of the wim ( which they identified as clouds of cooler denser gas entrained and accelerated by the wind fluid ) . in the case of m82 velocity of the wim component of
the wind appears to exceed the galactic escape velocity , although more generally the wnm and wim in superwinds has a speed more closely comparable to @xmath416 and hence its long term fate is ambiguous @xcite .
our results constrain the energy per particle within the starburst region , neglecting the possibility of radiative energy losses and additional mass - loading of the wind fluid exterior to the starburst region .
is it plausible that these effects could be of sufficient magnitude to reduce the terminal velocity of the wind fluid at large radius to a value below the escape velocity ? for a starburst with a sn thermalization efficiency of @xmath426 and a central mass loading factor @xmath392 the larger scale wind would have to lose @xmath427% of its remaining energy to reduce the final value of @xmath140 to @xmath422 in the absence of additional mass loading of the wind fluid ( @xmath428% energy loss to reduce @xmath140 to @xmath429 ) .
the required energy loss fraction is very similar for a central starburst with @xmath430 and @xmath431 .
we know that radiative energy losses in m82 s large - scale wind are small , @xmath374% @xcite .
it seems unlikely that the wind fluid could so effective in transferring enough of its energy to a different gaseous phase , such as the galactic halo medium , without a significant fraction of this energy being radiated and hence observed , unless the radiation is spread over a region much larger than the currently studied @xmath432 kpc extent of m82 s wind .
the remaining alternative is to strongly mass load the wind fluid once it has left the starburst region .
assuming that energy losses in the wind fluid are negligible then in the case of the starburst with @xmath426 and @xmath392 we would need to effectively mix an additional @xmath433 into the wind fluid outside the starburst region to reduce @xmath140 to @xmath422 ( @xmath434 to reduce @xmath140 to @xmath429 ) . for @xmath430 and @xmath431 then @xmath435 ( @xmath436 ) and @xmath437 ( @xmath438 ) .
even the smallest of these values is uncomfortably large in terms of the required gas reservoir , and furthermore the resulting x - ray emissivity of the volume filling wind fluid might then exceed the observed soft x - ray luminosity of the wind ( @xmath439 ) . given the preceding arguments we believe that is probable that the metal enriched wind fluid will ultimately escape m82 .
it is worth stressing that a multi - phase wind model leads to phase - dependent ejection efficiencies for the material in starburst - driven superwinds .
current cosmological models of galaxies that attempt to include superwinds invariably have single phase winds ( although the simulation may allow multiple ism phases , their galactic winds lack the fine - scale multiphase nature of real winds ) .
cosmological simulations lack the ability to separately model the differential ejection of the metal - enriched high - energy - per - particle wind fluid from the less enriched , lower velocity , but more massive wnm and wim .
their results on metal transport and feedback processes in galaxies are potentially misleading or incorrect
. in particular , simulations that add all the gas mass of the wnm and wim phases to the metal - enriched sn ejecta require nonphysically large fractions ( @xmath440% ) of all the available sn mechanical energy in star - forming galaxies ( normal galaxies in addition to starburst galaxies ) in order to match the observed degree of metal loss from galaxies and metal enrichment of the igm .
( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these papers interpret such problems as implying the physical inadequacy of purely sn - driven winds and need for addition energy or wind sources ( e.g. radiation - driven winds ) .
our interpretation is that the flaw lies in the inadequacy of their recipe for winds .
in contrast simulations that effectively treat the sn ejecta alone without additional wim / wnm mass loading achieve good matches to observed ly@xmath0 and absorber statistics while using reasonable sn efficiencies ( @xmath391% , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we further note that it is unrealistic to apply a single uniform feedback prescription to all star forming galaxies .
as discussed previously there are good reasons to expect that @xmath29 is a function of the local intensity of star formation .
furthermore only galaxies with average star formation rates is based on a salpeter imf with mass limits of 1 and 100 @xmath441 .
@xcite demonstrate that the star formation intensity threshold to establish extra - planar gaseous activity depends on the host galaxy mass , but as this mass dependence is relatively weak we have ignored it in presenting the simple superwind recipe . ] per unit area @xmath442 kpc@xmath443 show evidence for superwinds @xcite . in the spirit of @xcite we now present a simplified bimodal feedback and superwind recipe appropriate for use in semi - analytical and numerical simulations of galaxy formation .
this recipe is most applicable to simulations that aim to answer questions regarding the galaxy mass - metallicity relationship , and/or the metal content of the igm , as it primarily addresses the metal - enriched stellar wind and sn ejecta from starburst - driven superwinds .
the formulation presented below differs from most currently used recipes for winds .
it prevents massive galaxies with moderate star formation rates ( such as the milky - way ) and the bulk of the galaxy population at low redshift from having any wind - related metal - loss , but the chances of metal ejection from intensely star - forming galaxies ( starbursts ) is enhanced .
it is unlikely to act as an effective mechanism to suppress star formation in all star - forming galaxies , unlike the wind recipes in the simulations we criticized earlier . in galaxies with average star formation rates per units area greater than @xmath444
kpc@xmath443 ( i.e. starbursts ) feedback is efficient and leads to the creation of galactic superwinds .
the hot gas in a superwind can be characterized by a mass injection rate @xmath445 of material of metal abundance @xmath446 , with an associated mechanical energy injection rate @xmath447 . in terms of the star formation rate @xmath191 in units of @xmath448 for a salpeter imf between the mass limits of 1 and 100 @xmath441 , then : * @xmath449 where @xmath450 is in the range 1.0 2.8 .
* @xmath446 is given by equation [ equ : zwf ] , and is a function of the metal abundance in the merged sn and stellar wind eject @xmath384 , the metal abundance in the cool ambient ism @xmath451 and the mass loading factor @xmath450 .
* the effective mechanical energy injection rate , accounting for the efficiency of thermalization , @xmath452 where @xmath453 is in the range 0.3 1.0 .
* alternatively , if winds are implemented by imparting a fixed velocity to gas `` particles , '' then the appropriate velocity is @xmath454 , and should be in the range 1400 @xmath455 . * the values of @xmath453 and @xmath450
are correlated in the sense that low values of @xmath29 are associated with low values of @xmath30 , such that @xmath379 or @xmath140 lie in the range described in [ sec : discussion : eta_beta ] & [ sec : discussion : velocity ] .
* the participation factor @xmath226 is the fraction of the total star formation that can effectively participate in driving a superwind . in the case of m82 @xmath226
is unlikely to be lower than @xmath403 , and is more plausible in the range @xmath456 1.0 ( see appendix [ app : zeta ] ) . in more quiescent star forming galaxies with
@xmath457 kpc@xmath443 feedback is energetically inefficient and no true wind arises , although sn - heated and enriched gas may vent into the halo to form a galactic fountain .
galactic fountains are considerably less constrained theoretically than superwinds , but it is plausible that all the material eventually returns to the galactic disk after some delay .
this recipe captures two essential inter - related features of starburst and feedback physics that have been established in the local universe and that have not been applied in existing recipes for winds in cosmological simulations : the existence of a threshold in star formation activity per unit disk area necessary to drive global galactic winds , and that the quantitative properties of feedback are not the same in all galaxies but depend on the intensity of star formation .
one limitation of the superwind recipe as presented is that it treats all starbursts equally , and applies an m82-like wind to them all .
although simulations of the type made by @xcite suggest that the efficiency of supernova feedback may be bimodal , other evidence suggests that not all starburst - driven flows are equally energy - efficient ( see e.g. * ? ? ?
the techniques described in @xcite and in this paper can be applied to other starburst galaxies in order to measure their effective supernova thermalization efficiencies and mass - loading parameters .
in general one would be limited to a purely spectral x - ray analysis , as for more distant starbursts it would be difficult to spatially separate the hard diffuse x - ray emission from the emission from x - ray binaries and nuclear agn even with the _ chandra _ x - ray observatory .
if our model is correct then we expect that the luminosity of the diffuse hard x - ray ( hx ) emission of a starburst region of star formation rate @xmath191 and effective radius @xmath115 to scale roughly as @xmath458 where we have substituted the star formation rate into equation [ equ : ei ] , and have ignored the @xmath29 and @xmath30 dependence of @xmath286 .
this is notable as it predicts @xmath459 from the wind fluid has a stronger than linear dependence on the star formation rate .
it should therefore be possible to extract this emission component from the other hard x - ray emission components in star forming galaxies that have been shown to scale linearly with star formation rate @xcite .
the task will also be simplified in that it is the hard x - ray line emission that has proven itself to provide the strongest constraints on @xmath29 and @xmath30 in m82 , rather than the hard x - ray continuum emission that the x - ray binaries dominate in the integrated x - ray spectra of galaxies .
we note that fe he@xmath0 emission has been detected in the x - ray spectra of the nearby starburst ngc 253 @xcite , the merging system arp 299 @xcite , and the ultra luminous ir galaxies ( ulirgs ) arp 220 @xcite , mrk 273 @xcite , ngc 6240 @xcite and the superantennae ( braito et al.2008 , submitted ) .
previously this emission has been interpreted in terms of ionized reflection associated with an agn @xcite , and while this is likely true in objects with strong agn ( e.g. ngc 6240 ) , ionized reflection might not be the cause of the fe he@xmath0 emission in all of these objects .
these papers often do consider a starburst origin for the ionized iron emission , but dismiss it based on the expectation that its luminosity should scale linearly with the star formation rate . apparently diffuse hard x - ray emission has also been seen in the _ chandra _ observations of the starburst galaxies ngc 2146 @xcite and ngc 3256 @xcite , although these papers do not discuss hard x - ray line emission .
the _ xmm - newton _ x - ray spectrum of ngc 3256 presented as figure 2 of @xcite shows a hint of a hard x - ray iron line , but the paper does not quantify the energy and intensity of the feature . to crudely test the plausibility of a wind fluid origin for the fe he@xmath0 emission in these other starbursts we consider ngc 253 ( @xmath460 , @xcite ) , arp 299 ( @xmath461 , @xcite ) arp 220 ( @xmath462 , @xcite ) and mrk 273 ( @xmath463 , @xcite ) .
we will use the @xmath464
1000 @xmath465 m ir luminosities of these galaxies , taken from the data in @xcite , as proxies for the star formation rate @xmath191 , making no correction for the fraction of the ir luminosity contributed by any agn .
the @xmath290 luminosities given above are based on the 6.7 kev line photon fluxes quoted in the appropriate paper and scaled to the adopted distances given below .
@xcite found that the hard x - ray emission in ngc 253 ( d=2.6 mpc , @xmath466 ) comes from a 300 pc by 60 pc region associated with the nuclear starburst region and gas torus .
if we interpret this as a disk of that diameter and height then the equivalent 1-dimensional effective radius is @xmath467 pc .
scaling m82 s observed fe he@xmath0 emission by @xmath468 we obtain a predicted fe he@xmath0 luminosity associated with the starburst of @xmath469 , which is consistent with the observed value given above .
arp 299 ( d=48.5 mpc ) is a merging system with a total ir luminosity of @xmath470 , although the fe he@xmath0 emission is only associated with the eastern nucleus ( ic 694 ) .
arp 220 ( d=74 mpc ) and mrk 273 ( d=158 mpc ) have comparable ir luminosities of @xmath471 .
as the sizes and geometries of the starburst regions in these galaxies are not well established we will ignore the @xmath115 scaling and scale the observed m82 fe he@xmath0 luminosity only by the square of the ratio of the ir luminosities .
for arp 299 the predicted fe he@xmath0 luminosity is @xmath472 , a few times higher than the observed value of @xmath473 . for arp 220 and mrk 273
the predicted luminosity is @xmath474 .
this value is comparable to the ( poorly - constrained ) fe he@xmath0 luminosity of mrk 273 , and several times larger than the observed fe he@xmath0 luminosity of arp 220 .
as we would expect the effective size of the starburst regions in these objects to be larger than that of m82 we would expect the luminosity to somewhat less than the predictions given above , so these predictions are in reasonable agreement with the observations .
such back of the envelope calculations certainly do not prove that the fe he@xmath0 emission seen in these objects arises from the same processes as in m82 .
they do demonstrate that there is some hope of being able to test and explore this model in objects other than m82 with current x - ray data from the _ chandra _ and _ xmm - newton _ x - ray observatories .
in the longer term the superior spectral resolution and sensitivity of the calorimeter instrument on the international x - ray observatory will allow routine detection of hard x - ray line emission from a much larger sample of star forming galaxies .
furthermore , with an energy resolution of @xmath475 ev at @xmath476 kev it will allow us to test the assumption that these lines arise in collisionally heated gas .
we have measured the net energy efficiency of supernova and stellar wind feedback ( the thermalization efficiency @xmath29 ) in the starburst galaxy m82 and the degree of central mass - loading @xmath30 of its superwind by comparing a large suite of 1 and 2-dimensional hydrodynamical models to a set of observational constraints derived from hard x - ray observations of the starburst region ( the fluxes of the he@xmath0 and ly@xmath0-like lines of s , ar , ca and fe , along with the total diffuse @xmath1 8 kev x - ray luminosity ) .
the observational constraints are based on a combination of the best currently available observations of the central 500 pc of m82 with both the _ chandra _ and _ xmm - newton _ x - ray observatories .
we combine the results presented in @xcite on the diffuse hard x - ray fe line emission and the diffuse @xmath1 8 kev hard x - ray luminosity with new results ( this paper ) on the s , ar and ca lines . by simultaneously applying a large number of observational constraints ( we used eight different line fluxes or upper limits , an upper limit on the broad - band x - ray luminosity and a physical requirement that the flow not be strongly radiative ) to the simulated starbursts we have been able to constrain the physical properties of the starburst region more tightly than by using more standard x - ray diagnostics .
we modeled the physical properties of the sn - heated and enriched plasma within an assumed disk - like starburst region and the flow of gas outside this region out to a radius of 500 pc using either an appropriately scaled version of the 1-dimensional @xcite analytical superwind solution , or with a full hydrodynamical simulation using the vh-1 hydrodynamics code in 2-dimensional cylindrical symmetry .
we demonstrated a method for correcting the 1-dimensional radial cc model so as to apply to a disk - like starburst geometry with reasonable accuracy .
although this introduces errors of @xmath338% in the individual predicted hard x - ray line fluxes compared to the full 2-d hydrodynamical solution , such errors have little effect on the derived constraints on @xmath29 and @xmath30 as the uncertainties in the observed line fluxes can be larger than this .
we also considered the issue of the applicability of the adiabatic cc wind solution and find that for the study of the volume - filling gaseous component of the m82 superwind within individual star clusters and over the starburst region as a whole the use of adiabatic models is valid .
we considered a broad range of plausible parameters for the m82 starburst , varying the age and mode of star formation , the starburst region size and geometry , and supernova metal yields .
we further considered a correction accounting for the fraction of massive stars that occur in high density regions where high radiative energy losses will preclude their making any contribution to the superwind , which we term the participation fraction @xmath226 .
we find that over all these varied input parameters ( we calculated 663 full hydrodynamical simulations using different model parameters , and 4615 models using the scaled cc model ) , those models that do satisfy the existing observational constraints all have medium to high sn+sw thermalization efficiencies ( @xmath2% @xmath3% ) and have mildly centrally mass - loaded wind fluids ( generally @xmath477 , although in a few cases @xmath229 is allowed ) .
our results are ultimately based on the measured fe he@xmath0 emission of the sn heated and enriched plasma driving m82 s wind .
as such they represent the first direct measurement of the efficiency of sn feedback and mass - loading of the wind fluid in a starburst galaxy .
future observations with higher spectral resolution ( e.g. with _ astro - h _ and _ ixo _ ) will be required to test whether the observed line emission truly arises from a collisional plasma in or close to ionization equilibrium . that the majority of the models that successfully match our imposed hard x - ray observational constraints also predict thermal pressures in the starburst region ( @xmath404 ) that are consistent with prior independent estimates of the starburst region pressure increases our confidence in the validity of our method and results .
the allowed values of the sn and sw thermalization efficiency and the mass loading factor correspond to a temperature for the metal - enriched plasma within the starburst region of 30 80 million kelvin , a mass flow rate of the wind fluid out of the starburst region of @xmath5 @xmath6 and imply a terminal velocity of the wind ( in free expansion ) in the range @xmath7 @xmath8 .
this velocity is considerably larger than the gravitational escape velocity from m82 ( @xmath9 ) and the velocity of the h@xmath0 emitting clumps and filaments in the inner 2 kpc of m82 s wind ( @xmath10 ) .
reducing the velocity of the wind fluid once _ outside _ the starburst region to the escape velocity would require either unrealistically large mass loading ( @xmath478 @xmath479 ) or excessive energy losses ( @xmath427 98% ) that must also avoid raising the luminosity of the wind .
thus the hot gas driving m82 s wind will probably escape the galaxy s potential well .
this result also adds support to the idea that the long term fate of gas within superwinds is phase - dependent .
we show that the luminosity of the hard x - ray line emission produced by the wind fluid in a starburst region scales with the star formation rate @xmath480 and starburst size @xmath115 as @xmath481 .
this should serve to distinguish this emission from the other sources of both soft and hard x - ray emission in star forming galaxies that scale linearly with the star formation rate .
we show that existing measurements of fe he@xmath0 emission from other actively actively star - forming galaxies ( ngc 253 , arp 299 , arp 220 , mrk 273 ) , previously interpreted as ionized reflection from their agn , are also consistent with our prediction for the wind fluid given the existing observational uncertainties . drawing on our results for m82
we have also presented a simple prescription for implementing starburst - driven superwinds in cosmological models of galaxy formation and evolution that more accurately represents the energetics of the hot metal - enriched phases than do existing recipes in the literature .
we thank the anonymous referee for their careful reading of this manuscript and their helpful recommendations .
this work was funded through grant nng05gf62 g of the nasa astrophysics theory program .
it is based on observations obtained with the _ chandra _ x - ray observatory , which is operated by the smithsonian astrophysical observatory on behalf of nasa , and _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and nasa .
the basic adiabatic cc model is a function of only three parameters : @xmath482 , @xmath347 , and @xmath115 . @xcite
were the first to point out that for certain values of these parameters , in particular small @xmath115 and/or high @xmath347 , the total luminosity predicted by the cc model was comparable to the net energy injection rate and that the analytical cc solution was no longer self - consistent as it assumed energy losses to be negligible .
they developed a self - consistent radiative solution , later tested with numerical hydrodynamical simulations @xcite .
there in not an exact and easily evaluated function that may be used to assess whether the adiabatic assumption is valid for any particular combination of @xmath482 , @xmath347 , and @xmath115 .
nevertheless , @xcite state that the threshold between the adiabatic and the fully radiative solution occurs when the radiative energy losses per unit volume exceed @xmath155% of the energy injection rate per unit volume .
@xcite show that this is most likely to occur at the very center of the star cluster or starburst region . in terms of the variables defined in
[ sec : theory ] , the adiabatic solution becomes invalid when @xmath483 equivalently , if @xmath484 then radiative losses are high and the cc model is invalid .
note the strong dependence on the mass injection rate and any mass loading , and inverse dependence on the energy injection rate , thermalization efficiency and effective radius .
the higher the mean energy per particle , or the larger the starbursting region is , the less - likely significant radiative losses become .
@xcite and @xcite provide examples of star clusters with properties such that radiative losses are severe .
however , the mass and energy injection rates they generally use have significantly lower values of the energy per particle ( and hence higher radiative losses for a given assumed metal abundance ) than what we consider appropriate for clusters in a m82-like starburst .
equation [ equ : app : radthresh ] may be rewritten as an explicit function of the cluster mass .
we rewrite the energy and mass injection rates as a function of the star cluster mass assuming in this case an instantaneous starburst model , such that @xmath485 and @xmath486 , where @xmath487 is the total cluster mass in units of @xmath488 ( again using a salpeter imf between the lower and upper mass limits of 1 @xmath172 ) .
note that the value of @xmath489 and @xmath490 are a functions of the age of the cluster , and are largest in magnitude for a cluster age of @xmath491 myr in the starburst99 models .
the critical cluster mass above which cooling exceeds 30% of the energy injection and the cc model is no longer valid is then @xmath492 for the purposes of illustration we consider a cluster of age @xmath493 myr , for which @xmath494 and @xmath495 .
for unit @xmath29 , @xmath30 , and @xmath226 the central temperature in the cc model would be @xmath496 ( equivalent to @xmath497 ) .
assuming @xmath309 the cooling function is @xmath498 cm@xmath208 @xmath500 .
the massive star clusters observed by @xcite in the nuclear starburst region of m82 have half - light radii of between 1 3 pc .
we adopt a typical cluster radius of @xmath501 = 2 pc . substituting these values into equation [ equ : app : m6_v1 ]
we find @xmath502 for unit thermalization efficiency and no additional mass loading then no realistic cluster will lie in the radiative regime .
significant mass loading within the cluster ( e.g. @xmath503 ) or low thermalization efficiency ( @xmath504 ) are required to reduce @xmath487 to of order unity for a @xmath493 myr old cluster .
an older cluster is even less likely to be radiative . from the starburst99 models we find that at an age of 20 myr the mass injection from stellar winds has declined dramatically yielding a net mass injection rate a factor 5 lower than at an age of 5 myr .
the energy injection rate is also lower , but only by a 2.5 . at this age
even clusters of mass a few times @xmath505 are still adiabatic even if @xmath503 or @xmath504 .
we rewrite the energy and mass injection rates as a function of the star formation rate @xmath191 , so that @xmath506 and @xmath507
. if we use the values given in table [ tab : burst ] for constant star formation with an age of 30 myr , again assuming a metal abundance of @xmath309 , we find that the star formation rate must exceed @xmath508 before the adiabatic cc solution becomes invalid .
for m82 the star formation rate is @xmath509
@xmath192 for the salpeter imf assumed , and the starburst region has effective radius of @xmath510 pc at smallest ( and more likely @xmath511 pc ) . thus the cc model is valid unless mass loading is significant @xmath329 ) or thermalization is inefficient @xmath512 , or some combination of the two .
note that a low value for the participation fraction @xmath226 will reduce the tendency of the starburst region to be radiative .
we conclude that under most , but not all , conditions the adiabatic @xcite model is appropriate for m82 s nuclear starburst .
it is possible to use the spherical analytical cc model to predict the wind - fluid properties and emission from a disk - like starburst of diameter @xmath203 and total height @xmath204 by applying modest numerical corrections that are a fixed function of the geometry alone .
as long as the adiabatic assumption is valid these correction factors do not depend on the starburst energy or mass injection rates , and are thus independent of the thermalization efficiency @xmath29 and the mass loading factor @xmath30 .
in the analytical cc model the surface area of the starburst region @xmath513 controls the gas density within the starburst region by limiting the rate at which matter can flow out of it , as described in [ sec : ccmodel ] ( see in particular equ .
30 of @xcite ) .
note that other parameters , in particular the central temperature and the terminal velocity of the wind are independent of the starburst geometry .
thus , for a given fixed energy and mass injection rate , the central density and pressure in a disk - like starburst of surface area @xmath514 should be very similar or identical to values in a spherical starburst of the same surface area @xmath515 .
we tested whether this is indeed the case using the vh-1 hydrodynamical code run in 2-d cylindrical symmetry as described in [ sec : hydromodel ] .
this code reproduces the standard spherical cc solution to within an accuracy of 0.1% for an assumed spherical starburst . for a disk - like starburst of diameter @xmath221 pc and @xmath222 pc ( [ sec :
theory : applying : sbsize ] ) a spherical starburst of equivalent surface area has @xmath308 pc .
we find that the central density and pressure predicted by vh-1 for this disk - like starburst geometry differs from the values it predicts for a @xmath308 pc spherical starburst by only 3.1% .
adopting another disk - like geometry with the same surface area ( @xmath516 pc and @xmath517 pc ) yields central density and pressures within 4.2% of the spherical solution . for the other disk - like starburst geometry suggested by observations of m82 , with @xmath518 pc and @xmath519 pc
the 2-d simulations predict central densities and pressures within 3.7% of the spherical starburst with equivalent surface area ( @xmath520 pc ) .
these simulations demonstrate that we can accurately predict the gas properties _ within _ a disk - like starburst region using the spherical cc model of equivalent starburst surface area .
the flow pattern exterior to the starburst region is also influenced by the geometry of the starburst region .
a disk with @xmath521 will weakly collimate the flow along the minor axis and give the superwind a preferred direction even in the absence of external collimation by the ism . in fig .
[ fig:2dsim ] we illustrate this by plotting the density and velocity fields for equivalent spherical and disk - like starbursts . for our purposes
the main effect is that the resulting non - spherical flow alters the total volume emission integral within any fixed region that is larger than the starburst region . within the starburst region itself
this effect is not particularly significant when evaluating total gas masses or emission integral given the relative uniformity of the density and pressure .
however , outside and above a disk - like starburst region the gas density drops off with increasing height more slowly that the @xmath522 predicted by the cc model . calculating
the correct emission integral for a disk - like starburst based the results of the spherical cc model thus entails some geometry - based corrections because of the different starburst region volumes and secondly due to the non - spherical flow pattern outside the starburst region .
the first correction is trivial to evaluate analytically .
the second correction must be determined from a full hydrodynamical simulation of the specific disk - like starburst , and further depends on the size of region @xmath523 we wish to evaluate the emission integral over .
our procedure for calculating the gas properties and total emission of disk - like starburst regions using the spherical cc model is as follows : 1 . for a given starburst geometry evaluate the surface area
@xmath515 , and run a simulation using the cc model using the effective starburst radius @xmath524 .
this accurately predicts the values of fluid variables such as density , temperature and pressure _ within _ the starburst region .
2 . macroscopic variables , in particular the volume emission integral , require further geometrical corrections to make the emission integral predicted using the spherical cc model , @xmath525 , match the emission integral found in a full multi - dimensional simulation of a non - spherical starburst region @xmath526 .
3 . the volume of the starburst region in the spherical model is larger than the volume of the disk - like starburst region , so the volume emission integral predicted by the cc model within some radius @xmath527 must be scaled by a factor @xmath528 equal to the ratio of the starburst region volumes : @xmath529 4 .
a full hydrodynamical simulation for a given disk - like starburst must be run to evaluate the correction @xmath530 to the emission integral within the chosen @xmath523 due to the non - spherical flow pattern manifest outside the starburst region .
we use @xmath531 pc to match the region from which the m82 nuclear region spectra analyzed in paper i were extracted .
simulations of different disk - like starburst regions where we altered the thermalization efficiency and mass loading demonstrated that the correction @xmath532 to the total emission integral is a fixed function of the geometry , and is independent of the energy and mass injection rates .
the temperature - dependent differential emission measure for the scaled cc model @xmath533 is calculated using the factors discussed above : @xmath534 .
combining this with the temperature - dependent cooling functions and summing over all temperatures gives the final predicted luminosity for the chosen energy band or line . for @xmath531 pc and a disk - like starburst with @xmath221 pc and @xmath222 pc , the primary geometry we apply to m82 , the non - spherical emission integral correction factor @xmath535 .
for the alternate disk - like starburst geometry with @xmath518 pc and @xmath519 pc the correction factor is @xmath536 .
for the compact starbursting disk with @xmath216 pc and @xmath217 pc the correction factor is @xmath537 . with these scalings
the spherical cc model can be used to predict to high accuracy the total emission integral from the wind fluid within @xmath523 for a disk - like starburst .
this method is less accurate at reproducing the luminosity predicted by a full 2-dimensional hydrodynamical simulation because non - spherical flow pattern leads to subtle differences in the shape of the differential emission integral ei(t ) compared to a spherical cc solution .
we compared the luminosities predicted by a small suite of hydrodynamical models with different disk - like starburst geometries to the results from the scaled cc analytical model in order to assess the significance of this effect .
the error introduced by the effect is small in terms of computing the total cooling rate ( @xmath538% for the sutherland & dopita cooling curve ) , or even the the x - ray luminosity over the broad 0.3 2.0 kev or 2 8 kev energy ranges ( @xmath374% ) . for the high energy lines we are interested in because of their temperature sensitivity the effect is larger , and
the associated errors in predicted line luminosity can be as large as @xmath155% .
nevertheless these uncertainties do not compromise the usefulness of the scaled cc model as they are comparable to the observational uncertainties in the nuclear region s , ar and ca luminosities of the well detected lines from the _ xmm - newton _ data , and less than the uncertainties in the diffuse fe he@xmath0 line luminosity derived from the _ chandra _ acis - s data .
the relative effect of these uncertainties on the region of model parameter space matching our chosen constraints can be seen in fig . [
fig : vh1_vs_scaledcc ] .
there are two main environments that in principle might reduce the fraction @xmath226 of all the sne and stellar winds in a starburst region that can contribute to powering a superwind .
the first , likely to to be the most significant but also hardest to estimate , is physical confinement and increased radiative cooling of young supernova remnants and stellar winds occurring in regions of high ambient gas density such as molecular clouds .
the second is that radiative losses may be high for those sne and stellar winds associated with massive stars in massive and compact star clusters ( particularly super star clusters ) , where the adiabatic cc solution is no longer valid .
there are some observational findings that are suggest that some fraction of the sne in m82 s nuclear starburst occurred within regions of gas number density @xmath539 @xcite , although there are problems with this interpretation ( see * ? ? ?
* and references therein ) . however , based on the total molecular gas mass of @xmath540 within the starburst region , the fraction of the total starburst region volume occupied by such dense gas must be @xmath541 or lower @xcite .
massive stars will be born in regions of high gas density , so that if confinement does occur the magnitude of the drop in @xmath226 may well be larger the the filling factor of the dense gas . at present
we can not rigorously quantify the effectiveness of confinement , nor the fraction of stars that might be affected .
@xcite speculate that perhaps as much as 50% of the sne in m82 occur in dense regions , in which case the participation factor may be as low as @xmath542 .
it is possible to quantitatively estimate the role that over - cooling of the thermalized sn and stellar wind ejecta in dense star clusters have on @xmath226 .
based on the discussion in appendix [ app : radiative_case ] we conclude that only the compact clusters of radii a few parsecs and masses @xmath543 ( salpeter imf , mass limits 1 100 @xmath441 ) are likely to suffer significant radiative losses ( and then only under certain conditions ) . for a star cluster mass function of the form @xmath544 between the mass limits @xmath545 and @xmath546 and where @xmath547 @xcite , the fraction of the total stellar mass in clusters with @xmath548 is @xmath549 we adopt a minimum cluster mass of @xmath550 and a maximum cluster mass of @xmath551 @xcite . for @xmath552 then @xmath553 .
thus the magnitude of the reduction in @xmath226 due to radiative cluster winds is small , of order 15% , even if sn and stellar wind thermalization is inefficient , and/or there are long - lived sources of additional mass loading in the massive star clusters . , m. & chieffi , a. 2007 , in american institute of physics conference series , vol .
924 , the multicolored landscape of compact objects and their explosive origins , ed .
t. di salvo , g. l. israel , l. piersant , l. burderi , g. matt , a. tornambe , & m. t. menna , 226233 | we measure the net energy efficiency of supernova and stellar wind feedback in the starburst galaxy m82 and the degree of mass - loading of the hot gas piston driving its superwind by comparing a large suite of 1 and 2-dimensional hydrodynamical models to a set of observational constraints derived from hard x - ray observations of the starburst region ( the fluxes of the he@xmath0 and ly@xmath0-like lines of s , ar , ca and fe , along with the total diffuse @xmath1 8 kev x - ray luminosity ) .
these are the first direct measurements of the feedback efficiency and mass - loading of supernova heated and enriched plasma in a starburst galaxy .
we consider a broad range of plausible parameters for the m82 starburst , varying the age and mode of star formation , the starburst region size and geometry , and supernova metal yields .
over all these varied input parameters all the models that satisfy the existing observational constraints have medium to high thermalization efficiencies ( @xmath2% @xmath3% ) and the volume - filling wind fluid that flows out of the starburst region is only mildly centrally mass loaded ( @xmath4 ) .
these results imply a temperature of the plasma within the starburst region in the range 30 80 million kelvin , a mass flow rate of the wind fluid out of the starburst region of @xmath5 @xmath6 and a terminal velocity of the wind in the range @xmath7 @xmath8 .
this velocity is considerably larger than the escape velocity from m82 ( @xmath9 ) and the velocity of the h@xmath0 emitting clumps and filaments within m82 s wind ( @xmath10 ) .
drawing on these results we provide a prescription for implementing starburst - driven superwinds in cosmological models of galaxy formation and evolution that more accurately represents the energetics of the hot metal - enriched phases than the existing recipes do . |
public disclosure of datasets containing _ micro - data _ ,
i.e. , information on precise individuals , is an increasingly frequent practice .
such datasets are collected in a number of different ways , including surveys , transaction recorders , positioning data loggers , mobile applications , and communicaiton network probes .
they yield fine - grained data about large populations that has proven critical to seminal studies in a number of research fields . however , preserving user privacy in publicly accessible micro - data datasets is currently an open problem .
publishing an incorrectly anonymized dataset may disclose sensible information about specific users .
this has been repeatedly proven in the past .
one of the first and best known attempts at re - identification of badly anonymized datasets was carried out by then mit graduate student latanya sweeney @xcite in 1996 . by using a database of medical records released by an insurance company and the voter roll for the city of cambridge ( ma ) , purchased for 20 us dollars , dr .
sweeney could successfully re - identify the full medical history of the then governor of massachusetts , william weld .
she even sent the governor full health records , including diagnoses and prescriptions , to his office . a later , yet equally famous experiment was performed by narayanan _
et al . _
@xcite on a dataset released by netflix for a data - mining contest , which was cross - correlated with a web scraping of the popular imdb website .
the authors were able to match two users from both datasets revealing , e.g. , their political views .
mobile traffic datasets include micro - data collected at different locations of the cellular network infrastructure , concerning the movements and traffic generated by thousands to millions of subscribers , typically for long timespans in the order of weeks or months .
they have become a paramount instrument in large - scale analyses across disciplines such as sociology , demography , epidemiology , or computer science .
unfortunately , mobile traffic datasets may also be prone to attacks on individual privacy .
specifically , they suffer from the following two issues . 1 .
* elevate uniqueness . *
mobile subscribers have very distinctive patterns that often make them unique even within a very large population .
zang and bolot @xcite showed that 50% of the mobile subscribers in a 25 million - strong dataset could be uniquely detected with minimal knowledge about their movement patterns , namely the three locations they visit the most frequently .
the result was corroborated by de montjoye _ et al . _
@xcite , who demonstrated how an individual can be pinpointed among 1.5 million other mobile customers with a probability almost equal to one , by just knowing five random spatiotemporal points contained in his mobile traffic data .
uniqueness does not implies identifiability , since the sole knowledge of a unique subscriber trajectory can not disclose the subscriber s identity .
building that correspondence requires instead sensible side information and cross - database analyses similar to those carried out on medical or netflix records .
to date , there has been no actual demonstration of subscriber re - identification from mobile traffic datasets using such techniques and our study does not change that situation . still , uniqueness may be a first step towards re - identification , and whether this represents a threat to user privacy is an open topic for discussion @xcite .
in such a context , the standard , safe approach to ensure data confidentiality relies on non - technical solutions , i.e. , non - disclosure agreements that well define the scope of the activities ( e.g. , fundamental research only ) carried out on the datasets , and that prevent open disclosure of the data or results without prior verification by the relevant authorities .
this is , for instance , the solution adopted in the case of the mobile traffic information we will consider in sec.[sec : datasets ] .
clearly , this practice can strongly limit the availability of mobile traffic datasets , as well as the reproducibility of related research .
mitigating the uniqueness of subscriber trajectories becomes then a very desirable facility that can entail more privacy - preserving datasets , and favor their open circulation .
it is however at this point that the second problem of mobile traffic datasets comes into play . 1 .
* low anonymizability . *
the legacy solution to reduce uniqueness in micro - data datasets is generalization and suppression .
however , previous studies showed that blurring users in the crowd , by reducing the spatial and temporal granularity of the data , is hardly a solution in the case of mobile traffic datasets .
zang and bolot @xcite found that reliable anonymization is attained only under very coarse spatial aggregation , namely when the mobile subscriber location granularity is reduced to the city level .
similarly , de montjoye _ et al . _
@xcite proved that a power - law relationship exists between uniqueness and spatiotemporal aggregation of mobile traffic .
this implies that privacy is increasingly hard to ensure as the resolution of a dataset is reduced . in conclusion ,
not only mobile traffic datasets yield highly unique trjectories , but the latter are also hard to anonymize . ensuring individual privacy risks to lower the level of detail of such datasets to the point that they are not informative anymore . in this work ,
we aim at better investigating the reasons behind such inconvenient properties of mobile traffic datasets .
we focus on anonymizability , since it is a more revealing feature : multiple datasets that feature similar trajectory uniqueness may be more or less difficult to anonymize . attaining our objective brings along the following contributions : ( i ) we define a measure of the level of anonymizability of mobile traffic datasets , in sec.[sec : metric ] ; ( ii ) we provide a first assessment of the anonymizability of two large - scale mobile traffic datasets , in sec.[sec : datasets ] ; ( iii )
we unveil the cause of elevate uniqueness and poor anonymizability in such datasets , i.e. , the heavy tail of the temporal diversity among subscriber mobility patterns , in sec.[sec : results ] . finally , sec.[sec : conc ] concludes the paper .
in this section , we first define in a formal way the problem of user uniqueness in mobile traffic datasets , in sec.[sub : problem ] .
then , we introduce the proposed measure of anonymizability , in sec.[sub : measure ] . in order to properly define the problem we target , we need to introduce the notion of mobile traffic fingerprint that is at the base of the mobile traffic dataset format .
we also need to specify the type of anonymity we consider in our case , @xmath0-anonymity .
next , we discuss these aspects of the problem . [
tab : db_std ] .standard micro - data database format . [ cols="<,<,<,<,<,<,<",options="header " , ] [ tab : db_mt ] a & @xmath1,8 & @xmath2,14 & @xmath3,17 & + b & @xmath4,8 & @xmath5,15 & @xmath6,15 & & @xmath7,15 & @xmath8,16 & @xmath9,17 + c & @xmath10,7 & @xmath11,20 & + & & + traditional micro - data databases are structured into matrices where each row maps to one individual , and each column to an _
attribute_. an example is provided in tab.[tab : db_std ] .
individuals are associated to one _ identifier _ , i.e. , a value that uniquely pinpoints the user across datasets ( e.g. , his complete name , social number , or passport number ) .
since identifiers allow direct identification and immediate cross - database correlation , they are never disclosed .
instead , they are replaced by a _ pseudo - identifier _ , which is again unique for each individual , but changes across datasets ( e.g. , a random string substituting the actual identifier ) . then , standard re - identification attacks leverage _ quasi - identifiers _ , i.e. , a sequence of known attributes of one user ( e.g. , the age , gender , zip code , etc . ) to recognize the user in the dataset . if successful , the attacker has then access to the complete record of the target user .
this knowledge can directly include sensitive attributes , i.e. , items that should not be disclosed because they may pertain to the personal sphere of the individual ( e.g. , diseases , political or religious views , sexual orientation , etc . ) .
it can also be exploited for further cross - database correlation so as to extract additional private information about the user .
the same model directly applies to the case of mobile traffic datasets .
however , the database semantics make all the difference here : while mobile users are the obvious individuals whose privacy we want to protect , attributes are now sequences of spatiotemporal samples .
each sample is the result of an event that the cellular network associated to the user .
an illustration is provided in fig.[fig : map_def ] , which portrays the trajectories of three mobile customers , denoted with pseudo - identifiers @xmath12 , @xmath13 , and @xmath14 , respectively , across an urban area .
user @xmath12 interacts with the radio access infrastructure at 8 am , while he is in cell @xmath1 along his trajectory .
then , he triggers additional mobile traffic activities at 2 pm , while located in a cell @xmath2 in the city center , and at 5 pm , from a cell @xmath3 in the south - east city outskirts .
the same goes for users @xmath13 and @xmath14 .
all these spatiotemporal samples are recorded by the mobile operator and constitute the _ mobile traffic fingerprint _ of the user .
the resulting database has a format such as that in tab.[tab : db_mt ] , where subscriber identifiers are replaced by pseudo - identifiers , and each element of a user s fingerprint is a cell and hourly timestamp pair . in order to preserve user privacy in micro - data
, one has to ensure that no individual can be uniquely pinpointed in a dataset .
this principle has led to the definition of multiple notions of non - uniqueness , such as @xmath0-anonymity @xcite , @xmath15-diversity @xcite and @xmath16-closeness @xcite . among those ,
@xmath0-anonymity is the baseline criterion , to which @xmath15-diversity or @xmath16-closeness add further security layers that cope with sensitive attributes or cross - database correlation .
more precisely , @xmath0-anonymity ensures that , for each individual , the set of attributes ( or its quasi - identifier subset ) is identical to that of at least other @xmath0 - 1 users . in other words ,
each individual is always hidden in a crowd of @xmath0 , and thus he can not be uniquely identified among such other users .
granting @xmath0-anonymity in micro - data databases implies generalizing and suppressing data .
as an example , in order to ensure @xmath17-anonymity on the age and zip code attributes for the first user in tab.[tab : db_std ] , one can aggregate the age in twenty - year ranges , and the zip codes in three - number ranges : both the first and second user end up with a ` ( 20,40 ) ` age and ` 770 * * ` zip code , which makes them both @xmath17-anonymous . clearly , the process is lossy , since the information granularity is reduced .
many efficient algorithms have been proposed that achieve @xmath0-anonymity in legacy micro - data databases , while minimizing information loss @xcite .
also in mobile traffic datasets , @xmath0-anonymity is regarded as a best practice , and data aggregation is the common approach to achieve it @xcite . in this case
, one has to ensure that the fingerprint of each subscriber is identical to that of at least other @xmath0 - 1 mobile users in the same dataset .
we remark that previous works have typically considered a model of attacker who only has partial knowledge of the subscribers fingerprints , e.g. , most popular locations @xcite or random samples @xcite . in order to counter such a attack model , a partial @xmath0-anonymization , targeting the limited information owned be the attacker , would be sufficient .
however , we are interested in a general solution , so we do not make any assumption on the precise knowledge of the attacker , which can be diverse and possibly broad . thus
, @xmath0-anonymizing the whole fingerprint of each subscriber in the mobile traffic dataset is the only way to deterministically ensure mobile user privacy .
both spatial and temporal aggregations can be leveraged to attain this goal .
examples are provided in fig.[fig : map_arr-2h ] and fig.[fig : map_half-12h ] . in fig.[fig : map_arr-2h ] , cells are aggregated in large sets that roughly map to the nine major neighborhoods of the urban area ; also , time is aggregated in two - hour intervals . the reduction of spatiotemporal granularity allows @xmath17-anonymizing mobile users @xmath12 and @xmath13 : both have now a fingerprint composed by samples ` ( v,8 - 9 ) ` , ` ( iii,14 - 15 ) ` , and ` ( vii,16 - 17 ) ` .
user @xmath14 has instead a different footprint , with samples ` ( iv,6 - 7 ) ` and ` ( iii,20 - 21 ) ` .
if we need to @xmath18-anonymize all three mobile customers in the example , then a further generalization is required , as in fig.[fig : map_half-12h ] .
there , the metropolitan region is divided in west and east halves , and only two time intervals , before and after noon , are considered .
the result is that all subscribers @xmath12 , @xmath13 , and @xmath14 have identical fingerprints ` ( west,1 - 12 ) ` and ` ( east,13 - 24 ) ` .
clearly , this level of anonymization comes at a high cost in terms of information loss , as the location data is very coarse both in space and time .
this is precisely the problem of low anonymizability of mobile traffic datasets unveiled by previous works @xcite : even guaranteeing @xmath17-anonymization in a very large population requires severe reductions of the spatiotemporal granularity , which limits the usability of the data .
we intend to devise a measure of anonymizability that is based on the @xmath0-anonymity criterion .
thus , our proposed measure evaluates the effort , in terms of data aggregation , needed to make a user indistinguishable from @xmath0 - 1 other subscribers .
we start by defining the distance between two spatiotemporal samples in the mobile traffic fingerprints of two mobile users . each sample is composed of a spatial information ( e.g. , the cell location ) and a temporal information ( e.g. , the timestamp ) .
the distance must keep into account both dimensions . a generic formulation of the distance between the @xmath19-th sample of @xmath12 s fingerprint , @xmath20 , and the @xmath21-th sample of @xmath13 s fingerprint , @xmath22 , is @xmath23 here , @xmath24 and @xmath25 are functions that determine the distance along the spatial and temporal dimensions , respectively .
the former thus operates on the spatial information in the two samples , @xmath26 and @xmath27 , and the latter on the temporal information , @xmath28 and @xmath29 .
the factors @xmath30 and @xmath31 weigth the spatial and temporal contributions in ( [ eqn : sampledist ] ) . in the following , we will assume that the two dimensional have the same importance , thus @xmath32 .
we shape the @xmath24 and @xmath25 functions by considering that both spatial and temporal aggregations induce a loss of information that is linear with the decrease of granularity .
however , above a given spatial or temporal threshold , the information loss is so severe that the data is not usable anymore . as a result
, the functions can be expressed as @xmath33 and @xmath34 in ( [ eqn : deltas ] ) , @xmath35 is the _ taxicab distance _
@xcite between the spatial components of the samples , whose coordinates are denoted as @xmath36 and @xmath37 in a valid map projection system .
both functions fulfill the properties of distances , i.e. , are positive definite , symmetric , and satisfy the triangle inequality .
they range from @xmath38 ( samples are identical from a spatial or temporal viewpoint ) to @xmath39 ( samples are at or beyond the maximum meaningful aggregation threshold ) . concerning the values of the thresholds , in the following we will consider that the aggregation limits beyond which the information deprivation is excessive are 20 km for the spatial dimension ( i.e. , the size of a city , beyond which all intra - urban movements are lost ) and 8 hours ( beyond which the night , working hours , and evening periods are merged together ) . the sample distance in ( [ eqn : sampledist ] ) can be used to define the distance among the whole fingerprints of two mobile subscribers @xmath12 and @xmath13 , as @xmath40 here , @xmath41 and @xmath42 are the cardinalities of the fingerprints of @xmath12 and @xmath13 , respectively . the expression in ( [ eqn : fingerdist ] ) takes the longer fingerprint between the two , and finds , for each sample , the sample at minimum distance in the shorter fingerprint .
the resulting @xmath43 is the average among all such sample distances , and @xmath44 , @xmath45 .
the measure of anonymizability of a generic mobile user @xmath12 can be mapped , under the @xmath0-anonymity criterion , to the average distance of his fingerprint from those of the nearest @xmath0 - 1 other users .
formally @xmath46 where @xmath47 is the set of @xmath48 users @xmath13 with the smallest fingerprint distances to that of @xmath12 .
the expression in ( [ eqn : delta ] ) returns a measure @xmath49 $ ] that indicates how hard it is to hide subscriber @xmath12 in a the crowd of @xmath0 users .
if @xmath50 , then the user is already @xmath0-anonymized in the dataset . if @xmath51 , the user is completely isolated , i.e. , no sample in the fingerprints of all other subscribers is within the spatial and temporal thresholds , @xmath52 and @xmath53 , from any samples of @xmath12 s fingerprint .
we employ the proposed measure to assess the level of anonymizability of fingerprints present in two mobile traffic datasets released by orange in the framework of the data for development challenge . in order to allow for a fair comparison , we preprocessed the datasets so as to make them more homogeneous .
* * ivory coast .
* released for the 2012 challenge , this dataset describes five months of call detail records ( cdr ) over the whole the african nation of ivory coast .
we used the high spatial resolution dataset , containing the complete spatio temporal trajectories for a subset of 50,000 randomly selected users that are changed every two weeks .
thus , the dataset contains information about 10 2-weeks periods overall .
we performed a preliminary screening , discarding the most disperse trajectories , keeping the users that have at least one spatio - temporal point per day .
then , we merged all the user that met this criteria in a single dataset , so as to achieve a meaningful size of around 82,000 users .
this dataset is indicated as ` d4d - civ ` in the following . *
* senegal . *
the 2014 challenge dataset is derived from cdr collected over the whole senegal for one year .
we used the fine - grained mobility dataset , containing a randomly selected subset of around 300,000 users over a rolling 2-week period , for a total of 25 periods .
we did not filter out subscribers , since the dataset is already limited to users that are active for more than 75% of the 2-week time span . in our study
, we consider one representative 2-week period among those available . this dataset is referred to as ` d4d - sen ` in the following . in both the mobile traffic datasets ,
the information about the user position is provided as a latitude and longitude pair .
we projected the latter in a two - dimensional coordinate system using the lambert azimuthal equal - area projection .
we then discretize the resulting positions on a 100-m regular grid , which represents the maximum spatial granularity we consider .
as far as the temporal dimension is concerned , the maximum precision granted by both datasets is one minute , and this is also our finest time granularity .
the measure of anoymizability in ( [ eqn : delta ] ) can be intended as a dissimilarity measure , and employed in legacy definitions used to understand micro - data database sparsity , e.g. , ` ( , ) -sparsity ` @xcite . however , these definitions are less informative than the complete distribution of the anonymizability measure .
thus , in this section , we employ cumulative distribution functions ( cdf ) of the measure in ( [ eqn : delta ] ) in order to assess the anonymizability of the two datasets presented before .
-anonymity criterion , in the ` d4d - civ ` and ` d4d - sen ` mobile traffic datasets . ]
our basic result is shown in fig.[fig : d4d_cdf ] .
the plot portrays the cdf of the anonymizability measure computed on all users in the two reference mobile traffic datasets , ` d4d - civ ` and ` d4d - sen ` , when considering @xmath17-anonymity as the privacy criterion .
we observe that the two curves are quite similar , and both are at zero in the x - axis origin .
this means that no single mobile subscriber is @xmath17-anonymous in either of the original datasets .
since similar observations were made on different data @xcite , our results seem to confirm that the elevate uniqueness of subscriber trajectories is an intrinsic property of any mobile traffic dataset , and not just a specificity of those we analyse in this study .
more interestingly , the probability mass gathered in both cases in the @xmath54-@xmath55 range , i.e. , it is quite close to the origin .
this is good news , since it implies that the average aggregation effort needed to achieve @xmath17-anonymity is not elevate .
as an example , 50% of the users in the ` d4d - civ ` dataset have a measure @xmath56 or less , which maps , on average , to a combined spatiotemporal aggregation of less than one km and little more than 20 minutes .
in other words , the result seems to suggest that half of the individuals in the dataset can be @xmath17-anonymized if the spatial granularity is decreased to 1 km , and the temporal precision is reduced to around 20 minutes .
similar considerations hold in the ` d4d - sen ` case , where , e.g. , 80% of the dataset population has a measure @xmath57 or less .
such a measure is the result of average spatial and temporal distances of 1.7 km and 41 minutes from @xmath17-anonymity .
one may wonder how more stringent privacy requirements affect these results .
fig.[fig : d4d_vark ] shows the evolution of the anonymizability of the two datasets when @xmath0 varies from @xmath17 to @xmath58 .
as expected , higher values of @xmath0 require that a user is hidden in a larger crowd , and thus shift the distributions towards the right , implying the need for a more coarse aggregation .
however , quite surprisingly , the shift is not dramatic : @xmath58-anonymity does not appear much more difficult to reach than @xmath17-anonymity .
unfortunately , the easy anonymizability suggested by the distributions is only apparent .
fig.[fig : d4d_agg ] depicts the impact of spatiotemporal generalization on anonymizability : each curve maps to a different level of aggregation , from @xmath58 meters and @xmath39 minute ( the finer granularity ) to @xmath59 km and 8 hours .
as one could expect , the curves are pushed towards smaller values of the anonymizability measure .
however , the reduction of spatiotemporal precision does not have the desired magnitude , and even a coarse - grained citywide , 8-hour aggregation can not @xmath17-anonymize but 30% of the mobile users .
this result is again in agreement with previous studies @xcite , and confirms that mobile traffic datasets are difficult to anonymize .
we are interested in understanding the reasons behind the incongruity above , i.e. , the fact that spatiotemporal aggregation yields such poor performance , even if the average effort needed to attain @xmath0-anonymity is in theory not elevate . to attain our goal
, we proceed along two directions .
first , we separate the spatial and temporal dimensions of the measure in ( [ eqn : delta ] ) , so as to understand their precise contribution to the dataset anonymizability .
second , we measure the statistical dispersion of the fingerprint distances along the two dimensions : the rationale is that we observed the average distance among fingerprints to be quite small , thus the reason of the low anonymizability must lie in the deviation of sample distances around that mean . formally , we consider , for each user @xmath12 in the dataset , the set @xmath47 of @xmath0 - 1 other subscribers whose fingerprints are the closest to that of @xmath12 , according to ( [ eqn : delta ] ) . then , we disaggregate all the fingerprint distances @xmath43 between @xmath12 and the users @xmath60 into sample distances @xmath61 , as per ( [ eqn : fingerdist ] ) . finally , we separately collect the spatial and temporal components of all such sample distances , in ( [ eqn : sampledist ] ) , into ordered sets @xmath62 and @xmath63 .
the resulting sets can be treated as disjoint distributions of the distances , along the spatial and temporal dimensions , between the fingerprint of a generic individual @xmath12 and those of the @xmath0 - 1 other users that show the most similar patterns to his .
examples of the spatial and temporal distance distributions we obtain in the case of @xmath17-anonymity are shown in fig.[fig : d4d_dis1]-[fig : d4d_dis5 ] .
each plot refers to one random user in the ` d4d - civ ` or ` d4d - sen ` dataset , and portrays the cdf of the spatial ( @xmath64 ) and temporal ( @xmath65 ) component distance , as well as that of the total sample distance ( @xmath66 ) . we can remark that temporal components typically bring a significantly larger contribution to the total fingerprint distance than spatial ones .
in fact , a significant portion of the spatial components is at zero distance , i.e. , is immediately @xmath17-anonymous in the original dataset .
the same is not true for the temporal components .
a rigorous confirmation is provided in fig.[fig : d4d_diswei ] , which shows the distribution of the temporal - to - spatial component ratios , i.e. , @xmath67 , for all subscribers @xmath12 in the two reference datasets .
the cdf is skewed towards high values , and for half of mobile subscribers in both ` d4d - civ ` or ` d4d - sen ` datasets temporal components contribute to 80% or more of the total sample distance .
we conclude that the temporal component of a mobile traffic fingerprint is much harder to anonymize than the spatial one . in other words , _
an individual generates mobile traffic activity is easily masked , but hiding _ when _ he carries out such activity it is not so .
not only temporal components weight much more than spatial ones in the fingerprint distance , but they also seem to show longer tails in fig.[fig : d4d_dis1]-[fig : d4d_dis5 ] .
longer tails imply the presence of more samples with a large distance : this , in turn , significantly increases the level of aggregation needed to achieve @xmath0-anyonimity , as the latter is only granted once all samples in the fingerprint have zero distance from those in the second fingerprint .
we rigorously evaluate the presence of a long tail of hard - to - anonymize samples by means of two complementary metrics , still separating their spatial and temporal components .
the first metric is the gini coefficient , which measures the dispersion of a distribution around its mean .
considering an ordered set @xmath68 , the coefficient is computed as @xmath69 where @xmath70 is the cardinality of @xmath71 .
we compute the gini coefficient on the sets @xmath72 and @xmath73 , for all users @xmath12 .
the second metric is the tail weight index @xcite , which quantifies the weight of the tail of a distribution with empirical cdf @xmath74 as @xmath75 in the expression above , @xmath76 is the inverse function of the empirical cdf and @xmath77 is the inverse function of a standard normal cdf .
we compute again the tail weight index on the distributions obtained from both @xmath72 and @xmath73 , for all @xmath12 .
fig.[fig : d4d_ginitail ] shows the results returned by the two metrics in the ` d4d - civ ` or ` d4d - sen ` datasets .
no significant differences emerge among the two mobile traffic datasets . in both cases , the gini coefficient , in fig.[fig : d4d_gini_sen ] and fig.[fig : d4d_gini_civ ] ,
has , for all mobile user fingerprints ( @xmath66 ) , high values around @xmath78 that denote significant dispersion around the mean .
however , two opposite behaviors are observed for the spatial ( @xmath64 ) and temporal ( @xmath65 ) components .
the former show cases where no dispersion at all is recorded ( coefficient close to zero ) , and cases where the distribution is very sparse . the latter has the same behavior as the overall distance , with values clustered around @xmath78 .
the result ( i ) corroborates the observation that the overall anonymizability is driven by distances along the temporal dimension , and ( ii ) imputes the latter to the complete absence of easy - to - anonymize short tails in the distribution of temporal distances .
fig.[fig : d4d_tail_sen ] and fig.[fig : d4d_tail_civ ] show instead the cdf of tail weight indices . here , the result is even more clear : the tail of temporal component distances is typically much longer than that of spatial ones , and in between those of exponential and heavy - tailed distributions .
once more , the temporal component tail fundamentally shapes that of the overall fingerprint distance .
at the light of all previous observations , we confirm the findings of previous works on user privacy preservation in mobile traffic datasets .
namely , the two datasets we analysed do not grant @xmath0-anonymity , not even for the minimum @xmath0 = 2 .
moreover , our reference datasets show poor anonymizability , i.e. , require important spatial and temporal generalization in order to slightly improve user privacy .
the fact that these properties have been independently verified across diverse datasets of mobile traffic suggests that the elevate uniqueness of trajectories and low anonymizability are intrinsic properties of this type of datasets . in our case , even a citywide ,
8-hour aggregation is not sufficient to ensure complete @xmath17-anonymity to all subscribers .
the result is even worse than that observed in previous studies : the difference is due to the fact that we consider the anonymization of complete subscriber fingerprints , whereas past works focus on simpler obfuscation of summaries @xcite or subsets @xcite of the fingerprints . on the one hand ,
the typical mobile user fingerprint in such datasets is composed of many spatiotemporal samples that are easily hidden among those of other users in the dataset .
this leads to fingerprints that appear easily anonymizable , since their samples can be matched , _ on average _ , with minimal spatial and temporal aggregation .
on the other hand , mobile traffic fingerprints tend to have a non - negligible number of elements that are much more difficult to anonymize than the average sample .
these elements , which determine a characteristic dispersion and long - tail behavior in the distribution of fingerprint sample distances , are mainly due to a significant diversity along the temporal dimension . in other words
, mobile users may have similar spatial fingerprints , but their temporal patterns typically contain a non - negligible number of dissimilar points .
it is the presence of these hard - to - anonymize elements in the fingerprint that makes spatiotemporal aggregation scarcely effective in attaining anonymity .
indeed , in order to anonymize a user , one needs to aggregate over space and time , until all his long - tail samples are hidden within the fingerprints of other subscribers . as a result ,
even significant reductions of granularity ( and consequent information losses ) may not be sufficient to ensure non - uniqueness in mobile traffic datasets . as a concluding remark , we recall that such uniqueness does not implies direct identifiability of mobile users , which is much harder to achieve and requires , in any case , cross - correlation with non - anonymized datasets . instead , uniqueness is a first step towards re - identification .
understanding its nature can help developing mobile traffic datasets that are even more privacy - preserving , and thus more easily accessible .
h. zang and j. bolot _ anonymization of location data does not work : a large - scale measurement study_. proceedings of the 17th annual international conference on mobile computing and networking .
acm , 2011 . | preserving user privacy is paramount when it comes to publicly disclosed datasets that contain fine - grained data about large populations .
the problem is especially critical in the case of mobile traffic datasets collected by cellular operators , as they feature elevate subscriber trajectory uniqueness and they are resistant to anonymization through spatiotemporal generalization . in this work ,
we investigate the @xmath0-anonymizability of trajectories in two large - scale mobile traffic datasets , by means of a novel dedicated measure .
our results are in agreement with those of previous analyses , however they also provide additional insights on the reasons behind the poor anonimizability of mobile traffic datasets . as such , our study is a step forward in the direction of a more robust dataset anonymization . |
dwarf galaxies are believed to represent the dominant population , by number , of the present day universe , and a major constituent of groups and clusters .
studying their star formation and chemical enrichment histories is the key to understand the evolution of galaxies on cosmological time scales ( e.g. madau _ et al .
local group galaxies are ideal laboratories to detailed studies of dwarf galaxy properties : we can resolve their individual stars and thus learn about their star formation histories ( sfh ) by exploring ages , metallicities , and the spatial distribution of the stellar populations they contain . the color - magnitude diagram ( cmd )
is the best tool for retrieving the sfh of a stellar system .
cmds reaching at least the brightest part of the rgb or much better , the oldest main - sequence ( ms ) turnoffs , display stars born throughout the life - time of a galaxy and are fossil records of its sfh . at 50 and 60 kpc ,
our nearest irregular neighbours , the magellanic clouds ( mcs ) are an optimal workplace to carry out detailed sfh studies .
thanks to their proximity to the galaxy , it is possible to obtain deep cmds well down the ms turnoff stars ( m@xmath3@xmath0 + 4 ) . in spite of the proximity and the intrinsec interest ,
is remarcable that there are still important gaps in the knowledge of the mcs ( e.g. olszewski _ et al .
_ 1996 ) . maybe their huge projected size , and the considerably big number of stars to be analized explain the situation . regarding the field population of the small magellanic cloud ( smc ) , this galaxy lacks large area studies reaching faint limiting magnitudes , equivalent to those of gallart _ et al . _
( 2004 ) and gallart _ et al . _
( 2005 ) for the large magellanic cloud ( lmc ) .
gardiner & hatzidimitriou ( 1992 ) were pioneers in relatively deep wide field studies in the smc for ages greater than 0.1 gyr .
they covered the external region ( @xmath42@xmath1 from the center ) . from their cmds ( which reached the horizontal branch at r@xmath020 mag )
the authors infered an average age of @xmath010 gyr and interpreted the existence of a red horizontal branch as an evidence of a 15 - 16 gyr population , comprising around 7% of the total mass .
however this work is mainly limited by the fact that the cmds do not reach the old ms turnoffs . dolphin _
( 2001 ) presented a combination of hst and ground - based _
v _ and _ i _ images of a smc field near ngc 121 . from their ground - based images of the field
, they infered a peak of star formation between 5 and 8 gyr ago , with a medium age of 7.5 gyr , and that 14%@xmath55% of the stars were formed more than 11 gyr ago .
however , because of the small field analized these results should not be considered as representative of the whole smc .
more recently harris & zaritsky ( 2004 ) , presented the sfh and chemical enrichment history of an 4@xmath1@xmath64.@xmath15 area centered on the smc , based on _ ubvi _ photometry from the magellanic clouds photometric survey ( zaritsky _ et al .
they found that the smc formed @xmath050% of its total stellar population prior to 8.4 gyr ago , and that there was a quiescent epoch between 3 and 8.4 gyr ago , followed by more or less continuous star formation starting about 3 gyr ago and extending to the present .
we present here twelve unprecedented deep _ br _ smc cmds ranging from @xmath01@xmath1 to @xmath04@xmath1 from the center of the smc and located at different position angles around the galaxy .
these cmds clearly show the stellar population gradients in the smc .
their depth , reaching the oldest ms turnoffs with very good photometric accuracy , will allow us to obtain detailed sfhs from all of them , and to investigate , therefore , the variation of the sfh accross the whole smc field .
the observations were made at the 100-inch telescope at las campanas observatory . throughout our four years campaign ( 2001 - 2004 ) ,
_ b_-band and _ r_-band images of 12 fields in the smc were obtained .
each field cover 8.85@[email protected]@xmath7 with a scale of 0.259@xmath8/pixel .
the fields were chosen to span a wide range of galactocentric distance , from @xmath01 to @xmath04 from the center of the cloud .
_ figure [ campo ] _ shows the spatial distribution of the fields ( hexagons ) . in photometric nights , typically six _
ubvri _ standard star areas from landolt ( 1992 ) were observed multiple times to determine the transformation of our instrumental magnitudes to the standard _
br _ system .
most of these areas include stars of a wide range of colors . given the crowding in our smc fields , we performed profile - fitting photometry of the stars in them using the daophot
ii / allstar / allframe programs , following the steps recommended by stetson ( stetson 1987 , stetson 1990 , stetson 1994 ) . daophot
ii / allstar were used to create a master star list by combining , using daomaster ( stetson 1993 ) , the star list obtained for each individual image .
then , allframe was used to perform simultaneous , consistent reductions of all ccd images of each smc field . for the final photometry , we selected stars with average @[email protected] , [email protected] and -0.3@[email protected] .
we kept a total of 215121 stars , which have been measured in ( b - r ) down to _ r_@xmath1025 .
_ figures [ cuatro1 ] _ to _ _ [ cuatro3 ] _ _ show [ _ ( b - r ) , r _ ] cmds of the smc .
a set of teramo isochrones ( pietrinferni _ et al .
_ 2004 ) have been superposed for three different metallicities appropiate for stellar populations in the smc ( z=0.001 , z=0.002 , y z=0.004 ) .
we adopt the distance modulus ( m - m)@xmath11=18.9 ( see van den bergh 1999 ) .
iras/ cobe ( schlegel _ et al .
_ 1998 ) extinction maps were used to obtain our smc fields reddening , except for the inner fields smc0057 ( at 1.09@xmath12 ) , qj0037 ( at 1.31@xmath12 ) , qj0036 ( at 1.33@xmath12 ) , and qj0111 ( at 1.35@xmath12 ) for which schlegel _ et al . _
( 1998 ) estimated a typical reddening e@xmath13=0.037 , from the median dust emission in surrounding annuli . given that their quoted value is not accurate ,
we have estimated a mean value of the reddening for each of the four smc fields mencioned above . in all cases ,
relations a@xmath3=3.315e@xmath13 , a@xmath14=1.316a@xmath3 , and a@xmath15=0.758a@xmath3 were used .
_ figure [ cuatro1 ] _ shows smc fields qj0112 , qj0111 , qj0116 , and smc0057 corresponding to the eastern side , i.e. , facing the lmc , in order of increasing position angle , at distances from 1.09@xmath12 to 1.71@xmath12 .
these diagrams show a prominent , conspicuous ms , which appears populated quite smoothly from the oldest turnoff at r@xmath022 to the 0.03 gyr isochrone .
the four cmds present a substantial number of very young stars , which are well matched by isochrones as young as 0.03 and 0.01 gyr .
this very young population may be a consequence of star formation triggered by a recent interaction with the lmc and the milky way , consistent with the yoshizawa & noguchi ( 2003 ) models . _ figure [ cuatro2 ] _ presents four cmds for fields to the western side of the smc , located up to a distance of @xmath02.9@xmath1 ; they are shown in order of increasing distance from the smc center .
it is noticeable that these western fields show a much less populated young ms as compared with the eastern ones at similar galactocentric distances . in these fields , the intermediate - age population ( 1 gyr@xmath2age@xmath210 gyr )
is the dominant one .
fields qj0036 and qj0037 had star formation up to 0.1 gyr while the others two seem to lack stars younger than 1 gyr .
four cmds of southern smc fields , located at distances ranging from @xmath02.2@xmath1 to @xmath04@xmath1 , are depicted in _
figure [ cuatro3 ] _ ; they are presented in order of increasing distance from the center and show a progresively less populated ms in their youngest parts as we go farther south . a shared characteristic in all these cmds is the absense of a populated blue horizontal branch ( hb ) , which is consistent with the fact that the position of the @xmath013 gyr isochrone is not strongly populated .
these results clearly show the existence of important stellar population gradients as a function not only of the distance to the center of the smc , but also , as a function of the position angle .
thus , a complete description of the stellar populations and evolutive history of the smc requires further analysis involving spatially significant positions .
the authors acknowledge support by the plan nacional de investigacin cientfica , desarrollo , e investigacin tecnolgica , ( aya2004 - 06343 ) .
ec and ram acknowledge support by the fondo nacional de investigacin cientfica y tecnolgica ( no . 1050718 , fondecyt ) and by the chilean centro de astrofsica fondap ( no .
15010003 ) .
dolphin , a. e. , walker , a. r. , hodge , p. w. , mateo , m. , olszewski , e. w. , schommer , r. a. , & suntzeff , n. b. 2001 , , 562 , 303 gallart , c. , stetson , p. b. , hardy , e. , pont , f. , & zinn , r. 2004 , , 614 , 109l gallart , c. , stetson , p. b. , pont , f. , hardy , e. , & zinn , r. , in preparation gardiner , l. t. , & hatzidimitriou , d. 1992 , mnras , 257 , 195 harris , j. , & zaritsky , d. 2004 , aj , 127 , 1531 madau , p. , pozzetti , l. , & dickinson , m. 1998 , , 498 , 106 olszewski , e. w. , suntzeff , n. b. , & mateo , m. 1996 , ara&a , 34 , 511 pietrinferni , a. , cassisi , s. , salaris , m. , & castelli , f. 2004 , , 612 , 168 stetson , p. b. 1987 , pasp , 99 , 191 stetson , p. b. 1990 , pasp , 102 , 932 stetson , p. b. 1993 , user s manual for daophot ii stetson , p. b. 1994 , pasp , 106 , 250 van den bergh , s. 1999 , a&ar , 9 , 273 yoshizawa , a. m. , noguchi , m. 2003 , mnras , 339 , 1135 zaritsky , d. , harris , j. , & thompson , i. b. 1997 , aj , 114 , 1002 | we present deep color - magnitud diagrams ( cmds ) reaching the oldest main - sequence turnoffs for 12 fields in the smc .
the _ b_-band and _ r_-band observations were performed using the 100-inch irne du pont telescope at las campanas observatory , chile , during four different campaigns ( 2001 - 2004 ) .
our fields cover a wide range of galactocentric distance ranging from @xmath01@xmath1 to @xmath04@xmath1 from the center of the galaxy and are located a different position angles .
photometry was carried out using daophot
ii / allstar / allframe .
teramo isochrones have been overlapped .
all our unprecedented deep ground - based cmds reach the old ms turnoffs with very good photometric accuracy .
they clearly show stellar population gradients as a function of both galactocentric distance and position angle .
the most conspicuous difference involves the young population ( age@xmath21 gyr ) : the young ms is much more populated on the eastern fields , located on the smc wing area , than on the western fields located at similar galactocentric radius .
in addition , the main stellar population gets progresively older on average as we go to larger galactocentric radius . |
many modern theories of the universe assume the existence of various types of scalar fields .
such fields could explain the recently discovered acceleration of the expansion of the universe ( see e.g. ref .
@xcite ) , or the formation of clustered systems leading to gravitational walls for galaxies and galaxy clusters .
if such fields really exist , the universe could contain many compact star - like configurations of large total mass , called scalar stars . among the many possible scalar fields , the dilaton field has a special theoretical appeal .
it couples in a unique minimal way to electromagnetism to make the maxwell action dimensionless . this coupling leads to star - like objects which are composed of scalar and electromagnetic fields . that such objects can exist
was pointed out by many authors @xcite .
we argue that such electro - dilaton stars may be responsible for the strong gamma - ray bursts observed in the universe . at present there exists no simple conventional explanation for the origin of these events reaching us isotropically from all directions of the universe . for an comprehensive discussion of the subject ,
in particular for the failures of most theories , see the article by ruffini @xcite . at the same time , there exist models for the early universe where an initial state of large volume of relativistic plasma quickly expands as a spherical wave , like a critical bubble in an overheated liquid , causing a decay of a false vacuum , and creating the universe from such a bubble @xcite .
the electro - dilaton stars can supply such an explanation .
we assume that the dark matter in the universe contains a multitude of such objects , whose total mass exceeds by far the total mass of luminous matter and is responsible for the large - scale structure of the universe .
the luminous matter concentrates in the gravitational potential wells of the scalar fields .
let us imagine that collision of relativistic particles produce a fireball of critical size .
such bubbles have been investigated as possible triggers of phase transitions in the early universe @xcite . in the context of gamma - ray bursts ,
similar assumptions have been made in ref .
the critical bubble forms the seed for transferring effectively scalar fields into pairs of elementary particles and their antiparticles .
the transfer may be initiated by fast oscillation of a field on the outer boundary of the fireball .
the process causes a relativistic detonation . in a conventional detonation
, chemical energy is converted into kinetic energy . in a relativistic detonation of an electro - dilaton star , it is the energy of the electric and scalar fields which is rapidly converted in particle - antiparticle pairs .
the resulting fireball expands with relativistic velocity and will therefore not depend on the weak gravitational fields of a newtonian configuration
. we may thus study the process within special relativity .
we begin by deducing the self - similar solutions for a spherical relativistic detonation which goes over to the well - known zeldovich solution in the limit of small velocities @xcite .
the set of equations of relativistic hydrodynamics is conveniently described in a spherical coordinate system @xmath0 .
if @xmath1 denotes the radial velocity of the plasma in three - dimensions @xcite and @xmath2 is energy density , @xmath3 - pressure , the equation of motion reads @xmath4 while energy conservation requires that @xmath5 + \frac{1}{\gamma ^{2}}\left ( \frac{\partial v}{\partial r}+v\frac{\partial v}{\partial \tau } \right ) + \frac{2v}{r}=0,\ ] ] where @xmath6 , @xmath7 , and @xmath8 . as in the nonrelativistic case ,
the motion of the plasma behind the detonation front is considered as isentropic , such that ( [ eq1 ] ) and ( [ eq2 ] ) are the only relevant equations . the pairs of relativistic particles and antiparticles created behind the wave front generate a high - temperature plasma with the equation of state : @xmath9 where @xmath10 is the sound velocity . as in an ordinary spherical detonation problem , we search for a solution depending on the self - similar variable @xmath11 in which the differential equations ( [ eq1 ] ) and ( [ eq2 ] ) reduce to ordinary differential equations , which can be combined to the equation for @xmath1 : @xmath12 = \frac{2v}{\xi } \frac{\gamma ^{2}}{1-v\xi } .\ ] ] in the nonrelativistic limit where @xmath13 , this equation reduces to a textbook equation @xcite .
the qualitative analysis of the relativistic equation ( [ eq5 ] ) is similar to the one in the textbook : the solutions for @xmath1 and @xmath14 have a diverging derivative at the wave front .
such singularity of a derivative is defined by the following : the expression in parentheses is velocity of current of plasma in relation to the wave front . according to the theory of detonation , this velocity is equal to sound velocity @xmath15 .
therefore , at approaching to the wave front the expression in brackets tends to zero .
the coordinate of the wave front @xmath16 ( where @xmath17 is a velocity of detonation wave in central frame ) and the velocity of gas behind the wave in laboratory frame @xmath18 is uniquely determined from conservation laws at the wave front .
+ fig1.tps the discussion of possible mechanisms of `` recycling '' of the field behind the wave front into relativistic plasma is considered below .
let us specify here the following estimates of the value of @xmath19 and energy density behind the detonation wave .
consider a scalar field in the simplest form , with an energy - momentum tensor @xmath20 , \,\,\,\,\,\,\,\,\,v(\varphi ) = m^{2}\varphi ^{2}/2.\ ] ] in the scalaron " regime where it undergoes fast oscillations of frequency @xmath21 , i.e. where @xmath22 , the spatial gradients of the field can be ignored and the energy density in the laboratory frame ) is @xmath23 the expression for @xmath19 and the energy density of the plasma behind the wave front are determined from the conservation laws @xmath24 and @xmath25 for an observer moving with the wave front . recall that the plasma is emitted from the wave front with the velocity @xmath10 .
hence @xmath26 since a relativistic plasma has @xmath27 , we find @xmath28 and @xmath29 . the self - similar solutions for this case are presented in fig .
[ @fig1 ] .
we now turn to the mechanism of transition of the field energy into a relativistic plasma .
so far , the description of the conversion of the energy of a scalar field into relativistic plasma at the front of detonation " wave is purely phenomenological .
the physical properties of the front of the detonation wave are completely determined by the conservation laws for a general energy - momentum tensor @xmath30 .
one specific mechanism of such a conversion was considered in @xcite based on an analogy with a laser . here
we shall consider an alternative possibility where the relativistic plasma and radiation are obtained from the energy of a dilaton field .
the star - like configurations for such field with strong and weak ( newtonian ) gravitational field have been considered before @xcite . as in section [ sec1 ]
, we shall discuss only the case of a weak gravitational field and treat the problem within special relativity .
the lagrangian density of a system of dilaton and electromagnetic fields is @xcite : @xmath31 the parameter @xmath32 can have the values @xmath33 as will be explained below .
the normalization of the fields is the same as in @xcite .
the unique interaction of the scalar field @xmath34 with an electromagnetic field required by scale invariance allows for a nontrivial combined electro - dilaton configuration . in the front of the detonation wave ,
the electromagnetic field reduces the dilaton field strength by dissipation . depending on the intensity of the electric field
, the dissipation may be due to the creation of pairs of particles and antiparticles and to a heating of the plasma .
the combined process
generation of an electromagnetic field and its subsequent dissipation supplies the energy to the front of the detonation wave .
let us study the process in a simple plane - wave configuration .
the equations for @xmath34 and @xmath35 following from lagrangian ( [ lagr ] ) are @xmath36_{;k}&=&0 , \\
\quad \phi_{;i}^{;i}&=&-\frac{\alpha \zeta}{2}e^{-2\alpha\phi}(f_{lm}f^{lm } ) .
\label{cons1b}\end{aligned}\ ] ] the system is supplemented by the missing maxwell equations ( electromagnetic versions of the bianchi identities ) : @xmath37 the total energy - momentum tensor associated with the lagrange density ( [ lagr ] ) is @xmath38.\ ] ]
we now show that there exist plane electro - dilaton waves travelling along the @xmath39-axis in which the electric field has a component @xmath40 with all other components vanishing . whereas eq .
( [ maxv ] ) is fulfilled trivially , eq .
( [ cons1 ] ) yields @xmath41 thus we find a constant of motion @xmath42 the quadratic field combination @xmath43 has the same negative value in all systems of special relativity .
for this reason we follow ref .
@xcite in choosing the parameter @xmath44 in ( [ lagr ] ) .
the other field equation ( [ cons1b ] ) becomes @xmath45 or , introducing new variables @xmath46 , @xmath47 and function @xmath48 , we have @xmath49 this is one of form of liouville equation @xcite .
its possesses a steady - state wave solution which , after the redefinition of the variables @xmath50 takes the form @xmath51 the first integral of this equation leads to the differential equation @xmath52 whose solution @xmath53 describes a soliton . though the potential of dilaton field @xmath54 diverges at @xmath55 , where @xmath56 , the derivatives @xmath57 remain finite , thus ensuring a finite energy density of the field for all @xmath58 .
note that in contrast to a charged plane in electrostatics , the electric field of this solution has a zero flux @xmath59 at infinity .
asymptotically , no electric field is detectable .
there exists a general class of solutions of the liouville equation ( [ wave1 ] ) containing two arbitrary functions @xmath60 .
setting @xmath61 , the solution of eq .
( [ wave ] ) is @xmath62}\right]\ ] ] the primes denote derivatives .
especially simple solutions are obtained for the linear functions @xmath63 where @xmath64 is velocity of soliton in eq.([redef ] ) .
the following solutions are of special interest : 1 .
@xmath65 ) localized solution for @xmath59 with a fixed asymptotic energy density @xmath66 , where @xmath59 goes over into the previous soliton solution ( [ eqn10 ] ) . in these solutions , the functions @xmath67 and @xmath68 in ( [ solut ] ) and their derivatives are regular .
singular solutions of liouville equation are also known @xcite .
they may lead to a local catastrophic growth of electric field , and require special attention .
@xmath65 ) the main purpose of this section is to show that at different values of the energy of electro - dilaton wave before and behind the electric layer the electromagnetic energy may increase in time .
a similar distributions of the energy may be created by gravitational configuration .
+ let us specify the arbitrary functions in liouville solution as @xmath69 + + + ( 500,120 ) ( 20,-300 ) + ( 360,-00)@xmath70 ( 90,21.5)@xmath70 ( 192.5,165.8)@xmath70 ( 298.9,172.5)@xmath70 ( 82.5,152)@xmath71 + + where @xmath72 are determined by the initial velocity of the wave from ( [ func ] ) and ( [ func2 ] ) , and @xmath73 is an arbitrary constant .
now the profile of the electric field looks like @xmath74 } { \cosh^2[(\gamma+\beta)x - ( \gamma - \beta)\tau+\ln \cosh \mu ( \tau - x)]}\ ] ] + the profile of such a wave is shown on fig .
let us assume that the numerator of eq .
( [ elec ] ) slowly changes in comparison with the denominator . in that case , at given @xmath75 the maximum of electric soliton concentrates about a zero point of argument of @xmath76 . as we see from eq .
( [ func ] ) , this region moves with velocity @xmath77 , so that the argument of @xmath78 is always positive in this region .
the numerator and thus the electric field increase with time .
this increase comes to an end as @xmath79 $ ] reaches its asymptotical value 1 .
this is illustrated in fig .
[ fig3 ] . if instead of @xmath79 $ ] in ( [ func1 ] ) a function
is chosen which grows without bounds , then also the electric field will keep growing . + the growth of the energy of the electric field will be consumed by dissipation .
its influence on the electro - dilaton wave will be considered in section [ @diss ] .
let us use the above results to study a wave with a transverse magnetic wave . in this case
the maxwell equations ( [ maxv ] ) are not fulfilled identically but yield a relation between transverse components @xmath35 .
let us select for consideration the following nonzero components @xmath80 .
then the complete set of the equations is @xmath81 a travelling wave has the field components @xmath82 since @xmath77 and @xmath83 , the field is transverse magnetic .
it dictates in ( [ cons1 ] ) the choice of the sign @xmath84 .
further , using similar variables as in ( [ redef ] ) @xmath85 we obtain the solutions for dilaton and electromagnetic fields similar to ( [ eqn10 ] ) @xmath86 the principal difference with respect to ( [ eqn10 ] ) is the absence of a relativistic factor @xmath87 in the argument @xmath88 .
the transverse magnetic wave is focused in a band of width @xmath89 . outside of this
, the energy density lies mainly in the gradient part of the dilaton field . from eq .
( [ eqn14 ] ) it follows that energy flux density is independent of @xmath90 : @xmath91 this can be interpreted as follows : the conversion of the energy of the dilaton field from the gradient to the potential part implies a conversion of the gradient energy of the dilaton field to the energy of electromagnetic field .
certainly , this conversion is completely reversible .
the above electro - dilaton soliton appears at the front of the detonation wave .
if we use the plasma behind the front as a frame of reference , then this soliton will be move with a velocity @xmath92 in the positive @xmath39-direction .
such a movement generates heat in the plasma and creates of pairs of particles and antiparticles in high concentration .
this reduces the @xmath93 and @xmath94 fields . in the simplest description
, the decay wave can be taken into account in a travelling - wave approximation by replacing eqs .
( [ eqn11 ] ) by @xmath95,\ ] ] where @xmath96 is the friction constant with the dimension of a reverse length . while moving though the plasma with @xmath97 , the electromagnetic wave decays .
the equation for the dilaton field with @xmath84 becomes @xmath98 substitution here ( [ eqn16 ] ) and using the redefinitions ( [ eqn13 ] ) , we obtain @xmath99 after a change of the variable @xmath100 , this equation reduces to the liouville form ( [ eqn9 ] ) .
then subject to dissipation we have the following solution : @xmath101 ; \quad h_d(\eta)=h_0 e^{-\beta \eta}/\cosh^2(\eta/2).\ ] ] here , the subscript @xmath102 denotes a solution with dissipation .
thus electro - dilaton solitons with dissipation become asymmetric with a steeper front part .
the creation of particles and heat of plasma happens at decreasing of the energy density of dilaton field . for @xmath55
the energy of the system is concentrated only in the gradient part of dilaton field ( see ( [ ener]),([redef ] ) ) @xmath103 in the limit @xmath55 , the equations ( [ eqn18 ] ) becomes @xmath104 it is clear from here that the energy density of dilaton field before the wave is more than the energy density behind the wave ( @xmath105 ) .
the limiting value for @xmath106 in this example is @xmath107 .
this implies that the whole energy of the dilaton field transforms to the heat energy of plasma .
consider now the dissipation of energy by the induced fields at the wave front .
for this purpose we substitute a current on the right hand side of eq .
( [ eqn11 ] ) @xmath108 where @xmath109 is specific conductivity of the medium .
then we find the following system of dimensionless equations : @xmath110 with the parameter @xmath111 . at @xmath112 ,
this reduces to the previous electro - dilaton soliton ( [ eqn14 ] ) . to analyze eqs .
( [ eqn19 ] ) it is convenient to introduce the new variable @xmath113 which is unity in the absence of dissipation .
the first integral of the differential equations ( [ eqn19 ] ) leads to the solution : @xmath114 the integration constant is selected so that in the limit @xmath115 the solution tends to the dissipation - free soliton with @xmath116 for @xmath117 . in the opposite limit @xmath118
, the variable @xmath119 tends to zero and @xmath120 thus for @xmath121 , the solution represents a kink moving in the positive @xmath39-direction . by virtue of @xmath105 ,
the asymptotic value of the dilaton energy density before the front is less than behind it .
physically this means that the pressure of the dilaton field creates a plasma at the wave front pushing it ahead .
it follows from eq .
( [ eqn21 ] ) that the limiting value of @xmath106 is now @xmath122 implying that the electro - dilaton energy goes completely over into an energy flux of moving plasma .
the solutions are plotted in fig .
[ @fig2 ] .
b0.tps + [ 4 mm ]
> from our discussion it appears perfectly plausible that dark matter consists of electro - dilaton stars and is not dark at all , but has been showing its presence quite dramatically all along via the powerful gamma - ray bursts observed since .
vf and vg than the daad in bonn / germany for financial support .
we also thank dr .
a. pelster for discussions .
99 v. sahni and a. starobinsky , gr - qc/9904398 , int .
d * 9 * ( 2000 ) 373 ; a review for the older results is : s. gottlber , v. mller , h .- j .
schmidt , a. a. starobinsky , _ models of chaotic inflation _ ,
d * 1 * ( 1992 ) 257 - 279 .
+ see also the related discussion of _ quintessence _ in r.r .
caldwell , r. dave , p. j. steinhard , _ cosmological imprint of an energy component with general equation of state _ ,
lett . , 1582 ( 1998 ) ( astro - ph/9708069 ) ; + for experimental data see + p.m. garnavich et al . , _ supernova limits on the cosmic equation of state _ , astrophys . j. * 509 * , 74 ( 1998 ) ( astro - ph/9806396 ) ; + b. ratra , p.j.e .
peebles , _ cosmological consequences of a rolling homogeneous scalar field _ , phys .
d * 37 * , 3406 ( 1988 ) ; + c. kolda , w. lahneman , _ exponential quintessence and the end of acceleration _ , hep - ph/0105300 .
m. gasperini , _ dilatonic interpretation of the quintessence ?
_ , gr - qc/0105082 ; + j. m. cline , `` quintessence , cosmological horizons , and self - tuning '' , hep - ph/0105251 ; + ph .
brax , a. c. davis , _ cosmological evolution on self - tuned branes and the cosmological constant _ , jhep * 0105 * , 007 ( 2001 ) ( hep - th/0104023 ) ; + a. riazuclo and j.p .
uzan , astro - ph/0107386 ; + bin chen and feng - li lin , _ holographic cosmic quintessence on dilatonic brane world _ , hep - ph/0106054 ; m. gasperini , f. piazza , and g. veneziano , _ quintessence as a run - away dilaton _ , gr - qc/0108016 .
r. ruffini and s. bonazolla , _ phys .
rev . _ * 187 * , 1767 , ( 1969 ) .
t. piran , _ phys .
_ * d36 * , 3575 , ( 1987 ) .
e. mielke and f. schunck , gr - qc/9801063 ; _ gen .
grav . _ * 33 * , 805 , ( 2001 ) .
ruffini , _ analogies , new paradigms , and observational data as growing factors of relativistic astrophysics _ , in _ fluctuating paths and fields - dedicated to hagen kleinert on the occasion of his 60th birthday _ , eds .
w. janke , a. pelster , h .- j .
schmidt , and m. bachmann ( world scientific , singapore , 2001 ) , p. 771 .
+ see also + r. ruffini , c.l .
bianco , p. chardonnet , f. fraschetti , s .- s .
xue , _ on the interpretation of the burst structure of grbs _ , astrophys .
j. 555 ( 2001 ) , l107 , l117 ; nuovo cimento b , 116 , 99 ; + according to these works , gamma - ray bursts are correlated with supernovas .
the discussion in the present paper attempts to give an alternative interpretation .
+ g. mao , s. chiba , w. greiner , k. oyamatsu , _ vacuum discharge as a possible source of gamma - ray bursts _ , int
e * 9 * , 185 ( 2000 ) ( astro - ph/9912303 ) .
s. coleman and f. de luccia , _ phys .
rev . _ * d21 * , 3305 , ( 1980 ) .
k. postnov , _ physics uspekhi _ * 42 * , 469 , ( 1999 ) .
l. landau and e. lifshitz , _ hydrodynamics _
( nauka , moscow , 1988 ) , p. 680 .
( in russian ) f. baum , s. kaplan and k. staniukovich , _ introduction in cosmic gas dynamics _ ( moscow , 1958 ) .
a. linde , _ elementary particle physics and inflationary cosmology _
( nauka , moscow , 1990 ) .
a. migdal and v. krainov , _ approximate methods of quantum mechanics _ ( nauka , moscow , 1966 ) .
i. tkachev , _ phys .
lett . _ * b191 * , 41 , ( 1987 ) .
zhi - jian tao and xun xue , _ phys .
* d45 * , 1878 , ( 1992 ) .
r. gregory and j. harvey , _ phys .
* d47 * , 2411 , ( 1993 ) .
t. matos , d. nunez and h. quevedo _ phys .
rev . _ * d51 * , r310 , ( 1995 ) . s. yazadjiev , p. fiziev , t. boyadjiev and m. todorov , hep - th/0105165 .
j. liouville , j. math .
pures et appl . *
18 * , 71 , ( 1853 ) .
g. jorjadze , a. pogrebkov and m. polivanov , ictp prep .
( trieste , 1978 ) . | we speculate that the universe is filled with stars composed of electromagnetic and dilaton fields which are the sources of the powerful gamma - ray bursts impinging upon us from all directions of the universe .
we calculate soliton - like solutions of these fields and show that their energy can be converted into a relativistic plasma in an explosive way . as in classical detonation theory
the conversion proceeds by a relativistic self - similar solution for a spherical detonation wave which extracts the energy from the scalar field via a plasma in the wave front in the atmosphere of the star .
= -10.0 mm = 0.0 mm = 0.0 mm # 1 # 1 * |
in the paradigm of cosmological structure formation , gravitationally bound halos form at the peaks of the primordial density field as dark matter and baryons undergo nonlinear gravitational collapse .
dark matter , being collisionless and dissipationless , conserves its orbital energy , remaining in an extended dispersion - supported profile with overlapping inward- and outward - moving orbits @xcite .
in contrast , gas collides , shocks , mixes , and eventually dissipates energy via radiative cooling , causing it to collapse to the minimum of a halo s potential well and seed the formation of stars and galaxies @xcite . within this paradigm , debate persists about the most physically meaningful ways to describe the physical extent of a halo , the rate of cosmic accretion into a halo , and the amount of mass growth within a halo , including how these compare for dark matter versus baryons .
these are important questions because measurements of cosmic accretion and mass growth depend sensitively on how and where one measures them .
thus , understanding the evolution of halos requires a detailed understanding of the relevant physical scales across cosmic time , including the physical meaning ( if any ) of a choice for a halo s virial boundary / edge .
furthermore , because cosmic accretion into a halo feeds the growth of the galaxy inside , understanding the physics of all of these scales is necessary for developing a physical picture of galaxy evolution in a cosmological context .
many works have examined the nature of dark - matter accretion into halos .
most previous works measured cosmic accretion or mass growth according to some evolving virial radius , @xmath8 , that was linked to the background density of the universe ( for example , * ? ? ?
* ; * ? ? ?
however , recent works have questioned the physical meaning of the commonly used halo virial radius and its implications for inferred cosmological accretion and mass growth @xcite .
specifically , these works found that low - mass halos experience little - to - no significant physical growth of dark matter at fixed physical radii , especially at late cosmic time ( @xmath9 ) .
moreover , the recent work by @xcite found that most of the growth of dark - matter mass at halo masses @xmath10 arises because one measures halo mass within some virial radius that is tied to a reference background density that evolves with time , which in turn can lead to inferred growth of halo mass even if the physical density profile of the halo remains constant , an effect that they called `` pseudo - evolution '' .
this suggests that there is little - to - no _ physical _ accretion into lower - mass halos at late cosmic time , such that they effectively evolve as `` island '' halos , divorced from the cosmic background , at @xmath11 .
pseudo - evolution has a number of important implications for understanding phenomenological links between galaxies and dark - matter halos as well as for models of galaxy evolution .
for example , observations are probing the relation between the stellar mass of a galaxy and the virial mass of its host halo , including its evolution with time ( for example , * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ) , finding that the relation between galaxy mass and halo mass evolves only weakly at @xmath12 ( and possibly at higher @xmath13 ) , which suggests that galaxy mass evolves largely in sync with halo mass .
similarly , @xcite found a tight linear relation between the size of a galaxy and the size of its host halo across a wide range of masses at @xmath14 , despite the significantly varying ratio of stellar - to - halo mass across this range .
these studies suggest that the mass and size of a galaxy is set by , or at least responds to , that of its host halo , but these relations are meaningful only insofar as one uses a physically sensible radius ( and thus mass ) for a halo .
moreover , many ( semi - analytic ) models of galaxy evolution try to link the accretion rate of a halo in simulations to its baryonic accretion rate and in turn to the star formation rate of the galaxy ( for example , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
halo mass growth that is incorrectly attributed from pseudo - evolution would affect all of these analyses .
however , it remains unclear what role pseudo - evolution plays in the cosmic accretion of baryons , because gas dynamics can be markedly different : gas is collisional , so it can shock and mix , and it also can dissipate energy via radiative cooling .
a number of works have examined the accretion rates of gas into galaxies ( for example , * ? ? ?
* ; * ? ? ?
* ) and halos ( for example , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , though in almost all cases they measured mass growth / flux at some predefined and evolving virial radius .
few works have compared in detail the specific accretion rates of gas versus dark matter : @xcite and @xcite both found that specific accretion rates near the virial radius were broadly similar for baryons and dark matter , with some ( up to a factor of 2 ) reduction in the specific rate for baryons , depending on the model for stellar winds ; but they both found significantly reduced baryon accretion rates at small radii near the galaxy compared to that near the virial radius .
however , all of these works used differing techniques to measure accretion and at somewhat different choices for virial radii . furthermore , many of these works focused on the role of feedback from stars and/or black holes on baryonic accretion rates , using various phenomenological models for driving stellar winds .
while there is general consensus that stellar winds can alter accretion rates into the galaxies , results are mixed regarding the regulation of gas accretion at larger radii within a halo .
however , all of these works found _ some _ level of gas accretion into galaxies at late cosmic time , implying that the pseudo - evolution of dark matter may not extend to gas .
indeed , @xcite , examined baryonic accretion rates in a suite of cosmological zoom - in simulations at @xmath15 and found that the mass inflow rate at @xmath16 is broadly similar to ( @xmath17 ) that at @xmath8 ; they found similar trends examining the rate at fixed physical radii of 10 and @xmath18 , implying that baryon accretion to small radii is indeed physical .
more generally , the physical nature of cosmic accretion has a variety of implications for the evolution of gas in halos and galaxies at late cosmic time , especially for low - mass galaxies .
for example , galaxies at @xmath19 formed @xmath20 of their mass since @xmath21 @xcite , and the rate of decline of the cosmic density of star formation broadly mimics the decline of dark - matter accretion rates into halos at fixed mass ( for example , * ? ? ?
* ; * ? ? ?
. however , it is not clear how much star formation and galaxy growth at @xmath12 is linked to cosmic accretion , as opposed to the consumption of gas that already is within galaxies , given that ( molecular ) gas fractions are observed to decrease over time ( for example * ? ? ?
* ) , or the recycling of gas in galaxies from stellar winds ( for example , * ? ? ?
* ; * ? ? ?
additionally , cosmic gas accretion into halos drives the evolution of extended gas around galaxies , referred to as the circumgalactic medium , as many observations and surveys now are probing ( for example , * ? ? ?
* ; * ? ? ?
* ) . for such studies
it is important to understand both the most physically meaningful virial definition to use for a halo as well as the rate of accretion of ( relatively unenriched ) gas and how it propagates to smaller radii .
the primary goal of this work is to understand the physical nature of cosmic accretion into halos and how it propagates down to scales of the galaxy inside , for both baryons and dark matter . in particular
, we aim to bridge the gap between detailed studies of halo growth , typically based on dark - matter - only simulations , and detailed studies of cosmic gas accretion into galaxies .
more specifically , we seek to understand the significance of pseudo - evolution not only for dark - matter accretion , but also its role in baryon accretion and hence galaxy growth .
we focus on halos of mass @xmath22 at late cosmic time ( @xmath23 ) .
such halos are observed to host galaxies with @xmath24 ( @xmath25 in our simulation with star formation is @xmath26 higher because of overcooling ; see ) .
these mass and redshift regimes are where the effects of pseudo - evolution are particularly strong @xcite , where observations constrain well the relation between galaxies and their host halos as well as gas in / around galaxies .
we use simulations with varying treatments of gas physics , some including star formation and thermal feedback , though our simulations only marginally resolve the scales within galaxies , and our prescription for stellar feedback does not drive particularly strong winds out of galaxies , which play a strong role in regulating accretion into the galaxy itself .
thus , we focus on cosmic accretion and mass growth on scales within a halo , but we do not investigate accretion into the galaxy , or stellar mass growth , directly .
we defer such work to a follow - up analysis . throughout , we cite all masses using @xmath27 for the dimensionless hubble parameter .
we first review the basic theoretical framework for _ spherical _ collapse and virialization of a halo , to set the stage for and aid in the interpretation of our numerical results . in the standard model @xcite ,
if a spherical region is sufficiently overdense , its gravitational self - attraction overcomes the initial cosmological expansion , such that a mass shell will reach a maximum radius and then collapse .
specifically , for flat @xmath28cdm cosmology , the radial acceleration around some overdense region is @xmath29 in which @xmath30 is the physical radius from the center of the overdensity , @xmath31 is the enclosed mass , @xmath32 is the gravitational constant , and @xmath33 is the ( constant ) physical density of dark energy . around the overdense region , this acceleration counteracts the initial outward - moving velocity that is set by the hubble expansion , and if a shell experiences turn - around , it decouples from the hubble flow , at which point it no longer `` feels '' the expansion of the universe , modulo the acceleration from dark energy in .
we refer to the radius of this first turn - around as @xmath34 . in einstein - de sitter cosmology (
@xmath35 , @xmath36 ) , turn - around occurs when the average density within the sphere is @xmath37 that of the background . assuming the virial theorem for the equilibrium state of the halo , such that @xmath38 , for which @xmath39 and @xmath40 are the kinetic and potential energies , its final @xmath30 ( density ) will be @xmath41 that at @xmath34 . over this period of collapse
, the background universe has expanded to become @xmath42 less dense , so a `` virialized '' halo has an average density of @xmath43 that of background , a value that many authors round to 200 .
( see @xcite for a generalization of this model to @xmath44 . ) more generally , one can define a halo s virial radius , @xmath45 , such that the average interior density is @xmath46 times some reference density , @xmath47 : @xmath48 . in this work ,
we use @xmath49 m , that is , we define halos as containing @xmath50 the average matter density of the universe .
( we will compare other virial definitions in the context of the results of this paper in future work . ) while the above model describes halo collapse in terms of instantaneous energetics , the actual physics of collapse is more complicated , given that halos experience ongoing accretion , so they almost never are well - relaxed virialized systems , especially at @xmath51 .
first we consider dark matter . after reaching @xmath34 , a shell continues to collapse until reaching its first pericenter , after which it orbits back out to its apocenter , or secondary turn - around radius , @xmath52 .
if there were no change in @xmath53 throughout this full orbit , energy conservation implies that @xmath54 . in reality
, the continued accretion of mass shells from larger @xmath30 causes the potential to deepen , so @xmath55 : the higher the accretion rate , the more that @xmath55 @xcite . throughout this orbit
, a shell passes through other shells that collapsed at different times , because dark matter is collisionless .
thus , a halo represents a superposition of shells of inward- and outward - moving orbits that have collapsed at different epochs . at any time
, there is an outermost @xmath52 , which corresponds to the shell that is reaching its @xmath52 for the first time .
we refer to this as the splashback radius , @xmath56 , and it corresponds to the maximum @xmath30 of matter that has passed through the core of the halo .
thus , a halo s density profile declines rapidly beyond @xmath56 , and typically @xmath57 ; it is smaller for halos with higher rates of accretion @xcite .
the physics of gas accretion is different , because gas is a collisional fluid , so there are no shell crossings of orbits , and there is no splashback radius . thus , after reaching @xmath34 , gas collapses ( typically supersonically ) until it encounters a previously collapsed gas shell , at which point it shocks and heats @xcite . thus , while gas has no splashback radius , it can have a virial - shock radius , although this can occur at @xmath30 much smaller ( or larger ) than @xmath5 , including near the galaxy itself @xcite . if shocked , gas is heated to near the halo s virial temperature , at which point it is supported by thermal pressure .
if the timescale for gas cooling is longer than the dynamical time at the given radius , the gas will remain in near hydrostatic equilibrium , which typically is true at @xmath58 @xcite . in lower - mass halos , even if gas is shock - heated to near the virial temperature @xcite , it will cool rapidly and advect to smaller @xmath30 , while remaining relatively cold @xcite .
we emphasize that the above picture is valid for halo collapse that is purely spherical and smooth . in reality , cosmological collapse is triaxial and clumpy . for dark matter
, this means that @xmath56 is smeared out ( for example , * ? ? ?
* ) . for gas , triaxial collapse can drive turbulence with effective pressure support , and gas can mix via kelvin - helmholtz and rayleigh - taylor instabilities . for both components ,
angular - momentum support will regulate radial advection , and infalling satellite halos can persist as bound subhalos that behave neither purely collisional nor collisionless .
thus , while the above picture is informative , it is not necessarily true in detail , and one needs to use cosmological simulations to model these processes fully , as we now describe .
to study the accretion of dark matter and gas in a realistic cosmological context , we performed and analyzed a suite of cosmological simulations that we call _ `` omega25 '' _ using the adaptive refinement tree ( art ) eulerian @xmath59-body plus hydrodynamics code @xcite and the _ omega _ high - performance computing cluster at yale university .
these simulations include collisionless dynamics of dark matter and stars as well as gas dynamics in a cubical volume of comoving size @xmath60 in a flat @xmath28cdm cosmology : @xmath61 , @xmath62 , @xmath27 , @xmath63 , and @xmath64 .
art uses adaptive refinement in space and time to reach the high dynamic range required to resolve galaxies and their halos in cosmological simulations .
we use simulations at two different resolutions : @xmath65 and @xmath66 root mesh cells / dark - matter particles .
both runs use a maximum of 8 levels of adaptive refinement in the mesh . the dark matter particle masses are @xmath67 , and the minimum mesh cell sizes are @xmath68 comoving , respectively .
we generate initial conditions at @xmath69 , and the number of time steps in the root grid to @xmath70 is @xmath71 and @xmath72 for the @xmath65 and @xmath66 runs , respectively .
each level of adaptive refinement is a factor of 1 - 4 ( typically 2 ) higher in time resolution .
we save 60 snapshots spaced evenly in @xmath73 from @xmath74 to 0 , leading to a snapshot time resolution of @xmath75 .
we present results only from the higher - resolution @xmath66 simulations , having used the lower - resolution simulations to ensure that our results do not depend significantly on resolution within the ranges of halo mass ( @xmath76 ) and radius ( @xmath77 ) that we examine . in order to assess the effects of gas cooling and star formation on the dynamics of baryon accretion
, we conducted each simulation with four different prescriptions for the inclusion of gas dynamics , star formation , and feedback , as follows : 1 . including only dark matter , 2 .
additionally including gas without radiative cooling , 3 . additionally turning on radiative cooling for primordial gas in the presence of a cosmic ultraviolet background ( without star formation ) , 4 . additionally including star formation , thermal feedback , metal enrichment , and metal - line cooling , as detailed below .
we use equilibrium gas cooling and heating rates that incorporate compton heating and cooling , heating from a cosmic ultraviolet ionizing background @xcite , and atomic cooling including metals .
we use cloudy @xcite to tabulate these for the temperature range of @xmath78 and a grid of metallicities and ultraviolet intensities .
our model for star formation and feedback is an extension of that of previous works using art @xcite , based on an empirically motivated efficiency and dependence on gas density : @xmath79 with @xmath80 .
we allow star particles of minimum mass @xmath81 to form in cells with number density @xmath82 and temperature @xmath83 .
such temperature and density thresholds are reasonable for simulations at our moderate resolution @xcite .
we model each newly formed star particle as a single stellar population with an initial mass function from @xcite in the range of @xmath84 .
all stars with @xmath85 immediately deposit @xmath86 of thermal energy into the gas of their host cell , accounting for energy input by stellar winds and type ii ( core - collapse ) supernovae .
moreover , they deposit min(0.2 , @xmath87 ) of their mass as metals into their host cell . in addition
, we account for delayed supernova ia , assuming that a fraction of 0.015 of the stellar mass at @xmath88 explodes over the entire history of the population , with each event dumping @xmath86 of thermal energy and ejecting @xmath89 of metals into the gas of the host cell .
note that we do not add explicit momentum flux from supernovae or stellar winds .
finally , we model mass loss from stellar winds for each star particle assuming that the cumulative fraction of mass lost at time @xmath90 since its birth is @xmath91 $ ] . at each time step
, this mass loss is added to the gas mass of its host cell along with its energy and momentum . with these parameters ,
about 40% of initial stellar mass is lost over a hubble time ( see * ? ? ? * for more details ) our goal in this paper is to explore systematically the effects of gas and stellar physics on cosmological accretion rates .
our prescription for stellar feedback does not generate strong outflows that are encompassed in more detailed simulations of galaxy evolution ( for example , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . as a result ,
our galaxies experience overcooling and produce @xmath92 times too many stars compared to observational constraints at our halo masses .
however , the physics of these outflows and the coupling to the surrounding halo gas remain areas of active investigation and debate .
instead , our goal in this work is to understand the importance ( or lack thereof ) of cosmic accretion for dark matter and baryons absent strong outflows . in a follow - up analysis
, we will use higher - resolution zoom - in simulations to explore the role of more detailed outflows in regulating gas accretion .
we identify halos using a variant of the method in @xcite .
we first find peaks in the dark - matter distribution by measuring the local density of each particle using a smoothing kernel based on its 256 nearest neighboring particles . starting with the densest peak in the simulation ,
we grow a sphere around it until the average density , including all matter components ( dark matter , gas , and stars ) , interior to the radius @xmath5 is @xmath93 , the average matter density of the universe . excluding all other density peaks within @xmath5 , we repeat the procedure for the next densest dark matter particle until we have identified all isolated centers . for each halo , we identify its main progenitor at the previous simulation snapshot .
first , at all snapshots , we identify 30% of the most bound dark - matter particles in each halo .
then , we link halos across adjacent snapshots that share these particles .
we identify the main progenitor as the halo at the previous snapshot that shares the most number of particles , and we follow back this progenitor link back across each snapshot to identify the main progenitor at a given @xmath13 . for all analyses , we first select halos at @xmath70 in bins of @xmath94 .
we focus on @xmath95 and @xmath96 ; each simulation contains @xmath97 and 100 such halos , respectively .
each simulation also contains 12 group - mass halos with @xmath98 .
while we show results at this mass for our dark - matter simulation , we choose not to examine this mass range for our simulations that include hydrodynamics , given that the feedback from supermassive black holes , which we do not model , likely plays an increasingly important role in gas thermodynamics .
all of our halos contain at least 1200 dark - matter particles within @xmath5 at @xmath70 , and we show profiles down to a radius of @xmath99 , which corresponds to 14 mesh cells at the highest refinement .
we focus on the formation histories of isolated halos by selecting only halos at @xmath70 whose center lies @xmath100 from the center of any more massive halo ( for which @xmath5 is that of the neighboring halo ) .
thus , we seek to exclude strong environmental effects from neighboring halos , including tidal stripping , which typically starts at @xmath101 beyond a more massive halo @xcite , as well as contamination from halos that used to be satellite ( sub)halos but orbited beyond @xmath5 after infall and are highly stripped @xcite . in summary
, we examine how isolated halos at @xmath70 have evolved since @xmath102 , when the effects of pseudo - evolution in dark matter are strongest @xcite .
+ + we first examine the physical nature of cosmic accretion in the simulation with only dark matter , to set the stage to compare later with hydrodynamic simulations .
we select isolated halos in three bins of @xmath94 at @xmath70 and follow their progenitors back to @xmath102 . for each quantity , we compute the average value in bins of physical radius for each halo , and we show the median value across the sample of halos at each @xmath13 . we also show the 68% scatter for each quantity at / since @xmath21 .
( top ) shows profiles of the physical density of dark matter , @xmath103 .
upper sub - panels show @xmath104 versus physical radius , @xmath30 , while lower sub - panels show the ratio @xmath105 at fixed @xmath30 , that is , physical growth . for the latter ,
we compute this ratio for each halo and its progenitor , and panels show the median of this ratio across the sample . at large @xmath30 , @xmath106 around halos declines over time , following the declining average matter density of the expanding universe . within this expanding background
, gravitational attraction causes density to increase at smaller @xmath30 , according to .
however , the physical growth of @xmath106 at any fixed @xmath30 is modest , being @xmath107 since @xmath102 and only 10% since @xmath21 for @xmath108 .
higher - mass halos show slightly stronger physical growth .
moreover , as the 68% scatter in growth since @xmath21 ( shaded region ) demonstrates , a significant fraction of halos experienced _ no _ physical growth of dark matter over the last @xmath109 .
( bottom ) shows the corresponding profiles of the average radial velocity of dark matter , @xmath110 . at large @xmath30
, @xmath110 is positive ( net outflowing ) and increases with @xmath30 , following the hubble flow of the expanding universe .
the gravitational acceleration ( ) causes @xmath110 to decline over time at fixed @xmath30 .
the outermost @xmath30 where @xmath111 is where matter is experiencing its first turn - around at the given @xmath13 . moving to a slightly smaller @xmath30 , @xmath112 , that is , the majority of the mass is infalling .
however , @xmath110 , which is an average over all masses at a given @xmath30 , reaches a minimum value and turns up at a smaller @xmath30 .
we refer to this radius of minimum @xmath113 as @xmath114 .
physically , @xmath114 represents the characteristic radius where the _ average _ infall velocity is maximal , and it occurs because @xmath110 is an average of a juxtaposition of inward- and outward - moving orbits that overlap because dark matter is collisionless .
( though a given mass shell will continue to collapse to smaller @xmath30 at increasingly negative @xmath115 . ) thus , at @xmath116 , most mass is infalling for the first time , while at @xmath117 , most mass already has passed through the halo .
in other words , @xmath118 , corresponding to the outermost shell that is experiencing secondary turn - around at the given @xmath13 after passing through the halo core . for a given shell
is smeared over an extended range of @xmath30 @xcite .
therefore , the way that one measures @xmath56 can influence its value .
for example , using @xmath30 where the slope of the density profile is steepest , as those works did , can yield a @xmath56 that is smaller than @xmath114 .
nevertheless , @xmath114 does represent the @xmath30 where the splashback mass starts to dominate over the infalling mass .
] at smaller @xmath30 , @xmath119 , where there is equal mass in inward- and outward - moving orbits and little change in mass over time .
@xmath114 increases monotonically with time , while the minimum of @xmath110 tends to weaken over time .
these trends are linked , as governed by : infalling matter experiences a weaker halo potential at a larger @xmath30 ( for these relatively static dark - matter halos ) , because dark energy causes increasingly stronger positive acceleration at larger @xmath30 over time .
this fact , of decreasing inflow velocity , combined with the decreasing density of inflowing mass at later times , means that the cosmic accretion rate , and thus the growth of @xmath106 in the halo , declines over time , as ( top ) shows . at any @xmath13
, the most significant growth of @xmath106 occurs at @xmath120 .
however , ( top ) shows _ some _ growth of @xmath106 at smaller @xmath30 , even where @xmath121 , because ( collisionless and dissipationless ) mass inflowing from a larger @xmath30 does deposit itself , in a time - average sense , at smaller @xmath30 . at @xmath122 , halos in our mass range
typically have a clear infall region , where @xmath112 , but at @xmath14 , this infall region has disappeared for halos with @xmath123 , meaning that they experience _ no _ net physical growth of dark matter from cosmological accretion . halos with @xmath124 , on the other hand , experience weak physical accretion of dark matter even at @xmath70 . in
, vertical arrows show the median @xmath125 of halos and their progenitors at each @xmath13 . across all masses and @xmath13
, @xmath125 approximates the inner edge of the infall region , where @xmath121 , reasonably well .
similarly , @xmath125 approximates the @xmath30 beyond which the growth of @xmath106 is most significant .
we will explore in more detail such correlations with @xmath125 in .
solving for an navarro - frenk - white @xcite mass profile implies that the acceleration from dark energy equals the gravitational acceleration from the halo at @xmath126 at @xmath70 , independent of mass ( see also * ? ? ?
thus , dark energy largely accounts for the upturn in @xmath110 at large @xmath30 for the most massive halos in .
however , dark energy alone can not account for the upturn in @xmath110 just beyond @xmath5 in lower - mass halos at low @xmath13 .
the radial extent of the infall region for low - mass halos at low @xmath13 is in fact set by the ( 3-dimensional ) tidal motions and angular momenta around such halos ( see , also @xcite ) . another way to see
the nature of cosmic accretion is to examine the physical growth of cumulative mass , @xmath127 , within fixed @xmath30 .
this also allows us to compare with the commonly used mass growth that one infers from an evolving @xmath128 .
shows the median ratio @xmath129 at various @xmath30 for the same halos as in . at large @xmath30 ,
@xmath130 declines over time , in accord with the declining average density of matter in the expanding universe . at smaller @xmath30 , cumulative mass
instead grows over time , though the mass growth eventually stalls , which occurs at larger @xmath30 over time .
thus , the amount of physical growth within most @xmath30 is modest since @xmath102 , especially as compared with the mass growth inferred from @xmath131 ( thick black curve ) . indeed , as inferred from @xmath131 , our lowest - mass halos have doubled in dark - matter mass since @xmath21 , but the amount of physical growth at any @xmath30 is significantly lower ( @xmath132 , increasing with mass ) .
furthermore , the shaded region indicates the scatter in physical growth at @xmath133 , highlighting that a significant fraction of isolated halos experienced _ no _ physical growth since @xmath134 .
we find similar scatter for the growth of @xmath131 or in using a narrower bin of @xmath135 of 0.1 dex , so the scatter at @xmath133 is real and meaningful .
these trends are consistent with previous related works based on dark matter simulations @xcite .
+ having explored the physical nature of cosmic accretion in the simulation with only dark matter , we now turn to examining both dark matter and baryons in simulations that model both . in this section ,
we first explore systematically how gas physics alone affects cosmic accretion , to obtain a basic understanding of the underlying physics .
thus , we first examine two set of simulations with no star formation that are identical , except that the first includes non - radiative gas , while the second employs radiative cooling in a ultraviolet background . in the next section , we will show the simulation that additionally includes star formation and feedback . here , we report results for a single bin of @xmath136 and note that trends with gas physics are qualitatively similar across our range of @xmath137 .
shows the profiles of gas ( solid ) and dark matter ( dashed ) as a function of physical radius , @xmath30 , similar to .
top row shows @xmath138 and its growth history ; middle row shows the baryon fraction , @xmath139 , in units of the cosmic value , @xmath140 ; and bottom row shows the average radial velocity , @xmath113 .
first , ( left column ) shows gas without radiative cooling . while this is an unphysical scenario , especially for low - mass halos with virial temperatures that correspond to short cooling times compared to their dynamical times , it provides a benchmark for understanding the subsequent impact of gas cooling .
the inclusion of non - radiative gas imparts little change to the behavior of dark matter , which remains similar to that in the simulation with only dark matter ( ) . for gas accretion , shows that at large @xmath30 , @xmath141 and @xmath142 closely track those of dark matter , as may be expected for freely infalling , supersonic gas whose dynamics are governed primarily by gravity
. shows some decoupling of @xmath113 between non - radiative gas and dark matter just beyond @xmath5 , where @xmath142 is slightly weaker than @xmath110 .
such decoupling is not surprising , because the underlying dynamics of dark matter and non - radiative gas are distinct , as outlined in : @xmath114 for ( collisionless ) dark matter is set by @xmath56 , while for non - radiative gas , whose collisional nature means that orbits can not overlap , it is set by where gas shocks and/or starts to become supported by both thermal and turbulent pressure .
( though accretion can be clumpy , and some gas can orbit within a satellite subhalo in a neither purely collisional nor collisionless manner . ) thus , the addition of this pressure support is likely what causes weaker @xmath113 for gas as compared with dark matter . at similar @xmath30
, @xmath143 shows a corresponding , though weak , enhancement . at @xmath144 ,
non - radiative gas decouples from dark matter , as the gas density profile flattens in the core and @xmath143 decreases , a result of strong thermal and turbulent pressure support . here , @xmath141 decreases over time , because the higher rate of accretion at higher @xmath13 causes the non - radiative gas to be slightly overpressurized in the core .
this gas then expands to a somewhat lower density .
given the strongly _ differing _ physics that govern dark matter and non - radiative gas , we emphasize the strikingly _ similar _ behavior that they display at essentially _ all _ @xmath30 .
this simply may be a reflection that the dynamics of both components , though different , conserve energy and respond to the same underlying halo potential .
the similar behavior of non - radiative gas and dark matter means that essentially all subsequent differential effects between them are driven not by the collisional nature of gas , but rather , by its ability to cool radiatively .
( right column ) shows the same profiles , but from the simulation with radiative cooling .
this represents the opposite extreme from non - radiative , because gas is able to experience runaway cooling and collapse to small @xmath30 without any feedback .
thus , this case represents the maximal amount of gas cooling and cosmic accretion that is feasible for gas with primordial metallicity .
first , examining the impact on dark matter , the inclusion of radiative cooling does not change significantly the growth of @xmath106 at @xmath145 , but it reduces the physical growth rate at smaller @xmath30 , which may be because the runaway cooling of gas accelerates the growth of @xmath106 in the core at early times .. this introduces uncertainties in the exact center position , so one should interpret the evolution of the dark - matter density profile at @xmath146 in ( right ) with care .
] however , @xmath110 does not change significantly . for gas ,
the addition of radiative cooling causes it to decouple from dark matter starting at @xmath147 , because this cooling prevents the formation of strong virial shocks and pressure support @xcite at this @xmath30 . at @xmath148 ,
the evolution of @xmath141 is non - monotonic , as driven by the relative efficiency of cosmic accretion and cooling . at @xmath149
, @xmath141 increases significantly over time , at a rate many times that of dark matter . at @xmath150 ,
@xmath141 is nearly constant because of the balance of cosmic accretion and gas cooling .
the second row shows @xmath143 , as scaled to the cosmic value , @xmath151 . at large @xmath30 , baryons trace dark matter , but at @xmath152 , @xmath143 decreases , as gas cools and advects to smaller @xmath30 in a relatively static profile of dark matter . @xmath143
reaches a minimum at intermediate @xmath30 , which moves outward over time , driven primarily by the changing @xmath153 . at small @xmath30
, @xmath143 rises rapidly as gas cools among relatively static dark matter . as with non - radiative gas ,
@xmath142 for radiative gas tracks @xmath110 beyond dark matter s @xmath154
. however , in this case , infalling gas cools and continues to advect to smaller @xmath30 with relatively little pressure support , so @xmath114 is much smaller and less well - defined for gas .
nevertheless , gas does not experience runaway advection to small @xmath30 , because some shocking occurs at smaller @xmath30 that imparts some pressure support ( as we have checked explicitly ) , and gas starts to feel significant angular - momentum support , as we explore in the next section . in combination
, these processes decrease the magnitude of @xmath142 at fixed @xmath30 over time . to summarize , the cosmic accretion of gas is distinct from that of dark matter at @xmath155 .
while the collisional nature of gas leads to some differences , this effect is modest .
instead , the dissipational radiative cooling of gas drives strong differences , which lead to a significant physical advection / accretion of gas at all @xmath30 and @xmath13 atop a relatively static profile of dark matter .
+ having explored cosmic accretion in simulations with only dark matter , with non - radiative gas , and with radiatively cooling gas , we now examine our simulation that additionally includes star formation and feedback . here , the primary differences compared with the simulation with only cooling are ( 1 ) thermal energy injection from supernovae and stellar winds , which heat the gas at small radii ; ( 2 ) advection of metals into the halo gas , which enhances the efficiency of cooling ; are @xmath156 of solar , which implies at most a modest ( factors of a few ) increase to the cooling rate compared to primoridal gas in the previous section . ] and ( 3 ) the transition of some gas to stars , which behave as a collisionless fluid , similar to dark matter . henceforth , we compute all baryon masses including both gas and stars , to simplify the interpretation of cosmic accretion without the ambiguities of conversions between the two from star formation and stellar mass loss and to allow us to compare with our simulations without star formation .
however , we continue to compute velocities separately for gas
. shows profiles of baryons and dark matter as a function of @xmath30 , for halos selected in two bins of @xmath94 at @xmath70 and their progenitors , similar to figures [ fig : profile_dm ] and [ fig : profile_gas_physics ] .
( top row ) shows @xmath106 and @xmath153 ( upper sub - panels ) , as well as @xmath105 and @xmath157 , that is , physical growth ( lower sub - panels ) .
the injection of energy from feedback mitigates the effects of gas overcooling on dark matter at small @xmath30 , compared with the cooling - only simulation ( right ) .
however , this feedback also reduces the infall velocity of dark matter at @xmath158 at @xmath11 at both masses , as compared with simulation with only dark matter ( ) .
we think that this is caused by angular - momentum transfer from cooling gas to dark matter , because the angular momentum of dark matter at @xmath159 is @xmath160 higher in this simulation ( see ) than with only dark matter .
furthermore , this boost in dark - matter angular momentum is higher in lower - mass halos , likely because of their lower potential energy , which also explains the stronger differences in infall velocity at @xmath158 for lower - mass halos in . compared to @xmath106 ( dashed curves ) , @xmath153 ( solid curves ) increases much more over time , again because of the ability of gas to cool radiatively and advect to smaller @xmath30 .
specifically , at small @xmath30 , @xmath153 increases monotonically with time , @xmath161 since @xmath21 , with more growth at higher mass , although thermal feedback has reduced the growth of @xmath153 at @xmath162 noticeably compared to the case with only cooling .
( second row ) shows @xmath143 , as scaled to the cosmic value , @xmath151 .
the trends are similar to those in ( right ) . at large @xmath30 , baryons trace dark matter at the cosmic value . at @xmath152 ,
@xmath143 decreases , as gas cools and advects to smaller @xmath30 in a relatively static profile of dark matter .
@xmath143 reaches a minimum at intermediate @xmath30 , where gas dynamics are most decoupled from dark matter , and this @xmath30 moves outward over time , driven primarily by the changing @xmath153 , while @xmath106 remains nearly static . at small @xmath30
, @xmath143 rises rapidly as cooled gas settles near the galaxy .
thus , baryons advect to smaller @xmath30 even in the absence of the physical growth of dark matter .
( third row ) shows @xmath110 and @xmath142 .
again , the trends are similar to those in ( right ) . at @xmath163 , beyond @xmath114 of dark matter , @xmath142 closely tracks @xmath110 .
as stated above , the @xmath30 of this decoupling is set by @xmath164 of dark matter , whereas radiative gas continues to cool and advect to smaller @xmath30 .
( though the stars , being collisionless , can experience splashback similar to dark matter . )
gas reaches stronger @xmath165 at smaller @xmath114 , which corresponds to the minimum of @xmath143 .
this behavior is governed by the efficiency of gas cooling and the onset of rotational support , and indeed , at the smallest @xmath30 , @xmath142 moves toward 0 , where gas becomes strongly supported by angular momentum ( see below ) .
these trends for gas persist at all @xmath13 , though as with dark matter , the inflow velocity at fixed @xmath30 becomes weaker at later times .
finally , we emphasize that with the inclusion of gas , star formation , and feedback , halos at @xmath166 stop having physically meaningful infall regions for dark matter at @xmath12 . furthermore , while meaningful gas inflow does persist at all @xmath167 ( the largest @xmath30 where @xmath168 ) , note that @xmath34 does not increase at @xmath12 . in other words , at @xmath12 the physical size of the region from which low - mass halos accrete gas no longer grows , because these halos can not overcome both dark - energy acceleration ( ) and large - scale tidal motions ( see ) . in this sense , low - mass halos _
have _ decoupled from the cosmic background , but ( residual ) gas infall persists at all @xmath167 because of radiative cooling .
this trend is similar in the simulation with only cooling .
+ for halos across the mass range that we examine , the cooling time of the gas is similar to or shorter than their dynamical times , especially at our low - mass end @xcite , which naively implies that gas should be able to cool and advect efficiently to @xmath169 .
however , we have shown that the gas density profiles in even low - mass halos remain extended instead of accreting to @xmath169 on a cooling timescale . what then regulates the rate of gas accretion into the galaxy and causes the weaker infall velocity at small @xmath30 in ( also in ) ?
the answer is that cosmic accretion plus gas cooling are not the only drivers of gas flow into the galaxy , but rather , angular - momentum support also plays a critical role . to examine the level of angular - momentum support ,
we compute the _ magnitude _ of tangential velocity ( regardless of direction ) for each cell / particle , and we sum this _ scalar _ value to compute the average tangential velocity , @xmath170 , in bins of @xmath30 .
we then scale @xmath170 for each species to the circular velocity , @xmath171 , for which @xmath172 is the total mass within @xmath30 .
shows @xmath173 separately for baryons and dark matter .
virialized orbits are rotationally supported by angular momentum if @xmath174 .
shows that dark matter is not fully rotationally supported at any @xmath30 . in addition
, the level of rotational support within @xmath175 is nearly constant over time , analogous to the lack of evolution of @xmath106 . by contrast ,
gas , which cools while largely conserving angular momentum , becomes strongly rotationally supported at small @xmath30 ( @xmath176 ) , which moves outward over time .
, @xmath177 , because some gas and stars at the core are on high - energy unbound orbits for which @xmath178 . furthermore , some halos experience mergers and not fully relaxed , which leads to slight offsets in halo centering . ] note that these @xmath30 are much larger than the galactic stellar disk , whose size is @xmath179 @xcite .
thus , the rate of gas accretion into the disks of galaxies at late cosmic time is governed not just by cosmic accretion near @xmath5 , but rather by angular - momentum support and transport at @xmath180 .
while the cooling of gas causes it to have a higher angular - momentum than dark matter at small @xmath30 , note that at larger @xmath30 , dark matter has systematically higher angular - momentum support than gas .
this is because gas experiences thermal pressure , mixing , and shocking that cause it to settle into a more coherent velocity field ( stream ) , while dark matter , being collisionless , retains a higher dispersion in tangential velocity .
indeed , we find that the scatter in @xmath181 for dark matter is higher than that of gas at all @xmath30 ( see also * ? ? ? * ) .
other analyses ( e.g. , * ? ? ?
* ; * ? ? ?
* ) found that gas accretes into a halo with systematically _ higher _ angular momentum than dark matter .
these results seemingly contradict , but the reason lies in the way that one computes / examines angular momentum . @xcite and @xcite examined the net angular - momentum _ vector _ of a halo , that is , they computed the angular - momentum vector of each particle , summed these vectors , and then took the magnitude of this sum .
by contrast , we compute the _ magnitude _ of angular momentum for each particle / cell and sum this scalar .
our choice is motivated by examining the angular - momentum support for each cell / particle , as opposed to the net angular momentum of the halo .
all of the above trends are similar in the simulation with only cooling .
however , in the non - radiative simulation , gas has less angular - momentum support than dark matter at all @xmath30 , because gas shocking and pressure support reduce bulk orbital motions , and even gas with low angular momentum can not cool and advect to small @xmath30 .
+ + until now , we have shown that many features in the profiles of both baryons and dark matter are coincident with some fraction of a halo s @xmath125 .
we now examine the relation to @xmath125 in more detail in order to understand to what extent it is a physically meaningful radius . shows the same profiles as in , but scaled by each halo s @xmath125 .
while we are most interested in the degree to which this scaling to @xmath182 captures physical features in the profile , we also scale @xmath138 to the cosmic @xmath183 and @xmath113 to each halo s @xmath184 , to examine more clearly the degree of self - similarity in the profile .
thus , our primary interest is whether any features in the profiles occur at a fixed @xmath182 , and our secondary interest is whether the profiles normalized by virial values are self - similar . for both baryons and dark matter
, @xmath185 is self - similar to within a factor of @xmath186 , with a somewhat stronger degree of self - similarity at @xmath187 than at smaller @xmath30 ( see also * ? ? ?
* for similar results for massive galaxy clusters ) .
thus , @xmath104 at @xmath152 is more invariant over time at fixed @xmath30 than at fixed @xmath188 , a result of the dissipationless nature of dark matter .
however , @xmath104 at @xmath163 and @xmath189 at all @xmath30 are more invariant at fixed @xmath188 , because they respond to continued physical cosmic accretion .
@xmath190 shows similar invariance at fixed @xmath188 , with deviations of @xmath191 at @xmath192 .
@xmath190 shows larger deviations at smaller @xmath30 , where regulation by gas cooling and angular - momentum support are more important .
the @xmath30 of the minimum of @xmath190 occurs at @xmath193 , being smaller for higher - mass halos . at @xmath30 beyond this minimum , @xmath143 clearly responds to @xmath125 .
considering @xmath194 , @xmath154 for dark matter .
scaled to @xmath182 , this shows more clearly that @xmath125 roughly tracks the inner edge of the infall region for dark matter , where @xmath195 . for gas
, the velocity profile shows less self - similarity , though in all cases @xmath196 .
neither @xmath197 nor @xmath198 are particularly self - similar in terms of normalization , because the infall velocities of both components decrease over time , while we find that @xmath199 for these halos is nearly static ( it increases slightly with time as their potential deepens somewhat ) .
however , these scaled velocity profiles are somewhat more self - similar than the non - scaled ( physical ) profiles
. finally , shows the scaled profile of angular - momentum support , that is , @xmath200 .
recall from that the level of angular - momentum support for dark matter is nearly constant over time at fixed @xmath30 ( @xmath201 ) , similar to @xmath104 , while the level of angular - momentum support for gas grows significantly at fixed @xmath30 , and thus the @xmath30 out to which gas experiences strong rotational support increases with time .
however , shows that radial extent of angular - momentum supported gas is fixed at @xmath202 , that is , it responds to @xmath125 .
( similar results persist for our simulation with only radiative cooling . )
this is likely because , for a given halo , cosmic accretion at later times comes in with a higher impact parameter and thus a higher specific angular momentum @xcite . for dark matter , accreting mass
is deposited ( in a time - average sense ) at large @xmath30 , so it does not significantly change the angular momentum at small @xmath30 . however , as gas cools and advects to smaller @xmath30 , it largely conserves angular momentum , so it continues to advect efficiently until it becomes strongly rotationally supported , which occurs at larger @xmath30 over time as the accretion at @xmath125 has higher specific angular momentum .
thus , we conclude that , @xmath125 does have physical meaning , given its correlation with these profiles , for all properties except @xmath106 at @xmath203 . this does not mean necessarily that @xmath125 captures the exact location of any physically meaningful feature .
indeed , the above figures show that the exact @xmath30 depends on the property in question .
rather , the scaling of @xmath5 with @xmath13 captures the _ relative _ radius at which physical features occur .
we will pursue a more detailed investigation of various virial scalings in future work .
+ finally , summarizes this section and the most important results of this paper by showing the mass growth and specific accretion rate over time at fixed @xmath30 , similar to , but for both baryons and dark matter .
( top row ) shows the median ratio @xmath204 at various @xmath30 separately for both baryons ( solid ) and dark matter ( dashed ) , for the same halos as in .
scaling by @xmath205 of each component allows easy comparison of baryons and dark matter at different @xmath30 in the same panel .
the panel also shows both @xmath206 and @xmath207 as dashed and solid thick black curves . in the presence of star formation and feedback ,
the ( lack of ) growth of dark matter is qualitatively similar to . at large @xmath30 ,
@xmath130 declines over time , following the expanding universe .
after a shell reaches turn - around , the enclosed mass grows over time , but this growth quickly stalls because of the dissipationless nature of dark matter .
cosmic accretion deposits dark - matter mass at larger @xmath30 in a shell - like manner , with little growth of @xmath130 at smaller @xmath30 .
baryon mass closely tracks that of dark matter at large @xmath30 .
however , after turn - around , dissipative cooling causes baryon mass to decouple from dark matter , as @xmath208 grows significantly over time at fixed @xmath30 .
even at the smallest @xmath30 , the relative growth rate of baryon mass mimics the relative growth rate of @xmath131 ( thick curves ) , so the accretion at @xmath125 regulates baryonic mass growth at smaller @xmath30 . for our lowest - mass bin , note the relative flatness of mass growth at @xmath209 . at @xmath11
, this @xmath30 corresponds to the turn - around radius , @xmath34 , which as showed , does not grow for these low - mass halos .
shows more explicitly that @xmath210 does not change at @xmath11 , nor will it change in the future , because it corresponds to the asymptotic baryon mass of these low - mass halos . comparing the absolute growth at different @xmath30 in ( top row )
is difficult , because we normalize growth at each @xmath30 by a different @xmath205 .
thus , ( bottom row ) shows the specific accretion rate at fixed @xmath30 . here
, we measure the accretion rate by differencing @xmath31 at fixed @xmath30 across adjacent snapshots , and we show the specific accretion rate , @xmath211 , to allow easier comparison across mass bins ( the results do not change qualitatively if instead we examine the absolute accretion rate ) . finally , we multiply the specific accretion rate for dark matter by the cosmic @xmath151 to allow easy comparison . again
, thick black curves show the specific accretion rates inferred via @xmath212 for baryons ( solid ) and dark matter ( dashed ) . at large @xmath30 ,
the specific accretion rate is negative , corresponding to uncollapsed mass that is expanding with the universe .
after turn - around , the specific accretion rate becomes positive .
at all @xmath167 , including @xmath125 , the rate declines over time , as caused by the declining infall velocity and density of accreting matter , as demonstrated in previous sections . at fixed @xmath30 , the specific accretion rate of dark matter
declines over time and then stalls near zero . before stalling
, the rate of decline approximately tracks that at @xmath125 , so the mass flux at @xmath125 sets the accretion rate at a given @xmath30 for some time . for baryons , after turn - around , at essentially all fixed @xmath30 , the accretion rate tracks that at @xmath125 , though lower by a factor of @xmath186 , depending on mass .
thus , the gas accretion rate at @xmath125 governs that at _ all _ smaller @xmath30 , though at a reduced rate , because gas cooling is not instantaneous , and gas experiences significant rotational support , especially at smaller @xmath30 .
the accretion rate inferred from @xmath213 is not quite synonymous for dark matter and baryons ( dashed and solid thick black curves ) , because @xmath5 represents a radius at which the dynamics of the two components start to decouple , as we demonstrated above .
( we checked that using a larger @xmath30 yields accretion rates that are more synonymous . ) for this reason , the better agreement of baryonic accretion rates at fixed @xmath30 with @xmath213 of dark matter at @xmath108 is mostly a coincidence , though perhaps a fortuitous one , given that baryonic accretion rates often are inferred from dark - matter simulations via @xmath214 .
we conclude that the accretion rate and mass growth of both dark matter and baryons are governed by the cosmic accretion rate at @xmath159 .
however , at @xmath215 , these components differ significantly .
the accretion rate and mass growth of dark matter stalls , because it is dissipationless . because gas cools , the baryonic accretion rate and mass growth continue down to all @xmath30 , at a rate that tracks that at @xmath125 , but with a lower absolute value because of the additional physics of finite cooling efficiency and angular - momentum support .
in this work , we examined the _
physical _ nature of cosmic accretion and mass growth in halos with @xmath216 .
while cosmic accretion and mass growth are typically measured using some evolving virial relation , such as @xmath131 , we examined the _
physical _ growth of both dark matter and baryons at fixed physical radii over time at @xmath23 , including the relationship of this growth to @xmath131 .
we presented and analyzed a suite of cosmological simulations , _
omega25 _ , that incorporate both dark matter and gas dynamics with differing treatments of gas cooling , star formation , and thermal feedback ( though absent strong stellar winds ) to explore systematically the physics that governs the accretion of dark matter and baryons into halos and their galaxies .
we summarize our main results , first for the cosmic accretion of _ dark matter _ :
physically meaningful cosmic accretion occurs for dark matter at @xmath122 ( depending on mass ) . _ over time , the declining average density of matter in the universe , combined with the stronger dynamical effects of dark energy and large - scale tidal motions for accretion at larger physical radii , reduce the infall velocity of accreting matter and thus the accretion / growth rate .
at @xmath11 ( depending on mass ) , halos with @xmath217 do not experience significant physical accretion / growth of dark matter at _ any _ radius .
_ this is because dark matter is dissipationless , so it is deposited ( in a time - average sense ) at @xmath159 in a shell - like manner , not within the halo .
the most physically meaningful radius to measure cosmic accretion and mass growth of dark matter is at the radius of maximum average infall velocity , @xmath114 . _
physically , @xmath118 , the splashback ( or secondary turn - around ) radius of ( collisionless ) dark matter . for halos in our mass range , @xmath154 , though there is no meaningful @xmath114 for low - mass halos at @xmath11 .
while commonly used , @xmath5 represents an incomplete census of the mass that has passed through the halo , though @xmath5 approximately corresponds to the outer edge of where the profile is most static . additionally , we summarize our main results for the cosmic accretion of _ baryons _ :
physically meaningful cosmic accretion and mass growth of baryons persists at all radii and across all redshifts . _
the difference between dark matter and gas arises not because gas is collisional , but rather because gas can cool radiatively and advect to smaller radii .
while dark - matter growth is subject to pseudo - evolution , baryon growth is not .
_ the physical accretion rate of baryons at all radii inside of the halo roughly tracks the accretion rate into the halo measured at @xmath159 .
_ though the rate at smaller radii has a somewhat lower normalization as governed by the efficiency of gas cooling and angular - momentum support .
_ accreting gas becomes strongly rotationally supported at @xmath218 , independent of redshift .
_ thus , the rate of gas inflow into the galaxy is regulated by angular - momentum support .
4 . _ inflowing gas starts to decouple from dark matter at @xmath219_. for halos in our mass range , @xmath220 .
this is the smallest radius where the specific accretion rates are the same for both baryons and dark matter .
the physical size of the region that sources cosmic accretion into low - mass halos ( @xmath166 ) does not grow at @xmath12 .
_ matter decouples from the cosmic expansion and starts to fall into the halo at the turn - around radius , @xmath221 , but this radius stops increasing for low - mass halos at @xmath12 . in this sense ,
low - mass halos have decoupled from the cosmic background for most of their history , but physically meaningful infall of gas to smaller @xmath30 does persists because of gas cooling .
we have shown that scaling profiles of dark matter and baryons to @xmath188 preserves a strong degree of self - similarity in @xmath222 , @xmath190 , @xmath223 , and @xmath173 .
while this argues for the meaningfulness of the scaling of @xmath5 with @xmath13 , this does not argue necessarily for the absolute value of @xmath125 .
our results , and those of other works @xcite , indicate that commonly used virial radii , such as @xmath5 , represent an incomplete census of the physical mass associated with halos .
furthermore , other commonly used definitions , such as @xmath224 and @xmath8 @xcite are even smaller than @xmath5 . given that dark matter and
baryons are continuous density fields , that cosmic accretion is triaxial and clumpy , and that halos are not fully relaxed `` virialized '' systems , any definition of a virial radius / boundary for a halo is at some level an oversimplification .
that said , our results suggest that @xmath220 is the most physically meaningful radius to measure the mass growth and accretion of dark matter for halos in our mass range .
additionally , beyond just the splashback of dark matter , this radius also corresponds to where a significant fraction of ( observable ) galaxies around massive groups / clusters are splashback satellite galaxies that passed through the host ( for example , * ? ? ?
moreover , @xmath220 is the most natural radius for comparing dark matter with baryons because this is where their average infall dynamics starts to decouple . for baryons ,
their collisional and dissipational dynamics mean that there is no shell crossing ( with the exception of any gas that remains bound in massive subhalos ) , therefore there is no analogous @xmath56 . in sufficiently high - mass halos
, gas can experience an analogous virial - shock radius , though this can occur at a range of radii with respect to @xmath5 , depending on mass and @xmath13 @xcite .
thus , we suggest that @xmath147 has many advantages over @xmath5 ( or even smaller radii ) for measuring the physically meaningful extent , and thus the cosmic accretion rate and mass growth , of halos in our mass range .
nevertheless , the exact value of @xmath56 and thus @xmath114 , with respect to @xmath5 , depends on the specific accretion rate and therefore on halo mass and @xmath13 , such that @xmath56 and @xmath114 occur more typically at @xmath225 in massive galaxy clusters @xcite . in a follow - up analysis
, we will examine a broader range of virial definitions to examine which best capture physical scales of density , velocity , and thermal profiles for dark matter and gas , including dependence across a more comprehensive range of halo mass and @xmath13 . as discussed in the introduction ,
galaxies obey tight scaling relations with their host halos , including both mass and size .
this tightness over cosmic time may seem surprising in the context of the pseudo - evolution of halo mass , given that dark - matter mass and density do not change at small @xmath30 , on scales of the galaxy . however , our results provide some insights into this relation .
baryons continue to cool and advect to scales of the galaxy at a rate that roughly tracks the accretion rate at @xmath125 .
thus , we expect that @xmath226 _ approximately _ captures the collapsed mass of the halo , which also represents the baryons available to feed the galaxy .
in detail , our results imply that @xmath227 may provide a better correlation with the properties of the galaxy .
if true , this means that galaxy growth at @xmath23 is at some level limited and/or regulated by cosmic accretion at @xmath228 , despite the added complexity of feedback and wind recycling .
this picture is supported at least indirectly by the similarity of the decline in halo accretion rates and galaxy star - formation rates at @xmath23 ( for example , * ? ? ?
* ; * ? ? ?
* ) , though it requires more investigation in the context of simulations with more detailed and physically motivated feedback models . while we have shown that the accretion / growth rates of dark matter and baryons are decoupled at small @xmath30 within a halo , our results do support the methodology of using accretion rates from dark - matter simulations to estimate baryonic accretion rates into halos , provided that this is measured at @xmath187 , or more optimally , at @xmath147 .
furthermore , while the accretion rates are decoupled at small @xmath30 , they may still correlate with each other , in the sense that the halos with the highest accretion rates of dark matter may also have the highest rates for baryons .
if true , this may shed light on recent `` age matching '' models that assume a tight correlation between star formation in galaxies at @xmath229 and the formation timescales ( effectively measured at high @xmath13 ) of their host halos @xcite . in future work
, we will examine such correlations between rates of baryons and dark matter across cosmic time in more detail .
finally , our result that gas typically becomes strongly rotationally supported at @xmath218 , independent of redshift , has interesting implications for galaxy size growth .
this suggests that galactic disks are fed by torquing or viscous advection from much more extended pseudo - disks of rotationally supported gas .
indeed , several analyses of cosmological zoom - in simulations of individual @xmath230 galaxies found that they had extended thick and/or warped disks of cool accreting gas out to @xmath231 and beyond @xcite .
furthermore , observations of nearby galaxies show that disks of cold atomic gas often extend many times beyond the size of the stellar disk ( for example , * ? ? ?
if the resultant size of the galactic disk is governed by the size of the extended gaseous pseudo - disk , then this could explain the tight correlation between the observed size of a galaxy and the inferred size of its host halo as noted in @xcite .
nevertheless , we emphasize that the size of the stellar disk is much smaller at @xmath232 , so such a connection requires more detailed analysis using higher - resolution simulations .
our results do not change substantially in comparing simulations with only cooling to those with star formation and thermal feedback .
however , we do not include more detailed radiative and/or momentum feedback , so our simulations do not drive strong galactic outflows that are encapsulated in recent high - resolution simulations with more detailed treatments ( for example , * ? ? ?
* ) . in principle
, strong stellar winds could suppress gas accretion rates below what we find in this work .
@xcite examined cosmic accretion rates into halos in cosmological simulations using models for stellar and black - hole feedback in which winds are driven at a variety of fixed values for gas particle velocities , finding that the specific accretion rates into halos were broadly similar for baryons and dark matter , and that their feedback models change baryonic accretion rates only slightly for halos @xmath233 , though feedback yielded stronger reduction in gas accretion rates at smaller @xmath30 . using a different implementation of fixed - velocity wind models in a cosmological simulation , @xcite found that winds can reduce the rates of infalling gas , particularly in low - mass halos , by up to a factor of 2 for their strongest wind model , again with stronger reduction at smaller @xmath30 .
more recently , @xcite used a combination of delayed cooling supernova and early stellar feedback in a handful of cosmological zoom - in halos of mass @xmath234 at @xmath70 , finding that gas accretion rates into halos did not change with their feedback implementation .
similarly , @xcite examined gas accretion rates in cosmological simulations that generate galactic winds at a velocity that scales with the local velocity dispersion of dark matter , with additional thermal and ionizing feedback from black holes , and found that feedback reduced gas accretion rates only slightly at @xmath8 as compared with no feedback , but that feedback increased the gas inflow rates at @xmath235 as a result recycling from the winds .
overall , these works suggest that galactic winds do not substantially affect baryonic accretion rates at and beyond @xmath5 but that they do drive gas recycling that modulates inflow / outflow rates near the core of the halo .
furthermore , it is not clear how much galactic winds , launched from the galaxy with a low impact parameter and thus a low angular momentum , would change the angular - momentum distribution of extended halo gas .
however , we emphasize that almost all previous works examined winds with phenomenologically tuned prescriptions for wind velocities and its coupling ( or lack thereof ) with the surrounding halo gas .
such analyses should be revisited with more realistic and comprehensive treatments of stellar feedback @xcite .
we thank andrey kravtsov , surhud more , phil hopkins , frank van den bosch , andrew hearin , erik tollerud , erwin lau , and benedikt diemer for useful discussions and/or comments on an early draft .
we also thank the reviewer for useful comments .
a.r.w . acknowledges the hospitality and stimulating environment of the aspen center for physics , supported in part by the nsf . a.r.w .
also gratefully acknowledges support from the moore center for theoretical cosmology and physics at caltech . this work was supported in part by nsf grants ast-1412768 & 1009811 , nasa atp grant nnx11ae07 g , nasa chandra grants go213004b and tm4 - 15007x , the research corporation , and by the facilities and staff of the yale university faculty of arts and sciences high performance computing center | the cosmic accretion of both dark matter and baryons into halos typically is measured using some evolving virial relation , but recent work suggests that most halo growth at late cosmic time ( @xmath0 ) is not physical but is rather the by - product of an evolving virial radius ( `` pseudo - evolution '' ) . using _
omega25 _ , a suite of cosmological simulations that incorporate both dark matter and gas dynamics with differing treatments of gas cooling , star formation , and thermal feedback , we systematically explore the physics that governs cosmic accretion into halos and their galaxies .
physically meaningful cosmic accretion of both dark matter and baryons occurs at @xmath1 across our halo mass range : @xmath2 .
however , dark matter , because it is dissipationless , is deposited ( in a time - average sense ) at @xmath3 in a shell - like manner , such that dark - matter mass and density experience little - to - no physical growth at _ any _ radius within a halo at @xmath4 .
in contrast , gas , because it is able to cool radiatively , experiences significant accretion at all radii , at a rate that roughly tracks the accretion rate at @xmath5 , at all redshifts . infalling gas starts to decouple from dark matter at @xmath6 and continues to accrete to smaller radii until the onset of strong angular - momentum support at @xmath7 .
thus , while the growth of dark matter is subject to pseudo - evolution , the growth of baryons is not .
the fact that the accretion rate of gas on galactic scales tracks the accretion rate near @xmath5 provides insight into the tight relations between the masses / sizes of galaxies and those of their host halos across cosmic time . |
we show in this paper by elementary means that the support of the capacity achieving input measure for multiple - input multiple - output ( mimo ) rayleigh fading channels subject to average power constraint with coherence time @xmath0 is bounded .
a generalization of the result to coherence intervals of size @xmath1 seems to be highly non - trivial and will probably require a substantial extension of the techniques used here supplemented by some results and methods from the `` hard analysis '' .
+ previous fundamental achievements , e.g. @xcite , follow the same procedure which can be traced back to the classic paper @xcite by smith . the basic tools are the karush - kuhn - tucker ( kkt ) conditions from the theory of convex optimization supported by an application of the identity theorem ( also known as the uniqueness theorem ) from complex analysis . our approach is based on the kkt conditions too but avoids the usage of the identity theorem .
+ in @xcite abou - faycal , trott , and shamai proved , using these techniques , that for a one - dimensional rayleigh fading channel the optimal input measure subjected to an average power constraint to be discrete with a finite number of mass points . in @xcite chan , hranilovic , and
kschischang showed for a mimo rayleigh block - fading channel with i.i.d .
channel matrix coefficients that the optimum input distribution subjected to peak and average power constraint contains a finite number of mass points with respect to a specific norm .
in addition fozunbal , mclaughlin , and schafer argued in @xcite that a bounded support of the capacity maximizer implies its singularity with respect to the borel - lebesgue measure .
the approach in @xcite is based on the identity theorem for holomorphic functions in several complex variables and use the assumption that an open set in @xmath2 fulfills the hypothesis of the identity theorem in @xmath3 .
we show in section [ sec - discussion ] by a simple example that the conclusion of the identity theorem fails in this setting
. consequently , these results are not rigorous . since , in contrast to the complex analysis in one variable , it is still an open difficult problem to characterize the families of sets for which the identity theorem for holomorphic functions in several complex variables holds we can not hope to understand the properties of the capacity maximizers in the present setting by an reduction to uniqueness properties of holomorphic functions in higher dimensions .
therefore , it is likely that we will be forced to develop or apply `` real - analytic '' tools for tackling this important communication - theoretic problem .
+ the paper is organized as follows : section [ b ] provides some basic definitions and is followed by section [ sec - bounded - support ] which contains the main result of this paper . as mentioned above , in section [ sec - discussion ] we give an elementary example that shows that the application of the identity theorem in higher dimensions is , in general , not admissible if we want to understand the properties of capacity maximizers of rayleigh fading channels .
+ _ notation .
_ throughout the paper we will denote the set of complex @xmath4-by-@xmath5 matrices by @xmath6 and will freely identify this set with @xmath7 .
@xmath8 stands for the logarithm to the base @xmath9 .
capital letters @xmath10 are reserved for random variables .
we consider a rayleigh fading channel with the coherence time @xmath0 which is described by @xmath11 with coefficient matrices @xmath12 , + @xmath13 and @xmath14 , where the the channel h is assumed to be complex circularly symmetric gaussian with zero mean and with covariance matrix @xmath15 and the additive noise coefficients @xmath16 are assumed to be i.i.d .
complex circularly symmetric gaussian with @xmath17 .
let @xmath18 be the set of probability measures on + @xmath19 .
then the set @xmath20 with the average power constraint of the transmitted signal @xmath21 is weak * compact as it was shown in @xcite and @xcite .
+ if @xmath22 is the set of conditional probability measures on + @xmath23 we can determine the channel by a set @xmath24 , where @xmath25 is absolutely continuous with respect to borel - lebesgue measure . for the rayleigh fading channel the conditional probability density of the received signals @xmath26 conditioned on the input symbol @xmath27 is given by @xmath28}}{\pi^{m}\textrm{det } ( \sigma^2_z{\mathbf{1}}_{m } + ( { \mathbf{1}}_m\otimes x^h)\sigma ( { \mathbf{1}}_m\otimes x ) ) } \ ] ] with covariance matrix @xmath15 of @xmath29 @xmath30 let @xmath31 be a probability measure and define @xmath32 then the mutual information of the channel with no csi at the receiver is given by @xmath33 the mutual information is a weak * continuous functional on the weak * compact and convex set @xmath34 ( see @xcite ) . thus the functional @xmath35 achieves its maximum on @xmath34 by the following let @xmath36 be a weak * continuous real - valued functional on a weak * compact subset @xmath37 of @xmath38 .
then @xmath36 is bounded on @xmath37 and achieves its maximum on @xmath37
. the mutual information is strictly concave functional on @xmath34 up to _ equivalence _ of measures .
hereby , two measures @xmath39 are called equivalent if @xmath40 .
so its maximum on @xmath34 is achieved by a unique input distribution up to equivalence defined above @xcite .
hence , with @xmath41 there exists a measure @xmath42 that achieves the capacity of the channel and is unique up to equivalence of measures .
the aim of this paper is to show that subjected to an average power constraint the capacity achieving distribution of the channel has an bounded support .
the purpose of this section is to show that the support of the capacity achieving input measure for the channel given in ( [ channel - def ] ) , with coherence time @xmath0 , is bounded . + for @xmath43 with @xmath44 we set @xmath45 with @xmath46 .
+ [ lemma1 ] let @xmath43 with @xmath44 and @xmath47 with @xmath31 be given . then @xmath48 with @xmath49 and @xmath50 and @xmath51 are the minimum and maximum eigenvalues of the covariance matrix @xmath15 . by the defining relation ( [ int - measure ] ) we have @xmath52 next we define + @xmath53 whereas the maximum of the function is achieved on @xmath54 because of the compactness of @xmath54 . hence ,
+ for @xmath55 we obtain @xmath56}}{\pi^{m}\pi}.\ ] ] for every @xmath57 we have @xmath58 where @xmath50 and @xmath51 are the minimum and maximum eigenvalues of the hermitian and strictly positive covariance matrix @xmath15 . by the definition of @xmath54 we have @xmath59 hence
, it follows that @xmath60 for two operators @xmath61 with @xmath62 and a positive operator @xmath63 we have @xmath64 due to the fact that the operators in ( [ ineq ] ) are hermitian and positive and the same holds for @xmath65 and because the function @xmath66 is operator monotone for all positive operators @xcite , we have @xmath67 \geq \\ \textrm{tr}\left[((\sigma^2_z + \lambda_{min}x^hx ) { \mathbf{1}}_m)^{-1}yy^h\right ] \\
\geq \textrm{tr}\left[(\sigma^2_z{\mathbf{1}}_{m } + ( { \mathbf{1}}_m\otimes x^h)\sigma({\mathbf{1}}_m\otimes x ) ) ^{-1}yy^h \right].\end{gathered}\ ] ] with ( [ ineq - max ] ) it follows that for @xmath68 @xmath69}}{\pi^{m}\pi}\ ] ] inserting this into ( [ ineq - shell ] ) yields @xmath70}.\ ] ] therewith we get & p(y|x)f_(y)dy & + & p(y|x)dy & + & = - p(y|x ) dy & + & = - p(y|x ) dy & + & = - & + & - & + & = - & & a:=. & determining the capacity achieving input distribution subjected to average power constraint is a convex optimization problem .
necessary conditions for the optimal input distribution can be derived from the local _ karush - kuhn - tucker _ conditions .
together with the fact that the mutual information is a concave functional and the convexity of the constraint functional we obtain ( see @xcite and @xcite ) , that @xmath71 achieves capacity if and only if @xmath72 with equality if @xmath73 , where @xmath74 denotes the lagrange multiplier and @xmath75 is the constraint under consideration .
it is fairly standard fact that @xmath76 \nonumber\end{gathered}\ ] ] and ( [ loc - kkt ] ) can be therefore rewritten as [ loc - kkt3 ] & ( x^2 - a)+c(a)+ ( e)^m + & + & + ( ^2_z_m + ( _ m x^h ) ( _ m x ) ) + & + & + p(y|x)f_(y)dy 0 & with equality if @xmath73 . let [ kkt ] & kkt(x):= ( x^2 - a)+c(a)+ ( e)^m + & + & + ( ^2_z _
m + ( _ m x^h ) ( _ m x ) ) + & + & + p(y|x)f_(y)dy & then ( [ loc - kkt ] ) can be rephrased as @xmath77 for @xmath78 and @xmath79 if @xmath80 .
+ the following theorem gives a sufficient condition for the boundedness of the support of the capacity achieving measure in terms of the lagrange multiplier @xmath81 .
[ lemma - bounded - input ] let @xmath82 be given and let @xmath71 be a capacity achieving input measure subject to the average power constraint @xmath83 for the channel ( [ channel - def ] ) .
then @xmath84 implies that @xmath85 is bounded .
the proof is by contradiction .
suppose that @xmath86 and that @xmath85 is not bounded . by our assumptions we can find @xmath87 with the following properties : @xmath88 applying _ lemma _ [ lemma1 ] to the function @xmath89 defined in ( [ kkt ] ) we obtain the following inequality .
[ kktb ] & kkt(x ) ( x^2 - a)+c(a)+ ( e)^m + & + & + ( ^2_z_m + ( _ m x^h ) ( _ m x))+ & + & + a - & + & = x^2 ( - ) -a+ c(a)+(e)^m + & + & + ( ^2_z _ m + ( _ m x^h ) ( _ m x))+ & + & + a - & combining the karush - kuhn - tucker conditions and ( [ kktb ] ) we obtain that for any @xmath73 & 0=kkt(x ) & + & x^2 ( - ) -a+
c(a)+(e)^m + & + & + ( ^2_z_m + ( _ m x^h ) ( _ m x))+ & + & + a - & but this last inequality with our assumption that @xmath85 is not bounded , ( [ fac_greater_zero ] ) , and the fact that & x^2 ( - ) x & + & & + & ( ^2_z _ m + ( _ m x^h ) ( _ m x ) ) x & implies that @xmath90 , which is the desired contradiction . in view of lemma [ lemma - bounded - input ]
our remaining goal is to show that @xmath84 for each @xmath91 .
for example in @xcite abou - faycal , trott , and shamai showed this in the scalar case .
our proof of the corresponding result in mimo case below is strongly motivated by their approach via fano s inequality .
[ gamma>0 ] for the channel given in ( [ channel - def ] ) we have @xmath84 for each @xmath91 .
as mentioned above the proof is an extension of the argument given in @xcite .
the capacity functional @xmath92 is a non - decreasing and concave function of the argument @xmath82 .
it was observed in @xcite using _ global _ karush - kuhn - tucker conditions that @xmath93 is the slope of the tangent line to @xmath92 at @xmath83 ( cf .
@xcite , section iii.b and appendix ii.a ) .
thus , since @xmath94 is non - decreasing and concave , it can be shown that @xmath95 implies @xmath96 for all @xmath97 need not be differentiable .
however , @xmath92 is differentiable a.e . due to the monotonicity and concavity .
the proof that @xmath98 for all @xmath97 follows a standard line of reasoning from the real analysis and is skipped due to the space limitation .
the full argument will be given elsewhere . ] .
consequently , we can rule out the possibility that @xmath95 by showing the existence of a sequence of input measures such that the corresponding sequence of mutual informations approaches @xmath99 .
+ we will be done if there is @xmath100 such that for each @xmath101 we can find distinct @xmath102 and disjoint measurable sets @xmath103 such that @xmath104 for all @xmath105 . because a simple application of fano s inequality with block length @xmath5 shows then that for the input measures @xmath106 ( @xmath107 is the point measure concentrated on @xmath108 ) we have @xmath109 now we define @xmath110 where @xmath111 denotes the surface area of the unit sphere in @xmath112 and @xmath113 are the smallest and the largest eigenvalues of @xmath15 and let @xmath114 be given .
+ we will now present the construction of the vectors @xmath115 and the decoding sets @xmath116 . let @xmath117 with @xmath118 be fixed and consider a large positive real number @xmath119 that will be specified later .
set @xmath120 for @xmath105 where @xmath121 .
+ let @xmath122 denote the smallest eigenvalue of @xmath15 .
for @xmath105 we set @xmath123 and @xmath124 where @xmath125 as shown in the proof of lemma [ lemma1 ] we have @xmath126 using ( [ gamma-1 ] ) and transforming to spherical coordinates in @xmath127we obtain @xmath128 where @xmath111 denotes the surface area of the unit sphere in @xmath129 and @xmath130 .
after the substitution @xmath131 in the integral on the rhs of the inequality ( [ gamma-2 ] ) we arrive at @xmath132 in what follows we use the abbreviation @xmath133 the defining relation ( [ gamma-0 ] ) and our assumption that @xmath134 ensure that @xmath135 . using this and ( [ gamma-3 ] )
we are led to @xmath136 for all @xmath137 .
now , since @xmath138 , @xmath130 , and @xmath139 it is clear that @xmath140 @xmath141 and from ( [ gamma-4 ] ) we have @xmath142 for all @xmath137 .
thus if we choose our @xmath143 sufficiently large ( [ gamma-5 ] ) and these limit relations ensure that @xmath144 for all @xmath137 .
moreover it is clear that the sequence of second moments of the measures @xmath145 can be made arbitrarily large for large @xmath146 .
this concludes our proof by the remarks given at the beginning of the argument .
now , we can summarize our results obtained so far in the following fashion : [ theorem - bounded - input ] we consider the channel defined by ( [ channel - def ] ) .
then the support of the capacity achieving input measure is bounded .
simply apply lemma [ gamma>0 ] and lemma [ lemma - bounded - input ] .
with the embedding function @xmath147 with @xmath148 and @xmath149 and the transformed channel we get an extension of the function & kkt(x ) : m(n1,c ) r & + & & + & kkt(z ) : m(2n1,c ) c & + & & + & kkt(z):=(z^tz - a)+c(a)-(|z)d . & @xmath150 and @xmath151 are obtained by changing the channel matrix and the channel output according the transformation of the input under @xmath152 ( in @xcite p. 2081 ,
moreover it is easily seen using fubini s theorem from measure theory and morera s theorem from the complex analysis in several variables ( cf .
@xcite ) that this extension of the function @xmath153 is holomorphic .
but , unfortunately , it is _ not _ true that the identity theorem ( also known as the uniqueness theorem ) holds for open sets in @xmath154 as the following standard example shows : + _ example .
_ we consider the simplest non - trivial case @xmath155 .
let @xmath156 denote the standard basis of @xmath155 and let @xmath157 be defined as @xmath158 where @xmath159 denotes the transpose and @xmath160 are the coordinates of @xmath161 with respect to the basis @xmath156 .
clearly , @xmath36 is holomorphic and the set of zeros of @xmath36 is @xmath162 in what follows we identify @xmath163 with @xmath164 .
@xmath164 is , by definition , open in the natural topology on @xmath164 ( but it is _ not _ open in the natural topology of @xmath155 , it is a closed linear subspace of @xmath155 ) , and the function @xmath36 is , apparently , not identically zero on @xmath155 . + note that this example with the identical arguments shows also that the conclusion of the identity theorem is not valid for open balls , say , in @xmath165 . if @xmath166 is any open ball in @xmath164 then @xmath167 for all @xmath168 but , again , @xmath169 on @xmath155 .
the reason is , as before , that an open ball in @xmath164 ( with the natural topology of @xmath164 ) is not open in the topology of @xmath155 .
+ this last example shows that the proof of proposition 4.3 in @xcite is not correct , since it assumes the validity of the identity theorem in exactly this setting .
it is this proposition 4.3 in @xcite which would allow us to conclude that the support of the capacity achieving input measure contains no open sets ( in @xmath170 ) provided we know that this support is bounded .
+ actually , the authors of this paper are convinced that we need different mathematical techniques to tackle the problem of characterization of the optimal inputs for multiple antenna rayleigh fading systems not relying on the identity theorem .
one reason for this opinion is the fact that the characterization of sets for which the identity theorem holds ( so called sets of uniqueness ) in the setting of several complex variables is a long standing challenging open problem in complex analysis .
we have shown that for a rayleigh fading channel with coherence time @xmath0 the support of the capacity achieving input measure is bounded .
our method of proof does not allow to extend the results to the case @xmath1 .
in fact the techniques we have used have to be substantially sharpened and supplemented by additional new tools .
furthermore we have shown that the approach based on the application of the identity theorem from the complex analysis in several variables is not admissible .
therefore , it seems highly likely for us that the techniques needed should be `` real - analytic '' in spirit .
this work is supported by the deutsche forschungsgemeinschaft dfg via project bo 1734/16 - 1 entwurf von geometrisch - algebraischen und analytischen methoden zur optimierung von mimo kommunikationssystemen .
chan , s. hranilovic and f.r .
kschischang , `` capacity - achieving probability measure for conditionally gaussian channels with bounded inputs '' , _ ieee trans .
inform . theory _ , vol .
51(6 ) , pp . 2073 - 2088 , june 2005 m. fozunbal , s.w .
mclaughlin , r.w .
schafer , `` capacity analysis for continuous - alphabet channels with side information , part i : a general framework '' , _ ieee trans . inform .
51(9 ) , pp . 3075 - 3084 , sept .
2005 m. fozunbal , s.w .
mclaughlin , r.w .
schafer , `` capacity analysis for continuous - alphabet channels with side information , part ii : mimo channels '' , _ ieee trans .
inform . theory _
51(9 ) , pp . 3086 - 3101 , sept . | we consider transmission over a wireless multiple antenna communication system operating in a rayleigh flat fading environment with no channel state information at the receiver and the transmitter with coherence time @xmath0 .
we show that , subject to the average power constraint , the support of the capacity achieving input distribution is bounded .
moreover , we show by a simple example concerning the identity theorem ( or uniqueness theorem ) from the complex analysis in several variables that some of the existing results in the field are not rigorous . |
community detection in large graphs is getting attention as an important application of social network analysis ( sna ) , the ability to detect closely knit communities opens several applications from targeting ads to recommender systems . in this work
we try to derive a very simple and efficient algorithm for community detection based on a size parameter .
being able to specify the minimum and maximum size of communities to detect can be a critical factor in the sna area , some networks tend to form very small and dense communities while other networks form larger groups .
the first section of this report discusses some existing algorithms for community detection in social graphs , then we introduce the idea behind the entropy walker and present our algorithm .
the final sections show some examples of the algorithm being used in some toy examples and analyzes the scaling of the method for large graphs .
several algorithms have been developed for community detection in large graphs .
clutsering methods based in k - means need to know in advance the number of communities to find in the network . in practice
this is not possible as the number of communities is usually unknown and furthermore due to social interactions the number of communities in a network might change over time making it very hard to set up as a parameter .
the modularity optimization algorithm [ b08 ] automatically detects the number of communities but it does nt allow for overlapping communities .
this is also inpractical for social networks as most nodes will be members of several different social circles .
bigclam [ lesk13 ] is a fast algorithm to detect overlaping communities , it s based in non - negative matrix factorization but it needs to know the number of communities to detect , as mentioned before this is an important limitation .
[ mca13 ] presents an algorithm to find social circles in networks but is based on node parameters `` features '' , we would like to perform the extraction of communities based in network structure only .
the idea of random walks being used to detect communities is also used in the mcl algorithm [ vdon99 ] however mcl ca nt control the size of the communities being detected and it needs to perform operations on the complete matrix of the graph limiting its use to small and medium sized networks .
we define a `` tour '' as a random walk of length `` s '' .
the basic idea of the algorithm is to perform several tours starting from random nodes and to detect communities based on the result of those tours .
`` s '' should be longer than the minimum number of members that we want for a community and it serves as an upper bound for the maximum number of members in a community .
it is likely for a random walker to get `` trapped '' inside nodes of a community , going back and forth between them because there are more inter - community edges than edges that will take the walker outside of the community . even if the random walker goes outside the community chances are it might come back . the algorithm will filter the random walks that are nt likely to have found a community calculating the entropy of the tour [ sha48 ] . tours with high entropy are unlikely to contain a community because they visit mostly different nodes .
they are probably paths or bridges between communities and might be of interest for some other applications .
the entropy is computed using the very popular shannon formula : @xmath0 where @xmath1 is just the probability of the node in the tour , in other words its frequency in the tour over the sum of all node frequencies .
a threshold parameter establishes the maximum entropy for a tour to be accepted as a fraction of the maximum possible entropy that can be computed assuming a random walk that never visits the same node more than once .
we call this parameter @xmath2 for entropy threshold . when @xmath2 is 1 all the tours are accepted , lowering @xmath2 increases the amount of rejected tours .
the graph in figure 1 shows the percentage of accepted tours for different values of @xmath2 using the food network as an example . the @xmath2 parameter can be tuned based on two different goals .
one possibility is to use it to limit the total number of tours to store in memory for very large graphs , a second use , more logical , is to set how dense a community has to be to be considered .
this second use that is data dependant is probably the recommended one .
this is an example of a very low entropy tour from the food network : @xmath3 $ ] and this is an example of a high entropy tour from the same network : @xmath4 $ ] we can see how the first tour can be converted in a community with the top ingredients being used for the same kind of dishes , the second tour has a wide array of ingredients and ca nt be considered a community .
maybe a bridge between different communities . as we have mentioned extracting the high entropy tours from a network
might also be an interesting application . after accepting or rejecting a tour based on its entropy
the algorithm will try to see if this tour is new or if it is similar to an already seen tour .
locality sensitive hashing ( lsh ) can be used to make similar tours hash to the same bucket avoiding the need to compare new tours with the existing ones .
if lsh maps the tour to a bucket where a tour is already stored then both tours are merged adding the frequencies of the nodes present in both tours .
this greatly reduces the number of tours that need to be stored in memory and avoids the problem of two very similar tours being detected as different communities . in some applications
the @xmath5 most frequent nodes in a tour can be used as the key to a hash function to determine the bucket number for the node .
this is a simplification of lsh using only one minhash computed from the most frequent nodes in a tour .
when this is not possible or does nt work standard lsh can be used .
now we describe the parameters used in the algorithm : the algorithm uses several parameters to fine - tune its behaviour : 0.25 cm the algorithm will perform @xmath6 random tours and check the entropy of each tour . if the tour entropy is below the @xmath2 threshold then the tour will be stored in a hash table along with a counter merging the tour with the already existing one if the bucket is not empty .
it s easy to notice that this process can be parallelized and that several million tours can be performed efficiently .
the memory cost to store the tours depends on the algorithm parameters . when the @xmath2 ( entropy threshold ) parameter is low the algorithm with detect only a few very dense communities and tours with frequency 1 can be considered a community .
when the @xmath2 parameter is higher the algorithm will check many tours and it might make sense to discard the tours with lower frequencies keeping the ones that have been repetedly matched .
it is known that montecarlo random walks can be used to compute pagerank and/or eigenvector centrality , the procedure used to detect communities can be used to compute at the same time a centrality score for the network nodes .
so the first conclusion is that node centrality can be computed at the same time as the community detection algorithm runs , just adding 1 to a counter every time a node is visited by a tour and then normalizing the cummulative score .
the effect of entropy filter is show in fig2 .
we can see that some nodes produce peaks for entropy thresholds below 1.00 , this means that the centrality of those nodes is higher in the entropy filtered sets compared to the plain random walks without filtering .
these peaks can be detected computing the delta between the eigenvector centrality and the tour computed centrality . from these peaks
we can detect nodes that are both central to the network and to the small communities where they belong , this gives an index of in - community centrality .
testing the procedure on the facebook ego network the peaks matched nodes that had a high degree of connections with the members in their communities .
something interesting to notice is that the algorithm can be run starting always from the same node , in the style of a personalized pagerank , when that happens we get as a result the social circles of a given user .
this is in some way similar to the algorithm used by twitter to recommend users to follow[gup12 ] the difference is that instead of computing a score for each node we compute scores for each random walk ( tour ) performed by the simulation . for example we can run the algorithm from the tomato ingredient to see what goes well with tomato : instantaneous delicious recipes !
this section presents some analysis and graphs about the behaviour of the algorithm .
it is interesting to analyze the number of tours that the algorithm will keep in memory as the network grows larger for a constant fixed entropy threshold .
we found that the number of tours analyzed does not grow as the size of the network and is strongly dependant on network structure . with only a few nodes small communities
are common in a graph with high clustering , as the network grows larger the number of small communities quickly goes down .
this can be explained because a random walker has now more options and is less likely to get trapped inside a community .
then after more nodes are added a threshold is passed and small communities emerge again .
this curious behaviour in the formation of small communities as the network grows larger resulted an interesting find and can be useful to refine generic models for network growth .
the emergence of small communities in large networks is strongly related to the clustering coefficient of the network . when the clustering coefficient is very los there are not enough edges to form dense communities so small communities will not form in random networks . in the same way
if the clustering coefficient is too high then the random walker can visit almost any node from any node and thus will not get trapped inside a small community , the whole network is the only existing community .
the following graph shows the number of tours detected for a fixed entropy threshold depending on the clustering coefficient of networks synthetically generated using the barabasi - albert model[bar99 ] .
as the clustering coefficient gets larger the number of nodes in a tour has to be increased to detect communities .
in our example we run the algorithm against the eastern food network composed by different ingredients using in the eastern cuisine .
the idea is that the algorithm should be able to find groups of ingredients that are frequently used together .
using @xmath2 at 0.75 and simulating 150.000 tours of 30 hops the algorithm processed a total of 8308 tours to find clusters with 5 to 10 nodes in less than 5 seconds and these were the top results .
the number between parentheses reflects the number of times the same community was detected , so the higher the number the stronger the community .
we can see that the algorithm quickly detects the ingredients for most deserts or breakfast - type preparations .
in total the algorithm detected 141 overlapping communities .
the following result looks like a good recipe to try : as a point of comparision we run the modularity optimization algorithm [ blon08 ] as implemented in gephi and got the following communities : [ lemon , egg , orange , almond , orangejuice , cream , raisin , cinnamon , honey , butter , milk , vanilla , walnut ] [ coriander , pepper , blackpepper , chicken , thyme , cayenne , cilantro , dill , cumin , bellpepper , chickenbroth , ginger , turmeric , carrot ] [ garlic , parsley , onion , lemonjuice , beef , lamb , tomato , cucumber , bread , oliveoil , mint , vinegar , yogurt , potato ] as we can see the modularity algorithm does a very good job but it lists all the ingredients that are similar together and is not very helpful to detect smaller groups that go very well together , for example communities of 3 or 4 ingredients . the algorithm presented here would create the following top 10 communities of 3 ingredients : the graph of the communities found by gephi looks like this [ figure2 ] as we can see the results help to create new recipes starting with ingredients that go well together frequently .
something interesting is that by allowing overlapping communities we can see that some ingredients are partially in different groups .
for example ginger is used for both savory and deserts .
the modularity algorithm is forced to choose only one cluster for ginger but in our algorithm we can find it in different communities .
we also run the algorithm in a very large dump of a social network with a total of about 5 million nodes .
the algorithm runs in constant time regardless of the size of the graph as it always simulates a constant number of random walks , the only difference in runtime is due to the time needed to access the adjacency list of each node and that is independant of the clustering algorithm .
besides the runtime analysis we weree curious to investigate what kind of small communities the algorithm would find in a large social network .
we run a modularity clustering phase first and then the entropy walker algorithm . after running the entropy walker algorithm we found that 100 we see that the entropy walker algorithm finds small dense communities inside the big communities created by the modularity algorithm .
figure 4 shows an accepted random walk inside a modularity class .
figure 5 shows the shape of one of the accepted random walk , we can see the community is actually a clique so the algorithm is finding cliques or structures similar to cliques for the parametrized size of components that depend on the length of the random walks .
in a streaming model the graph is constantly updated via the addition and deletion of nodes and edges . in this model
the algorithm can be kept running continuously producing `` infinite '' tours .
as the graph is updated communities that were previously detected might disappear and new communities can emerge .
an algorithm like the count - min sketch [ mutu05 ] can be used to keep in memory a list of only the top @xmath5 communities discovered so far . if a new very tight community forms it will be eventually found by the algorithm several times entering the top @xmath5 ranking . besides keeping the top @xmath5 communities
the streaming model can be used to detect communities that pass the entropy filter and the count - min sketch can be used to only list those communities that have repeated a number of times .
several strategies to prune old communities from memory can be used .
the entropy walker is a very simple algorithm , the core is just a montecarlo simulation of random walks in a graph .
the algorithm uses two very simple tricks to be able to compute communities from these random walks , first it is able to keep or discard a tour by calculating its entropy reasoning that a tour that gets trapped inside a community will visit several times the same nodes resulting in a low - entropy tour .
the second trick is the use of lsh and the ability to merge similar tours into a single one to reduce memory consumption and be able to detect the same community even if the nodes have been visited in different order and with different frequencies .
the algorithm can run very quickly consuming very little memory even for massive graphs , it can be kept running continusly in a streamming model where the graph is constantly updated , this setup is perfect for the anlysis of large social networks . | this report presents a very simple algorithm for overlaping community - detection in large graphs under constraints such as the minimum and maximum number of members allowed .
the algorithm is based on the simulation of random walks and measures the entropy of each random walk to detect the discovery of a community . |
cooper pair can carry finite total momentum . just after the celebrated bardeen - cooper - schrieffer ( bcs ) theory of superconductivity @xcite , in which the cooper pair is formed by two fermions with opposite momenta , fulde and ferrell ( ff ) predicted that finite momentum pairing may occur in some type - ii superconductors at strong magnetic field @xcite , where the order parameter @xmath0 , with @xmath1 the total momentum of the cooper pair and @xmath2 a constant .
it corresponds to pairing between @xmath3 and @xmath4 in the momentum space . a similar state , which may have lower energy than the ff phase ,
was predicted by larkin and ovchinnikov ( lo ) @xcite , with @xmath5 .
the fflo state admits the coexistence of magnetism and superconductivity and is a key concept to understand the superconducting behavior in some type - ii superconductors , e.g. , layered @xcite , heavy - fermion @xcite and organic @xcite superconductors . in the past five decades ,
great endeavors have been paid trying to unveil this novel phase , unfortunately , only indirect evidences related to the possible finite momentum pairing have been reported .
the basic reason is that the reported experimental tools , including andreev reflection ( ar ) @xcite , specific heat @xcite , nuclear magnetic resonance @xcite and ultrasound velocities @xcite , only measure some anomalous properties of the superconductors , which may be caused by other phase transitions @xcite . directly detecting the cooper pair momentum , without doubt ,
provides the most convincing evidence for finite momentum pairing . in this paper
, we propose an andreev interferometer based on a branched y - junction for this particular purpose . in our interferometer
, the subgap conductance oscillation is uniquely determined by the phase modulation of the order parameter , thus provides a distinctive method to detect the cooper pair momentum in ff superconductors .
the device is very robust because all the uncontrollable phases during the multiple scatterings at the y - junction are exactly canceled out .
this interferometer may have intriguing applications in non - centrosymmetric superconductors @xcite , including li@xmath6(pd@xmath7pt@xmath8)@xmath9b @xcite , cept@xmath9si @xcite , cerh(ir)si@xmath9 @xcite , cecoin@xmath10/ybcoin@xmath10 superlattice @xcite and srtio@xmath9/laalo@xmath9 interface @xcite _ etc._. generally , the interplay between the spin - orbit coupling ( induced by either the bulk or the structure inversion symmetry breaking ) and the in - plane zeeman field not only stabilizes the ff phase against the formation of the lo phase @xcite , but also greatly enlarges its phase size in the parameter space @xcite . in this sense ,
our interferometer is best suitable to facilitate the identification of the possible finite momentum pairing in these materials .
-direction is sketched by the color bar .
the electron and the andreev reflected hole are sketched by the filled and open circles , with their propagations represented by the solid and dashed lines , respectively .
( b ) analogy between an andreev interferometer and an aharonov - bohm interferometer .
( c ) schematics of the dispersion of the ff superconductor and the scattering processes at the ns interfaces .
the quasiparticle excitation spectra near the fermi wave vectors ( @xmath11 ) are shifted by the values of @xmath12 .
processes a , b , c and d denote the ar , the normal reflection , and the electron - like and hole - like quasiparticle transmissions , respectively.,scaledwidth=80.0% ]
the periodic phase modulation in the order parameter is essential to construct the andreev interferometer , as shown in fig .
[ fig1](a ) .
we consider a nanowire y - junction in proximity to a two dimensional ff superconductor , which introduces a finite pair potential in the two channels . to be specific , the two channels are set along the @xmath13-direction , with a spatial separation of @xmath14 in the @xmath15-direction .
the pairing in the ff superconductor occurs with a finite momentum @xmath16 = @xmath17 , and the order parameter takes the form of @xmath18 for the @xmath19-wave pairing . in this work
, @xmath1 is assumed to be changed _ in situ _ by an external magnetic field .
the proposed interferometer still works with the unconventional @xmath20-wave pairing .
this is because the projection of the order parameter of the ff superconductor into one dimensional channels eliminates the effect of the internal phase difference in @xmath20-wave superconductors . in other words ,
the underlying mechanism for ff pairing is not essential in our interferometer .
the basic idea is as follows .
an electron injected from channel @xmath21 of the y - junction is split into channels @xmath22 and @xmath23 . for a transparent normal metal - superconductor ( ns )
interface , the incident electron penetrates into the superconductor and forms a cooper pair with another electron , leaving back an andreev reflected hole , which acquires an extra phase equal to the macroscopic phase of the superconductor @xcite .
thus , the phase difference @xmath24 between the two channels due to the phase modulation in the ff superconductor can be registered accurately by the reflected hole , leading to the interference when the hole paths finally get combined at the left channel @xmath21 . as a result
, a total ar probability proportional to @xmath25 is expected for a symmetric y - junction .
it is interesting to note that the picture is resemblant to the well - known aharonov - bohm interferometer , if we sketch the hole paths as a mirror symmetry of the electron paths about the ns interfaces in fig . [
fig1](b ) . in the practical case , channels 1 and 2 may have different lengths , there are
multiple scatterings at the y - junction , and barriers may exist at the ns interfaces .
these effects may complicate the interference between the two channels , however , as shown below , the basic idea and qualitative results still hold .
in order to find the total ar amplitude in channel 0 , we first need to find the scattering coefficients at the ns interfaces and then combine them with the scattering matrix at the y - junction . in fig .
[ fig1](a ) , the y - junction is located at @xmath26 and the ns interface at @xmath27 . the nanowire can be described by the bogoliubov - de gennes ( bdg ) equation , @xmath28 where @xmath29 defines the free electron in channel @xmath30 ( @xmath31 ) with a chemical potential @xmath32 and a barrier @xmath33 at the ns interface .
the effects due to the land @xmath34-factor and the spin - orbit coupling are assumed to be negligible in the nanowires , so that the bdg equation is written in a @xmath35 form with the external magnetic field being absent , which although is necessary for the ff superconductor . due to the proximity effect with an ff superconductor ,
a finite pair potential is induced in the nanowire , which is represented by the step function @xmath36 with the phase @xmath37 and @xmath38 being the vertical ordinate of the channel @xmath30 .
a standard green function approach shows that , although the induced pairing strength @xmath2 may be smaller than that in the bulk , the phase modulation of the order parameter in the ff superconductor is always accurately registered by the induced pair potential in the nanowire ( see appendix 1 for details ) .
the bdg equation ( [ bdg ] ) can be solved by a galilean transformation @xmath39 and @xmath40 , and the quasiparticle excitation spectra around the fermi wave vectors ( @xmath11 ) are given by @xmath41 where @xmath42 is a small wave vector measured from @xmath11 , @xmath43 is the fermi velocity , and the conditions @xmath44 and @xmath45 have already been taken into account .
the quasiparticle energy around @xmath11 is shifted by a value of @xmath46 due to the cooper pair momentum , which is sketched in fig . [ fig1](c ) .
we find that the @xmath13-component of the cooper pair momentum results in an energy split at @xmath11 , while the @xmath15-component contributes to a phase difference @xmath47 between the two channels .
the scattering amplitudes at the ns interface can be obtained by the blonder - tinkham - klapwijk ( btk ) approach @xcite . by matching the wave functions at the ns interface ,
the ar amplitude for an incident electron is obtained as @xmath48 , where @xmath49 , @xmath50 is the dimensionless barrier strength , and @xmath51/2}$ ] and @xmath52 are the electron and hole components of the wave functions around @xmath11 , respectively . the ar amplitude for an incident hole
is denoted as @xmath53 and can be obtained similarly ( see appendix 2 for details ) . since the incident electron and
the andreev reflected hole move with the same momentum but opposite directions , the phases accumulated by them during the propagation are canceled out exactly .
consequently , the ar amplitudes @xmath54 and @xmath53 only depend on the macroscopic phase of the superconductor ( @xmath55 ) and have nothing to do with the channel lengths .
the phase difference @xmath56 is defined by the cooper pair momentum @xmath57 .
a symmetric y - junction is generally favorable to observe the strongest interference effect .
such that the scattering amplitudes for the electron at the y - junction can be fully parameterized as follows @xcite : @xmath58 for the transmission between channel 0 and channel @xmath30 , @xmath59 for the transmission between channel 1 and channel 2 , and @xmath60 for the reflection within channel @xmath30 . since there is no ar process at the y - junction
, the energy dependence of @xmath58 , @xmath59 , and @xmath60 in scale of @xmath2 can be neglected .
the scattering amplitudes for the hole are just the complex conjugate of those for the electron due to the particle - hole symmetry of the bdg equation .
when a bias voltage @xmath61 is utilized in channel 0 and the superconductor is grounded , the incident electron is multiply scattered between the y - junction and the ns interfaces before it finally goes back into channel 0 as an electron or a hole .
the total ar amplitude @xmath62 in channel 0 can be obtained by combining the scattering amplitudes at the ns interfaces and those at the y - junction in a standard scattering matrix approach ( see appendix 3 for details ) .
we note that , since the ns interface here is actually consisted of the same nanowire and the ar signal is strongest when the ns interface is transparent , the limiting case of @xmath63 is particularly important and practically useful . in this simple case
, the combination of the scattering coefficients can be fulfilled in a physically transparent way by a feynman path summation .
the basic scattering loops between the y - junction and the ns interfaces are sketched in fig .
[ fig2 ] , where @xmath64 represents the scattering amplitude for a backward hole ( moving towards @xmath26 ) in channel @xmath65 scattered into a backward hole in channel @xmath30 . from fig .
[ fig2 ] , we obtain @xmath66 and @xmath67 . similarly , @xmath68 and @xmath69 .
actually , the hole can have multiple scatterings which are the repeats of the basic loops .
the total scattering amplitude @xmath70 , representing an andreev reflected hole in channel @xmath30 scattered back into channel 0 , should satisfy the self - consistent equations @xmath71 the total ar amplitude can be written as @xmath72 , where @xmath73 is contributed by the initial electron transmission and the final hole transmission across the y - junction . with the help of the self - consistent equation ( [ selfr ] ) , we find @xmath74/(1+\lambda),\ ] ] where @xmath75 . in eq .
( [ ra ] ) , the condition @xmath76 is utilized , which is obliged by the unitarity of the scattering matrix at the y - junction .
interestingly , we find that , although @xmath58 , @xmath59 , and @xmath60 are all complex numbers , their phases are precisely canceled out and @xmath62 is determined only by the macroscopic phases of the superconductor .
experimentally , the ar can be measured by the differential conductance at low temperature .
utilizing the btk formula @xcite , the subgap conductance ( @xmath77 ) is twice as much as the ar probability , and is given by @xmath78 ^ 2},\ ] ] where @xmath79 is the unit conductance , @xmath80 , and @xmath81 $ ] .
( [ ra ] ) is the major finding of this work .
the factor @xmath25 in the numerator of eq .
( [ ra ] ) signifies the interference oscillation of the conductance , from which the cooper pair momentum in ff superconductors can be directly resolved .
the scattering paths for electron and hole are always in pairs ( see fig . [ fig2 ] ) , therefore all the uncontrollable phases during the scattering at the y - junction can not affect the conductance , guaranteed by the particle - hole symmetry .
this point marks the major advantage of our andreev interferometer . when @xmath82 ( @xmath83 ) and @xmath84 ( no transmission between the two channels ) , the conductance is reduced to @xmath85 , which is the standard result for an aharonov - bohm interferometer , as expected from our basic idea in fig .
1(b ) . .
the red solid , blue dotted and black dashed lines are corresponding to the barrier strengths of @xmath86 , @xmath87 and @xmath88 , @xmath89 , respectively .
see text for values of other parameters , scaledwidth=80.0% ] for the general ns interface with a finite barrier , the conductance can be obtained numerically by combining the scattering matrices at the y - junction and the ns interfaces in a standard way ( see appendix 3 for details ) . in order to present the numerical results ,
the cooper pair momentum is expressed in polar coordinates as @xmath90 , with @xmath91 being the superconducting coherence length .
the direction angle @xmath92 of the cooper pair momentum can be precisely determined by carefully tuning the direction of the external zeeman field @xcite . in the following ,
we adopt the parameters as @xmath93 , and @xmath94 , where @xmath95 is the length of channel @xmath30 in unit of @xmath96 , to illustrate our major results . the zero bias conductance @xmath97 as a function of @xmath92 is plotted in fig .
we find that the positions for the minimal conductance are irrelevant to the barriers , which always occur at @xmath98 .
for the symmetric ns interfaces ( @xmath99 ) the minimal conductance reaches zero , while for the unbalanced ns interfaces , it is lifted from zero .
given @xmath100 being two adjacent minimums , the magnitude of the cooper pair momentum is given by @xmath101 $ ] . and @xmath92 .
@xmath102 , and see text for values of other parameters.,scaledwidth=80.0% ] at finite bias , the contour plot of @xmath97 in @xmath61 and @xmath92 plane is shown in fig .
we find that the conductance oscillation with @xmath92 still exists .
moreover , the minimal conductance is insensitive to the bias when @xmath103 , which greatly relaxes the condition to determine the cooper pair momentum in the proposed interferometer .
in addition , there is a notable boundary in fig .
[ fig4 ] described by @xmath104 representing the conductance singularities , which is the evidence of the energy spectra split caused by @xmath105 .
it is worthwhile to compare the present andreev interferometer with the conventional ones @xcite . in a conventional interferometer
, the conductance oscillation is driven by a supercurrent @xcite or an external flux @xcite , while in our proposal , the conductance oscillates with the direction of the pairing momentum .
the momentum direction can be tuned by an in - plane zeeman field , which does not introduce any orbital effects or fluxes for a two dimensional ff superconductor . in the conventional interferometers ,
the zeeman splitting of the conductance is independent of the direction of the magnetic field , while in the present case , the conductance splitting is caused by @xmath105 and strongly depends on the direction angle @xmath92 of the in - plane zeeman field . as a result ,
the conductance structure in the present model is specific to the ff states and can be easily distinguished from that in conventional interferometers .
several important issues related to the experimental implementation of the interferometer should be discussed .
the diameter of the nanowires should be smaller than the superconducting coherence length in order to eliminate the superconducting phase fluctuations in the one dimensional channels . for superconductors with a large coherence length
( @xmath106 nm in cept@xmath9si @xcite ) , the y - junction may be routinely fabricated using the lithography method .
however , for superconductors with a small coherence length ( @xmath107 nm @xcite ) , the junction can be realized by carbon nanotubes or other metallic nanowires using the state - of - the - art chemical growing method @xcite . for such nanowires ,
the transverse modes have an energy spacing about 10 - 100 mev , therefore the electron and hole can transport within a single transverse mode in the energy scale of @xmath108 kelvin .
to observe at least one period of the subgap conductance oscillation , @xmath109 is required , which can be satisfied in the practical experiments @xcite .
the necessary proximity effects and the associated ar in nanowires have already been demonstrated experimentally by different groups@xcite . to ensure that the phase informations are not lost during the multiple scatterings , the nanowires need a large phase coherence length @xmath110 . in carbon nanotubes
@xmath111 250 nm @xcite and in other semiconductor or metallic nanowires with high mobilities , @xmath112 @xmath32 m , which is sufficient for our andreev interferometer .
we need to point out that the proposed interferometer can resolve the ff superconductors with even small pairing momentum , which may not be distinguished by the conventional measurements on their anomalous properties .
this interferometer opens the exciting perspectives to detect the cooper pair momentum in non - centrosymmetric superconductors , which not only can facilitate the unambiguous identification of finite momentum pairing in inhomogeneous superconductors , but also greatly deepen our understanding of fermionic pairing in modern physics .
we thank q. h. cui , s. raymond , a. akbari and j. kaczmarczyk for valuable communications .
w. c , r. s , and d. x are supported by 973 program ( grants no .
2011cb922100 , no .
2011cba00205 and no .
2013cb921804 ) , by nsfc ( grants no . 11074111 , no .
11174125 , and no .
11023002 ) , by papd of jiangsu higher education institutions , by ncet , and by the fundamental research funds for the central universities .
m. g is supported by hong kong rgc / grf projects ( no .
401011 and no . 2130352 ) and the chinese university of hong kong ( cuhk ) focused investments scheme .
we employ a tunneling model to calculate the effective pair potential in the nanowire induced by the two dimensional ff superconductor deposited above . the nanowire is set along @xmath13-direction , with @xmath15 being its lateral location .
the hamiltonian of the whole system is composed of three terms , @xmath113 , where @xmath114 and @xmath115 describe the hamiltonian in the ff superconductor and that in the nanowire , respectively , and @xmath116 describes the tunneling between them .
we have @xmath117 where @xmath118 and @xmath119 are electron operators for the ff superconductor and the nanowire , respectively , the single - particle energy @xmath120 and @xmath121 are both measured from the fermi level , @xmath2 is the pair potential in the superconductor and @xmath122 is the tunneling amplitude between the superconductor and the nanowire .
the energy dependence of @xmath122 is neglected in the energy scale of @xmath2 , which is much smaller than the fermi energy .
a clean contact between the superconductor and the nanowire ( without any scattering caused by impurities ) is assumed , so that the momentum component along the nanowire ( @xmath123 ) is conserved during the tunneling processes .
the pairing in the superconductor occurs with a finite cooper pair momentum @xmath16 .
the self - energy @xmath124 for the nanowire is given by @xmath125 where @xmath126 is the green function in the ff superconductor and @xmath127 represents the tunneling terms from the nanowire to the superconductor in @xmath116 .
given that the cooper pair momentum is much smaller than the fermi wave vector of the superconductor @xmath128 , the green function for the ff superconductor can be expressed in the nambu representation @xmath129 as @xmath130 where @xmath131 are the pauli matrices in the nambu space .
inserting eq .
( [ gf ] ) into eq .
( [ se ] ) , one obtains the self energy of the nanowire under the basis @xmath132 as @xmath133 two important conclusions can be drawn from eq .
( [ se2 ] ) : ( a ) the proximity effect leads to the finite momentum pairing in the nanowire with @xmath105 equal to that in the ff superconductor ;
( b ) the information of @xmath57 is registered by the phase @xmath134 , which defines the macroscopic phases of two channels of the proposed andreev interferometer .
the electron density in the nanowire is much lower than that in the superconductor so that the @xmath123-dependence in eq .
( [ se2 ] ) can be dropped .
the summation in eq .
( [ se2 ] ) can be transformed into a integral over energy @xmath135 .
the density of states @xmath136^{-1}$ ] is approximated by its value at the fermi energy @xmath137 , and one obtains , @xmath138,\ ] ] where @xmath139 , @xmath140 , and @xmath141 .
the green function of the nanowire is given by , @xmath142 where @xmath143 + ( 1-\mathcal{z})\delta(\cos\beta\tau_{x}-\sin\beta\tau_{y})$ ] , @xmath144^{-1}$ ] , and @xmath145 is the hamiltonian @xmath115 in eq .
( [ smh ] ) written in the nambu space .
since the green function is given under the basis of @xmath132 , after a fourier transformation , one finds that the effective pair potential in the nanowire is @xmath146 , with the phase modulation being the same as that in the bulk ff superconductor and the strength renormalized by the factor @xmath147 . under the strong coupling limit @xmath148 , we have @xmath149 . the normal part of the hamiltonian of the nanowire is also modified by the factor @xmath150 , indicating a renormalization of the fermi velocity .
although such a renormalization may result in a velocity mismatch at the ns interface , it can be compensated by utilizing a gate voltage or simulated by a barrier potential .
thus , the effective hamiltonian in the spatial space for the nanowire can be written as @xmath151\ ] ] with @xmath152 and @xmath153 defined in the main text . from eq .
( [ sheff ] ) , the bdg equation in the main text is restored .
after the transformation @xmath154 and @xmath155 , the bdg equation is reduced to @xmath156 where @xmath157 .
the eigenvalues of the bdg equation are found as @xmath158 under the general condition @xmath44 , the dispersion eq .
( 2 ) in the main text for the particles near the fermi level ( @xmath159 ) is obtained . for an electron incident from the normal region with an energy @xmath160
, the wave functions take the form @xmath161 where @xmath51/2}$ ] , @xmath52 , phase @xmath162 , and all the wave vectors are approximated by @xmath163 .
all the scattering states in eq .
( [ state ] ) are illustrated in fig .
1(c ) in the main text . the amplitudes @xmath164 and @xmath165 denote the amplitudes of the ar , the normal reflection , and the electron - like and hole - like quasiparticle transmissions , respectively . utilizing the boundary conditions @xmath166 and @xmath167 at @xmath27 , the ar and the normal reflection amplitudes are obtained as @xmath168 where @xmath49
, @xmath95 is the length of channel @xmath30 in unit of @xmath96 , and @xmath50 is the dimensionless barrier strength . similarly , the scattering amplitudes for a incident hole are obtained as @xmath169
it is convenient to write all the scattering coefficients at the ns interfaces into a reflection matrix as @xmath170 , with @xmath171 .
the scattering matrix at a symmetric y - junction takes the form @xmath172 where the submatrices @xmath173 describe the quasiparticle transmission between chanel 0 and channel @xmath30 , @xmath174 represents the reflection within channel 0 , and @xmath175 represents the reflection within channel @xmath30 and the transmission between channel 1 and channel 2 .
the scattering matrix @xmath176 is written in the particle - hole space , however , all the scattering amplitudes representing the electron - hole conversion are zero at the y - junction .
the scattering amplitudes for electrons are given as @xmath60 for the reflection within channel @xmath30 , @xmath59 for the transmission between channel 1 and channel 2 , and @xmath58 for the transmission between channel 0 and channel @xmath30 , respectively .
the scattering amplitudes for holes are just the complex conjugate of those for electrons , due to the particle - hole symmetry of the bdg equation . due to the time reversal symmetry of the y - junction
, the scattering matrix @xmath176 is symmetric .
then , the scattering matrix for the whole andreev interferometer can be obtained by combining @xmath176 and @xmath177 .
the unitarity of @xmath176 gives the restrictions @xmath178 , @xmath179 , and @xmath76 , which are useful to simplify the matrix combination .
the composite scattering matrix , describing an electron or a hole in channel 0 is reflected back as an electron or a hole , is given by @xmath180 specifically , the total ar amplitude in channel 0 for a incident electron is @xmath181 and the normal reflection amplitude is @xmath182 .
the dimensionless conductance is obtained as @xmath183
99 bardeen j , cooper l n and schrieffer j r 1957 _ phys .
rev . _ * 108 * 1175 fulde p and ferrell r a 1964 _ phys . rev . _ * 135 * a550 larkin a i and ovchinnikov yu n 1965 _ sov . phys .
* 20 * 762 croitoru m d , houzet m and buzdin a i 2012 _ phys .
lett . _ * 108 * 207005 radovan h a , fortune n a , murphy t p , hannahs s t , palm e c , tozer s w and hall d 2003 _ nature _ ( london ) * 425 * 51 matsuda y and shimahara h 2007 _ j. phys . soc . jpn . _ * 76 * 051005 gloos k , modler r , schimanski h , bredl c d , geibel c , steglich f , buzdin a i , sato n and komatsubara t 1993 _ phys .
lett . _ * 70 * 501 bianchi a , movshovich r , capan c , pagliuso
p g and sarrao j l 2003 _ phys .
lett . _ * 91 * 187004 kenzelmann m , strssle th , niedermayer c , sigrist m , padmanabhan b , zolliker m , bianchi a d , movshovich r , bauer e d , sarrao j l and thompson j d 2008 _ science _ * 321 * 1652 singleton j , symington j a , nam m - s , ardavan a , kurmoo m and day p 2000 _ j. phys . :
condens . matter _
* 12 * l641 lortz r , wang y , demuer a , bttger p h m , bergk b , zwicknagl g , nakazawa y and wosnitza j 2007 _ phys . rev . lett . _
* 99 * 187002 kontos t , aprili m , lesueur j and grison x 2001 _ phys .
lett . _ * 86 * 304 park w k , sarrao j l , thompson j d and greene l h 2008 _ phys . rev .
lett . _ * 100 * 177001 park w k and greene l h 2009 _ j. phys . : condens .
matter _ * 21 * 103203 krawiec m , gyrffy b l and annett j f 2004 _ phys .
b _ * 70 * 134519 bianchi a , movshovich r , oeschler n , gegenwart p , steglich f , thompson j d , pagliuso p g and sarrao j l 2002 _ phys . rev .
lett . _ * 89 * 137002 movshovich r , jaime m , thompson j d , petrovic c , fisk z , pagliuso p g and sarrao j l 2001 _ phys .
lett . _ * 86 * 5152 kakuyanagi k , saitoh m , kumagai k , takashima s , nohara m , takagi h and matsuda y 2005 _ phys . rev
. lett . _ * 94 * 047602 watanabe t , kasahara y , izawa k , sakakibara t , matsuda y , van der beek c j , hanaguri t , shishido h , settai r and onuki y 2004 _ phys .
b _ * 70 * 020506(r ) yang kun and agterberg d f 2000 _ phys . rev .
_ * 84 * 4970 yokoyama t , onari s and tanaka y 2008 _ j. phys .
jpn . _ * 77 * 064711 yuan h q , agterberg d f , hayashi n , badica p , vandervelde d , togano k , sigrist m and salamon m b 2006 _ phys .
lett . _ * 97 * 017006 lee k w and pickett w e 2005 _ phys .
b _ * 72 * 174505 frigeri p a , agterberg d f , koga a and sigrist m 2004 _ phys .
* 92 * 097001 tada y , kawakami n and fujimoto s 2008 _ phys .
lett . _ * 101 * 267006 goh s k , mizukami y , shishido h , watanabe d , yasumoto s , shimozawa m , yamashita m , terashima t , yanase y , shibauchi t , buzdin a i and matsuda y 2012 _ phys .
* 109 * 157006 michaeli k , potter a c and lee p a 2012 _ phys .
lett . _ * 108 * 117003 dimitrova o v and feigelman m v 2003 _ pisma zh
fiz . _ * 78 * 1132 [ 2003 _ jetp lett .
_ * 78 * 637 ] samokhin k v 2004 _ phys .
b _ * 70 * 104521 kaur r p , agterberg d f and sigrist m 2005 _ phys .
_ * 94 * 137002 agterberg d f and kaur r p 2007 _ phys . rev .
b _ * 75 * 064511 dimitrova o and feigelman m v 2007 _ phys . rev . b _ * 76 * 014522 zheng z , gong m , zou x , zhang c , guo g 2013 _ phys . rev
. a _ * 87 * 031602(r ) qu c , zheng z , gong m , xu y , mao l , zou x , guo g and zhang c 2013 _ nature communications _ * 4 * 2710 spivak b z and khmelnitskii d e 1982 _ pisma zh .
* 35 * 334 [ 2003 _ jetp lett . _
* 35 * 412 ] .
nakano h and takayanagi h 1993 _ phys .
b _ * 47 * 7986 blonder g e , tinkham m and klapwijk t m 1982 _ phys . rev .
* 4515 bttiker m , imry y and azbel m ya 1984 _ phys . rev .
a _ * 30 * 1982 lambert c and raimondi r 1998 _ j. phys .
condens . matter _ *
10 * 901 samuelsson p , lantz j , shumeiko v s and wendin g 2000 _ phys .
b _ * 62 * 1319 papadopoulos c , rakitin a , li j , vedeneev a s and xu j m 2000 _ phys .
_ * 85 * 3476 li j , papadopoulos c and xu j 1999 _ nature _ ( london ) * 402 * 253 andriotis a n , menon m , srivastava d and chernozatonskii l 2001 _ phys . rev . lett . _ * 87 * 066802 li w z , wen j g and ren z f 2001 _ appl .
lett . _ * 79 * 1879 satishkumar b c , thomas p j , govindaraj a and rao c n r 2000 _ appl .
lett . _ * 77 * 2530 meng g , jung y j , cao a , vajtai r and ajayan p m 2005 _ proc .
usa _ * 102 * 7074 mourik v , zuo k , frolov s m , plissard s r , bakkers e p a m and kouwenhoven l p 2012 _ sciece _ * 336 * 1003 doh y j , dam j a , roest a l , bakkers e p a m , kouwenhoven l p and franceschi s d 2005 _ science _ * 309 * 272 xiang j , vidan a , tinkham m , westervelt r m and lieber c m 2006 _ nat .
_ * 1 * 208 tsukagoshi k , alphenaar b w and ago h 1999 _ nature _ ( london ) * 401 * 572 | we propose an andreev interferometer , based on a branched y - junction , to detect the finite momentum pairing in fulde - ferrell ( ff ) superconductors . in this interferometer ,
the oscillation of subgap conductance is a unique function of phase difference between the two channels of the y - junction , which is determined by the phase modulation of the order parameter in the ff superconductors .
this interferometer has the potential not only to determine the magnitude but also the direction of the momentum of cooper pairs in the ff superconductor .
the possible applications of the interferometer in the identification of the finite momentum pairing in non - centrosymmetric superconductors are also discussed . |
ever since the advent of quantum chromodynamics ( qcd ) it has been popular to describe the nucleon in terms of bag or soliton models .
there are many versions of such models , characterized by two extremes : the mit bag model @xcite , where the nucleon consists of just constituent quarks arbitrarily restricted in a given volume , and the skyrme model @xcite , where there are no quarks and the nucleon is instead a topological soliton of the pion field .
other models interpolate between these two extremes , seeking to combine the obvious emphases on the structure of nucleons within the mit bag model and on nuclear interactions via meson exchange within the skyrme model . for example , chiral bag models @xcite surround an mit - like bag of quarks with a skyrme - like cloud of pions ( and perhaps other mesons as well ) . clearly , each such model attempts to balance between low - energy degrees of freedom the mesons and high - energy degrees of freedom the quarks ( and possibly gluons as well ) .
our aim here is to develop a model that as simply as possible , but without losing essential physics , combines quark and meson degrees of freedom .
this task is made more difficult by the `` cheshire cat principle '' @xcite , an extrapolation of results found from chiral bag models , which find that low - energy nucleon properties are largely insensitive to the size of the quark bag . as the bag shrinks , the meson cloud forms more and more of the nuclear structure . to select a model ,
we must study nuclear properties that distinguish between a bag of quarks and a cloud of mesons .
an obvious testing ground for any bag or soliton model is dense baryon matter , where the structure of individual nucleons will become as important as the interactions between neighboring nucleons .
preferably , the model will have a dynamical formation of the bag or soliton , for then one can treat the transition to a quark - gluon plasma consistently .
surprisingly enough , the skyrme model , which may be thought of as a chiral bag model with the quark bag shrunk to zero size , does predict a dynamical transition to a phase of solitons of fractional baryon number at high densities @xcite ; however , identifying this phase with a quark - gluon plasma is certainly problematic .
thus in this paper we study dense nuclear matter in a set of models known as non - topological soliton ( nts ) models .
in particular , we study the friedberg - lee ( fl ) model @xcite and a class of related models , the chiral chromodielectric ( @xmath1cd ) models @xcite , as well as extensions of these models ( or , one might argue , approximations to these models ) that explicitly include couplings to mesons .
these models , based upon general arguments from qcd , are characterized by the coupling of quarks to a scalar field @xmath0 that has a nonzero vacuum expectation field .
this field is understood to be a composite gluon field .
the interaction between the quarks and scalar field leads to a dynamical confinement mechanism , with the quarks carving a hole in the background scalar field .
the structure of this bag depends precisely on how the quarks couple to the scalar field , and it is this coupling that distinguishes the various models we study here .
all these models reproduce single nucleon properties reasonably well indeed , chiral bag models show that single nucleon properties are relatively insensitive to the structure of the quark bag .
we must look at high densities , where the bags begin to overlap , to see the differences between models with different quark-@xmath0 couplings . in the last few years there
also has been considerable interest in the application of the quark - meson coupling ( qmc ) model to study the nuclear matter equation of state , as well as medium effects on the nucleon structure and nucleon - meson coupling , motivated by the apparent success of the walecka qhd models @xcite in describing the properties of nuclear matter .
the qmc model , initially suggested by guichon @xcite , consists of non - overlapping nucleon ( mit ) bags that interact through the exchange of scalar and vector mesons in the mean - field approximation ( mfa ) .
this model has been generalized to include fermi motion and center of mass corrections @xcite and was applied to nuclear matter ( see @xcite and references therein ) and also to finite nuclei @xcite .
of course , the assumption that the nucleons can be regarded as non - overlapping bags is only valid at low density , where these models seem to capture the essential physics . already at nuclear saturation density , however , the internucleon separation is comparable to the nucleon radius , and at higher densities the assumption that the bags do not overlap clearly breaks down .
this assumption becomes even more questionable when a modification of the bag constant in nuclear matter is taken into account .
jin and jennings @xcite and mller and jennings @xcite have recently shown that the introduction of a density - dependent bag constant can reproduce the emc effect and reconcile the qmc results with those of the walecka qhd - i model @xcite .
as a result @xcite the bag radius grows with increasing density and the overlapping of the bags starts just above nuclear saturation density . to take into account the effects from overlapping nucleon bags
that is , to study a nuclear liquid rather a nuclear gas it is clearly of interest to introduce some dynamics in the confining mechanism .
thus we are led to replace the mit bag model by a nts model .
although the study of nuclear matter properties was begun long ago ( see @xcite and references therein ) and was carried out with different soliton models and with different levels of sophistication ( see @xcite and references therein ) , none of these calculations included the effect of background meson fields on nuclear matter . now , in principle , the non - topological soliton models in particular , the @xmath1cd model , which in its full form includes perturbative gluon exchange contain sea quarks and meson exchange explicitly .
however , in practice it is very difficult to deal with anything besides the nuclear constituent quarks . in the interest of simplicity
( and at the sacrifice of consistency ) , we extend our nts models to include explicitly meson degrees of freedom .
( such an extension has been referred to as the local uniform approximation to @xmath1cd in @xcite . ) in the following , then , we shall study a system of non - topological solitons interacting via the exchange of scalar and vector mesons within the mfa .
in modeling nuclear matter each soliton is centered in a spherical wigner - seitz ( ws ) cell .
it has been argued that the choice of a spherical ws cell is more appropriate for a fluid phase than a crystal as it represents an angular average over neighbouring sites .
we consider two further approximations : in one , the ws calculation is simply used to provide an effective nucleon mass and the kinetic energy is then taken to be that of a fermi gas , thus providing the correct low - density limit .
a second approximation uses a bloch - like boundary condition to calculate the band structure of the quark states .
neither of these approximations is quite satisfactory , and the second paper of this series is devoted to improving the modeling of a liquid of solitons .
nevertheless , we can still hope that the major qualitive features of dense nuclear matter will be reproduced even given the two approximations used here .
indeed , we find some encouraging results such as nuclear saturation and an increase of the proton rms with nuclear density . these effects are shown to derive from the background meson fields , which have a considerable effect on the formation of energy bands in nuclear matter . in this paper
we limit our attention to friedberg - lee soliton type models extended to include the meson fields .
the non - topological soliton models are presented in sec .
[ sec : models ] .
the wigner - seitz approximation is then discussed in sec .
[ sec : ws ] . in sec .
[ sec : trivial ] we study in some detail the trivial solution of the model . using these solutions
we show that the high density limit of the friedberg - lee type models depends on the leading power of the quark-@xmath0 coupling vertex .
the numerical results are presented in sec .
[ sec : results ]
our starting point for studying the high density behavior of soliton matter is the friedberg - lee @xcite non - topological soliton model , and we also consider related models like the chiral chromodielectric model of fai , perry and wilets @xcite . in their simplest versions
, these models include only constituent quarks and a single scalar field @xmath0 that couples to the quarks . for the light
quarks we shall assume @xmath2 .
extending these models in the spirit of the quark - meson coupling model , we introduce in addition two meson fields : namely , a scalar meson @xmath3 and a vector meson @xmath4 , which play important roles in quantum hadrodynamics .
we assume these mesons couple linearly to the quarks .
there is some freedom in the structure of the quark - meson vertex , in regard to its dependence on the soliton field @xmath0 . using the nielsen - patkos lagrangian @xcite ,
banerjee and tjon @xcite have recently argued that in nts models the quark - meson coupling should also depend on the scalar soliton field .
similar conclusions are reached by krein , _ et al .
however , we have found that this causes unwanted behavior within the mean field and wigner - seitz approximations used here ( see sec .
4 ) , and so we report calculations that use @xmath0-independent quark - meson couplings .
as we shall see , this choice has the advantage of reproducing the quantum hadrodynamics equation of state at low densities .
thus we take the lagrangian density to have the form @xmath5
\psi + \frac{1}{2}\partial_{\mu}\sigma\partial^{\mu}\sigma - u(\sigma ) \nonumber \\ & & \qquad + \frac{1}{2}\partial_{\mu}\phi\partial^{\mu}\phi - \frac{1}{2}m_{s}^2 \phi^2 - \frac{1}{4}f_{\mu\nu}f^{\mu\nu } + \frac{1}{2}m_{v}^2 v_{\mu}v^{\mu}\end{aligned}\ ] ] where the @xmath0 field self interaction is assumed to be @xmath6 the constants @xmath7 , @xmath8 and @xmath9 are fixed so that @xmath10 has a local minimum at @xmath11 ( inflection point if @xmath7=0 ) and a global minimum at @xmath12 , the vacuum value .
the mass of the glueball excitation associated with the @xmath0 field is given by @xmath13 .
the quark-@xmath0 coupling is taken to be @xmath14 } } & \mbox{for the $ \chi$cdm } \end{array } \right.\ ] ] where the chromodielectric function @xmath15 has the form @xmath16 \ ; \ ; ; \ ; x=\sigma/\sigma_v \ ; , \ ] ] in the following we will take @xmath17 .
in perturbative calculations that include gluons the dielectric function @xmath15 is regularized in order to handle infinities in the one gluon exchange diagrams associated with a vanishing dielectric constant @xcite .
such a regulation is also useful for our numerical work .
@xcite use the prescription @xmath18 where @xmath19 is a constant often simply fixed at @xmath20 .
however , this regulation forces @xmath21 , and we prefer instead to regulate as follows : @xmath22.\ ] ] we find that properties of isolated solitons are independent of @xmath19 for values even as large as .1 , but results become sensitive to @xmath19 at high densities . here
, we shall treat @xmath19 as an additional parameter .
the euler - lagrange equations corresponding to ( [ lagrange ] ) are given by @xmath23 \psi = 0\ ] ] @xmath24 @xmath25 @xmath26 where @xmath27 and @xmath28 .
we solve eqs .
( [ el1]-[el4 ] ) in the mean field approximation : we replace the soliton field @xmath0 by a _ c_-number @xmath29 and the meson fields by their expectation values in the nuclear medium @xmath30 and @xmath31 , with @xmath32 and @xmath33 constants .
the approximation that the scalar and vector mesons fields can be regarded as constants while the soliton field is to depend on spatial coordinates stems from the long range nature of the light mesons and the short range nature of the soliton field due to the large glueball mass .
in essence , the mesons are fast degrees of freedom and the glueball is slow , and we use a born - oppenheimer approximation .
the resulting equations for the quark and the scalar soliton field are , @xmath34 \psi_k = \epsilon_k \psi_k\ ] ] @xmath35 we consider only valence quarks in our calculations .
in order to use eqs .
( [ mf1]-[mf2 ] ) for the study of high density nuclear matter , it is sufficient to concentrate on a unit cell and solve these equations with the appropriate boundary conditions . in this paper
we shall assume that each unit cell contains a single nucleon , that is , our unit cell coincides with a wigner - seitz cell .
the single nucleon energy @xmath36 is a sum of two terms , the energy of the @xmath37 quarks and the energy carried by the scalar soliton field @xmath0 : @xmath38 \ ; .\ ] ] the quark energy @xmath39 should be regarded as the eigenvalue of eq .
( [ mf1 ] ) for isolated bags , or as the average energy in the band for dense nuclear matter .
it should be noted that @xmath36 can not be identified with the nucleon mass as it contains spurious center of mass motion @xcite . in order to correct the center of mass motion in the wigner - seitz cell the nucleon mass at rest
is taken to be @xmath40 where @xmath41 .
the notation @xmath42 stands for an average over the wigner - seitz cell .
thus @xmath43 is the expectation value of the quark momentum squared and @xmath44 is the scalar soliton momentum squared .
the latter average is obtained using a coherent state approximation @xcite , which is essentially a single mode correction to the classical soliton mass .
we find this latter correction to be small with respect to the total mass of the soliton and ignore it henceforth ( it is not clear how valid this single - mode approximation is : at higher density we find that this correction exceeds the energy of the @xmath0 field , with both going to zero ) .
at low density the band width vanishes and the quarks are confined in separate bags .
then we expect the following approximation to be most accurate : we assume the individual nucleons move around as a gas of fermions with effective mass @xmath45 given by eq .
( [ m_n ] ) , and so we get the following estimate for the total energy density at nuclear density @xmath46 : @xmath47 where @xmath48 is the spin - isospin degeneracy of the nucleons .
the fermi momentum of the nucleons is related to the baryon density through the relation @xmath49 the total energy per baryon is given by @xmath50 .
the constant scalar meson field @xmath32 is determined by the thermodynamic demand of minimizing @xmath51 , which gives @xmath52 the vector meson field is determined by averaging the euler - lagrange equation , eq .
( [ el4 ] ) , on a wigner - seitz cell yielding @xmath53 these equations are similar to those of quantum hadrodynamics , the difference here being that the nucleon now has structure and thus the meson fields couple to the nucleon through its quarks . at low density the nucleon mass approaches its free value , and our mean field equations reduce to those of quantum hadrodynamics . the approximation used here for the nuclear system has often been adopted in modeling soliton matter , and is generally known as the wigner - seitz approximation .
each soliton is enclosed in a sphere of radius @xmath54 such that @xmath55 . on a periodic lattice
the quark functions should satisfy bloch s theorem @xmath56 for the lattice vectors @xmath57 . concentrating on a single cell
the bloch theorem gives boundary conditions for the quark spinors in that cell .
although one can solve these boundary condition in a self consistent manner @xcite , we shall make the simplifying assumption of identifying the bottom of the lowest band by the demand that the derivative of the upper component of the dirac function disappears at @xmath54 , and the top of that band by the demand that the value of the upper component is zero at @xmath54 @xcite .
the quark spinor in the lowest band is assumed to be an _
s_-state @xmath58 and the resulting euler - lagrange equations for the spinor components are @xmath59 v_k = 0\ ] ] @xmath60 u_k = 0 \ ; .\ ] ] the corresponding equation , ( [ mf2 ] ) , for the soliton field assumes the form @xmath61 the quark density @xmath62 and the quark scalar density @xmath63 are given by @xmath64 \ ; , \ ] ] @xmath65 \ ; , \ ] ] where the band is filled up to @xmath66 .
the quark functions are normalized so that there are three quarks in the wigner - seitz cell .
the boundary conditions for the soliton field are @xmath67 .
the boundary conditions for the quark functions at the origin are given by @xmath68 and @xmath69 , where @xmath70 is determined by the normalization condition @xmath71 the boundary conditions at @xmath72 are given by @xmath73 for the bottom of the lowest band , and @xmath74 for the top of the band .
using these equations we can solve for the corresponding @xmath75 and @xmath76 .
we assume the tight - binding dispersion relation @xmath77 and that the band is filled right to the top @xmath78 .
the assumption of such a dilute filling has been made previously@xcite , and we do not discuss it further .
the quark functions corresponding to the energy @xmath79 can be simply obtained by integrating eqs .
( [ u_r ] ) and ( [ v_r ] ) for each intermediate value @xmath79 .
substituting the dispersion relation into eq .
( [ e_n ] ) , the nucleon energy is given by @xmath80 \ ; , \ ] ] which is then used in ( [ m_n ] ) to determine the equation of state ( [ eeos ] ) . at higher density , where the bags begin to overlap
, there is no reason to assume each quark is tightly bound to a single bag nor to impose that 3@xmath81 groups move collectively with a well - defined momentum .
our equation of state can not then be considered a good approximation at higher densities , and we must then find a different approximation for handling the kinetic energy .
this is the subject of the next paper in this series .
for now we are content with studying low density behavior , with a special interest in whether our approximations can still produce saturation at the expected density .
in fact , as one approaches the empirical nuclear saturation density , the different models begin to distinguish themselves .
this is discussed in detail in the next section .
the nts models we are considering have a uniform plasma phase that is preferred at high densities .
this corresponds to the solution @xmath11 , so that the soliton bags `` dissolve '' and the quarks are free .
for the original fl model , this solution is favored at unreasonably low densities . moreover , in the ws calculations @xcite , for cell radii below @xmath820.8 - 0.9 fm only a trivial constant solution can be found . for the more sophisticated analysis reported in @xcite ,
which includes higher partial waves in the quark wave functions and direct computation of the bands , the calculation breaks down already at a cell radius @xmath821.4 fm .
we view this as a fault of the model and not of the ws approximation , for regardless of whether our restricting the soliton to a single spherical cell represents a dense system well or not , our model of the nucleon should nevertheless allow the soliton to be squeezed to volumes well below nuclear density before it breaks apart . ( the saturation density of nuclear matter is @xmath83=0.17@xmath84 , which corresponds to a ws cell radius @xmath85=1.12fm . )
this leads us to look for improvements upon the original fl model , and the two particular extensions studied in this paper are generalizations of the quark - glueball coupling @xmath86 and the addition of explicit meson degrees of freedom . before proceeding to the numerical results ,
let us look at how these extensions affect the trivial solution .
the trivial solution to the ws equations ( [ u_r]-[sigma_r ] ) is @xmath87 , @xmath88 and @xmath89 , where the constants @xmath70 and @xmath90 are independent of both @xmath91 and @xmath92 .
strictly , for @xmath93 this solution does not satisfy our boundary conditions ; however , we find that below a given cell volume the numerical solution develops a singularity at the boundary in order to reproduce this preferred solution . moreover , the more accurate self - consistent calculations of @xcite for the fl model ( which do not assume just @xmath94-wave states in the quark wave function ) show that the width of the band also narrows sharply at the onset of the trivial solution , indicating that the @xmath92-dependence is not important . thus in order to find the trivial solution that characterizes the breaking down of the ws calculation , we may assume that all the quarks are at the bottom of the band .
this corresponds to squeezing an isolated soliton , and we insist that our model favors a nontrivial solution until the cell radius gets well below that corresponding to nuclear density . in the present section we shall consider altering the quark - meson couplings in our models as suggested in lagrangian @xcite , namely : @xmath95 this will allow us to investigate whether the coupling to meson fields can help cure the breaking down of the fl model at too low density .
for the trivial solution , then , the normalization condition ( [ norm ] ) gives @xmath96 from eq .
( [ v_r ] ) we find the quark eigenenergy is @xmath97,\ ] ] where @xmath90 obeys eq .
( [ sigma_r ] ) , modified to account for the change in @xmath81-@xmath0 coupling , which leads to @xmath98.\ ] ] solving for the mean field values of @xmath32 and @xmath33 , we have @xmath99 yielding finally ( for @xmath100=0 ) @xmath101\right\}\ ] ] and the total energy density @xmath102\right\ } + u(\sigma_0).\ ] ] this is clearly a spurious solution corresponding to putting all the quarks in the lowest level .
one needs to add the kinetic energy correctly to get the true energy of the quark plasma .
we would like to select a model for which the trivial solution is found only at densities much higher than nuclear density , and we would like the quark mass @xmath103 to be zero in the preferred phase .
this latter condition implies @xmath90=0 .
however , for the fl model @xmath104 , and therefore @xmath90=0 is not a solution of eq .
( [ sig_tr ] ) . using eqs .
( [ phi_tr ] ) and substituting for @xmath105 and @xmath106 , for the fl model eq .
( [ sig_tr ] ) becomes @xmath107.\ ] ] if we turn off the meson fields @xmath108 , we see that the solution to ( [ sig2_tr ] ) goes as @xmath109 at large densities .
not only does this blow up as @xmath110 , but it gives the quarks an unphysical negative mass @xmath111 .
on the other hand , if we turn on the meson fields and make the usual choice of parameters so that @xmath112 ( which is necessary for saturation ) , there is now a positive high density solution that goes as @xmath113 .
however , there is still an unphysical negative solution , now diverging as @xmath46 .
thus we do not expect the inclusion of the meson fields to cure the problem of the fl model .
now let us consider a modified fl model in which we take the quark - glueball coupling to be @xmath114 .
we shall call this the fl@xmath115 model . then instead of eq .
( [ sig2_tr ] ) we have @xmath116.\ ] ] for this model the trivial solution @xmath117 exists in the wigner - seitz approximation
. there are also nonzero solutions , however . with the meson fields switched
off there are two nontrivial solutions @xmath118 which exist only at densities @xmath119 . for the parameter choice used here
, this corresponds to cell radii @xmath120 fm .
these solutions give the quarks a positive mass and do not necessarily signal any problems with the model . for nonzero @xmath81-@xmath0 coupling ,
the solutions exist also at high density , behaving as @xmath121 .
the quark effective mass ( [ eps2_tr ] ) vanishes in this limit .
thus the fl@xmath115 does not seem to have the problems of the fl model when using the ws approximation .
finally , let us consider the @xmath1cd models . from eqs .
( [ g(s ) ] ) and ( [ kappa(s ) ] ) we see that for small @xmath0 the coupling @xmath86 is of leading order @xmath122 in @xmath0 .
this means that , regardless of whether the mesons are present , for the @xmath1cd model with @xmath123 there will be a solution to eq .
( [ sig_tr ] ) such that @xmath117 .
this corresponds to the desired free massless quark plasma phase , with vanishing meson field averages , since then @xmath124 .
furthermore , one can show that as @xmath110 the only other solution has @xmath125 and gives the quarks a positive mass that also vanishes in the high density limit .
thus we expect the @xmath1cd models to provide a more reasonable description of dense nuclear matter than does the fl model , for there is no unphysical trivial solution to the ws equations that will cause problems . indeed , the type of behavior predicted by our analysis of the trivial solution is reflected in our wigner - seitz calculations . as we show in fig .
1 , the @xmath1cd model exhibits wigner - seitz solutions down to very small cell radii ( we never reached a breakdown point , taking @xmath54 as low as @xmath126fm ) , whereas the fl model breaks down at the expected point @xmath127fm .
the fl@xmath115 model can be taken down to lower cell radii than the fl , breaking down at @xmath128fm , well above the transition to the uniform plasma phase .
however , the behavior of @xmath75 below @xmath129fm for the fl@xmath115 model does not seem desirable .
this could be cured by allowing the quark - meson coupling to depend on @xmath0 , but we find then non - smooth transitions to new non - trivial solutions in all the nts models discussed here .
we find more reasonable results if we keep the quark - meson coupling independent of @xmath0 , and thus the @xmath1cd seems the preferable model .
we shall present results only for this latter model in the next section .
we have performed the calculations using a straightforward numerical integration of the equations of motion for the @xmath81 and @xmath0 fields .
the mean field @xmath32 is found by directly minimizing the free energy ( [ eeos ] ) . as a check ,
we have also performed some calculations with a relaxation routine , but this is far more time consuming .
the numerical calculations with the @xmath1cd model were carried out using the following set of parameters @xmath130 the parameters of the potential @xmath10 are chosen to give a reasonable bag constant @xmath131mev/@xmath132 and glueball mass @xmath133gev .
the value of @xmath134 is calculated to be @xmath135@xmath136 , the single soliton energy is @xmath137 mev and the nucleon mass is @xmath138 mev .
the nucleon rms radius is @xmath139 fm .
the parameters were chosen to fit the rms radius and nuclear matter properties , resulting in a somewhat high value of the nucleon mass .
we did not spend much effort in fine - tuning the parameters , however .
as the short range repulsion provided by the vector field in qhd is provided by the wigner - seitz boundary conditions in the liquid soliton model , in what follows we shall take the coupling constant between the vector meson and the quarks to be zero , @xmath140 . on the other hand we shall treat @xmath141 as a free parameter in order to study its effect on the soliton matter . for completeness
, we also list the parameter sets used for the other models studied in the previous section . for the fl model ,
we take the parameters as in birse et .
al . @xcite .
( this set was also used in ref . @xcite . )
the full set of parameters for the soliton model is , @xmath142 these parameters correspond to a single soliton energy @xmath143 mev , and nucleon mass @xmath144 mev after removing the center of mass motion .
for the fl2 model we use the same potential parameters and a coupling @xmath145@xmath136 .
we proceed to our presentation of the results for the @xmath1cd model . to our knowledge , this is the first application of this particular model to the study of dense matter .
previous work has dealt with few nucleon systems @xcite . in figs . 2 and 3 we show the fields @xmath146 and @xmath147 for several values of the cell radius @xmath54 .
as can be seen from its behavior at the cell boundary , the quark field @xmath148 displayed in fig .
2 is that corresponding to the bottom of the band @xmath149 .
3 shows that the depth of the bag decreases only slighty as the cell radius is lowered , indicating that the quarks remain tightly bound .
( note that for the fl model , however , the bag depth actually increases with smaller @xmath54 @xcite . ) that the quarks are tightly bound in the bag can be seen clearly in fig .
4 , where we display the quark density @xmath150 given by eq .
( [ rho_q ] ) .
there is only a small overlap with quarks from neighboring cells even at radii below 1 fm . in figs . 1 - 4 ,
the quark - meson coupling has been set to zero . in fig .
5 the top and bottom of the lowest energy band is shown as a function of @xmath54 for several values of the quark - meson coupling @xmath141 .
the point at which the band begins to form , @xmath151fm , is not very sensitive to the value of @xmath141 . on the other hand ,
the structure of the band depends strongly on the coupling to the scalar meson , becoming wider as @xmath141 is increased . in fig .
6 we present the energy of the soliton @xmath36 given by eq .
( [ e_nr ] ) , and in fig .
7 we show the total energy per baryon @xmath152 of the system , derived from eq .
( [ eeos ] ) . as expected , the introduction of the scalar meson to the the soliton matter results in attraction and saturation .
the empirical nuclear saturation density corresponds to @xmath153 fm , whereas from fig .
7 we see , for example , that our nts nuclear matter energy per baryon has a minimum at @xmath154fm for @xmath155 .
the saturation energy is @xmath156 mev , as compared to the empirical value @xmath157 mev .
the compression modulus of nuclear matter is @xmath158 at the equilibrium point : for the present model we find @xmath159mev .
this is to be compared to empirical estimates that usually lie in the range @xmath160-@xmath161mev , with @xmath162mev the generally accepted value@xcite .
these results should be regarded as only order of magnitude , as at higher density we are surely underestimating the kinetic energy of the system .
we shall develop a more reasonable model of the liquid state , which leads to a more quantitatively accurate eos for solitonic nuclear matter , in the next paper of this series .
the solution for the @xmath1cd model is stable down to very low values of @xmath54 ( at least as low as @xmath126 fm ) , and it is apparent that with proper calibration this model can be used as an excellent starting point for the study of nuclear matter . another important feature of this model is the increase of the nucleon rms with density , fig .
this unexpected outcome of the model is in accord with the emc effect .
checking the sensitivity of this effect to model parameters , we found that the increase in the rms is insensitive to recoil corrections , but might disappear with different choices of the soliton parameters .
in this paper we have investigated a class of non - topological soliton models that generalize the friedberg - lee model .
we have studied nuclear matter in the wigner - seitz approximation , where neighboring bags begin to overlap at higher density , as a means of distinguishing between the various models .
the models differ in the precise form of the coupling between the quarks and the scalar gluon field , and in the presence and form of couplings to explicit meson fields . of the models
studied , we have found the chiral chromodielectric model to exhibit behavior best in line with phenomenological expectations . for this model
, there is no breakdown in the wigner - seitz calculation as the cell radius is decreased , thus overcoming a serious shortcoming of the original fl model .
we consider it a computational necessity to consider only constituent quarks and ignore colored gluons and sea quarks , and thus these restrictions along with the addition of explicit meson fields may be seen as an approximation to the original theory .
the original @xmath1cd model is closely based upon qcd , exhibiting absolute confinement , and it is therefore satisfying to see that this model is selected by our studies of nuclear matter .
we include mesons along the lines of quantum hadrodynamics , and we use the mean field approximation . in our final calculations
we have considered only the scalar meson , but it is straightforward to include also the vector meson , which should be done for fine - tuning the parameters .
we note that we found it best to use a quark - meson coupling that is not modulated by the quark - glueball coupling , in contrast to that argued for in refs .
@xcite . indeed , when using a @xmath163 coupling proportional to @xmath86 , we find jumps to qualitatively different solutions as density is increased .
the resulting curve of energy as a function of cell radius has several bumps , which is clearly an undesirable feature .
however , this should not necessarily be taken as an argument against models that employ quark - meson couplings that depend on @xmath0 , but rather merely a statement that when using the mean field approximation it is more consistent to use a @xmath0-independent @xmath81-@xmath3 coupling @xmath141 .
indeed , one can view this coupling as a `` mean field approximation '' to a more fundamental coupling namely , @xmath164 . in the end , then , we have used a @xmath1cd model with @xmath0-independent coupling between quarks and the scalar meson .
we find that the inclusion of the scalar meson provides a clear saturation point for nuclear matter and that a rough fit to both empirical nuclear matter and single nucleon properties can be obtained .
one of the most interesting features found is an increase in the nucleon rms radius at intermediate densities , in line with the emc effect .
this is dependent on the presence of the scalar meson .
there are clearly several refinements to our model and our calculations that must be considered before fine - tuning the parameters . among these are the inclusion of other mesons in the model , the addition of perturbative gluonic effects , and an improved calculation of the quark wave function along the lines of ref .
one can also improve our handling of the spurious cm motion .
foremost , however , we need to improve our modeling of the liquid state .
in this paper we have used a low - density approximation in modeling nuclear matter with solitons .
this consists in using a wigner - seitz approximation to calculate an effective nucleon mass .
then the kinetic energy is added to the system by taking the motion of the nucleons to be that of a fermi gas .
the assumptions of this picture include : the nuclear medium restricts any given nuclear bag to a spherical cell ; any given nucleon moves slowly , so that it can be constructed at rest and then boosted ; the quarks remain tightly bound inside the bag ; the nucleons move independently within the medium
. only the first of these assumptions can remain valid at high densities , where the nuclear medium will form a liquid in which each nucleon s motion is localized on short time scales and quarks need not remain tightly bound inside the bag .
since we are ultimately interested in studying the transition to a uniform quark plasma within our non - topological soliton model , we must do a better job of modeling the liquid state at high densities .
this is the subject of our next paper .
we thank j.a . tjon for helpful discussions . | a system of nontopological solitons interacting through scalar and vector meson exchange is used to model dense nuclear matter .
the models studied are of the friedberg - lee type , which exhibit dynamical bag formation due to the coupling of quarks to a scalar composite gluon field @xmath0 .
it is shown that the high density behavior of such models in the wigner - seitz approximation depends essentially on the leading power of the quark-@xmath0 coupling vertex . by insisting that the parameters of any soliton model be chosen to reproduce single nucleon properties
, this high - density behavior then selects a promising class of models that better fit the empirical results the chiral chromodielectric models .
it is shown that one can adjust the model to obtain saturation as well as an increase of the proton charge radius with nuclear density .
these two phenomena depend on the presence of a scalar meson and were not found in other nuclear matter calculations based on friedberg - lee type models . _
pacs : _
24.85.+p , 12.39.ki , 21.65.+f _ keywords : _ nontopological soliton , chiral chromodielectric model , friedberg - lee |
the ultra - relativistic heavy - ion collisions are expected to produce a hot and dense form of matter called quark gluon plasma ( qgp ) @xcite .
the fireball produced in these collisions thermalizes rapidly leading to expansion and cooling of the system .
subsequently , the hadronization takes place and the particles get detected in the detectors . during this process , two important stages occur as described below . the point in time after the collisions when the inelastic interactions among the particles stop is referred to as chemical freeze - out .
the yields of most of the produced particles get fixed at chemical freeze - out .
the statistical thermal models have successfully described the chemical freeze - out stage with unique system parameters such as chemical freeze - out temperature @xmath4 and baryon chemical potential @xmath5 @xcite . even after the chemical freeze - out
, the elastic interactions among the particles are still ongoing which could lead to change in the momentum of the particles . after some time , when the inter - particle distance becomes so large that the elastic interactions stop , the system is said to have undergone kinetic freeze - out . at this stage ,
the transverse momentum @xmath6 spectra of the produced particles get fixed .
the hydrodynamics inspired models such as the blast wave model @xcite have described the kinetic freeze - out scenario with a common temperature @xmath2 and average transverse radial flow velocity @xmath7 which reflects the expansion in transverse direction .
quantum chromo dynamics ( qcd ) , a theory of strong interactions , suggests that the phase diagram has two main phases : qgp and hadron gas .
lattice qcd predicts that the transition between hadron gas and qgp is a crossover @xcite at @xmath8 0 . at high @xmath5 ,
the transition is expected to be a first order @xcite . in between
, one expects the position of the critical point , where the first order phase transition line ends @xcite .
the experiments at rhic focus on exploring the qcd phase diagram , locating a critical point and determining the phase boundary between hadron and qgp phase . in view of these , a beam energy scan ( bes ) program was started at rhic @xcite .
the first phase of the bes program yielded many interesting results as a function of energy or @xmath5 related to the search for critical point and phase boundary @xcite .
the kinetic freeze - out parameters provide important information about the collision dynamics .
the vast range of data collected in the bes program allows for the systematic study of kinetic freeze - out parameters and to see their energy dependence trends .
the bes results presented here are obtained for au+au collisions at @xmath9 7.7 , 11.5 , 19.6 , 27 , and 39 gev at mid - rapidity ( @xmath100.1 ) using both star time projection chamber ( tpc ) and time of flight ( tof ) detectors @xcite .
the error bars shown in figures represent the statistical and systematic errors added in quadrature .
the pion spectra presented here are corrected for the weak decay feed - down and muon contamination while proton and anti - proton spectra are not corrected for feed - down effects .
as mentioned earlier the kinetic freeze - out parameters @xmath2 and @xmath3 of the system can be obtained using the hydrodynamics - motivated blast wave model .
the model assumes that the particles are locally thermalized at kinetic freeze - out temperature and are moving with a common transverse collective flow velocity @xcite . assuming a radially boosted thermal source , with a kinetic freeze - out temperature @xmath2 and a transverse radial flow velocity @xmath11 , the transverse momentum @xmath6 distribution of the particles can be given by @xmath12 where @xmath13 , @xmath14 being mass of a hadron , @xmath15 , and @xmath16 and @xmath17 are the modified bessel functions .
we use the flow velocity profile of the form @xmath18 , where @xmath19 is the surface velocity , @xmath20 is the relative radial position in the thermal source , and @xmath21 is the exponent of flow velocity profile .
average transverse radial flow velocity @xmath22 can then be obtained as : @xmath23 .
, @xmath24 , and @xmath25 versus transverse momentum in 05% central au+au collisions at @xmath9 7.7 , 11.5 , 19.6 , 27 , and 39 gev .
curves represent blast wave model fits . ]
[ fig : pt_spectra ] for ( a ) @xmath26 , ( b ) @xmath24 , and ( c ) @xmath25 in central heavy - ion collisions . ] figure [ fig : pt_spectra ] shows the invariant yields of @xmath26 , @xmath24 , and @xmath25 versus @xmath6 for @xmath10 0.1 in 05% central au+au collisions at @xmath9 7.7 , 11.5 , 19.6 , 27 , and 39 gev .
these distributions are fitted with the blast wave model which can be seen to reasonably describe the @xmath6 spectra of @xmath27 at all energies studied .
the low @xmath6 part of pion spectra is affected by resonance decays due to which the pion spectra is fitted above @xmath28 0.5 gev/@xmath29 .
the fit parameters are @xmath2 , @xmath30 , and @xmath21 .
with @xmath30 for different energies and centralities .
the centrality increases from left to right for a given energy .
the data points other than bes energies are taken from refs .
@xcite.,width=207 ] the @xmath6 spectra can also be characterized by obtaining @xmath31 or @xmath32 , where @xmath33 is the transverse mass of the particles .
figure [ fig : mt ] shows the energy dependence of @xmath34 for @xmath26 , @xmath24 , and @xmath25 .
the star symbols show results from the bes while the data points for ags , sps , top rhic , and lhc energies ( represented by open circles and squares ) , are taken from the references @xcite .
it can be seen that @xmath35 increases with energy at lower energies , remains almost constant at the bes energies and then increases again towards higher energies up to the lhc .
if the system is assumed to be in a thermodynamic equilibrium , @xmath35 can be related to temperature and @xmath36 can be related to entropy of the system ( @xmath37 @xmath38 @xmath39 ) . in view of this , the constant value of @xmath35 can be interpreted as a signature of first order phase transition @xcite .
however , more studies may be needed to understand this behavior @xcite .
figure [ fig : tkin_beta ] shows the variation of @xmath2 with @xmath30 for different energies and centralities .
@xmath2 increases from central to peripheral collisions suggesting a longer lived fireball in central collisions , while @xmath30 decreases from central to peripheral collisions suggesting more rapid expansion in central collisions .
furthermore , we observe that these parameters show a two - dimensional anti - correlation band .
higher values of @xmath2 correspond to lower values of @xmath30 and vice - versa .
[ fig : tb_en ] figure [ fig : tb_en ] ( left panel ) shows the energy dependence of kinetic and chemical freeze - out temperatures for central heavy - ion collisions .
we observe that the values of kinetic and chemical freeze - out temperatures are similar around @xmath94 - 5 gev .
if the collision energy is increased , the chemical freeze - out temperature increases and becomes constant after the @xmath911.5 gev .
however , the @xmath2 is almost constant around the 7.739 gev and then decreases up to the lhc energies .
the separation between @xmath4 and @xmath2 increases with increasing energy .
this might suggest the effect of increasing hadronic interactions between chemical and kinetic freeze - out when we go towards higher energies @xcite .
figure [ fig : tb_en ] ( right panel ) shows the average transverse radial flow velocity plotted as a function of @xmath36 .
the @xmath3 increases very rapidly at lower energies , remains almost constant for @xmath36=7.719.6 gev , and then increases again up to the lhc energies . since @xmath3 reflects the expansion in the transverse direction , it is observed that this expansion velocity is constant around @xmath36=7.719.6 gev .
we have presented a systematic study of kinetic freeze - out parameters in heavy - ion collisions with results from the rhic bes program .
the @xmath2 increases from central to peripheral collisions suggesting a longer lived fireball in central collisions , while @xmath30 decreases from central to peripheral collisions suggesting more rapid expansion in central collisions .
the kinetic freeze - out temperature suggests a decrease from lower ( @xmath40 45 gev ) to higher ( @xmath40 2.76 tev ) energies . the separation between chemical and kinetic freeze - out temperatures increases while going towards higher energies , suggesting that hadronic interactions increase between chemical and kinetic freeze - out at higher energies .
the average transverse radial flow velocity increases rapidly at lower energies ( @xmath40 23 gev ) .
the expansion in the radial direction remains similar around @xmath97.719.6 gev and then increases towards higher energies where the initial energy density produced is higher .
the energy dependence of @xmath34 for @xmath41 , @xmath42 , @xmath43 , and @xmath44 also shows similar constant behavior around lower bes energies which could be related to first - order phase transition signature .
s. v. afanasiev _
( na49 collaboration ) , phys . rev .
c * 66 * , 054902 ( 2002 ) ; c. alt _ et al . _
( na49 collaboration ) , phys .
c * 77 * , 024903 ( 2008 ) ; _ ibid . _
* 73 * , 044910 ( 2006 ) ; t. anticic _
( na49 collaboration ) , phys .
c * 69 * , 024902 ( 2004 ) .
j. cleymans , h. oeschler , k. redlich and s. wheaton , phys . rev .
c * 73 * , 034905 ( 2006 ) [ hep - ph/0511094 ] .
a. andronic , p. braun - munzinger and j. stachel , nucl .
a * 834 * , 237c ( 2010 ) [ arxiv:0911.4931 [ nucl - th ] ] . | the main aim of the rhic beam energy scan ( bes ) program is to explore the qcd phase diagram which includes search for a possible qcd critical point and the phase boundary between qgp and hadronic phase .
we report the collision energy and centrality dependence of kinetic freeze - out properties from the measured mid - rapidity ( @xmath0 light hadrons ( pions , kaons , protons and their anti - particles ) for au+au collisions at the center - of - mass energy @xmath1 7.7 , 11.5 , 19.6 , 27 , and 39 gev .
the star detector , with a large uniform acceptance and excellent particle identification is used in the data collection and analysis .
the kinetic freeze - out temperature @xmath2 and average collective velocity @xmath3 parameters are extracted from blast - wave fits to the identified hadron spectra and systematically compared with the results from other collision energies including those at ags , sps and lhc .
it is found that all results fall into an anti - correlation band in the 2-dimension ( @xmath2 , @xmath3 ) distribution : the largest value of collective velocity and lowest temperature is reached in the most central collisions at the highest collision energy . the energy dependence of these freeze - out parameters are discussed .
freeze - out , transverse momentum spectra , beam energy scan |
in their survey for ob stars in the field of the carina nebula ( ngc3372 ) , forte & orsatti ( 1981 ) discovered an early o type star in the darkest region of the nebula .
this star ( @xmath5 ; @xmath6 ; v=12.05 ) , number 15 in their list of new ob stars , was assigned a spectral classification o4v on photographic image tube spectrograms .
the star has been named in subsequent literature . in an infrared study of the stellar population in the direction of the carina nebula , smith ( 1987 )
identifies as a member of a group of heavily reddened ob stars in the southeast border of the open cluster trumpler 16 , considering that these stars most probably also are members of this cluster .
an anomalous reddening law characterized by a value of r = av / e(b - v)=4.8 is obtained by smith ( 1987 ) . in fig .
[ mapa ] we illustrate the location of in the carina nebula , inside the v shaped dust lane dividing the brightest part of the nebula , between the open clusters trumpler 16 and collinder 228 . the h@xmath7 image shown in fig .
[ mapa ] was obtained in 1999 , may , with the curtis - schmidt telescope at the cerro tololo inter american observatory ( ctio ) , chile .
fo15 was observed as a faint x - ray source in einstein satellite x - ray images of the carina nebula ( chlebowski et al .
because one of the most often proposed mechanisms for producing x - rays from o - type stars are colliding stellar winds in binary systems , we decided to include fo15 in our ongoing optical spectroscopic observations in search for o - type binaries . in this paper
, we will present our spectroscopic observations of fo15 showing this star to be a double lined binary system with a short orbital period .
after our spectroscopic analysis was essentially complete , fo15 also appeared as an eclipsing binary in the all sky automated survey ( asas ) ( cf .
pojmaski , 2003 ) .
we have analyzed the asas light curve together with our radial velocity study of the binary orbit .
twenty five spectral images of fo15 were obtained with the 2.15-m telescope at complejo astronmico el leoncito ( casleo ) in san juan , argentina , during several observing runs between february 1996 and march 2003 .
we used , alternatively , the cassegrain boller & chivens ( b&c ) spectrograph , with a 600 lmm@xmath8 grating and a pm 516@xmath9516 20@xmath10 m pixel ccd as detector , and the reosc echelle spectrograph , in its simple dispersion mode .
for the reosc spectra a tek 1024@xmath91024 pixel ccd , with pixel size of 24@xmath10 m was used as detector .
these instrumental configurations provide reciprocal dispersions of @xmath112.3 and 1.6 px@xmath8 , respectively .
the wavelength region observed was roughly @xmath123900
5040 .
we used a slit width of 2 arcsec for all our casleo spectra .
exposure times for the stellar images ranged between 30 minutes and 1 hour , resulting in spectra of signal - to - noise ratio between 50 and 150 , depending on seeing and transparency .
he - ar and cu - ar comparison arc images were observed with the b&c and reosc spectrographs , respectively , at the same telescope position as the stellar images immediately after or before the stellar exposures .
the usual series of bias and flat - field frames were also obtained for each observing night .
one high resolution spectral image of was obtained with eso - vlt ut2 ( kueyen ) and the uv - visual echelle spectrograph ( uves ) in december 2001 .
the setting used produced a blue spectrum covering the region 3260 - 4520 with a resolution ( 2 pixels ) r = 67000 and a red one spanning from 4580 to 6690 at a spectral resolution ( 2 pixels ) r = 58000 .
high gain , 2 @xmath9 2 binning mode was used .
a signal - to - noise of @xmath1150 was achieved in this observation .
further three high resolution spectra were obtained at las campanas observatory ( lco ) , as follows : one in december 2002 , with the magellan ii ( clay ) 6.5-m telescope and magellan inamori kyocera echelle ( mike ) double echelle spectrograph ( bernstein et al . 2003 ) .
a 1 @xmath9 3 binning was applied achieving a resolution of 19000 in the red spectrum ( 4500 - 7200 ) and 29000 in the blue spectrum ( 3200 - 4700 ) , at a signal - to - noise ranging from 60 to 200 in the red and from 40 to 100 in the blue in a 1200s exposure using a 1 arcsec slit width .
another echelle observation was obtained in may 2003 with the magellan i ( baade ) 6.5-m telescope and mike .
no binning was applied to these data yielding a resolution @xmath13 in the blue spectrum ( 3200 - 4800 ) and 27000 in the red spectrum ( 4800 - 8800 ) , with a 0.7 arcsec slit width .
the signal - to - noise achieved in this 600 seconds exposure is 60 - 130 in the red spectrum and 10 - 60 in the blue spectrum .
a third echelle spectrum of was obtained with the 2.5 m du pont telescope and its echelle spectrograph in july 2003 .
the signal - to - noise achieved in the 3 @xmath9 1200 seconds exposure ranges from 10 to 140 from the blue ( 3700 ) to the red end ( 10000 ) of the spectrum at a resolution @xmath14 using a 0.75 arcsec slit .
th - ar comparison lamps were secured for every high resolution observation as well as the usual series of bias and flat - field exposures .
all the spectral images were processed and analyzed with standard iraf routines .
mike spectra were extracted using the iraf tasks contained in the _ mtools _ package , developed by j. baldwin and available for downloading from lco website .
our spectra of confirm that the stellar spectrum corresponds to an o type star , as illustrated in fig.[fig02 ] . in this figure
we also observe that the heii absorption at @xmath154686 is stronger than any of the other heii lines , a proposed signature of a zero age main sequence luminosity class ( vz ) for o stars ( cf . walborn & blades 1997 and references therein ) .
a visual inspection of our first spectra of which we obtained at casleo in 1996 , december , during successive nights , showed that the stellar hydrogen absorption lines moved from the blueside to the redside of the nebular emissions , a signature of a rather high amplitude orbital motion in a binary system .
this was confirmed with further observations , which also showed that the neutral helium lines appeared double at maximum velocities , but ionized helium lines always appeared single , as shown in fig .
[ fig03 ] , which depicts two spectra obtained at casleo during approximately opposite orbital phases of the binary ( see below ) .
the spectrum of appears somewhat variable , as occasionally faint niii emisson at @xmath15 4634 - 40 appears , as seen in the upper spectrum of fig.[fig03 ] .
however , the radial velocities of this emission do not clearly correspond to any of the binary components , and it may arise in the zone of interaction of the close components .
nebular emission lines are observed in our spectra , as well as nebular hei @xmath15 3888 absorption .
the interstellar absorption lines of caii and nai appear multiple . these are common features observed towards the giant hii region of carina ( e.g. walborn & hesser 1975 , walborn et al .
the main components of caii absorption in our echelle spectra have a heliocentric radial velocity of @xmath16 , the nebular [ oiii ] @xmath15 5007 emission a velocity of @xmath17 , and the nebular hei @xmath15 3888 absorption of @xmath18 .
we have used these velocities to check the consistency of the stellar radial velocities derived from our lower resolution spectra .
spectral classification of components in a close binary systems is not as straightforward task as classifying spectra of single stars . in very close systems of hot stars , such as fo15 ,
mutual heating effects may introduce a spectral appearance with hotter effective temperature .
furthermore , spectral variations due to the interaction of stellar winds of the components are often observed .
we have chosen the usual approach to classify the binary components of in the spectrograms observed when the spectral lines have their maximum separation . to determine the spectral types
, we measured in our higher resolution spectra the equivalent widths of the hei and heii lines .
we then used the cuantitative classification criteria for o - type stars as described by conti & alschuler ( 1971 ) , comparing the equivalent widths of hei@xmath154471 vs. heii@xmath154542 .
we also compared hei@xmath154922 vs. heii@xmath155411 , as suggested by kerton et al .
according to these criteria , we obtained for the primary component of a spectral classifications of o5.5v .
we note however , that the spectrum observed near the orbital phase when the primary star is in front of the system ( see fig . [ fig02 ] ) and the contribution of the secondary to the spectrum is expected to be minor , corresponds to a spectral type about one subtype later than o5.5 , when compared with the digital spectral atlas of o - type stars published by walborn & fitzpatrick ( 1990 ) .
the spectral type of the secondary component is more difficult to ascertain .
heii absorptions correponding to the secondary component is not observed in our lower resolution spectra , as is evident in fig .
[ fig03 ] . in our high resolution
spectra all the hei absorptions appear as double lines , but only the heii@xmath154686 line corresponding to the secondary component is clearly observed .
both components of this line are shown in fig .
[ fighe2 ] .
heii@xmath155411 absorption of the secondary component may also be present , but it appears blended with the diffuse interstellar bands at @xmath125404 and 5420 .
the spectral type of the secondary component probably is o9.5v , but could be slightly later .
we do not observe in our optical spectra of any mass - loss indicators ( signs of strong stellar winds ) , such as those listed e.g. by hutchings ( 1978 ) .
the balmer h@xmath7 line is observed in absorption in both binary components , as illustrated in fig .
[ figha ] .
the radial velocity of this absorption agrees with the values derived from he lines , indicating that the h@xmath7 absorption is mostly photospheric , and not formed in the accelerated part of an expanding atmosphere .
thus may be composed of weak - wind and low mass - loss rate o type dwarfs , as those recently discussed by martins et al .
( 2005b ) .
to determine the radial velocity orbit of the binary , we measured the radial velocities of the he lines in the spectra of fitting gaussian profiles to the spectral lines within the iraf routine splot .
the results are listed in table [ vr ] .
the radial velocities of the primary component are mostly based on the velocities of heii absorptions , and the velocities of the secondary on those of hei . to calculate the orbital parameters , we used a modified version of the program originally written by bertiau & grobben ( 1976 ) , introducing the period determined by asas photometry , namely 1.4136 days , as an initial value .
we assigned higher weight to the higher resolution observations and to those obtained near quadratures .
the orbital parameters are listed in table 2 , along with their formal standard errors as calculated by the above mentioned program .
.observed heliocentric radial velocities for the primary and secondary components of .
radial velocities and ( o - c ) values are in [ kms@xmath8 ] .
[ cols="<,^ , < , < , < , < , < " , ] [ tab02 ]
photometric v filter data of fo15 were reported by the all sky automated survey ( asas ) in pojmaski ( 2003 ) .
this star is catalogued as asas 104536 - 5948.4 in the asas catalog of variable stars , beeing classified as an eclipsing binary with a period of p=1@xmath204136 .
a visual inspection of the asas light curve of fo15 shows periodic light variations with a rather large apparent scatter of data , almost @xmath21 over all orbital phases . in order to obtain a first estimation of the orbital inclination @xmath22 of the binary system , we have attempted to fit a numerical eclipsing binary model to the asas observations , using the wilson - devinney ( w - d ) code ( wilson & devinney 1971 , wilson 1990 , wilson & van hamme 2004 ) .
asas photometric data and our radial velocities were used to compare with the results of the model fitting .
we set the w
d code in mode 2 for detached binaries with no constraints on the potentials ( except the luminosity of the secondary ) .
the simplest considerations were applied for the emission parameters of the stars in the model , i.e. the stars as black bodies , approximate reflection model ( mref=1 ) .
no third light or spots were included .
we adopted gravity darkening exponents @xmath23 , and bolometric albedos @xmath24 were set for radiative envelopes .
we used the square root limb darkening law .
limb darkening coefficients for visual wavelengths were taken from daz - cordovs , claret & gimnez ( 1995 ) , and bolometric limb darkening coefficients from van hamme ( 1993 ) .
we adopted the period p = 1@xmath2041356 and mass ratio @xmath25 from the radial velocity orbit ( cf .
table 2 ) . for both binary components ,
effective temperatures corresponding to their spectral types were adopted from the compilation by martins et al .
( 2005a ) , i.e. @xmath26 and @xmath27 for the primary and secondary components , respectively .
we assumed that the system has a circular orbit ( @xmath28 ) with both components rotating synchronously ( @xmath29 ) , as suggested by the observations and as expected for a massive binary with a short period , as is the case for . with all of these parameters
fixed , we generated synthetic light and radial velocity curves adjusting them to the observations .
we used the phoebe package ( pra & zwitter 2005 ) which incorporates numerical innovations and technical aspects to the w
the results for the best fit , optimised by the code , are shown in figures [ cl ] and [ vr ] and listed in table 3 .
c c c parameters & + & prim . & sec .
+ p ( @xmath30 ) & + @xmath22 ( @xmath31 ) & + a ( @xmath32 ) & + m ( @xmath33 ) & 30.4 @xmath34 1 & 15.8 @xmath34 1 + @xmath35/@xmath36 & + r mean ( @xmath32 ) & 7.5 @xmath34 0.5 & 5.3 @xmath34 0.5 + @xmath37 ( @xmath32 ) & 8.3 & 6.15 + teff ( @xmath38)@xmath39 & 40000 & 32000 + mbol & [email protected] & [email protected] + @xmath40 & + log g ( cgs ) & [email protected] & [email protected] + with the inclination of the orbital plane , which we found to be @xmath41 , the values of the stellar masses for the components of @xmath42 and @xmath43 , are in fair agreement with the tabulations by martins et al .
( 2005a ) based on models of stellar atmospheres .
however , the stellar mean radii that we have obtained , namely @xmath44 and @xmath45 for the primary and secondary components , respectively , are about 30% smaller than the tabulated values , hence the bolometric luminosities implied from our light
curve solution are also lower . for the secondary component
, would it be e.g. of spectral type b0v , does not modify this result beyond the quoted error bars . ]
this is similar to what was observed for another o type binary system in the carina nebula , namely tr16 - 104 by rauw et al .
( 2001 ) , who interpreted the observations as due to fainter absolute magnitudes for stars just entering to the main sequence , i.e. stars of luminosity class vz .
in fact the spectrum of tr16 - 104 also exhibits the spectral signature of zams o - type stars , namely that heii@xmath154686 absorption is stronger than other heii lines , as is illustrated in fig .
[ sp104 ] .
the case may apply also to but for a more reliable assesment of this problem , an improved and less noisy light curve is needed , and is currently being obtained ( fernndez lajs et al .
2006 , in preparation ) .
based on their proximity in the sky , forte & orsatti ( 1981 ) assumed that belongs to the open cluster tr 16 , of which @xmath46 car is the brightest member .
however , the line of sight in this direction of the galaxy goes almost parallel to the sagittarius carina spiral arm , which is rich in young stellar population .
an o5v type star with the colors and apparent magnitude as those observed for would have a spectrophotometric distance of @xmath115 kpc , if normal interstellar extinction is assumed .
this would place well behind the carina nebula and clusters embedded within . on the other hand
, it is well known that the total to selective extinction ratio ( r ) in the direction of carina nebula is anomalous ( e.g. smith 1987 , tapia et al .
therefore , we decided to determine the value of r for using the published ubv photometry from forte & orsatti ( 1981 ) and ir magnitudes from the 2mass all - sky catalog of point sources ( cutri et al .
2003 ) .
we used the code chorizos developed by maz - apellniz ( 2004 ) to derive the value of r for .
chorizos is a code that uses @xmath47 minimization to find all models of energy distribution compatible with an observed data set in the @xmath48-dimensional model parameter space , which in our case are broadband photometry and spectral type . for a complete description of the method , see maz - apellniz ( 2004 ) .
we considered tlusty ( lanz & hubeny , 2002 ) atmosphere models for the spectral energy distribution ( sed ) of o - type stars .
an effective temperature of @xmath49 and @xmath50 were adopted to constrain the models .
using the six color photometry ( @xmath51 ) , we derived a colour excess @xmath52 and a ratio of total to selective extinction @xmath53 , which are the monocromatic equivalents to the usual @xmath54 and @xmath55 , respectively .
the main source of error comes from the adopted values of the magnitudes , which were obtained at different epochs , and therefore they could correspond to different orbital phases and be affected by the photometric variations of the binary system . with the bolometric magnitudes derived from our light - curve analysis , and assuming that the bolometric corrections corresponding for the spectral types as listed by martins et al .
( 2005a ) hold for the binary components , we obtain absolute visual magnitudes of -4.2 and -3.2 for the primary and secondary , respectively .
these values are about 1 magnitude fainter than those tabulated by martins et al .
( 2005a ) for stars of spectral types
o5.5v and o9.5v , but are in closer agreement with the absolute magnitudes for zams o type stars as tabulated by hanson et al .
the values of absolute magnitudes of the binary components that we have obtained coupled with the total to selective extinction ratio r=4.15 derived above , would locate at a distance of 2.2kpc .
this is coincident with the distance derived for @xmath46 car by davidson et al .
( 2001 ) based on spatially resolved doppler velocities of the bipolar ejecta . is located in the region of the carina nebula named south pillars by smith et al .
( 2000 ) ( see fig . [ mapa ] )
recently imaged by the spitzer space telescope .
newborn stars in dust pillars pointing to @xmath46 car are observed in the infrared image of spitzer ( smith et al . 2005 ) .
if is at the same distance , it can be considered as a zams star embedded in an active star formation region .
as mentioned previously , was detected as an x - ray source in einstein observations of the carina nebula .
subsequent x - ray data of were observed by the satellite in the context of the x - mega international campaign ( cf .
corcoran 1996 ) which also involves spectroscopic observations of hot stars with detectable x - ray emission .
the -hri image shows a weak x - ray emission at the position of , considerably weaker than expected on the basis of the first observation of this star , thus suggesting a variable x ray source . in further x ray observations by the satellite , appeared as a hard x - ray source ( albacete - colombo et al .
however , in their analysis of an x - ray image of region obtained by the chandra satellite , evans et al .
( 2003 ) did not mention in their list of sources with ob optical counterparts . in their list of detected sources without optical counterparts ,
the source number 106 appears close to but with a difference of over 10 arcsec in the published positional coordinates .
although is observed near the borders of the x - ray images , which all had as the aimpoint , a positional difference larger than 10 arcsecs seemed improbable .
therefore , we decided to re - examine the chandra x - ray image in order to see if there might be some instrumental problem in the non - detection of in this image .
the result is depicted in fig .
[ xx ] , which shows two very close x - ray sources , the southern one is coincident with and the northern with source 106 of evans et al .
this source does not appear in the x - ray image of xmm satellite used by albacete - colombo et al .
( 2003 ) , as seen in fig .
therefore , it most probably is a variable source due to a pre - main sequence star located in the vicinity of .
indeed , evans et al .
( 2003 ) already have suggested that the x - ray sources without optical counterparts , such as their source 106 close to , probably are pre - main sequence stars . as mentioned above ,
x - ray flux variability seem to be present in this system .
albacete - colombo & micela ( 2005 ) have studied the x - ray emission of in the 0.4 - 10 kev energy range , revealing the existence of long term x - ray variability .
higher and lower un - absorbed x - ray flux limits are between 15.8@xmath910@xmath56 to 0.96@xmath910@xmath56 ergs@xmath57 , implying l@xmath58/l@xmath59 ranges from 9.0@xmath910@xmath60 to 0.54@xmath910@xmath60 , respectively .
these authors also discuss the origin of the observed hard x - ray photons as inverse compton scattering , and confirm the existence of soft ( 0.2 - 1.2 kev ) short - term variability ( @xmath1125% of the total flux ) .
* we have discovered that the o - type star immersed in the active star formation site called southern pillars in the carina nebula , is a short period eclipsing binary . *
both binary components are visible in the spectrum , the secondary component showing considerably weaker lines .
* we classify the primary spectrum as o5.5vz , i.e. as an early type zero - age - main - sequence star .
the secondary seems to be of spectral type o9.5v . *
analysis of the asas light curve of fitting a binary model by the wilson - devinney method , yields an orbital inclination of @xmath1180@xmath31 .
* the stellar masses of the components are @xmath11 30 and 16 @xmath61 . * simultaneous light and radial velocity curve analysis yields components with smaller radii and fainter absolute magnitudes when compared with normal galactic o - type stars .
these values are in agreement with recently born zams o type stars . * an individual determination of total - to - selective extinction ratio ( r ) for yields a value of 4.15 , which coupled with the values of absolute magnitudes determined from the light curve locate at a distance of 2.2kpc , coincident with that of @xmath46 car . *
a chandra x - ray image shows two close sources at the position of . presumably the northern source is a pre - main - sequence star with an ocassional high x - ray state , as this source apparently is not visible in the x - ray image of the same location observed by the xmm satellite
we are indebted to roberto gamen for kindly obtaining 4 spectra of fo15 for this study .
we thank the directors and staff of casleo , lco , ctio and eso , for the use of their facilities .
we also acknowledge the use at casleo of the ccd and data acquisition system partly financed by u.s .
nsf grant ast-90 - 15827 to r. m. rich .
this research has received financial support from ialp , conicet , argentina .
vn and efl thank cic - ba for travel support .
rb has received financial support from fondecyt no 1050052 .
mo is grateful for financial support from unlp in the form of a research studentship .
we thank the referee , dr . c. evans , for useful comments which have improved the presentation of this paper . | we report the discovery of a new o type double lined spectroscopic binary with a short orbital period of 1.4 days .
we find the primary component of this binary , , to have an approximate spectral type o5.5vz , i.e. a zero age
main sequence star .
the secondary appears to be of spectral type o9.5v .
we have performed a numerical model fit to the public asas photometry , which shows that is also an eclipsing binary .
we find an orbital inclination of @xmath0 . from a simultaneous light - curve and radial velocity solution
we find the masses and radii of the two components to be @xmath1 and @xmath2 solar masses and @xmath3 and @xmath4 solar radii .
these radii , and hence also the luminosities , are smaller than those of normal o - type stars , but similar to recently born zams o - type stars .
the absolute magnitudes derived from our analysis locate at the same distance as . from chandra and xmm x - ray images we also find that there are two close x - ray sources , one coincident with and another one without optical counterpart .
this latter seems to be a highly variable source , presumably due to a pre
main sequence stellar neighbour of .
[ firstpage ] stars : binaries , spectroscopic , eclipsing stars : o type - stars : fundamental parameters - stars : individual : - x - rays : stars |
the study of liquid crystal phases benefits from geometrical reasoning in two important ways .
firstly , liquid crystal elasticity can often be cast in terms of the curvature of equipotential lines ( or surfaces ) that map out the corresponding textures .
second , the observed textures are strongly affected by geometric and topological constraints imposed by the presence of boundaries confining the system .
the liquid crystal ground state results from the competition between the energetic requirement of minimizing the `` curvature of the texture '' and the geometric frustration introduced by boundaries that impart a preferred curvature at the edge of the sample that often can not propagate across the system @xcite .
the boundary conditions can be controlled experimentally with the possibility of designing molecular systems with intriguing technological applications @xcite .
for example , colloidal particles coated by a very thin nematic layer have in their ground state four disclinations sitting at the vertices of a tetrahedron .
each coated colloidal particle can then be viewed as the fundamental building block of a self assembled lattice with tetravalent coordination .
the `` bonds '' between the colloidal particles could be provided by chemical linkers attached at the four `` bald spots '' at the cores of the four disclinations present in each colloid @xcite .
a second example , is provided by self - assembled systems of block copolymers @xcite which are a promising tool for `` soft lithography '' on both flat and curved substrates @xcite .
in addition , liquid crystals in confined geometries provide an arena for physicists and mathematicians interested in applications of geometrical and topological ideas to material science @xcite . in this work we present a theoretical study of liquid crystal phases ( focusing on vector , nematic and hexatic order ) confined in a spherical shell of varying thickness with
the director assumed to be tangent to the two interfaces .
we first consider the two dimensional regime where a nematic film coats a quenched spherical surface such as a colloidal particle in solution or the interface of , say , a water droplet in oil .
the presence of topological defects in the ground state for ordered states on spherical surfaces is unavoidable @xcite .
recent experimental and theoretical investigations of spherical crystallography have provided an alternative context to study the constraints posed by the compactness of the underlying curved space @xcite .
more recent explorations have concentrated on 2d ordered phases confined to interfaces of varying gaussian curvature @xcite as well as dynamically fluctuating surfaces @xcite . as the thickness of a nematic film increases , an escaped three dimensional texture ,
also strongly influenced by the spherical topology and the boundary conditions , become energetically favored with respect to planar textures .
this instability destabilizes the tetravalent nematic texture on colloids . in this paper
, we estimate the thickness of the nematic film above which a texture with four radial disclination lines of charge @xmath1 becomes unstable to four half - hedgehogs .
the two competing textures studied in this paper are shown in fig .
[ fig : summary ] .
we also discuss the possibility of hysteresis between the two textures .
( top panel ) two - dimensional texture characterized by four short disclination lines at the vertices of a tetrahedron inscribed in the sphere .
the surface texture shown ( inscribed on the surface of a baseball ) is invariant throughout the thickness of the shell .
( bottom panel ) cut view of the escaped three dimensional texture given by two pairs of half hedgehogs located at the north and south poles of the sphere.,scaledwidth=30.0% ] the organization of this paper is as follows . in section [ sec : texture ] we derive exact solutions for the ground state of spherical films of tilted molecules and nematogens within isotropic elasticity by using the method of conformal mappings . in the notation of references @xcite
, these situations correspond to order parameters described by a bond angle with @xmath2 fold symmetry in the tangent plane of the sphere ( see appendix [ appa ] ) .
a mathematical justification for our approach is provided in appendix [ appb ] , where the same technique is illustrated in the context of a more familiar flat space problem . in section [ sec : stability ]
we study the stability of liquid crystal textures to thermal fluctuations by means of a normal mode analysis whose details are relegated to appendices [ appa ] and [ appc ] .
the stability of the valence - four texture against escaped solutions is considered in section [ sec : valence ] where a phase diagram is derived with the thickness as a control parameter .
the texture distortions caused by the elastic anisotropy between bend and splay deformations are briefly considered in section [ sec : conc ] .
the liquid crystal free energy for molecules embedded in an arbitrary frozen surface can be written in the one constant approximation as @xmath3 where @xmath4 is a set of internal coordinates , @xmath5 is the liquid crystal director defined in the tangent plane , @xmath6 is the covariant derivative with respect to the metric of the surface and @xmath7 is the infinitesimal surface area @xcite .
the free energy of eq.([free energy ] ) is invariant upon rotating each molecule * * n**@xmath8 by the same ( arbitrary ) angle with respect to any axis of rotation perpendicular to the local tangent plane .
the treatment of systems with a p - fold symmetry is straightforward provided that the one frank constant approximation is used for @xmath9 and @xmath10 and the consequences of any additional couplings to curvature neglected @xcite .
this choice of free energy implies that the minimal energy configuration will be given locally by neighboring @xmath5 vectors which differ only by parallel transport .
the curvature of the surface induces `` frustration '' in the texture .
in fact , by gauss `` theorema egregium '' @xcite , tangent vectors parallel transported along a closed loop are rotated by an amount equal to the gaussian curvature integrated over the enclosed area . on a sphere
, this theorem insures that the nematic ground state always has four excess disclinations @xcite .
more generally , the sum of the topological charges on any closed surface is equal to the integrated gaussian curvature , implying a minimum of @xmath11 and @xmath12 disclinations in the ground state of tilted molecules and hexatics , respectively .
we introduce a local angle field @xmath13 , corresponding to the angle between @xmath5 and an arbitrary local reference frame , we can rewrite the free energy introduced in eq.([free energy ] ) as : @xmath14 where @xmath15 , @xmath16 is the determinant of the metric tensor @xmath17 and @xmath18 is the spin - connection whose curl is the gaussian curvature @xmath19 @xcite . on a sphere of radius @xmath20 parametrized by polar coordinates
@xmath21 , the only non vanishing components of the ( inverse ) metric tensor are @xmath22 and @xmath23 . a convenient choice of of the spin connection ( which plays the role of the vector potential ) is discussed in appendix [ appa ] . the simplified free energy in eq.([eq : patic - ener ] ) is the starting point of our analysis
. the orientational order of molecules tilted by a constant angle with respect to a spherical interface can be modelled by a vector field @xmath24 defined in the local tangent plane on which the molecule has a fixed length projection @xcite . to determine the ground state of the liquid crystal texture , we minimize the frank free energy of eq.([eq : patic - ener ] ) .
as discussed above , the topological charges must sum up to @xmath25 , the integrated gaussian curvature of the sphere @xcite . for a vector field ( @xmath9 ) the texture with only two defects of charges @xmath26 minimizes the frank free energy and satisfies the topological constraint .
since the defects repel each other they preferentially sit at two antipodal points that we can designate as the north and south pole of the sphere . if the splay and bend coupling constants of the nematic are equal , then there is a large degeneracy in the ground state arising from the invariance of the vector free energy in eq.([eq : patic - ener ] ) under global rotations @xmath27@xmath28@xmath27+c , where @xmath29 .
one representative texture is a `` sink '' and a `` source '' of @xmath30 at the two poles . in this splay rich texture @xmath30
is parallel to the lines of longitude on a sphere . in a bend rich texture , related to the previous by a @xmath31 rotation about the local normal to the surface ,
@xmath30 is everywhere parallel to the lines of latitude .
any other rotation of @xmath30 that makes an arbitrary constant angle with respect to this texture is an acceptable solution for the ground state of the molecules . as we now show ,
this degeneracy is lifted when @xmath32 .
indeed the effect of distinct splay and bend elastic constants @xmath33 and @xmath34 ( the twist elastic constant @xmath35 is absent in two dimensions ) is to select the bend - rich texture if @xmath36 or the splay - rich one if @xmath37 .
the intermediate configurations obtained by a global rotation of the director are now unstable .
assume for simplicity that @xmath37 .
in this case , it is convenient to recast the frank free energy ( see appendix a ) as follows : @xmath38 \!\!\!\ ! \quad , \label{eq : frank1}\end{aligned}\ ] ] where the covariant derivatives is expressed in terms of the christoffel connection , @xmath39 , @xmath40 and the covariant form of the curl squared is @xcite , @xmath41 the first term in eq.([eq : frank1 ] ) resembles the frank free energy in the one coupling constant approximation and is minimized by choosing the sink - source ( or `` lines of longitude '' ) solution .
the second term ( which is positive definite ) will vanish for this texture since the sink - source texture is bend free .
all other textures have a higher energy .
a similar argument can be used to prove that the two vortex - configuration which follows the lines of latitude is the minimum of the free energy when @xmath36 by rewriting the frank free energy as @xmath42 \!\!\!\ ! \quad , \label{eq : frank3}\end{aligned}\ ] ] where the covariant form of the divergence reads @xmath43 the latitudinal texture minimizes the first term of eq.([eq : frank3 ] ) while the second vanishes because this texture is splay free .
any deviation from the splay - free latitudinal texture will only increase the energy .
the energy of both textures can be expressed as a function of the anisotropy parameter , @xmath44 , and the mean of the elastic constants , @xmath45 , @xmath46 and the radius of the sphere , @xmath20 , scaled by the short distance cutoff @xmath47 . the resulting free energy for arbitrary @xmath44 reads ( see eq.([eq : frankexplicit3 ] ) ) @xmath48 the conclusions of this section are summarized in fig .
[ fig : tilted ] which suggests that there is a discontinuous first order transition when @xmath44 passes through zero .
this analysis mirrors similar arguments valid in the plane @xcite .
schematic illustration of the phase diagram of the texture of tilted molecules on a sphere as a function of the anisotropy parameter @xmath44 superimposed on a plot of the free energy , @xmath49 , stored in the texture versus @xmath44 .
the two competing ground states ( top view ) are characterized by either pure bend ( lines of longitude configuration at left ) or pure splay ( lines of latitude configuration at right).,scaledwidth=45.0% ] the nematic texture of very thin spherical shells of nematic liquid crystal with tangential boundary conditions can be analyzed within the one frank constant approximation by using the method of conformal mappings whose mathematical justification is illustrated in appendix [ appb ] by means of a simpler example .
an elegant argument introduced by lubensky and prost in ref.@xcite shows that the ground state of nematogens on a sphere is given by @xmath50 disclinations of topological charge @xmath51 sitting at the vertexes of a tetrahedron .
the energy of single disclinations is proportional to the square of its strength . as a result ,
the longitudinal and latitudinal textures derived for tilted molecules in section [ tilt ] are unstable since their energies can be lowered by splitting each @xmath52 defect at the north and south pole into two @xmath51 disclinations and letting them relax to their equilibrium positions at the vertexes of a tetrahedron where they are as far away from each other as possible . according to a calculation in ref.@xcite , the energy @xmath53 of a sphere of radius @xmath20 with in plane orientational order and @xmath54 interacting minimal disclinations for a p - fold order parameter
is given by @xmath55 \!\!\!\ ! \quad .
\label{eq : e4bis}\end{aligned}\ ] ] where the @xmath56 are constants depending on the symmetry of the order parameter and the defect core energy while @xmath57 is the thickness of the liquid crystal layer .
the numerical values of the relevant constants are @xmath58 and @xmath59 . when @xmath10 is chosen in eq.([eq : e4bis ] ) the elastic energy is indeed smaller than the corresponding value for @xmath9 in the limit @xmath60 , in agreement with related arguments given in ref.@xcite . to obtain an algebraic expression for the texture we proceed as illustrated in appendix [ appb ] and seek a function @xmath61 which is harmonic on the sphere except for two arcs connecting the defects in pairs .
the calculation for nematogens described below was suggested to us by f. dyson @xcite
. the function @xmath62 , which we can interpret as an electrostatic potential , takes equal and opposite values on the two arcs and is equal to zero on a baseball - like seam ( see fig .
[ fig : baseball ] ) which divides the sphere into two congruent regions .
schematic illustration of the baseball texture of a thin nematic shell .
the same texture is reproduced from a different perspective in the top panel of fig .
[ fig : summary].,scaledwidth=30.0% ] the nematic director is then oriented ( up to a global rotation ) along the contour lines of @xmath62 , that is the equipotential lines of this `` curved space capacitor '' . in this analogy ,
the contour lines of @xmath63 are electric field lines , hence they correspond to a valid texture where the director is rotated locally by @xmath31 with respect to the equipotential lines .
the arcs can be either great - circle arcs extending more than half way round the sphere or short great - circles arcs connecting the same pair of defects along the shortest path .
the first choice leads to equipotential lines whose seam resembles in shape that of a baseball .
if the second choice is made the pattern of equipotential lines would not deviate much from concentric circles and the seam would look more like the seam of a cricket ball .
we will explicitly show that the two choices are equivalent since the equipotential lines of the first solution are field lines of the second and vice versa .
we choose the arcs connecting the defect pairs along great circles and we take the four defects labelled by a , b , c , d to lie at the vertices of a tetrahedron inscribed on a sphere of radius @xmath64 and whose north and south poles are @xmath65 and @xmath66 respectively @xmath67 we now perform a stereographic projection ( see fig .
[ fig : stereo ] ) graphical construction of the stereographic projection .
regions close to the north pole have larger images in the conformal plane than regions of equal areas close to the south pole .
the stereographic projection preserves the topology of the surface provided all points at infinity are identified with the north pole.,scaledwidth=45.0% ] that maps every point on a unit sphere centered on the origin onto the plane @xmath68 according to the rule @xmath69 the coordinates of the image points ( connected to points on the sphere by dashed lines in fig . [ fig : stereo ] ) are given by @xmath70 upon transforming to a complex coordinate @xmath71 , the four tetrahedral points of eq.([eq : points1 ] ) are mapped onto @xmath72 where @xmath73 ( in this section @xmath74 does not refer to the symmetry of the order parameter ) . the great arc passing through the south pole ( corresponding to one capacitor plate in the electrostatic analogy ) maps onto the segment @xmath75 , as illustrated schematically in the top panel of fig . [ fig : mapping1 ] , while the great arc through the north pole illustration of the change in the branch cut structures of the complex function @xmath76 describing the nematic texture after performing a series of conformal transformations . in the top panel
we have simply performed a stereographic projection from the sphere .
the middle panel shows the `` fold up '' transformation of eq.([eq : map1 ] ) whereas the bottom panel corresponds to the transformation in eq.([eq : bilinear1 ] ) that symmetrizes the positions of the cuts.,scaledwidth=45.0% ] maps onto the two semi - infinite segments of the line @xmath77 which bracket @xmath78 .
we can fold the two cuts in the @xmath79 plane back on top of each other by mapping the @xmath79 plane onto the @xmath80 plane via @xmath81 as shown in the middle panel of fig .
[ fig : mapping1 ] , the images and b of a and b now both lie on the real axis at @xmath82 while the images of c and d now lie at @xmath83 .
the two cuts in the @xmath80 plane are both on the real axis , running from zero to @xmath82 and from @xmath83 to minus infinity .
on the sphere , these correspond to geodesics connecting defects which stretch more than halfway around the sphere . in order to make the cuts symmetric with respect to the imaginary axis ( see the bottom panel of fig .
[ fig : mapping1 ] ) we search for a conformal transformation that maps the following four points in the complex @xmath80-plane to four points on the real axis of a complex @xmath84-plane @xmath85 in order to fully determine the conformal transformation we need to determine the value of @xmath86 .
this can be done by using a standard relation in the theory of conformal transformations @xcite @xmath87 upon inserting the points of eq.([eq : points3 ] ) into eq.([eq : formula ] ) , we determine the value of @xmath86 ( less than one ) @xmath88 equation ( [ eq : points3 ] ) contains four independent relations so we are still left with three conditions to determine the three independent coefficients @xmath89 of the bilinear conformal transformation that implements the mapping illustrated pictorially in the bottom plate of fig .
[ fig : mapping1 ] @xmath90 the required coefficients needed to implement the mapping in eq.([eq : points3 ] ) are @xmath91 to solve laplace s equation , we desire a function @xmath76 which is analytic except on the two cuts on the real axis , and whose real part takes constant values on the cuts . by symmetry , @xmath76 is an odd function of v , and its real part @xmath92 is zero when the real part of v is zero .
therefore the image of the seam in the v plane is simply the imaginary axis @xmath93 . upon substituting for @xmath84 using equations ( [ eq : bilinear1 ] ) and ( [ eq : map1 ] ) ,
the condition @xmath93 becomes @xmath94 with the help of equations ( [ eq : ab ] ) and ( [ eq : points2 ] ) , we can now write down the equation of the seam explicitly in the original cartesian coordinates @xcite @xmath95 or in spherical polar coordinates @xmath96 as @xmath97 the seam defined by the line of zero potential , is represented for different orientations of the sphere in fig .
[ fig : seam3d ] . different views of a track of parallel nematogens which partitions the sphere into two equal areas , each containing two @xmath1 disclination defects .
it resembles a `` fattened '' version of the seam of a baseball.,scaledwidth=45.0% ] its contour length @xmath98 , on a unit radius ball , is readily calculated upon integrating the expression for the infinitesimal arc of the seam @xmath99 from @xmath100 radians to @xmath101 radians and multiplying the result by four in view of the symmetry of the seam .
the values of @xmath102 and @xmath103 are obtained from eq.([eq : seam2 ] ) by setting @xmath104 equal to @xmath105 and @xmath106 respectively . upon using eq.([eq : seam2 ] ) to substitute @xmath107 in eq.([eq : lineint ] ) , we obtain @xmath108 for a sphere of unit radius . the seam is longer than the equatorial circumference by slightly less than @xmath109 .
the branch cut structure in the @xmath84 plane is sufficiently simple to allow a guess of the corresponding analytic function @xmath76 . a function with cuts from @xmath86 to @xmath64 and @xmath110 to @xmath111 , whose real part is equal and opposite on the two cuts and with a single imaginary period around any curve separating the cuts
is easily identified to be a standard elliptic integral @xmath112^{- \frac{1}{2 } } dt \!\!\ ! \quad .
\label{eq : omega - nem}\end{aligned}\ ] ] with @xmath84 given in terms of @xmath79 by equations ( [ eq : map1 ] ) and ( [ eq : bilinear1 ] ) .
the nematic director is oriented ( up to a global rotation ) along the contour lines of the imaginary or ( real part ) of @xmath113 .
the equipotential ( red ) and field lines ( black ) of @xmath76 are conveniently plotted using the stereographic projection plane @xmath71 in fig .
[ fig : nematic - plot ] along with the positions of the disclinations ( green dots ) .
illustration of the nematic texture in stereographic projection ( color online ) .
the red and black field lines correspond to two energetically degenerate ( in the one frank constant approximation ) families of bend and splay rich textures .
the dots indicate the four tetrahedral @xmath1 disclination defects.,scaledwidth=45.0% ] it is easy to switch from the stereographic - projection plane @xmath71 of fig .
[ fig : nematic - plot ] to spherical polar coordinates @xmath21 by using the relation ( reviewed in appendix [ appb ] ) , @xmath114 where @xmath20 is the radius of the sphere .
if we had constructed the base - ball with cuts along the @xmath115 geodesics connecting the defects , then the form of the texture given in eq.([eq : omega - nem ] ) would be the same , but the parameter @xmath86 in the elliptic integral would be given by @xmath116 instead of eq.([eq : k ] ) .
the equation for the seam becomes @xmath117 and the corresponding equipotential and field lines are plotted in black and red respectively in fig .
[ fig : nematic - plot ] .
the expression reads @xmath118 in spherical polar coordinates , and leads to the same contour length @xmath98 as eq.([eq : seam2 ] ) .
thus the two choices of arcs lead to two equivalent textures differing only by a @xmath31 about the local normal . as in section [ tilt ] , the degeneracy in energy between the red and black flow lines in fig .
[ fig : nematic - plot ] is lifted upon considering the effect of elastic anisotropy generated by a different energy cost for bend and splay .
in this section , we study the stability of liquid crystal ground states to thermal fluctuations @xcite . to explore the fidelity of directional bonds at finite temperatures , we employ a coulomb gas representation of the liquid crystal free energy ( in the one frank constant approximation ) obtained by substituting in eq.([eq : patic - ener ] ) the relation @xmath119 where @xmath120 is the covariant antisymmetric tensor @xcite
, @xmath19 is the gaussian curvature and @xmath121 is the disclination density with @xmath122 defects of charge @xmath123 at positions @xmath124 .
the final result is an effective free energy whose basic degrees of freedom are the defects themselves @xcite : @xmath125 the green s function @xmath126 is calculated ( see appendix [ appa ] ) by inverting the laplacian defined on the sphere @xmath127 and we have suppressed for now defect core energy contributions which reflect the physics at microscopic length scales .
equations ( [ eq : curl - vel - d ] ) and ( [ coulomb model - energy - d ] ) can be understood by analogy to two dimensional electrostatics , with the gaussian curvature @xmath19 ( with sign reversed ) playing the role of a uniform background charge distribution and the topological defects appearing as point - like sources with electrostatic charges equal to their topological charge @xmath123 .
the charge @xmath128 can be defined by the amount @xmath129 increases along a counterclockwise path enclosing the defect s core . on a generic surface ,
the defects tend to position themselves so that the gaussian curvature is screened : typically , the positive ones are attracted to peaks and valleys while the negative ones to the saddles of the surface @xcite .
this geometric potential is ruled out by symmetry on an undeformed sphere since the gaussian curvature is constant .
the gaussian curvature plays the role of a uniform background charge fixing the net charge of the defects consistent with the topological constraint imposed by the poincare - hopf theorem ( see section [ sec : texture ] and references @xcite ) .
the equilibrium positions of the defects are then determined only by defect - defect interactions which are proportional to the logarithm of their chordal distance ( see appendix [ appa ] ) according to @xmath130 \!\!\!\ ! \quad , \label{eq : defectint1}\ ] ] where the integers @xmath131 and @xmath132 describing the singularities associated with each defect , and the integer @xmath74 controls the period @xmath133 of the orientational order parameter .
the `` topological charge '' describing the rotation of the order parameter around each defect is given by @xmath134 .
the geodesic angle @xmath135 subtended by the two defects at positions @xmath136 and @xmath137 can be conveniently recast in terms of their spherical polar coordinates @xmath138 \!\!\!\ !
\label{eq : defectint2}\ ] ] as a simple example , we first consider the case of a @xmath139 colloidal particle ( p=1 ) with two antipodal defects of index @xmath64 .
we can study the effect of thermal disruption of the ground state by setting @xmath140 in eq.([eq : defectint1 ] ) and expanding in the bending angle @xmath129 .
the resulting free energy , apart from an additive constant , reads @xmath141 upon applying the equipartition theorem we obtain in the limit @xmath142
@xcite @xmath143 which describes the fidelity of @xmath144 antipodal `` bonds '' of a divalent colloidal particle .
the effect of thermal fluctuations on the tetrahedral ground state of nematic molecules confined on the sphere ( p=2 and all @xmath145 ) can be studied by means of a normal mode analysis .
the basic results sketched in @xcite were obtained by a slightly different method .
here we describe an alternative treatment in some detail and extend our analysis to hexatic and tetratic defect arrays ( see appendix [ appc ] ) .
we start by defining a generalized array of defect coordinates @xmath146 as a @xmath147 dimensional vector , where @xmath148 is the number of defects in the ground state ( or , equivalently , the valence of the colloidal molecule ) .
@xmath149 in the case of the tetrahedron .
the first @xmath148 entries of the vector @xmath150 are the longitudinal deviations of the @xmath148 defects from a perfect tetrahedral configuration while the remaining @xmath148 components describe defect displacements along the lines of latitude of a sphere of unit radius . as a result
, the deviations of the @xmath151 defect from its equilibrium configuration @xmath152 are parameterized by the two independent components of the vector @xmath150 @xmath153 the relations in eq.([eq : coords ] ) can be used to reexpress eq.([eq : defectint2 ] ) in terms of the components of the displacements vector @xmath123 , with the result , @xmath154 \!\!\!\ ! \quad .
\label{eq : defectint3}\end{aligned}\ ] ] upon substituting eq.([eq : defectint3 ] ) in eq.([eq : defectint1 ] ) , the free energy @xmath49 can be expanded around the equilibrium configuration to quadratic order in @xmath123 with the result ( apart from an additive constant ) @xmath155 where the matrix , @xmath156 , describing the deformation of the tetrahedral molecule is naturally defined as @xmath157_{q_{i } , q_{j } = 0 } \!\!\ ! \quad .
\label{eq : fexp2}\ ] ] the eigenvalues of this matrix can be classified according to the irreducible representation of the symmetry group of the tetrahedron ; their degeneracies can be determined purely from the group theoretical relation @xcite @xmath158 where @xmath159 is the number of frequency degenerate normal modes that transform like the irreducible representation labeled by @xmath160 , @xmath161 is the number of symmetry operations of the tetrahedral point group in the i@xmath162 class , @xmath163 is the total number of symmetry operation in the group , @xmath164 is the character of the i@xmath162 class in the irreducible representation labelled by @xmath160 while @xmath165 is the corresponding character for the reducible representation formed by the defects displacements . the information necessary to apply eq.([eq : group1 ] ) to a tetravalent colloid
is collected in table [ tab : tetrahedron ] .
the top row contains the five symmetry class @xmath166 contained in the tetrahedral point group @xmath167 , corresponding respectively to the identity , three and two fold rotations , four fold rotatory - reflections and reflection through a plane of symmetry @xcite . the number of symmetry operations @xmath161 included in the i@xmath162 class also appears in the top row : thus , @xmath168 where the same ordering used above to list the classes has been adopted .
.[tab : tetrahedron ] character for the irreducible representations of the tetrahedral point group together with the character of the eight - dimensional representation @xmath169 generated by the defect displacements of a tetravalent colloid . [ cols="<,^,^,^,^,>",options="header " , ] upon using eq.([eq : group1 ] ) and the character table [ tab : icosahedron ] we can decompose the 24 dimensional representation , @xmath170 , formed by the displacements from an icosahedral equilibrium configuration into irreducible representations .
the result reads @xmath171 the three rigid body rotations correspond to one of the two triplets in @xmath172 while the remaining 21 independent normal coordinates form energetically degenerate multiplets with the following degeneracy factors : 2 quintets , 2 quartets and 1 triplet .
this analysis is confirmed upon direct diagonalization of the representation @xmath170 which leads the 21 non - vanishing eigenvalues @xmath173 , with the multiplicities shown bold in parenthesis @xmath174 note that the normal modes in the second quintet are much `` softer '' than the rest .
a similar analysis applied to the twisted cube configuration of defects leads to the following decomposition of the ( defects displacements ) representation , @xmath175 , @xmath176 where the three rigid body rotations are contained in one of the two doublets @xmath177 and in the singlet @xmath178 , which leaves five doublets and three singlets for the eigenvalues @xmath179 .
direct diagonalization of @xmath175 leads four doublets , one triplet and two singlets of non - vanishing eigenvalues , @xmath180 the discrepancy between the degeneracies found by direct diagonalization on one hand and group theory on the other is caused by an accidental symmetry of the potential energy of the tilted - cube arrangement of defects .
hence the first triplet is to be interpreted as the missing doublet and singlet that happen to have the same energy even if there is no symmetry reasons to expect so .
the modes in the last doublet of eq.([eq : decomposetilt ] ) are the softest . | the equilibrium texture of nematic shells is studied as a function of their thickness . for ultrathin shells
the ground state has four short @xmath0 disclination lines but , as the thickness of the film increases , a three dimensional escaped configuration composed of two pairs of half - hedgehogs becomes energetically favorable .
we derive an exact solution for the nematic ground state in the one frank constant approximation and study the stability of the corresponding texture against thermal fluctuations . |
the phase diagram of high temperature superconductors in the mixed state has provided an astonishingly broad field to workers in -among other fields- many body problems , polymer physics , low dimensional systems , critical phenomena , and statistical physics in general .
the main reason of this situation is the great number of parameters that define the behavior of the vortex structure . on the other hand
, the same abundance of parameters defining the system turns difficult to find a unified description of all features observed in experiments .
some of the main parameters that define the behavior of the vortex structure are the external magnetic field @xmath4 , temperature @xmath5 , anisotropy @xmath6 , and the disorder ( which produces a non - homogeneous pinning potential for the vortices ) that at this moment we loosely characterize by a parameter @xmath7 .
there are convincing explanations of the main characteristics of different sectors of this multi - dimensional phase diagram , such as the first order melting of the vortex - lattice in clean samples , the continuous melting of a glassy phase in disordered samples , or the existence of two different superconducting transitions ( perpendicular and parallel to @xmath4 ) in some cases ( for a review see ref
however a unified , consistent with experiments description of the problem , even in a qualitative level , is still lacking . in this paper
we propose a qualitative @xmath4-@xmath5-@xmath7-@xmath6 phase diagram of high-@xmath8s , that reproduces most of the available experimental results .
our approach is twofold : we use numerical simulation on the three - dimensional josephson junction array model to study the behavior of the system as a function of @xmath7 and @xmath6 , and show that the dependence on @xmath4 can be deduced from a rescaling of @xmath7 and @xmath6 .
the obtained phase diagram is then rederived using a phenomenological estimation of the free energy @xmath9 of the system for different values of @xmath7 and @xmath6 .
this estimation relies on the existence of two characteristic lengths @xmath10 and @xmath11 parallel and perpendicular to the applied field @xmath4 which are supposed to govern the behavior of the system .
the minimizing of @xmath9 with respect to @xmath10 and @xmath11 allows one to obtain the @xmath12 and @xmath13 functions , which in turn are used to detect the superconducting transitions .
the reminder of the paper is organized as follows . in the next section we present the results of the numerical simulations , and discuss the @xmath7-@xmath6 phase diagram emerging from them . in sec .
iii this phase diagram is qualitatively re - obtained using a proposal for the free energy of the system . in sec .
iv we indicate that a change in the external magnetic field can be interpreted as a movements in the @xmath7-@xmath6 plane , and so the @xmath4-@xmath5 phase diagram for samples with different @xmath7 and @xmath6 can be obtained from the results of the previous sections .
we also compare our results with those found in experimental studies . finally in sec .
v , we summarize and conclude .
our numerical results are based on simulations performed on the three dimensional ( 3d ) josephson junction array ( jja ) model on a stacked triangular network .
each junction is modelled by an ideal josephson junction with critical current @xmath14 shunted by a normal resistance @xmath15 and its attached johnson noise generator , which accounts for the effects of temperature .
an external _ c _ directed magnetic field is included .
the variables that characterize the model are the phase of a superconducting order parameter defined on the nodes of the lattice .
vortices form in the system as singularities in the distribution of these phases .
the details of the model have been discussed elsewhere.@xcite the 3d jja model has been previously used to show the first order melting of the vortex lattice in clean systems.@xcite both thermodynamical and transport signatures of this first order transition were obtained , in close relation to experimental results.@xcite in addition , using the same model we have shown that disorder can destroy the first order transition.@xcite we study here the model further by systematically exploring the case of anisotropic and disordered samples .
we introduce anisotropy by reducing the critical current of the _ c _ axis directed junctions by a factor @xmath16 , and at the same time increasing the _ c _ axis normal resistance by the same factor .
disorder is introduced by randomly varying the critical current of the junctions through the lattice . as vortices gain energy when close to a low critical current region , the effect of randomizing the critical currents is to provide a nearly random pinning potential for vortices .
we characterize the disorder by a parameter @xmath7 which is defined as @xmath17 , where @xmath18 and @xmath19 are the maximum and minimum value of the critical current of the junctions through the sample . the probability distribution between @xmath18 and @xmath19 is taken flat .
we carried out simulations for @xmath20 flux quanta per plaquette .
this is the value used in refs . 4 and 5 .
it produces a ground state ( for a clean sample ) which is commensurate with the subjacent triangular lattice , so no frustration effects are expected .
although the value @xmath21 is rather large and effects of the substrate may be observable , we expect the physics of the problem to be qualitatively well described .
in particular , we assume that for a clean sample the first order melting observed in simulations is in fact the counterpart of the experimental observations.@xcite it would be interesting to perform simulations at lower ( commensurate ) fields , such as @xmath22 or @xmath23 . however , the sample size needed to minimize size effects make the computing time be exceedingly large .
all simulations presented here were performed for @xmath20 , with boundary conditions as in ref .
7 , and for a sample of @xmath24 junctions .
we characterize the superconducting transitions by measuring the resistivity of the sample when a small current ( typically 1/100 of the mean critical current in the corresponding direction ) is applied along the _ ab _ or _ c _ direction .
we will observe two well different behaviors of the resistivities as a function of temperature : in some cases resistivities have a jump from zero to a finite value at a given temperature .
this jump in the resistivity corresponds ( see the discussions in sec .
iv ) to a first order phase transition of the vortex lattice . in some other cases
we will obtain that resistivities as a function of temperature smoothly depart from zero .
we will refer to this behavior as a _
continuous transition_. we do not claim at this point about whether these continuous transitions are or are not real phase transitions , they can be crossovers as well . in sec .
v we present a discussion on this point .
we start by showing in fig .
[ uno](a ) and ( b ) ( full symbols ) the results for the _ ab _ plane and _ c _ axis resistivity of an isotropic sample as a function of temperature for different values of the disorder @xmath7 .
for @xmath25 we obtain a jump in the resistivities at a well defined temperature ( the low temperature tails in @xmath26 are due to surface effects ) .
this temperature is the ( first order ) melting temperature @xmath27 of the system . at @xmath27 the superconducting coherence
is lost discontinuously in all directions .
in other words , vortices passes from a solid phase to an entangled liquid phase .
this situation persists for low values of disorder .
when disorder increases further ( @xmath28 in our simulations ) , the first order transition is lost , and two continuous transitions at different temperatures @xmath29 -for @xmath30- and @xmath31 -for @xmath32- are obtained .
superconducting coherence along the _ ab _ plane is lost at @xmath29 when increasing the temperature , but the system is still superconducting along the _ c _ direction . at the higher temperature @xmath31 the parallel to field superconducting coherence
for @xmath33 the vortex structure corresponds to that of a disentangled vortex liquid .
= 8.7truecm thermodynamical measurements agree with this picture .
we calculated the helicity modulus parallel ( @xmath34 ) and perpendicular ( @xmath35 ) to the field .
helicity modulus measures the influence on the energy of the system of a twist in the boundary conditions , and has to be different from zero to indicate superconducting coherence.@xcite open symbols in fig .
[ uno](a ) and ( b ) show the values @xmath36 that correspond to the resistivity curves .
in the case of the first order transition both @xmath37 and @xmath38 have an abrupt drop at the melting temperature . for the highly disordered case
@xmath38 has a ( smoother ) decrease at @xmath29 , whereas @xmath34 becomes nearly zero at @xmath31 .
the transition at @xmath31 is a percolation phase transition of the vortex structure as discussed in refs .
14 and 12 , that can be characterized by a percolation probability @xmath39 the value of @xmath40 is the probability that a vortex path traversing the sample perpendicularly to the applied field exists .
this value is zero below the transition and one above , with a transition zone that becomes narrower when @xmath41 increases .
the plot of @xmath40 vs temperature is shown in fig .
[ uno](c ) .
we see that @xmath32 is different from ( equal to ) zero if @xmath40 is different from ( equal to ) zero .
= 8.7truecm in fig . [ dos ]
we give the corresponding results for anisotropy @xmath42 .
the first order melting transition occurring for low disorder is similar to that observed for @xmath43 , i.e. , the inter - plane coherence is lost at the same temperature than the in - plane coherence .
however , for higher values of disorder some differences occur : the inter - plane resistive transition occurs ( to our numerical precision ) at the same temperature as the in - plane transition @xmath29 , but the percolation transition temperature @xmath44 ( which for low anisotropy coincides with the resistive inter - plane transition ) occurs for lower temperatures , i.e. , @xmath45 .
this indicates that for high anisotropy , there is a temperature range @xmath46 for which percolation paths across the samples exist , however these path are not mobile ( presumably because they are pinned to the planes , as @xmath47 ) and dissipation is not observed.@xcite from the above discussion on figs .
[ uno ] and [ dos ] , the following picture emerges : at low disorder the vortex lattice melts through a first order phase transition . when disorder increases this transition breaks into two continuous ones , where in - plane and inter - plane coherence are lost at different temperatures @xmath29 and @xmath31 .
whether @xmath29 is larger or lower than @xmath31 depends on the anisotropy of the system . for nearly isotropic samples @xmath29 is lower than @xmath31 , and a disentangled vortex liquid phase exists for @xmath33 . for highly anisotropic samples
@xmath31 is lower than @xmath48 although dissipation along the _ c _ axis is observed only for @xmath49 . in fig .
[ tres ] we show the numerical @xmath7-@xmath6 phase diagram as obtained from simulations .
stars indicate points where the melting transition is first order .
full circles are points where @xmath50 , and hollow circles are points where @xmath45 .
this diagram shows the three different zones discussed above . the continuous line is a sketch of the frontiers between the different zones . in the next section
we show that the main characteristics of this phase diagram can be derived from a simplified description of the vortex structure .
we saw in the previous section that a single first order transition is observed at low disorder , whereas two continuous ones are obtained in disordered samples .
these features suggest that the system can be described as having in general two different transitions , but that in certain cases they can merge onto a single first order transition due to some kind of ` interaction ' . here
we show that this idea can be formulated more precisely by estimating the free energy of the system . to make this estimation we will consider first a single clean plane .
we suppose that the thermodynamics of that plane can be phenomenologically described by a quantity @xmath11 , which is a correlation length : for distances shorter than @xmath11 the system has superconducting coherence , whereas this coherence is lost for distances larger than @xmath11 .
the free energy functional @xmath51 of the plane has the form @xmath52 ( the thermodynamical free energy @xmath9 is obtained by minimizing with respect to @xmath11 ) . for a system of @xmath53 completely decoupled planes , we would have @xmath54 on the other hand , if the coupling between planes is infinite the vortices are rigid lines and we get @xmath55 note that in this case the entropy term does not have the factor @xmath53 because giving the position of the vortices on one plane automatically determines the position of vortices in all other planes .
the term @xmath56 accounts for the energy gain due to the coupling of the planes , @xmath57 being a numerical constant and @xmath6 the anisotropy parameter defined before . in an intermediate situation ( @xmath58 )
the system can be thought as formed by @xmath59 layers ( @xmath10 is a number that satisfies @xmath60 ) . within each layer
the vortices are almost straight lines , whereas correlation is small between different layers . within this picture
the free energy of the system is @xmath61 in the first term , @xmath62 is the number of sites along the _ c_-direction at which the system gains an energy @xmath63 .
the length @xmath10 is a correlation length along the _ c-_direction .
for distances shorter than @xmath10 the system possesses superconducting coherence , whereas this coherence is lost at distances greater than @xmath10 .
the previous estimation of the free energy of the system is too crude . in particular , in the form given by eq .
( [ f4 ] ) , it leads to some unphysical results .
we must modify eq .
( [ f4 ] ) slightly in order to obtain the correct behavior in some limiting cases .
however , we will keep a fundamental property of eq .
( [ f4 ] ) and make the guess that the entropy can be written for the real system as @xmath64 i.e. , as a product of independent functions of @xmath10 and @xmath11 , with @xmath65 and @xmath66 two yet unknown functions .
the basic assumption contained in ( [ entropia ] ) is the following : if the value of @xmath10 is kept fixed , then the system behaves as a two dimensional system with a renormalized temperature .
although this assumption can not be fully justified a priori , it is a natural starting point , and gives sensible results as we will show soon .
we still have to add a term to the free energy which accounts for the effect of impurities .
impurities decrease the energy of the system when vortices pin to them .
if pinning is uncorrelated the energy gain due to pinning becomes lower when the vortex positions are more correlated .
so we add a term to the free energy that increases the energy of the system when @xmath11 and @xmath10 increase .
the most simple term of this type is of the form @xmath67 .
finally , the free energy functional of the system @xmath68 ( dropping an irrelevant constant term ) is @xmath69 the true expressions for the functions @xmath70 , @xmath71 and @xmath72 are difficult to establish .
however , it is not our aim to give a complete quantitative description of the free energy of the system but only a qualitative description of the phase diagram .
we will only ask the functions @xmath70 , @xmath71 and @xmath72 to reproduce some known limiting cases : if @xmath73 ( @xmath11 ) is kept fix we expect the value of @xmath13 ( @xmath74 ) obtained by minimizing @xmath75 , to be smoothly dependent on temperature .
this corresponds to the absence of first order transitions in the dynamics of a single plane ( a single vortex line).@xcite however , when @xmath75 is minimized with respect to _ both _ @xmath10 and @xmath76 a _ discontinuity _ in @xmath12 and @xmath13 can appear , as we show below .
just to give an example we use for the function @xmath70 the form @xmath77 .
the term @xmath78 ( @xmath79 is a numerical constant ) is proportional to the entropy of a single ( isolated ) vortex line .
the constant added assures that the two - dimensional limiting case is reobtained when @xmath80 .
in addition , we will take for the functions @xmath71 and @xmath72 the form @xmath81 and @xmath82 , in such a way that we obtain a form for the free energy functional that is ( unphysically ! ) symmetric ( for @xmath83 ) between _ ab_- and _ c_-directions .
we have tested other forms of the functions @xmath70 , @xmath71 and @xmath72 ( giving the same limiting behavior discussed above ) and found results qualitatively similar to those shown here .
we arrive to our final working expression of the free energy functional ( we will measure @xmath11 and @xmath10 in units of @xmath41 and @xmath53 , respectively , take @xmath84 by rescaling @xmath85 and also rescale @xmath7 ) @xmath86 \left [ \gamma \ln \left ( 1/\xi _ { ab}\right ) + 1\right ] + d\xi _ { ab}\xi _ c \label{f6}\ ] ] 2 the thermodynamical free energy @xmath9 is obtained by minimizing with respect to @xmath11 and @xmath10 : @xmath87 we stress again we do not claim expression ( [ f6 ] ) is a good detailed description of the free energy of the system , but only a expression ( which main characteristic is given by eq .
( [ entropia ] ) ) , that will help us to understand different sectors of the @xmath7-@xmath6 phase diagram of high-@xmath88 s .
we present now the results obtained by minimizing the free energy functional given by expression ( [ f6 ] ) . when doing this minimization we obtain the free energy @xmath9 and the lengths @xmath11 and @xmath10 as a function of temperature .
a first order transition is identified as a discontinuous derivative of @xmath9 , or equivalently a jump in the values of @xmath11 and @xmath89 .
when no jump in @xmath11 and @xmath10 is obtained , the dependence of @xmath11 and @xmath10 with temperature gives a clue on how the superconducting coherence is lost when raising temperature .
however , in this case the identification of a phase transition is not simple , and we can only identify temperature ranges where coherence along _
ab-_directions is high or low .
let us start with the case @xmath25 and @xmath43 ( note that we are using a renormalized anisotropy parameter , so @xmath43 does not imply necessarily an isotropic system ) . in this case eq . ( [ f6 ] ) is symmetric in @xmath76 and @xmath10 , and in fact
the minima of @xmath75 are on the line @xmath90 , so in fig .
[ cuatro ] we plot the function @xmath91 for different values of the temperature .
we see that when @xmath92 the minimum free energy state corresponds to the maximum possible value of @xmath93 , i.e. , the system is in the ordered state . when temperature increases a _ first order _ transition to a disordered state occurs .
this is also seen from the behavior of @xmath93 as a function of temperature , as depicted in fig .
[ cuatro](b ) .
thus we see that the coupling of two continuous transition ( through the entropy term in ( [ f6 ] ) ) can merge them onto a single one , which is first order .
in fact , such sort of mechanism has been previously proposed to occur in other cases , such as two dimensional melting , in which the continuous dislocation - unbinding and disclination - unbinding transitions of the kosterlitsz - thouless - halperin - nelson - young melting theory@xcite can collapse onto a first order melting transition.@xcite = 8.7truecm if disorder is present in the system it will tend to destroy the melting transition . in fig .
[ cinco ] we show results as those of fig .
[ cuatro ] but for a value of @xmath94 . as we see
the jumps in @xmath11 and @xmath10 have disappeared , indicating that the transition is not first order .
if the anisotropy had been chosen different from one , then the temperature at which @xmath11 and @xmath10 take a given value would have been different .
although we can not characterize from our simplified model a phase transition when @xmath11 and @xmath10 are continuous functions of temperature , it is tempting to say that if @xmath11 ( @xmath10 ) drops to zero at lower temperature than @xmath10 ( @xmath11 ) , then we are in a sector of the phase diagram where @xmath50 ( @xmath45 ) . in this way
we generate the phase diagram depicted in fig .
[ seis ] . as indicated , it is qualitatively similar to the one obtained from the numerical simulation , and it gives support _ a posteriori _ to the proposal of an entropy of the system of the form given by eq .
( [ entropia ] ) .
some characteristics of this phase diagram can be analyzed in simple terms .
for example , when @xmath95 the system is a set of decoupled planes , and no first order transition is obtained for any value of @xmath96 when @xmath97 vortices are rigid lines and in fact effectively two - dimensional , so a first order transition is not obtained in this case either .
this limiting cases give some insight on the form of the border between first - order and continuous transitions in figs .
[ tres ] and [ seis ] .
there is an optimum value of the anisotropy , at which the first order transition persists up to a highest value of disorder .
this optimum value depends on the thickness of the sample .
in fact , as we discussed in a previous work@xcite the temperature @xmath31 logaritmically _ decreases _ when the thickness of the sample increases .
this means that in fig .
[ tres ] the border between the zones with @xmath98 and @xmath99 moves to lower values of @xmath6 .
it is thus likely that the optimum value of @xmath6 for the occurrence of the first order transition also decreases with sample thickness .
having discussed the @xmath7-@xmath6 phase diagram for a fixed magnetic field @xmath4 , we turn now to the discussion of the dependence on @xmath4 . from the numerical point of view
the direct approach would be to do simulations at different fields .
however , as we discussed above , to reduce the magnetic field to other commensurate values would require exceedingly large computing time . fortunately , there are arguments that suggest that a change of @xmath4 can be mapped onto a change of @xmath7 and @xmath6 .
= 8.7truecm the scaling combination between magnetic field and anisotropy has been given by chen and teitel.@xcite we generalize here the argument to include the disorder parameter @xmath7 .
our dimensionless temperature ( measured in units of the mean josephson energy of the in - plane junctions ) can be only a function of the other dimensionless parameters of the system .
these are @xmath7 , @xmath4 , and @xmath6 .
these parameters have different dependences on the coherence length @xmath100 .
if we identify the discretization parameter in the _ ab _ plane with a distance of the order of the coherence length @xmath101 then the critical current of the josephson junctions along the _ c _ direction is proportional to @xmath102 so the anisotropy parameter @xmath6 behaves as @xmath103 on the other hand , our dimensionless magnetic field @xmath4 is given in terms of the real external magnetic field @xmath104 by the expression @xmath105 ( @xmath106 is the flux quantum ) . for the parameter @xmath7
, we note that @xmath7 is proportional to the amplitude of the pinning potential . a vortex averages
this random function on an area @xmath107 considering the case of random ( uncorrelated ) pinning we find that @xmath7 depends on @xmath100 as @xmath108 .
since we are ignoring details of the vortex cores , we expect the @xmath100 dependence to cancel out , and our temperature transition to be only a function of the @xmath100-independent quantities @xmath109 and @xmath110 .
we conclude that we can obtain the behavior of the system as a function of magnetic field from the results of fig .
[ tres ] on lines with constant @xmath111 .
a sketch of the different possibilities is shown in fig .
[ siete ] .
the general prediction from fig .
[ tres ] is depicted in fig .
[ siete](a ) .
the scales on the axis as well as the extent of the first order transition depend on the particular value of @xmath112 .
this general picture has to be modified at very low fields .
in fact , a minimum crossover field@xcite ( given essentially as the field at which the vortex lattice parameter matches the thickness of the sample ) exists , below which @xmath29 and @xmath31 are essentially the same ( this is because in this case there are so few vortices in the sample that the transition is enterally due to thermally generated vortex loops ) .
different possibilities are depicted in fig .
[ siete ] ( b ) , and ( c ) .
they correspond to different ranges of values of @xmath111 .
[ siete](b ) correspond to @xmath113 small , so fig . [ tres ] predicts @xmath45 at high fields , and a first order zone at intermediate and low fields .
this phase diagram corresponds to that experimentally obtained for bi@xmath0sr@xmath0cacu@xmath0o,@xcite which in fact has the largest value of @xmath85 and even to the case of clean yba@xmath0cu@xmath2o@xmath3,@xcite which has a very low @xmath7 .
[ siete](c ) shows the expected phase diagram for samples with a high value of @xmath113 . at low fields
@xmath29 is lower than @xmath114 whereas at high fields a cross to a zone with @xmath98 is possible .
no first order zone is shown because a curve defined by a high value of @xmath113 in fig .
[ tres ] do not pass through the first order zone .
the low field part of this phase diagram corresponds to the one obtained for yba@xmath0cu@xmath2o@xmath3 samples with defects.@xcite the crossover to a case with @xmath50 has not been observed , presumably because of the high fields needed .
= 8.7truecm as we indicated above , when @xmath50 superconducting coherence is lost along the _ ab-_plane at lower temperatures than along the _ c_-axis .
for @xmath33 finite resistance within the planes and zero resistance in the _
c_-direction is observed .
when @xmath98 and in the case that @xmath115 we potentially expect finite resistance in the _ c_-direction and zero resistance within the planes .
this has turned to be difficult to find , both experimentally @xcite and in our simulations .
the resistance seems rather to go to zero at the same temperature @xmath29 in all directions.@xcite this is due to the fact that vortex paths crossing the sample for @xmath116 are pinned to the _ ab_-planes as long as @xmath47 , preventing their movement ( and thus dissipation ) .
however , it is worth noting that some other experimental measurements of coherence ( ac magnetization ) indicate@xcite that in fact , _ c_-axis coherence is lost at lower temperatures than in - plane coherence for the case of bi@xmath0sr@xmath0cacu@xmath0o .
the transitions observed in our simulations are of different character , and we want to discuss the point a bit further .
the first order transition is the easiest to characterize numerically .
although we do not show all the results , we observed that when the resistivity has a jump other indicators point to a first order phase transition , among them the existence of hysteresis in the resistivity curves upon heating and cooling , and the fact that right at the transition temperature , the energy histogram of the system has two peaks , indicating two coexisting phases with an energy barrier separating them.@xcite the continuous transitions are more difficult to characterize .
the transition at @xmath29 _ is not _ a phase transition in our model .
in fact , it is a crossover due to thermal deppining of rather independent vortices.@xcite however , in a real sample it may correspond to the vortex glass transition , depending on the strength of the disorder.@xcite when @xmath99 , we have previously characterized the transition at @xmath31 as a percolation phase transition of the vortex structure perpendicularly to the applied field.@xcite in the thermodynamic limit for @xmath41 ( @xmath117 ) the system does not have any vortex line running perpendicularly to the applied field for @xmath118 , whereas for @xmath116 these paths extend all over the _ ab _ plane with probability one . in references
[ 14,12 ] we showed numerical evidence suggesting that this transition is a second order phase transition and gave its critical exponents as found from simulations .
in this paper we presented numerical evidence that supports an anisotropy - disorder phase diagram of the vortex structure of high-@xmath119 with the following characteristics : for clean samples the vortex lattice melts through a first order phase transition for a wide range of anisotropies . when disorder is included
the behavior of the system is strongly dependent of the anisotropy . for low anisotropies
the in - plane coherence is lost at a temperature @xmath29 lower than the temperature @xmath31 at which inter - plane coherence is lost , and a zone of disentangled vortex lines is observed for @xmath33 . for highly anisotropic samples the superconducting coherence as deduced from simulations of the resistivity
is lost at the same temperature @xmath29 within the planes and perpendicularly to the planes . however , in this case the vortex structure percolates at a temperature @xmath31 well below @xmath29 . in this case
the system for @xmath46 is in an ` entangled solid ' phase .
these features are also obtained from an estimation of the free energy of the system which is mainly based on a proposal for the entropy of the system ( eq .
( [ entropia ] ) ) .
we showed that the magnetic field - temperature behavior of the system can be deduced from results obtained from a fixed magnetic field provided the anisotropy and disorder present in the system are properly rescaled .
our results present in a unified way , different characteristics of the vortex structure that had been previously found in partial studies .
the analysis is in agreement with a variety of experiments performed on different materials with a broad range of parameters such as disorder , anisotropy and magnetic field .
it could prove to be useful to find a more solid base of our proposal for the free energy of the system -that we showed is qualitative good- in order to obtain more detailed analytical results .
we thank helpful discussions with d. lpez , e. f. righi and f. de la cruz . e.a.j . acknowledges financial support by conicet .
is partially supported by conicet . in ref .
12 we showed some results on anisotropic cubic lattices where a zone with @xmath120 and @xmath121 was observed .
this dissipation is probably due to the movement of percolation paths lying entirely between two consecutive planes , and thus being unaffected by the pinning of the _ ab _ planes .
we did not observe clearly the same effect on triangular lattices up to now .
a first order transition has been reported for a single plane ( w. yu , ph d thesis , ohio state university ( 1994 ) , and also m. franz and s. teitel , phys .
b * 51 * , 6551 ( 1995 ) ) . however , this transition is very weakly first order , and does not have inmediate relation with the one occurring on three - dimensional systems , where a strong first order transition is observed only if the thichness of the sample is higher than a critical value ( see reference @xcite and also g. carneiro , phys .
lett . * 75 * , 521 ( 1995 ) ) . | we propose a phase diagram for the vortex structure of high temperature superconductors which incorporates the effects of anisotropy and disorder .
it is based on numerical simulations using the three - dimensional josephson junction array model .
we support the results with an estimation of the internal energy and configurational entropy of the system .
our results give a unified picture of the behavior of the vortex lattice , covering from the very anysotropic bi@xmath0sr@xmath0cacu@xmath0o@xmath1 to the less anisotropic yba@xmath0cu@xmath2o@xmath3 , and from the first order melting ocurring in clean samples to the continuous transitions observed in samples with defects . 2 |
the science surrounding the origin of life presents an obvious difficulty to scientists : our data consist of a single example .
it is unknown whether the events that followed the origin of life are deterministic consequences of life or unique to this specific origin .
one question of interest is whether all life , or specifically all self - replicators , are evolvable .
evolvability ( the ability to evolve ) has many similar , but differing definitions @xcite ; here , we define it as the ability to increase in fitness .
we also distinguish between two possibilities for such fitness increases : _ optimization_defined as the ability to improve an already - present phenotypic trait , and _ innovation_the ability to evolve novel phenotypic traits .
a common assumption is that life originated with self - replicating rna molecules @xcite .
thus , most empirical studies have focused on rna replicators , either emergent replicators @xcite or those created by experimenters @xcite .
experiments ( both computational and biochemical ) have also explored the evolvability of rna replicators , usually involving extensive mapping of their fitness landscapes @xcite .
while rna is an enticing candidate for a pre - biotic molecule , the so - called rna world hypothesis " has its own problems foremost the asphaltization " that tends to befall formose carbohydrates under the expected conditions of an early earth @xcite which has led researchers to explore other origins of life not necessarily rna - based @xcite , or even to move the origin of life from earth to mars @xcite . in recent years
, more and more theoretical models concerning the origin of life have focused on exploring the abstract concepts that could possibly be involved in any potential origination , independent of a particular biochemical system @xcite .
the field of artificial life is ideally suited to study various possibilities for the origin of life as it imagines life as it could be , not just as it is .
the question of the random emergence of replicators has been addressed in various digital systems before @xcite , while theoretical models have explored the factors that lead to differing evolvability @xcite .
artificial life tools have also been used to explore the potential of evolvability in different systems @xcite .
recently , adami used information theory to calculate the likelihood of the random emergence of a self - replicator in a sense , the progenitor to life without regards to a specific biochemistry @xcite .
adami and labar tested this theory with avida by generating billions of random avidian sequences and checking for their ability to replicate themselves @xcite . such an investigation is akin to studying the chance emergence of self - replicating rna molecules @xcite . in previous work
we explored the relationship between evolvability and self - replication using these emergent replicators @xcite , and found that almost all emergent replicators were evolvable , both in terms of optimization for replication and in terms of evolving beneficial phenotypic innovations .
we also discovered that these replicators came in two forms : proto - replicators and self - replicators .
proto - replicators deterministically copy themselves inaccurately , but eventually evolve into self - replicators ; self - replicators , on the other hand , produce an exact copy of themselves in the absence of stochastic mutation .
we also noted the possibility of an optimization - innovation trade - off in some of these replicators , especially the default avida ancestor ( a hand - written self - replicator specifically designed for evolution experiments ) . here , we extend our previous study and test a fundamental question concerning life s origins : how does the genetic composition of the first replicator determine the future evolution of life ?
one extreme possibility is that all emergent replicators have similar genetic compositions and thus the future evolution of life will occur in a similar manner , no matter the progenitor . on the opposite end of the spectrum is the possibility that every emergent replicator is sufficiently different from every other replicator in genetic composition . in this case , the future outcome of life may be entirely dependent on which replicator emerges first . in experiments with the artificial life system avida , we find that the interplay between the genetic composition of the first emergent replicator and future evolutionary outcomes is between these two extremes .
emergent replicators can be classified into two distinct classes based on a functional analysis of their replication machinery .
these classes differ in their ability to optimize their replication ability .
however , those replication classes that display high evolvability towards optimizing replication also demonstrate low evolvability towards evolving novel phenotypic traits , and vice versa .
finally , we show that this difference in evolvability is due to differences in the replication machinery between the different replicator classes
in order to study the interplay between emergent replicators and evolvability , we used the avida digital evolution platform @xcite . in avida , simple computer programs ( avidians " ) in a population compete for the memory space and cpu time needed to replicate themselves .
each avidian consists of a genome of computer instructions , where each locus in the genome can be one of 26 possible instructions .
contained with each genome are the instructions necessary for the avidian to allocate a daughter genome , copy its genome into this new daughter genome , and divide off the daughter genome .
as the replication process is mechanistic ( i.e. , avidians execute their genome s instructions , including those to replicate , sequentially ) , the speed at which replication occurs is also genome dependent
. therefore , fitness is genome - dependent , as an organism s ultimate success is determined by its replication speed in these simple environments .
as avidians directly copy and pass their genomes to their daughters , avida populations also have heredity . during the copying process
, errors may be introduced , resulting in mutations and population variation .
therefore , avida populations possess heredity , variation , and differential fitness : they are an _ instantiation _ ( as opposed to a simulation ) of darwinian evolution @xcite this simpler instantiation of evolution by natural selection has allowed for the exploration of many topics hard to test in biological systems , see e.g. @xcite . avida has been explained in greater detail elsewhere ( see @xcite for a full description ) ; here , we will cover the details relevant for this study .
an avidian consists of a variety of elements : a genome of instructions , a
read - head " that marks which instruction should be copied , a
write - head " that denotes where the read - head - marked instruction should be copied to , and three registers to hold integer numbers ( ax , bx , and cx ) , among other elements . in order to undergo reproduction
, an avidian needs to perform four operations in the following order .
first , it must execute an _ h - allocate _ instruction which allocates a fixed number of _ nop - a _ instructions to the end of its genome ( here 15 , as we work exclusively with length 15 genomes ) .
these _ nop - a _ instructions are inert by themselves , and serve as placeholders to be replaced by actual information . allocation of this memory prepares the daughter genome space to receive the information from the parent .
next , the read - head and write - head must be set 15 instructions apart , allowing for the instructions in the parent genome to be copied into the daughter genome .
these operations are algorithmically similar to creating a dna replication fork at the origin of replication , preparing for the assembly of a copied sequence . following this step
, the genome must find a way of looping over the _ h - copy _ instruction in the genome to actually copy instructions from the parent into the daughter genome , similar to the action of a dna polymerase fusing the new ( daughter ) nucleotides on the former parent strand .
this copying can be done by either looping through the entire genome ( using the circular nature of the genome to re - use a single _ h - copy _ command as many times as necessary to copy all instructions ) or else by continuously looping over a smaller set of instructions in the genome ( called the copy loop " ) .
the latter algorithm requires marking the set of instructions ( the replication gene " ) so as to control the forking of execution flow .
finally , an avidian must execute an _ h - divide _ instruction to divide the duplicated genome into two avidians , and thus successfully reproduce . in the experimental design used here , we used two different mechanisms to produce mutations during replication .
the first mechanism produces at most one _ point mutation _ at a random locus in the genome at a rate of 0.15 mutations per genome per generation upon successful division .
the second mechanism is a deterministic ( or incipient " ) mutation , which occurs when the instructions in an avidian cause it to copy instructions into the daughter genome in the wrong place . often , this results in a daughter genome with one or two _ nop - a _ instructions at the beginning if these two loci were skipped over during the copying process , remaining instead in their pristine form .
in many cases , such a faulty copy algorithm results in the offspring being non - viable ; however , in some cases viable offspring are produced . sometimes , this incipient mutation mechanism results in the phenomenon of _ proto - replicators _ , defined as those replicators that deterministically ( i.e. , reproducibly because genetically controlled ) make an offspring different from itself @xcite .
replicators that instead deterministically make an identical offspring are called self - replicators . in the standard experimental evolution scenario ,
avidians are placed into a landscape that rewards a variety of phenotypic traits .
these traits commonly refer to the ability to perform certain boolean logic calculations on binary numbers that the environment provides . during an avidian s lifespan , they can input ( read from ) and output ( write to ) binary numbers from / to the environment . whenever a number is written , the avida world checks if a boolean logic calculation was performed .
successful performance results in an increase in the replication speed of that individual s descendants . in the standard avida design ( also the one used here in the experiments to explore evolvability in the sense of innovation ) that environment rewards the performance of nine different calculations .
this environment is usually referred to as the logic-9 " environment and rewards calculations such as not and equals @xcite .
the more complex the calculation performed , the greater the replication speed increase . the performance of such logic calculations can be viewed as an algorithmic analogue of performing different metabolizing reactions using different sugar resources ( the binary numbers provided by the environment ) . to study the evolvability of emergent replicators , we first had to generate a collection of emergent replicators .
we re - analyzed a list of 10@xmath6 random avidian length-15 genomes generated previously @xcite . in order to discover replicators of fixed genome size
, we set the offspring size range parameter in the main avida configuration file to 1.0 , which guarantees that the _ allocate _ command allocates exactly as much space as needed by the daughter genome . for our focus on replicators of length 15
, this guaranteed that exactly 15 instructions are allocated for the daughter to receive next , we looked at the relative abundance of proto - replicators compared to true self - replicators . we re - analyzed the above set of replicators , but made one additional parameter change : we set require exact copy to 1 .
any replicator that registered as having zero fitness in this analysis would deterministically copy its genome inaccurately , and would be classified as proto - replicators . any replicators that could reproduce themselves accurately under this treatment
were classified as self - replicators . in order to test the evolvability of the these replicators , we performed three different experimental tests . for all experiments
, we used a population size of 3600 .
individuals offspring were placed in any random cell in the 60x60 grid , thus mimicking a well - mixed environment .
point mutations were applied upon division and the genomic mutation rate was 0.15 mutations per generation .
genome size was fixed at 15 instructions .
the first experiment tested the replicators evolvability in the sense of replication optimization ( optimization experiments ) . in this experiment , each replicator was used to seed 10 populations ; these populations evolved for 10@xmath7 generations .
phenotypic traits were not under selection in these experiments , that is , evolution only optimized the replication machinery .
next , we performed evolution experiments ( referred to as the innovation " experiments ) where we rewarded individuals for evolved phenotypic traits ( i.e. , logic tasks ) besides any increase in replicatory prowess .
we ran the innovation experiments for 10@xmath8 generations ( an order of magnitude longer than the optimization experiments ) .
these experiments were designed to decrease the likelihood that our data resulted from one of the replicator classes taking longer to evolve complex traits . finally , we repeated the innovation experiments , but used each ancestral replicator s fittest descendant from the optimization experiments to initialize the populations ( trade - off experiments ) .
these experiments tested the presence of an optimization innovation trade - off .
once we determined the set of replicators , we first tried to determine if there were similarities in the genomes of the replicators .
we realized that many of the genomes had similar instruction motifs within the sequence . to test whether different replication algorithms existed in different replicators , we used avida s trace function to analyze the step - by - step execution of the replicators genomes .
this allowed us to cluster the emergent replicators into two main distinct functional replicator classes . to examine the results from the evolution treatments
, we analyzed the most abundant genotype at the end of each experiment .
the data collected included the final evolved fitness and the number of evolved phenotypic traits .
all avida analysis was performed using avida s analyze mode with settings matching those under which the relevant experiments were performed .
statistical analyses were performed using the numpy , scipy , and pandas python modules @xcite .
figures were generated using the matplotlib python module @xcite .
among the one billion randomly - generated genomes , we found 75 genomes that could replicate themselves when genome size was fixed .
of these replicators , 22 were true self - replicators in the sense that they could perfectly copy their genomes even when mutations were turned off .
the remaining 53 replicators were proto - replicators " in the sense that they could _ not _ produce a perfect copy of their genome when the mutation rate was set to zero .
this deterministic miscopying is due to the specific nature of a proto - replicator s genome ( see methods ) .
however , these replicators still produced viable offspring that would eventually lead to a self - replicator .
the discovery of these proto - replicators extends the definition of proto - replicators from @xcite to include fixed - length proto - replicators . upon examination of these replicators
, we detected the presence of distinct instruction motifs in their genomes .
these motifs consist of instructions involved in genome replication and suggested that different replicators used different replication mechanisms . to explore this possibility
, we performed a step - by - step analysis of each replicator s lifestyle by looking at the execution of their genome s instructions .
avidians must perform four steps to successfully reproduce : allocate a blank daughter genome , separate their read- and write - heads , copy their genome into the daughter genome through some looping process , and divide off the daughter genome ( see methods for a fuller description ) .
we were able to cluster the replicators into two replication classes based on a difference in two traits : ( 1 ) how they separated their read- and write - heads to copy their genome and ( 2 ) how they looped through their genome in order to copy their genome .
we named the first class of replicators hc " replicators because of the hc instruction motif they all share ( see table [ avidainst ] for the avida instructions and their descriptions ) .
this class contains 62 replicators ( 9 self - replicators ) .
only 8 of these replicators have a standard copy loop ( see methods for a definition of a copy loop ) , which appears at the end of their genomes ; these copy loops were marked by the presence of a _ mov - head _
( g ) instruction . | the role of historical contingency in the origin of life is one of the great unknowns in modern science .
only one example of life exists one that proceeded from a single self - replicating organism ( or a set of replicating hyper - cycles ) to the vast complexity we see today in earth s biosphere .
we know that emergent life has the potential to evolve great increases in complexity , but it is unknown if evolvability is automatic given any self - replicating organism . at the same time , it is difficult to test such questions in biochemical systems . laboratory studies with rna replicators have had some success with exploring the capacities of simple self - replicators , but these experiments are still limited in both capabilities and scope . here , we use the digital evolution system avida to explore the interaction between emergent replicators ( rare randomly - assembled self - replicators ) and evolvability .
we find that we can classify fixed - length emergent replicators in avida into two classes based on functional analysis .
one class is more evolvable in the sense of optimizing their replication abilities .
however , the other class is more evolvable in the sense of acquiring evolutionary innovations .
we tie this trade - off in evolvability to the structure of the respective classes replication machinery , and speculate on the relevance of these results to biochemical replicators .
@xmath0 department of microbiology & molecular genetics + @xmath1 beacon center for the study of evolution in action + @xmath2 program in ecology , evolutionary biology , and behavior + @xmath3 department of integrative biology + @xmath4 department of computer science and engineering + @xmath5 department of physics and astronomy + michigan state university , east lansing , mi 48824 + |
open clusters provide unique opportunities in astrophysics .
stars in open clusters are believed to be formed from the same cloud of gas at roughly the same time .
the fewer free parameters available to model cluster stars make them interesting targets to analyze as a uniform ensemble , especially for asteroseismic studies .
asteroseismology is an elegant tool based on the simple principle that the frequency of a standing acoustic wave inside a star depends on the sound speed , which in turn depends on the physical properties of the interior .
this technique applied to the sun ( helioseismology ) has provided extremely detailed knowledge about the physics that governs the solar interior , ( e.g. , * ? ? ? * ) .
all cool stars are expected to exhibit solar - like oscillations of standing acoustic waves called p modes that are stochastically driven by surface convection . using asteroseismology to probe the interiors of cool stars in clusters , therefore ,
holds promise of rewarding scientific return @xcite .
this potential has resulted in several attempts to detect solar - like oscillations in clusters using time - series photometry .
these attempts were often aimed at red giants , since their oscillation amplitudes are expected to be larger than those of main - sequence or subgiant stars due to more vigorous surface convection . despite these attempts ,
only marginal detections have been attained so far , limited either by the length of the time series usually achievable through observations with the _ hubble space telescope _
@xcite or by the difficulty in attaining high precision from ground - based campaigns ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in this letter
we report clear detections of solar - like oscillations in red - giant stars in the open cluster ngc 6819 using photometry from nasa s _ kepler mission _ @xcite . this cluster , one of four in the _ kepler _ field , is about 2.5 gyr old .
it is at a distance of 2.3 kpc , and has a metallicity of [ fe / h ] @xmath2 ( see * ? ? ?
* and references herein ) .
the data were obtained between 2009 may 12 and june 14 , i.e. , the first 34 days of continuous science observations by _ kepler _ ( q1 phase ) .
the spacecraft s long - cadence mode ( @xmath3minutes ) used in this investigation provided a total of 1639 data points in the time series of each observed star .
for this letter we selected 47 stars in the field of the open cluster ngc 6819 with membership probability @xmath4% from radial velocity measurements @xcite .
figure [ f1 ] shows the color - magnitude diagram ( cmd ) of the cluster with the selected stars indicated by green symbols .
the eleven annotated stars form a representative subset , which we will use to illustrate our analyses in sections [ extract ] and [ membership ] .
we selected the stars in this subset to cover the same brightness range as our full sample , while giving high weight to stars that appear to be photometric non - members ( i.e. , stars located far from the isochrone in the cmd ) .
data for each target were checked carefully to ensure that the time - series photometry was not contaminated significantly by other stars in the field , which could otherwise complicate the interpretation of the oscillation signal .
fourteen data points affected by the momentum dumping of the spacecraft were removed from the time series of each star .
in addition , we removed points that showed a point - to - point deviation greater than @xmath5 , where @xmath6 is the local rms of the point - to - point scatter within a 24 hour window .
this process removed on average one data - point per time series .
finally , we removed a linear trend from each time series and then calculated the discrete fourier transform .
the fourier spectra at high frequency have mean levels below 5 parts per million ( ppm ) in amplitude , allowing us to search for low - amplitude solar - like oscillations .
figure [ f2 ] shows the fourier spectra ( in power ) of 9 stars from our subset . these range from the lower red - giant branch to the tip of the branch ( see figure [ f1 ] ) .
the stars are sorted by apparent magnitude , which for a cluster is indicative of luminosity , with brightest at the top .
note that the red giants in ngc 6819 are significantly fainter ( @xmath7 ) than the sample of _ kepler _ field red giants ( @xmath8 ) studied by @xcite . nevertheless , it is clear from figure [ f2 ] that we can detect oscillations for stars that span about two orders of magnitude in luminosity along the cluster sequence .
we used four different pipelines @xcite to extract the average frequency separation between modes of the same degree ( the so - called large frequency separation , @xmath0 ) .
we have also obtained the frequency of maximum oscillation power , @xmath9 , and the oscillation amplitude .
the measured values of @xmath0 are indicated by vertical dotted lines in figure [ f2 ] centered on the highest oscillation peaks near @xmath9 . while the stars in figure [ f2 ] , particularly in the lower panels , show the regular series of peaks expected for solar - like oscillations , the limited length of the time - series data does not allow such structure to be clearly resolved for the most luminous stars in our sample
those with @xmath9 @xmath10@xmath11hz .
we do , however , see humps of excess power in the fourier spectra ( see figure [ f2 ] star no . 2 and 8) with @xmath9 and amplitude in mutual agreement with oscillations . with longer time series we expect more firm results for these high - luminosity giants .
it is immediately clear from figure [ f2 ] that not all stars follow the expected trend of increasing @xmath9 with decreasing apparent magnitude , suggesting that some of the stars might be intrinsically brighter or fainter than expected .
since oscillations in a star only depend on the physical properties of the star , we can use asteroseismology to judge whether or not a star is likely to be a cluster member independently of its distance and of interstellar absorption and reddening . for cool stars , @xmath9 scales with the acoustic cut - off frequency , and it is well established that we can estimate @xmath9 by scaling from the solar value @xcite : @xmath12 where @xmath13hz .
the accuracy of such estimates is good to within 5% @xcite assuming we have good estimates of the stellar parameters @xmath14 , @xmath15 , and @xmath16 . in the following
we assume the idealistic scenario where all cluster members follow standard stellar evolution described by the isochrone .
stellar mass along the red giant branch of the cluster isochrone varies by less than 1% .
the variation is less than 5% even if we also consider the asymptotic giant branch . for simplicity
, we therefore adopt a mass of @xmath17 for all stars , which is representative for the isochrone from @xcite ( figure [ f1 ] ) and a similar isochrone by @xcite . neglecting binarity ( see table [ tab1 ] ) , we derive the luminosity of each star in our subset from its @xmath18-band apparent magnitude , adopting reddening and distance modulus of @xmath19 and @xmath20 , respectively ( obtained from simple isochrone fitting , see @xcite ) .
we used the calibration of @xcite to convert the stellar @xmath21 color to @xmath16 .
bolometric corrections were also taken from @xcite .
the derived quantities were then used to estimate @xmath9 for each star ( eq . [ numaxsc ] ) , and compared with the observed value ( see figure [ f3 ] ) .
figure [ f3 ] shows four obvious outliers ( no . 1 , 3 , 8 and 11 ) , three of which are also outliers in the cmd ( no . 1 , 3 , and 11 ) .
for the rest of the stars we see good agreement between the expected and observed value , indicating that the uncertainty on the @xmath9 estimates are relatively small .
since the variations in mass and effective temperature among the cluster giant stars are small , deviations from the dotted line must be caused by an incorrect estimate of the luminosity .
this implies that the luminosities of stars falling significantly above or below the line have been over- or underestimated , respectively .
the simplest interpretation is that these outliers are fore- or background stars , and hence not members of the cluster . to explain the differences between the observed and expected value of @xmath9 would require the deviant stars to have @xmath18 errors of more than 1 magnitude , and in some cases @xmath22 errors of about 0.2 magnitude if they were cluster members .
binarity may explain deviations above the dotted line , but only by up to a factor of two in @xmath15 ( and hence , in the ratio of the observed to expected @xmath9 ) .
the deviation of only one star ( no.1 ) could potentially be explained this way .
however , that would be in disagreement with its single - star classification from multi - epoch radial velocity measurements , assuming it is not a binary viewed pole - on ( see table [ tab1 ] ) . hence , under the assumption of a standard stellar evolution , the most likely explanation for all four outliers in figure [ f3 ] is therefore that these stars are not cluster members .
this conclusion is , however , in disagreement with their high membership probability from measurements of radial velocity @xcite and proper motion @xcite ( see table [ tab1 ] ) .
another interesting possibility is that the anomalous pulsation properties might be explained by more exotic stellar evolution scenarios than is generally anticipated for open - cluster stars .
it is clear from figure [ f2 ] that the amplitudes of the oscillations increase with luminosity for the seismically determined cluster members .
based on calculations by @xcite , @xcite have suggested that the photometric oscillation amplitude of p modes scale as @xmath23@xmath16@xmath24 , with @xmath25 ( the velocity amplitudes , meanwhile , would scale as @xmath26 ) .
this was revised by @xcite to @xmath27 based on models of main sequence stars .
taking advantage of the fewer free parameters within this ensemble of stars , our observations allow us to make some progress towards extrapolating this scaling to red giants and determining the value of @xmath28 . in figure [ f4 ]
we introduce a new type of diagram that is similar to a cmd , but with magnitude replaced by an asteroseismic parameter in this case , the measured oscillation amplitude .
amplitudes were estimated for all stars in our sample ( except for the four outliers ) using methods similar to that of @xcite ( see also * ? ? ?
* ) , which assume that the relative power between radial and non - radial modes is the same as in the sun .
this diagram confirms the relationship between amplitude and luminosity . despite a large scatter , which is not surprising from this relatively short time series ,
we see that @xmath27 provides a much better match than @xmath29 .
once verified with more data , this relation will allow the use of the measured amplitude as an additional asteroseismic diagnostic for testing cluster membership and for isochrone fitting in general .
we note that the other clusters observed by _ kepler _ have different metallicities than ngc 6819 , which will allow future investigation on the metallicity dependence of the oscillation amplitudes .
we expect to obtain less scatter in the asteroseismic measurements when longer time series become available .
that will enable us to expand classical isochrone fitting techniques to include diagrams like this , where amplitude could also be replaced by @xmath9 or @xmath0 .
in particular , we should be able to determine the absolute radii aided by @xmath0 of the red giant branch stars , which would be an important calibrator for theoretical isochrones .
additionally , the distributions of the asteroseismic parameters such as @xmath9 can potentially be used to test stellar population synthesis models @xcite . applying this approach to clusters
could lead to further progress in understanding of physical processes such as mass loss during the red - giant phase ( see e.g. , * ? ? ?
note that a few clear outliers are indicative of non - membership or exotic stellar evolution , as a result of factors such as stellar collisions or heavy mass loss , while a general deviation from the theoretical predictions by a large group of stars would suggest that the standard theory may need revision .
finally , we note that ngc 6819 and another _ kepler _ cluster , ngc 6791 , contain detached eclipsing binaries @xcite . for these stars masses and radii can be determined independently @xcite , which will further strengthen results of asteroseismic analyses .
photometric data of red giants in ngc 6819 obtained by nasa s _ kepler mission _ have enabled us to make the first clear detection of solar - like oscillations in cluster stars .
the general properties of the oscillations ( @xmath0 , @xmath9 , and amplitudes ) agree well with results of field red giants made by _ kepler _
@xcite and corot @xcite .
we find that the oscillation amplitudes of the observed stars scale as @xmath30@xmath16@xmath24 , suggesting that previous attempts to detect oscillations in clusters from ground were at the limit of detection .
we find that the oscillation properties provide additional tests for cluster membership , allowing us to identify four stars that are either non - members or exotic stars .
all four stars have membership probability higher than 80% from radial - velocity measurements , but three of them appear to be photometric non - members .
we further point out that deviations from the theoretical predictions of the asteroseismic parameters among a large sample of cluster stars have the potential of being used as additional constraints in the isochrone fitting process , which can lead to improved stellar models .
our results , based on limited data of about one month , highlight the unique potential of asteroseismology on the brightest stars in the stellar clusters observed by _
kepler_. with longer series sampled at the spacecraft s short cadence ( @xmath31 1 minute ) , we expect to detect oscillations in the subgiants and turn - off stars , as well as in the blue stragglers in this cluster .
funding for this discovery mission is provided by nasa s science mission directorate .
the authors would like to thank the entire _ kepler _ team without whom this investigation would not have been possible .
the authors also thank all funding councils and agencies that have supported the activities of working group 2 of the _ kepler _ asteroseismic science consortium ( kasc ) . , a. , montalbn , j. , eggenberger , p. , hekker , s. , & noels , a. 2009 , in american institute of physics conference series , vol . 1170 , american institute of physics conference series , ed . j. a. guzik & p. a. bradley , 132 lcccccc i d & i d & wocs i d & i d & mem.ship & mem.ship & mem.ship + this work & kic@xmath32 & hole et al . & sanders & hole et al.@xmath33 & sanders@xmath34 & this work + 1 & 5024272 & 003003 & & sm 95% & & no + 2 & 5024750 & 001004 & 141 & sm 93% & 83% & yes + 3 & 5023889 & 004014 & 42 & sm 95% & 90% & no + 4 & 5023732 & 005014 & 27 & sm 94% & 90% & yes + 5 & 5112950 & 003005 & 148 & sm 95% & 92% & yes + 6 & 5112387 & 003007 & 73 & sm 95% & 88% & yes + 7 & 5024512 & 003001 & 116 & sm 93% & 90% & yes + 8 & 4936335 & 007021 & 9 & sm 95% & 68% & no + 9 & 5024405 & 004001 & 100 & sm 93% & 91% & yes + 10 & 5112072 & 009010 & 39 & sm 95% & 91% & yes + 11 & 4937257 & 009015 & 144 & sm 88% & 80% & no + | asteroseismology of stars in clusters has been a long - sought goal because the assumption of a common age , distance and initial chemical composition allows strong tests of the theory of stellar evolution .
we report results from the first 34 days of science data from the _ kepler mission _ for the open cluster ngc 6819 one of four clusters in the field of view .
we obtain the first clear detections of solar - like oscillations in the cluster red giants and are able to measure the large frequency separation , @xmath0 , and the frequency of maximum oscillation power , @xmath1 .
we find that the asteroseismic parameters allow us to test cluster - membership of the stars , and even with the limited seismic data in hand , we can already identify four possible non - members despite their having a better than 80% membership probability from radial velocity measurements .
we are also able to determine the oscillation amplitudes for stars that span about two orders of magnitude in luminosity and find good agreement with the prediction that oscillation amplitudes scale as the luminosity to the power of 0.7 .
these early results demonstrate the unique potential of asteroseismology of the stellar clusters observed by _ |
inducing magnetism in non - magnetic oxides is one of the main areas of interest in the research community . generally , many oxides are insulator and , without magnetic impurity atoms , non - magnetic .
however , recent theoretical and experimental research of oxides has show ferromagnetism with defects or light elements , e.g. , c , n , li.@xcite in the past decade , density functional theory ( dft ) has played a vital role in either proposing new magnetic materials or elucidating the origins of defects - driven magnetism in oxides.@xcite to date , there are a few oxides that show room temperature ( rt ) ferromagnetism , e.g. , zno , sno@xmath1 , in@xmath1o@xmath2 , ceo@xmath1.@xcite light elements can not only be used to develop magnetism in oxides , but can also be used to stabilize intrinsic defects in the host materials.@xcite through the combined efforts of theoreticians and experimentalists , we are now certain that magnetism in oxides induced by non - magnetic impurities .
@xcite now it is known that magnetism develops in the nonmagnetic oxides where the impurity atoms have a finite local magnetic moment and these local magnetic moments will interact with each other to form a net magnetic moment in the host material .
those materials in which magnetism is induced by doping nonmagnetic impurities , for which the substitutional atom can have a finite magnetic moment and the @xmath3-electrons of the doped atom plays an essential role in governing magnetism in the host material , are generally considered as @xmath4 magnetic materials .
these nonmagnetic impurity atoms can also form impurity bands in the bandgap of host material , and ferromagnetic behavior can be expected if the fermi energy lies within these impurity bands .
bao is an oxide with interesting structural and electronic properties .
it can be used as a no@xmath2 storage device for catalysis.@xcite bao is also considered as a precursor to the well known ferroelectric perovskite oxide batio@xmath2 , which can have either tio@xmath1 or bao terminated surface when batio@xmath2 is grown on a suitable substrate as a thin film .
possible magnetism in tio@xmath1.@xcite has been extensively studied , however , less attention has been given to impurities in bao and there are no detailed theoretical and experimental studies on inducing magnetism in it .
it has been experimentally observed that bulk bao naturally occurs in a b1 ( nacl ) structure.@xcite recently , experimental reports have claimed the growth of ultrathin bao films on srtio@xmath2(001 ) substrate , and the formation of bao nanoparticles on reconstructed srtio@xmath2(001 ) , while a locally ordered c(@xmath5 ) bao structure is observed on the disordered sample surface.@xcite tan _
et.al_ , @xcite also observed rt ferromagnetism in n - doped batio@xmath2 and the origin of the magnetism was correlated with the presence of n. therefore , studying the electronic properties of bao is of great importance for the development of a new applications , specially in the area of magnetism .
hence , we propose to add a new functionality , i.e. magnetism , in bao .
we show that n in bao has a ferromagnetic ( fm ) ground state and the transition temperature @xmath6 , which is the temperature at which a material goes from a paramagnetic(disorder phase)state to a magnetic phase(order phase ) state , is above room temperature .
to study the magnetism of n - doped bao , we performed calculations in the framework of density functional theory , @xcite using linear combination of atomic orbital basis as implemented in the siesta code @xcite .
a double-@xmath7 polarized basis set for all atoms was used .
the local spin density approximation @xcite(lsda ) was adopted for describing exchange - correlation interactions .
we used standard norm - conserving pseudopotentials @xcite in their fully nonlocal form @xcite .
a cutoff energy of 400 ry for the real - space grid was adopted .
the sampling of @xmath8-space was performed with monkhorst and pack ( mp ) scheme with a regularly spaced mesh of @xmath9 .
convergence with respect to @xmath8-point sampling and cutoff energy was carefully checked . to investigate the magnetism and electronic structures of n - doped bao , we considered the @xmath10 ( ba@xmath11o@xmath2n@xmath12 ) , @xmath13 ( ba@xmath14o@xmath15n@xmath12 ) , @xmath16(ba@xmath17o@xmath18n@xmath12 ) supercells of the primitive unit cell of nacl - type bao .
calculations were also carried out using a @xmath19 ( ba@xmath20o@xmath21n@xmath12 ) supercell . in all these supercells , n was doped at o site due to smaller difference in their atomic sizes and electro - negativities . for the magnetic properties of bao ( 001)surface , we considered the conventional unit cell of bao , and used different @xmath22 symmetric slabs of thickness @xmath23 , defined as a monolayer of bao .
we investigated @xmath24 ( surf@xmath12 ) and @xmath25 ( surf@xmath1 ) surfaces of bao and doped n at o on both sides of the bao surface .
we added a vacuum region of about @xmath26 10 , so that the two surfaces do not interact with each other through the vacuum region .
additional simulations , using plane waves plus pseudopotentials as implemented in the quantum espresso ( qe ) code@xcite , were also carried out to further test the validity of our results .
atomic positions were relaxed , using conjugate - gradient algorithm , @xcite until the residual hellmann - feynman force on single atom converges to less then 0.05 ev / .
test calculations , using lsda@xmath27 , were also carried out by considering the on - site coulomb correction ( @xmath28 ev , our previously optimized value@xcite ) between the @xmath29-orbital electrons of o.@xcite
first , we calculated the optimized lattice constant of nacl - type bao , which was found to be @xmath30 , as shown in fig .
[ band_bao](a ) .
the calculated lattice constant is comparable to the previous calculated @xmath31,@xcite and experimental @xmath32@xcite values . our dft estimated value is smaller than the experimental value due to the underestimation problem of dft - lda . using the optimized lattice constant , we calculated the band structure , which is shown in fig .
[ band_bao](b ) .
we see that bao is a wide bandgap semiconductor and the calculated bandgap at the @xmath33 point is @xmath34 ev . the band is degenerate at @xmath35 point [ @xmath36 , but this degeneracy is removed at @xmath33 point where the band has less curvature ( large effective mass ) compared with point @xmath35 .
the electronic density of states ( dos ) shows that bands near the fermi energy ( @xmath37 ) are mainly derived from the o @xmath29-orbitals [ see fig .
[ fm - dos](a ) ] .
the lsda calculated indirect ( @xmath38-@xmath33)bandgap is @xmath39 ev , in agreement with the previous lda calculation.@xcite however , the experimental and gw calculated value is 3.88 ev @xcite .
we then used lsda@xmath27 , and our lsda@xmath27 calculated bandgap is @xmath40 , which is comparable with the gw calculated and experimental values .
it is important to consider atomic relaxation in n - doped bao .
usually , the doping impurity at the host site can either compress or elongate the bond length depending on the atomic size of the impurity atom . in some cases
the impurity atom can also be located at an interstitial site and form a defect complex.@xcite we therefore relaxed all the atomic coordinates of bao : n and no significant changes in the atomic position of n at o sites were found . to further confirm this ,
both ba@xmath14o@xmath15n@xmath12 and ba@xmath17o@xmath18n@xmath12 were considered , and calculated their optimized lattice constants , which are @xmath41 for both the systems [ see fig . [ band_bao](a ) ] .
this lattice constant is approximately equal to our optimized lattice constant of pristine bao , which shows that n can easily dope the o sites without any structural distortion .
figure [ fm - dos](a ) shows the calculated total and atom projected ( p ) dos of n - doped bao in the non - magnetic state . for comparison purposes , the total pdos of pristine bao
is also shown .
it can be clearly seen that the dos of doped and pristine bao are the same in the conduction and valance bands .
however , near the fermi energy ( @xmath42 ) the dos of the doped system is drastically changed and the material shows metallic behavior , in contrast to pristine bao , which is an insulator .
the pdos demonstrates that the main peak at the fermi energy arises mainly from contributions from the @xmath29-orbitals of n. the n atom in bao forms an impurity band in the bandgap of bao .
a small impurity - derived peak can also be seen at 2.0 ev just below @xmath37 , and this peak hybridizes with the @xmath29-orbitals of o atoms .
therefore , the o and n atoms are only hybridize near the fermi energy .
no significant changes in the pdos of o can be seen when the o sites are doped with n , except a small shoulder , derived by the n @xmath29-orbitals , near the fermi energy .
the calculated results show that the top of the valence band consists of n @xmath3 electrons , which are higher than o @xmath3 electrons .
the large dos at the fermi energy @xmath43 in the n - doped bao shows that there is an instability towards magnetism within the stoner mean field theory of magnetism.@xcite within the stoner model , which was mainly proposed for itinerant electron system , the large @xmath43 can lead to a large pauli susceptibility which is large enough for the band to split spontaneously , and magnetism in n - doped bao can be expected . to confirm
that n - doped bao can present magnetism within the stoner mean field theory , we further carried out spin - polarized calculations .
the spin - polarized electronic density of state of ba@xmath11o@xmath2n@xmath12 is shown in fig.[fm - dos](b ) .
the total dos clearly shows that n induces magnetism in bao when doped at o sites .
the spin - polarized structure shows that the non - magnetic impurity band in the bandgap [ see fig.[fm - dos](a ) ] is mainly due to the contribution from minority spins .
large exchange splitting at n site can also be seen , as expected in the stoner mean field theory .
for comparison , we also show the total dos of pristine bao and one can easily judge that the conduction band of ba@xmath11o@xmath2n@xmath12 in the spin - up state is identical to that of bao .
the @xmath3 states of nitrogen are located at the top of the valence band , which are mainly derived from the anion @xmath29-states , resulting in the hybridization of electron wave functions of the @xmath3 orbitals .
the majority spin - states of nitrogen also hybridize with oxygen @xmath3 states , promoting the the impurity band which connects to the top of the valence band .
such chemical process rise to a bandgap in the majority spins states .
the minority spins on the other hand create an impurity band in the bandgap which includes @xmath37 . as a result
this impurity band is broadened having about 0.38 ev half - width , and the approximated exchange splitting is 0.80 ev .
the majority spins are completely occupied and behaving just like an insulator , whereas the minority spins are partially occupied and have a metallic nature .
such electronic structure is the fingerprint of a half - metal ferromagnet , @xcite which has applications in the area of spin electronics .
the total and local magnetic moments of ba@xmath11o@xmath2n@xmath12 were also calculated , and the calculated total magnetic moment is @xmath44 per unit cell .
the pdos clearly shows that the magnetic moment arises mainly from contributions from the n @xmath29-orbitals and the local magnetic moment at the n site is @xmath45 . a small induced magnetic moment ( @xmath46 ) at the nearby o site is also observed .
the pdos of o shows some unoccupied minority spin states , which were occupied in the pristine bao , suggesting that the holes induced by n also localized at the o @xmath29-orbitals .
we repeated the same calculations by doping n at o site in @xmath13 and @xmath16 supercells , and interestingly we found half - metallic behavior in all these doped systems [ see fig .
[ fm - dos](d and e ) ] and the magnetic moments were mainly localized at the n sites . it is also noticeable that n does not polarize the whole valance band , it polarizes only the valance electronic states near the fermi energy
. such behavior of n in bao is different from that of transition metals in oxides / semiconductors , which usually polarize the whole valance band.@xcite the band - width of n decreases and the band is narrowed , as the concentration of n is reduced in bao [ see fig .
[ fm - dos](e ) ] , which further increases the localization of @xmath3 states of n. this localization of the n @xmath3 states may further increase the observation / suppression of ferromagnetism / antiferromagnetism in bao , which is discussed in the following paragraph .
the total magnetic moment per n atom is @xmath44 , which indicates that the total magnetic moment is independent of the n concentrations . in all these n - doped bao systems ,
the calculated magnetic moment is consistent with hund s rules indicating that the n dopant exists as a n@xmath47 ( @xmath48 ) anion illustrating that each n impurity introduces one @xmath3 hole and the strong hund - type exchange makes the nitrogen @xmath3 spin - up orbitals completely occupied , and the spin - down states are partially filled [ fig.[fm - dos](b - e ) ] . as a result
, one can expect the ( ferro)magnetism in n - doped bao to be stabilized mainly by the predominant double - exchange mechanism.@xcite before we address the question of fm or antiferromagnetic ( afm ) coupling between the n atoms , the above calculations were also performed for ba@xmath20o@xmath21n@xmath12 and ba@xmath49o@xmath50n@xmath12 , and we confirmed that n in bao has spin - polarized band structure .
we also confirmed the above conclusions using the qe code .
it is also important to comment on the magnetism of n - doped bao using lsda@xmath27 .
the lsda@xmath27 calculated total and pdos of ba@xmath11o@xmath2n@xmath12 are shown in fig .
[ fm - dos](c ) .
as expected , the n forms an impurity band in the bandgap of bao and n - driven oxygen @xmath29 states can also be seen in the gap region of bao .
similar observations have also been observed in self - interaction correction ( sic ) for c - doped bao.@xcite the local magnetic moment of n ( o)is increased ( decreased ) , but the total magnetic moment of the unit cell remains unchanged , i.e. , @xmath51 .
similar conclusions were also drawn when lsda@xmath27 calculations were performed for ba@xmath14o@xmath15n@xmath12 , and ba@xmath17o@xmath18n@xmath12 consistent with our previous work .
@xcite as predicted , n can induce half - metallic magnetism in insulating bao .
however , there remains a question on whether n - doped bao can have fm order above room temperature , and whether the exchange coupling sufficiently large to give rise to possible rt ferromagnetism .
magnetism is a cooperative phenomenon and a single n in bao can not determine the true magnetic ground state .
therefore , we consider @xmath16 and @xmath19 supercells of bao and doped two n atoms at different o sites .
the distance between the two n atoms @xmath52 was varied , as it was the fm and afm coupling between them . for comparison purposes
, we kept the same n - n separation in both supercells , and the results are summarized in table [ table ] .
it is enchanting to see that both bao : n systems have fm ground state , and the total magnetic moment of the unit cell is @xmath53 ( i.e. , @xmath44 per n ) .
all the systems remained half - metallic as well .
the strength of the exchange interactions @xmath54 can be judged from @xmath55 , where @xmath56(@xmath57 ) is the total energy of the supercell in the afm ( fm ) state .
this exchange energy can be further used to estimate @xmath0 . using the heisenberg mean field model ( @xmath58/3),@xcite the estimated @xmath0
is found to be close to room temperature , indicating the possibility of room temperature ferromagnetism in bao : n systems .
one can also see that both the systems have different transition temperatures due to different n concentrations in bao .
note that , usually , mean field theory overestimates @xmath0 but it can give an indication of possible rt fm .
therefore , rt ferromagnetism in n - doped bao is expected , if properly prepared .
it is encouraging to mention that our estimated @xmath0 matches with n - doped batio@xmath2 @xcite and other magnetic oxides @xcite indicating that the magnetism is mainly attributed to n impurities .
the formation enthalpy @xmath59 of bao is calculated using @xmath60 $ ] , where @xmath61 , @xmath62 , and @xmath63 are the total energies of nacl - type bao , bcc ba , and oxygen molecule , respectively .
the calculated @xmath59 was found to be @xmath64 ev , which comparable with the previous calculated ( -@xmath65@xcite , -@xmath66@xcite ) , and experimental ( -@xmath67@xcite ) values .
we followed our previous approach @xcite and calculated the defect formation energy @xmath68 of n in bao under ba - rich and o - rich conditions because the defect formation energy strongly depends on the growth condition and chemical potential .
@xcite the calculated @xmath68 of ba@xmath11o@xmath2n@xmath12 , ba@xmath14o@xmath15n@xmath12 , and ba@xmath17o@xmath18n@xmath12 in ba - rich ( o - rich ) conditions are -1.15 ( 5.33 ) , -1.14 ( 5.34 ) , and -1.15 ( 5.45 ) ev , respectively .
it is clear to see that n in bao has a negative defect formation energy under ba - rich condition , which further increases the possibility of introducing n in bao .
note that lsda@xmath27 calculated @xmath68 of ba@xmath11o@xmath2n@xmath12 is about 6.6 ev in o - rich condition , which is increased by including @xmath69 parameter similar
behvior has also been observed in other oxides.@xcite we then varied the distance @xmath52 between the two n atoms , using @xmath70 and @xmath71 supercells , and calculated @xmath68 under different growth conditions .
the calculated @xmath68 per n atom of both systems in the fm states is shown in table [ table ] .
the calculated @xmath68 values are in the range of a single n doped bao , and therefore we can not expect clustering of n in bao when doping the o sites .
the formation energy depends on the n concentration in bao .
we may safely conclude that all the studied systems have negative formation energies under ba - rich environments .
such calculated results can help experimentalists to dope n in bao .
the role of intrinsic defects ( vacancies ) can not be ignored when a sample is synthesised , these intrinsic defects may interact with the impurity atom , and such interaction may further decrease @xmath68 of the impurity atom.@xcite there is also an experimental report on n - induced ferromagnetism in batio@xmath2 , @xcite which suggests the possibility of n - doping at the o sites in bao .
experimental work will be very helpful to further analyse the possibilities of n - doped bao . to propose bao for practical practical applications , e.g. , in the area of thin films , it is essential to study the electronic and magnetic properties of n doped at surface o site in bao , surf@xmath12 .
we found that n can also induce magnetism in bao ( 001 ) surface , and the magnetism is mainly localized at the n site as shown in fig.[surface](a , b ) .
surface oxygen atoms have negligible magnetic moment and n has a large magnetic moment ( @xmath72 when doped at surface o sites , as expected for surface n atoms .
no induced magnetic moment at the subsurface atoms were found , which is different from other doped oxides.@xcite the pdos of surface o atoms are shifted towards the fermi energy compared to clean surface o atoms , indicating that the band gap of bao(001 ) surface is reduced due to surface states driven by n - o hybridization .
the majority spins of n are occupied and the minority spins are partially occupied .
the calculated surface spin densities of n , o , and ba atoms are shown in fig.[surface](c ) , and it is evident to see no spin - polarization at the surface o and ba atoms is observed .
the spin density is mainly localized around the n atoms .
we then used @xmath22 surface ( surf@xmath1 ) and doped two n atoms at two different o sites and considered the fm and afm coupling between the n atoms . here
, we have also found that the fm coupling between the surface n atoms has lower energy than that of afm coupling .
as the concentration of n atom is increases , a small magnetic moments at the surface o atoms were also induced , but the sub - surface o atoms do not show any induced magnetic moment[fig.[surface](b ) ] . no surface relaxation or reconstruction
was observed when all the atoms were fully relaxed .
the calculated @xmath68 of n at surface o in ba - rich ( o - rich ) condition is @xmath73 ( @xmath74 ) ev for surf@xmath12 . for surf@xmath1 , when the two n atoms were doped at the surface o sites ,
the @xmath68 per n in ba - rich ( o - rich ) conditions is @xmath75 ( 5.00 ) ev when the distance between the two surface n atom is 3.82 , whereas when the separation between the two n atoms is 5.40 , the calculated @xmath68 per n in ba - rich ( o - rich ) condition is @xmath76 ( 4.88 ) ev .
surface n - doped bao has a smaller defect formation energy than the bulk n - doped bao , similar to other oxides.@xcite note that a smaller surface @xmath68 corresponds to a larger n - n bond distance suggesting that n prefers isolation than clustering in bao(001 ) surface .
we have shown that n in bao can display possible room temperature ferromagnetism either in bulk or surface doped system , and finally we would like to compare it with the co - doped bao .
our preliminary results clearly indicate that a co atom at either o or ba site induces magnetism with a larger(about 3 times ) @xmath68 than bao : n systems .
therefore , we believe that n doping in bao has the potential to easily tune magnetism and may be superior to co doped bao . further experimental work would be needed to demonstrate the potential of bao : n for possible applications in the area of spin electronics .
we also believe that our prediction can also be helpful for bao terminated surface of batio@xmath2 .
to summarize , dft was used to investigate potential magnetism in n - doped nacl - type bao .
pristine bao is known to be non - magnetic wide bandgap semiconductor , and n at o sites induce an impurity band in the bandgap of bao .
it has been discussed that n in bao can also induce magnetism and the magnetism arises mainly from contributions by the n atoms .
the minority impurity - driven band is narrowed with decreasing the n concentrations in bao .
the origin of magnetism was elucidated from the atom projected density of states .
all the considered scenarios with different n concentrations in bao showed half - metallic behavior .
the ferromagnetic and antiferomagnetic interactions between the n atoms were also considered , and they revealed that n - doped bao is a ferromagnetic material .
the transition temperature was also calculated using heisenberg mean field model , and it was found to be above room temperature .
the magnetic interaction between the n atoms is limited to the next - nearest - neighbor interactions .
the surface magnetism of n - doped bao(001 ) was also studied by doping surface o sites with n. n atoms at surface o sites also showed ferromagnetic ground state , and the magnetism was mainly limited to the bao surface .
no induced magnetic moments were observed at the sub - surface o sites .
the defect formation energy showed that n has a smaller formation energy than the bulk n doped bao .
the magnetism of n - doped bao was also compared to that co - doped bao , and it was concluded n - doped bao can easily be achieved .
lsda@xmath27 calculations also supported ferromagnetism in n - doped bao .
m. khalid , m. ziese , a. setzer , p. esquinazi , m. lorenz , h. hochmuth , m. grundmann , d. spemann , t. butz , g. brauer , w. anwand , g. fischer , w. a. adeagbo , w. hergert , and a. ernst , phy .
b * 80 * , 035331 ( 2009 ) . j. b. yi , c. c. lim , g. z. xing , h. m. fan , l. h. van , s. l. huang , k. s. yang , x. l. huang , x. b. qin , b. y. wang , t. wu , l. wang , h. t. zhang , x. y. gao , t. liu , a. t. s. wee , y. p. feng , and j. ding , phys . rev . lett . *
104 * , 137201 ( 2010 ) .
xiao , l .- l .
wang , l. xu , q. wan , and b. s. zou , solid state commun . *
149 * , 1304 ( 2009 ) .
w. zhou , l. liu , and p. wu , j. magn .
magn . mater . *
321 * , 3356 ( 2009 ) .
h. wang , y. yan , y. s. mohammed , x. du , k. li , and h. jin , j. magn .
magn . mater . *
321 * , 337 ( 2009 ) .
x. l. wang , z. zeng , and x. h. zheng , j. appl .
* 101 * , 09h104 ( 2007 ) . s .-
hu , s .- s .
yan , x .- x .
yao , y .- x .
chen , g .-
liu , and l .- m .
mei , phys .
b * 75 * , 094412 ( 2007 ) .
g. rahman , n. ud din , v. m. garc ' ia - su ' arez and erjun kan , phy .
b * 87 * , 205205 ( 2013 ) .
p. giannozzi , s. baroni , n. bonini , m. calandra , r. car , c. cavazzoni , d. ceresoli , g. l. chiarotti , m. cococcioni , i. dabo , a. dal corso , s. fabris , g. fratesi , s. de gironcoli , r. gebauer , u. gerstmann , c. gougoussis , a. kokalj , m. lazzeri , l. martin - samos , n. marzari , f. mauri , r. mazzarello , s. paolini , a. pasquarello , l. paulatto , c. sbraccia , s. scandolo , g. sclauzero , a. p. seitsonen , a. smogunov , p. umari , r. m. wentzcovitch , j.phys .
: condens.matter * 21 * , 395502 ( 2009 ) . .calculated total magnetic moment @xmath77 ( in @xmath78 ) , @xmath55 per cell ( in ev ) , transition temperature @xmath0 ( in k ) , and defect formation energy @xmath68 per n ( in ev ) of n - doped bao ( @xmath71 ) . @xmath79
and o@xmath80 shows @xmath68 under ba - rich and o - rich conditions , respectively .
the first column shows n
n separation @xmath52 ( in ) .
values in the brackets correspond to the @xmath70 supercell . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] o@xmath15n@xmath12 ( blue line ) , and ba@xmath17o@xmath18n@xmath12 ( green line ) .
( b ) the band structure of pristine bao , and the horizontal line ( black ) shows the fermi energy , which is set to zero .
, title="fig:",scaledwidth=80.0% ] o@xmath15n@xmath12 ( blue line ) , and ba@xmath17o@xmath18n@xmath12 ( green line ) .
( b ) the band structure of pristine bao , and the horizontal line ( black ) shows the fermi energy , which is set to zero .
, title="fig:",scaledwidth=60.0% ] o@xmath2n@xmath12 in the non - magnetic ( a ) , where the green dotted lines show the pdos(o - pris ) and total dos ( pris ) of pristine bao . the spin - polarized ( magnetic ) total and pdos of ba@xmath11o@xmath2n@xmath12 , ba@xmath14o@xmath15n@xmath12 , and ba@xmath17o@xmath18n@xmath12
are shown in ( b ) , ( d ) , and ( e ) , respectively .
panel ( c ) shows lsda@xmath27 dos of ba@xmath11o@xmath2n@xmath12 .
the green dotted lines in the total dos panels show the total dos of the pristine bao ( pris ) of the same supercells .
the total magnetic moments per n atom and local magnetic moments of n and o are also shown .
the o and n pdos are represented by solid ( red ) lines .
the vertical lines represent the fermi energy , which is set to zero .
, title="fig:",scaledwidth=65.0% ] o@xmath2n@xmath12 in the non - magnetic ( a ) , where the green dotted lines show the pdos(o - pris ) and total dos ( pris ) of pristine bao .
the spin - polarized ( magnetic ) total and pdos of ba@xmath11o@xmath2n@xmath12 , ba@xmath14o@xmath15n@xmath12 , and ba@xmath17o@xmath18n@xmath12 are shown in ( b ) , ( d ) , and ( e ) , respectively .
panel ( c ) shows lsda@xmath27 dos of ba@xmath11o@xmath2n@xmath12 .
the green dotted lines in the total dos panels show the total dos of the pristine bao ( pris ) of the same supercells .
the total magnetic moments per n atom and local magnetic moments of n and o are also shown .
the o and n pdos are represented by solid ( red ) lines .
the vertical lines represent the fermi energy , which is set to zero .
, title="fig : " ] o@xmath2n@xmath12 in the non - magnetic ( a ) , where the green dotted lines show the pdos(o - pris ) and total dos ( pris ) of pristine bao . the spin - polarized ( magnetic ) total and pdos of ba@xmath11o@xmath2n@xmath12 , ba@xmath14o@xmath15n@xmath12 , and ba@xmath17o@xmath18n@xmath12
are shown in ( b ) , ( d ) , and ( e ) , respectively .
panel ( c ) shows lsda@xmath27 dos of ba@xmath11o@xmath2n@xmath12 .
the green dotted lines in the total dos panels show the total dos of the pristine bao ( pris ) of the same supercells .
the total magnetic moments per n atom and local magnetic moments of n and o are also shown .
the o and n pdos are represented by solid ( red ) lines .
the vertical lines represent the fermi energy , which is set to zero .
, title="fig : " ] o@xmath2n@xmath12 in the non - magnetic ( a ) , where the green dotted lines show the pdos(o - pris ) and total dos ( pris ) of pristine bao . the spin - polarized ( magnetic ) total and pdos of ba@xmath11o@xmath2n@xmath12 , ba@xmath14o@xmath15n@xmath12 , and ba@xmath17o@xmath18n@xmath12
are shown in ( b ) , ( d ) , and ( e ) , respectively .
panel ( c ) shows lsda@xmath27 dos of ba@xmath11o@xmath2n@xmath12 .
the green dotted lines in the total dos panels show the total dos of the pristine bao ( pris ) of the same supercells .
the total magnetic moments per n atom and local magnetic moments of n and o are also shown .
the o and n pdos are represented by solid ( red ) lines .
the vertical lines represent the fermi energy , which is set to zero . , title="fig : " ] o@xmath2n@xmath12 in the non - magnetic ( a ) , where the green dotted lines show the pdos(o - pris ) and total dos ( pris ) of pristine bao .
the spin - polarized ( magnetic ) total and pdos of ba@xmath11o@xmath2n@xmath12 , ba@xmath14o@xmath15n@xmath12 , and ba@xmath17o@xmath18n@xmath12 are shown in ( b ) , ( d ) , and ( e ) , respectively .
panel ( c ) shows lsda@xmath27 dos of ba@xmath11o@xmath2n@xmath12 .
the green dotted lines in the total dos panels show the total dos of the pristine bao ( pris ) of the same supercells .
the total magnetic moments per n atom and local magnetic moments of n and o are also shown .
the o and n pdos are represented by solid ( red ) lines .
the vertical lines represent the fermi energy , which is set to zero .
, title="fig : " ] ( a ) and surf@xmath1(b ) .
the top panels show the pdos of surface o atoms ( red lines ) along with the pdos of clean surface o atoms ( green lines ) .
the middle panels show the pdos of surface n atoms ( red lines ) .
the bottom panels show the pdos of sub - surface o atoms denoted as o@xmath81 .
the local magnetic moments of n and o atoms are also shown .
the vertical lines show the fermi energy , which is set to zero . the calculated surface spin density of surf@xmath12(c ) , where red , green , and yellow balls represent ba , o , and n atoms , respectively .
the spin density of surf@xmath12 is calculated on a large surface to show zero polarization at surface o atoms .
the inset shows the scale of the spin density .
, title="fig : " ] ( a ) and surf@xmath1(b ) .
the top panels show the pdos of surface o atoms ( red lines ) along with the pdos of clean surface o atoms ( green lines ) .
the middle panels show the pdos of surface n atoms ( red lines ) .
the bottom panels show the pdos of sub - surface o atoms denoted as o@xmath81 .
the local magnetic moments of n and o atoms are also shown .
the vertical lines show the fermi energy , which is set to zero . the calculated surface spin density of surf@xmath12(c ) , where red , green , and yellow balls represent ba , o , and n atoms , respectively .
the spin density of surf@xmath12 is calculated on a large surface to show zero polarization at surface o atoms .
the inset shows the scale of the spin density .
, title="fig : " ] ( a ) and surf@xmath1(b ) .
the top panels show the pdos of surface o atoms ( red lines ) along with the pdos of clean surface o atoms ( green lines ) . the middle panels show the pdos of surface n atoms ( red lines ) .
the bottom panels show the pdos of sub - surface o atoms denoted as o@xmath81 .
the local magnetic moments of n and o atoms are also shown .
the vertical lines show the fermi energy , which is set to zero . the calculated surface spin density of surf@xmath12(c ) , where red , green , and yellow balls represent ba , o , and n atoms , respectively .
the spin density of surf@xmath12 is calculated on a large surface to show zero polarization at surface o atoms .
the inset shows the scale of the spin density . ,
title="fig:"](a ) and surf@xmath1(b ) .
the top panels show the pdos of surface o atoms ( red lines ) along with the pdos of clean surface o atoms ( green lines ) .
the middle panels show the pdos of surface n atoms ( red lines ) . the bottom panels show the pdos of sub - surface o atoms denoted as o@xmath81 .
the local magnetic moments of n and o atoms are also shown .
the vertical lines show the fermi energy , which is set to zero . the calculated surface spin density of surf@xmath12(c ) , where red , green , and yellow balls represent ba , o , and n atoms , respectively .
the spin density of surf@xmath12 is calculated on a large surface to show zero polarization at surface o atoms .
the inset shows the scale of the spin density .
, title="fig : " ] | density functional theory with local spin density approximation has been used to propose possible room temperature ferromagnetism in n - doped nacl - type bao .
pristine bao is a wide bandgap semiconductor , however , n induces a large density of states at the fermi level in the nonmagnetic state , which suggests magnetic instability within the stoner mean field model .
the spin - polarized calculations show that n - doped bao is a true half- metal , where n has a large magnetic moment , which is mainly localized around the n atoms and a small polarization at the o sites is also observed .
the origin of magnetism is linked to the electronic structure .
the ferromagnetic(fm ) and antiferromagnetic ( afm ) coupling between the n atoms in bao reveal that doping n atoms have a fm ground state , and the calculated transition temperature ( @xmath0 ) , within the heisenberg mean field theory , theorizes possible room temperature fm in n - doped bao .
nitrogen also induces ferromagnetism when doping occurs at surface o site and has a smaller defect formation energy than the bulk n - doped bao .
the magnetism of n - doped bao is also compared with co - doped bao , and we believe that n has a greater potential for tuning magnetism in bao than co. |
the koszul duality theory originates from the work of s. priddy @xcite , inspired by ideas of koszul @xcite .
priddy s theory describes `` small '' resolutions for associative algebras . to compute derived functors , like the @xmath1 and the @xmath2 functors
, one needs projective resolutions of modules .
there always exist such resolutions , which are built functorially from the bar construction , for instance .
this functoriality has a price : such a resolution is pretty big . for a given algebra
, one can find a ( generally ) smaller resolution by describing step by step an economic set of generators , called the syzygies .
this inductive process is now called the koszul tate resolution after @xcite .
the same type of process allows one to build a model for a given algebra @xmath3 , that is a differential graded free algebra whose homology is isomorphic to @xmath3 .
such a model , according to the general results of homotopical algebra @xcite , can be used to compute the quillen homology @xmath4 , for instance . the original koszul duality theory developed by priddy starts from an algebra @xmath3 given with a quadratic presentation @xmath5 and produces a quadratic koszul dual coalgebra @xmath6 generated by the same presentation . under this algebraic structure ,
it produces , at once , a good candidate for the space of syzygies .
all the three graded vector spaces @xmath7 , @xmath8 , @xmath4 can be expressed naturally in terms of @xmath6 by re - grading or passing to the graded dual @xmath9 . the algebra @xmath10 admits a simple presentation as @xmath11 .
the relationship between the differentials in the corresponding resolutions is more subtle .
the higher case , when a given algebra @xmath3 has a homogeneous presentation @xmath12 of weight @xmath13 , has been first studied by r. berger in @xcite ( see also @xcite ) , where a candidate for the dual @xmath10 of the space of syzygies was described , and where a condition for that candidate to work was given .
each of the equivalent ways to look at the space of syzygies ( @xmath2 , @xmath1 , quillen homology ) exhibits a rich structure of an associative algebra up to homotopy , or @xmath14-algebra .
this viewpoint has become more common in ring theory in the past decade , once it was understood that higher operations on the @xmath2-algebras of associative algebras allow one to keep track of all the homotopy information of those algebras that would otherwise be lost if we restrict ourselves to the yoneda product only .
this circle of ideas is presented in a sequence of papers of lu , palmieri , wu , and zhang @xcite . in the case of @xmath0-koszul algebras , the @xmath14-algebra structure of the yoneda algebras
was computed by he and lu in @xcite .
they proved that most of the higher operations for such an algebra vanish , and that the non - vanishing ones admit very explicit formulas . in a sense ,
their theorems provide an exhaustive description of the @xmath2-algebras as @xmath14-algebras . however , those theorems have a slightly mysterious feature which makes one hope for a different way to view the same result .
namely , in the quadratic case the @xmath2-algebra of a koszul algebra is an @xmath14-algebra with all the higher operations vanishing , that is an associative algebra , presented by generators and relations . in the @xmath0-koszul case
, the @xmath2-algebra carries an @xmath14-algebra structure with all the higher operations except for the @xmath0th one vanishing , that is , an @xmath15-algebra .
it is also proved to be a reduced @xmath15-algebra , meaning that it satisfies additional vanishing conditions .
however to describe the higher operations explicitly , this algebra has to be realised as a subspace of a usual associative algebra . in this paper
, we suggest to replace the condition of being reduced by a more functorial condition involving properties of operations only .
more precisely , we consider a new type of algebras which we call @xmath16-algebras .
an @xmath16-algebra is a graded vector space with an associative product @xmath17 of degree zero , and an @xmath0-ary operation @xmath18 of degree @xmath19 which satisfy the identity @xmath20 and the identities @xmath21 it turns out that the berger s
candidate @xmath10 for the linear dual of space of syzygies admits the following presentation via generators and relations as a @xmath16-algebra .
let @xmath22 be a finitely generated @xmath0-homogeneous algebra .
then @xmath23 as @xmath16-algebras , where @xmath24 is viewed as a subspace of @xmath25 . to prove this theorem , and in general to get a better understanding of @xmath16-algebras , we introduce a general notion of grbner bases for algebras over non - symmetric operads
. we also exhibit a precise equivalence between describing a free @xmath3-module resolution of the ground field @xmath26 and describing a model for @xmath3 .
overall , this demonstrates a clear analogy between the @xmath27 koszul duality theory of priddy and the higher koszul property for @xmath0-homogeneous algebras with @xmath13 .
we were motivated to develop the higher koszul duality theory for @xmath0-homogeneous algebras in a way similar to the koszul duality for quadratic algebras in order to generalise it even further .
some related work was done in @xcite for relations of weights @xmath28 and @xmath0 , and in @xcite for relations of weights @xmath29 and @xmath30 , both greater than @xmath28 .
however , the results of @xcite demonstrate that even `` small '' resolutions may yield `` big '' @xmath14-algebra structures .
and so it might make sense to refine the corresponding class of koszul algebras .
we shall address this question in a future work .
the paper is organised as follows . in section [ sec : recoll ] , we recall some basic notions and results we use throughout the paper . in section [ sec : grobner ] , we introduce grbner bases for non - symmetric operads and for algebras over a non - symmetric operad , and we formulate the key results about them . in section [ sec : operadb ] , we introduce @xmath16-algebras and compute its reduced grbner basis . in section [ sec : present ] , we prove the theorem stated above , and use it to formulate an equivalent criterion for an @xmath0-homogeneous algebra to be @xmath0-koszul . in section [ sec : higherkoszul ] , we relate the construction of the free resolution of the trivial @xmath3-module and the construction of a dg algebra resolution of @xmath3 , using the language of twisting morphisms .
we work over a ground field @xmath26 .
the word `` algebra '' , unless otherwise specified , refers to an associative algebra .
as above , @xmath0 is a positive integer greater than @xmath28 .
most of vector spaces that we use are bi - graded with finite - dimensional bi - graded components .
the _ weight _
grading is coming from the convention that we consider algebras with generators of weight @xmath31 subject to homogeneous relations , and the _ ( homological ) degree _ is coming from the fact that our constructions produce chain complexes .
the first author acknowledges partial support of grant geoalgphys 20112013 awarded by the university of luxembourg .
we would like to express our appreciation to the max - planck institute for mathematics in bonn , where the work on this paper began , and to the university lyon 1 , where the paper was completed , for the joint invitation and the excellent working conditions .
we are grateful to eric hoffbeck for many useful comments on the first draft of this paper .
last but not the least , the first author wishes to thank the organisers of the workshop `` new developments in noncommutative algebra '' for a great opportunity to discuss some maths in the most beautiful scenery of the isle of skye ; he benefited from a few conversations with participants of that workshop , and is especially thankful to andrea solotar for explaining the key results of @xcite .
we shall recall the basic definitions and results used throughout the paper , however trying to keep this section reasonably concise . for details on associative algebras and operads and their homotopy theory we refer the reader to the book @xcite .
an associative algebra @xmath3 presented by generators and relations , @xmath22 is said to be _ @xmath0-homogeneous _ if @xmath32 . in this case , the algebra @xmath3 is weight graded : @xmath33 .
when @xmath34 is finite dimensional , the _ @xmath0-homogeneous dual algebra _ @xmath35 of an @xmath0-homogenenous algebra is defined by the formula @xmath36 where @xmath37 is the annihilator of @xmath38 under the natural pairing of vector spaces @xmath39 . in the case
@xmath27 , it turns out that various homological questions of @xmath3 admit natural answers in terms of the algebra @xmath35 which is , in that case , usually denoted by @xmath10 .
however , in the case @xmath13 , only some of the homogeneous components of @xmath35 are relevant . we define the graded vector space @xmath10 by the formula @xmath40 where in addition to being of weight @xmath41 , @xmath42 is assigned the homological degree @xmath43 if @xmath44 and the homological degree @xmath45 is @xmath46 .
the associative product on @xmath35 induces an associative product on @xmath10 ( whenever the product in @xmath35 ends up in a component absent in @xmath10 , the corresponding product is defined to be equal to zero ) .
the algebra @xmath10 is called the _ koszul dual algebra _ of @xmath3 .
note that our notational convention differs somewhat from the one adopted in earlier literature @xcite ; we believe that it is strongly beneficial to use , for any homogeneous algebra @xmath3 , the notation @xmath6 for the space of syzygies of @xmath3 , hence the change .
we also put @xmath47 ( we use the convention that the dual of a graded space with finite - dimensional component is the direct sum of component - wise duals ) . since @xmath10 is weight - wise finite dimensional , @xmath6 is a coalgebra .
note that for @xmath27 , we have @xmath48 .
recall from berger @xcite that the _ koszul complex _ of an @xmath0-homogeneous algebra @xmath3 is @xmath49 with the boundary maps @xmath50 being , respectively , the composites @xmath51 the maps coming from the coproduct of @xmath6 , the identification @xmath52 , and the product of @xmath3 . note that we assume elements of @xmath3 to be of homological degree @xmath53 , so the elements of @xmath54 ( respectively , @xmath55 ) are , according to the convention we adopted above , assigned the homological degree @xmath56 ( respectively , @xmath57 ) .
an @xmath0-homogeneous algebra @xmath3 is said to be _
@xmath0-koszul _ if its koszul complex is acyclic in positive homological degrees .
let @xmath3 be a graded vector space , and @xmath58 , @xmath59 be two linear maps of degrees @xmath60 and @xmath61 respectively .
we define their pre - lie product @xmath62 by the formula @xmath63 where @xmath64 stands for @xmath65 .
let @xmath66 be an integer .
@xmath15-algebra _ is a triple @xmath67 where @xmath3 is a graded vector space and where @xmath68 and @xmath69 are linear maps of degree @xmath53 and @xmath19 respectively .
they are required to satisfy the relations @xmath70 an @xmath15-algebra is a particular case of an @xmath14-algebra ( see , e. g. , @xcite and ( * ? ? ?
* chapter @xmath71 ) ) , whose differential vanishes and whose structure maps @xmath72 vanish as well for @xmath73 .
dualising the above definitions , we can define the corresponding notion of coalgebras .
if @xmath74 is a graded vector space , and @xmath75 , @xmath76 be two linear maps of degrees @xmath60 and @xmath61 respectively , we define their pre - lie product @xmath77 by the formula @xmath78 where @xmath79 stands for @xmath80 .
an _ @xmath15-coalgebra _ is a triple @xmath81 where @xmath74 is a graded vector space and where @xmath82 and @xmath83 are linear maps of degree @xmath53 and @xmath84 respectively .
they are required to satisfy the relations @xmath85 this section is a brief summary of the results obtained in @xcite .
let @xmath3 be an augmented algebra .
recall that the @xmath2-algebra @xmath86 of @xmath3 is defined as the derived functor @xmath87 ; it can be computed in many ways , for instance , via the bar construction .
the _ bar construction _
@xmath88 is a quasi - free coalgebra @xmath89 on the suspension of the augmentation ideal of @xmath3 .
its differential is the unique coderivation extending the product of @xmath3 .
hence , the linear dual @xmath90 of the bar construction is a differential graded algebra , whose underlying cohomology groups are equal to the @xmath2-functor : @xmath91 as the homology of a dg algebra , the @xmath2-functor acquires a canonical associative product ; this algebra is called the _
yoneda algebra_. moreover , using the homotopy transfer theorem @xcite , one can endow the @xmath2-functor with an @xmath14-algebra structure , which extends the yoneda product .
we call it the _ yoneda @xmath14-algebra_. in the @xmath14-setting , we shall , for simplicity , mostly work with non - unital algebras , and therefore consider the augmentation ideal of the usual yoneda algebra .
an @xmath15-algebra @xmath92 is said to be _ reduced _ if @xmath93 when both @xmath94 and @xmath95 are of odd degree , and @xmath96 when at least one of @xmath97 is _ not _ of odd degree .
one of the reasons for the notion of a reduced @xmath15-algebra to be important is explained by the following theorem .
[ thm : reduced ] let @xmath3 be an @xmath0-homogeneous algebra . it is @xmath0-koszul if and only if its yoneda @xmath14-algebra is a reduced @xmath15-algebra generated in degree @xmath31 . the higher operations of the yoneda algebra of an @xmath0-koszul algebra can be described as follows .
[ computeproducts ] the yoneda @xmath14-algebra @xmath8 of an @xmath0-koszul algebra @xmath3 satisfies @xmath98 the operations @xmath17 and @xmath18 on that algebra can be computed using the product on @xmath35 as follows : @xmath99 and are zero for all other choices of arguments .
we denote by @xmath100 the category whose objects are finite ordered sets ( with order - preserving bijections as morphisms ) , and by @xmath101 the tensor category of graded vector spaces ( by definition , morphisms in that category are degree zero linear operators , and the symmetry isomorphism is given by the formula @xmath102 ) .
a _ ( non - symmetric ) collection _ is a contravariant functor from the category @xmath100 to the category @xmath101 .
we shall refer to images of individual sets as _
components _ of our collection .
we denote the set @xmath103:=\ { 1 , 2 , \ldots , k \}$ ] .
let @xmath104 and @xmath105 be two collections .
the _ composition product _ of @xmath104 and @xmath105 is the collection @xmath106 defined by the formula @xmath107)\otimes\left(\bigoplus_{f\colon i\twoheadrightarrow[k]}\q(f^{-1}(1))\otimes\cdots\otimes\q(f^{-1}(k))\right),\ ] ] where the sum is taken over all non - decreasing surjections @xmath108 .
the composition product equips the category of collections with a structure of a monoidal category .
the unit object is the collection @xmath109 with @xmath110 a _ non - symmetric operad _ is a monoid in the monoidal category of collections equipped with the composition product . by functoriality , for every collection @xmath104 and every finite ordered set @xmath111 , we have the corresponding isomorphism @xmath112 , so to define a collection it is sufficient to define a graded vector space @xmath113 the component @xmath114 of this graded vector space may be viewed as the space of @xmath115-ary operations of some sort .
the associativity and the unit condition for a monoid ensure that to know the non - symmetric operad structure on a collection it is sufficient to know all the _ partial compositions _ : those correspond to the surjections @xmath108 in formula ( which we now apply in the case @xmath116 ) for which all the preimages except for @xmath117 for some fixed @xmath118 consist of one element , and such that @xmath119 is equal to the unit , for @xmath120 .
such a partial composition is denoted by @xmath121 .
these partial compositions satisfy the appropriate associativity relations ; namely , if @xmath122 , @xmath123 , @xmath124 , we have @xmath125 these associativity relations include signs coming from the symmetry isomorphisms ; these signs will be of crucial importance in our subsequent computations .
similarly to the case of associative algebras , a non - symmetric operad can be presented via generators and relations , that is , as a quotient of the free non - symmetric operad @xmath126 for some collection @xmath127 .
let us describe an explicit construction of @xmath126 .
suppose that @xmath128 is a collection of ordered sets , that is a functor from @xmath100 to @xmath100 , that provides bases for the components of @xmath127 .
a _ ( planar ) tree monomial _ is a planar rooted tree with labelled vertices : `` planar '' means that for each vertex @xmath129 of a tree , the set of its inputs @xmath130 is ordered , and `` labelled '' means that each vertex @xmath129 carries a label from the set @xmath131 .
we use the notation @xmath132 for the underlying tree of a tree monomial @xmath133 .
for example , if @xmath134 for @xmath135 , then the component @xmath136 has basis elements indexed by pairs @xmath137 .
the free non - symmetric operad @xmath126 admits a basis made up of ( planar ) tree monomials .
an important property of the aforementioned basis is that each composition of tree monomials is again a tree monomial . to be precise
, a tree monomial should be viewed as an equivalence class : to determine the actual composition product in the operad , one has to introduce levels of trees , i. e. decide what should be composed first ( ignoring that would only compute the result up to a sign ) .
a basic example of a non - symmetric operad is the _ endomorphism operad _
@xmath138 of a vector space @xmath34 , whose components @xmath139 is @xmath140 with the partial composition of @xmath141 and @xmath142 given by the composition of functions : @xmath143 . by definition ,
a structure of an _ algebra _ over a non - symmetric operad @xmath104 on a vector space @xmath34 is a morphism of operads @xmath144 .
free @xmath104-algebra _ @xmath145 generated by a vector space @xmath127 is given by the formula @xmath146
in this section , we shall define the notions of grbner bases for non - symmetric operads and their algebras .
some related work for `` free nonassociative algebras '' has been done in @xcite ; a much more general approach , on which our exposition is modelled , is discussed in @xcite within the framework of shuffle operads .
let us note , however , that there is a very important feature of non - symmetric operads that symmetric and shuffle operads do not exhibit : the machinery of grbner bases for non - symmetric operads includes @xmath53-ary operations .
the definitions and results below are completely parallel to those for associative algebras
. basically , grbner bases allow one to find for each algebra @xmath3 a `` monomial replacement '' , that is , an algebra @xmath147 with the same generators and with monomial relations for which the natural monomial basis ( of monomials not divisible by any of relations ) is also a basis for the vector space @xmath3 .
a total ordering of the set of tree monomials is said to be _
admissible _ if the operadic composition of several tree monomials is an increasing function in each argument .
let us describe one of the orderings of tree monomials that we shall use throughout the paper .
this ordering was discovered and used by e. hoffbeck in @xcite . the _ path - lexicographic ordering _ of tree monomials
is defined as follows .
we first assign , to each tree monomial @xmath133 , a sequence of words in the alphabet @xmath148 one word for each leaf of @xmath132 ( here by leaves we mean both the inputs and the vertices that have no inputs but however have labels corresponding to operations with no arguments ) . for each of those leaves , starting from the leftmost one and moving from left to right , we walk along the unique path from the root of @xmath132 to that leaf , and write down the labels of vertices encountered along the way .
we form a word in the alphabet @xmath149 from these labels using the isomorphisms @xmath150 . for each total
ordering of @xmath149 , we are now able to define the corresponding path - lexicographic ordering of tree monomials . to compare two monomials @xmath151 and @xmath133
, we first compare the numbers of leaves ; a tree with the larger number of leaves is , by definition , larger . if two trees have the same number of leaves , we compare the corresponding sequences word by word , comparing words degree - lexicographically . for example , we have and therefore whenever @xmath152 , that is if @xmath153 , or @xmath154 and @xmath155 .
the path - lexicographic ordering is admissible .
in what follows , we fix some admissible ordering of tree monomials in @xmath126 . all our definitions and statements are valid for any choice of such ordering .
a tree monomial @xmath151 is said to be _ divisible _ by a tree monomial @xmath133 if @xmath156 contains @xmath132 as a subtree with matching labels . for
a fixed admissible ordering , we define the _ leading term _ @xmath157 of an element @xmath158 to be the largest tree monomial @xmath133 that appears in @xmath159 with a non - zero coefficient . this nonzero coefficient ( the leading coefficient of @xmath159 ) is denoted by @xmath160 .
assume that a tree monomial @xmath151 is divisible by a tree monomial @xmath133 .
for every element @xmath158 of the same arity as @xmath133 , we can consider the operation that inserts @xmath159 in the place of @xmath133 in @xmath151 ; this produces an element of the free operad of the same arity as @xmath151 .
we denote that operation by @xmath161 .
note that by the construction @xmath162 , and since we work with admissible orderings , it is clear that if @xmath163 , then @xmath164 .
assume that @xmath108 and @xmath159 are two elements of @xmath126 for which @xmath165 is divisible by @xmath157 .
the element @xmath166 is called the _ reduction of @xmath108 modulo @xmath159_. we shall often use the notation @xmath167 for reductions . two tree monomials @xmath151 and @xmath133 are said to have a _ small common multiple _
, if there exists a tree monomial @xmath168 which is divisible by both @xmath151 and @xmath133 for which each of its vertices is contained in either @xmath151 or @xmath133 , and such that the number of vertices of @xmath169 is less than the total number of vertices of @xmath156 and @xmath132 .
assume that the leading terms of two elements @xmath170 have a small common multiple @xmath168 .
the element @xmath171 is called the _ @xmath151-polynomial of @xmath108 and @xmath159 _ ( corresponding to the small common multiple @xmath168 ) .
a _ grbner basis _ of an ideal @xmath172 in the free non - symmetric operad is a system @xmath173 of generators of @xmath172 for which the leading tree monomial of every element of @xmath172 is divisible by one of the leading terms of elements of @xmath173 . a grbner basis @xmath173 is said to be _ reduced _ if none of the leading terms of elements of @xmath173 are divisible by leading terms of other elements of @xmath173 .
one can prove that every ideal of the free non - symmetric operad admits a unique reduced grbner basis .
a tree monomial is said to be _ normal _ with respect to a system of elements @xmath173 if it is not divisible by any of the leading terms of elements of @xmath173 .
it is easy to see that iterated reductions modulo elements of some set @xmath173 allow one to replace every element of a free operad by a linear combinations of tree monomials that are normal with respect to @xmath173 . in the case of a grbner basis
, this representation produces what is natural to think of as a canonical form of an element , as the following proposition shows .
suppose that @xmath173 is a grbner basis of an ideal @xmath174
. then normal tree monomials with respect to @xmath173 form a basis of the quotient @xmath175 . a useful criterion for a set of generators @xmath173 to form a grbner basis is given by the following version of bergman s diamond lemma @xcite .
the proof is analogous to that for diamond lemma and is omitted .
[ diamondlemma ] suppose that @xmath173 is a system of generators of @xmath172 .
then @xmath173 forms a grbner basis if and only if each such s - polynomial of two elements of @xmath173 can be reduced to zero modulo @xmath173 .
this theorem allows one to come up with an algorithm , analogous to buchberger s algorithm in the commutative case @xcite , that computes for each operad presented by generators and relations its reduced grbner basis . in what follows ,
a computation of a grbner basis of a certain non - symmetric operad will be done from the beginning to the end , which will hopefully be instructional enough to the reader . unlike the case of symmetric operads where
first steps in studying an operad can be done without symmetries by passing to shuffle operads @xcite but for algebras over a given operad there is no clear way to define a theory of grbner bases , for non - symmetric operads one can handle the corresponding algebras as well .
let @xmath176 be a non - symmetric operad , and @xmath3 be an algebra over @xmath104 . by a slight abuse of notation , we think of @xmath3 as of a collection concentrated in arity @xmath53 .
the following construction is an explicit adaptation for the non - symmetric case of the construction of the _ enveloping operad _ for the pair @xmath177 , see @xcite and references therein .
the _ extension of constants in @xmath104 by @xmath3 _ , denoted @xmath178 , is the operad @xmath179 where @xmath180 consists of all the relations @xmath181 with @xmath182 , and the elements @xmath183 viewed as operations of arity @xmath53 in @xmath178 on the left - hand side ( so that the non - symmetric composition @xmath184 can be evaluated ) , and as elements of the algebra @xmath3 on the right - hand side ( so that @xmath185 acts on these elements via the homomorphism @xmath186 ) .
we have @xmath187 .
consequently , it becomes clear how to develop grbner bases for algebras over any non - symmetric operad : one should pass to the extension of constants , compute the operadic grbner basis , and restrict it to arity @xmath53 .
this results in the following theorem .
[ basis - extension ] let @xmath176 be a non - symmetric operad , and let @xmath3 be an algebra over @xmath104 .
if @xmath173 is the reduced grbner basis of the operad @xmath178 , then the normal tree monomials of arity zero not belonging to @xmath104 form a basis of @xmath3 . in the case of the operad of associative algebras ,
this leads to the standard notion of a monomial replacement coming from the theory of grbner shirshov bases @xcite , whereas in various other case this produces new definitions that are to be explored .
to the best of our knowledge , even for the simplest case of algebras with two associative products satisfying no identities with each other , the corresponding grbner bases theory is not known .
the general approach of the operad theory is to express universal results in terms of relations that operations satisfy when they are evaluated on any elements . in this section ,
we suggest relations to provide a replacement for the property of an @xmath15-algebra to be reduced .
namely , we introduce a quotient of the operad @xmath15 that naturally acts on all reduced @xmath15-algebras .
we conclude with the description of its reduced grbner basis .
the operad @xmath16 is generated by two operations , a binary operation @xmath17 and an @xmath0-ary operation @xmath18 that satisfy the identities @xmath188 the operations of any reduced @xmath15-algebra satisfy the relations of the operad @xmath16 .
indeed , @xmath96 when at least one of @xmath97 is not of odd degree .
also , in the case when all the elements @xmath97 are of odd degree , the element @xmath189 is of even degree . we immediately conclude that the operation @xmath190 is identically zero . for the condition of being a reduced @xmath15-algebra in the sense of @xcite ( which is similar to the one used in this paper but includes congruences modulo @xmath191 ) relevant for the concept of `` piecewise - koszul algebras ''
, one can show in the exact same way that the corresponding algebras carry canonical @xmath16-algebra structures .
[ a2n - constr ] for every weight graded algebra @xmath3 concentrated in degree zero , we consider the vector space @xmath92 whose component of degree @xmath192 is defined as @xmath193 the operations @xmath17 and @xmath18 given by the product on @xmath3 by @xmath194 and defined to be zero for all other choices of arguments , endow @xmath92 with a @xmath16-algebra structure .
it is clear , since this algebra is easily seen to be a reduced @xmath15-algebra . for every @xmath0-koszul algebra , the operations on its yoneda @xmath14-algebra satisfy the relations of the operad @xmath16 . indeed , by theorem [ computeproducts ] , the yoneda @xmath14-algebra of an @xmath0-koszul algebra @xmath3 is the algebra @xmath92 obtained from the algebra @xmath35 by the construction we just described above .
the yoneda @xmath14-algebra of the algebra @xmath195 turns out to be a @xmath16-algebra which is not reduced as an @xmath15-algebra ( in any sense of the word `` reduced '' used in the literature ) , contrary to the feeling that the previous statements might create .
this example suggests that the notion of a @xmath16-algebra is well suited for generalisations of the @xmath0-koszul duality theory including relations of weight @xmath28 and @xmath0 .
this algebra is indeed @xmath28-@xmath0-koszul in the sense of green and marcos @xcite .
we shall explore this observation further in a sequel paper .
it turns out that yoneda @xmath14-algebras of @xmath0-koszul algebras admit compact presentations as algebras over the operad @xmath16 . in order to prove that
, we shall study the operad @xmath16 in more detail , exhibiting its reduced grbner basis . in what follows ,
we use the path - lexicographic ordering on tree monomials , assuming that @xmath17 is greater than @xmath18 lexicographically .
[ thm : gbb2n ] the reduced grbner basis of the operad @xmath16 is obtained from its defining relations , , and by adjoining , for each @xmath196 and for each @xmath197 , the relation @xmath198 with the operations @xmath199 defined inductively by @xmath200 , @xmath201 . by theorem [ diamondlemma ] ,
it is enough to prove that all s - polynomials coming from small common multiples of leading terms of the given relations can be reduced to zero using the same system of relations .
we shall explain how the proof goes , and demonstrate it in the case of low arities , since in further arities all the computations one has to performed are modelled on those for low arities .
the s - polynomial that arises from the small common multiple the leading term @xmath202 of the relation @xmath203 has with itself can be reduced to zero using just that relation ; this can be interpreted in many different ways including the setup of the maclane coherence axiom for monoidal categories ; see ( * ? ? ?
* section @xmath204 ) for details .
the s - polynomial that arises from the small common multiple of the leading terms @xmath202 and @xmath205 of the relations @xmath203 and @xmath206 respectively can as well be reduced to zero using only those operations and nothing else ; this is a result of a more tedious but straightforward computation .
however , the leading term @xmath205 of the relation @xmath206 forms @xmath115 small common multiples with the relations @xmath190 , and the corresponding s - polynomials can not be reduced using only the defining relations of @xmath16 ; that is where new elements of the reduced grbner basis start showing up .
we shall exhibit here the first steps of the respective computations , the further steps being completely analogous .
in what follows , we use the notation @xmath207 ( @xmath208 , @xmath197 ) for the relation given by formula .
for the small common multiple @xmath209 the corresponding s - polynomial can be , using the relations @xmath210 , be reduced to @xmath211 where the first equality comes from the graded associativity of the operad composition , and the second one uses reductions modulo the defining relations and . for @xmath212 ,
the small common multiple @xmath213 gives rise to the s - polynomial @xmath214 which can be reduced modulo the defining relations and to @xmath215 altogether these elements can be reduced to zero using the relations @xmath216 which are precisely the relations @xmath217 of what we want to prove to be the reduced grbner basis .
the s - polynomials corresponding to the small common multiples of the monomial relations @xmath210 and @xmath218 are trivially zero , so there is nothing to check .
all the s - polynomials corresponding to the small common multiples of the leading terms the defining relations have now been treated .
let us now study the small common multiples of the leading terms of the defining relations with the leading terms of relations .
all the small common multiples of the leading terms of the relations @xmath217 with the monomial relations @xmath219 can be easily reduced to zero without using any new relations , as well as the small common multiple @xmath220 of the leading term of the relation @xmath217 with the leading term of @xmath206 .
it is also easy to see that there are no new relations needed to reduce all the s - polynomials arising from the small common multiples of the leading terms of relations @xmath217 and @xmath221 . indeed , there are two combinatorially different kinds of small common multiples of that sort .
the small common multiple @xmath222 for @xmath223 leads to the s - polynomial @xmath224 which can be reduced to zero modulo the relations @xmath225 for various @xmath192 ( taking into the account the graded associativity of the operad composition ) .
the small common multiple @xmath226 for @xmath227 leads to the s - polynomial @xmath228 where the second term is immediately reduced to @xmath229 using the relation @xmath217 twice , while the first term is reduced to the same result through a lengthier sequence of reductions ( depending on @xmath230 being less than , equal to or greater to @xmath0 ) .
the small common multiple @xmath231 of the leading term of the relation @xmath217 with the leading term of @xmath203 creates an s - polynomial that can be reduced to @xmath232 , and hence can be reduced to zero using all the elements we have .
the last computation at this stage is that for the s - polynomial coming from a yet another small common multiple of the leading terms of the relations @xmath217 and @xmath206 , that is @xmath233 .
however , it is true that this s - polynomial can be reduced completely using the defining relations of the operad together with the relations @xmath234 and @xmath232 . the way the elements @xmath235 etc . arise is exactly similar , and our system of elements can be shown to be sufficient to reduce to zero all arising s - polynomials .
[ basis - b ] a basis for the operad @xmath16 can be defined inductively as follows .
the identity map @xmath236 is a basis element , for every basis element @xmath237 , @xmath238 is a basis element , and also for each choice of nonnegative integers @xmath239 , @xmath240 is a basis element
. indeed , these are precisely the normal tree monomials with respect to the leading terms of the elements of the grbner basis we computed .
in this section , we use the aforementioned operadic notions to provide a natural framework to describe the koszul dual algebras of @xmath0-koszul algebras functorially via generators and relations . [
thm : presentationkdinfty ] let @xmath22 be a finitely generated @xmath0-homogeneous algebra .
then @xmath241 as @xmath16-algebras , where @xmath24 is viewed as a subspace of @xmath25 .
let us first prove a particular case of this result , and then use it in the general case .
this theorem holds for @xmath242 .
let us remark that , in this case , the algebra on the right - hand side of is the algebra @xmath243 of which all the algebras of the right - hand sides of for various @xmath38 are quotients .
applying the methods we explained in the end of section [ sec : grobner ] , we shall study the corresponding extension @xmath244 .
if we fix some basis @xmath245 of @xmath246 , the relations needed to present the extension are @xmath247 . the small common multiple of the associativity relation @xmath203 and the relation @xmath247 produces the relation @xmath248 .
this extra relation has no small common multiples with the other relations .
applying theorem [ basis - extension ] , in the view of theorem [ thm : gbb2n ] and corollary [ basis - b ] , we immediately conclude that a basis for the algebra @xmath249 can be defined inductively as follows .
it has even and odd elements , all generators @xmath250 , , @xmath251 are odd basis elements , for every even basis element @xmath237 , and each @xmath252 , the element @xmath253 is an odd basis element , and also for each odd basis element @xmath237 , and for each @xmath254 , the element @xmath255 is an even basis element . in other words , our algebra has nonzero elements only of weight divisible by @xmath0 or congruent to @xmath31 modulo @xmath0 , and for each such weight there exists exactly one type of basis element of that weight .
combinatorially , the corresponding trees are alternating `` towers '' of operations with all the compositions using the last slot of an operation only .
this gives a vector space identification of @xmath256 with the yoneda @xmath14-algebra @xmath257 of the algebra @xmath258 .
comparing the operations of @xmath249 with those given by corollary [ a2n - constr ] , we see that the corresponding @xmath16-algebras are isomorphic .
let us prove the theorem for a general set of relations @xmath38 . in the proof above for the case of the algebra @xmath249 , corresponding to @xmath242 , we obtained a basis where we alternate the operations @xmath17 and @xmath18 , computing all compositions at the last slot , and then substitute into the resulting operation an arbitrary word in @xmath250 , , @xmath251 .
now , the defining relation @xmath206 , together with the vanishing of all the elements , , and mean that in our alternating towers , the only operation we plug in at each level can be freely moved between the slots .
if we now impose the additional relations @xmath259 and use the identification @xmath260 discussed above , it becomes clear that the underlying vector space of the quotient that we are studying is the direct sum of the appropriate homogeneous components of the algebra @xmath261 .
the isomorphism of @xmath16-algebras follows , yet again , from corollary [ a2n - constr ] .
an @xmath0-homogeneous algebra @xmath22 is @xmath0-koszul if and only if the yoneda @xmath14-structure on its @xmath2-algebra factors through the operad @xmath16 , and @xmath262 as @xmath16-algebras .
this is a direct corollary of theorem [ computeproducts ] and theorem [ thm : presentationkdinfty ] .
this section provides a conceptual framework for the main constructions and statements of the @xmath0-koszul duality theory .
it proves the equivalence between finding a resolution of the trivial @xmath3-module and finding a quasi - free algebra resolution of @xmath3 .
the _ convolution @xmath15-algebra _ is defined by the space @xmath265 of linear maps from @xmath74 to @xmath3 endowed with the following operations @xmath266 where @xmath267 stands for any @xmath268 iterations of the associative product @xmath269 . inside the convolution @xmath15-algebra
, we consider the _ maurer
cartan equation _ : @xmath270 solutions of degree @xmath271 to the maurer
cartan equation are called _ twisting morphisms_. the associated set is denoted by @xmath272 . the chain complex @xmath279 associated to a twisting morphism @xmath280 is called a _ twisted tensor product_. when this chain complex is acyclic , the twisting morphism @xmath280 is called a _
koszul morphism_. the set of all koszul morphisms is denoted by @xmath281 .
the data of an @xmath15-coalgebra is equivalent to a square - zero derivation @xmath282 on a free algebra @xmath283 , on the homological desuspension of @xmath74 , such that the components of the restriction @xmath284 vanish for @xmath73 . because of the universal property of free algebras ,
defining a morphism of graded algebras from @xmath286 to @xmath3 is equivalent to defining a degree @xmath271 map of graded vector spaces from @xmath74 to @xmath3 .
inside the space @xmath287 of all morphisms of graded algebras , there is a subspace @xmath288 consisting of morphisms of dg algebras .
it includes an even smaller subspace @xmath289 consisting of quasi - isomorphisms of dg algebras .
we denote the unique morphism of graded algebras associated to a degree @xmath271 linear map @xmath273 by @xmath292 .
it is easy to check that it commutes with the differential of the cobar construction if and only if @xmath280 is a twisting morphism .
let us prove the last equivalence .
first , any @xmath293-coalgebra @xmath263 admits a canonical twisting morphism @xmath294 it satisfies the following universal property : any twisting morphism @xmath295 factors through it @xmath296^{g_\alpha } & \\ c \ar[rr]^{\alpha}\ar[ur]^{\iota } & & a \ . } \ ] ] hence , the tensor product of the map @xmath297 with the identity map on @xmath74 is a map of chain complexes @xmath298 under the connectedness assumptions , we can apply the same arguments as the comparison lemma for twisted tensor products ( * ? ? ?
* lemma @xmath299 ) .
the twisted tensor product @xmath300 is acyclic by ( * ? ? ?
* theorem @xmath301 ) .
therefore , the map @xmath297 is a quasi - isomorphism if and only if the twisted tensor product @xmath302 is acyclic .
this theorem demonstrates that the existence of a quasi - free algebra resolution @xmath303 of an algebra @xmath3 is equivalent to the existence of a free @xmath3-module resolution @xmath304 of the ground field @xmath26 .
so , given an algebra @xmath3 , for instance an @xmath0-homogeneous algebra , one has to look for an @xmath15-coalgebra @xmath74 ( or more generally an @xmath14-coalgebra ) together with a koszul morphism @xmath305 to solve these two questions .
[ @xcite ] for any augmented algebra @xmath3 , the bar - cobar construction @xmath306 is a functorial resolution .
the twisted tensor product @xmath307 , associated to the twisting morphism @xmath308 , is a resolution of the ground field @xmath26 .
this general idea to reduce to size of these two resolutions is to consider the underlying homology groups @xmath309 instead of the bar construction @xmath88 together with the transferred @xmath14-coalgebra structure .
this is precisely what the higher koszul duality does , by proposing a candidate for @xmath309 . since @xmath313
, we automatically have @xmath314 .
the image of @xmath315 under @xmath316 is equal to zero , since the following composite vanishes @xmath317 for weight grading reasons , the other components of @xmath318 vanish , which concludes the proof .
this theorem demonstrates that the koszul dual @xmath15-coalgebra construction of theorem [ thm : presentationkdinfty ] provides a good candidate to get a `` small '' quasi - free algebra resolution of an @xmath0-homogeneous algebra @xmath3 and to get an equally `` small '' free @xmath3-module resolution of the ground field @xmath26 . in the koszul duality theory ,
this latter resolution is used to compute @xmath1 and @xmath2 functors , see @xcite and to establish equivalence between derived categories , see , e. g. , @xcite and @xcite .
the data of a quasi - free algebra resolution is used in the homotopy transfer theorem ( * ? ? ?
* section @xmath323 ) .
bruno buchberger , _ an algorithm for finding a basis for the residue class ring of a zero - dimensional polynomial ideal _ ,
thesis , university of innsbruck , 1965 ( german ) , journal of symbolic computation , * 41 * ( 2006 ) , 471511 .
di ming lu , john h. palmieri , quan shui wu , and james j. zhang , _
@xmath3-infinity algebras for ring theorists _ , in : `` proceedings of the international conference on algebra '' , algebra colloq .
* 11 * ( 2004 ) , no . 1 , 91128 . | we present a unifying framework for the key concepts and results of higher koszul duality theory for @xmath0-homogeneous algebras : the koszul complex , the candidate for the space of syzygies , and the higher operations on the yoneda algebra .
we give a universal description of the koszul dual algebra under a new algebraic structure .
for that we introduce a general notion : grbner bases for algebras over non - symmetric operads . |
the galactic center ( gc ) of our galaxy found in 1974 as a strong radio source called sgr a@xmath10 is the site a supermassive black hole weighing 2.6 millions solar mass is present ( e.g. melia and falcke 2001 ) .
morever the total mass of stars enclosing the inner 20 pc of the galaxy is @xmath11 m@xmath12 , or @xmath13% of the galactic stellar mass ( baganoff et al .
therefore , the x - ray observations of the galactic center can potentially reveal accreting black holes and neutron stars . during five years of chandra observations , baganoff et al . identified seven x - ray transients located within 20 pc of sgr a@xmath10 .
remarkably four of these transients lie within only 1 pc of sgr a@xmath10 ( muno et al .
this implies that , compared to the numbers of similar systems located between 1 and 20 pc , transients are over - abundant by a factor of @xmath14 per unit stellar mass within 1 pc of sgr a@xmath10 .
it is also now a subject of intense activity due to the possible presence of a dark matter cusp inside the galactic center .
numerous space missions have observed the center of our galaxy in radio , infrared , x and @xmath15 rays .
thanks to these measurements we have a more precise picture of this region . the center of our galaxy presents the characteristic that there is a strong activity both in x and @xmath15 rays . a recent hess ( high energy stereoscopic system ) experiment reports a detection of very high energy gamma rays from the galactic center within @xmath16 of sgr a@xmath10 . the flux above 165 gev is order of @xmath17 m@xmath18 s@xmath2 and consistent with a luminosity of @xmath19 ergs s@xmath2 ( aharonian , et al .
2004 ) .
the most intense line of our galaxy is coming also from gc : the 511 kev annihilation line .
balloon flight observations reported in 1972 showed first evidence for the existence of an annihilation line from the galactic center region ( johnson , harnden & haymes 1972 ; see also johnson & haymes , 1973 ) .
since that time the galactic center has been the object of intense observations from space missions and balloon experiments ( leventhal , maccallum & stang 1978 ; purcell et al .
1993 ; ramaty , skibo & lingenfelter 1994 ; purcell et al . 1997 ; and cheng et al . 1997 ; share et al . 1990
; teegarden et al . 1996 ; gehrels et al . 1991 ; leventhal et al . 1993 ; niel et al . 1990 ; chapuis et al .
the observed emission of the brightest gamma - ray line in our galaxy is due to two separate sources : a diffuse one and a point source near the galactic center .
for the point source we use the value given by integral from recent observations ( churazov et al . 2005 ) .
the point source line flux is : @xmath20 ph @xmath1 s@xmath2 , for which the corresponding luminosity of the line emission is about @xmath21 ergs s@xmath2 for the distance of about 8 kpc
. then the positron sources are produced at a rate given by : @xmath22s@xmath2 , where @xmath23 is the distance expressed in unit of 8 kpc .
this observation is in a good agreement with the previous measurements and is well fitted by a gaussian with the full width at half maximum of @xmath24 10 degrees , with a 2 @xmath25 confidence interval of 6 - 18 degrees.the integral measurement providing the line width of @xmath26 kev exlcudes velocity @xmath27 of the interstellar medium ( ism ) greater than @xmath28 km s@xmath2 , i.e. @xmath29 .
several possible sources of the observed line related to the positron sources have been proposed .
for example , dermer & skibo ( 1997 ) suggest that these sources could be associated with galactic fountains of radioactive elements ejected from supernova activity in the galactic center .
boehm , fayet & silk ( 2004 ) offer an alternative explanation that the observed annihilation line could be related to the annihilation of dark light matter particles into @xmath30 .
in fact , guessoum , jean & gillard ( 2005 ) present the list of the main processes that led to the positron production .
they are : ( i ) .
the @xmath31 decay of a radioactive nucleous which is widespread in the explosive and/or hydrostatic nucleosynthesis enviroments of novae , supernovae , wolf - rayet , and asymptotic giant branch stars , ( ii ) . @xmath32 decay into @xmath33 and gives off a positron taking place where the pion is produced in collisions of highly energetic ( @xmath34 mev ) cosmic rays with interstellar material , ( iii ) .
pair production by interaction of an electron with a strong magnetic field and ( iv ) the process of the photon - photon pair production which requires photons of high energies occuring in the enviroment of luminous compact objects , black holes , active nuclei ( agn ) an etc .
thus one can not exclude _ that pair production is a result of the interaction of x - ray and @xmath35 emission in the very compact shell near galactic center_. x - ray observations of the galactic center have been made with several instruments ( watson et al .
1981 ; skinner et al . 1987 ; kawai et al . 1988 ; takeshima et al . 1991
; sunyaev et al .
1993 ; markevitch et al . 1993 ; pavlinsky et al .
1993 ; preddehl , & trumper 1994 ) .
high - sensitivity imaging observations with the einstein observatory ( watson et al . ) and rosat ( preddehl , & trumper ) resolved several discrete sources within @xmath36 of galactic center .
one of the rosat sources , rxj 1745.6 - 2900 ( preddehl & trumper ) , is within @xmath37 coincident of the nonthermal radio source sgr a @xmath10 , which is believed to be the dynamical center of our galaxy .
the japanese satellite ginga having detected k@xmath38 lines from highly ionized iron , found a large amount of hot plasma which is strongly concentrated towards the galactic center ( koyama et al .
1989 ; yamauchi et al . 1990 ) .
the asca image ( tanaka et al . 1995 ) has revealed extended diffuse emission over the galactic center region .
the observed spectrum shows the k@xmath39 lines from highly - ionized ions of silicon , sulphur , argon , calcium , in addition to the high - ionization iron lines .
this gives firm evidence for the presence of high - temperature ( @xmath40 kev ) plasma .
a large energy generation rate of @xmath41 ergs s@xmath2 is required to produce the observed plasma .
in fact , koyama et al .
( 1996 ) argue for a source at the galactic center .
the luminosity required to account for the 6.6-kev line from sgr b2 ( the prominent molecular cloud ) is @xmath42 ergs s@xmath2 ( in 2 - 10 kev band ) .
while the sgr a region has an x - ray luminosity of @xmath43 ergs s@xmath2 at present , it could have been much more luminous @xmath44 yr ago .
the distinctly bright region inside the sgr a shell is probably due to this recent activity and the x - ray luminosity may have been as high as several @xmath45 ergs s@xmath2 .
koyama et al .
conclude that the asca results suggest the presence of an active galactic nucleus and together form the evidence for a large mass concentration at the galactic center and transient activities due to sporadic mass accretion rate .
muno et al .
( 2004a ) also presented the study of the spectrum of diffuse emission in @xmath46 field around sgr a@xmath10 during 625 ks of chandra observations .
they confirmed and extended the results of the previous study ( see e.g. koyama et al . ) .
in particular , they showed that the contiuum and line spectra are consistent with an origin in a two - temperature plasma .
as the cooler , @xmath47 kev plasma differs in the surface brightness , the hotter , @xmath48 kev component is more spatially uniform .
recently revnivtsev et al .
( 2004 ) have published integral data on giant molecular cloud sgr b2 s ( igr j17475 - 2822 ) fluorescence .
they have constructed the broad band ( 3 - 200 kev ) spectrum from the asca - granat - integral data .
it appears that the x - ray irradiating spectrum of the cloud is , in fact , not free - free , ( as early proposed irradiating spectrum , see above ) and it is rather quite hard , extending to over 100 kev and probably a power law with photon index 1.8 .
revnivtsev et al
. concluded that 300 - 400 years ago sgr a@xmath10 was a low luminosity ( of @xmath49 ergs s@xmath2 at 2 - 200 kev ) agn with characteristic hard x - ray comptonization spectrum with photon index of about 1.8 and plasma temperature is about 50 kev .
the @xmath35 emission between 1 mev and 30 mev has been measured by comptel .
using the comptel data points ( hunter et al .
1997 ) one can calculate that the luminosity in 1 - 30 mev energy band is about @xmath50 ergs s@xmath2 with an assumption that the distance to the galactic center is about 8 kpc .
in this paper , we study a possibility of pair creation due to x - ray and @xmath35 photon interaction inside the compact galactic center shell ( the sgr a shell ) and we discuss a possibility of annihilation line formation when these pairs propagate through the galaxy from the galactic center region . in
2 we present a pair creation model . in
3 we compare our theoretical predictions for the pair and annihilation photon production with observations .
we determine the size of the pair production area using the observable fluxes in x - ray and @xmath35ray energy ranges . in
4 we investigate the outward propagation of positrons from the galactic center region . in
5 we present arguments for the origin of the @xmath51 mev emission near galactic center .
we discuss our results in 6 and we make our final conclusions in 7 .
we assume that the photon density of x - ray photons emitted by the inner shell near the central bh within @xmath52 @xmath53 ( schwarzschild radii ) is described by the following distribution over radius @xmath54 ( see fig . 1 for the geometrical description of the model ) @xmath55 where @xmath56 is x - ray luminosity of the central source per kev ( x - ray energy spectrum )
, @xmath57 is the energy of x - ray photon and @xmath58 is the speed of light .
also we consider that the spherical shell between the inner radius @xmath59 and @xmath60 ( @xmath61 and @xmath62 ) is uniformly filled by @xmath35radiation sources where photon density production @xmath63 is @xmath64 where @xmath65 is the @xmath15-radiation luminosity of the spherical shell per kev ( @xmath15-ray energy spectrum ) and @xmath66 is the energy of the @xmath15-ray photon .
then the pair production by the photons ( for which the pair production condition @xmath67 is satisfied ) is @xmath68 where @xmath69 is the optical path for @xmath15-ray photon propagating in the x - ray background and incoming at radius @xmath54 at given direction @xmath70 .
the multiple integral in equation ( 3 ) can be calculated analytically .
in fact , @xmath71 @xmath72 + \int_{-1}^{0}d\mu \int_{r}^{\infty}n(r^{\prime})ds\right\}\ ] ] where @xmath73^{1/2}}\ ] ] and @xmath74 ; where @xmath75 is a pair creation cross section given by ( dirac , 1930 , heitler , 1954 ) @xmath76\ ] ] and @xmath77^{1/2}$ ] , and where the classical electron radius @xmath78 cm .
the analytical integration of @xmath79 @xmath80 allows us to calculate the second integral of @xmath79 over @xmath81 : @xmath82 finally , the expression for the green s function @xmath83 can be written as @xmath84 in order to obtain the total pair production rate @xmath85 one should integrate @xmath83 over x - ray and @xmath35 energies , namely @xmath86\sigma_{x,\gamma}l_xl_{\gamma}}{e_xe_{\gamma}}de_{\gamma } \frac{\sigma_{x,\gamma}{\cal l}_x{\cal l}_{\gamma}}{e_xe_{\gamma}}de_{\gamma}.\ ] ]
to compute the positron flux one needs the x and @xmath87 ray energy spectra , @xmath56 and @xmath65 respectively .
we will use for the total x - ray and @xmath15 luminosities of the inner gc region , @xmath88 ergs s@xmath2 and @xmath89 ergs s@xmath2 respectively ( see a review of the observations in the introduction ) .
we assume that the @xmath35 radiation emanates from a optically thick medium and consequently the emergent spectrum has a blackbody like shape that color temperature @xmath90 mev ( see fig .
1 ) . then for a normalization constant @xmath91 we have an equation as follows @xmath92 where @xmath93 . for the x - ray photons
, we assume a power law with an exponential cutoff @xmath94 where @xmath95 , @xmath96 kev , the spectral index @xmath97 and @xmath98 is the normalization constant . from the previous section we saw that the process of pair creation is driven by the cross section given by equation ( 6 ) . the energy dependence , represented by the parameter @xmath99 can be expressed using the dimesionless energies @xmath95 and @xmath100 , namely @xmath77^{1/2 } = [ 1 - 3.26/xy]^{1/2}$ ] .
then the formula for the pair flux @xmath101 can be modified using the variables @xmath102 and @xmath103 as follows @xmath104 we compute this double integral and obtain the following result : @xmath105 to reproduce the value of @xmath106s@xmath2 for the positrons rate , one needs a value of the radius @xmath60 , given by : @xmath107 it is worth noting that the intrinsic pair flux @xmath108 s@xmath2 , formed as a result of @xmath109 interaction ( to calculate this flux one should replace @xmath110 by @xmath111 in formula 14 ) in this @xmath35emission region of size @xmath112 cm , is six orders of magnitude less than the pair flux @xmath113 s@xmath2 required to reproduce the observed annihilation line strength .
a positron , @xmath7 may collide with an electron @xmath114 , to produce two gamma ray photons according to the reaction @xmath115 . before decaying
, the positron in the free space interacts with an electron and forms the positronium , a bound state with lifetime of order of @xmath116 s. then particles annihilate into two or three photons .
one photon will have a high energy and , if the electron is at rest ; the other photon will have an energy of the order @xmath117 kev . in our calculations
we assume that most of the produced positrons are thermalized in the surrounding thermal plasma and thus we can consider the annihilation of thermal positrons at a temperature of order @xmath118 k , i.e. @xmath119 , where @xmath120 is the thermal velocity of electrons .
in fact , the recent integral observations of 511 kev annihilation line ( churazov et al . 2005 ) exclude velocity of the interstellar medium greater than @xmath121 . for @xmath122 , the cross - section @xmath25 , for positron annihilation with a free electron at rest
is given by dirac ( 1930 ) @xmath123 murphy , dermer & ramaty ( 1987 ) showed that the fraction of @xmath7 that annihilites in flight prior to thermalizing usually amounts to less that @xmath124 % , and that these positrons do not contribute to the 511 kev line emission . after entering the thermal pool , the positron annihilate in a fully ionized thermal plasma through direct annihilation .
the problem of the positron thermalization and annihilation photon production in the interstellar medium has been recently studied in details by guessoum , jean , & gillard ( 2005 ) ( see for a review of this subject ) .
they reexamine in the utmost detail the various processes undergone by positron annihilation using most recent interaction cross section with atomic and molecular hydrogen , as well as helium . putting all the new calculations together
, they constructed annihilation spectra of ism . the fair question is how far the positrons of energy about one mev can travel from the galactic central region to be annihilated . in order to answer to this question one has to compare the diffusion ( travel ) time with the energy loss ( thermalization ) time with taking into the main energy loss mechanisms . in general terms , the particle diffusion is related to escape of a magnetized turbulent plasma .
jean et al .
( 2005 ) present a spectral analyzis of the @xmath125 annihilation emission from the gc region based on the first year of measurements made with the spectrometer spi of the integral mission .
they also analyze the positron diffusion propagation in the gc region in detail .
the quasilinear diffusion coefficient the diffusion coefficient @xmath126 ( melrose 1980 ) can be expressed as : @xmath127 with @xmath128 the bohm diffusion coefficient , @xmath129 is the maximum scale of the turbulence , @xmath130 for a kolmogorov turbulent spectrum , @xmath131 is relative perturbation of the magnetic field pressure which is often approximated to 1 , @xmath27 is the positron velocity and @xmath132 is the larmor radius ( gyroradius ) @xmath133 @xmath134 is the positron ( electron ) charge , @xmath135 is the magnetic field strength , @xmath136^{-1/2}$ ] .
the maximum scale @xmath129 was estimated to be @xmath137 pc from the measurements of ism turbulence ( armstrong et al . 1995 ) .
recently the ( mean ) magnetic field strength in the gc region was measured by la rosa et al .
( 2005 ) , hereafter lar05 , to be @xmath6 @xmath81 g .
lar05 have used the very large array in all four configurations to image the gc region at 74 mhz .
the resulting image of large scale ( @xmath138 ) of nonthermal synchrotron emission presented in lar05 has a resolution of @xmath139 . at distance of 8 kpc
the angular scales of @xmath138 corresponds to region @xmath140 pc and that of @xmath139 corresponds to 0.09 pc .
it has be noted that in the past several authors ( see morris & serabyn et al .
1996 and references therein ) estimated magnetic fields values of @xmath141 g in the gc region .
this is 2 orders of magnitude larger than the value ( @xmath6 @xmath81 g ) obtained by lar05 .
thus for the mean b of @xmath124 @xmath81 g and for 1 mev positrons @xmath142 @xmath143s@xmath2 . the low limit time scale for 1 mev positrons to thermalize in a the @xmath144 @xmath81 g , and @xmath145 cm @xmath146 density region due to ionization , bremsstrahlung , synchrotron and other processes is @xmath147 years
. then the distance travelled by diffusion in a time of @xmath148 years ( slowing down time plus annihilation time in the standard grain model ) is @xmath149 which is much greater than typical half - size of the hot gc region .
the volume filling factor of the hot gc region 0.72 is highest among all phases of the galactic bulge ( see more details of these estimates in jean et al . 2005 ) . using the spectral data and positron propagation analysis jean et al .
( 2005 ) _ come to conclusion that the sources of annihilation emission are diffusively distributed over the galaxy_. they also explain that the lack of annihilation emission from the gc hot gas is due to its low density , which allows positrons to escape this phase .
it is worth noting liu , petrosian & melia ( 2004 ) find that an outflow of low energy electrons of order 1 mev are distributed over a spatially larger scale than that of the bh inner region ( see more discussion of their work in section 6.2 ) .
namely , the escape time of these electrons is more than a factor of 2 shorter than the time scales for acceleration and losses . on the hand
one can argue that the positrons can be diffusely trapped , thermalized and annihilated in the region that size @xmath150 is of order a few parsec and where the magnetic field can be much higher than @xmath6 @xmath81 g .
melia & falcke ( 2001 ) suggest that out to @xmath151 pc ( essentially inside the cavity surrounded by the circumnuclear disk ) , the field is turbulent ( matching the turbulent plasma generated by wind - wind interactions from the wolf - rayet and o / b stars in this region ) , can be as high as @xmath152 mg .
so the problem is that the gyroradius of the electrons and positrons produced in gamma - gamma interactions , will be far smaller than the region over which the positrons are believed to be annihilating .
consequently , the pairs should be radiating profusely via synchrotron , bremsstrahlung , and inverse compton .
the photon energy density in this region is about @xmath153 ev cm@xmath146 . to obtain this number
we assume that the bolometric luminosity of the source in the galactic center is of order @xmath155 erg s@xmath2 ( see e.g. narayan et al .
thus the derived magnetic strength @xmath156 using equipartition between photons and magnetic field energies is about @xmath124 @xmath81 g .
for such a low magnetic field we have already demonstrated that the positrons can escape from the gc region .
in fact , longair ( 1994 ) derived the minimum magnetic energy requirement for a given luminosity @xmath157 and photon volume v per unit time ( in our case @xmath158 .
he shows that @xmath159 which value is higher than that obtained using relation @xmath160 .
but the ratio of @xmath161 is not orders of magnitude .
the difference of the indicies of ratio @xmath162 is @xmath163 vs @xmath164 .
furthermore longair found that the magnetic field strength , @xmath165 , corresponds to approximate equality of the energies in the particles and magnetic field .
lar05 confirm this conclusion .
they show that the mean magnetic field in the gc region @xmath124 @xmath81 g inferred from their observations is consistent with the particle energy density of @xmath166 ev found in cosmic - ray data .
the turbulent diffusion time scale @xmath167 is not determined by the magnetic field only .
in fact for @xmath130 the diffusion coefficient @xmath126 is proportional to @xmath168 ( see eq.[d_ql ] ) .
even if the magnetic field is about @xmath153 mg the diffusion coefficient @xmath126 decreases by factor 4.6 only , namely @xmath169 @xmath143s@xmath2 .
the related difussion time through this region @xmath170 is about @xmath171 years which is much less than the thermalization and annihilation time @xmath118 years .
one of the possible origins for mev - emission is a disk emission from mass accretion by small mass black holes ( smmbhs ) .
the bh mass can be evaluated using the color disk temperature @xmath173 . using formula ( 5 ) in shrader & titarchuk ( 1999 ) , hereafter sht99
, we obtain that a black hole mass in solar units ( @xmath174 ) @xmath175 ^ 3 } \label{bhmass}\ ] ] where @xmath176 is the dimensionless mass accretion rate in units of @xmath177 , related to the eddington luminosty @xmath178 , @xmath179 is the maximum temperature in the disk in kev , @xmath180 is a disk color factor and @xmath181 is the inner disk radius calculated in schwarzchild units . analyzing quite a few bh sources , sht99 and borozdin et al .
( 1999 ) , hereafter brt99 , established that in the soft state the inner disk radius is very close to the last stable keplerian orbit , namely @xmath182 .
the disk color factor @xmath183 was calculated by brt99 using the known contraints on bh mass and the distance to the source for gro j1655 - 40 .
this value of the disk color factor @xmath183 obtained for gro j1655 - 40 was recently confirmed for other galatic black hole candidates sources .
shrader & titarchuk ( 2003 ) have made use of improved galactic black hole binary dynamical mass determinations to derive , in effect , an empirical calibration of this factor . in the soft state
the mass accretion rate in the disk @xmath176 is in the order of eddington , i.e @xmath184 . with assumption regarding @xmath181 , @xmath185 and @xmath186
we obtain @xmath187 ^ 3}.\ ] ] the smmbhs can be exposed through the accretion , if the mass accretion is at the eddington rate and higher .
the eddington accretion rate by definition is @xmath188 thus for @xmath189 the eddington rate is about 52 g s@xmath2 .
the gravitation attraction of the black hole in the galactic center causes a substantial mass accretion in its proximity .
this accretion flow passing through smmbh which are probably distributed uniformly around the central black hole , provides enough material for mass supply of the smmbh objects .
sgr a@xmath10 , the massive bh in the galactic center was recently found to be surrounded by a cluster of young , massive stars . over the past year , three of these stars named so-2 , so-16 and so-19 by ghez et al ( 2004 ) and schodel et al ( 2003 ) are at @xmath190 schwarzschild radii from the central black hole .
it was also found that these stars can be classified as o stars of the main sequence ( ghez et al .
2003 ; eisenhauer et al .
the winds from these stars have a typical mass loss @xmath191 yr@xmath2 and speed in the range @xmath192 km s@xmath2 ( puls et al .
1996 ; repolust , puls & herrero 2004 ) .
loeb ( 2004 ) showed that so-2 , so-16 , so-19 can supply required mass flow that fuels the emission from sgr a@xmath10 .
now we can check if this mass accretion rate @xmath191 yr@xmath193 g s@xmath2 can provide the eddington mass accretion into smmbh of @xmath194 g s@xmath2 . in order to do
this one should calculate the smmbh luminosity of @xmath195 for a given color temperature @xmath196 kev as follows ( brt99 ) @xmath197
^ 4~{\rm ergs~s}^{-1}\approx 2\times10^{21}{\rm ergs~s}^{-1}\ ] ] where @xmath198 is the disk effective radius in schwarzschild units , and the color factor @xmath199 ( see brt99 )
. then we can evaluate the number of smmbh @xmath200 in the gc region if we assume they are responsible for @xmath35ray emission about @xmath201 ergs s@xmath2 , namely @xmath202 .
one can conclude that the mass flow rate @xmath203 g s@xmath2 needed to fuel smmbh in the galactic center region within @xmath204 cm ( or within @xmath205 schwarzschild radii from the central bh ) is much smaller then the mass flow @xmath206 g s@xmath2 supplied by the young massive o - stars in this central region .
consequently , _ we show how the small mass black holes could be exposed when there is enough material around them_.
revnivtsev et al . ( 2004 ) claimed that the inferred x - ray spectrum and luminosity of the emission illuminating the molecular cloud sgr b2 s , strongly support the idea that the x - ray emission of sgr b2 is compton scattered and reprocessed radiation emitted in the past by the sgr a@xmath10 source .
we have calculated the pair creation using the inferred luminosity and the inferred spectrum that is a power - law photon spectrum of index 1.8 with an exponential cutoff at 100 kev .
in fact , this is a typical spectrum of bh sources at low / hard state which is presumably a result of comptonization of the soft ( disk ) radiation in the hot electrons of temperature 50 kev in compton cloud around central bh .
it has already been shown that the spectral shape in the low hard state is independent of the luminosity ( e.g. laurent & titarchuk 1999 , titarchuk & fiorito 2004 ) .
we can propose that sgr a * is now in a quiescent state , presumably a result of the gravitational energy release in the advection dominated flow ( see e.g narayan & yi 1994 ) .
the luminosity of the x - ray spectrum of the x - ray source igr j17475 - 2822 does not change since granat / art - p and asca era , i.e. during last 10 years .
the clear match of asca / gis , granat / art - p and integral / ibis ( fig .
2 , revnivtsev et al .
2004 ) strongly supports this claim .
thus the proposed incident spectrum emanating from sgr a * in the past ( 300 years ago ) should not be vary either during the period at least of 10 years .
it means that the inferred e+e- rate should be quite stable during the same period .
the constraints on the `` breathtaking variability on a time scale of only several hundred years '' was proposed by sunyaev et al .
they argue that x - ray echo from sgr b2 should be delayed by 300 - 400 years relative to the direct signal from sgr a@xmath10 due to the light travel time from sgr a@xmath10 to sgr b2 .
one can put the fair question regarding such a short recurrence time for the spectral transition , as 300 - 400 years for a black hole of mass @xmath207 solar masses . on
the other hand one can suggest the more likely scenario for irradiating sgr b2 was the interaction between the supernova remnant , we now see as sgr a east and the @xmath208 km s@xmath2 molecular cloud behind sgr a@xmath10 , rather than sgr a@xmath10 itself .
the luminosity , time scale , and spectrum all fit the requirements rather well , so the need for a high x - ray activity in sgr a@xmath10 can be weaker now .
this effect of supenova ( sn ) ejecta and its interactions with molecular clouds proposed by bykov ( 2002 , 2003 ) is recently discussed in the literature ( see e.g. muno et al .
2004b , and park et al .
one can be right that `` the luminosity , time scale , and spectrum all fit the requirements rather well '' .
however park et al . pointed out that although the sn ejecta contribution to the observed neutral fe line emission appears to be plausible , they noted a caveat , that nonthermal radio emission features have not been reported in the northeast regions of sgr a@xmath10 .
this absence of nonthermal radio features is admittedly a difficult problem in terms of the snr interpretation .
it means that the supernova ejecta model needs the nonthermal distribution of electrons in order to explain the hard x - ray energy spectrum of the sgr b2 molecular cloud ( revnivtsev et al . 2004 ) .
we want to emphasize once again that the spectrum seen from sgr b2 is a typical spectrum of the low / hard state seen in many bh sources .
it has already been proven long ago ( see e.g. sunyaev & titarchuk 1980 ) that this hard spectrum is a result of thermal comptonization of soft ( disk ) photons in the hot compton cloud surrounding the central part of a bh .
thus one does not need any fine tuning or nonthermal electron distribution to produce such a spectrum , because in the outskirts of the gc bh there are plenty of molecular clouds , and we should see the effect of the reflection of the bh x - ray spectrum from these configurations .
thus the spectra of molecular clouds are naturally formed as a result of reflection of bh emission from molecular clouds . in connection with the previous points
, one can raise a question related to expected rapid variability in the @xmath209 annihilation line flux for such a compact bh region .
however the observations over several decades seem to not show any variability at all .
it is correct to say that the variability in the @xmath209 annihilation line flux would be quite high if this line was formed in the compact central region .
however , we propose that the pairs are produced in this compact region and only then do they propagate through the galaxy and the e+e- annihilation line is formed through the galaxy .
thus variability of the e+e- annihilation line flux is related to the scale of a few kpc and it is not by chance that observations over several decades seem to not show any variability at all .
we infer that the @xmath24mev gamma ray region , where the pairs are created due to x - ray illuminations by the central source , is quite compact .
its radius is @xmath112 cm .
this region within @xmath210 cm , is consistent with the comptel map ( schonfelder et al .
2000 , strong et al . 1998 ) .
the full sky intensity map , longtitude and latitude profiles for 1 - 3 , 3 - 10 and 10 - 30 mev show the strong peaks within a few degrees near gc . however , the spatial resolution of the comptel map is not better than one degree and thus the compact region of @xmath211 cm can not be resolved in comptel map .
it is worth noting that strong & moskalenko ( 1999 ) argue that there is an unresolved point - source population in the inner galaxy and that this source is an important contribution to the emission around 10 mev .
producing @xmath212 s@xmath2 ( or more , depending on the duty cycle ) has consequences beyond simply calculating the annihilation rate .
the region surrounding sgr a@xmath10 is magnetized , with magnetic field strength @xmath135 of order 10 g inside @xmath204 cm . out to @xmath151 pc
the field is turbulent ( matching the turbulent plasma generated by wind - wind interactions from the wolf - rayet and o / b stars in this region ) .
melia & falcke ( 2001 ) have suggested that the magnetic field strength of this region should be in the range of @xmath152 mg .
in section 4.2 we review the positron propagation in the galactic center region and we find the positrons can pass through there .
consequently the sources of annihilation emission are presumably distributed over the galaxy .
recently liu , petrosian & melia ( 2004 ) , hereafter lpm04 , formulated and solved the propagation and acceleration problem in the turbulent magnetized plasma for a given pair flux @xmath213 , magnetic field strength @xmath135 , plasma interaction time @xmath214 , density @xmath215 , the size of the acceleration site @xmath60 and particle distribution of the background plasma .. they showed that acceleration of electrons by plasma wave turbulence in hot gas in the inner - most part of the black hole reasonably account for sgr a@xmath10 s millimeter and shorter wavelength emission in the quiescent state and for the infrared and x - ray flares . in their model
a ( @xmath216 s@xmath2 , @xmath217 g ) they found the power carried away by accelerated electrons , with @xmath218 , is about @xmath219 ergs s@xmath2 ; which is more than enough to power observed radio emission , whose liminosity of @xmath220 ergs s@xmath2 . for the pair flux of order
@xmath106 s@xmath2 , the `` collateral '' spectrum ( and flux ) arising from these particles is very consistent with sgr a@xmath10 s millimeter and shorter wavelength emission in the quiescent state and for the infrared and x - ray flares ( lpm04 ) .
it is worth noting that the low energy electron flux @xmath221 calculated by lpm04 , is comparable with the injected flux @xmath213 .
it means that a significant fraction of the low enery electrons escapes from this region without any noticeable acceleration .
in fact , the escape time of low energy electrons is more than a factor of 2 shorter than the time scales for acceleration and losses .
the exact value of the electron flux depends on the model assumptions ( see lpm04 ) ; and thus these theoretical uncertainties of the flux estimate ( within a factor of a few ) are unavoidable .
the lpm04 s model also suggested that an outflow of high energy electrons ( liu & melia 2002 ) are distributed over a spatially larger scale than that of the bh inner region .
finally the escaping high energy electron gas ( cooled through the radiation ) could form the annihilation line emission in the large scale of the galaxy .
the lpm04 s injected flux @xmath222 s@xmath2 , is related to the mass accretion rate of order @xmath223 of the eddington value ( @xmath177 ) .
if the pairs are accelerated to @xmath218 then their energy would be enough to power the observed radio emission whose luminosity is about @xmath220 ergs s@xmath2 .
it is worth noting that within our model , the same flux of pairs ( @xmath222 s@xmath2 ) is generated in the smmbh site as a result of x - ray illumination emitted by surrounding material .
if at the present time the x - ray flux is of order @xmath224 erg s@xmath2 , then the pair flux would be @xmath225 s@xmath2 .
dark matter is a long standing puzzle starting from le verrier with his discovery of neptune in 1846 as a perturbation of uranus ( le verrier 1846 ) .
our galaxy is surrounded by an extended halo of unseen material .
so far undetected , that dark matter induces a flat rotation curve in the galactic plane ( van albada & sancisi 1986 ) .
its presence has been noticed since many decades ( see oort 1932 ) . however , its nature is still unknown .
the dark matter problem consists simply , in the existence of invisible mass showing its presence by gravitational effects .
now , it is widely accepted that dark matter exists .
in fact , there is evidence for dark matter on scales from galaxy to the cluster of galaxies and to the whole universe itself .
zwicky ( 1937 ) measured the dispersion velocity in the coma cluster and found that the dynamical mass was hundred times more than the luminosity mass .
moreover , the dark matter seems essential : the growth of structure in the universe by hierachical merging of the dark matter halo is a master piece of moderm cosmology .
recently , wmap has established the presence of non baryonic dark matter with the density reported in terms of critical density of order @xmath226 .
this is now the standard value of the modern cosmology for dark matter density . on the galactic scale , one can interprete the rotation curve measurements by the presence of the dark matter halo and then determine the mass density profile of this distribution .
the average density is @xmath227 .
the core of this halo seems to be extremely packed . in the galactic center , numerical simulations of structure formation in the non - linear regime have shown the presence of a singular power law cusp @xmath228 , where @xmath229 and @xmath230 is the dark matter density at radius @xmath231 pc ( gnedin & primack 2004 ) .
the central density @xmath232 is not uniquely determined and its value depends on models and it varies from 60 to 600 @xmath233 pc@xmath146 . on the other hand up to now there is no identification for it despite the large number of candidates proposed from masses in the range between @xmath223 ev to @xmath234 . while there is no a priori reason to expect that massive compact halo objects ( machos ) in the mass range @xmath235 comprise a significant fraction of the density of the universe ,
neither is there any definitive argument ruling them out .
nemiroff & gould ( 1995 ) and marani et al .
( 1999 ) discuss the possibility of identification of smmbhs by gravitational lensing .
the basic idea is to study the deflection of light by measuring the amplification due to gravitational bending of compact objects . for black hole mass in the range @xmath236 to @xmath237 ,
the expected image separation is of order of @xmath238 arcseconds and the average time separation between the images is order of @xmath239 s. they applied this delay as an interference process using gamma ray bursters . on the other hand ,
the absolute value of the light deflection by small black holes is extremely small @xmath240 , where r is the characteristic distance in the galaxy and therefore not observable .
it is easy to show that the optical depth for the photons scattering at small angles by smmbh is small , ( @xmath241 ) and consequently the light multiple scattering by the small black holes effect is also ruled out .
the supersymmetry theory as an extension of the standard model , predicts that one stable particle should be around today as a cosmological relic . among these new particles , the neutralinos ,
could be the dark matter .
if the dark matter halo of our galaxy is made up of such particles , they will annihilate each other into fermion - antifermions pairs , and this will produce stable particles which could be detected as anti - protons , electrons and photons . another way is to look for direct detection via laboratory experiments .
such new particles are scrutinized by direct and indirect searches since more than ten years .
our galaxy is indeed a subject of intense activities because it could be a direct search for dark matter in studying astrophysical signatures . among all the signatures , our galaxy presents three strong signatures in x - ray , @xmath35ray backgounds and a typical @xmath242 annihilation source in the galactic center . among the other possibilities ,
the micro - lensing searches have been developed and used to probe the dark matter halo of our galaxy ( see above ) .
the possibility that the halo could be made of primordial black holes is not ruled out by observations .
recently , seto and cooray ( 2004 ) discussed also the future possibility of searching dark matter candidates using the space - based gravitationnal wave detectors .
they used this idea to test the possibility that the halo could be made of primordial black holes . according to hawking predictions such primordial black holes will not suffer evaporation
if their mass is greater than @xmath243 g. while the consequences of evaporating black holes has been widely analyzed in terms of particles production , the case for stable smmbhs s has not been considered so far . in literature , the problem is analyzed generically and the mass range used are often very large from @xmath5 g to @xmath206 g.
in this paper , we show that there is a particular black hole mass which is very interesting for the dark matter problem .
this mass of order of @xmath5 g , corresponds to a typical schwarzschild radius proton size order .
we explore this possibility in the light of existing galactic diffuse x and @xmath15 backgrounds and with the strong source of annihilation lines at 511 kev in the galactic center . our approach is to find a natural explanation of such astrophysical backgrounds within an astrophysical dark matter scenario .
we also offer a model for the origin of the annihilation line appearance in the our galaxy which has been detected in several balloon and space experiments .
we combine x - rays and @xmath15-observations of the galactic center region to show that pair production can be a result of interactions of x - rays and gamma photons emitted in this region .
pairs production strongly depends on the compactness of the x - rays and @xmath15 emission region which is determined by x - rays and @xmath15 luminosity and size of the region . for velocity of the interstellar medium of much less than @xmath244 cm s@xmath2 and the value of density column a few times @xmath245 , the optical depth for 511 kev line is more than one .
we found for the observed x - rays and @xmath15 luminosities and for the observed annihilation lines flux , the size of the x - ray and @xmath15 emission area is @xmath246 cm .
this size is consistent with the size of the area where the ob stars supply sufficient amount of fuel required for the x - rays and @xmath15 emissions from the galactic center region .
if we compute the total black hole mass inside this region of radius r=@xmath247 pc , we obtain that this mass is about two solar masses .
we can then naturally ask if smmbhs can be a candidate for dark matter as already proposed in literature ( e.g. nemiroff & gould 1995 , marani et al .
it is interesting to compare this number to the dark matter mass computed using the singular power law cusp @xmath228 often used in the dark matter model for the galactic center ( see moore et al .
1998 and gnedin & primack 2004 ) .
we found that with a set of parameters ( @xmath248 pc@xmath146 , @xmath249 pc , @xmath250 ) , we can obtain a mass of @xmath251 inside the same radius r , which is compatible to our number within a factor of a few .
koyama ( 1996 ) and muno et al . (
2004a ) , revnivtsev et al .
( 2004 ) noted that the galactic center exhibited intermittent activities with time - averaged energy generation rate @xmath41 ergs s@xmath2 comparable to seyfert nuclei .
muno et al .
( 2004a ) are left to conclude either that there is a significant shortcoming in understanding of mechanisms that heat the interstellar medium in galactic center region or that of a population of faint ( @xmath252 ergs s@xmath2 ) hard x - ray sources .
muno et al .
( 2004b ) using a deep chandra observations study to detail the origin of the point like sources near the galactic center .
possibly a single population of sources may dominate in this cluster .
there are certainly many classes of objects present in this set of faint sources .
we may suggest that some part of x - ray and @xmath35radiation of the galactic center , can originate in accreting flows into smmbhs ( that are really faint sources with their bolometric luminosity @xmath253 ergs s@xmath2 ) in which outer parts of flow emit x - rays and the innermost part of the flow emit @xmath35rays .
finally we can conclude , that the only way to identify small mass black holes ( smmhhs ) is by accretion mechanism in presence of a surrounding dense material . in the absence of this matter
these smmbhs are invisible and can be treated as dark matter , or invisible matter , i.e. without any observational appearance in the radiation .
we appreciate productive discussions with gerald shore and robert duffin . for this work , l.t .
was partly supported by french ministre de leducation nationale et de la recherche with grant sshn2004 - 413077d .
we thank the referee for his / her profound questions and we acknowledge the referee s contribution in the discussion section of this paper .
aharonian , f. , et al .
2004 , a&a , 495 , l13 armstrong , j.w . ,
rickett , b.j . & spangler , s.r .
1995 , , 443 , 209 baganoff , f.k . , 2003 , , 591 , 891 boehm , c , fayet , p. , & silk , j. 2004 , phys .
rev , d69 , 101302 bykov , a.m. 2003 , a&a , 410 , l5 bykov , a.m. 2002 , a&a , 390 , 327 borozdin , k. , revnivtsev , m. , trudolyubov , s. , shrader , c , & titarchuk , l. 1999 , apj , 517 , 367 ( brt99 ) chakrabarti s.k .
& titarchuk , l. g. 1995 , apj , 455 , 623 chapuis , c.g.l .
1991 , gamma - ray line astrophysics , ed .
p. durouchoux & n. prantos ( new york : aip ) , 54 cheng , l.x . , 1997 , , 481 , l43 chez , a.m. , et al . 2004 , , 601 , l159 chez , a.m. , et al .
2003 , , 586 , l127 churazov , e. , sunyaev , r. , sazonov , s. & varshalovich , d. 2005 , mnras , 357 , 1377 dermer , c.d . & skibo , j.g .
1997 , , 487 , 57 dirac , p.a.m . 1930 , proc .
camb , phil .
26 , 361 eisenhauer , f. , et al .
2003 , , 597 , l121 gehrels , n. , et al .
1991 , , 375 , l13 gnedin , o.y . , & primack , j.r . 2004 , _ phys .
_ , 61 , 302 guessoum , n. , jean , p & gillard , w. 2005 , a&a , 436,171 heitler , w. 1954 , the quantum theory of radiation , oxford university press , oxford hunter , s.d . , bertsch , j.r . ,
catelli , j.r .
1997 , in aip conf proc .
410 , 4th compton symposium , new york , 1193 jean , p. 2005 , a&a in press ( astro - ph/0509298 ) kawai , n. , et al .
1988 , , 330 , 130 koyama , k. , et al .
1996 , pasj , 48 , 249 koyama , k. , et al .
1989 , nature , 339 , 603 laurent , p. & titarchuk , l. 1999 , , 511 , 289 le verrier , 1846 , _ astronomische nachrichten _ , 25 , 65 leventhal , m. et al .
1993 , , 405 , l25 leventhal , m. , maccallum , c.j .
, & stang , p.d .
1978 , , 225 , l11 liu , s. , petrosian , v. , & melia , f. 2004 , , 611 , l101 liu , s. , & melia , f. 2002 , , 573 , l23 loeb , a. 2004 , , 350 , 725 longair , m.s .
1994 , high energy astrophysics , cambridge : university press marani , g.f . ,
nemiroff , r.j . , norris , k. & bonnell , j.t .
1999 , , 512 , l13 markevitch , m. , sunyaev , r.a . , & pavlinsky , m. 1993 , nature 364 , 40 melia , f. & falcke , h. 2001 , araa , 39 , 309 melrose , d.b .
1980 , plasma astrophysics .
nonthermal processes in duffuse magnetized plasmas , new york : gordon and breach moore , b. , et al .
1998 , , 499 , l5 morris , m. , & serabyn , e. 1996 , ara&a , 34 , 645 muno , m. , et al . , 2005 , , 622 , l113 muno , m. , baganoff , f.k , bautz , m. w. et al . 2004a , , 613 , 326 muno , m. , arabadjis , f. k , baganoff , f.k , et
al . 2004b , , 613 , 1179 murphy , r.j . ,
dermer , c.d . , & ramaty , r. 1987 , apjs , 63 , 721 oort j. h. , 1932 , _ bull .
netherlands _ 6 , 349 narayan , r. , et al . 1998 , , 492 , 554 narayan , r. & yi , i. 1994 , , 428 , l13 nemiroff , r.j . & gould , a. 1995 , , 452 , l111 niel , m. et al . 1990 , , 356 , l21 park , s. et al . 2004 , , 603 , 548 parker , e. n. 1979 , cosmical magnetic fields , clarendon press , oxford pavlinsky , m. , grebenev , s.a . , &
sunyaev , r.a . , &
1993 , , 425 , 110 preddehl , p. , & trumper , j .
1994 , a&a , 290 , l29 puls et al .
1996 , a&a , 305 , 171 purcell , w.r .
, et al . 1997 , , 491 , 725 purcell , w.r . , et al .
1993 , , 413 , l85 ramaty , r. , skibo , j.g . &
lingenfelter 1994 , apjs , 92 , 393 repolust , puls & herrero 2004 , a&a , 415 , 349 revnivtsev , m.g .
2004 , a&a , 425 , l49 schodel , r. , et al .
2003 , , 596 , 1015 schonfelder , v. , et al .
2000 , a&ass , 143 , 145 seto , n. & a. cooray , a. 2004 , _ phys .
_ * d * , 70 , 63512 share , g.h .
, et al . 1990 , , 358 , l45 shrader , c. , & titarchuk , l.g . 2003 , apj , 598 , 168 shrader , c. , & titarchuk , l.g . 1999 ,
apj , 521 , l121 ( sht99 ) skinner , g.k . , et al .
1987 , nature , 330 , 544 smith , d.m . , heindl , w.a . & swank , j.h .
2002 , , 569 , 362 strong , a.m. , bloemen , h. , diehl , r. , hermsen , w. , & schonfelder , v. 1998 , proc .
integral workshop ( eds .
g. palumbo , a.bazano , & c. winkler ) astrophys .
39 , 209 strong , a.m. & moskalenko , i.v .
1999 , proc .
5th compton symposium ( eds . m. l. mcconnell and j. m. ryan ) aip conference proceedings , 510 , 291 sunyaev , r.a . ,
markevitch , m. , & pavlinsky , m. 1993 , , 407 , 606 sunyaev , r.a . & titarchuk , l.g .
1980 , a&a , 86 , 121 takeshima , t. et al .
1991 , in proc . of 28th yamada conf . ,
frontiers of x - ray astronomy , ed .
y. tanaka , k. koyama ( universal academy press , tokyo ) , 421 teegarden , b.j .
, et al . 1996 , , 463 , l75 tanaka , y. , inoue , h. , & holt , s.s .
1995 , pasj , 46 , l37 titarchuk , l. & fiorito , r. 2004 , , 612 , 988 van albada t. s. , & sancisi r. , 1986 , _ phil .
a _ , 320 , 447 watson , m.g .
, willingate , r. , grindlay , j.e . , hertz , p. 1981
, , 250 , 142 wood , k.s . , et al . 2001 , , 563 , 246 yamauchi , s. et al . 1990 , , 365 , 532 zwicky , f. 1937 , , 86 217 | compton gamma ray observatory , osse , smm , tgrs , balloon and recent integral data reveal a feature of the 0.511 mev annihilation radiation of the galactic center with a flux of approximately @xmath0 mev photons @xmath1s@xmath2 .
we argue that @xmath3 pairs can be generated when the x - ray radiation photons and @xmath4 mev photons interact with each other in the compact region in the proximity of the galactic center black hole .
in fact , disks formed near black holes of @xmath5 g mass should emit the @xmath6 mev temperature blackbody radiation . if positron ( @xmath7 ) sources are producing about @xmath8s@xmath2 near the galactic center they would annihilate on the way out and result in @xmath9 mev emission .
we suggest that the annihilation radiation can be an observational consequence of the interaction of the accretion disk radiation of the small mass black holes ( smmbhs ) with x - ray radiation in the galactic center .
this is probably the only way to identify and observe these smmbhs . |
one of the most promising configurations for thermoacoustic devices is the traveling wave configuration which generally consists of a closed - loop tube.@xcite this design however , has one disadvantage : the possibility of a time - averaged mass flow , or gedeon streaming , to occur due to the looped geometry@xcite .
this streaming leads to unwanted convective heat transport and reduces the efficiency of closed - loop thermoacoustic devices .
a commonly used solution is the application of a jet pump.@xcite this is a tapered tube section which , due to the asymmetry in the hydrodynamic end effects , establishes a time - averaged pressure drop . by balancing the time - averaged pressure drop across the jet pump with that which exists across the regenerator of a thermoacoustic device
, the gedeon streaming can be canceled.@xcite fig .
[ fig : jetpumpgeom ] shows a schematic of a typical conical jet pump geometry with its corresponding dimensions .
the two openings both have a different radius : @xmath0 for the big opening ( left side ) and @xmath1 for the small opening .
together with the jet pump length , @xmath2 , the jet pump taper angle @xmath3 is defined .
furthermore , at the small opening a curvature @xmath4 is applied to increase the time - averaged pressure drop compared to a sharp contraction . to estimate the performance of a jet pump ,
a quasi steady model has been proposed by backhaus and swift.@xcite this model is based on minor losses in steady flow and assumes that the oscillatory flow can be approximated as two steady flows.@xcite given the pressure drop generated by a pipe transition in steady flow,@xcite @xmath5 and assuming a pure sinusoidal velocity in the jet pump , the time - averaged pressure drop across the jet pump can be estimated as @xmath6 , \label{eq : backhaus}\end{aligned}\ ] ] where @xmath7 is the velocity amplitude at the small exit of the jet pump .
the subscripts `` @xmath8 '' and `` @xmath9 '' indicate the small and big opening of the jet pump , respectively .
@xmath10 is the minor loss coefficient for abrupt expansion and @xmath11 is the minor loss coefficient for contraction , both are well documented for steady flows.@xcite with the jet pump being an additional flow resistance , acoustic power will be dissipated due to viscous dissipation and vortex formation .
using the same quasi steady approach ,
backhaus and swift derived an equation to estimate the amount of acoustic power dissipation associated with the jet pump as@xcite @xmath12 .
\label{eq : backhaus_de}\end{aligned}\ ] ] an optimal jet pump should establish the required amount of time - averaged pressure drop to cancel any gedeon streaming with minimal acoustic power dissipation .
this requires maximizing the difference in minor losses due to contraction and expansion while at the same time minimizing the sum of the minor loss coefficients .
previous studies have shown however , that the accuracy of this quasi
steady approximation is limited and that there are other factors influencing the performance of a jet pump.@xcite petculescu and wilen experimentally studied the influence of taper angle and curvature on minor loss coefficients.@xcite good agreement was obtained between steady flow and oscillatory experiments for taper angles up to .
however , the minor loss coefficients determined were found to be strongly dependent on the taper angle used .
a qualitative comparison between their findings and the current results will be provided in section [ sec : varalpha_varrs ] .
smith and swift investigated a single diameter transition , corresponding to one end of a jet pump.@xcite the measured pressure drop and acoustic power dissipation was found to be dependent on the dimensionless stroke length , the dimensionless curvature and the acoustic reynolds number . recently , tang et al . investigated the performance of jet pumps numerically,@xcite but assumed a priori the quasi
steady approximation to be valid by modeling the flow as two separate steady flows .
the negative effect of flow separation on the jet pump performance was identified which is in line with the current work .
although some jet pump measurements are available in literature , a systematic parameter study that directly relates variations in wave amplitude and geometry to a jet pump s performance has , to the authors knowledge , not yet been addressed .
a first step towards the investigation of a jet pump s performance in oscillatory flows has been made in previous work by scaling the jet pump geometry using two different keulegan carpenter numbers and correlating this to the jet pump performance.@xcite four different flow regimes were distinguished as a function of the wave amplitude .
[ fig : flowregimes ] shows the observed vorticity fields in the vicinity of a jet pump at @xmath13 when @xmath14 . in order of increasing wave amplitude
these flow regimes can be described as follows . at low wave amplitudes ,
an oscillatory vortex pair exists on both sides of the jet pump but no vortex shedding is observed ( fig .
[ fig : flowregime1 ] ) . when the amplitude is increased , one - sided vortex propagation at the location of the small jet pump opening starts ( fig .
[ fig : flowregime2 ] ) . in this flow regime ,
vortex rings are shed from the jet pump waist to the right while on the big side of the jet pump an oscillatory vortex pair is still visible .
vortex shedding from the big side of the jet pump begins to occur at increased wave amplitudes ( fig .
[ fig : flowregime3 ] ) . here
vortex rings are shed from both jet pump openings directed outwards , but no interaction between the two sides is visible .
ultimately , a further increase in wave amplitude leads to interaction between the two jet pump openings ( fig .
[ fig : flowregime4 ] ) .
vortices being shed from the small jet pump opening now propagate in alternating directions and flow separation inside the jet pump occurs . albeit the described flow regimes were distinguished using the proposed scaling , the data set was too limited to determine whether the observed flow separation ( fig . [ fig : flowregime4 ] ) was geometrically initiated by an increase in the taper angle or by a decrease in the jet pump length . in this paper ,
a parameter study is performed using a computational fluid dynamics ( cfd ) model to further identify the relation between the four different flow regimes and the jet pump geometry .
the influence of various geometric parameters including the jet pump taper angle , length , curvature and waist diameter on the jet pump performance is investigated .
based on the presented results , the existing scaling parameters are further extended .
furthermore , a comparison with the quasi steady model is provided to determine under what conditions the approximation is applicable . based on this parameter study ,
design guidelines for future jet pump design are proposed . after a description of the used cfd model in section [ sec : modeling ] ,
the various investigated jet pump geometries are introduced in section [ sec : geom_varalpha_varrs ] .
the resulting flow regimes will be distinguished in section [ sec : flowregimes ] and subsequently linked to the jet pump performance in section [ sec : varalpha_varrs ] .
finally , the influence of the jet pump curvature is investigated in section [ sec : var_rc ] before drawing final conclusions on the design guidelines in section [ sec : conclusions ] .
a two - dimensional axisymmetric cfd model is developed using the commercial software package ansys cfx version 14.5,@xcite identical to the model used in previous work.@xcite in the following , the numerical model with the corresponding boundary conditions will be repeated briefly and the methods to derive the jet pump performance will be explained .
the different jet pump geometries are confined in an outer tube with radius @xmath15 .
the length of the outer tube on either side of the jet pump is @xmath16 when @xmath17 .
this length is scaled relative to the wavelength to avoid the jet pump being placed at a velocity node for simulations performed at other frequencies . in all cases , air at a mean pressure of @xmath18 and a mean temperature of @xmath19 is used as the working fluid .
the unsteady , fully compressible navier - stokes equations are discretized using a high resolution advection scheme in space and a second order backward euler scheme in time.@xcite the system of equations is closed using the ideal gas law as the equation of state and the energy transport is calculated using the total energy equation , including viscous work terms .
based on the critical reynolds number defined in section [ sec : results ] , all presented results fall within the laminar regime so no additional turbulence modeling is applied .
the time - step @xmath20 is chosen such that each wave period is discretized using 1000 time - steps .
a total of ten wave periods are simulated to achieve a steady periodic solution and the last five wave periods are used for further analysis .
the acoustic wave is generated on the left boundary of the domain using a sinusoidal velocity boundary condition with a specified frequency and velocity amplitude . to control the wave propagation over the right boundary of the computational domain , a dedicated time - domain impedance boundary condition is developed and implemented in ansys cfx .
the approach is based on the work of polifke et al . and allows the application of a complex reflection coefficient at the boundary.@xcite a detailed validation and explanation of the implementation can be found in the work of van der poel , which was carried out as part of the current research.@xcite in all cases described here , a reflection coefficient of @xmath21 is specified to simulate a traveling wave on the right side of the jet pump . the combination of the velocity boundary condition and the time - domain impedance boundary condition results in a time - averaged volume flow that is on average less than of the acoustic volume flow rate . on the radial boundary of the outer tube ( at @xmath22 ) , a slip adiabatic wall boundary condition is imposed as the pipe losses in this part of the domain are currently not of interest . to correctly simulate the minor losses in the jet pump , a no - slip adiabatic wall boundary condition is used at the walls of the jet pump .
the choice for a two - dimensional axisymmetric model to simulate flow separation might need some additional explanation as in planar diffusers the flow separation can be asymmetric.@xcite a clear distinction should be made between flow separation in planar and conical diffusers .
the steady flow separation in conical diffusers has been investigated previously and no visible asymmetry of the flow was reported under laminar conditions.@xcite furthermore , the oscillatory nature of the flow prevents the flow from developing asymmetric instabilities , even in planar diffuser geometries.@xcite these observations motivate the applicability of a two - dimensional axisymmetric model for the current situation of pure oscillatory flow through a conical geometry .
the used spatial discretization is validated in previous work and will be only briefly described here.@xcite the computational mesh consists of an unstructured part within from the jet pump and a structured mesh in the rest of the domain . in both parts
quadrilateral elements are used . in the jet pump region , a maximum element size of is used which is refined up to near the jet pump waist . moreover , to be able to accurately resolve the flow separation , a refinement in the viscous boundary layer is applied such that a minimum of 10 elements are located within one viscous penetration depth distance @xmath23 from the jet pump wall . for each simulated wave frequency
, the mesh distribution is adjusted according to the viscous penetration depth , @xmath24 . in the quadrilateral part of the mesh , a fixed radial mesh resolution of
is used whereas the mesh size in the axial direction grows from at a distance of from the jet pump up to at the extremities of the domain . in additional to the mesh validation performed in previous work,@xcite the accuracy of the employed computational mesh has been investigated specifically for the prediction of flow separation . the jet pump geometry with a taper angle of @xmath25 ( no . 4 in table [ tab : geom_varalpha ] ) has been used at a frequency of and at a wave amplitude where flow separation is expected ( @xmath26 , eq . [ eq : x1dsalpha ] ) .
the mesh size in the jet pump region is in two steps refined from to . at the same time , the mesh size in the jet pump waist region is refined from to .
additionally , the effect of a mesh refinement in the viscous boundary layer is investigated by increasing the number of elements from 10 to 20 .
the difference in the dimensionless pressure drop ( eq .
[ eq : dp2_star ] ) and dimensionless acoustic power dissipation ( eq . [ eq : de2_star ] ) between the investigated meshes is less than 0.04 and 0.05 , respectively .
furthermore , both the time and location where the flow first separates do not vary more than @xmath27 and @xmath28 between the different mesh resolutions . in order to determine the jet pump performance from the transient cfd solution ,
some additional analysis steps are performed .
the first order amplitudes of all physical quantities are calculated by a discrete fourier transformation for the specified wave frequency using the solution from the last five simulated wave periods . to obtain the time - averaged quantities ,
the cfd solution is averaged over the last five wave periods in order to eliminate the first order contribution .
the streaming velocity is calculated using a density - weighted time - average : @xmath29 .
the velocity amplitude in the jet pump waist , @xmath7 , is calculated using an area - weighted average over the cross - section .
subsequently , the local acoustic displacement amplitude in the jet pump waist @xmath30 can be calculated under the assumption of a sinusoidal jet pump velocity : @xmath31 to scale the displacement amplitude to the jet pump geometry , a keulegan
carpenter number is defined based on the jet pump waist diameter : @xmath32 where @xmath33 is the jet pump waist diameter .
this @xmath34 is also referred to as a `` dimensionless stroke length '' or `` dimensionless displacement amplitude '' and can be rewritten to @xmath35 where @xmath36 is an acoustic reynolds number based on diameter and @xmath37 the stokes number @xcite .
the described computational model is used to investigate a large variety of conical jet pump geometries and wave amplitudes . by varying either the jet pump length , the size of one of the two openings or the local curvature at the jet pump waist , 19 different geometries are considered .
the different investigated geometries are all variations based on one reference geometry ( denoted by `` _ _ ref _ _ '' in all tables ) .
the reference geometry is a jet pump with a taper angle , all other reference dimensions are denoted in table [ tab : geom_ref ] .
.dimensions of the reference jet pump geometry .
[ cols= " < , < " , ] + when comparing the measured dimensionless pressure drop using these jet pump samples , it is found that an additional contribution to the keulegan
carpenter number is required to properly scale the results from different curvatures .
[ fig : dp2_kcd[varrc ] ] shows the dimensionless pressure drop as a function of @xmath34 for the geometries with varied curvature .
the investigated curvatures are indicated by different symbols depicted in the corresponding legend .
the crosses ( @xmath38 ) represent the reference geometry where @xmath39 .
the shape of the dimensionless pressure drop curve is found to be inversely proportional to the dimensionless curvature , especially for low values of @xmath40 . for large curvatures , the effect on
the dimensionless pressure drop becomes less apparent .
holman et al . studied the effect of edge curvature on the onset of a synthetic jet.@xcite a formation criterion of the form @xmath41 was derived based on the vorticity flux through the jet .
the applicability of this criterion to jet pumps with variable curvature is investigated in the following .
it is observed that especially the initial increase in @xmath42 , corresponding to the onset of vortex shedding , can be properly scaled using @xmath43 with @xmath44 .
given that the onset of vortex shedding for a jet pump geometry with @xmath39 is found at @xmath45 , the additional scaling parameter @xmath46 can be estimated .
[ fig : dp2_kcd - alt1[varrc ] ] shows the dimensionless pressure drop as a function of the adjusted scaling parameter .
while the initial increase in @xmath42 for the various curvatures all collapse to a single curve , the decay in @xmath42 due to the occurrence of flow separation can not properly be accounted for by any scaling of the form @xmath47 . to overcome this , an alternative scaling strategy for @xmath40
is proposed of the form @xmath48 . using a least
squares approach , the fitting parameter @xmath49 is determined .
note that @xmath50 represents the fraction of the dimensionless curvature with respect to a `` smooth '' contraction .
[ fig : dp2_kcd - alt2[varrc ] ] shows again the dimensionless pressure drop data , but now plotted against @xmath51 .
the case with no curvature ( @xmath52 ) does not appear in this figure as it will lead to infinite values on the horizontal axis .
both the maxima of the dimensionless pressure drop as well as the decay at higher amplitudes are well aligned using this alternative scaling approach .
what is remarkable is that the maximum measured dimensionless pressure drop decreases with increasing curvature , which is in contrast with the steady flow theory where the minor loss coefficient for contraction is expected to fall until @xmath53 .
the maximum achieved dimensionless pressure drop ranges from @xmath54 for a sharp contraction ( @xmath52 ) to @xmath55 for @xmath56 in an almost linear manner .
furthermore , the effect of curvature on the dimensionless pressure drop is not only reversed , but the variation is larger than what is predicted by the quasi steady model ( dashed lines ) .
an explanation of this behavior is the effect of curvature on the expansion phase , which is typically neglected in a quasi
steady approximation of the jet pump performance .
not only will the minor loss coefficient for contraction at the jet pump waist decrease , but so does the minor loss coefficient for expansion .
this was first experimentally determined by smith under steady flow conditions.@xcite a decrease in the expansion minor loss coefficient at the jet pump waist will lead to a decrease in the dimensionless pressure drop according to eq .
[ eq : backhaus ] .
additionally , this is reflected in the measured dimensionless acoustic power dissipation , which is shown in fig .
[ fig : de2[varrc ] ] .
the minimum achieved dimensionless acoustic power dissipation decreases more than six times between @xmath52 ( not shown ) and @xmath56 .
a similar effect has been reported when rounding the orifice of an synthetic jet.@xcite combining the effect the curvature has on both the pressure drop and acoustic power dissipation using the jet pump effectiveness , shows that a higher dimensionless curvature leads to a better jet pump performance .
this is depicted in fig .
[ fig : eta[varrc ] ] where the jet pump effectiveness for five investigated curvatures is shown .
similar to what was presented in section [ sec : flowregimes ] , the observed jet pump behavior can be explained by studying the flow field and distinguishing between the four different flow regimes .
[ fig : varrc_flowregimes ] shows the distribution of the flow regimes in the ( @xmath40 , @xmath34 ) space .
the four flow regimes are represented by different symbols in accordance to the symbols used in fig .
[ fig : flowregimes_varalpha+varrs ] . a clear influence of the dimensionless curvature is visible on the value of @xmath34 where vortex shedding is first observed ( transition from @xmath57 to @xmath58 ) as well as on the value where flow separation is initiated ( transition from @xmath59 to @xmath60 ) .
the boundary between one - sided and two - sided vortex shedding ( transition from @xmath58 to @xmath59 ) is not determined by the jet pump waist curvature at all . as the local curvature at the big opening is not changed
, no dependency on @xmath40 is to be expected . by substituting the original defined flow regime bounds from section [ sec : flowregimes ] in the adjusted scaling parameter @xmath51 , the bounds between the flow regimes ( indicated by symbols as used in fig .
[ fig : flowregimes_varalpha+varrs ] and fig .
[ fig : varrc_flowregimes ] ) can be represented in the ( @xmath40 , @xmath34 ) space : @xmath61 where the first expression corresponds to @xmath45 , the second expression corresponds to @xmath62 and the last expression corresponds to @xmath63 for the cases where @xmath39 .
the bounds are shown in fig .
[ fig : varrc_flowregimes ] by the dashed lines .
the bottom curve represents the initiation of vortex shedding ( eq . [ eq : flowbound1 ] ) and the upper curve the transition to flow separation ( eq . [ eq : flowbound4 ] ) , both a function of @xmath40 .
the horizontal dashed line in fig . [ fig : varrc_flowregimes ] indicates the boundary between one - sided and two - sided vortex shedding .
based on the data available the determined bounds do separate the flow regimes depicted in fig .
[ fig : varrc_flowregimes ] well , although the defined onset of vortex shedding ( @xmath57 @xmath64 @xmath58 ) is subject to discussion .
given the fact that the initial increase in @xmath42 caused by the onset of vortex shedding was well described by adopting a scaling of the form proposed by holman et al.,@xcite a flow regime bound based on the latter could be more appropriate : @xmath65 this is represented in fig .
[ fig : varrc_flowregimes ] by a gray solid line and does indeed separate the two flow regimes .
it should be noted that the data available is not sufficient to determine whether eq .
[ eq : flowbound1 ] or eq .
[ eq : flowbound1a ] is more appropriate .
nevertheless , eq . [ eq : flowbound1a ] predicts vortex shedding for @xmath52 at @xmath66 which is also the formation criterion observed experimentally for a sharp edged synthetic jet.@xcite this exactly describes the difference in the transitional keulegan
carpenter number , observed in section [ sec : flowregimes ] , between vortex shedding from the jet pump waist ( @xmath67 ) and vortex shedding from the big jet pump opening ( @xmath68 ) .
a lower dimensionless curvature leads to a lower value of @xmath34 where vortex shedding is initiated .
consequently , vortex shedding at the sharp edged big opening will take place at a lower value of the _ local _ keulegan carpenter number than the vortex shedding from the jet pump waist . , @xmath34 ) space for jet pump geometries with varied curvature ( table [ tab : geom_varrc ] ) and a taper angle .
symbols are in accordance with fig .
[ fig : flowregimes_varalpha+varrs ] with the open circles showing the position where @xmath42 is maximum for each investigated curvature .
the dashed lines represent the bounds of the different flow regimes determined according to eq .
[ eq : flowbound1][eq : flowbound4 ] and the gray solid line represents the onset of vortex shedding according to eq .
[ eq : flowbound1a ] ( color online ) . ]
a cfd based parameter study is performed to investigate the influence of various geometric parameters on the performance of jet pumps for thermoacoustic applications .
a total of 19 different jet pump geometries are used and for each geometry a number of wave amplitudes are simulated , resulting in a total of 197 simulations . in correspondence with previous work ,
four flow regimes are distinguished and separated in a fixed variable space.@xcite this space spans the jet pump taper angle @xmath3 , the dimensionless curvature @xmath40 and the keulegan carpenter number @xmath34 . at a certain value of @xmath34 , single - sided vortex propagation from the jet pump waist
is initiated and the flow field becomes asymmetric between the left and right side of the jet pump .
this onset is found at @xmath45 for a dimensionless curvature of @xmath39 .
two different expressions are proposed to account for the effect of @xmath40 , but more data is required to decisively choose one of the two . at some point ,
flow separation inside the jet pump can be distinguished and vortices are shed from the jet pump waist in alternating directions leading to a flow field that is more symmetric .
this transition is found to be strongly dependent on the jet pump taper angle and occurs when @xmath69 .
the amount of asymmetry in the flow field has a direct consequence on the measured time - averaged pressure drop . when no vortex shedding occurs , no dimensionless time - averaged pressure drop is observed .
furthermore , as soon as the flow separation is observed , the dimensionless pressure drop has already decayed significantly compared to the maximum value .
for all the performed simulations , the maximum jet pump effectiveness is observed when either single - sided or two - sided vortex propagation occurs .
consequently , the practical operation area of jet pumps with @xmath70 and @xmath71 is bound between the onset of vortex shedding from the jet pump waist and the occurrence of flow separation .
this can be used as a guideline for future jet pump design .
the design considerations presented in this paper , are supposed to provide better insight into the validity of the quasi
steady approximation that is widely used for the design of jet pumps for thermoacoustic applications .
the quasi
steady approximation is in most cases an ideal representation of the jet pump performance and is valid only in a small operation area . outside this region
, the actual jet pump performance is expected to be significantly lower and a correction using the presented results is advised .
although the current parameter study represents a wide range of jet pump geometries , it is restricted to jet pumps with a single , linear tapered hole . as such , this work can serve as a basis for further jet pump geometry optimization in terms of effectiveness , compactness and robustness .
extensions to multiple holes or different shaped jet pump walls might affect the jet pump performance and the occurrence of flow separation .
furthermore , the influence of turbulence on the flow separation requires further attention . if the flow separation regime could be shifted to higher keulegan
carpenter numbers without a decrease in effectiveness , the operation area of a jet pump can be extended , greatly enhancing its robustness .
i. idelchik , `` resistance to flow through orifices with sudden change in velocity and flow area . '' in a. ginevskiy and a. kolesnikov , editors , _ handbook of hydraulic resistance _
, chapter 4 , 223275 , ( begell house , new york , 2007 ) , 4th edition .
j. p. oosterhuis , s. bhler , d. wilcox and t. h. van der meer , `` computational fluid dynamics analysis of the oscillatory flow in a jet pump : the influence of taper angle . '' , 391395 , ( cnrs ipul , riga , latvia , 2014 ) .
j. p. oosterhuis , s. bhler , d. wilcox and t. h. van der meer , `` a numerical investigation on the vortex formation and flow separation of the oscillatory flow in jet pumps . ''
j. acoust .
, * 137 * , 17221731 ( 2015 ) .
k. tang , y. feng , s.h .
jin , t. jin , and m. li , `` performance comparison of jet pumps with rectangular and circular tapered channels for a loop - structured traveling - wave thermoacoustic engine . ''
energ . , * 148 * , 305313 ( 2015 ) . | the oscillatory flow through tapered cylindrical tube sections ( jet pumps ) is characterized by a numerical parameter study .
the shape of a jet pump results in asymmetric hydrodynamic end effects which cause a time - averaged pressure drop to occur under oscillatory flow conditions .
hence , jet pumps are used as streaming suppressors in closed - loop thermoacoustic devices .
a two - dimensional axisymmetric computational fluid dynamics model is used to calculate the performance of a large number of conical jet pump geometries in terms of time - averaged pressure drop and acoustic power dissipation .
the investigated geometrical parameters include the jet pump length , taper angle , waist diameter and waist curvature . in correspondence with previous work ,
four flow regimes are observed which characterize the jet pump performance and dimensionless parameters are introduced to scale the performance of the various jet pump geometries .
the simulation results are compared to an existing quasi steady theory and it is shown that this theory is only applicable in a small operation region .
based on the scaling parameters , an optimum operation region is defined and design guidelines are proposed which can be directly used for future jet pump design . + * pacs numbers : * 43.35.ud , 43.25.nm , 43.20.mv , 47.32.ff = 1 |
chiral effective field theory ( @xmath4eft ) , pioneered by weinberg in a series of seminal papers almost two decades ago @xcite , has led to a novel understanding of strong interactions in nuclei by providing a direct link between these interactions and the symmetries of quantum chromodynamics , including chiral symmetry with its explicit and dynamical breaking mechanisms ( see review papers in refs .
@xcite ) . in its original form , @xmath4eft
is formulated in terms of pions and ( non - relativistic ) nucleons , whose interactions , strongly constrained by chiral symmetry , are organized as an expansion in powers of small momenta @xmath1 .
all heavier degrees of freedom are `` integrated out '' , and their effects are implicitly subsumed in the coupling constants accompanying local vertices . at sufficiently low energy ,
even the pions can be integrated out , and the nucleons only interact through contact vertices . in either case , two - nucleon ( @xmath5 )
contact interactions are an important aspect of eft descriptions . in the present paper
we examine the constraints that relativistic covariance imposes on the resulting @xmath5 potential up to order @xmath0 ( or next - to - next leading order , n@xmath6lo ) . at lo ( @xmath2 ) in the low energy expansion
there are only 2 independent contact interactions @xcite @xmath7 where @xmath8 and @xmath9 denote low - energy constants ( lec s ) , and the operators @xmath10 and @xmath11 are defined in terms of the non - relativistic nucleon field @xmath12 , @xmath13 and its adjoint in table [ tb : tab1 ] . here @xmath14 and @xmath15 are annihilation and creation operators for a nucleon in spin state @xmath16 , satisfying standard anticommutation relations , _
i.e. _ @xmath17_+ = ( 2\pi)^3\delta({\bf p}-{\bf p}^\prime)\ , \delta_{ss^\prime}$ ] . a sum over the repeated index @xmath18
is implied , and it is understood that field operator products are normal - ordered in @xmath10 and @xmath11 ( as well as in the @xmath19 s defined below ) .
we have suppressed isospin indices , since they will not enter in the discussion to follow ( see sec .
[ sec : nr1 ] ) .
the corresponding @xmath5 potential reads @xmath20 .operators entering the lo ( @xmath2 ) and n@xmath6lo ( @xmath0 ) contact interactions .
the left ( right ) arrow on @xmath21 indicates that the gradient acts on the left ( right ) field .
normal - ordering of the field operator products is understood .
[ cols="^,^ " , ] inspection of table [ tb : ops2 ] shows that a set of linearly independent operator combinations can be defined as @xmath22 consisting of 2 leading ( of order @xmath2 ) and 7 sub - leading ( @xmath0 ) ones , in agreement with the results of an analysis based on the heavy - baryon formulation of eft @xcite , so that the effective lagrangian can be written as @xmath23 - \frac{1}{2 } c_t \bigg [ o_t - \frac{1}{4 m^2 } \bigg(o_5 + o_6 - o_7 + o_8 \nonumber \\ & & + 2\ , o_{12}+ o_{14 } \bigg ) \bigg ] - \frac{1}{2 } c_1 ( o_1 + 2\ , o_2 ) + \frac{1}{8 } c_2 ( 2\ , o_2 + o_3 ) - \frac{1}{2 } c_3 ( o_9 + 2\ , o_{12 } ) \nonumber \\ & & - \frac{1}{8 } c_4 ( o_9 + o_{14})+ \frac{1}{4 } c_5 ( o_6 - o_5 ) - \frac{1}{2 } c_6 ( o_7 + 2 \ , o_{10 } ) -\frac{1}{16 } c_7 ( o_7 + o_8 + 2\ , o_{13})\ , .
\nonumber \\ & & \end{aligned}\ ] ] evaluation of the matrix elements of the operators @xmath19 between initial and final two - nucleon states with momenta @xmath24 and @xmath25 , _ i.e. _ @xmath26 shows that the 7 sub - leading combinations above give vanishing @xmath27-dependent contributions , and in fact lead , in the center - of - mass frame , to the 7 @xmath28- and @xmath29-dependent terms occurring in @xmath30 , eq .
( [ eq : ct2 ] ) .
similarly , the 2 leading combinations and associated @xmath31 corrections first 2 lines of eq .
( [ eq:7op])give rise , respectively , to the @xmath27-dependent terms @xmath32 and @xmath33 which , after multiplication by @xmath34 and @xmath35 , are precisely the terms entering the potential @xmath36 in eq .
( [ eq : ct2p ] ) .
as an alternative to the procedure discussed in the previous section , one can impose the poincar algebra constraints on the hamiltonians derived from the lagrangians in eqs .
( [ eq : op0lagr ] ) and ( [ eq : op2lagr ] ) , @xmath37 . in the instant form of relativistic dynamics ,
the interactions affect not only the hamiltonian @xmath38 but also the boost generators @xmath29 .
we write @xmath39 to distinguish between the operators in the absence ( with subscript 0 ) and in the presence ( without subscript ) of interactions , and impose the following commutation relations : @xmath40 = i\ , \epsilon^{ijk } j^k\ , & \left[k^i\ , , \ , k^j\right ] = - i\ , \epsilon^{ijk } j^k\ , & \left[j^i\ , , \ ,
k^j\right ] = i\ , \epsilon^{ijk } k^k \ , & \left[p^\mu\ , , \ , p^\nu\right]=0 \ , \\
\left[k^{i}\ , , \ , p^j\right ] = i\ , \delta^{ij } h\ , & \left[j^i \ , , \ , p^j\right ] = i \ , \epsilon^{ijk } p^k\ , & { \left[k^i\ , , \ , h\right ] = i\ , p^i } , & \left[j^i \ , , \ , h \right]=0\ .
\end{array}\ ] ] the free lorentz boost generators are derived from the energy - momentum tensor of the free fermionic theory @xmath41 as @xmath42 where , for the time being , @xmath43 denotes the field with both positive- and negative - energy components .
the use of the symmetric energy - momentum tensor , the belinfante tensor @xcite , @xmath44_+ \psi\ ] ] would make no difference .
insertion of the field expansions in terms of normal modes in the equation above and manipulations of the resulting expressions lead to @xmath45 \!\ ! & + & \!\!m\ , \widetilde{b}^\dagger_s ( { \bf p})\ , \widetilde{b}_{s ' } ( { \bf p } ) \left [ u^{(s ) \dagger } ( { \bf p } ) \overleftrightarrow{\nabla}^i_{\!\!\bf p}\ , u^{(s ' ) } ( { \bf p})\right ] \nonumber \\
+ \,e_p \left[\widetilde{d}^{\ , \dagger}_s ( { \bf p } ) \overleftrightarrow{\nabla}_{\!\!\bf p}^i \ , \widetilde{d}_s ( { \bf p})\right]\!\ ! & -&\!\!m\ , \widetilde{d}^{\,\dagger}_{s ' } ( { \bf p})\ , \widetilde{d}_s ( { \bf p } ) \left [ v^{(s)\dagger } ( { \bf p } ) \overleftrightarrow{\nabla}^i_{\!\ ! \bf p } \ , v^{(s ' ) } ( { \bf p } ) \right ] \bigg]\ , \end{aligned}\ ] ] where @xmath46 and @xmath47 are annihilation and creation operators for antinucleons , and @xmath48 denotes a derivative with respect to @xmath49 .
note that @xmath50 is time independent , since it is the spatial integral of the time component ( @xmath51 ) of a conserved current , @xmath52 with @xmath53 . by making use of @xmath54 = - i\ ,
e_p { \bm \nabla}_{\bf p}\ , \widetilde{b}_s ( { \bf p } ) - \frac{1}{2\ , ( m + e_p ) } \ , { \bf p } \times { \bm \sigma}_{s s ' } \ , \widetilde{b}_{s ' } ( { \bf p } ) \ , \label{eq : bk0}\ ] ] and a similar relation for @xmath55 , in which the only difference is the sign of the second term on the right - hand - side of eq .
( [ eq : bk0 ] ) , one can show that the following commutation relations between the `` free '' generators are fulfilled @xmath56 = i\ , \delta^{ij } h_0 \ , \qquad \left[{\bf k}_0\ , , \ , h_0\right ] = i\ , { \bf p } \ , \ ] ] where @xmath57 \ .\ ] ] we now turn our attention to the interacting theory .
the addition of an interaction term @xmath58 requires the addition of a corresponding term @xmath59 in the boost generators , as inspection of the first commutator on the second line of eq .
( [ eq : poincare ] ) makes clear .
quite generally , this term can be expressed as @xmath60 , where @xmath61 and @xmath62 is translationally invariant , _
i.e. _ @xmath63=0 $ ] .
this latter condition ensures that the commutator @xmath64=i \ , \delta^{ij}\ , h$ ] is satisfied , since @xmath65=i \,\delta^{ij}\ , h_i$ ] .
a `` minimal '' choice would correspond to the case @xmath66 @xcite . in order to proceed systematically ,
it is useful to introduce the following low - energy power counting @xmath67 which follows by observing that @xmath68 and @xmath69 , as well as their non - relativistic counterparts @xmath70 and @xmath71 , each scale as @xmath72 , and by expanding @xmath73 and the dirac spinors in powers of @xmath74 .
we now require that the commutation relations among the poincar group generators be satisfied order by order in this power counting . to this end , it is useful to express @xmath75 where the superscript @xmath76 denotes the order in our power counting , that is @xmath77
. then the commutators of @xmath78 and @xmath79 with the non - relativistic annihilation operator @xmath14 read at leading order as @xmath80 = -i \,m\ , { \bm \nabla}_{\bf p } \ , b_s ( { \bf p } ) \ , \qquad \left [ h^{(0)}_0 \ , , \ , b_s ( { \bf p } ) \right ] = -m\ , b_s ( { \bf p})\ , \ ] ] and at next to leading order as @xmath81 & = & -i\ , \frac{p^2}{2 m } \,{\bm \nabla}_{\bf p } \ , b_s ( { \bf p } ) - \frac{1}{4 m } { \bf p } \times { \bm \sigma}_{s s ' } \ , b_{s'}({\bf p } ) -i\ , \frac{{\bf p}}{2m}b_s({\bf p})\ , \\ \left [ h^{(2)}_0 \ , , \ , b_s ( { \bf p})\right ] & = & -\frac{p^2}{2 m } \ , b_s ( { \bf p})\ , \end{aligned}\ ] ] where the last term in eq
. ( [ eq : k0b ] ) comes from the gradient @xmath82 acting on the factor @xmath83 relating @xmath84 to @xmath14 . it can now be shown that only the sub - leading terms of @xmath85 $ ] and @xmath86 $ ] , respectively of order @xmath2 and @xmath87 , are non - vanishing , consistently with the power counting for the angular momentum ( @xmath88 ) and linear momentum ( @xmath27 ) operators , established in eq .
( [ eq : hcnt ] ) .
we write the interaction hamiltonian as @xmath89 where @xmath90 and @xmath91 are obtained from the lagrangians in eqs .
( [ eq : op0lagr ] ) and ( [ eq : op2lagr ] ) , and the superscripts denote the order in our power counting . correspondingly , we have @xmath92 assuming , for the time being , @xmath93 , the relations to satisfy are @xmath94\!\ ! & = & \!\ ! -i\ , \epsilon_{ijk}\ , j^k = \left [ k_0^i \ , , \ , k_0^j \right ] \ , \label{eq : kkrel}\\ \!\!\!\ ! \left [ { \bf k}_0 + { \bf w}^{(2 ) } + { \bf w}^{(4 ) } + \dots\ , , \ , h_0 + h^{(3 ) } + h^{(5 ) } + \dots \right]\!\ ! & = & \!\ ! i\ , { \bf p } = \left [ { \bf k}_0 \ , , \ , h_0\right ]
\label{eq : khrel}\ , \end{aligned}\ ] ] where the @xmath95 represent additional terms to be determined below .
these relations impose non trivial constraints on @xmath90 and @xmath91 .
we first examine those on @xmath90 . by expanding @xmath96 and @xmath97 as in eq .
( [ eq : ek0h0 ] ) , we find that the leading order relations @xmath98
+ \left[w^{(2)i}\ , , \ , k_0^{(-1)j } \right]&=&0 \ , \\ \label{eq : comm43a } \left [ { \bf k}^{(-1)}_0\ , , \ , h^{(3 ) } \right ] + \left [ { \bf w}^{(2)}\ , , \ , h^{(0)}_0 \right ] & = & 0 \ , \end{aligned}\ ] ] are fulfilled ( see appendix [ app : a1 ] ) , so that inclusion of the ( leading ) contact hamiltonian @xmath99 does not spoil the poincar covariance of the theory ( in leading order)the operators @xmath10 , @xmath11 , and @xmath19 are those defined in table [ tb : tab1 ] . at
next - to - leading order , we may split @xmath91 as @xmath100 where @xmath101 and the corresponding @xmath102 are found by imposing the relations @xmath103
+ \left[w^{(2)i}\ , , \ , k_0^{(1)j } \right ] + \left[k_0^{(-1)i}\ , , \ ,
w_1^{(4)j}\right ] + \left[w_1^{(4)i}\ , , \ , k_0^{(-1)j } \right ] & = & 0\ , \\
\label{eq : comm44a } \left [ { \bf k}^{(1)}_0\ , , \ , h^{(3 ) } \right ] + \left [ { \bf k}^{(-1)}_0\ , , \ , h_1^{(5 ) } \right ] + \left [ { \bf w}^{(2)}\ , , \ , h^{(2)}_0 \right ] + \left [ { \bf w}_1^{(4)}\ , , \ , h^{(0)}_0\right ] & = & 0 \ .\end{aligned}\ ] ] after some algebra ( see appendix [ app : a1 ] ) , we find that @xmath104 with @xmath105 the constraints involving @xmath106 and the corresponding @xmath107 , @xmath108 + \left [ w_2^{(4)i}\ , , \ , k_0^{(-1)j } \right]&=&0 \ , \\ \label{eq : cr2b } \left[{\bf k}^{(-1)}_0\ , , \ , h_2^{(5)}\right ] + \left [ { \bf w}_2^{(4)}\ , , \ , h_0^{(0 ) } \right]&= & 0 \ , \end{aligned}\ ] ] are fulfilled as long as the hamiltonian @xmath109 is constructed out of the 7 sub - leading operators listed at the end of sec .
[ sec : nr2 ] , or combinations thereof .
this is also shown in appendix [ app : a1 ] .
the hamiltonian @xmath110 leads to the @xmath27-dependent and @xmath27-independent potentials in eqs .
( [ eq : ct2 ] ) and ( [ eq : ct2p ] ) , in accordance with the derivation presented in sec .
[ sec : nr ] . in closing
, we note that , although these constraints correspond to the `` minimal choice '' @xmath93 , the result holds in the general case .
indeed , the requirement that @xmath62 commute with the three - momentum operator implies that it be constructed as a spatial integral of fields and their derivatives only .
no factors of @xmath111 , which would lower the counting power , are allowed inside the integral .
the minimal power , in our counting , of an interacting ( two - body ) boost operator @xmath62 is therefore 3 , but is actually 4 if the relations ( [ eq : kkrel])([eq : comm44a ] ) have to be fulfilled order by order . as a result , the only contributions of @xmath62 to eqs .
( [ eq : comm44 ] ) and ( [ eq : comm44a ] ) are given by @xmath112 $ ] and @xmath113 $ ] , both of which vanish , since @xmath62 , being hermitian , must contain an equal number of creation and annihilation operators .
one of the authors ( r.s . ) would like to thank the physics department of the university of pisa , the infn pisa branch , and especially the pisa group for the support and warm hospitality extended to him on several occasions .
the work of r.s . is supported by the u.s .
department of energy , office of nuclear physics , under contract de - ac05 - 06or23177 .
in this appendix we outline the derivation of the leading- and next - to - leading order relations in eqs .
( [ eq : comm43])([eq : comm43a ] ) and eqs .
( [ eq : comm44])([eq : comm44a ] ) involving @xmath90 , as well as of the leading order relations in eqs .
( [ eq : cr2a])([eq : cr2b ] ) involving @xmath91 . for brevity , we suppress spin indices , and introduce the notation @xmath114 , @xmath115 , and @xmath116 consider the commutator @xmath117=\left [ k_0^{(-1)i}\ , , \ , c_s w_{s}^{(2)j}+ c_t w_{t}^{(2)j } \right ] \ , \ ] ] where the terms @xmath118 and @xmath119 correspond to the interactions @xmath120 and @xmath121 [ see eq .
( [ eq : leadingh ] ) ] , for example @xmath122 making use of @xmath123 = -i\ , m\ , \left ( \sum_{k=1}^4{\bm \nabla}_k\right)\ , b_1^\dagger\ , b_2\ , b_3^\dagger\ , b_4 \ , \ ] ] we find that @xmath124=-i\ , \frac{m}{2 } \int d{\bf x}\,x^j\ , \int_{\bf p } \left ( \sum_{k=1}^4\nabla_k^{\,i } \right)\ , b_1^\dagger\ , b_2\ , b_3^\dagger\ , b_4 \ , \ ] ] which vanishes after partial integrations with respect to the @xmath125s note the exponential factor in eq .
( [ eq : adf1 ] ) . the terms involving @xmath126 as well as those occurring in
@xmath127 $ ] can be worked out similarly , while those in @xmath128 $ ] vanish , since @xmath129 = 0 \ .\ ] ] thus each of the commutators entering the leading order relations vanishes . moving on to the next - to - leading order relations ,
consider first @xmath130 .
we obtain @xmath131 = \frac{1}{2 m } \int d{\bf x } \int_{\bf p } \left [ \big [ i\ , ( { \bf p}_1 + { \bf p}_2 ) + ( p_1 ^ 2 - p_2 ^ 2 ) \ , { \bf x } \big ] b_1^\dagger\ , b_2\ , b_3^\dagger\ , b_4 + b^\dagger_1\ , \frac{{\bf p}_1 - { \bf p}_2}{2 } \times { \bm \sigma}\ , b_2 \ , b^\dagger_3 \ , b_4 \right ] \ , \ ] ] where the terms involving the momenta @xmath132 and @xmath133 reduce to those with @xmath134 and @xmath135 after exchanging @xmath136 and @xmath137 . the linear term in @xmath111
is canceled by @xmath138 $ ] . in order to cancel the rest
, one requires an interaction term @xmath139 , given by @xmath140 \nonumber \\ & = & \frac{1}{8 m^2 } \int d{\bf x } \big [ ( n^\dagger\overleftarrow{\nabla}^2 n + n^\dagger \overrightarrow{\nabla}^2 n ) \,(n^\dagger n ) + ( n^\dagger \overrightarrow{{\bm \nabla } } n)^2 + ( n^\dagger \overleftarrow{{\bm \nabla } } n)^2 \nonumber \\ & & \ , + i\ , ( n^\dagger \overleftarrow{\bm \nabla } \cdot { \bm \sigma } \times \overrightarrow{\bm \nabla } n ) \ , ( n^\dagger n ) + i \ , ( n^\dagger { \bm \sigma } n ) \cdot ( n^\dagger \overleftarrow{\bm \nabla } \times \overrightarrow{\bm \nabla } n ) \big]\ , \end{aligned}\ ] ] and a corresponding @xmath141 , which , however , commutes with @xmath142 . proceeding in a similar fashion for @xmath121
, one finds that an interaction term @xmath143 , @xmath144 \ , b^\dagger_1\ , \sigma^j \ , b_2 \ , b^\dagger_3\ , \sigma^k \ , b_4 \nonumber \\ & & \ , + i\ , b^\dagger_1\ , ( { \bf p}_3 \times { \bf p}_4 - { \bf p}_1 \times { \bf p}_2 ) \cdot { \bm \sigma } \ , b_2\ , b^\dagger_3\ , b _ 4 \big ] \nonumber \\ & = & \!\!- \frac{1}{8m^2}\int d{\bf x } \big [ - ( n^\dagger { \bm \sigma } \cdot \overrightarrow{\bm \nabla } n ) ( n^\dagger { \bm \sigma}\cdot \overrightarrow{\bm \nabla } n ) - ( n^\dagger { \bm \sigma } \cdot \overleftarrow{\bm \nabla } n ) ( n^\dagger { \bm \sigma}\cdot \overleftarrow{\bm \nabla } n ) \nonumber \\ & & \ , + ( n^\dagger \sigma^j \overrightarrow{\nabla^k } n)(n^\dagger \sigma^k \overrightarrow{\nabla^j } n ) + ( n^\dagger \sigma^j \overleftarrow{\nabla^k } n)(n^\dagger \sigma^k \overleftarrow{\nabla^j } n ) \nonumber \\ & & \ , + 2 ( n^\dagger \sigma^j \overrightarrow{\nabla^k } n ) ( n^\dagger \overleftarrow{\nabla^k } \sigma^j n ) + 2\ , ( n^\dagger \overleftarrow{\bm \nabla } \sigma^j \cdot \overrightarrow{\bm \nabla } n ) ( n^\dagger \sigma^j n ) \nonumber \\ & & \ , + i\ , ( n^\dagger \overleftarrow{\bm \nabla } \cdot { \bm \sigma } \times \overrightarrow{\bm \nabla } n ) \ , ( n^\dagger n ) + i \ , ( n^\dagger { \bm \sigma } n ) \cdot ( n^\dagger \overleftarrow{\bm \nabla } \times \overrightarrow{\bm \nabla } n ) \big ] \ , \end{aligned}\ ] ] and a corresponding boost operator @xmath145 , are required in order to satisfy the @xmath146-piece of the commutators . thus eq .
( [ eq : comm44a ] ) holds .
similarly , eq . ( [ eq : comm44 ] ) can also be shown to hold .
the expressions for @xmath139 and @xmath143 correspond to those listed in eqs .
( [ eq : h5s])([eq : h5 t ] ) .
we now turn our attention to the constraint on @xmath147 implied by eq .
( [ eq : cr2b ] ) .
we first observe that the commutator @xmath148 $ ] vanishes .
defining @xmath149 \equiv \int d{\bf x}\ , \left [ { \bf k}^{(-1)}_0 , o_i\right]\ , \ ] ] we find @xmath150 & = -2\,i\ , m \int d{\bf x } \int_{\bf p } ( { \bf p}_1 + { \bf p}_2)\ , b_1^\dagger\ , b_2\ , b_3^\dagger\ , b_4\ , \\ \left [ o_2\right ] & = i\ , m \int d{\bf x } \int_{\bf p } ( { \bf p}_1 + { \bf p}_2)\ , b_1^\dagger\ , b_2\ , b_3^\dagger\ , b_4\ , \\ \left [ o_3\right ] & = -2\,i\ , m \int d{\bf x } \int_{\bf p } ( { \bf p}_1 + { \bf p}_2)\ , b_1^\dagger\ , b_2\ , b_3^\dagger\ , b_4\ , \\ \left [ o_4\right ] & = 0 \\ \left [ o_5\right ] & = - m \int d{\bf x } \int_{\bf p } ( { \bf p}_1 - { \bf p}_2)\ , \times b^\dagger_1\ , { \bm \sigma } \,b_2\ , b^\dagger_3\ , b_4 \ , \\ \left [ o_6\right ] & = m \int d{\bf x } \int_{\bf p } ( { \bf p}_3 - { \bf p}_4)\ , \times b^\dagger_1\ , { \bm \sigma } \,b_2\ , b^\dagger_3\ , b_4\ , \\
\left [ o_7\right ] & = -2\ , i \ , m \int d{\bf x } \int_{\bf p } b^\dagger_1\ , { \bm \sigma}\ , b_2 \ , b^\dagger_3\ , ( { \bf p}_3 + { \bf p}_4 ) \cdot { \bm \sigma } \ , b_4\ , \\
\left [ o_8\right ] & = -2\ , i \ , m \int d{\bf x } \int_{\bf p }
b^\dagger_1\ , { \bm \sigma}\ , b_2 \ , b^\dagger_3\ , ( { \bf p}_1 + { \bf p}_2 ) \cdot { \bm \sigma } \ , b_4\ , \\
\left [ o_9\right ] & = -2\ , i \ , m \int d{\bf x }
\int_{\bf p}({\bf p}_1+{\bf p}_2)\ , b^\dagger_1{\bm \sigma}\ , b_2\cdot b^\dagger_3\ , { \bm \sigma } \ , b_4\ , \\
% \end{eqnarray } % \begin{eqnarray } \left [ o_{10}\right ] & = i \ , m \int d{\bf x } \int_{\bf p } b^\dagger_1{\bm \sigma}\ , b_2\ , b^\dagger_3\ , ( { \bf p}_3+{\bf p}_4)\cdot{\bm \sigma } \ , b_4\ , \\ \left [ o_{11}\right ] & = i \ , m \int d{\bf x } \int_{\bf p } b^\dagger_1{\bm \sigma}\ , b_2\ , b^\dagger_3\ , ( { \bf p}_1+{\bf p}_2)\cdot{\bm \sigma } \ , b_4 \ , \\ \left [ o_{12}\right ] & = i \ , m \int d{\bf x } \int_{\bf p}({\bf p}_1+{\bf p}_2)\ , b^\dagger_1{\bm \sigma}\ , b_2\cdot b^\dagger_3\ , { \bm \sigma } \ ,
\left [ o_{13}\right ] & = i \ , m \int d{\bf x } \int_{\bf p } b^\dagger_1{\bm \sigma}\ , b_2\ , b^\dagger_3\ , ( { \bf p}_1+{\bf p}_2+{\bf p}_3+{\bf p}_4)\cdot{\bm \sigma } \ , b_4\ ,
\\ \left [ o_{14}\right ] & = 2\ , i \ , m \int d{\bf x }
\int_{\bf p}({\bf p}_1+{\bf p}_2)\ ,
b^\dagger_1{\bm \sigma}\ , b_2\cdot b^\dagger_3\ , { \bm \sigma } \ ,
b_4 \ , % \end{eqnarray}\end{aligned}\ ] ] and only 7 combinations of these operators satisfy the constraint in eq .
( [ eq : cr2b ] ) , such as those in eq .
( [ eq:7op ] ) .
it is possible to show that these 7 combinations also satisfy the constraint of eq .
( [ eq : cr2a ] ) .
100 s. weinberg , phys .
b * 251 * , 288 ( 1990 ) ; nucl .
b * 363 * , 3 ( 1991 ) ; phys .
b * 295 * , 114 ( 1992 ) .
v. bernard , n. kaiser and u. g. meissner , int .
j. mod .
e * 4 * , 193 ( 1995 ) ; u. van kolck , prog . part .
phys . * 43 * , 337 ( 1999 ) ; p. f. bedaque and u. van kolck , ann .
* 52 * , 339 ( 2002 ) ; e. epelbaum , prog .
* 57 * , 654 ( 2006 ) ; e. epelbaum , h. w. hammer and u. g. meissner , rev .
* 81 * , 1773 ( 2009 ) . c. ordonez , l. ray and u. van kolck , phys .
c * 53 * , 2086 ( 1996 ) .
s. pastore , l. girlanda , r. schiavilla , m. viviani and r. b. wiringa , phys .
c * 80 * , 034004 ( 2009 ) .
e. epelbaum , w. gloeckle and u. g. meissner , nucl .
a * 637 * , 107 ( 1998 ) .
a. m. dirac , rev .
phys . * 21 * , 392 ( 1949 ) .
r. a. krajcik and l. l. foldy , phys .
d * 10 * , 1777 ( 1974 ) ; j. l. friar , phys . rev .
c * 12 * , 695 ( 1975 ) .
j. carlson , v. r. pandharipande and r. schiavilla , phys .
c * 47 * , 484 ( 1993 ) ; j. l. forest , v. r. pandharipande , j. carlson and r. schiavilla , phys .
c * 52 * , 576 ( 1995 ) .
e. epelbaum , ph.d .
thesis ( bochum , 2000 ) unpublished ; e. epelbaum , u. g. meissner , w. gloeckle and c. elster , phys . rev .
c * 65 * , 044001 ( 2002 ) [ arxiv : nucl - th/0106007 ] .
h. georgi , phys .
b * 240 * , 447 ( 1990 ) ; e. e. jenkins and a. v. manohar , phys .
b * 255 * , 558 ( 1991 ) ; v. bernard , n. kaiser , j. kambor and u. g. meissner , nucl .
b * 388 * , 315 ( 1992 ) .
m. e. luke and a. v. manohar , phys .
b * 286 * , 348 ( 1992 ) .
n. brambilla , d. gromes and a. vairo , phys .
d * 64 * , 076010 ( 2001 ) ; phys .
b * 576 * , 314 ( 2003 ) ; a. vairo , nucl .
suppl . * 133 * , 196 ( 2004 ) .
h. witala , j. golak , r. skibinski , w. glockle , w. n. polyzou and h. kamada , phys .
c * 77 * , 034004 ( 2008 ) .
l. girlanda , s. pastore , r. schiavilla , and m. viviani , unpublished .
s. weinberg , _ the quantum theory of fields _ ( cambridge university press , 1995 ) , vol . | we construct the most general , relativistically invariant , contact lagrangian at order @xmath0 in the power counting , @xmath1 denoting the low momentum scale . a complete , but non - minimal , set of ( contact ) interaction terms is identified , which upon non - relativistic reduction generate 2 leading independent operator combinations of order @xmath2 and 7 sub - leading ones of order @xmath0a result derived previously in the heavy - baryon formulation of effective field theories ( eft s ) .
we show that poincar covariance of the theory requires that additional terms with fixed coefficients be included , in order to describe the two - nucleon potential in reference frames other than the center - of - mass frame .
these terms will contribute in systems with mass number @xmath3 , and their impact on eft calculations of binding energies and scattering observables in these systems should be studied . |
black hole thermodynamics has been an exciting and challenging topic in theoretical physics .
inspired especially by the ads / cft correspondence @xcite , the anti - de sitter ( ads ) black hole thermodynamics has attracted much attention .
it was first noted that in ads space , there exists hawking - page phase transition between a stable large black hole and thermal gas @xcite , which can be explained as the confinement / deconfinement phase transition of the gauge field @xcite
. further study of charged ads black holes revealed that there is a first - order phase transition between small and large black holes in the canonical ensemble @xcite . with an increase of the temperature , the phase transition point terminates at a critical point , where the first - order phase transition becomes a second - order one .
the behavior of such a phase transition is quite similar to the van der waals ( vdw ) fluid .
most recently , the investigation of the ads black hole thermodynamics has been generalized to the extended phase space , where the cosmological constant is treated as a dynamical pressure and its conjugate quantity is the thermodynamic volume of the black hole @xcite .
phase transition in the charged ads black hole was reexamined in ref .
it was found that the black hole system and the vdw system share the same pressure - thermodynamic ( @xmath0)oscillatory behavior , critical exponents , and scaling laws .
thus , the precise analogy between the charged ads black hole and the vdw system was established .
later , the study of the ads black hole phase transition in extended phase space was generalized to different ads black hole backgrounds @xcite . in these studies , it was shown that besides the small / large black hole first - order phase transition reminiscent of the liquid / gas transition of the vdw fluid , there were also some new interesting phenomena , such as reentrant phase transitions of multicomponent liquids , multiple first - order solid / liquid / gas phase transitions , and liquid / gas phase transitions of the vdw type .
for example , the ( @xmath1)-dimensional single spinning vacuum kerr - ads black hole demonstrates the peculiar behavior of the large / small / large black hole phase transitions @xcite , and the multiply rotating kerr - ads black hole system displays a small / intermediate / large black hole phase transition with one tricritical and two critical points in some range of the parameters @xcite .
the doubly spinning myers - perry black hole also has a similar reentrant phase transition structure @xcite .
such a reentrant phase transition is also observed in modified gravity theories , such as born - infeld gravity @xcite , gauss - bonnet gravity @xcite , and lovelock gravity @xcite .
on the other hand , understanding the thermodynamic properties of a black hole system along or across the coexistence curve is also an interesting subject .
the authors of ref .
@xcite calculated the quasinormal modes of massless scalar perturbations around the small and large four - dimensional charged ads black holes .
the result showed that the quasinormal modes can be a dynamic probe of the thermodynamic phase transition .
toward understanding the thermodynamic properties of the charged ads black hole system along or across the coexistence curve , one expects an analytic formula of the coexistence curve .
however , there is only a numerical calculation of it . in this paper , we reconsider the small / large black hole phase transition and present the fitting formula of the coexistence curve in the reduced parameter space for ( @xmath2)-dimensional charged ads black holes .
the result confirms our analysis that the fitting formula is independent of charge for any dimension @xmath3 .
along the coexistence curve , the clapeyron equations are also reexamined . besides the conventional one ,
two additional equations are obtained in the extended phase space .
based on the fitting formula of the coexistence curve , these clapeyron equations are numerically checked .
this paper is organized as follows . in sec .
[ dddd ] , we investigate the equal area laws and the clapeyron equations in the extended phase space for the charged ads black hole . in sec .
[ equalarealaw ] , we present the fitting formula of the coexistence curve and check the clapeyron equations
. we will summarize our results in sec .
[ conclusion ] .
recently , there has been great interest in studying the phase transition by treating the cosmological constant as a dynamical pressure , i.e. , @xmath4 with such a new interpretation , the cosmological constant can be regarded as a new thermodynamic variable . in this extended phase space , the black hole mass @xmath5 is considered as the enthalpy @xmath6 rather than the internal energy of the gravitational system @xcite , and the smarr relation for a @xmath3-dimensional charged ads black hole is modified as @xmath7 differentiating it , we get the first law : @xmath8 with @xmath9 being the thermodynamic volume @xmath10 furthermore , the gibbs free energy can be obtained by the legendre transformations @xmath11 for a specific spacetime background , one can clearly get its gibbs free energy . considering that the thermodynamic system prefers a low gibbs free energy ,
we then can obtain its phase transition information .
now the study of phase transition has been applied to a number of ads black holes , and richer phase structures are displayed .
many ads black hole systems exhibit the small / large black hole phase transition , which is reminiscent of the vdw liquid / gas phase transition .
the state equation for them can be expressed as @xmath12 analogous to the vdw fluid , here @xmath13 can be interpreted as the specific volume of the black hole fluid .
study shows that both of them share the similar physical properties , such as the @xmath14 ( pressure - specific volume ) oscillatory behavior , the critical exponents , and scaling laws near the critical point
. moreover the reentrant phase transitions were also found in the black hole system .
in general , the phase diagram or the coexistence curve can be obtained by analyzing the characteristic swallow tail behavior of the gibbs free energy , and it is extensively used in the related papers . on the other hand ,
it can also be obtained by constructing the maxwell s equal area law . however , after a simple check , we arrive at the result that the phase diagrams obtained from the gibbs free energy and the equal area law in @xmath14 behavior are not consistent with each other .
we suggest that the problem originates from the confusion of thermodynamic volume @xmath9 and specific volume @xmath13 . for the vdw fluid , the equal area law is effective for both thermodynamic and specific volumes , because @xmath15 , with @xmath16 being a fixed constant denoting the total number of the fluid molecules . for the black hole case ,
however , such a linear relation does not hold anymore .
thus , considering eq .
( [ gibss ] ) , the maxwell s equal area law is effective only for the thermodynamic volume @xmath9 in a black hole system . before discussing the equal area law
, we plot a sketch picture of the phase diagram in fig .
the red line denotes the coexistence curve , above and below which there are two different system phases .
each point located on the curve corresponds to two different phases while having the same gibbs free energy .
for example , states @xmath17 and @xmath18 are related by one system phase , while states @xmath19 and @xmath20 by another phase . between them , states @xmath17 @xmath21 and @xmath19 @xmath22 have the same gibbs free energy . considering that a system transforms state @xmath17 to @xmath19 , i.e. , @xmath23 , we have @xmath24 now
, we integrate it : @xmath25 * case i : * charge @xmath26 and temperature @xmath27 fixed . for this case
, we consider the isotherm in the @xmath0 plane .
therefore , the first and second terms in eq .
( [ int ] ) vanish , and we get @xmath28 along the coexistence curve , we know that the two states have the same pressure , i.e. , @xmath29 . here
we show the isotherm in the @xmath30 plane in fig .
it is clear that oscillatory behavior exists .
then the integral in eq . ( [ vp0 ] )
can be reexpressed as @xmath31 the pressure @xmath32 is equivalent to that of the phase transition pressure . from fig .
[ pvt ] , we are see clearly that the first bracket in ( [ interalvdp ] ) denotes the area of region ii and the second one the negative area of region i. thus , we get @xmath33 this is in fact the maxwell s equal area law , and we are familiar with it as showed in the @xmath0 plane [ see fig . [ ptv ] ] .
therefore , we see that this equal area law holds during the phase transition .
on the other hand , we can also use such an area law to find the value of the phase transition parameter .
letting the temperature @xmath27 take all possible values , we will obtain the phase diagram for each charge @xmath26 using such an area law .
it is also worthwhile to note that this equal area law is effective for the thermodynamic volume @xmath9 rather the specific volume @xmath13 . and
@xmath19 , @xmath18 and @xmath20 , respectively , can coexist with each other.,width=302,height=226 ] * case ii : * charge @xmath26 and pressure @xmath34 fixed . in the above case
, we see that the equal area law holds in the @xmath0 plane with @xmath26 and @xmath27 fixed .
there is an equivalent choice in which @xmath26 and @xmath34 are fixed .
under such a choice , one can carry out the above analysis and the equal area law will be found in the temperature - entropy ( @xmath35 ) plane .
this generalized area law was first studied in refs .
@xcite .
* case iii : * pressure @xmath34 and temperature @xmath27 fixed .
during the phase transition , eq .
( [ int ] ) reduces to @xmath36 which can be reexpressed as @xmath37 where @xmath38 is considered .
this equation is just another expression of the equal area law . here
, it is clear that the maxwell s equal area law for the charged ads black hole gets a generalization . and the three area laws are consistent with each other . here
we continue to consider the manner in which the black hole system passes from state @xmath17 to @xmath19 .
the gibbs free energies related to the two states are @xmath39 taking charge @xmath26 as fixed and noting that @xmath40 , we have @xmath41 then the phase transition pressure can be expressed as @xmath42 a similar calculation can be applied to the case of fixed @xmath43 , which corresponds to the grand canonical ensemble , and eq .
( [ pa ] ) becomes @xmath44 the clapeyron equation is a useful equation for studying the liquid / gas phase transition of the vdw fluid .
here we will review it and make a generalization to the charged black hole system among the small / large black hole phase transition .
since states @xmath17 and @xmath19 , @xmath18 and @xmath20 are in phase equilibrium , @xmath45 thus @xmath46 if states @xmath17 ( @xmath19 ) and @xmath18 ( @xmath20 ) are very close , we then have @xmath47 by inserting them into eq .
( [ gg ] ) , we easily get @xmath48 taking @xmath26 , @xmath34 , and @xmath27 as fixed , we obtain , respectively @xmath49 the first one is just the conventional clapeyron equation .
the other two are the generalized clapeyron equations in the extended phase space of the charged ads black hole system . in ref .
@xcite , the author has checked eq .
( [ clapeyron ] ) for the black hole / ads space phase transition in the nonrotating and rotating black hole systems .
however , checking these equations during the small / large black hole phase transition requires a numerical calculation , and we will carry that out in the next section . using these equations , one can easily check the maxwell s relation @xmath50 . on the other hand , the pressure @xmath51 in eq .
( [ pa ] ) can be reexpressed as @xmath52
here we first give a brief review of the charged ads black hole .
its line element is given by @xmath53 the parameters @xmath54 and @xmath55 , relate to the enthalpy and charge of the black hole , respectively @xmath56 with @xmath57 the volume of the unit @xmath3 sphere .
the black hole temperature , entropy , electric potential , and thermodynamic volume are given by @xmath58 the gibbs free energy @xmath59 reads @xmath60 we plot in fig .
[ pgtq ] the characteristic swallow tail behavior of the gibbs free energy with fixed charge @xmath26 and pressure @xmath34 in the @xmath61 plane . at the point @xmath16 , it can be seen that the gibbs free energy of the charged black hole vanishes .
note that the gibbs free energy for ads space also vanishes .
hence it seems that at that point the black hole phase and the ads space phase can coexist .
however , the concept that empty charged space " is unphysical .
thus we could not compare these two phases with fixed charge together , and the ads phase will not be considered in the following discussion . after excluding the ads space phase , the system will first go along the small black hole branch with the increasing of @xmath27 until the point @xmath5 is arrived , where the small and large black holes have the same gibbs free energy , and the small and large black holes can coexist at that point . with the temperature further increasing , the system will prefer the large black hole with lower gibbs free energy over the small and intermediate black holes .
that is the small / large black hole phase transition . in the extended phase space ,
the small / large black hole phase transition was well studied in ref .
this phase transition is reminiscent of the liquid / gas phase transition of the vdw fluid .
this coexistence curve has positive slope everywhere and terminates at the critical point @xmath62 , above which the small and large black hole phases can not be clearly distinguished .
the critical point is determined by @xmath63 by solving the above two equations , we get the critical point : @xmath64 with the critical specific volume @xmath65 .
here we would like to give a brief discussion of the thermodynamic quantities and phase transition in the reduced parameter space .
the reduced thermodynamic quantity corresponding to @xmath17 can be defined as @xmath66 with @xmath67 being its critical value . then the reduced temperature and
the gibbs free energy can be expressed as @xmath68 from which we can see that @xmath69 and @xmath70 are independent of the black hole charge @xmath26
. using the fact that the phase transition point is determined by the swallow tail behavior of the gibbs free energy , we can get the natural result that the phase transition in the reduced parameter space is charge independent .
this can also be found from the equal area law .
for example , the equal area law on the @xmath0 line is described by eq .
( [ vp0 ] ) with fixed @xmath26 and @xmath27 .
a simple calculation shows that @xmath71 .
then we have the equal area law in the reduced parameter space , @xmath72 since the reduced temperature in ( [ tp ] ) is charge independent , the phase transition point determined by constructing the equal area law on the ( reduced ) isotherm is also charge independent .
it is also worthwhile to point out that , in the reduced parameter space , all of the reduced physical quantities have a value of order 1 . a word to conclusion is that there is a charge independent coexistence curve for any @xmath3 . for the case of @xmath73 ,
the values of these critical parameters reduce to @xmath74 and the reduced temperature and gibbs free energy are @xmath75{\tilde{v}}+\frac{3}{\sqrt[3]{\tilde{v}}}\right).\end{aligned}\ ] ] according to the swallow tail behavior of @xmath70 or the equal area law , the phase diagram in the @xmath76 plane ( or the coexistence curve ) can be obtained . according to our above analysis
, the coexistence curve is charge independent .
on the other hand , as shown in many works , this coexistence curve is of the vdw type , and it has positive slope everywhere and terminates at the critical point .
thus , we can fit it using the polynomial of @xmath69 . combined with the calculated values ,
the parametrization form of it reads @xmath77 we plot the numerical values ( the discrete points ) and the fitting formula ( the solid line ) of the coexistence curve in fig .
it is quite clear that the numerical values and the fitting formula match well with each other .
inverting @xmath78 to @xmath79 , we will obtain a coexistence surface , which is charge dependent , and we show it in fig .
[ pqt4b ] .
+ with the help of the fitting formula of the coexistence curve ( [ coex4 ] ) , we would like to check the equal area law and the clapeyron equations obtained in sec .
[ dddd ] .
the equal area law is shown in fig .
[ parealaw ] .
these figures show that the three equal area laws discussed above hold .
it is also worthwhile to note that the equal area law in the @xmath14 plane fails . from fig .
[ ppv2 ] , it can be seen that these two areas in the shadow are not equal to each other , and a difference of 0.0023 exists .
this supports our result that the equal area law holds only for the isotherm in the @xmath0 plane rather the @xmath14 plane .
this result implies a difference between the thermodynamic and specific volumes . on the other hand ,
the critical point can be equivalently determined by one of the following four conditions : ( i ) @xmath80 , ( ii ) @xmath81 , ( iii ) @xmath82 , or ( iv ) @xmath83 . in ref .
@xcite , the author pointed out that the clapeyron equation ( [ clapeyron ] ) holds for the black hole / ads phase transition in the nonrotating and rotating black hole systems . in the following
, we will numerically check eqs .
( [ clapeyron])-([clapeyron4 ] ) for the small / large black hole phase transition using the parametrization form of the coexistence curve eq .
( [ coex4 ] ) . in order to make a clear comparison ,
we define two relative deviations @xmath84 where @xmath85 and @xmath86 are numerical values , and @xmath87 and @xmath88 are calculated with eq .
( [ coex4 ] ) . @xmath89
measures the deviation of @xmath87 from @xmath85 , and @xmath90 the deviation of @xmath88 from @xmath86 .
the numerical check in high temperature is listed in table [ tab1 ] , which shows that the relative deviations @xmath91 and @xmath92 are @xmath93 and @xmath94 , respectively .
they are also shown in fig .
[ pdelta ] for @xmath95 and @xmath96 , respectively . with a fixed @xmath26
, @xmath91 shows an oscillatory behavior . at low @xmath97
, @xmath91 is about @xmath98 and quickly decreases below @xmath99 with the temperature increasing . near the critical value of @xmath100 , @xmath91 blows up , which is caused by the fitting method .
@xmath92 also shares the oscillatory behavior . at small charge @xmath26 , it is about @xmath101 , while it decreases to @xmath102 at large @xmath26 .
combining with the above results , we can conclude that the clapeyron equations ( [ clapeyron ] ) and ( [ clapeyron2 ] ) hold among the small / large black hole phase transition of the charged ads black hole .
the third clapeyron equation ( [ clapeyron3 ] ) will naturally hold when we consider the maxwell s relation @xmath50 .
it is also straightforward to check that the pressure calculated from eq .
( [ clapeyron4 ] ) is exactly consistent with eq .
( [ coex4 ] ) .
.values of the thermodynamic quantities and deviations @xmath91 and @xmath92 at the phase transition points at high temperature .
@xmath103 and @xmath104 correspond to the radius of the large and small black holes .
[ cols="^,^,^,^,^,^,^,^,^",options="header " , ] from bottom to top.,width=302,height=226 ]
in this paper , we studied the thermodynamic properties of the charged ads black hole along the coexistence curve of the small and large black hole phases .
we first reviewed the classical @xmath0 oscillatory behavior , which implies the phase transition taking place . by means of the maxwell s equal area law
, the unphysical branch of the system can be excluded , and the phase transition point can also be determined . for the charged ads black hole , we showed that such @xmath0 oscillatory behavior is also obvious in the charge - electric potential ( @xmath105 ) and @xmath35 planes . at each oscillatory line ,
the equal area law is useful for finding the phase transition point , and which is exactly consistent with the analysis of the gibbs free energy .
it is worthwhile to point out that the oscillatory behavior also appears in the @xmath14 plane .
however , its equal area law fails to produce the correct phase transition point , which reminds us of the difference between the thermodynamic and specific volumes . employing the first law
, we get three clapeyron equations at the coexistence curve .
the first one is the conventional clapeyron equation with fixed charge , and the other two are its generalizations with , respectively , fixed pressure and temperature . before carrying out the numerical check of the clapeyron equations , we examined the thermodynamic quantities in the reduced parameter space .
the reduced temperature @xmath69 and the gibbs free energy @xmath70 are found to be charge independent , which will produce a charge independent phase transition , as well as the coexistence curve .
the charge independent property of the coexistence curve can also be found from the state equation expressed with the reduced parameters ; see ( [ tp ] ) . as a consequence ,
it is a universal property for the charged ads black hole of any dimension @xmath3 .
the numerical calculation also confirms this result .
however , after converting to the parameters @xmath34 and @xmath27 , the coexistence surface is obviously charge dependent . making use of the fitting formula
, we checked that these clapeyron equations hold exactly when the black hole system varies along the coexistence curve .
for example , at low temperature , the deviation is of @xmath98 , and it quickly decreases to @xmath99 for high temperature .
a simple check shows that our fitting formulas of the coexistence curve agree with the calculated values to @xmath93 . at last
, we would like to make a few comments .
recent studies mainly focus on finding the vdw - type phase transition , reentrant phase transitions or thermodynamic properties near the critical point , while only a few papers address the properties of the black hole system going along or crossing the coexistence curve .
we argue that studying the physics near the coexistence curve may help us to reveal the information of the microscopic structure of a thermodynamic system , which is still unclear from the gravity side .
so we hope that this study of the phase transition can provide us with some information about the microscopic structure of a gravitational system from the thermodynamic viewpoint . on the other hand , we presented in this paper the highly accurate fitting formulas of the coexistence curves [ see eqs .
( [ coex4 ] ) and ( [ tildeptd ] ) as well as table [ tab2 ] ] .
further research , such as the exploration of the thermodynamic geometry and examination of the quasinormal modes along the coexistence curves can be easily carried out based on them . on the other hand , seeking the fitting formula of the coexistence curve for the ads black hole systems with the rich phase structure , such as the higher - dimensional kerr - ads black hole system or those in modified gravity , is also an interesting issue .
we thank r. b. mann for providing the maple file and for the valuable discussions on the equal area law .
the authors would especially like to thank the anonymous referees whose comments helped us greatly in improving the original manuscript .
this work was supported by the national natural science foundation of china ( grants no .
11205074 and no .
11375075 ) .
b. p. dolan , d. kastor , d. kubiznak , r. b. mann , and j. traschen , _ thermodynamic volumes and isoperimetric inequalities for de sitter black holes _
, phys . rev . * d 87 * , 104017 ( 2013 ) , [ arxiv:1301.5926[hep - th ] ] .
s. gunasekaran , d. kubiznak , and r. mann , _ extended phase space thermodynamics for charged and rotating black holes and born - infeld vacuum polarization _ , jhep * 1211 * , 110 ( 2012 ) , [ arxiv:1208.6251[hep - th ] ] .
r. zhao , h. h. zhao , m. s. ma , and l. c. zhang , _ on the critical phenomena and thermodynamics of charged topological dilaton ads black holes _ , eur .
j. * c 73 * , 2645 ( 2013 ) , [ arxiv:1305.3725[gr - qc ] ] .
n. altamirano , d. kubiznak , r. b. mann , and z. sherkatghanad , _ kerr - ads analogue of triple point and solid / liquid / gas phase transition _ class .
* 31 * , 042001 ( 2014 ) , [ arxiv:1308.2672[hep - th ] ] .
n. altamirano , d. kubiznak , r. b. mann , and z. sherkatghanad , _ thermodynamics of rotating black holes and black rings : phase transitions and thermodynamic volume _ , galaxies * 2 * , 89 ( 2014 ) , [ arxiv:1401.2586[hep - th ] ] . l. c. zhang , m. s. ma , h. h. zhao , and r. zhao , _ thermodynamics of phase transition in higher dimensional reissner - nordstrom - de sitter black hole _ , eur
j. c * 74 * , 3052 ( 2014 ) , [ arxiv:1403.2151[gr - qc ] .
j. x. mo and w. b. liu , _ ehrenfest scheme for @xmath0 criticality of higher dimensional charged black holes , rotating black holes and gauss - bonnet ads black holes _ , phys
d 89 * , 084057 ( 2014 ) , [ arxiv:1404.3872[gr - qc ] ] .
b. mirza and z. sherkatghanad , _ phase transitions of hairy black holes in massive gravity and thermodynamic behavior of charged ads black holes in an extended phase space _ , phys
d 90 * , 084006 ( 2014 ) , [ arxiv:1409.6839[gr - qc ] ] . | in this paper , we first review the equal area laws and clapeyron equations in the extended phase space of the charged anti - de sitter black holes . with different fixed parameters , the maxwell s equal area law holds not only in the pressure - thermodynamic volume oscillatory line , but also in the charge - electric potential and temperature - entropy oscillatory lines .
the conventional clapeyron equation is generalized and two extra equations are found .
moreover , we show that the coexistence curve of the small and large charged black holes is charge independent in the reduced parameter space for any dimension of spacetime . the highly accurate fitting formula for the coexistence curve is also presented .
using this fitting formula of the coexistence curve , we find that the clapeyron equations are highly consistent with the calculated values .
the fitting formula is also very useful for further study on the thermodynamic property of the system varying along the coexistence curve . |
distributions are subbundles of the tangent bundle of a manifold . it is natural not to consider general distributions but to make geometric assumptions , for example integrability . in this case
the distribution defines a foliation on the manifold .
another possibility is to assume that a distribution is nowhere integrable .
important examples of this type are contact structures on odd dimensional manifolds . on @xmath3dimensional manifolds , properties of contact structures reflect topological features of the underlying manifold in a surprising way .
engel structures form another class of non integrable distributions which is closely related to contact structures . by definition ,
an engel structure is a smooth distribution @xmath4 of rank @xmath5 on a manifold @xmath6 of dimension @xmath0 which satisfies the non integrability conditions @xmath7&=3 \\ \mathrm{rank}[{\mathcal { d}},[{\mathcal { d}},{\mathcal { d}}]]&=4~ , \end{aligned}\ ] ] where @xmath8 $ ] consists of those tangent vectors which can be obtained by taking commutators of local sections of @xmath4
. examples of engel structures can be obtained from contact structures on @xmath3manifolds . in this article
we use results from contact topology to investigate engel structures .
this shows that engel structures and contact structures are closely related . similarly to contact structures and symplectic structures , all engel structures are locally isomorphic .
every point of an engel manifold has a neighborhood with local coordinates @xmath9 such that the engel structure is the intersection of the kernels of the @xmath10forms @xmath11 this normal form was obtained first by f. engel in @xcite .
together with the fact that a @xmath12small perturbation of an engel structure is again an engel structure , this implies that engel structures are stable in the sense of singularity theory . in @xcite r. montgomery
classified all distribution types with this stability property .
they belong to one of the following classes : * foliations of rank one on manifolds of arbitrary dimension * contact structures on manifolds of odd dimension * even contact structures on manifolds of even dimension * engel structures on manifolds of dimension @xmath0
. thus engel structures are special among general distributions and even among the stable distribution types they seem to be exceptional since they are a peculiarity of dimension @xmath0 . on the other hand engel structures
appear quite naturally .
for example , a generic plane field on a @xmath0manifold satisfies the engel conditions almost everywhere and one can construct an engel structure from a contact structure on a @xmath3manifold .
certain non holonomic constraints studied in classical mechanics also lead to engel structures .
an engel structure @xmath4 induces a distribution of hyperplanes @xmath13 $ ] which is an even contact structure , i.e. @xmath14 $ ] is the whole tangent bundle .
moreover , to the even contact structure @xmath15 one can associate its characteristic foliation @xmath16 and it turns out that @xmath16 is tangent to @xmath4 .
this is explained in section [ s : engeldef ] .
thus an engel structure @xmath4 on @xmath6 induces a flag of distributions @xmath17\subset tm\ ] ] such that each distribution has corank one in the next one .
hypersurfaces transverse to @xmath16 carry a contact structure together with a legendrian line field .
we will refer to this line field as the intersection line field .
the presence of the distributions in leads to strong restrictions for the topology of engel manifolds . the following result can be found in @xcite where it is attributed to v. gershkovich .
[ p : trivial intro ] if an orientable @xmath0manifold admits an orientable engel structure , then it has trivial tangent bundle . we develop a construction of engel manifolds which allows us to prove the converse of proposition [ p : trivial intro ] .
[ t : exist intro ] every parallelizable @xmath0manifold admits an orientable engel structure .
this solves a problem from @xcite and @xcite . for open manifolds ,
theorem [ t : exist intro ] can be proved using the @xmath18principle for open , @xmath19invariant relations , cf .
@xcite . in the literature
one can find only two other constructions of engel structures on closed manifolds .
the first one is called prolongation and was introduced by e. cartan . with this method one
gets engel structures on certain @xmath20bundles over @xmath3dimensional contact manifolds .
the second construction is due to h.j .
geiges , cf .
it yields engel structures on parallelizable mapping tori .
these two constructions cover only a small portion of all parallelizable @xmath0manifolds .
our construction is based on decompositions of manifolds into round handles .
a round handle of dimension @xmath21 and index @xmath22 is @xmath23 it is attached to a manifold with boundary using an embedding of @xmath24 into the boundary of @xmath6 .
y a theorem of d. asimov in @xcite every @xmath0manifold with vanishing euler characteristic can be obtained from @xmath25 by attaching round handles of higher index successively .
we fix a set of model engel structures on round handles such that the characteristic foliation is transverse to @xmath26 and @xmath27 .
these model engel structures are constructed using a perturbed version of the prolongation construction .
the contact structure on each @xmath26 depends essentially only on @xmath28 while the intersection line field and the orientation of the contact structure are different for different model engel structures on @xmath29 . when we attach a round handle @xmath29 with a model engel structure to an engel manifold @xmath30 with transverse boundary we have to ensure that the attaching map preserves the oriented contact structures and the intersection line field if we want to extend the engel structure from @xmath30 to @xmath31 by the model engel structure .
in order to satisfy these conditions we isotope the attaching map and we choose a model engel structure suitably .
for this we use several constructions of contact topology .
it turns out that it is convenient to ensure that the contact structures on boundaries transverse to the characteristic foliation are overtwisted throughout the construction .
since @xmath31 is again an engel manifold with transverse boundary we can iterate this construction . for the construction of an engel structure on a manifold @xmath6 with trivial tangent bundle
we fix a round handle decomposition of @xmath6 and we use the model engel structures to extend the engel structure when the round handles are attached successively .
the condition that @xmath6 has trivial tangent bundle will be used to show that there is a model engel structure with the desired properties in our collection of model engel structures on @xmath29 for @xmath32 .
another result which can be obtained using round handles is the following theorem .
[ t : m+n intro ] let @xmath33 be manifolds with engel structures @xmath34 such that the characteristic foliations admit closed transversals @xmath35 .
then @xmath36 admits an engel structure which coincides with @xmath37 and @xmath38 away from a neighborhood of @xmath35 where all connected sums are performed .
the characteristic foliation of the new engel structure again admits a closed transversal . this theorem can be used to construct engel manifolds like for example @xmath39 for @xmath40 which are not covered by the construction of geiges or by prolongation .
this article is organized as follows .
section [ s : defengel ] contains several definitions and properties of engel structures and related distributions . in section [ ss : vert mod ] we introduce vertical modifications of the boundary of engel manifolds .
this construction will be used frequently in later sections .
our construction relies on several facts from @xmath3dimensional contact topology .
we summarize them in section [ s : contact top ] for the convenience of the reader .
the theorems discussed here will be used later in order to bring the attaching maps of round handles in a suitable form . in section [
s : round model ] we discuss round handle decompositions of manifolds and model engel structures on round handles .
the various isotopies of attaching maps for round handles and the choice of the right model engel structure are explained in section [ s : isotopy+choice ] .
we give the proofs of theorem [ t : exist intro ] and theorem [ t : m+n intro ] in section [ s : proofs ] . _
acknowledgements : _ this article contains the main results of the authors thesis .
it is a pleasure for me to thank my advisor dieter kotschick for his continuous support and help .
i would like to thank kai cieliebak , yakov eliashberg and paolo ghiggini for helpful discussions .
i am also grateful to the studienstiftung des deutschen volkes for their financial support .
[ s : defengel ] we first define all the distributions we will encounter in our constructions and the relations between them .
this will lead to a proof of proposition [ p : trivial intro ] . to every engel structure
one can associate a foliation of rank @xmath10 and hypersurfaces transverse to the this foliation carry a contact structure together with a legendrian line field . in section [ ss : vert mod ]
we explain vertical modifications of the boundary , a construction which will be used frequently .
contact structures and even contact structures arise naturally on engel manifolds .
they will play an important role in all our constructions .
here we summarize the definitions and elementary properties .
[ d : cont str ] a _ contact structure _
@xmath41 on a @xmath42dimensional manifold @xmath43 is a smooth subbundle of @xmath44 with corank @xmath10 such that for every local defining @xmath10form @xmath45 the rank of @xmath46 is maximal , or equivalently @xmath47 .
if @xmath21 is even the sign of @xmath48 is independent of the choice of @xmath45 .
thus a contact manifold of dimension @xmath49 has a preferred orientation . in dimension
@xmath3 the orientability of @xmath6 is the only obstruction to the existence of a contact structure by a result of j. martinet in @xcite .
the following two theorems will be used later .
we sketch the proofs because we will apply them in concrete situations .
first we explain gray s theorem which states a remarkable stability property of contact structures .
[ t : gray ] let @xmath50 $ ] be a family of contact structures on @xmath43 which is constant outside a compact subset of @xmath43 .
then there is an isotopy @xmath51 such that @xmath52 .
we assume that @xmath53 is globally defined by a smooth family of @xmath10forms @xmath54 .
the proof without this assumption is slightly more complicated , cf .
the desired isotopy is the flow of a time
dependent vector field @xmath55 .
this is the unique vector field which is tangent to @xmath56 and satisfies @xmath57 because @xmath58 is a non
degenerate @xmath5form on @xmath53 , such a vector field exists and is uniquely determined .
one can show by a direct calculation that the flow of @xmath55 has the desired properties .
an immediate consequence of gray s theorem is a normal form for the contact structure on neighborhoods of curves tangent to the contact structure , i.e. legendrian curves .
[ c : tube leg curve ] if @xmath59 is a closed legendrian curve in a @xmath3manifold with contact structure then we can choose coordinates @xmath60 on a tubular neighborhood of @xmath59 such that @xmath61 and the contact structure is defined by @xmath62 .
in particular , every point of a contact manifold admits a neighborhood with coordinates @xmath63 such that the contact structure is defined by @xmath64 .
next we consider vector fields which preserve a given contact structure , so called contact vector fields .
first recall that to each contact form @xmath45 one can associate a unique contact vector field @xmath65 , the _ reeb vector field _ , such that @xmath66 and @xmath67 .
[ p : convectfunctions ] the map which assigns to each contact vector field @xmath68 the function @xmath69 is an isomorphism between the space of contact vector fields and @xmath70 .
injectivity follows immediately from definition [ d : cont str ] .
now let @xmath71 be a smooth function on @xmath6 and @xmath65 the reeb vector field .
since @xmath72 is non degenerate there is a unique vector field @xmath73 tangent to @xmath41 such that @xmath74 .
then @xmath75 has the desired properties .
this proves surjectivity .
we will discuss several theorems from @xmath3dimensional contact topology in section [ s : contact top ] .
now we state the definition and some elementary properties of even contact structures .
[ d : evencontact ] let @xmath6 be a @xmath76dimensional manifold and @xmath15 a distribution on @xmath6 of corank one .
if for every local defining @xmath10form @xmath45 of @xmath15 the @xmath5form @xmath77 has maximal rank on @xmath15 , then @xmath15 is an even contact structure .
equivalently , @xmath15 is an even contact structure if for every local defining form @xmath45 , the @xmath42form @xmath48 has no zeroes . in dimension @xmath0
this condition may be rephrased as @xmath14=tm$ ] .
the definitions of contact structures and even contact structures are very similar .
still there are significant differences .
one of them is the presence of a distinguished line field on a manifold with an even contact structure .
since @xmath15 has rank @xmath78 , the rank of @xmath79 is @xmath80 .
hence @xmath79 has a kernel @xmath81 of dimension @xmath10 and it is easy to show that @xmath16 does not depend on the choice of the defining form @xmath45 .
all flows which are tangent to @xmath16 preserve @xmath15 .
[ d : charlinefield ] the line field @xmath16 is the _
characteristic line field _ of @xmath15 .
the foliation induced by this line field is called the _
characteristic foliation_. a hypersurface is _ transverse _ if it is transverse to the characteristic foliation .
the following lemma is a simple consequence of the definition of an even contact structures and its characteristic foliation .
[ l : transverse contact structure ] let @xmath15 be an even contact structure on @xmath6 and @xmath16 be the characteristic line foliation of @xmath15 . if @xmath43 is a transverse hypersurface , then @xmath82 is a contact structure on @xmath43 . if @xmath83 is another transversal such that two interior points @xmath84 and @xmath85 lie on the same leaf @xmath86 of the characteristic foliation , then the map obtained by following nearby leaves preserves the induced contact structures on neighborhoods of @xmath87 and @xmath88 as we have explained above a contact structure on a manifold of dimension @xmath78 induces an orientation of this manifold if @xmath21 is even .
in this situation an orientation of the characteristic foliation together with the contact orientation on a transversal define an orientation of @xmath6 .
[ p : even contact orientations ] let @xmath15 be an even contact structure on a @xmath89manifold @xmath6 .
then an orientation of @xmath6 induces an orientation of the characteristic line field @xmath16 and vice versa .
[ s : engeldef ] if @xmath4 is a distribution we denote the sheaf of all vectors which can be obtained as commutators of pairs of local sections of @xmath4 by @xmath8 $ ] . [
d : engel ] an engel structure is a distribution @xmath4 of rank two on a manifold @xmath6 of dimension four with the following properties .
* @xmath13\subset tm$ ] is a subbundle of rank @xmath3 .
* @xmath90 $ ] .
the second condition in definition [ d : engel ] implies that @xmath13 $ ] is an even contact structure .
to @xmath15 we associate the characteristic foliation @xmath16 of @xmath15 . by the definition of @xmath16 we have @xmath81 .
[ l : el ] if @xmath13 $ ] is an even contact structure induced by engel structure @xmath4 , then @xmath91 .
assume that @xmath86 is not contained in @xmath92 at @xmath93 .
let @xmath45 be a local defining form for @xmath15 on a neighborhood of @xmath87 .
if @xmath94 are two linearly independent local sections of @xmath4 around @xmath87 , then @xmath95 by the assumption on @xmath16 . on the other hand
@xmath96)=0 $ ] since @xmath97 $ ] is a local section of @xmath13 $ ] .
thus the assumption @xmath98 leads to a contradiction .
together with lemma [ l : transverse contact structure ] this allows us to give the following interpretation of the condition @xmath13 $ ] .
as one moves along the leaves of @xmath16 , the plane field @xmath4 rotates in @xmath15 around the characteristic foliation without stopping .
we have shown that an engel structure @xmath4 on @xmath6 induces a flag of distributions @xmath99 each of these distributions has corank one in the distribution containing it . by proposition [ p :
even contact orientations ] , an orientation of the characteristic foliation of an engel structure induces an orientation of the underlying manifold and vice versa .
in addition , @xmath13 $ ] is oriented by the following proposition .
[ p : e is oriented ] if @xmath4 is an engel structure , then the even contact structure @xmath13 $ ] has a distinguished orientation .
let @xmath94 be local sections of @xmath4 around @xmath93 such that @xmath100 and @xmath101 are linearly independent .
then @xmath102(p)$ ] is an orientation of @xmath103 which is independent of the choice of @xmath94 .
this leads to the following orientation conventions . *
if @xmath16 is oriented we orient @xmath6 by proposition [ p : even contact orientations ] . *
@xmath13 $ ] is oriented according to proposition [ p : e is oriented ] . *
transverse hypersurfaces are oriented by the induced contact structure . in particular transverse boundaries
are oriented by this convention . *
if @xmath16 is oriented , then we orient the contact structure on a transverse hypersurface such that the orientation of the contact structure followed by the orientation of @xmath16 is the orientation of the even contact structure .
let @xmath6 be an engel manifold with oriented characteristic foliation and transverse boundary .
we write @xmath104 for those connected components of @xmath105 where the characteristic foliation points out of @xmath6 .
the remaining boundary components are @xmath106 .
it is clear from that the existence of an engel structure has strong implications for the topology of the underlying manifold .
a proof of the following result can be found in @xcite .
it was known earlier by v. gershkovic .
[ p : trivial tm ] let @xmath4 be an oriented engel structure on an oriented @xmath0manifold @xmath6 .
then the tangent bundle of @xmath6 is trivial .
consider the flag @xmath107 of subbundles of @xmath108 . according to our orientation conventions each of the quotient bundles @xmath109
, @xmath110 and @xmath111 is oriented .
thus @xmath112 is isomorphic to the sum of four trivial line bundles . under the assumptions of proposition [ p : trivial tm ] we can choose oriented orthogonal complements in after having fixed an auxiliary riemannian metric and we obtain a trivialization of @xmath108 .
since we only chose the riemannian metric , this framing is well defined up to homotopy .
we will refer to such framings as _
engel framings_. a very simple example of an engel structure is the standard engel structure in the following darboux theorem for engel structures .
every point of an engel manifold has a neighborhood with coordinates @xmath113 such that the engel structure is given by @xmath114 here , the characteristic foliation is spanned by @xmath115 and the even contact structure is defined by @xmath116 .
we now discuss more interesting examples . apart from the construction we present later
, two other constructions of closed engel manifolds can be found in the literature .
[ e : prolong ] the starting point of the _ prolongation _ construction is a contact structure @xmath41 on a @xmath3manifold @xmath43 .
the projectivization @xmath117 of @xmath41 is a circle bundle over @xmath43 .
the projection @xmath118 of this fibration maps each legendrian line to its base point in @xmath43 .
one can show that @xmath119}{\mathbb { p}}{\mathcal { c}}\ : \big|\ : \operatorname{pr}_*(v)\in [ l ] \right\}\ ] ] is an engel structure whose characteristic foliation is tangent to the fibers of @xmath120 .
the associated even contact structure is @xmath121 .
similarly , one obtains _
orientable _ engel structures on the space of _ oriented _ legendrian lines .
the second previously known construction of closed engel manifolds is due to h.j
. geiges .
it yields engel structures on parallelizable mapping tori of compact @xmath3manifolds , cf .
@xcite .
[ s : int line field ] according to lemma [ l : transverse contact structure ] transverse hypersurfaces carry a natural contact structure . in this section
we discuss a legendrian line field induced by the engel structure on a transverse hypersurface .
this line field can be used to obtain a normal form for the engel structure near a transverse hypersurface @xmath43 .
the _ intersection line field _ of @xmath4 on @xmath43 is the legendrian line field @xmath122 .
if the engel structure @xmath4 is oriented , then we orient the intersection line field such that the orientation of @xmath16 followed by the orientation of the intersection line field is the orientation of @xmath4 .
the following theorem , as well as proposition [ p : engel diffeos ] , can be found in @xcite but according to that article they were known before .
both deal with engel structures obtained by prolongation .
if one applies the prolongation construction to the contact structure @xmath41 on a transverse hypersurface @xmath43 in an engel manifold , then one obtains the manifold @xmath117 with its canonical engel structure .
let @xmath123 be the section of @xmath117 which assigns to each point @xmath87 of @xmath43 the legendrian line @xmath124 .
[ t : normal forms for transversals ] any sufficiently small tubular neighborhood of @xmath43 in @xmath6 is canonically diffeomorphic as an engel manifold to a tubular neighborhood of @xmath123 in @xmath117 .
this means that the intersection line field and the contact structure @xmath41 on a transverse hypersurface @xmath43 suffice to recover the engel structure on a small neighborhood of @xmath43 .
another property of prolonged engel structures is that they admit many diffeomorphisms which preserve the engel structure , cf .
@xcite .
[ p : engel diffeos ] let @xmath125 be @xmath3manifolds with contact structures @xmath126 and let @xmath127 be a contact diffeomorphism .
the diffeomorphism between @xmath128 and @xmath129 which is induced by the action of @xmath130 on @xmath131 preserves the engel structures .
every diffeomorphism @xmath132 with this property is of this form .
we will mainly be concerned with the case that @xmath43 is the transverse boundary of a compact engel manifold . in the following section we will explain a modification of the engel structure on a collar of the boundary which allows us to vary the intersection line field within its homotopy class . therefore it suffices to control the homotopy class of the intersection line field and not the induced foliation . in the remaining part of this section we define simple invariants which determine the intersection line field on transverse boundaries of engel manifolds . for this
we assume that the characteristic foliation and the engel structure are oriented .
this induces orientations of the contact structure and the intersection line field on @xmath43 .
if we fix an oriented framing @xmath133 of the contact structure , we can assign an element of @xmath134 to a nowhere vanishing legendrian line field @xmath68 as follows .
there are unique functions @xmath135 such that @xmath136 .
the element of @xmath134 corresponding to @xmath68 assigns the winding number of @xmath137 around @xmath138 to the homology class represented by @xmath59 .
this depends only on the homotopy class of @xmath68 as nowhere vanishing legendrian line field .
in general we will not use a distinguished global framing but we will compare the homotopy classes of nowhere vanishing legendrian vector fields using rotation numbers along legendrian curves .
let @xmath68 be a nowhere vanishing legendrian vector field on @xmath43 and let @xmath94 be an oriented framing of the contact structure on @xmath43 . if @xmath59 is an immersed , closed legendrian curve , then @xmath139 is a legendrian vector field along @xmath59 and there are unique functions @xmath135 such that @xmath140 .
[ d : rotzahl ] the winding number of the map @xmath141 around zero is the _ rotation number _ of @xmath68 along @xmath59 . the sign of the rotation number changes if the orientation of @xmath59 , or if the orientation of the contact structure is reversed . in particular , its sign changes
if we change the orientation of the characteristic foliation of @xmath4 but it is independent of the orientation of @xmath4 . in order to determine the homotopy class of the intersection line field , it suffices to know the rotation number along sufficiently many legendrian curves .
[ ss : vert mod ] in this section we consider a manifold @xmath6 with oriented engel structure @xmath4 with oriented characteristic foliation such that @xmath105 is transverse and compact .
vertical modifications of the boundary allow us to change the intersection line field on @xmath105 . in this construction
we modify the engel structure near the boundary such that the even contact structure remains unchanged .
assume that @xmath142 is a legendrian line field which is homotopic to the intersection line field @xmath143 of @xmath4 .
then we can modify @xmath4 such that the intersection line field of the new engel structure is @xmath142 .
this can be generalized to the case of a transverse hypersurface @xmath43 , but then the possible changes of the intersection line fields depend on the behavior of the engel structure along the leaves of the characteristic foliation through @xmath43
. this problem will arise only in theorem [ t : m+n zweite vers .
] at the end of this article .
let @xmath144 be a legendrian vector field which spans and orients the intersection line field on @xmath105 .
assume that the legendrian vector field @xmath145 on @xmath105 is homotopic to @xmath144 through a family @xmath146 $ ] of nowhere vanishing legendrian vector fields .
our aim is to construct an engel structure on @xmath6 such that the intersection line field on @xmath105 is spanned and oriented by @xmath145 without changing the even contact structure .
we treat the boundary components @xmath104 where @xmath16 points out of @xmath6 .
the components @xmath147 can be treated similarly .
choose a collar @xmath148 $ ] of @xmath104 such that the second factor with its standard orientation corresponds to the oriented characteristic foliation of @xmath4 .
we write @xmath149 for the coordinate on @xmath150 $ ] .
furthermore let @xmath151 be a vector field such that @xmath152 is an oriented trivialization of the contact structure on @xmath104 .
the even contact structure on @xmath153 is spanned and oriented by @xmath154 .
there is a unique smooth function @xmath155{\longrightarrow}{\mathbb { r}}$ ] which vanishes along @xmath104 such that @xmath115 and @xmath156 span and orient @xmath157 . because @xmath4 is an engel structure @xmath158 $ ] and @xmath115 are linearly independent everywhere
this implies that @xmath71 is strictly monotone along the leaves of @xmath16 . by our orientation convention for the contact structure on transverse boundaries
, @xmath71 is strictly increasing .
there is a unique family of real functions @xmath159 such that @xmath160 is a positive multiple of @xmath161 and @xmath162 . since @xmath104 is compact we can choose @xmath163 such that @xmath164 is at least @xmath165 .
now extend @xmath71 from @xmath166 $ ] to a smooth function @xmath167 on @xmath168 $ ] such that @xmath167 is strictly increasing and coincides with @xmath164 along @xmath169 .
we extend @xmath4 from @xmath6 to a distribution on @xmath170 $ ] by the span of the vector field @xmath115 and @xmath171 the fact that @xmath172 implies that @xmath68 and @xmath173 $ ] are linearly independent .
moreover , @xmath115 together with @xmath158 $ ] , respectively the horizontal lifts of @xmath152 , span the same even contact structure . thus the extended distribution is an engel structure whose intersection line field on @xmath169 is spanned by @xmath145 .
using a flow tangent to the characteristic foliation one can identify @xmath6 and @xmath174 $ ] such that the even contact structures are preserved .
the modifications of an engel manifold with boundary described above will be called a _ vertical modifications of the boundary_.
let us summarize the discussion in the following theorem .
[ t : inters fol ] let @xmath175 be an oriented engel manifold such that @xmath105 is transverse and compact .
assume that the legendrian line field @xmath142 on @xmath104 is homotopic to the intersection line field @xmath143 of @xmath4 .
then @xmath142 is the intersection line field of a suitable vertical modification of @xmath104 .
there is an analogous construction in the context of contact structures .
let @xmath43 be a compact @xmath3manifold carrying a contact structure @xmath41 and a contact vector field @xmath176 transverse to the boundary .
for this , let @xmath68 be a nowhere vanishing legendrian vector field such that @xmath177 $ ] and @xmath68 are linearly independent .
assume that @xmath145 is a legendrian vector field along @xmath178 which is homotopic to the restriction of @xmath68 to @xmath178 through nowhere vanishing legendrian vector fields .
then we can modify @xmath68 near the boundary of @xmath43 such that the resulting legendrian vector field @xmath179 coincides with @xmath145 along @xmath180 and @xmath181 $ ] are linearly independent .
[ r : suitable vert mod 1 ] in our applications the legendrian vector field @xmath145 is given only on a compact embedded submanifold @xmath182 .
we then pick a positive function @xmath164 as above and extend it to a positive function @xmath183 which equals @xmath165 outside of a small neighborhood of @xmath153 .
if @xmath153 is not connected we can choose different values for @xmath184 on each connected component of @xmath153 .
[ s : contact top ] in this section we summarize facts from the theory of contact structures . section [ s : stabilization ] deals with legendrian curves in contact manifolds .
we discuss in particular stabilization of legendrian curves .
this will play an important role in the construction of attaching maps for round handles of index @xmath10 carrying model engel structures . in section [ s : contact facts ]
we collect those facts about convex surfaces in contact manifolds which we will use in section [ s : bypasses ] for the construction of bypasses in overtwisted contact manifolds as well as in section [ s : modelle index 2 ] and section [ ss : model 3-handles ] for the construction of model engel structures on round handles of index @xmath5 and @xmath3 .
most of the material presented in this section ( with the exception of the construction of bypasses in overtwisted contact manifolds in section [ s : bypasses ] ) can be found in @xcite .
[ s : stabilization ] throughout this section we consider an oriented contact structure @xmath41 on a @xmath3manifold @xmath43 .
a curve is _ legendrian _ if it is tangent to the contact structure .
[ d : stand framing ] a framing @xmath185 of an embedded closed legendrian curve @xmath59 is an _ oriented contact framing _ if * @xmath186 is tangent to the contact structure and @xmath187 represents the orientation of the contact structure and * @xmath188 is transverse to the contact structure and @xmath189 represents the orientation induced by the contact structure .
every legendrian curve has a contact framing which is well defined up to homotopy .
the homotopy class of framings of @xmath59 represented by contact framings is denoted by @xmath190 . on the set of framings of @xmath59
there is a @xmath191action defined by @xmath192 this action is transitive on the set of homotopy classes of framings such that @xmath139 followed by the framing induces the contact orientation . when we reverse the orientation of @xmath59 the coorientation of @xmath59 changes .
therefore the @xmath191action on the framings does not depend on the orientation of @xmath59 .
contact framings are related to the thurston
bennequin invariant of null homologous legendrian knots .
[ d : tb ] let @xmath59 be a null homologous legendrian curve in @xmath43 .
fix a relative homology class @xmath193\in h_2(n,\gamma;{\mathbb { z}})$ ] which is represented by an oriented surface @xmath194 such that @xmath195 and @xmath59 is oriented as the boundary of @xmath194 . a new curve @xmath196 is obtained by pushing @xmath59 slightly along a vector field which is transverse to @xmath41 .
bennequin invariant _
@xmath197)$ ] is the homological intersection number of @xmath196 with @xmath194 . in definition
[ d : tb ] the surface @xmath194 singles out those framings of @xmath198 whose first component is tangent to @xmath194 . we write @xmath199 for the corresponding homotopy class of framings . the thurston
bennequin invariant measures the difference between @xmath199 and @xmath190 of @xmath59 @xmath200)\cdot \operatorname{fr}_\sigma(\gamma ) = \operatorname{fr}(\gamma)~.\ ] ] a simple application of gray s theorem ( theorem [ t : gray ] ) shows that the contact framing of a legendrian curve determines the isotopy type of the contact structure on a neighborhood of the legendrian curve .
[ l : tub nbhd of leg curves ] let @xmath201 be two contact structures on @xmath43 which induce the same orientation of @xmath6 and @xmath59 a curve tangent to both @xmath202 and @xmath131 . if the contact framings of @xmath59 with respect to @xmath202 and @xmath131 are homotopic , then there is an isotopy @xmath203,$ ] of @xmath43 relative @xmath59 such that @xmath204 on a tubular neighborhood of @xmath59 .
if @xmath202 and @xmath131 are oriented , then one can choose @xmath51 such that @xmath205 preserves also the orientations of the contact structures .
now we discuss the other classical invariant of null homologous legendrian curves .
[ d : rot zahl klassisch ] let @xmath194 be a connected orientable surface with @xmath206 .
fix an oriented trivialization @xmath94 of @xmath41 along @xmath194 .
there are unique functions @xmath207 such that @xmath208 .
the winding number of @xmath209 around @xmath138 is the _ rotation number _ @xmath210)$ ] .
[ r : rotzahlvergleich ] in definition [ d : rotzahl ] we defined the rotation number for legendrian curves on transverse hypersurfaces in oriented manifolds with oriented engel structures . in definition
[ d : rot zahl klassisch ] we fix an oriented trivialization of the oriented contact structure on a seifert surface @xmath194 of the legendrian knot @xmath206 and compare @xmath139 with this trivialization .
now assume that @xmath41 is the oriented contact structure on a transverse hypersurface of an oriented manifold with oriented engel structure . in this case
we can use the intersection line field as the first component of the trivialization of @xmath41 over @xmath194 .
thus in this situation the two rotation numbers in definition [ d : rotzahl ] and definition [ d : rot zahl klassisch ] are equivalent . using
contact framings and rotation numbers one can distinguish isotopy classes of legendrian curves .
this isotopy type can be changed using the following construction .
let @xmath59 be a closed legendrian curve in a contact manifold .
according to corollary [ c : tube leg curve ] there are coordinates @xmath60 on a neighborhood of @xmath59 such that the contact structure is defined by @xmath62 and @xmath211 .
the @xmath212axis in both parts of figure [ b : stab ] represents the projection of a neighborhood of a segment of @xmath59 to the @xmath213plane and @xmath214 is pointing outwards .
let @xmath123 be another smooth curve in the @xmath215plane that coincides with the @xmath212axis near its endpoints .
since the contact structure near @xmath59 is defined by @xmath62 the segment @xmath123 lifts to a legendrian curve whose endpoints lie on @xmath59 if and only if the integral of the @xmath10form @xmath216 along @xmath123 is zero
. then the lifts of curves @xmath217 , respectively @xmath218 as in figure [ b : stab ] , yield new closed legendrian curves which are close to @xmath59 .
they represent the positive and the stabilization of @xmath59 .
, scaledwidth=96.0% ] up to isotopy through legendrian curves the stabilized curves do not depend on the choices made above .
in particular , positive and negative stabilization commute up to legendrian isotopy .
the effect of stabilization on the rotation number with respect to a nowhere vanishing legendrian vector field @xmath68 is given by @xmath219 hence @xmath220 and @xmath59 are pairwise non isotopic as legendrian curves . from the construction of @xmath221
it follows that @xmath222 from it follows that this is coherent with the well known formula @xmath223 for null homologous legendrian curves .
[ s : contact facts ] let @xmath43 be a @xmath3manifold with an oriented contact structure @xmath41 . consider a properly embedded oriented surface @xmath194 .
the contact structure induces the _ singular foliation _
@xmath224 on @xmath194 .
usually this is called the characteristic foliation of @xmath194 .
we avoid this terminology since there is already a characteristic ( non singular ) foliation in the context of engel structures .
the singularities of @xmath225 are those points @xmath226 where @xmath227 . from the orientations of @xmath194 and @xmath41 we obtain an orientation of @xmath225 away from the singularities using the following convention : for each non singular point @xmath87 of @xmath225 we choose @xmath228 such that @xmath229 is the contact orientation , @xmath230 orients @xmath194 and @xmath231 orients @xmath41 .
then @xmath232 represents the orientation of @xmath233 .
generically , singular points are non degenerate .
we say that a singular point is _ elliptic _ if its index is @xmath234 and _ hyperbolic _ if the index is @xmath235 .
when the orientation of @xmath41 and the orientation of the surface coincide at a singular point of @xmath225 , we say that this singularity is _ positive _ , otherwise it is _
negative_. if we orient @xmath225 according to our conventions , positive elliptic points are sources and negative elliptic points are sinks .
@xmath194 is called _ convex _ if there is a contact vector field transverse to @xmath194 .
giroux studied closed convex surfaces in @xcite . if @xmath194 has boundary we will usually assume that @xmath236 is legendrian . in particular he showed that a closed embedded surface is generically convex ( with respect to the @xmath237topology ) .
the analogous statement is in general not true when @xmath194 has boundary ( even when @xmath238 is legendrian ) .
for each boundary component @xmath239 let @xmath240 be the unique integer such that @xmath241 if @xmath59 is a legendrian knot and @xmath194 is a seifert surface for @xmath59 , then @xmath240 is the thurston
bennequin invariant by .
[ p : convexity with boundary ] let @xmath194 be a compact , oriented , properly embedded surface with legendrian boundary , and assume @xmath242 for all boundary components @xmath59 of @xmath194 .
there exists a @xmath243small perturbation near the boundary ( fixing @xmath236 ) which puts an annular neighborhood @xmath244 of @xmath238 into a standard form , and a subsequent @xmath237small perturbation keeping @xmath244 fixed which makes @xmath194 convex .
moreover , if @xmath176 is a contact vector field defined on a neighborhood of @xmath244 and transverse to @xmath245 , then @xmath176 can be extended to a contact vector field transverse to all of @xmath194 .
the singular foliation is enough to determine the contact structure on a small neighborhood of a convex surface .
[ t : giroux unique ] let @xmath194 be a compact orientable convex surface with legendrian boundary .
two @xmath246invariant contact structures on @xmath247 which induce the same orientation and the same singular foliation on @xmath248 are isotopic .
they are conjugate by a diffeomorphism @xmath249 , and @xmath130 is isotopic to the identity through diffeomorphisms of @xmath194 that preserve the singular foliation . for a convex surface @xmath194 with legendrian boundary
we fix a contact vector field @xmath176 transverse to @xmath194 .
[ d : dividing set ] the _ dividing set _
@xmath250 of @xmath194 is the set of points where @xmath176 is tangent to the contact structure .
the dividing set contains much information about the contact structure near @xmath194 .
giroux showed in @xcite that @xmath250 is a submanifold of @xmath194 which is transverse to the singular foliation .
its isotopy class depends only on @xmath194 and @xmath41 but not on @xmath176 .
let @xmath225 be a singular foliation on @xmath194 such that @xmath236 is tangent to @xmath225 .
a collection @xmath251 of closed curves and arcs with end points on @xmath236 is said to _ divide _
@xmath225 if on each connected component of the closure of @xmath252 there is a smooth volume form @xmath253 and a vector field @xmath68 tangent to @xmath225 such that * the divergence of @xmath68 with respect to @xmath253 is positive everywhere * @xmath68 points outwards along those parts of the boundary which correspond to @xmath254 .
[ t : divide impera ] if @xmath194 is a convex surface with legendrian boundary , then @xmath250 divides the singular foliation on @xmath194 .
conversely , let @xmath225 be a singular foliation on a compact oriented surface @xmath194 such that @xmath236 is tangent to @xmath225 and @xmath254 divides @xmath225 .
then there is a positive @xmath246invariant contact structure on @xmath247 such that @xmath248 is convex , the induced singular foliation on @xmath248 is precisely @xmath225 and such that @xmath254 consists of those points where the contact structure is tangent to the second factor in @xmath247 .
this theorem implies in particular that the dividing set of a convex surface with legendrian boundary is not empty .
assume that @xmath41 is cooriented by @xmath45 and that the orientation @xmath194 followed by @xmath176 is the contact orientation .
then the dividing set @xmath254 separates the region @xmath255 where @xmath256 is positive from the region @xmath257 where @xmath256 is negative .
if @xmath194 is closed , then the euler characteristic @xmath258 of @xmath41 viewed as oriented bundle satisfies @xmath259 if @xmath194 is a seifert surface for a legendrian knot @xmath59 , then one can determine the thurston
bennequin invariant and the rotation number of @xmath59 using the following formulas from @xcite @xmath260 next we consider deformations of the singular foliation .
let @xmath194 be a convex surface with legendrian boundary and fix a transverse contact vector field @xmath176 .
an isotopy @xmath261 of a surface @xmath194 is called _ admissible _ if @xmath262 is transverse to @xmath176 for all @xmath263 .
the following theorem is a generalization of the giroux flexibility theorem in @xcite where @xmath194 is assumed to be closed .
[ t : giroux flex ] let @xmath264 be the singular foliation on @xmath194 induced by the contact structure and @xmath265 a singular foliation which is divided by @xmath250 . then there is an admissible isotopy @xmath266 $ ] , of @xmath194 such that @xmath265 is the singular foliation on @xmath267 . in this example
we want to fix some terminology .
consider the @xmath246invariant contact form @xmath268 on @xmath269 where @xmath270 is the coordinate on the @xmath246factor while @xmath130 and @xmath212 correspond to @xmath271 .
we say that the singular foliation @xmath225 on @xmath272 is in _
standard form_. the singularities of the singular foliation form two circles @xmath273 .
these are called the _
legendrian divides_. the dividing set of @xmath272 has two connected components @xmath274 the curves tangent to @xmath275 form the _
legendrian ruling_. by theorem [ t : giroux flex ] , the legendrian ruling can be changed into any foliation by straight lines as long as the straight lines remain transverse to the dividing set .
using theorem [ t : giroux flex ] we can isotope @xmath194 such that a submanifold with certain properties becomes legendrian .
a union @xmath276 of disjoint properly embedded arcs and closed curves on @xmath194 is called _ non isolating _ if * @xmath276 is transverse to @xmath254 and every arc begins and ends on @xmath254 * the boundary of each component of @xmath277 contains a segment of @xmath254 .
[ t : lerp ] consider a collection @xmath276 of properly embedded closed curves and arcs on a convex surface @xmath194 with legendrian boundary .
if @xmath276 is non isolating there exists a @xmath243small admissible isotopy @xmath278 $ ] such that * @xmath279 and @xmath280 * @xmath281 is legendrian .
we finish this section with an important dichotomy of contact structures on @xmath3manifolds .
an embedded disc with legendrian boundary is _ overtwisted disc _ if all singularities on the boundary have the same sign .
a contact structure is called _ overtwisted _ if it there is an overtwisted disc .
a contact structure is _ tight _ if it is not overtwisted .
the following theorem shows that tight contact structures are more interesting than overtwisted ones in many aspects .
[ t : eliashbergs wahnsinn ] if two overtwisted contact structures on a closed manifold are homotopic as plane fields and induce the same orientation , then they are isotopic . for our purposes however
, the flexibility of overtwisted contact structures will be very useful .
regions where a given contact structure is tight can be found using the following theorem .
[ t : giroux tight ] if @xmath282 is a convex surface ( closed or compact with legendrian boundary ) in a contact manifold @xmath283 , then @xmath194 has a tight neighborhood if and only if the dividing set of @xmath194 has no homotopically trivial closed curves . if @xmath284 , then @xmath194 has a tight neighborhood if and only if @xmath250 is connected . [
s : bypasses ] consider a contact manifold @xmath285 and a convex surface @xmath286 which has empty or legendrian boundary .
let @xmath287 be a dividing set of @xmath194 .
[ d : bypasses ] a _ bypass _ for @xmath194 is an embedded half disc @xmath288 in @xmath43
whose singular foliation has the following properties : * @xmath289 is the union of two legendrian arcs @xmath290 which intersect at their endpoints .
* @xmath288 intersects @xmath194 transversally along @xmath291 .
there are no other intersection points .
* @xmath288 admits an orientation such that * * there are exactly two positive singularities along @xmath291 .
these are the endpoints of @xmath291 .
they are elliptic . * * there is exactly one negative singularity on @xmath291 .
it is elliptic . *
* there are only positive singularities along @xmath292 .
they alternate between elliptic and hyperbolic .
* @xmath291 intersects @xmath250 in exactly three points .
the intersections are transverse and correspond to the singularities of the singular foliation on @xmath288 along @xmath291 . *
the dividing set of @xmath288 has exactly one connected component .
a bypass for @xmath194 allows us to isotope @xmath194 such that the isotopy type of the dividing set changes as in lemma [ l : attach ] .
they where introduced by k. honda in @xcite .
requirement ( v ) in definition [ d : bypasses ] does not appear in @xcite since all contact structures considered in that article are tight .
in this situation , the dividing set @xmath293 of @xmath288 is determined ( up to isotopy ) by ( i)(iv ) while this is not true for overtwisted contact structures .
requirement ( v ) is necessary for the following bypass attachment lemma .
[ l : attach ] let @xmath288 be a bypass for a convex surface @xmath194 .
there exists a neighborhood of @xmath294 which is diffeomorphic to @xmath295 $ ] such that @xmath296 is convex for @xmath297 .
the dividing set of @xmath298 can be obtained from the dividing set of @xmath248 as in figure [ b : bypass ] .
( the bypass is attached to the front .
figure [ b : bypass ] represents only a neighborhood of the attaching region of @xmath288 . ) ] gluing together two bypasses along the boundary component containing the singularities with alternating signs one obtains a disc with a singular foliation like the singular foliation on an overtwisted disc .
thus one can think of a bypass as one half of an overtwisted disc and it is not surprising that it is much easier to find bypasses if the contact structure is overtwisted rather than tight .
in the following proposition we assume that @xmath291 is a legendrian arc contained in the convex surface @xmath194 .
since every arc @xmath299 which is transverse to @xmath254 and intersects @xmath254 as @xmath291 in proposition [ p : bypass existence ] is non isolating , we can apply theorem [ t : lerp ] to find a @xmath243 small admissible isotopy of @xmath194 such that @xmath299 becomes legendrian , i.e. the assumption that @xmath291 is legendrian is actually not restrictive .
[ p : bypass existence ] let @xmath194 be a convex surface in a contact manifold and @xmath300 an overtwisted disc disjoint from @xmath194 .
assume that @xmath301 is an embedded legendrian arc such that the endpoints lie on @xmath254 and @xmath302 contains three points .
then there is a bypass for @xmath194 which intersects @xmath194 along the @xmath291 .
note that @xmath291 is automatically transverse to @xmath254 .
let @xmath176 be a contact vector field transverse to @xmath194 .
consider the image @xmath65 of @xmath291 under the flow @xmath303 of @xmath176 for @xmath304 such that @xmath305 .
the curves @xmath306 are legendrian since @xmath303 is the flow of a contact vector field . if @xmath307 , then the segment @xmath308 is legendrian .
if @xmath309 are points in different components of @xmath310 , then the union @xmath311 of the curves @xmath312 with @xmath313 divide the singular foliation on @xmath65 . since @xmath65 has legendrian boundary , theorem [ t : divide impera ] implies that @xmath65 is convex .
we orient @xmath65 such that the singularities at the endpoints of @xmath291 are positive .
by @xmath314 the idea is to perform a legendrian connected sum of the knots @xmath315 and @xmath316 together with a boundary connected sum of @xmath65 and @xmath300 .
doing this carefully enough , one obtains a bypass .
let us first explain the legendrian connected sum of null homologous legendrian knots @xmath317 in a contact manifold , cf .
@xcite .
let @xmath318 and @xmath319 .
choose a legendrian curve @xmath59 between @xmath320 and @xmath321 such that @xmath59 does not meet the knots anywhere else and @xmath59 is not tangent to @xmath322 in @xmath323 for @xmath324 . by corollary [ c : tube leg curve ]
there are coordinates @xmath60 on a tubular neighborhood of @xmath59 such that @xmath59 corresponds to the @xmath212axis and the contact structure is defined by @xmath62 . after a @xmath243small legendrian isotopy of @xmath325 and @xmath326 we may assume that @xmath327 are cusp points of the front projection , i.e the projection to the @xmath328plane , and that the knots are oriented as in figure [ b : confront ] .
the legendrian connected sum @xmath329 is then formed using the dashed curves in figure [ b : confront ] . ]
now consider seifert surfaces @xmath330 of @xmath325 and @xmath331 of @xmath326 .
we assume that @xmath330 , respectively @xmath331 , coincides with translates of @xmath325 in the negative @xmath212direction , respectively of @xmath326 in the positive @xmath212direction , on a neighborhood of @xmath320 , respectively @xmath321 .
such neighborhoods are shown in figure [ b : confront ] .
if we orient @xmath330 and @xmath331 such that @xmath325 and @xmath326 are oriented as boundaries , then @xmath320 is a negative singularity and @xmath321 is a positive singularity .
we use the ribbon which is bounded by the dashed curves in figure [ b : confront ] to construct a seifert surface @xmath332 for the knot @xmath329 . according to @xcite @xmath333 ) & = \operatorname{tb}(k_1,[\sigma_1 ] ) + \operatorname{tb}(k_2,[\sigma_2 ] ) + 1 \\ \label{e : consumrot } \operatorname{rot}(k_1\ # k_2,[\sigma_1\#\sigma_2 ] ) & = \operatorname{rot}(k_1,[\sigma_1 ] ) + \operatorname{rot}(k_2,[\sigma_2])~.\end{aligned}\ ] ] now we apply this to the seifert surfaces @xmath334 of @xmath335 and @xmath336 of @xmath337 where @xmath300 is oriented such that @xmath338 and @xmath339
. then we perturb @xmath340 to a convex surface with boundary @xmath341 . the difficulty in showing that the perturbed surface is a bypass is to establish ( v ) of definition [ d : bypasses ] . in order to do this
we reduce the region where the legendrian connected sum is performed such that the contact structure on this region is tight . by theorem [ t : lerp ] and theorem [ t : giroux
flex ] we can assume that the singular foliation on @xmath300 is of the form indicated in figure [ b : otdisc+ ] where the thickened circle is the dividing set .
then we can decompose @xmath300 into two half
discs bounded by legendrian unknots with thurston bennequin invariant @xmath235 and rotation number @xmath138 . the half
discs are separated by straight legendrian arcs .
the singular foliation near the unknots is in the standard form used in proposition [ p : convexity with boundary ] . ]
by the last part of proposition [ p : convexity with boundary ] we can pretend that we form a boundary connected sum of the surfaces @xmath65 with the left part @xmath342 of @xmath300 and that the perturbation of @xmath340 only changes the half disc @xmath342 .
the presence of the legendrian curves in the middle of @xmath300 prevents an interaction between the left and the right part of @xmath300 .
the union of tubular neighborhoods of @xmath65 , of the legendrian arc connecting @xmath65 with @xmath300 and of the left part @xmath342 of @xmath300 can be recovered in tight contact manifolds : @xmath342 can be obtained applying theorem [ t : giroux flex ] to a bypass in a tight contact manifold , hence a neighborhood of @xmath343 is tight .
the thurston
bennequin invariant of @xmath344 is @xmath345 by . from and theorem [ t : giroux tight ]
it follows that the dividing set on @xmath346 ( after this surface is perturbed to a convex surface ) consists of exactly two arcs with endpoints on @xmath347 and no closed components .
the notations @xmath346 and @xmath340 are slightly misleading because @xmath342 , respectively @xmath300 , is _ not _ a subset of @xmath346 , respectively @xmath340 , after these surfaces are smoothed and made convex .
we are left with the two possibilities for the isotopy type of the dividing set of @xmath340 shown in figure [ b : divbypass ] . ] according to the boundary of the left part of figure [ b : divbypass ] has rotation number @xmath10 while in the right part its rotation number is @xmath345 . by the rotation number of @xmath348
is @xmath10 .
this shows that the dividing set on the seifert surface of @xmath348 contains no closed components . the remaining conditions ( i ) , ( ii ) and ( iv ) in definition [ d : bypasses ] are satisfied by construction .
condition ( iii ) can be achieved using theorem [ t : giroux flex ] .
thus @xmath340 yields a bypass .
[ s : round model ] this section provides the building blocks for our construction of engel manifolds .
we first recall several facts from @xcite about round handle decompositions . in section [ s : perturbed prolong ] we describe a perturbed version of the prolongation construction and first examples of engel structures which are compatible with a round handle decomposition of the underlying manifold .
perturbed prolongation will be used for the construction of model engel structures on round handles in sections [ s : index zero models ] to [ ss : model 3-handles ] .
round handles were used by d. asimov ( @xcite ) for the study of flow manifolds .
a flow manifold is a manifold @xmath6 with a non singular vector field @xmath349 transverse to the boundary which points outwards along @xmath104 and inwards along @xmath350 .
[ d : rk ] @xmath351 is a _ round handle _ of dimension @xmath21 and index @xmath22 .
the boundary @xmath352 is the union of the two subsets @xmath353 a round handle is attached to a flow manifold using an embedding @xmath354 .
one can extend @xmath349 to a non singular vector field on @xmath355 which is again transverse to the boundary .
this is impossible if one attaches an ordinary handle @xmath356 of index @xmath357 to @xmath6 since this changes the euler characteristic by @xmath358
. now assume that @xmath349 spans the characteristic foliation of an engel structure on @xmath6 .
if one extends the engel structure from @xmath6 to @xmath355 it is clear from the properties of transverse hypersurfaces that it is useful to ensure that the boundary of the new engel manifold is again transverse to the characteristic foliation . therefore round
handles are suitable building blocks for the construction of engel structures .
[ d : round decomp ] if @xmath6 is obtained from the disjoint union of finitely many round handles of index @xmath138 by attaching round handles of higher index successively , i.e. @xmath359 with @xmath360 for @xmath361 , then we have a _ round handle decomposition _ of @xmath6 .
we can rearrange a given round handle decomposition of a manifold such that the round handles are ordered according to their index .
contrary to the case of ordinary handles , two round handles of the same index can not be interchanged in general .
if a closed manifold @xmath6 admits a round handle decomposition , then its euler characteristic @xmath362 has to vanish because we can use the round handle decomposition to find a non singular vector field on @xmath6 .
the converse direction is covered by the following theorem .
[ t : round decomp euler 0 ] a closed , connected manifold of dimension @xmath363 admits a decomposition into round handles if and only if @xmath364 .
this decomposition can be chosen such that there is only one round @xmath138handle and one round @xmath365handle .
the analogous statement is wrong for manifolds of dimension @xmath3 , cf .
@xcite . in order to prove the last part of theorem [ t : round decomp euler 0 ] one follows the proof in @xcite starting with a decomposition of @xmath6 into ordinary handles with only one handle of index @xmath138 , respectively @xmath21
. one can decompose @xmath29 into one ordinary handle of index @xmath28 and another one of index @xmath366 as follows @xmath367 this allows us to obtain decompositions into ordinary handles from round handle decompositions .
[ s : perturbed prolong ] from now on we consider only @xmath0dimensional round handles . for the construction of engel structures we will fix a set of particular engel structures on @xmath29 for each @xmath368 . a model engel structure on a round handle of index @xmath28 is an engel structure on @xmath29 such that the characteristic foliation is oriented and transverse to @xmath369 and @xmath26 .
it points outwards along @xmath369 and inwards along @xmath26 .
the usual prolongation construction is described in example [ e : prolong ] and deformations of certain prolonged engel structures are discussed in @xcite .
we now describe perturbations of engel structures which we will use later for the explicit construction of model engel structures .
let @xmath45 be a contact form on a compact manifold @xmath43 such that there is a contact vector field @xmath176 for @xmath370 which is transverse to @xmath178 .
moreover we assume that @xmath41 is trivial as a bundle and we fix a trivialization @xmath133 of @xmath41 .
we use the same notation for the horizontal lifts to @xmath371 .
[ p : perturbed prolong ] for @xmath163 the distribution @xmath372 on @xmath371 spanned by @xmath373 is an engel structure on @xmath371 for @xmath374 small enough and @xmath375 .
for @xmath376 we obtain an engel structure if and only if @xmath377 and @xmath378 $ ] are linearly independent everywhere .
then the characteristic foliation of @xmath372 is spanned by @xmath349 . in particular , the boundary is transverse and both @xmath41 and the contact structure on @xmath379 induce the same singular foliation on @xmath380 . in order to show that @xmath372 is an engel structure we first calculate @xmath381 & = -k\sin(kt)c_1 + k\cos(kt)c_2 \\ & \quad + { \varepsilon}(\cos(kt)[v , c_1]+\sin(kt)[v , c_2])~. \end{split}\end{aligned}\ ] ] if @xmath375 and @xmath374 is small enough , then @xmath382 and @xmath383 $ ] are linearly independent .
the case @xmath376 is obvious .
since @xmath176 is a contact vector field @xmath384 $ ] is generated by @xmath385 and @xmath386 .
hence @xmath387 $ ] is defined by @xmath388 from the fact that @xmath45 is a contact form it follows that @xmath389 defines an even contact structure for all @xmath374 .
this shows that @xmath372 is an engel structure .
notice that @xmath390 .
let @xmath18 be the function with the property @xmath391 .
then @xmath392 implies that @xmath349 spans the characteristic foliation of @xmath372 . if @xmath393 , then we obtain a model engel structure on @xmath394 if @xmath176 points outwards along @xmath395 and inwards along @xmath396 .
let us describe the orientation of @xmath387 $ ] : if @xmath376 we obtain an engel structure only if @xmath377 and @xmath378 $ ] are linearly independent and then @xmath15 is oriented by @xmath397 $ ] . for @xmath398 the even contact structure @xmath15 is oriented by @xmath399 . if @xmath400 , then @xmath15 has the opposite orientation . if @xmath401 lies on the characteristic surface @xmath402 of @xmath176 , then @xmath403 is tangent to the contact structure on @xmath379 . for later use
we determine the rotation number along these legendrian curves . if @xmath376 the rotation number is zero since the intersection line field along @xmath379 is @xmath20invariant .
for @xmath375 we may assume that @xmath404 at @xmath87 .
clearly @xmath405 + r_{\varepsilon}\ ] ] where the correction term @xmath406 satisfies @xmath407 .
if we project all vector fields in along @xmath349 to @xmath379 we obtain an analogous expression for @xmath408 .
notice that the projection of @xmath409 to @xmath379 spans the intersection line field .
hence the rotation number along @xmath403 is @xmath410 ( cf .
definition [ d : rotzahl ] ) . in order to obtain model engel structures with positive rotation number along @xmath403 one
can replace @xmath349 by @xmath411 , but doing so one changes the contact structure on @xmath379 . in the cases
@xmath393 for @xmath412 considered later we will construct a diffeomorphism @xmath71 of @xmath394 which preserves the contact structure on large parts of @xmath26 and reverses the orientation of @xmath403 . pushing forward the model engel structures with @xmath71 one obtains model engel structures with non negative rotation numbers along @xmath403 .
before we discuss explicit model engel structures let us first describe examples of engel manifolds with a decomposition into round handles which carry model engel structures .
if @xmath43 is decomposed into ordinary handles we obtain a round handle decomposition of @xmath371 such that the round handles are products of @xmath20 with ordinary handles from the decomposition of @xmath43 .
if @xmath176 is transverse to the boundaries of all handles and points into the right directions , then @xmath371 carries an engel structure such that all round handles carry model engel structures .
this situation arises in the context of convex contact structures defined in @xcite .
[ d : convcontstr ] a contact structure @xmath41 on a manifold @xmath43 is _ convex _ if there is a proper morse function @xmath413 and a complete contact vector field @xmath176 which is a pseudo
gradient for @xmath414 , i.e. there is a riemannian metric and a positive function @xmath263 such that @xmath415 . according to @xcite every contact structure on a @xmath3manifold is convex . from @xmath414
one obtains a handle decomposition of @xmath43 and applying proposition [ p : perturbed prolong ] to convex contact structures one obtains examples of engel structures with adapted round handle decomposition .
[ s : index zero models ] for our proof of theorem [ t : exist ] it is useful to have overtwisted contact structures on transverse boundaries of engel manifolds .
since we will start our construction of engel manifolds with a round handle of index @xmath138 we define the model engel structures on @xmath25 such that the contact structure on @xmath416 is overtwisted .
we use cylindrical coordinates on @xmath417 . consider the overtwisted contact structure @xmath418 and the ball of radius @xmath419 .
the singular foliation on the boundary @xmath420 admits a dividing set with three connected components . from theorem [ t : divide impera ] it follows that @xmath420 is convex , i.e. there is a contact vector field @xmath176 transverse to @xmath420 which points out of the ball . since the ball is contractible the contact structure admits a trivialization @xmath133 . by proposition [ p : perturbed prolong ]
we obtain model engel structures @xmath421 on @xmath25 . at this point
we do not need to deal with the case @xmath376 .
the singular foliation on @xmath422 is the same as the original singular foliation on @xmath420 . by theorem [ t : giroux tight ]
the contact structure on @xmath416 is overtwisted . since @xmath423 there are exactly two homotopy classes of oriented framings on @xmath25 .
the homotopy class of the engel framing of @xmath372 depends only on the parity of @xmath184 .
so we have shown [ p : index 0 models ] for each of the two homotopy classes of oriented framings on @xmath25 there is a model engel structure on @xmath25 such that the engel framing is homotopic to the given framing of the tangent bundle of @xmath25 . moreover , the induced contact structure on the boundary is overtwisted .
j. adachi asks in @xcite whether it is possible to find an engel structure on @xmath424 $ ] such that the boundary is transverse and the induced contact structures @xmath202 on @xmath425 and @xmath131 on @xmath426 are equivalent to given contact structures on @xmath43 .
one can show that @xmath202 and @xmath131 have to be homotopic
as plane fields on @xmath43 as in lemma [ l : unique homotopy class of plane fields ] .
the following example shows that @xmath202 and @xmath131 do not have to be equivalent .
consider the contact structure from above on @xmath427 .
we remove a ball contained in a darboux chart with convex boundary @xmath428 from the interior of @xmath427 and choose the contact vector field @xmath176 such that it is transverse to both @xmath428 and @xmath420 . by proposition [ p : perturbed prolong ] we obtain an engel structure on a manifold diffeomorphic to @xmath429{\times}s^1 $ ] such that * the characteristic foliation is transverse to the boundary * the contact structure on one boundary component is tight while it is overtwisted on the other boundary component .
this answers the question completely for @xmath430 by the classification of contact structures on @xmath431 .
[ s : index one models ] on @xmath432 we denote the coordinate on @xmath433 by @xmath270 , the coordinates on @xmath434 by @xmath435 and the coordinate on @xmath20 by @xmath212 . on @xmath436 consider @xmath437 these choices are motivated by @xcite : the resulting even contact structure on @xmath438 is defined by the contraction of the symplectic form @xmath439 with the liouville vector field @xmath440 .
notice that @xmath176 is transverse to @xmath441 and points outwards along @xmath442 and inwards along @xmath443 .
we apply proposition [ p : perturbed prolong ] to @xmath444=-{\frac{\partial}{\partial{y_2}}}-1/2{\frac{\partial}{\partial{x}}}+1/2(y_2+x){\frac{\partial}{\partial{y_1}}}~. \end{split}\end{aligned}\ ] ] with @xmath445 .
the properties of the resulting distributions @xmath372 are summarized in the following proposition .
[ p:1 modelle ] the distributions @xmath372 are model engel structures on @xmath438 for all @xmath163 .
they have the following properties . *
the even contact structure @xmath387 $ ] is defined by @xmath446 .
the orientation of the contact structure on @xmath447 and @xmath448 with respect to the restriction of @xmath449 is positive if @xmath450 and negative if @xmath400 . *
the curves @xmath451 are legendrian .
the rotation number along them is @xmath410 .
* the rotation number of the intersection line field with respect to the legendrian vector field @xmath452 along @xmath453 is @xmath410 and it equals @xmath138 along @xmath454 .
this follows from the discussion of proposition [ p : perturbed prolong ] and calculations using and .
[ r : contact contraction ] the contact structure on @xmath447 is defined by @xmath455 with @xmath456 .
it is invariant under the map @xmath457 induced by @xmath458 for @xmath459 . in particular ,
if @xmath460 is an embedding preserving contact structures , then the same is true for @xmath461 .
we can choose @xmath263 so small that the image of @xmath461 is contained in a given tubular neighborhood of @xmath462 .
up to now we have a model engel structure on @xmath438 for each orientation of the contact structure on @xmath447 and each homotopy class of intersection line fields with negative rotation number along @xmath463 .
moreover , we have a model engel structure with rotation number @xmath138 along @xmath463 for one of the two possible orientations of the contact structure on @xmath447 . in order to obtain the missing possibilities we extend the contact diffeomorphism @xmath464 of a neighborhood of @xmath463 in @xmath447 to a diffeomorphism of @xmath438 .
this map preserves @xmath463 and the contact structure near @xmath447 but it reverses the orientations of @xmath463 and of the contact structure .
we push forward the engel structures @xmath372 we have obtained so far using the extension of @xmath71 .
then the rotation number along @xmath463 changes its sign . up to now all model
engel structures induce the same contact framing along @xmath463 . in order to realize other contact framings
one can push forward the model engel structures obtained so far with the self diffeomorphisms @xmath465 of @xmath438 @xmath466 with @xmath467 .
this map preserves @xmath463 as well as @xmath447 .
now we have shown the following proposition .
for each oriented framing of @xmath468 and @xmath163 there is a model engel structure on @xmath438 such that the contact framing is homotopic to the given framing and the rotation number along @xmath463 is @xmath184 .
given one such model engel structure , there is another model engel structure which induces the same contact structure on @xmath447 with the same rotation number along @xmath463 but which induces the opposite orientation of the contact structure on @xmath447 . for @xmath469 let @xmath470 for @xmath469 be the push forward with @xmath465 of a model engel structure with rotation number @xmath184 along @xmath463 .
the orientation of the contact structure on @xmath471 does not appear in this notation .
[ s : modelle index 2 ] the model engel structures on @xmath472 depend on two parameters @xmath473 which correspond to the homotopy class of the intersection line field near the torus @xmath474 .
the contact structure on @xmath475 is essentially independent of the model engel structure .
[ p : zwei henkel modell ] given integers @xmath476 and @xmath477 there is a model engel structure @xmath478 on @xmath479 with the following properties . *
the characteristic foliation of @xmath4 can be oriented such that it points outwards along @xmath480 and inwards along @xmath475 . *
the singular foliation on @xmath474 is in standard form . in particular , @xmath481 is convex .
the legendrian ruling corresponds to the first factor of @xmath482 .
the dividing curves are tangent to the last factor . *
the rotation number of the intersection line field along @xmath483 ( with its orientation as boundary @xmath484 ) is @xmath76 . *
the rotation number of the intersection line field along the legendrian divides ( with the canonical orientation of the last factor of @xmath485 ) is @xmath375 . *
the orientation of the contact structure on @xmath480 can be chosen freely .
all model engel structures induce the same contact structure on a neighborhood of @xmath486 .
we choose the even contact structure on @xmath472 first .
the starting point is a singular foliation @xmath225 on a disc @xmath434 where we use polar coordinates @xmath487 .
the coordinate on @xmath433 is @xmath270 .
on @xmath488 we define @xmath225 by @xmath489 .
in addition we require that @xmath225 admits a dividing set @xmath254 with the following properties . *
the straight arc @xmath490 from @xmath491 to @xmath492 lies in @xmath254 .
in addition , @xmath254 contains @xmath493 closed circles . * except for @xmath490 , every component of @xmath254 is closed and bounds a disc containing no other components of @xmath254 . if @xmath494 all closed components lie below @xmath490 in @xmath495 and if @xmath496 they lie in the upper half disc
. figure [ b : r2 ] shows a possible @xmath225 for @xmath497 .
the thickened curves divide @xmath225 .
similar singular foliations can be found other @xmath21 . ] by theorem [ t : divide impera ] there is an @xmath246invariant positive contact structure @xmath41 on @xmath498 such that the induced singular foliation on @xmath499 is @xmath225 .
we choose an @xmath246invariant contact form @xmath45 for @xmath41 .
such that @xmath500 on @xmath501 .
the coordinate corresponding to the @xmath246factor is @xmath270 .
this choice fixes an orientation of the contact structure . in order to find a contact vector field @xmath176 and a @xmath5handle @xmath502
such that @xmath176 is transverse to @xmath503 we need to take some care since we know nothing about the region @xmath504 except that @xmath505 is a contact vector field everywhere .
we focus first on @xmath506 .
let @xmath507 be smooth functions on @xmath506 depending only on @xmath270 . by the proof of proposition [ p : convectfunctions ]
@xmath508 is the contact vector field associated to the function @xmath509 we choose the functions @xmath507 such that @xmath510 for @xmath511 . for these choices
the contact vector field @xmath176 associated to @xmath18 can be extended by @xmath512 on @xmath513 to a smooth contact vector field which we still denote by @xmath176 .
finally , we extend @xmath176 to a contact vector field on the whole of @xmath498 .
( for this it is enough @xmath256 and to apply proposition [ p : convectfunctions ] . ) the resulting contact vector field is transverse to @xmath514 $ ] and points inwards there . now consider the pair of hypersurfaces defined by the equation @xmath515 where @xmath516 depending on the sign of @xmath270 .
if we fix @xmath517 big enough , @xmath176 is transverse to the hypersurfaces @xmath518 and it points outwards
. then @xmath519 is diffeomorphic to an ordinary handle of index @xmath5 and @xmath176 is transverse to both boundary components , cf .
figure [ b : henkel ] .
moreover @xmath176 has the desired orientations along @xmath520 . ]
we chose @xmath225 and @xmath41 such that @xmath521 is legendrian .
using we can determine the rotation number of @xmath484 .
we obtain @xmath522 by remark [ r : rotzahlvergleich ] this is ( up to sign ) the rotation number of any model engel structure on @xmath479 whose even contact structure induces the singular foliation @xmath225 on @xmath523 .
now fix an oriented trivialization @xmath133 .
we denote the horizontal lifts of @xmath524 on @xmath525 by the same symbols . as usual @xmath212
is the coordinate on @xmath20 . by proposition [ p : perturbed prolong ]
the distribution @xmath372 spanned by @xmath526 is a model engel structure if @xmath374 is small enough and @xmath477 .
the characteristic foliation of @xmath372 is spanned by @xmath349 .
this vector field is transverse to @xmath527 and it points in the desired directions .
the even contact structure @xmath387 $ ] is defined by @xmath528 . using the expressions for @xmath529 and our choices of @xmath507 we obtain @xmath530 on @xmath531 .
the contact structure on @xmath475 is defined by @xmath532 restricting @xmath389 to @xmath533 we see that the singular foliation on @xmath481 is in standard form .
the curves @xmath534 and @xmath535 are the legendrian divides and the legendrian ruling is tangent to the foliation given by the first factor in @xmath533 . for @xmath398 ,
the orientation of the even contact structure is @xmath399 .
this is the orientation used in , hence the rotation number of the intersection line field along @xmath484 is @xmath76 . if @xmath400 we obtain the opposite orientation @xmath536 and therefore the rotation number of the intersection line field on @xmath475 along @xmath484 has now the opposite sign , i.e. @xmath537 . the rotation number along the legendrian divide is @xmath410 by proposition [ p : perturbed prolong ] .
let us summarize the properties of the model engel structures we have obtained up to now . recall that @xmath372 depends not only on @xmath184 but also on the choice of @xmath225 at the beginning of the proof and that @xmath493 is the number of closed components of the dividing set of @xmath225
. in the following table @xmath493 is the number of closed components of components . [
cols="^ " , ] + @xmath398 & @xmath133 & @xmath76 & @xmath538 + @xmath400 & @xmath539 & @xmath540 & @xmath410 + notice that since we fixed @xmath176 and @xmath45 on @xmath506 , the contact structure on @xmath475 is independent from @xmath541 .
the model engel structures with positive rotation numbers along the legendrian divides can be obtained by applying the involution @xmath542 to the model engel structures we have obtained so far .
the contact structure on @xmath543 is preserved by @xmath71 but @xmath71 reverses the orientation of the legendrian divides , cf . .
in particular , we can compare the orientations of the contact structure and the homotopy class of the intersection line fields with the corresponding properties of @xmath372 .
the model engel structures @xmath544 cover the cases missing in the table above ( with the exception of @xmath376 ) . using a singular foliation
whose dividing set has more non closed components than @xmath225 at the beginning of the proof of proposition [ p : zwei henkel modell ] one can construct model engel structures on @xmath479 such that the dividing set of @xmath481 has @xmath545 components .
[ r : contact contraction 2 ] this remark is the analogue of remark [ r : contact contraction ] for round handles of index @xmath5 .
equip @xmath479 with a model engel structure from proposition [ p : zwei henkel modell ] .
if @xmath546 is an embedding such that the image of @xmath547 is convex and such that @xmath548 preserves the isotopy class of the dividing set , then one can isotope @xmath548 such that it preserves the contact structure .
this follows almost immediately from theorem [ t : giroux flex ] and theorem [ t : giroux unique ] , but notice that the contact structure on @xmath475 is not invariant under @xmath505 . since it is homotopic through contact structures to a @xmath505invariant contact structure , we can use gray s theorem ( theorem [ t : gray ] ) to circumvent this problem .
[ ss : model 3-handles ] we now come to the construction of model engel structures on round handles of index @xmath3 . in the proof of theorem [ t :
exist ] we will attach @xmath549 to an engel manifold @xmath6 with transverse boundary such that the contact structure on @xmath104 is overtwisted .
hence the contact structure on @xmath550 should also be overtwisted .
contrary to round handles of index @xmath138 we will need _ all _ possible homotopy classes of intersection line fields on @xmath550 .
in particular we also need the case @xmath376 in proposition [ p : perturbed prolong ] . if it were enough to have the unique positive tight contact structure on @xmath551 it would be easy to describe the model engel structures explicitly . in the overtwisted case we were not able to find explicit formulas but we still can construct the model engel structures .
we use ordinary handles of dimension @xmath3 to construct a contact structure together with a contact vector field @xmath176 and a legendrian vector field @xmath68 such that @xmath552 $ ] are linearly independent .
the following construction is up to a small modification an example of giroux s construction of convex contact structures in @xcite .
on @xmath436 let @xmath270 be the coordinate on the first factor and @xmath553 the coordinates on the second factor .
consider the contact structure @xmath131 defined by @xmath554 and the contact vector field @xmath555 a simple calculation shows that the legendrian vector field @xmath556 has the property that @xmath557 $ ] are linearly independent everywhere .
the characteristic surface ( i.e. the points where @xmath176 is tangent to the contact structure ) of @xmath176 is the strip @xmath558 the contact vector field @xmath176 is transverse to @xmath441 and it points outwards ( respectively inwards ) along @xmath442 ( respectively @xmath443 ) .
the intersection of @xmath186 with @xmath559 , respectively @xmath560 , consists of two arcs .
on @xmath561 we use the same contact structure as on @xmath562 and let @xmath563 .
the contact vector field is reversed , i.e. @xmath564 , so it points outwards along @xmath565 and inwards along @xmath566 .
the characteristic surface @xmath567 of @xmath568 satisfies @xmath569 .
the singular foliation on @xmath559 consists of parallel straight segments and figure [ b : charaom ] shows the singular foliation on @xmath570 . ] on @xmath571 consider the standard contact structure defined by @xmath572 and the contact vector field @xmath573 .
let @xmath574 . again
a simple calculation shows that @xmath575 $ ] and @xmath576 are linearly independent everywhere .
the characteristic surface of @xmath577 is @xmath578 in particular , the dividing set @xmath254 on @xmath579 is connected .
now we combine the handles with the contact vector fields and the legendrian vector fields described above in order to construct a contact structure on @xmath580 together with a contact vector field @xmath176 and a legendrian vector field @xmath377 as in proposition [ p : perturbed prolong ] .
we orient @xmath581 such that the orientation of @xmath581 followed by @xmath582 gives the orientation of the contact structure .
moreover , we can associate a sign to each component of @xmath583 as follows : if the orientation of @xmath584,v_i$ ] at @xmath87 is the contact orientation , then the component containing @xmath87 is positive .
otherwise this region is negative .
notice that along @xmath585 , the vector fields @xmath584,v_i$ ] are not linearly independent .
[ p : d3 v x ] there is a contact structure @xmath41 on @xmath580 together with a contact vector field @xmath176 and a legendrian vector field @xmath68 without zeroes such that @xmath177 $ ] and @xmath68 are linearly independent everywhere and @xmath176 is transverse to @xmath586 .
moreover , the dividing set on @xmath586 has three connected components .
consider the overtwisted contact structure on @xmath417 which is defined by @xmath587 ( in polar coordinates ) . by theorem [ t
: divide impera ] the sphere @xmath420 with radius @xmath588 around the origin is convex , i.e. there is a contact vector field @xmath176 transverse to @xmath420 .
we assume that @xmath176 points inwards .
consider a collar @xmath589 , { \varepsilon}>0 $ ] of @xmath420 such that @xmath176 corresponds to the standard vector field @xmath590 induced by the second factor . clearly , one can choose a nowhere vanishing section of @xmath591 along @xmath420 and extend it to a legendrian vector field @xmath592 on @xmath153 with the desired property .
since @xmath420 is simply connected , all nowhere vanishing sections of @xmath41 along @xmath420 are homotopic .
we orient the boundary of @xmath153 such that the orientation of the boundary followed by @xmath176 gives the orientation of the contact structure . in order to prove the proposition we want to extend the contact structure , @xmath592 and @xmath590 to the interior of the sphere .
to do so we attach the handles @xmath593 .
the dividing set @xmath594 on @xmath595 consists of three connected components @xmath596 , as shown in the left part of figure [ b : surg ] .
regions of @xmath597 where @xmath552,v$ ] is the contact orientation are called positive and if @xmath598,v$ ] is he opposite orientation they are called negative . , scaledwidth=96.0% ] we attach @xmath562 to @xmath153 using an orientation preserving embedding @xmath599 such that @xmath600 and such that @xmath205 preserves the singular foliation .
moreover , we assume that @xmath205 maps the positive , respectively negative , regions of @xmath559 to the positive , respectively negative , regions of @xmath597 .
such a map exists because the dividing set on @xmath601 is transverse to the singular foliation .
it maps points where @xmath602 is tangent to the contact structure to @xmath594 . following @xcite one obtains a smooth contact structure @xmath603 and a smooth contact vector field on @xmath604 and we can modify the boundary of @xmath153 before we attach @xmath562 such that we obtain a smooth section @xmath179 of @xmath603 with the property that @xmath605 $ ] are linearly independent .
here we use that @xmath205 preserves positive and negative regions .
the dividing set @xmath606 on @xmath607 contains @xmath490 .
the other component consists of @xmath608 together with the parts of @xmath609 which do not lie in the attaching region of @xmath562 .
the right part of figure [ b : surg ] shows @xmath606 .
in particular @xmath606 has two connected components .
the homotopy class of @xmath179 as a nowhere vanishing section of @xmath603 along the torus @xmath607 is not fully determined by this construction .
the restriction of @xmath179 to @xmath490 depends only on @xmath592 and it is therefore uniquely determined up to homotopy . on the other hand ,
the homotopy class of @xmath179 along a curve which intersects the belt sphere of @xmath562 in exactly one point depends on the vertical modification of the boundary , cf .
remark [ r : suitable vert mod 1 ] .
we will use this flexibility when we attach @xmath610 . using theorem [ t : divide impera ] we first deform the torus @xmath607 by an admissible isotopy such that it is in standard form . the deformed space is still denoted by @xmath611 . we may vary the slope of the legendrian ruling such that there is an orientation preserving embedding @xmath612 which preserves singular foliations and which maps positive , respectively negative , regions of @xmath613 to positive , respectively negative , regions of the boundary .
now we can extend the contact structure @xmath603 and the contact vector field @xmath614 from @xmath611 to @xmath615 using @xmath616 and @xmath568 .
we denote the extensions by @xmath617 and @xmath618 . when we want to extend @xmath179 to @xmath619 using @xmath620 we have to ensure that @xmath621 is homotopic to @xmath179 and that the orientations of the contact structures given by @xmath605 $ ] and @xmath622 $ ] are coherent .
the latter requirement is satisfied since @xmath623 maps positive , respectively negative , regions of @xmath613 to positive , respectively negative , regions of @xmath607 . in order to ensure that @xmath624 is homotopic to @xmath179 we use the fact that the homotopy class of @xmath179 along @xmath607 is not uniquely determined . choosing different values for @xmath184 in remark [ r :
suitable vert mod 1 ] on the two connected components of the attaching region of @xmath562 we can vary the homotopy class of @xmath179 along closed curves which intersect the belt sphere of @xmath562 exactly once . in this way we can ensure the @xmath621 is homotopic to @xmath179 along the attaching curve of @xmath610 . now we can use vertical modifications of the boundary and we obtain a legendrian vector field @xmath625 on @xmath619 such that @xmath625 and @xmath626 $ ] are linearly independent . the attaching curve of @xmath610 intersects each component of @xmath606 exactly once .
the dividing set @xmath627 of @xmath628 contains both components of @xmath606 with the segments contained in the image of @xmath623 removed .
the endpoints of the remaining curves are connected by the two segments @xmath629 .
hence @xmath627 is connected and @xmath628 is diffeomorphic to a sphere .
all nowhere vanishing legendrian vector fields along @xmath628 are homotopic .
using theorem [ t : divide impera ] again , we change the singular foliation on @xmath628 such that it becomes equivalent to the singular foliation on @xmath630 .
we attach @xmath631 using an embedding of @xmath630 into @xmath628 which satisfies the same conditions on the orientations as @xmath205 and @xmath623 . after a vertical modification of the boundary we can attach @xmath631 to @xmath619 such that @xmath632 extend to a contact structure @xmath41 on a ball , a contact vector field @xmath176 and a nowhere vanishing section @xmath68 of @xmath41 with the desired properties .
we apply proposition [ p : perturbed prolong ] to @xmath633 and @xmath634 constructed in the proof of proposition [ p : d3 v x ] .
this yields model engel structures @xmath635 , on @xmath549 such that the contact structure on @xmath636 is overtwisted by theorem [ t : giroux tight ] .
because @xmath420 and the contact form @xmath45 used in the proof of proposition [ p : d3 v x ] are invariant under rotations around the @xmath637axis we may assume that the same is true for @xmath176 and @xmath256 .
for @xmath638 $ ] let @xmath639 on @xmath640 in spherical coordinates @xmath641 on @xmath579 . according to
, the contact structure on @xmath550 is defined by @xmath642 and @xmath643 is a contact form for all @xmath263 . by theorem [ t : gray ]
the contact structures defined by @xmath644 and @xmath642 are isotopic .
the self diffeomorphism @xmath71 of @xmath431 with @xmath645 preserves @xmath644 .
moreover @xmath71 preserves the legendrian curves @xmath403 when @xmath646 and @xmath256 vanishes at @xmath87 .
the orientations are reversed by @xmath71 .
hence @xmath71 extends to a diffeomorphism @xmath647 of @xmath549 which preserves the contact structure on @xmath550 but reverses its orientation and which preserves a legendrian curve on @xmath550 but reverses its orientation .
the rotation number of the intersection line field on @xmath550 along @xmath403 is @xmath410 for the model engel structures we have obtained so far .
if we push forward the model engel structure we obtained so far using @xmath647 we obtain model engel structures which represent the missing homotopy classes of intersection line fields on @xmath550 .
thus we have proved the following theorem .
[ t : models on r3 ] there are model engel structures on @xmath549 such that the induced contact structure on @xmath550 is @xmath648 and all possible orientations of @xmath41 and homotopy classes of intersection line fields are realized .
[ s : isotopy+choice ] in the following sections we explain how to extend engel structures from @xmath6 to @xmath649 using a suitable model engel structure on @xmath29 if @xmath548 is an attaching map for a round handle of index @xmath32 . for this
@xmath548 has to have certain properties explained in section [ s : extend engel structure ] and we will isotope @xmath548 and apply vertical modification of the boundary to @xmath6 to bring @xmath548 into the desired form .
this is discussed in section [ s : attach 1 ] to section [ s : attach 3 ] .
let @xmath6 be an engel manifold with oriented characteristic foliation and transverse boundaries and @xmath29 a round handle of index @xmath650 with a model engel structure .
[ p : gluing ] assume that an embedding @xmath651 maps the oriented contact structure @xmath652 on @xmath26 to the oriented contact structure @xmath653 on @xmath104 and preserves intersection line fields .
then the engel structure extends from @xmath6 to @xmath654 by the model engel structure on @xmath29 .
if the engel structures are oriented , then one obtains an oriented engel structure on @xmath30 if @xmath548 preserves the orientation of the intersection line fields . using the normal form theorem ( theorem [ t : normal forms for transversals ] ) for neighborhoods of transverse hypersurfaces in engel manifolds
we can embed collars @xmath655 of @xmath26 , respectively @xmath656 of @xmath104 , into @xmath657 , respectively @xmath658 .
we identify @xmath655 and @xmath656 with their image . by the second part of proposition [ p : engel diffeos ] we obtain an embedding @xmath659 from @xmath548 .
since @xmath130 preserves intersection line fields , @xmath660 maps @xmath661 to @xmath662 and because @xmath548 preserves the orientation of the contact structure , @xmath663 and @xmath656 lie on opposite sides of the hypersurface @xmath664 .
this proves the claim . strictly speaking
@xmath30 is a manifold with corners .
we smooth these corners by cutting out a piece of @xmath29 such that the boundary of the resulting manifold is still transverse .
proposition [ p : gluing ] can of course be formulated with an engel manifold instead of the round handle @xmath29 .
the following three sections describe how to choose model engel structures on @xmath665 and how to isotope the attaching map in order to satisfy the conditions in proposition [ p : gluing ] .
it is enough to ensure that the attaching map preserves to homotopy class of the intersection line field since we can then apply vertical modifications of the boundary . before we continue let us remark that we will always assume that the attaching map @xmath354 preserves the contact orientations .
if this is not the case we can replace @xmath548 by @xmath666 where @xmath71 is the diffeomorphism of @xmath26 induced by complex conjugation of @xmath20 .
we obtain diffeomorphic manifolds when we attach @xmath29 to @xmath6 using @xmath548 or @xmath666 .
[ s : attach 1 ] the model engel structures on round handles of index @xmath10 induce contact structures on @xmath447 such that the attaching curves @xmath451 are legendrian .
let @xmath667 be an embedding preserving the contact orientation .
we want to isotope @xmath548 such that the resulting map preserves oriented contact structures and the homotopy class of intersection line fields for a suitable choice of model engel structure @xmath470 . after a vertical modification of @xmath104
we then obtain an engel structure on @xmath6 with @xmath438 attached .
it is a standard fact in contact topology that every embedded curve in a contact manifold can by isotoped to a legendrian curve such that the isotopy never leaves an arbitrarily small tubular neighborhood of the original curve , @xcite .
hence we may assume that @xmath462 are legendrian curves .
next we want to isotope @xmath548 such that the resulting map preserves oriented contact structures and the homotopy class of intersection line fields along @xmath463 for a suitable choice of model engel structures on @xmath438 . for this we stabilize the curves @xmath462 .
we may assume that all stabilizations of @xmath462 are carried out in disjoint tubular neighborhoods of the original curves .
then one can extend the isotopy of @xmath548 from @xmath463 to @xmath447 and one obtains a stabilized attaching map .
let @xmath668 and @xmath669 be the restrictions of @xmath548 to @xmath670 and @xmath671 . by @xmath672
we denote a contact framing of @xmath463 if @xmath438 carries the model engel structure @xmath470 .
since @xmath548 is orientation preserving there exist integers @xmath673 such that @xmath548 maps a contact framing along @xmath463 to a framing representing @xmath674 when @xmath438 carries the model engel structure @xmath675 .
[ p : adapt frame and rot ] we can choose a model engel structure on @xmath438 and stabilize @xmath676 such that the stabilized maps * send contact framings of @xmath463 to framings of @xmath677 which are homotopic to a contact framing , and * the rotation numbers of @xmath4 along the stabilized legendrian curves obtained from @xmath678 and @xmath679 are both equal , if and only if @xmath680 we equip @xmath438 with the model engel structure @xmath675 .
let @xmath185 be a framing of @xmath681 . because @xmath548 is orientation preserving , the framings @xmath682 are homotopic .
if we use the model engel structure @xmath470 instead of @xmath675 on @xmath438 this implies @xmath683 in we have determined the effect of positive and negative stabilization on contact framings .
since we want the stabilized embeddings @xmath684 to map contact framings of @xmath463 to framings of @xmath685 which are homotopic to contact framings , we have to apply positive or negative stabilization @xmath686times .
hence @xmath687 has to be non negative .
if @xmath688 satisfy @xmath689 then it follows from that the rotation numbers of the stabilized legendrian curves are given by @xmath690 after sufficiently many stabilizations we also want the rotation numbers along the image of @xmath691 and @xmath692 to be equal .
this can be achieved if and only if we can solve and @xmath693 with nonnegative integers @xmath694 and @xmath467 .
then we can take @xmath695 from and we obtain the condition
. therefore this condition is necessary .
conversely , if is satisfied , then the under determined system of equations and admits solutions in @xmath191 . if we choose @xmath696 large enough we can achieve @xmath697 .
note that in this proof we used only stabilization to modify legendrian curves . attaching a bypass to a convex surface along a segment of a legendrian knot contained in the surface can be viewed as an inverse procedure of stabilization . under the assumption that the contact structure on @xmath104 is overtwisted it is possible to prove proposition [ p : adapt frame and rot ] for model engel structures with a fixed contact framing , i.e. without choosing @xmath696 big enough as we did above .
let us explain the meaning of in more topological terms .
consider an orientation preserving attaching map @xmath698 .
on @xmath108 we have the engel framing and we may assume that its components except of the component spanning @xmath16 form a framing of the tangent bundle of @xmath104 . using @xmath548
we pull back this trivialization to @xmath447 and add a vector field which points inwards and is transverse to @xmath447 .
we obtain a framing of @xmath699 along @xmath447 .
it is possible to isotope @xmath548 such that we can extend the engel structure from @xmath6 to @xmath6 with @xmath438 attached only if we can extend the engel framing from @xmath6 to @xmath700 .
this condition depends only on the engel framing and the isotopy class of @xmath548 .
if @xmath462 are legendrian one can show that such an extension exists if and only if is satisfied .
on the other hand extending a framing from @xmath6 to @xmath700 is a purely topological problem .
thus if we start with an attaching map @xmath676 and end up with a map violating then it is not possible to construct an engel structure on @xmath700 such that the engel framing on @xmath6 is homotopic to the engel framing induced by @xmath4 . now if @xmath667 is an embedding such that @xmath462 are legendrian curves and @xmath130 preserves contact framings and rotation numbers along @xmath463 then by lemma [ l : tub nbhd of leg curves ] we can isotope @xmath548 relative to @xmath463 such that the resulting map preserves the contact structure on a tubular neighborhood of @xmath463 . using remark [ r : contact contraction ]
we can isotope the new attaching map such that its image lies in this tubular neighborhood .
[ t : construct attaching maps ] let @xmath6 be an oriented engel manifold with transverse boundary and assume that @xmath667 is an embedding and the trivialization of @xmath108 induced by the engel structure can be extended to @xmath701 . then there is a model engel structure on @xmath438 such that @xmath548 is isotopic to an embedding @xmath660 which preserves oriented contact structures and the homotopy class of the intersection line field .
[ s : flex tori ] all our model engel structures on round @xmath5handles induce the same singular foliation on @xmath474 .
now suppose that @xmath6 is an engel manifold with transverse boundary and @xmath546 is an attaching map which preserves the orientations induced by the contact structures .
if we want to attach @xmath479 to @xmath6 and extend the engel structure from @xmath6 to @xmath702 , we have to ensure that the attaching map preserves contact structures . by theorem [ t
: divide impera ] and theorem [ t : giroux unique ] together with remark [ r : contact contraction 2 ] it suffices to modify @xmath548 such that after the deformation , the image of @xmath547 is convex and the attaching map preserves singular foliation . recall that the dividing set of @xmath481 consists of two homotopically non trivial circles .
let @xmath703 and @xmath704 .
we assume that the contact structure @xmath41 on @xmath43 is overtwisted .
using lemma [ l : attach ] and proposition [ p : bypass existence ] we can bring @xmath705 into the desired form .
it is clear how to obtain the desired isotopy of @xmath548 from this .
[ t : flexible tori ] let @xmath705 be an embedded torus in an overtwisted contact manifold @xmath285 .
assume that @xmath41 is orientable and that the euler class of the restriction of @xmath41 to @xmath705 is zero .
then we can isotope @xmath705 such that after the isotopy the singular foliation on the torus is in standard form .
moreover we can prescribe the slope of the dividing curves .
after the isotopy , the complement of a tubular neighborhood of @xmath705 contains an overtwisted disc .
it suffices to find a convex torus which is isotopic to the original one such that the dividing set consists of two homotopically non
trivial components which have the desired slope . using the giroux flexibility theorem ( theorem [ t : giroux flex ] )
one can arrange the singular foliation on @xmath705 such that it has standard form .
we will frequently use proposition [ p : bypass existence ] .
the following figures represent the dividing set on a torus before the bypass attachment .
the thickened arc represents the attaching curve @xmath291 of the bypass .
@xmath706 step : let @xmath300 be a convex overtwisted disc .
we perturb the embedding of @xmath705 such that it becomes transverse to @xmath300 . using an extension of a radial vector field on @xmath300 we can isotope @xmath705 such that after the isotopy @xmath707 and the resulting torus is convex .
since @xmath300 is convex there is a neighborhood which is foliated by overtwisted discs . in the following we will always ensure that after each modification of the embedding of @xmath705 , there is an overtwisted disc disjoint from the deformed torus : if @xmath288 is a bypass for @xmath705 , we choose the neighborhood of @xmath708 in lemma [ l : attach ] so small that its complement still contains overtwisted discs .
@xmath709 step : in this step we remove all homotopically non
trivial components of the dividing set . if there are no such components we continue with step 3 .
if the dividing set contains more than two homotopically non
trivial components , then we reduce the number of its components of the dividing set using the bypass attachments in the left part of figure [ b : weg ] often enough . ]
we end up with a dividing set which contains two homotopically non trivial curves .
we remove these components with the bypass attachment in the right part of figure [ b : weg ] @xmath710 step : using the bypass attachment in figure [ b : neuecomp ] , we obtain two new components of the dividing set .
their slope depends on the attaching curve of the bypass . for a given identification @xmath711
, we can achieve that the new components of the dividing set are isotopic to @xmath403 for @xmath712 .
the dashed curve represents this circle . ]
@xmath713 step : we are left with a convex torus whose dividing set contains exactly two homotopically non trivial dividing curves @xmath714 with the desired slope .
if this is the entire dividing set we are done .
otherwise we consider the two annuli @xmath715 .
we claim that if only one of these annuli contains other components of the dividing set @xmath254 , then there is at least one component of @xmath254 which bounds a disc @xmath716 containing another component of @xmath254 .
assume that this is not true .
then @xmath717 contains @xmath718 discs , one annulus and one annulus with @xmath719 holes .
the euler number @xmath720 of the restriction of @xmath41 to @xmath705 is @xmath721 by .
the sign depends on the orientations of @xmath705 and of the contact structure . but contradicts our assumption on the euler class of @xmath41 . in order to reduce the number of connected components of @xmath254 we perform a bypass attachment as indicated in the left part of figure [ b : otweg ] .
notice that this does not affect the homotopically non trivial dividing curves . ]
if both annuli @xmath715 contain connected components of @xmath254 we reduce the number of components using the bypass attachment in the right part of figure [ b : otweg ] .
if do this often enough we end up with the desired configuration of dividing curves on @xmath705 .
a bypass attachment also affects framings . for our purpose , it is enough to show this for a particular bypass attachment .
k. honda described this effect in more detail , cf .
proposition 4.7 in @xcite where one can find a proof of the following lemma . ]
[ l : dehn ] let @xmath271 be a torus in standard form contained in a contact manifold such that the legendrian divides are isotopic to @xmath722 .
let @xmath68 be a nowhere vanishing legendrian vector field on the contact manifold .
assume that the rotation number of @xmath68 along the legendrian divides is zero and that it is even along @xmath723 .
we attach a bypass as in figure [ b : dehn ] to @xmath705 and bring the characteristic foliation in standard form such that the legendrian ruling is still tangent to the foliation from the first factor in @xmath724 .
then the rotation number along the legendrian divides in the isotoped torus is odd ( and therefore non zero ) while the rotation number along the legendrian curve @xmath723 remains even .
[ p : attaching round 2-handles ] assume that @xmath6 carries an oriented engel structure with oriented characteristic foliation such that the boundary of @xmath6 is transverse .
moreover , assume that the contact structure on @xmath104 is overtwisted .
let @xmath546 be an attaching map .
we can extend the engel structure from @xmath6 to @xmath702 using a model engel structure if and only if the engel framing on @xmath6 extends to a trivialization of @xmath702 over the @xmath5cell @xmath725 . the engel structure on @xmath702
can be chosen such that the contact structure on @xmath726 again overtwisted .
the @xmath5cell @xmath727 is of course attached using the restriction of @xmath548 to @xmath728 . clearly , if the engel trivialization can not be extended over @xmath727 , then it is impossible to extend the engel structure to @xmath729 .
the contact structure @xmath41 on @xmath104 is oriented and the intersection line field yields a nowhere vanishing legendrian vector field .
therefore the euler characteristic of @xmath41 viewed as a bundle is zero .
we equip @xmath479 with one of the model engel structures from section [ s : modelle index 2 ] . by theorem [ t : flexible tori ] and
remark [ r : contact contraction 2 ] we can assume that @xmath548 preserves contact structures . in particular , @xmath730 is legendrian .
we first show that under the assumption of the proposition , the rotation number along @xmath730 is even . for this
we homotop the engel trivialization on @xmath6 such that the only component @xmath349 of the framing which is not tangent to @xmath104 spans the characteristic foliation . in this way we obtain a framing of @xmath104 . using the attaching map to pull back this framing , we obtain a framing on @xmath475 . if we add a vector field along @xmath475 which spans and orients the characteristic foliation we obtain a framing of @xmath731 along @xmath475 .
since @xmath548 preserves contact structures , the pull back of the component of the engel framing which is orthogonal to the contact structure on @xmath104 is transverse to the contact structure on @xmath475 . without loss of generality
, we assume that these components of the engel framings are preserved by the attaching map .
then the pullback framing and the engel framing on @xmath479 have two components in common .
when we want to compare the pull back framing with the engel trivialization along @xmath727 it is therefore enough to consider the rotation numbers along @xmath728 . by ( iii ) of proposition
[ p : zwei henkel modell ] , the rotation number along @xmath728 induced by the model engel structure on @xmath479 is even .
if the rotation number of the pull back framing along @xmath728 is odd , then the pull back framing and the engel framing are not homotopic along @xmath728 and the pull back framing can not be extended over the disc @xmath732 .
this contradicts the assumption .
hence the rotation number along @xmath730 is even .
if the rotation number along the legendrian divides of @xmath733 is not zero , then by proposition [ p : zwei henkel modell ] we can choose a suitable model engel structure such that the isotoped attaching map preserves the homotopy class of the intersection line fields .
if the rotation number along the legendrian divides of @xmath733 is zero , then we attach a bypass as in figure [ b : dehn ] to isotope @xmath548 once again . by lemma [ l : dehn ] ,
the rotation number along the legendrian rulings of the image of @xmath734 remains even and the rotation number along the legendrian divides is non zero .
the new attaching map preserves the singular foliation on @xmath547 after one pushes forward the model engel structures from proposition [ p : zwei henkel modell ] with the diffeomorphism @xmath735 from .
when we restrict this diffeomorphism to @xmath485 then we obtain a right handed dehn twist .
now we can choose the model engel structure such that the new attaching map preserves oriented contact structures and the homotopy class of the intersection line field .
after a vertical modification of the boundary of @xmath6 , we can extend the engel structure over @xmath479 using a model engel structure from section [ s : modelle index 2 ] .
[ s : attach 3 ] let @xmath736 be an orientation preserving attaching map for a round @xmath3handle carrying a model engel structure from section [ ss : model 3-handles ] .
the model engel structures on @xmath549 induce overtwisted contact structures on @xmath550 . by theorem [ t : eliashbergs wahnsinn ]
we can isotope @xmath548 to a contact diffeomorphism if and only if @xmath548 is orientation preserving and the image of the contact structure on @xmath550 is homotopic to the contact structure on @xmath104 as a plane field .
the contact structure on @xmath550 is orientable and it extends to @xmath549 as a plane field .
the following lemma shows that this determines the homotopy class of the plane field completely .
[ l : unique homotopy class of plane fields ] there is a unique homotopy class of orientable plane fields on @xmath737 which extends to @xmath738
. recall from @xcite that the grassmannian of oriented planes in @xmath417 , respectively @xmath739 , is @xmath740 , respectively @xmath741 and that the inclusion @xmath742 induces the diagonal map @xmath743 let @xmath202 and @xmath131 be two plane fields on @xmath431 that extend to the interior of @xmath738 .
we view @xmath201 as maps from @xmath431 to @xmath744 and their extensions as maps from @xmath738 to @xmath745 . because @xmath746 is a strong deformation retract of @xmath738 and @xmath745
is simply connected , there is a homotopy between the extensions of @xmath202 and @xmath131 through plane fields in @xmath747 . using the projection of @xmath741 onto the first factor
, we obtain a homotopy between @xmath202 and @xmath131 in @xmath748 .
after we have isotoped @xmath548 to a contact diffeomorphism we can choose a model engel structure on @xmath549 such that the orientation of the contact structure and the homotopy class of the intersection line field are preserved by the isotoped attaching map , cf .
theorem [ t : models on r3 ] .
[ p : extending over r3 ] the engel structure extends from @xmath6 to @xmath749 if and only if the contact structure on @xmath104 extends to @xmath749 as a plane field .
[ s : proofs ] now we combine the tools from the previous sections to prove existence theorems for engel structures . in section [
s : end proof ] we show that every parallelizable manifold admits an orientable engel structure . in section [ s : conn sum ] we explain the construction of engel structures on the connected sum of two engel manifolds with @xmath750 if the characteristic foliations of the original engel structures satisfy certain conditions . in the proof of our main result theorem [ t : exist ] we use the model engel structures from section [ s : round model ] and the results from section [ s : isotopy+choice ] showing that we can extend an engel structure from @xmath751 to @xmath31
if @xmath30 has transverse boundary and the engel framing extends to a framing of @xmath31 .
the remaining problem is to ensure that after we have constructed an engel structure on @xmath31 , the new engel framing extends from @xmath31 to the whole of @xmath6 .
this allows us to perform the handle attachments successively .
let @xmath6 be a closed parallelizable manifold of dimension @xmath0 and fix a trivialization @xmath752 of @xmath108 .
we consider a round handle decomposition @xmath753 of @xmath6 such that there is exactly one round @xmath3handle and one round @xmath138handle .
such a decomposition exists by theorem [ t : round decomp euler 0 ] .
we will frequently isotope the attaching maps but this will not be reflected in the notation .
we start with the round handle of index @xmath138 . by proposition
[ p : index 0 models ] we can choose a model engel structure on @xmath25 such that the engel framing on @xmath25 is homotopic to the original framing .
in particular the engel framing extends from @xmath25 to @xmath6 . the contact structure on @xmath416 is overtwisted by construction .
let @xmath754 be the round handle body obtained from @xmath755 .
assume that we have constructed an engel structure on @xmath754 such that the contact structure on @xmath756 is overtwisted .
assume moreover that throughout this process we have homotoped the trivialization of @xmath6 we chose at the beginning such that it coincides with the engel trivialization on @xmath754 .
then the engel trivialization on @xmath754 can be extended to @xmath757 . by theorem [ t :
construct attaching maps ] , we can isotope @xmath758 to an attaching map such that the engel structure on @xmath754 extends to an engel structure on @xmath759 using a model engel structure on @xmath438 from section [ s : index one models ] . in order to ensure that the contact structure on @xmath760 is again overtwisted , we isotope @xmath758 before the application of theorem [ t : construct attaching maps ] such that its image is disjoint from an overtwisted disc in @xmath756 .
let @xmath451 be the attaching curves of @xmath761 .
assume that @xmath762 is transverse to an overtwisted disc @xmath300 and let @xmath87 be a point on @xmath300 which does not lie on @xmath762 .
then use the flow of a radial vector field centered at @xmath87 to isotope @xmath758 such that the image of @xmath463 becomes disjoint from @xmath300 .
the remaining steps in the modification of @xmath548 , like making the attaching curves legendrian and performing stabilizations , can be carried out in a small tubular neighborhood which is also disjoint from @xmath300 .
next we compare the engel trivialization and the original trivialization of @xmath6 on @xmath763 relative to @xmath754 .
the cylinder @xmath764 can be decomposed into a @xmath10cell @xmath765 and a @xmath5cell @xmath727 .
the @xmath10cell is attached to @xmath766 using the restriction of @xmath758 to @xmath767 and @xmath727 is attached to @xmath768 . since @xmath423 there are two homotopy classes of oriented framings of @xmath108 along @xmath769 . if necessary
, we modify the model engel structure on @xmath761 such that the new engel framing is homotopic to the given trivialization along @xmath769 relative to the endpoints of @xmath769 . for
this let @xmath770 $ ] be a smooth function which is constant near the boundary , @xmath771 and @xmath772 .
if we push forward the model engel structure on @xmath438 using the diffeomorphism @xmath773 then we still obtain a smooth engel structure on @xmath763 by the choice of @xmath774 . because @xmath775 interchanges the two homotopy classes of framings along @xmath769 relative @xmath767 ,
the trivialization induced by the new engel structure on @xmath438 is now homotopic to the given trivialization of @xmath108 along @xmath769 relative @xmath767 .
we homotop the original framing such it coincides with the engel framing on @xmath768 . since @xmath776 is trivial the same is true for @xmath777 .
thus the engel framing on @xmath778 is now homotopic to the original framing relative to @xmath754 . in particular
it extends from @xmath778 to @xmath6 .
hence we can iterate the procedure to attach round handles of index @xmath10 . during this process
we ensure that the contact structure on @xmath760 is overtwisted for all @xmath779 by choosing the attaching regions disjoint from @xmath300 . in the next step
we attach round @xmath5handles .
assume that we have already attached the first @xmath780 round @xmath5handles such that on the resulting handle body @xmath781 there is an engel structure extending the engel structure on @xmath782 .
in addition we assume that the contact structure on @xmath783 is overtwisted .
consider the attaching map @xmath784 of @xmath785 .
the contact structure on @xmath783 is oriented and it has an orientable section , namely the intersection line field .
thus the euler class of the contact structure , viewed as a bundle , vanishes . by assumption ,
the contact structure is overtwisted .
according to theorem [ t : flexible tori ] and theorem [ t : giroux flex ] we can isotope @xmath786 such that the resulting map preserves singular foliations .
recall that in theorem [ t : flexible tori ] we also ensured that the attaching region of @xmath785 is contained in a neighborhood of @xmath787 which is disjoint from some overtwisted disc .
first we decompose each round @xmath5handle @xmath790 into one ordinary handle @xmath791 of index @xmath5 and one ordinary handle @xmath792 of index @xmath3 as in .
then we rearrange the handles @xmath793 such that the ordinary handles of index @xmath5 are attached to @xmath782 independently and the remaining ordinary @xmath3handles are attached to the resulting handle body .
the engel trivialization on @xmath781 extends over @xmath794 _ before _ we rearrange the handle decomposition if and only if the engel trivialization extends from @xmath782 over @xmath794 _
after _ the rearrangement . by construction
the engel framing on @xmath782 is homotopic to the framing of @xmath6 we chose at the beginning of the proof .
in particular the engel framing extends from @xmath782 to @xmath794 after the rearrangement of the handles @xmath795 . by proposition [ p : attaching round 2-handles ] we can extend the engel structure from @xmath781 to @xmath799 using a model engel structure from section [ s : modelle index 2 ] on @xmath785 . since the contact structure on @xmath800 is again overtwisted we can iterate the procedure .
when the last round @xmath5handle is attached we have constructed an engel structure on @xmath801 . in order to finish the construction we have to extend the engel structure over the round @xmath3handle @xmath549 .
we decompose all round @xmath5handles into ordinary handles @xmath802 for @xmath803 of index @xmath5 and @xmath3 and we rearrange the handles such that the @xmath5handles are attached to @xmath782 .
we also decompose @xmath549 into an ordinary @xmath3handle @xmath804 and one ordinary @xmath0handle @xmath805 . in claim 1
we have shown that the engel trivialization on @xmath782 extends to @xmath806 and that all such extensions are homotopic .
therefore , the engel trivialization on @xmath807 also extends to @xmath6 .
next we reduce the problem to trivializations of bundles of rank @xmath3 .
the first component @xmath349 of the engel trivialization is transverse to @xmath808 by construction .
thus @xmath349 extends to a vector field without zeroes on @xmath6 .
we equip @xmath6 with an almost quaternionic structure such that the engel framing and @xmath809 coincide on @xmath810 .
then we can choose a trivialization of the orthogonal complement @xmath811 of @xmath349 in @xmath6 .
( this trick can be found in @xcite . ) for the proof of claim 2 it suffices to show that we can extend the trivialization of @xmath811 induced by the engel structure . on the @xmath5skeleton of @xmath6
the @xmath812bundle @xmath811 is trivial .
therefore we can lift it to an @xmath813bundle .
( recall that @xmath814 . )
since @xmath815 is trivial , the trivialization of @xmath811 induced by the engel structure extends from @xmath801 to @xmath804 .
we fix such an extension .
the obstruction for the extension of the trivialization from @xmath816 to @xmath6 of @xmath811 induced by the engel structure is a cocycle @xmath270 in the cellular cochain group @xmath817 and @xmath270 may depend on the choice of extensions of the trivialization over the @xmath3handles .
the cocycle @xmath270 represents a class @xmath818\in h^4(m , m_2\cup\widehat{h}_3;{\mathbb { z}})$ ] which does not depend on the choice of trivializations on the @xmath3handles because it is a primary obstruction , cf .
theorem 34.2 in @xcite .
we have already shown that the engel trivialization extends from the @xmath5skeleton to the whole of @xmath6 .
thus @xmath818=0 $ ] . since there is exactly one @xmath0handle we have @xmath819 and hence @xmath820 .
this implies that the engel trivialization on @xmath801 extends to @xmath6 .
since the engel framing extends from @xmath801 to @xmath6 the same is true for the contact structure on @xmath821 viewed as a plane field .
the contact structure on @xmath821 is overtwisted by construction . by proposition [ p : extending over r3 ] we can extend the engel structure from @xmath801 to @xmath6 .
this finishes the proof of the theorem .
our construction relies on round handle decompositions and therefore condition ( i ) in theorem [ t : round decomp euler 0 ] is used at all stages of the proof . the second stiefel
whitney class @xmath822 of an orientable @xmath0manifold @xmath6 is zero if and only if @xmath108 is trivial on the @xmath5skeleton of @xmath6 .
thus condition ( ii ) of theorem [ t : trivial charkl ] is used when we attach round handles of index @xmath10 and @xmath5 in order to establish certain properties of rotation numbers .
it is also used when we lift @xmath811 to an @xmath813bundle in claim 2 .
finally the obstruction for the extension of a section of the @xmath813bundle appearing at the end of the proof of theorem [ t : exist ] can be viewed as the primary obstruction to the construction of a section of the @xmath813bundle over @xmath6 since all sections of a @xmath813bundle over @xmath810 are homotopic . according to @xcite p. 31 ,
an @xmath823bundle is trivial if and only if its second chern class vanishes . on the other hand the second chern class of the @xmath813bundle
is @xmath824 .
since @xmath825 , the vanishing of @xmath270 corresponds to ( iii ) by the signature theorem of hirzebruch .
let @xmath33 be manifolds with engel structures @xmath34 .
the connected sum @xmath826 does not admit an engel structure because its euler characteristic is @xmath345 . on the other hand
the euler characteristic of @xmath2 is zero .
moreover , if @xmath33 have trivial tangent bundle , then the same is true for @xmath827 by theorem [ t : trivial charkl ] .
therefore it is natural to try to construct an engel structure on @xmath827 from @xmath37 and @xmath38 such that the resulting engel structure coincides with the original engel structures one large parts of @xmath782 and @xmath801 . in this section
we show that this is possible under certain assumptions on @xmath37 and @xmath38 . for this
we use model engel structures on round handles of index @xmath10 and @xmath5 . as model engel structure @xmath828 on @xmath438
we use @xmath37 from proposition [ p:1 modelle ] .
on @xmath479 we will use a model engel structure @xmath829 which did not appear yet .
we define @xmath829 as the span of @xmath830 this model engel structure can be obtained using proposition [ p : perturbed prolong ] .
its characteristic foliation is spanned and oriented by @xmath349 .
the model engel structures @xmath828 and @xmath829 are very similar when one identifies @xmath438 and @xmath479 in the obvious way .
the curves @xmath831 in @xmath832 are legendrian and the rotation number of the intersection line field along these curves is @xmath235 for both @xmath828 on @xmath438 and @xmath829 on @xmath479 . on a neighborhood of the boundary of @xmath833
the even contact structures induced by @xmath828 and @xmath829 are homotopic through the family of even contact structures defined by @xmath834 with @xmath638 $ ] . by the even contact structure on @xmath438
is defined by @xmath644 while the even contact structure on @xmath479 is defined by @xmath642 .
the characteristic foliations of the even contact structures defined by @xmath643 are transverse to @xmath835 and @xmath836 for all @xmath263 .
hence @xmath643 induces a family of contact forms on @xmath447 and @xmath448 .
we can also obtain contact embeddings of @xmath480 from attaching maps of @xmath438 .
consider an engel manifold @xmath6 with transverse boundary and let @xmath837 be an embedding which preserves oriented contact structures . by remark [ r : contact contraction ]
we can isotope @xmath205 such that the image of the resulting embedding lies in a small neighbourhood of @xmath838 . the new embedding is again called @xmath205 . for a suitable cut off function
@xmath774 @xmath839 defines a family of contact structures on @xmath6 which is constant away from a neighborhood of @xmath838 and which is defined by @xmath643 near @xmath838 .
this homotopy through contact structures has compact support .
using gray s theorem ( theorem [ t : gray ] ) we obtain an isotopy @xmath457 of @xmath6 such that @xmath840 is a positive multiple of @xmath841 .
moreover , @xmath457 maps the legendrian curves @xmath838 to themselves for all @xmath263 .
the orientation of the contact structure induced by @xmath828 on @xmath447 , respectively @xmath829 on @xmath480 , is given by the restriction of @xmath842 , respectively @xmath843 , to the contact structure .
this implies that @xmath844 preserves _ oriented _ contact structures ( but the orientations of the characteristic foliations do not match ) .
since @xmath457 preserves the legendrian curves @xmath838 , the composition @xmath845 preserves the homotopy class of the intersection line field of @xmath829 .
[ t : m+n ] let @xmath1 be manifolds with engel structures @xmath34 such that both characteristic foliations admit closed transversals .
then @xmath36 carries an engel structure which coincides with @xmath37 , respectively @xmath38 , away from a neighborhood of the transversals where all connected sums are performed .
the characteristic foliation of the new engel structure again admits a closed transversal .
let us assume for the moment that the manifolds are oriented .
then the characteristic foliations are canonically oriented . for @xmath324 we cut @xmath846 along a closed transversal @xmath847 of the characteristic foliation .
the resulting manifolds are still denoted by @xmath782 and @xmath801 .
the boundary of @xmath848 has two connected components @xmath849 and there is a natural identification @xmath850 preserving oriented contact structures and the intersection line fields together with their orientations if @xmath34 are both oriented .
let @xmath851 and @xmath852 be two balls .
we will only need to modify intersection line fields on @xmath853 and @xmath854 .
therefore it is sufficient to orient @xmath34 only along @xmath853 .
from this we obtain coherent orientations of the intersection line fields on @xmath855 and @xmath856 . in @xmath853
we choose legendrian unknots @xmath317 with thurston
bennequin invariant @xmath345 and rotation number @xmath235 .
one can obtain @xmath317 by negative stabilization of the standard legendrian unknot with thurston
bennequin invariant @xmath235 , cf . .
let @xmath205 be an embedding of @xmath447 into @xmath857 which preserves oriented contact structures such that @xmath205 maps @xmath691 to @xmath325 and @xmath692 to @xmath326 .
because the rotation numbers along @xmath463 and @xmath858 equal @xmath235 , we can apply vertical modification to @xmath857 such that @xmath205 preserves oriented intersection line fields . using the discussion above
we obtain an attaching map @xmath859 for @xmath479 with the model engel structure @xmath829 which preserves oriented contact structures and the homotopy type of the intersection line field . here
we do not stick to the convention that @xmath479 is attached using an embedding of @xmath475 but notice that the orientation of the characteristic foliations match .
after a suitable vertical modification of the boundary , we obtain an engel structure on @xmath860 next we want to identify @xmath104 with @xmath350 such that oriented contact structures and the homotopy class of the intersection line fields are preserved . away from @xmath861
we can use @xmath130 .
this map has an obvious extension to a map @xmath862 which coincides with the natural identification of @xmath863 with @xmath864 .
unfortunately the contact structures on @xmath448 and @xmath475 are not preserved by this identification . however , the family of distributions defined by away from @xmath448 together with the restriction of to @xmath865 , is a family of contact structures . by theorem [ t :
gray ] there is an identification @xmath866 which preserves oriented contact structures and which coincides with @xmath130 away from a neighborhood of the attaching region of @xmath438 .
next we compare the intersection line fields on @xmath104 and @xmath350 . near @xmath867
the gray isotopy constructed in the proof of theorem [ t : gray ] is induced by the legendrian vector field @xmath868 from one sees that @xmath869 maps the legendrian curves @xmath870 to themselves .
the rotation number along these curves is @xmath235 for both model engel structures @xmath828 and @xmath829 .
the rotation number of the intersection line field with respect to the legendrian line field @xmath142 spanned by @xmath871 is zero for both @xmath828 and @xmath829 ( cf .
proposition [ p:1 modelle ] ) . since @xmath872 is tangent to @xmath142 , the flow of this vector field preserves @xmath142 .
this implies that the identification @xmath869 preserves the homotopy class of the intersection line field along @xmath481 .
now let @xmath59 be a curve in @xmath104 which intersects @xmath873 exactly once such that the endpoints of @xmath59 lie outside of @xmath874 and @xmath875 .
we have to ensure that the intersection line fields are homotopic along @xmath59 relative to the endpoints . for this
we use the flexibility mentioned in remark [ r : suitable vert mod 1 ] : choosing suitable values of @xmath184 for the vertical modification of the boundary when we attach @xmath479 , we can ensure that @xmath869 preserves the homotopy class of the intersection line field along @xmath59 relative to the endpoints .
then @xmath869 preserves the homotopy class of the intersection line field . after a vertical modification of the boundary we can identify @xmath104 with @xmath350 using @xmath869 such that we obtain an engel structure on the resulting manifold @xmath876 .
the new engel structure coincides with @xmath37 and @xmath38 away from sufficiently big neighborhoods of @xmath877 and @xmath878 .
these transverse hypersurfaces are also contained in @xmath876 .
it remains to show that @xmath876 is diffeomorphic to @xmath879 . in order to prove this
, we decompose @xmath880 and @xmath881 , where @xmath882 are ordinary handles of index @xmath883 , as in .
the left part of figure [ b : unknotkirby ] shows the attaching curve and the framing of @xmath610 after @xmath562 is attached .
one end of @xmath562 lies in @xmath874 and the other end lies in @xmath875 .
recall that @xmath438 and @xmath479 are attached to @xmath884 in a symmetric way .
if we discard @xmath610 and @xmath885 and identify the boundary components of @xmath884 with @xmath886 attached , then we obtain @xmath826 . using an isotopy of the attaching curves of @xmath610 we can separate @xmath610 , respectively @xmath885 , from @xmath562 , respectively @xmath631 , such that @xmath610 and @xmath885 are attached along symmetric unknots in @xmath857 , respectively @xmath887 , which do not meet the attaching region of @xmath562 , respectively @xmath631 . since the thurston
bennequin invariant of @xmath325 and @xmath326 is @xmath345 , the framing of the attaching curve of @xmath610 is @xmath888 with respect to the obvious seifert surface of the attaching curve .
the @xmath5handles @xmath889 are attached by a doubling construction .
the kirby diagram therefore consists of spheres representing the attaching regions of @xmath562 , the attaching curve of @xmath610 and a zero framed meridian of the attaching curve of @xmath610 , cf .
@xcite , p. 133 .
the ordinary handles of index @xmath5 therefore account for the summand @xmath750 .
the right part of figure [ b : unknotkirby ] shows the kirby diagram of @xmath884 with @xmath890 attached .
this proves the claim under the assumption that @xmath782 and @xmath801 are oriented .
if this is not the case we can nevertheless orient @xmath891 along the components of @xmath892 after we have cut along the hypersurfaces . for the proof it suffices to orient @xmath891 along @xmath893 such that it points outwards and along @xmath894 such that it points inwards .
all constructions carry over to this situation . in order to apply theorem [ t : m+n ]
, one has to find engel structures whose characteristic foliation admits a closed transversal .
this is true for the engel structures from the proof of theorem [ t : exist ]
. we can also apply theorem [ t : m+n ] to engel structures obtained by prolongation after we perturb them slightly in the neighborhood of a leaf of the characteristic foliation ( as in proposition [ p : perturbed prolong ] ) . using this corollary
one obtains engel structures on @xmath898 one can show that it is impossible to construct an engel structure on @xmath899 using the method of geiges or prolongation , although @xmath899 is actually the total space of an orientable circle bundle over a @xmath3manifold .
however , the euler class of all circle bundles with total space @xmath899 is odd ( notice that since @xmath899 is orientable , the characteristic foliation of all engel structures on @xmath899 are orientable ) .
since the tangent bundle of an orientable @xmath3manifold is trivial , @xmath899 can not arise as circle bundle of a subbundle of rank @xmath5 of the @xmath3manifold .
let us return to the proof of theorem [ t : m+n ] and discuss the meaning of the assumption that both engel structures have characteristic foliations which admit a closed transversal .
we do not make explicit use of the fact that @xmath877 and @xmath878 are _ closed _ transversals in the proof .
but implicitly , this assumption is used when we apply vertical modification of the boundary .
if @xmath877 or @xmath878 have boundary , then we have to ensure that everything is well defined at boundary points of @xmath847 when we identify the boundary components of @xmath6 .
one can replace the existence of closed transversals in theorem [ t : m+n ] by an assumption on the behavior of the engel structure along leaves of the characteristic foliation .
let @xmath900 be a leaf of the characteristic foliation of an engel structure @xmath4 .
we assume that @xmath900 is not closed .
using nearby leaves of the characteristic foliation we obtain identifications of @xmath901 with @xmath902 for all @xmath903 on @xmath900 by lemma [ l : transverse contact structure ] .
this allows us to define @xmath904\end{aligned}\ ] ] where we identify @xmath902 with @xmath901 .
this map is called _ development map _ of @xmath900 and it is an immersion since @xmath8={\mathcal { e}}$ ] , cf . @xcite . for a given orientation of @xmath900 let @xmath905 be the segments of @xmath16 which lie on the two sides of @xmath87 .
now we define the _ twisting number _ @xmath906)\cap{\mathcal { w}}^+(p)\right|~,\ ] ] i.e. @xmath907 is the number of half twists of @xmath4 around @xmath16 in @xmath15 when one moves from @xmath87 in the sense of the orientation along the leaf .
we define @xmath908 similarly .
finally , if @xmath900 is closed we define @xmath909 as the degree of @xmath910 .
now we can state a modified version of theorem [ t : m+n ] .
[ t : m+n zweite vers .
] let @xmath1 carry engel structures @xmath34 such that the characteristic foliation has non closed leaves @xmath911 through @xmath912 and @xmath913 through @xmath914 with the property @xmath915 for some positive constant @xmath276 which is independent of the engel structures .
then there is an engel structure @xmath4 on @xmath879 which coincides with @xmath34 outside of neighborhoods of @xmath916 and @xmath917 .
moreover one can choose @xmath4 such that there is a leaf of the characteristic foliation which satisfies .
if @xmath918 is closed one can replace by @xmath919 for @xmath324 .
we just give a sketch of the proof and omit the details .
let @xmath327 satisfy the assumption of the theorem .
for @xmath324 we cut @xmath846 along an embedded closed @xmath3ball @xmath847 transverse to @xmath891 , which is so small that the twisting numbers of all leaves of the characteristic foliation satisfy after we have cut out @xmath847 .
away from the boundary of @xmath847 , the resulting space is a manifold with boundary .
we can perform all constructions from the proof of theorem [ t : m+n ] as long as we do not change anything near @xmath920 .
vertical modifications of the boundary are the only operations in the proof of theorem [ t : m+n ] which affect @xmath920 .
the problematic step is the vertical modification of the boundary used for the identification of the boundary components of @xmath6 .
the vertical modification of the boundary needed at this point is not allowed to change anything near @xmath920 .
this means that the function @xmath167 used in section [ ss : vert mod ] for the description of a vertical modification of the boundary has to be zero near boundary points of @xmath847 .
the condition allows us to choose @xmath921 on a collar of the boundary of @xmath847 .
we did not determine bounds for @xmath276 .
since there are engel structures ( e.g. the standard engel structure on @xmath739 ) where the rotation number along all leaves of the characteristic foliation is zero , condition is not fulfilled in general . | we develop a construction of engel structures on @xmath0manifolds based on decompositions of manifolds into round handles .
this allows us to show that all parallelizable @xmath0manifolds admit an engel structure .
we also show that , given two engel manifolds @xmath1 satisfying a certain condition on the characteristic foliations , there is an engel structure on @xmath2 which is closely related to the original engel structures . |
one of the aims of the novel field of relativistic quantum information is to study the correlations present between field modes as seen by both inertial and accelerated observers . in this context
, there have appeared a number of works studying fermionic fields in non - inertial frames @xcite .
one of the most well - known scenarios consists in the analysis of the quantum correlations shared between an inertial mode of a fermionic field and an accelerated one , as a function of the latter s acceleration .
most of the previous literature was centred only in the analysis of pure states from the inertial perspective , where some entanglement was always found to survive in the infinite acceleration limit ( see previous citations ) .
this contrasts with the case of bosonic fields , where no field entanglement has ever been found to survive at infinite acceleration limit , and where it is possible even to completely cancel entanglement at any finite acceleration @xcite .
this survival of fermionic entanglement at finite acceleration has been usually linked to the pauli exclusion principle , which would arguably maintain some entanglement of statistical nature at any value of the acceleration @xcite .
this has led to some works showing that , for some field states , the entanglement at infinite acceleration is not usable for quantum information @xcite .
nevertheless , in some recent works , fermionic entanglement behaviour for a family of werner states has been studied @xcite finding fermionic entanglement extinction at finite acceleration for some states of this 1-parametric family . on the other hand , some other works @xcite found that if we consider a fermionic tripartite field w - state shared by two accelerated observers , @xmath0 and @xmath1 and one inertial , a , i.e. @xmath2 and then for whatever reason we trace out the inertial observer ( which is equivalent to considering a system composed of two accelerated partners ) , then the resulting density matrix @xmath3 is entangled at zero acceleration , but shows no entanglement survival for sufficiently high accelerations of the observers .
these particular cases show that the phenomenon of fermionic entanglement survival does not happen for any entangled state .
summarising , working with fermionic fields is a necessary , but not sufficient , condition to observe entanglement survival . in this work ,
we study a setting composed of two accelerated observers who watch a general mixed state of two modes of a fermionic field .
we generalise previous results about vanishing fermionic entanglement @xcite beyond the single mode approximation @xcite and their proper setting , considering both one and two accelereated observers .
we also characterise how frequently this phenomenon of entanglement extinction arises by probing the whole space of density matrices rather than restricting to particular families of entangled states .
we find that the abundance of mixed states that present the phenomenon is not negligible at all but , instead , it can be about 50% .
finally we provide a measure of the degree of purity for which the phenomenon starts to manifest .
our work is structured as follows : in section [ set ] , we present the setting under consideration and the necessary expressions for the study of field entanglement . section [ res ] contains our results about the distribution and characteristics of the phenomenon of vanishing entanglement at the infinite acceleration in general mixed states .
finally , section [ cojoniak ] contains our conclusions .
we consider two non - inertial observers in two causally disconnected patches of minkowski spacetime .
these observers move with uniform proper accelerations , which are are not necessarily equal . in other words
, they travel along lines of constant spacelike rindler coordinate .
we will call the observers rob or rodney if they are in region i , and antirob or antirodney if they are in region ii , as depicted in fig.[rindler ] . note that although we will refer to the observers by different names depending on the spacetime wedge on which they are , we will always consider only two observers at a time .
each observer probes a single mode of a grassmann scalar field @xcite , an anticommuting field with only one degree of freedom .
we will analyse entanglement between these modes for different field states associated with different observers accelerations .
the most natural way to describe the quantum field from the viewpoint of these uniformly accelerated observers is through field quantisation in the rindler basis .
this means that , whereas an inertial observer would quantise the field using minkowski modes ( positive frequency solutions to the relevant field equation in minkowski coordinates ) to build up the fock space , it is more appropriate for our observers to quantise the field using rindler modes ( positive frequency solutions in rindler coordinates ) .
this is so because any particle detector carried by the accelerated observers would couple directly to rindler rather than minkowski modes .
the existence of two causally disconnected regions as depicted in fig .
[ rindler ] implies that , for each rindler frequency , there are two different rindler modes , each of them having support only in either region i or ii . when considering a general field state both kinds of modes have to be taken into account .
to study entanglement in non - inertial frames it is convenient to use of the so - called unruh modes @xcite .
the annihilation operators associated with these modes have simple expressions in terms of rindler creation and annihilation operators for particles ( @xmath4 ) , and antiparticles ( @xmath5 ) , namely @xmath6 where @xmath7 , and @xmath8 indicates rindler frequency as measured by a rindler observer with proper acceleration @xmath9 .
this acceleration is merely a convention which amounts to the choice of a specific rindler observer in order to label frequencies ; alternatively , we could index the modes by their dimensionless rindler frequency @xmath10 .
the subscripts l and r stand for ` left ' and ` right ' modes , which are related to each other by a reversal of regions i and ii . these modes have the particularity of being linear combinations of purely positive - frequency minkowski modes .
this implies that the minkowski and unruh vacua are the same .
therefore , the hilbert space factorises as @xmath11 where each @xmath12 is annihilated by @xmath13 and @xmath14 . given ( [ product ] ) , we only need to study detailedly two of the factors in the tensor product , one for each observer , and hence from now on we will consider only the frequencies @xmath15 for rob and @xmath16 for rodney . the relevant part of the vacuum will be written as @xmath17 we will consider arbitrary unruh excitations of the form @xcite @xmath18 the case @xmath19 corresponds to the choice previously known as the single mode approximation @xcite . notice that we have endowed the fermionic fock space with a particular tensor product structure , one must be careful since the choice of a particular tensor product structure may affect entanglement .
we will assume the most extended way of building the fermionic fock @xcite space as it allows us to easily find the relevant expressions for the field states and the main results about entanglement vanishment remain true .
for a detailed study of these issues , see @xcite .
to study field entanglement , we need to express the minkowski vacuum and the particle excitations in the rindler basis .
this is easily done from eqs .
( [ umodes ] ) and ( [ reduc ] ) once we impose that the vacuum be annihilated by all four relevant unruh operators @xmath20 , @xmath21 , @xmath22 , @xmath23 .
if , following previous notation @xcite , we choose a basis for each factor space in ( [ reduc ] ) by @xmath24 where @xmath25 due to pauli exclusion principle , the unruh vacuum may be expressed in the rindler basis through ( [ reduc ] ) with ( see @xcite ) @xmath26 the excitations can now be straightforwardly computed using ( [ reduc ] ) , ( [ excs ] ) and ( [ gvac ] ) .
now we have the states necessary to study field entanglement as seen by two accelerated observers .
one last point remains before the discussion of the formalism is complete : as it is obvious from fig.[rindler ] , an uniformly accelerated observer lying in one wedge of spacetime is causally disconnected from the other .
it is therefore necessary for him to trace out the field modes which are unobservable .
since we have two such observers which can lie in either regions of rindler spacetime , two partial traces need to be taken , resulting in one of the four following situations : * rob - rodney , in which the region ii modes are traced out for both the @xmath15 and @xmath16 subspaces . *
rob - antirodney , in which region ii modes are traced for @xmath15 and region i modes are traced for @xmath16 .
* antirob - rodney , in which region i modes are traced for @xmath15 and region ii modes are traced for @xmath16 .
* antirob - antirodney , in which region i modes are traced for both @xmath15 and @xmath16 . in the next section
we will study the behaviour of entanglement in all these situations , for a general mixed state of the two field modes under consideration .
we note that , unlike in previous works @xcite where field entanglement between an inertial and an accelerated observer was considered , the inertial entanglement is not always recovered in the zero acceleration limit of our setting .
although a naive expectation might be that this should be the case , it is not because of the discontinuity in localisation of the rindler and minkowski modes : whereas the former are always localised in region i or ii of spacetime , the latter are supported in the whole spacetime .
we consider a general state described by a @xmath27 density matrix @xmath28 written in the basis @xmath29 this basis of unruh excitations is to be expressed in terms of the rindler basis by means of the results of section [ set ] .
subsequently , the appropriate partial traces are taken , leaving us with a reduced state @xmath30 containing all the observable correlations between field modes . as a distillable entanglement measure valid for mixed states ,
we employ the negativity @xcite . from now on when we refer to the negativity of a bipartition we mean the negativity of the corresponding @xmath30 matrix .
we will first study the phenomenon of entanglement extinction with a monte carlo method to probe the space of density matrices and find out the abundance of states whose entanglement dissapears for finite accelerations .
then , motivated by the observation that entanglement extinction in the fermionic case appears only for non - pure states , we will obtain a typical distance ( in the sense of degree of purity ) between maximally entangled states and the first mixed states where the phenomenon appears .
the relevant parameter space ( 4x4 state matrices ) is a 15-dimensional manifold , @xmath31 .
we relied on a monte carlo method to numerically estimate the abundance of states showing some inertial entanglement which vanishes at infinite acceleration of both partners in the rob - rodney bipartition .
our algorithm works as follows : first , we randomly generate a 4x4 matrix @xmath32 drawn from the ginibre ensemble @xcite ( normally distributed with identical variance in both real and imaginary parts of all its entries ) .
then , a state matrix @xmath28 is obtained as @xmath33 and we checked wether if it had some inertial entanglement by directly computing the negativity of the matrix @xmath28 .
if so , entanglement at infinite acceleration of both partners was studied .
the use of a matrix @xmath32 in the ginibre ensemble , together with the correspondence ( [ ensemble ] ) , induces the hilbert - schmidt measure in @xmath31 @xcite , a measure both unitarily invariant ( meaning that any two measurable subsets @xmath34 and @xmath35 related through a unitary @xmath36 such that @xmath37 have the same measure ) and obtained from a metric .
once we have a measure on @xmath31 , such questions as ` how many states show entanglement survival at infinite acceleration ? '
become meaningful . though this unitarily invariant measure is not an unique choice @xcite , its behaviour
is well understood under certain circumstances @xcite and is simple enough to generate states with . to gain some physical insight on the behaviour of the induced measure in the space of density matrices
, we generated @xmath38 states , plotting an histogram of their inertial negativities , which can be seen in figure [ histo ] .
approximately @xmath39 of these states were entangled , distributing themselves over a gaussian .
mixed states drawed from the ginibre ensemble and corresponding gaussian fit .
note that only entangled states are shown ; the fraction of separable states ( zero negativity ) is much higher ( approximately @xmath40 ) . ]
we found that from a sample of @xmath41 points , @xmath42 showed non - zero negativity , and that * for @xmath43 ( the single mode approximation ) , @xmath44 of the states with nonvanishing inertial entanglement had zero entanglement in the limit of infinite acceleration for both observers . * for @xmath45
, @xmath46 of the states with inertial nonvanishing entanglement had zero entanglement in the limit of infinite acceleration for both observers . * for @xmath47
, @xmath48 of the states with nonvanishing inertial entanglement had zero entanglement in the limit of infinite acceleration for both observers .
the results above show that the phenomenon of entanglement extinction at infinite acceleration ( and , as we shall see in subsection [ graf ] , at finite acceleration ) is fairly common and not a peculiarity of some specific family of states .
also , the percentage of states showing the phenomenon drops with @xmath49 , which is consistent with previous results that going beyond the single mode approximation could result in entanglement amplification @xcite . in this subsection ,
some of the states showing the phenomenon of vanishing entanglement at infinite acceleration are analysed thoroughly . more specifically , we computed the negativity for all four possible bipartitions for the state ( [ ira ] ) which first showed the phenomenon . considering previous results in @xcite
, it might have been reasonable to guess that entanglement is not destroyed , but rather transferred to the other bipartitions .
this would have been expectable if we considered the results obtained when analysing maximally entangled states .
we present our results in figure [ stats ] . and @xmath50 for the state ( [ ira ] ) .
the upper part of the graph corresponds to the rob - rodney bipartition , and the lower one , to rob - antirodney .
the negativity of the antirob - rodney bipartition is diagonal - symmetric to that of rob - antirodney , and the antirob - antirodney bipartition is always zero . ]
we remark that , whenever there is entanglement vanishing at some acceleration in the rob - rodney bipartition , the entanglement vanishes in all other bipartitions and for all values of acceleration as well .
we have thus established that entanglement is not merely being redistributed among the various possible bipartitions , but rather that it is lost .
although true for pure entangled states , it is not true in general that fermionic statistics directly implies entanglement survival .
we have seen that for some non - pure fermionic states , entanglement extinction due to acceleration might happen . however , it is true that entanglement survival in the infinite acceleration limit is only possible if there is some entanglement transfer to other bipartitions , as it was concluded in @xcite . note that the correlations between rob and antirodney only show up when @xmath51 is far from @xmath50 . in other words , there are only correlations in this bipartition if the accelerations or the field modes they watch are very different .
the region where entanglement is zero for all bipartitions does not include points with @xmath52 or @xmath53 .
however , for other states , such as the one given by the density matrix @xmath54,\nonumber\\\beta&= \left [ \begin{array}{cccc } 0 & -0.0089 & -0.0645 & -0.0339 \\
0.0089 & 0 & -0.0233 & 0.0149 \\ 0.0645 & 0.0233 & 0 & 0.0962 \\ 0.0339 & -0.0149 & -0.0962 & 0 \end{array }
\right].\end{aligned}\ ] ] the region of vanishing entanglement does indeed include such points .
since in the single mode approximation the limit of the unruh modes for zero acceleration yields modes confined to region i , this means that this behaviour is also present where there was only one accelerated observer .
the only simple requirement is thus to consider mixed states , as pure states do not present this behaviour . as mentioned above
, there is some entanglement in the rob - antirodney bipartition whenever rob s acceleration is higher than antirodney s and these accelerations are not close . the corresponding result but with antirob - rodney correlations is also true .
we also point out that all the states studied under the single mode approximation show the same qualitative behaviour : entanglement decreases monotonically along the diagonal up to a threshold value of @xmath55 where it hits zero , staying there onwards .
higher values of inertial entanglement correspond to higher values of this threshold .
it seems likely that the threshold can achieve all the values of @xmath55 if we allow arbitrarily low inertial entanglement , although we have not proved this .
these results are derived assuming the single mode approximation @xmath43 .
this assumption introduces a strong asymmetry between region i and region ii modes , which explains why there are no states showing the same behavior for the antirob - antirodney bipartition .
figure [ stats2 ] shows the same plot for negativity as a function of acceleration for state ( [ ira ] ) , when @xmath47 .
the behaviour is qualitatively different once the single mode approximation is relaxed ; there can be entanglement amplification maxima ( see @xcite ) and the negativity no longer gives rise to a smooth surface .
also , due to the symmetry between regions present when @xmath47 , all four bipartitions result in the same surface . and
@xmath50 for the state ( [ ira ] ) and @xmath47 .
the negativities of all bipartitions are equal since the use of unruh modes with @xmath47 makes regions i and ii completely equivalent . ]
this phenomenon of entanglement destruction at infinite acceleration has only been observed for mixed states .
since maximally entangled states are pure , this means that entanglement survival at infinite acceleration should be expected as long as we do not stray ` too far ' from maximally entangled states . to determine the precise meaning of
` too far ' , we developed the following scheme : after generating a random state matrix @xmath56 in the same way as for the monte carlo algorithm above , we computed the negativity at @xmath57 under the single mode approximation as a function of @xmath58 for the family of states @xmath59,\label{fam}\\{\left| { \psi } \right\rangle}&=\frac{1}{\sqrt{2}}\left({\left| { 00 } \right\rangle}+{\left| { 11 } \right\rangle}\right ) .\end{aligned}\ ] ] until a value of @xmath58 for which the negativity dropped to zero was found .
this defines a function @xmath60 from the space of state matrices to the real numbers .
since @xmath31 is compact it follows that @xmath61 has a minimum value , which may be regarded as a measure of how far from the maximally entangled state of our choice is the set of states with zero entanglement in the infinite acceleration limit .
the process was repeated for @xmath38 random matrices , using the same generating method as in subsection [ mcsurv ] .
the minimum value of @xmath58 was found to be @xmath62 .
this means that for any one - parameter family of states of the form ( [ fam ] ) , with a high degree of certainty , the states never become mixed enough to wipe out entanglement at infinite acceleration before this value of @xmath58 is attained .
this further confirms that the phenomenon is linked to non - coherent states , since it does not show up near entangled states .
we have studied exhaustively the phenomenon reported in @xcite where it was shown that when one or two observers are accelerating some states show no entanglement survival in the infinite acceleration limit , contrary to previous insights that entanglement survival was the hallmark and a direct consequence of fermionic statistics .
we have found that the phenomenon does not require two accelerated observers to show up , in accordance with @xcite , and that the only fundamental requisite is to consider mixed states .
this general study is the first work in relativistic quantum information which explicitly considers aribitrary mixed states from the inertial perspective .
we also performed our study beyond the single mode approximation .
we characterised the phenomenon in a number of ways , finding it to be fairly common in the space of states of 4x4 density matrices , analysing its general features and providing a reasonable measure on the degree of nonpurity necessary to observe the effect .
we found that entanglement is not transferred from one bipartition to another , but rather is completely erased .
there are fundamental differences in the way bosonic and fermionic fields behave with acceleration , which have traditionally regarded as having a profound impact on the more detailed aspects of field entanglement . here
we have studied all the possible bipartitions to which entanglement could have relocated , showing that entanglement completely disappears from our states and thus the phenomenon of fermionic entanglement survival is dependent on the detailed features of the field state , and not only on the form of the change of basis between minkowski and rindler bases .
e. m - m was supported by the csic jae - predoc2007 grant scheme .
e. m - m and j. l are partially supported by the spanish micinn project fis2008 - 05705/fis and the quitemad consortium .
27ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' ( ) , @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we study families of fermionic field states in non - inertial frames which show no entanglement survival in the infinite acceleration limit .
we generalise some recent results where some particular examples of such states where found .
we analyse the abundance and characteristics of the states showing this behaviour and discuss its relation with the statistics of the field .
we also consider the phenomenon beyond the single mode approximation . |
fundamental determination of stellar masses from binary orbits is a most classical astrophysical discipline , last comprehensively reviewed by andersen ( 1991 ; 1998 ) . besides defining a mean mass - luminosity relation , used at many places in astronomy to approximately convert the observable stellar light to the underlying mass ( for instance to derive an initial mass function ) , accurate stellar masses in multiple systems
provide what is perhaps the most demanding and fundamental test of stellar evolution theory ( e.g. andersen 1991 ) .
mass , the basic input of evolutionary models , is directly measured and the models must reproduce the effective temperatures and radii ( or luminosities ) of both components , for a single age and a single chemical composition . given the strong dependence of all stellar parameters on mass , this discriminating diagnostic however only shows its power for relative mass errors @xmath01 - 2% . in practice
this has up to now restricted its use to double - lined detached eclipsing binaries .
these systems are however relatively rare : only 44 have yielded masses accurate enough to be included in andersen s ( 1991 ) critical compilation , and few have appeared in the literature since then .
tidally induced rotational mixing may in addition affect the evolution of the short period eclipsing systems , perhaps sufficiently that they are not completely representative of isolated stars .
more seriously however , detached eclipsing systems poorly fill some interesting areas in the hr diagram .
the lower main sequence is one major region with very few known eclipsing systems , as a result of the strong observational and intrinsic biases against observing eclipses in faint and physically small stars .
only three well detached eclipsing binaries are known with significantly subsolar component masses : yy gem ( m0ve , 0.6 + 0.6 ; bopp 1974 ; leung & schneider 1978 ) , cm dra ( m4ve , 0.2 + 0.2 ; lacy 1977 , metcalfe et al .
1996 ) , and the recently identified gj 2069a ( m3.5ve , 0.4 + 0.4 ; delfosse et al .
detailed observational checks of evolutionary models ( e.g. paczynski & sienkiewicz 1984 ; chabrier & baraffe 1995 ) have therefore heavily relied on the first two of these systems , even though they are in some respects non - ideal for this purpose : both binaries have two nearly equal mass components , so that the strength of the differential comparison of two stars with different masses but otherwise equal parameters is largely lost . also , all three are chromospherically very active , due to tidal synchronisation of their rotation with the short orbital period . as a consequence
, they may have untypical colours for their bolometric luminosity .
angularly resolved spectroscopic binaries provide stellar masses in parts of the hr diagram where eclipsing systems are missing , though to date these systems have not matched the @xmath11% accuracy which can be obtained in detached eclipsing systems . for m dwarfs in particular , the best representation to date of the empirical m - l relation ( henry & mc carthy 1993 , hereafter hmcc ; henry , franz , wasserman et al . , 1999 ) still has to rely in part on some fairly noisy mass determinations . for several years ( perrier et al .
1992 ) we have therefore been following up with high angular resolution some low mass spectroscopic systems found with the coravel or elodie radial velocity spectrographs .
this follow - up initially used one - dimensional ( 1d ) ir speckle , then two - dimensional ( 2d ) ir speckle , and now uses adaptive optics imaging .
as a progress report on this program and as an illustration of our methods , we present here much improved parameters for the double - lined interferometric binary gl 570bc .
the @xmath11% accuracy for the 2 masses improves by an order of magnitude on our earlier measurements ( mariotti et al .
1990 ) of this system , and is getting close to what is obtainable for eclipsing systems .
the gl 570 system comprises the _
v_=5.7 k4v star gl 570a ( also hr 5568 , hd 131977 , hip 73184 , fk5 1391 ) , and at a projected distance of 25@xmath2 the close gl 570bc pair ( also hd 131976 , hip 73182 ) which is the subject of the present paper . as discussed below , the orbital period of the close pair is fairly short , only 10 months .
thanks to its small distance of only 6 pc it is nonetheless usually well resolved by the diffraction limit of 4m - class telescopes . the separation within gl 570bc on the other hand always remains less than 0.2@xmath2 , so that all seeing - limited mesurements refer to integrated properties of the close pair .
its integrated magnitude is _
v_=8.09 , and its joint spectral type is m1v ( henry et al .
1994 ; reid et al .
the three components have common parallax , proper motion , and radial velocity .
they are therefore gravitationnally bound , with the projected separation of the wide pair ( about 120 au at the distance of the system ) corresponding to p @xmath1 500 years .
formation of the system could either result from the fragmentation of a single gas clump or have involved some capture(s ) .
the latter process however only remains efficient at the high densities which characterize very rich star forming clusters , whose lifetimes are very much shorter than the hydrogen burning timescale in a k4 dwarf . for all practical purposes , the three stars can thus in both cases be considered as coeval and as formed from the same gas .
a variety of astrometric observations of different kinds and of different quality have been collected for gliese 570bc over the years .
the close pair was first resolved by mariotti et al .
( 1990 ) who obtained infrared 1d speckle observations on three occasions and derived a first visual orbit .
we refer to their paper for the description of the instrument and the observing and data reduction procedures .
hmcc later measured gl 570bc once with a 2d infrared speckle instrument . for readers convenience , we list these published measurements in table [ tab_oa ] together with our new 2d speckle and adaptive optics ( hereafter ao ) observations .
two measurements were obtained in february 1991 and april 1991 using the speckle mode of the 2d infrared imagers then installed respectively at the kpno 4.2 m telescope and the cfht 3.6 m telescope .
each imager was designed to permit acquisition of exposures short enough to substantially freeze the seeing under standard atmospheric conditions ( @xmath3 50 to 100 ms ) in bands _
h _ and _ k _ , independently of the overhead due to read - out time and data transfer time .
several sequences of a few hundred such short exposures were obtained , alternating every few minutes between the source and a nearby unresolved star ( usually gl 570a ) used as a point spread function ( psf ) calibrator .
the whole observation took one hour or less . in principle , this observing procedure allows an almost simultaneous psf calibration , and consequently estimates the visibility modulus with 1 to 5% accuracy .
we used a software package written specifically for this type of data reduction by e. tessier ( tessier et al .
it produces an unbiased visibility estimator for the source and then extracts the binary parameters and their estimated variance from this visibility .
the actual detector scale and position angle ( p.a . ) origin for those observations were calibrated from observations of the astrometric binary @xmath4 aqr ( heintz 1989 ) . 0.1 cm .angular separation measurements [ cols="^,>,^,^ " , ] [ tab_mag ] the difference in brightness between the components of gl 570bc has been measured on several occasions in different bandpasses .
mariotti et al . ( 1990 ) obtained magnitude differences of @[email protected] , @[email protected] , @[email protected] , and @[email protected] from their 1d scanning speckle observations .
hmcc published another estimate in the j band , @[email protected] .
our own adaptive optics measurements provide @[email protected] , as well as @[email protected] in a narrow band br@xmath11 ( 2.166@xmath12 ) filter .
we estimate a correction of @xmath13=0.04 , using synthetic photometry generated from nextgen ( hauschild , allard , & baron 1999 ) model spectra of effective temperatures that bracket those apropriate for m dwarfs of the luminosities of the two components of gl 570bc .
this provides a second determination of the k band flux ratio , @[email protected] .
while those measurements are mutually consistent for every filter , their run with wavelenghth is only marginally compatible with the known ir colours of m dwarfs ( leggett 1992 ) .
the @xmath14 and ( particularly ) @xmath15 colours of early and mid - m dwarfs define a remarkably flat plateau ( leggett 1992 ) : when the spectral type of solar metallicity stars ( apropriate for gl 570bc ) varies between m0v ( [email protected] ) and m5.5v ( [email protected] ) , @xmath14 only changes from 0.68 to 0.57 , and @xmath15 just from 0.85 to 0.87 ( leggett 1992 ) the two components of gl 570bc respectively have [email protected] and [email protected] and their @xmath14 and @xmath15 colours are therefore expected to only differ by about -0.04 and + 0.01 .
@xmath5 should therefore be almost identical to @xmath8 , whereas the observations give @xmath5-@[email protected] , and @xmath5-@xmath7 should be -0.04 , whereas the observations indicate @xmath5-@[email protected] . as the integrated colours of the system
match the expected values , this inconsistency must come from some of the measured magnitude differences .
we suspect that it traces back to an overestimated contrast in the j and h band observations of both mariotti et al .
( 1990 ) and hmcc , since speckle techniques have a known bias towards underestimating the relative flux of faint components ( perrier 1988 ) .
the k band adaptive optics observations are in principle immune to this bias , and the k band speckle observations of mariotti et al .
( 1990 ) should be relatively unaffected , thanks to the low d/@xmath17 ( where d is the telescope diameter and @xmath17 is the fried parameter of the atmosphere ) at this longer wavelength .
we therefore tentatively adopt as the basis of our magnitude difference system the mean of the three k band measurements , @[email protected] . from this value
we then derive preferred values of @xmath5=1.19 and @xmath7=1.15 , but we will also consider the published j and k flux ratios as an alternative . this discrepancy contributes significant uncertainty to the analysis , and better measurements of the flux ratios at @xmath18 and @xmath19 would be of considerable interest . to date all our adaptive optics measurements in the @xmath18 and @xmath19 bands
have unfortunately been obtained at phases when the secondary star overlaps the first airy ring of the primary for these wavelengths .
these circumstances maximize the uncertainties in differential photometry from partially corrected adaptive optics images ( veran et al . 1999 ) , and these data therefore contribute no useful information on the flux density ratio .
absolute magnitudes were derived from the parallax and the apparent magnitudes of the individual stars .
the absolute magnitudes of the brighter star have uncertainties which are dominated by those of the photometry of the system , while those for the secondary have some contribution from the magnitude difference .
the parallax does nt appreciably contribute in either case .
the accurate masses and absolute magnitudes that we have obtained for the gl 570bc system represent a new benchmark for model calculations ( e.g. baraffe et al . 1998 ) and an independent check of the empirical mass - luminosity relations ( hmcc ) . the constraints which they bring to the models are largely complementary to those coming from the eclipsing binaries , whose absolute radii can be determined very accurately but whose larger distances on the other hand contribute significant uncertainties to the absolute magnitudes . as long emphasized by theoreticians , and by observers of more massive stars ( e.g. andersen 1991 )
, there is however no such thing as one single mass - luminosity relation : stellar luminosities depend on chemical composition as well as on mass ( in general they depend on age too , though not in the age and mass range discussed here ) .
quantitative metallicity determinations however are notoriously difficult for m dwarfs ( e.g. viti et al . 1997 ; valenti et al .
observers in this field usually have to resort to photometric metallicity estimators ( leggett 1992 ) which are only approximately calibrated , or otherwise assume by default a solar metallicity .
thanks to its physical association with the hotter gl 570a ( k4v ) , the gl 570bc pair represents a rare case of two m dwarfs with a very well determined spectroscopic metallicity .
its accurate masses are thus fortunately matched with excellent metallicities .
hearnshaw ( 1976 ) first measured the metallicity of gl 570a from high resolution electronographic spectra and obtained [ fe / h][email protected] .
more recently , feltzing & gustafsson ( 1998 ) measured [ fe / h][email protected](random)@xmath60.1(systematic ) from high s / n r=10@xmath20 echelle ccd spectra .
these authors find some evidence for nlte overionisation of fe into fe@xmath21 , but the derived elemental fe abundance is unaffected , as fe is overwhelmingly neutral in the photosphere of a k4 dwarf .
quite conveniently for comparison with published models , the gl 570 system thus has a truly solar metallicity .
figure [ ml_plot ] compares the mass and luminosity of the two components of gl 570bc with the baraffe et al .
( 1998 ) 10 gyr solar - metallicity isochrone .
these models consistently combine stellar evolution models ( e.g. chabrier & baraffe 1997 ) and non - grey atmospheric models ( allard & hauschild 1995 ; hauschild et al .
the present generation of these evolutionary models still uses non - dusty atmospheres , but dust only becomes relevant at effective temperatures significantly lower than those of gl 570bc ( allard 1998 ) .
the baraffe et al .
models are consistently slightly brighter in all 3 bands than the two stars , by 0.08 to 0.15 magnitude . while this level of agreement is already very comforting , the discrepancies are significant at the @xmath13@xmath22 levels and may point towards remaining low level deficiencies of the theoretical models . if one adopts the measured j and h band flux ratios , rather than the smaller values that we have estimated from the k band measurement , the agreement with the models improves only slightly for the primary star , as its magnitude only weakly depends on the exact value of the large flux ratio . at the same time this choice degrades
the agreement for the fainter star by a larger factor , and the overall agreement with the models is significantly worse .
figure [ ml_plot ] also shows the data points of hmcc , as well as their analytic representation of those data .
the agreement is essentially perfect with the @xmath18 band hmcc relation , while the @xmath19 and @xmath23 band relations are slightly discrepant , by respectively 0.1 and 0.15 magnitudes .
we note that the hmcc mass - luminosity relations are only consistent with the empirical m dwarf colours of leggett ( 1992 ) at this 0.100.15 magnitude level , even though the hmcc photometry mostly traces back to leggett ( 1992 ) .
this is because the hmcc mass - magnitude relations for @xmath18 , @xmath19 and @xmath23 were adjusted independently , without explicit forcing of colour consistency .
the perfect agreement with the @xmath18 band hmcc fit is therefore probably fortuitous to some extent , and the 0.10 to 0.15 discrepancy for the @xmath19 and @xmath23 bands probably represents a more realistic estimate of the accuracy of those analytic fits around 0.5 .
we dedicate this paper to the memory of our two late colleagues , antoine duquennoy and jean marie mariotti , who started this program with two of us ( cp , mm ) and pushed it forward until their untimely death .
we are indebted to the many coravel observers who contributed to measurements of gl 570bc over the years , and to didier queloz , dominique naef and nuno santos who obtained some elodie and coralie radial velocities for us at critical orbital phases .
we also warmly thank the colleagues who assisted us for the 2d speckle or ao observations : steve ridgway and julian christou who participated in the kpno 2d speckle observation and ron probst for the operation of the infrared imager , daniel rouan , franois lacombe and their group for the assistance with the cfht circus infrared camera and e. tessier who wrote and operated its speckle acquisition and reduction software , the operators and the support and infrared teams of eso , la silla for their help during come - on+ observations .
we are grateful to the anonymous referee for constructive comments and clarification requests , which led to an improved paper .
beuzit j .-
l . , demailly l. , gendron e. , gigan p. , lacombe f. , rouan d. , hubin n. , bonaccini d. , prieto e. , chazallet f. , rabaud d. , madec p .- y .
, rousset g. , hofmann r. , eisenhauer f. , 1997 , experimental astronomy 7 , 285 doyon r. , nadeau d. , valle p. , starr b.m .
, cuillandre j .- c .
, beuzit j .-
l . , beigbeder f. , brau - nogu s. , 1998 , in spie proceedings 3354 , `` infrared astronomical instrumentation '' , ed .
a.m. fowler , 760 .
hummel c.a .
, armstrong j.t .
1992 , in `` eso conference on high - resolution imaging by interferometry ii .
ground - based interferometry at visible and infrared wavelengths '' , j.m .
beckers , f. merkle eds , european southern observatory , garching bei munchen , germany .
mariotti , j .-
m . , perrier , c. , duquennoy , a. , rigaut , f. , gehring , g. , gallais , p. 1991
, eso conf . on high - resolution imaging by interferometry .
ii p99 - 107 .
mc caughrean m.j . ,
stauffer j.r .
1994 , aj 108 , 1382 .
perrier c. , duquennoy a. , mariotti j.m . ,
mayor m. , morbey c. , 1992 , p. 543 . in `` complementary approaches to double and multiple star research '' , asp conference series , vol .
32 , iau colloquium 135 , mcalister h.a . and hartkopf w.i .
, eds . tessier , e. , bouvier , j. , lacombe f. 1994 , a&a 283 , 827 .
tokovinin a. 1992 , in complementary approaches to double and multiple stars research , iau colloquium 135 , asp conf .
series , vol 32 , edited by h.a .
mc alister and w.i .
hartkopf ( asp , san francisco ) , p573 .
2400000 + & ( in km / s ) & ( in km / s ) + + 43578.641 & 27.250 @xmath6 .680 & + 43591.613 & 28.040 @xmath6 .610 & + 43631.521 & 30.800 @xmath6 .830 & + 43686.390 & 33.750 @xmath6 .750 & + 43908.694 & 29.470 @xmath6 .790 & + 43996.501 & 34.810 @xmath6 .780 & + 44024.440 & 36.310 @xmath6 .570 & + 44042.420 & 37.070 @xmath6 .760 & + 44053.380 & 37.910 @xmath6 .830 & + 44307.656 & 34.990 @xmath6 .770 & + 44341.571 & 36.520 @xmath6 .780 & + 43918.672 & 28.960 @xmath6 .650 & + 44690.552 & 38.240 @xmath6 .510 & + 45001.730 & 36.270 @xmath6 .660 & + 45044.641 & 13.930 @xmath6 .560 & + 45130.422 & 27.490 @xmath6 .730 & + 45152.383 & 30.470 @xmath6 .590 & + 45413.611 & 26.180 @xmath6 .460 & + 45437.585 & 27.680 @xmath6 .520 & + 45449.524 & 28.050 @xmath6 .620 & + 45466.479 & 28.900 @xmath6 .450 & + 45476.530 & 30.800 @xmath6 .670 & + 45728.740 & 26.210 @xmath6 .470 & + 45763.694 & 28.520 @xmath6 .620 & + 45782.615 & 30.130 @xmath6 .590 & + 46133.650 & 31.070 @xmath6 .500 & + 46212.426 & 37.910 @xmath6 .700 & + 46220.405 & 37.840 @xmath6 .630 & + 46267.353 & 6.500 @xmath6 .920 & + 46268.346 & 6.580 @xmath6 .770 & + 46268.356 & 6.050 @xmath6 .770 & + 46271.357 & 9.630 @xmath6 .570 & + 46272.355 & 9.400 @xmath6 .520 & + 46274.347 & 11.930 @xmath6 .610 & + 46278.334 & 12.640 @xmath6 .620 & + 46279.342 & 13.270 @xmath6 .680 & + 46269.345 & 7.290 @xmath6 .630 & + 46519.571 & 36.560 @xmath6 .540 & + 46521.618 & 37.110 @xmath6 .500 & + 46540.557 & 38.410 @xmath6 .510 & + 46551.513 & 36.080 @xmath6 .530 & + 46552.492 & 34.020 @xmath6 .450 & + 46560.473 & 23.410 @xmath6 .440 & + 46561.475 & 19.070 @xmath6 .540 & + 46563.485 & 11.780 @xmath6 .410 & + 46564.489 & 8.180 @xmath6 .450 & + 46568.454 & 1.200 @xmath6 .380 & + 46569.457 & 1.170 @xmath6 .430 & + 2400000 + & ( in km / s ) & ( in km / s ) + + 46570.436 & 1.460 @xmath6 .460 & + 46572.456 & .580 @xmath6 .640 & + 46574.456 & 4.100 @xmath6 .590 & + 46576.449 & 3.310 @xmath6 .780 & + 46582.437 & 10.310 @xmath6 .520 & + 46587.414 & 12.590 @xmath6 .520 & + 46595.357 & 15.960 @xmath6 .810 & + 46602.395 & 18.200 @xmath6 .470 & + 46602.423 & 17.860 @xmath6 .470 & + 46611.413 & 19.290 @xmath6 .520 & + 46621.389 & 21.320 @xmath6 .490 & + 46883.575 & 4.330 @xmath6 .680 & + 46886.604 & 6.730 @xmath6 .570 & + 46908.527 & 17.470 @xmath6 .480 & + 46305.497 & 20.700 @xmath6 .480 & + 46316.488 & 22.340 @xmath6 .490 & + 48024.672 & 35.460 @xmath6 .410 & + 48448.664 & 16.380 @xmath6 .310 & + 48449.603 & 16.360 @xmath6 .290 & + 48463.505 & 19.090 @xmath6 .450 & + 48469.573 & 21.680 @xmath6 .310 & + 48696.881 & 37.040 @xmath6 .430 & + 48704.848 & 37.380 @xmath6 .330 & + 48732.747 & 1.630 @xmath6 .320 & + 48871.507 & 30.790 @xmath6 .470 & + 48878.514 & 30.270 @xmath6 .540 & + 49059.893 & 14.070 @xmath6 .310 & + 49066.821 & 16.650 @xmath6 .300 & + 49082.824 & 20.190 @xmath6 .450 & + 49116.709 & 25.390 @xmath6 .340 & + + 50524.647 & 37.252 @xmath6 .029 & 15.423 @xmath6 .124 + 50561.530 & 37.843 @xmath6 .026 & 14.718 @xmath6 .099 + 50576.491 & 20.394 @xmath6 .027 & 41.514 @xmath6 .093 + 50587.458 & 3.679 @xmath6 .034 & 65.622 @xmath6 .148 + 50588.455 & 4.366 @xmath6 .038 & 64.358 @xmath6 .156 + 50627.353 & 21.035 @xmath6 .028 & 40.644 @xmath6 .097 + 50837.711 & 37.502 @xmath6 .031 & 15.162 @xmath6 .144 + 50839.715 & 37.627 @xmath6 .033 & 14.939 @xmath6 .140 + 50851.704 & 38.237 @xmath6 .031 & 14.096 @xmath6 .117 + 50852.707 & 38.283 @xmath6 .035 & 14.000 @xmath6 .126 + 50853.704 & 38.273 @xmath6 .033 & 13.919 @xmath6 .127 + 50854.704 & 38.309 @xmath6 .030 & 13.906 @xmath6 .121 + 50857.713 & 38.417 @xmath6 .033 & 13.769 @xmath6 .129 + 50886.622 & 15.867 @xmath6 .023 & 47.349 @xmath6 .086 + 50889.637 & 5.941 @xmath6 .036 & 62.600 @xmath6
.125 + 50890.645 & 3.907 @xmath6 .033 & 65.285 @xmath6 .131 + 2400000 + & ( in km / s ) & ( in km / s ) + + 51197.874 & 7.546 @xmath6 .030 & 60.549 @xmath6 .120 + 51198.870 & 5.057 @xmath6 .030 & 63.986 @xmath6 .130 + 51200.880 & 2.391 @xmath6 .030 & 68.186 @xmath6 .120 + 51207.859 & 5.868 @xmath6 .030 & 62.853 @xmath6 .120 + 51209.886 & 7.488 @xmath6 .030 & 60.661 @xmath6 .130 + 51211.778 & 8.883 @xmath6 .030 & 58.694 @xmath6 .120 + | we present very accurate individual masses ( 1.2% relative accuracy ) for the two components of gl 570bc , an interferometric and double - lined spectroscopic binary system .
they were obtained from new high accuracy radial velocity and angular separation measurements , analysed together with previously published measurements . from those data we determine a much improved orbit through a simultaneous least square fit to the radial velocity , visual , and parallax information .
the derived masses and absolute magnitudes generally validate the theoretical and empirical mass - luminosity relations around 0.5 , but point towards some low level discrepancies at the 0.1 to 0.15 magnitude level .
forthcoming results of this observing program will extend the comparison to much lower masses with similar accuracy . |
the knapsack problem is one of the famous tasks in combinatorial optimization .
items and each of them has a profit of @xmath2 measuring the usefulness of this item during the trip and a size @xmath3 .
a natural constraint is that the total size of all selected items must not exceed the capacity @xmath4 of the knapsack .
the aim of the hitchhiker is to select a subset of items while maximizing the overall profit under the capacity constraint . ] in the knapsack problem ( kp ) we are given a set @xmath5 of @xmath6 items .
every item @xmath7 has a profit @xmath2 and a size @xmath3 .
further there is a capacity @xmath4 of the knapsack .
the task is to choose a subset @xmath8 of @xmath5 , such that the total profit of @xmath8 is maximized and the total size of @xmath8 is at most @xmath4 . within the d - dimensional knapsack problem ( d - kp ) a set a of @xmath6 items and
a number @xmath0 of dimensions is given .
every item @xmath7 has a profit @xmath2 and for dimension @xmath9 the size @xmath10 .
further for every dimension @xmath9 there is a capacity @xmath11 .
the goal is to find a subset @xmath8 of @xmath5 , such that the total profit of @xmath8 is maximized and for every dimension @xmath9 the total size of @xmath8 is at most the capacity @xmath11 .
further we consider the multiple knapsack problem ( mkp ) where beside @xmath6 items a number @xmath1 of knapsacks is given .
every item @xmath7 has a profit @xmath2 and a size @xmath3 and each knapsack @xmath9 has a capacity @xmath11 .
the task is to choose @xmath1 disjoint subsets of @xmath5 such that the total profit of the selected items is maximized and each subset can be assigned to a different knapsack @xmath9 without exceeding its capacity @xmath11 by the sizes of the selected items .
surveys on the knapsack problem and several of its variants can be found in books by kellerer et al .
@xcite and by martello et al . @xcite .
the knapsack problem arises in resource allocation where there are financial constraints , e.g. capital budgeting .
capital budgeting problems have been introduced in the 1950s by lorie and savage @xcite and also by manne and markowitz @xcite and a survey can be found in @xcite . from a computational point of view
the knapsack problem is intractable @xcite .
this motivates us to consider the fixed - parameter tractability and the existence of kernelizations of knapsack problems . beside the standard parameter @xmath12 ,
i.e. the threshold value for the profit in the decision version of these problems , and the number of items , knapsack problems offer a large number of interesting parameters . among these
are the sizes , the profits , the number of different sizes , the number of different profits , the number @xmath0 of dimensions , the number @xmath1 of knapsacks , and combined parameters on these .
such parameters were considered for fixed - parameter tractability of the subset sum problem , which can be regarded as a special case of the knapsack problem , in @xcite and in the field of kernelizaton in @xcite .
this paper is organized as follows . in section [ sec - pre ] , we give preliminaries on fixed - parameter tractability and kernelizations , which are two equivalent concepts within parameterized complexity theory .
we give a characterization for the special case of polynomial fixed - parameter tractability , which in the case of integer - valued problems is a super - class of the set of problems allowing polynomial time algorithms .
we show that a parameterized problem can be solved by a polynomial fpt - algorithm if and only if it is decidable and has a kernel of constant size .
this implies a tool to show kernels of several knapsack problems .
further we cite a useful theorem for finding kernels of knapsack problems with respect to parameter @xmath6 by compressing large integer values to smaller ones .
we also give results on the connection between the existence of parameterized algorithms , approximation algorithms , and pseudo - polynomial algorithms . in section [ sec - kp ]
, we consider the knapsack problem .
we apply known results as well as our characterizations to show fixed - parameter tractability and the existence of kernelizations . in section [ sec - mkp ] ,
we look at the d - dimensional knapsack problem .
we show that the problem is not pseudo - polynomial in general by a pseudo - polynomial reduction from independent set , but pseudo - polynomial for every fixed number @xmath0 of dimensions .
we give several parameterized algorithms and conclude bounds on possible kernelizations . in section [ def - sec - mkp ] , we consider the multiple knapsack problem .
we give a dynamic programming solution and a pseudo - polynomial reduction from 3-partition in order to show that the problem is not pseudo - polynomial in general , but for every fixed number @xmath1 of knapsacks .
further we give parameterized algorithms and bounds on possible kernelizations for several parameters . in the final section [ sec - con ]
we give some conclusions and an outlook for further research directions .
in this section we recall basic notations for common algorithm design techniques for hard problems from the textbooks @xcite , @xcite , @xcite , and @xcite . within parameterized complexity
we consider a two dimensional analysis of the computational complexity of a problem . denoting the input by @xmath13 , the two considered dimensions are the input size @xmath14 and the value of a parameter @xmath15 , see @xcite and @xcite for surveys .
let @xmath16 be a decision problem and @xmath17 the set of all instances of @xmath16 .
parameterization _ or _ parameter _ of @xmath16 is a mapping @xmath18 that is polynomial time computable .
the value of the parameter @xmath15 is expected to be small for all inputs @xmath19 .
parameterized problem _ is a pair @xmath20 , where @xmath16 is a decision problem and @xmath21 is a parameterization of @xmath16 .
for @xmath20 we will also use the abbreviation @xmath21-@xmath16 .
an algorithm @xmath5 is an _ fpt - algorithm with respect to @xmath21 _ , if there is a computable function @xmath22 and a constant @xmath23 such that for every instance @xmath24 the running time of @xmath5 on @xmath13 is at most @xmath25 or equivalently at most @xmath26 , see @xcite . for the case
where @xmath27 is also a polynomial , @xmath5 is denoted as _
polynomial fpt - algorithm with respect to @xmath21_. a parameterized problem @xmath20 belongs to the class @xmath28 and is called _ fixed - parameter tractable _ , if there is an fpt - algorithm with respect to @xmath21 which decides @xmath16 .
typical running times of an fpt - algorithm w.r.t .
parameter @xmath21 are @xmath29 and @xmath30 .
a parameterized problem @xmath20 belongs to the class @xmath31 and is _ polynomial fixed - parameter tractable _
@xcite ) , if there is a polynomial fpt - algorithm with respect to @xmath21 which decides @xmath16 .
please note that polynomial fixed - parameter tractability does not necessarily imply polynomial time computability for the decision problem in general .
a reason for this is that within integer - valued problems there are parameter values @xmath15 which are larger than any polynomial in the instance size @xmath14 .
an example is parameter @xmath32 for problem knapsack in section [ sec - para - kp ] . on the other hand , for small parameters polynomial
fixed - parameter tractability leads to polynomial time computability .
let @xmath20 be some parameterized problem and @xmath4 be some constant such that for every instance @xmath13 of @xmath16 it holds @xmath33 .
then the existence of a polynomial fpt - algorithm with respect to @xmath21 implies a polynomial time algorithm for @xmath16 .
let @xmath5 be some polynomial fpt - algorithm with respect to @xmath21 for @xmath20 .
then @xmath5 has a running time of @xmath34 for two constants @xmath0 , @xmath35 .
since @xmath33 we obtain a running time which is polynomial in @xmath14 . in order to state lower bounds we give the following corollary .
[ cor - pfpt - p ] let @xmath20 be some parameterized problem and @xmath4 be some constant such that @xmath16 is np - hard and for every instance @xmath13 of @xmath16 it holds @xmath33 . then there is no polynomial fpt - algorithm with respect to @xmath21 for @xmath20 , unless @xmath36 .
an algorithm @xmath5 is an _ xp - algorithm with respect to @xmath21 _ , if there are two computable functions @xmath37 such that for every instance @xmath24 the running time of @xmath5 on @xmath13 is at most @xmath38 a parameterized problem @xmath20 belongs to the class @xmath39 and is called _ slicewise polynomial _ , if there is an xp - algorithm with respect to @xmath21 which decides @xmath16 .
typical running times of an xp - algorithm w.r.t .
parameter @xmath21 are @xmath40 and @xmath41 . in order to show fixed - parameter intractability
, it is useful to show the hardness with respect to one of the classes @xmath42 $ ] for some @xmath43 , which were introduced by downey and fellows @xcite in terms of weighted satisfiability problems on classes of circuits . the following relations
the so called _ @xmath44-hierarchy _ hold and all inclusions are assumed to be strict .
@xmath45 \subseteq { \mbox{w}}[2 ] \subseteq \ldots \subseteq
{ \mbox{xp}}\ ] ] in the case of hardness with respect to some parameter @xmath21 a natural question is whether the problem remains hard for _ combined _ parameters , i.e.parameters @xmath46 that consists of @xmath47 parts of the input .
the given notations can be carried over to combined parameters , e.g. an fpt - algorithm with respect to @xmath46 is an algorithm of running time @xmath48 for some constant @xmath4 and some computable function depending only on @xmath49 .
next we consider the question , whether the sizes and the values can be reduced , such that their bit - length is bounded polynomially in a given parameter .
let @xmath20 be a parameterized problem , @xmath17 the set of all instances of @xmath16 and @xmath18 a parameterization for @xmath16 .
a polynomial time transformation @xmath50 is called a _ kernelization _ for @xmath20 , if @xmath27 maps a pair @xmath51 to a pair @xmath52 , such that the following three properties hold . * for all @xmath24 it holds @xmath13 is a yes - instance for @xmath16 if and only if @xmath53 is a yes - instance for @xmath16 . *
* there is some function @xmath55 , such that @xmath56 . depends only on the parameter @xmath15 and not on the size of @xmath13 . ]
the pair @xmath52 is called _ kernel _ for @xmath20 and @xmath57 is the _ size of the kernel_.
if @xmath58 is a polynomial , linear , or constant function of @xmath21 , we say @xmath52 is a _ polynomial _ , _ linear _ , or _ constant kernel _ , respectively , for @xmath20 .
next we show that fpt - algorithms lead to kernels .
although the existence is well known ( theorem 1.39 in @xcite ) , we give a proof since we need the kernel size later on .
[ th - ker ] let @xmath20 be some parameterized problem .
if there is an fpt - algorithm that solves @xmath20 for every instance @xmath13 in time @xmath59 , then @xmath16 is decidable and there is a kernel of size @xmath60 for @xmath20 .
let @xmath61 and @xmath5 be an fpt - algorithm with respect to @xmath21 which runs on input @xmath13 in time @xmath62 for some function @xmath27 and some constant @xmath4 . w.l.o.g .
we assume that there is one constant size yes - instance @xmath63 and one constant size no - instance @xmath64 of @xmath16 .
the following algorithm @xmath8 computes a kernelization for @xmath20 .
algorithm @xmath8 simulates @xmath65 steps of algorithm @xmath5 . if during this time @xmath5 stops and accepts or rejects , then @xmath8 chooses @xmath63 or @xmath64 , respectively , as the kernel . otherwise we know that @xmath66 and thus @xmath67 and @xmath8 states @xmath51 as kernel .
algorithm @xmath8 has a running time in @xmath68 and leads to a kernel of size @xmath69 .
the existence of an fpt - algorithm is even equivalent to the existence of a kernelization for decidable problems @xcite .
[ th - fg ] for every parameterized problem @xmath20 the following properties are equivalent : 1 .
@xmath61 2 .
@xmath16 is decidable and @xmath20 has a kernelization .
thus for fixed - parameter tractable problems the existence of kernels of _ polynomial _ size are of special interest . for a long time polynomial kernels
only were known for parameterized problems obtained from optimization problems with the standard parameterization ( i.e. problem @xmath12-@xmath16 defined in section [ sec - approx ] ) . in this paper
we will give a lot of examples for further parameters which lead to polynomial kernels for knapsack problems .
for the special case where @xmath27 is a polynomial theorem [ th - ker ] implies that polynomial fpt - algorithms lead to polynomial kernels .
[ th - polyker ] let @xmath20 be some parameterized problem . if @xmath70 , then @xmath16 is decidable and there is a polynomial kernel for @xmath20 .
a remarkable difference to the relation of theorem [ th - fg ] is that the reverse direction of corollary [ th - polyker ] does not hold true , unless @xmath36 .
this can be shown by the knapsack problem parameterized by the number of items @xmath6 , @xmath6-kp for short . by theorem [ maintheorem2 ]
there is a polynomial kernel of size @xmath71 for @xmath6-kp but by corollary [ cor - pfpt - p ] there is no polynomial fpt - algorithm for @xmath6-kp .
the used transformation @xmath27 for the proof of corollary [ th - polyker ] runs in time @xmath72 , while a kernelization may have a transformation @xmath27 which runs in polynomial time in @xmath14 _ and _ @xmath15 .
this will be exploited within the following characterization of problems allowing kernels of constant size .
[ th - me ] for every parameterized problem @xmath20 the following properties are equivalent : 1 .
@xmath70 2 .
@xmath16 is decidable and @xmath20 has a kernel of @xmath73 size .
\(1 ) @xmath74 ( 2 ) : let @xmath70 and @xmath5 be a polynomial fpt - algorithm with respect to @xmath21 which runs on input @xmath13 in time @xmath75 for two constants @xmath0 and @xmath35 .
w.l.o.g .
we assume that there is one constant size yes - instance @xmath63 and one constant size no - instance @xmath64 of @xmath16 .
we run @xmath5 on input @xmath13 and instead of deciding we transform the input to the yes- or no - instance of bounded size .
this leads to a kernel of constant size and the algorithm uses polynomial time in @xmath14 and @xmath15 .
\(2 ) @xmath74 ( 1 ) : if @xmath20 has a kernel of size @xmath76 , we can solve the problem on input @xmath13 as follows .
first we compute the kernel @xmath52 in polynomial time w.r.t . @xmath14 and @xmath15
. then we check , whether the instance @xmath53 belongs to the set of yes - instances of size at most @xmath4 , which does not dependent on the input .
the special case that we take the parameter in unary , i.e. @xmath77 , was mentioned in @xcite .
then for some parameterized problem @xmath20 it holds that @xmath16 belongs to @xmath78 if and only if it has a kernel of size @xmath73 . by theorem
[ th - me ] and corollary [ cor - pfpt - p ] we obtain the following result .
[ cor - pfpt - p2 ] let @xmath20 be some parameterized problem and @xmath4 be some constant such that @xmath16 is np - hard and for every instance @xmath13 of @xmath16 it holds @xmath33 .
then there is no kernel of @xmath73 size with respect to @xmath21 for @xmath20 , unless @xmath36 .
we have shown that every @xmath79 has for every instance @xmath13 a kernel of size @xmath80 using a kernelization of running time @xmath81 .
further we have shown that every @xmath79 has for every instance @xmath13 a kernel of size @xmath73 using a kernelization of running time @xmath82 .
we want to have a closer look at the differences between these two types of kernelizations .
[ th - me1 ] for every parameterized problem @xmath20 the following properties are equivalent : 1 .
@xmath61 2 .
@xmath16 is decidable and @xmath20 has a kernelization of running time @xmath81 .
@xmath16 is decidable and @xmath20 has a kernelization of running time @xmath82 .
\(1 ) @xmath83 ( 2 ) : proof of theorem 1.39 in @xcite .
( 1 ) @xmath83 ( 3 ) : proof of theorem 1 in @xcite . that is , the existence of a kernel found by a kernelization of running time which is polynomial in @xmath14 is equivalent to the existence of a kernelization of running time which is polynomial in @xmath14 and @xmath15 .
it remains open whether this is also the case for the existence of polynomial kernels . for the case of constant kernels
the proof of theorem [ th - me ] uses a kernelization of running time which is polynomial in @xmath14 and @xmath15 .
this is really necessary , which can be seen as follows .
the knapsack problem parameterized by the capacity @xmath4 , @xmath4-kp for short , is in @xmath31 by theorem [ maintheorem ] .
but if there would be a kernel for @xmath4-kp of size @xmath73 found by a kernelization of running time @xmath81 then the part ( 2 ) @xmath74 ( 1 ) of the proof of theorem [ th - me ] implies a polynomial algorithm for knapsack . a very useful theorem for finding kernels of knapsack problems with respect to parameter number of items @xmath6 is the following result of frank and tardos on compressing large integer values to smaller ones .
[ th - ft87 ] given a vector @xmath84 and an integer @xmath85 , there exists an algorithm that computes a vector @xmath86 in polynomial time , such that for some number @xmath87 , the notation @xmath88 gives its absolute value . ]
@xmath89 and @xmath90 for all @xmath91 with @xmath92 . by choosing vectors
@xmath93 @xmath94 @xmath95 we immediately see that for each @xmath96 there also is @xmath97 .
this result can be used to equivalently replace equations and inequalities for @xmath98 @xmath99 with @xmath100 by choosing vector @xmath101 and @xmath102 such that @xmath103 .
let @xmath16 be some optimization problem and @xmath13 be some instance of @xmath16 .
by @xmath104 we denote the value of an optimal solution for @xmath16 on input @xmath13 . an _ approximation algorithm _ @xmath5 for @xmath16 is an algorithm which returns a feasible solution for @xmath16 .
the value of the solution of @xmath5 on input @xmath13 is denoted by @xmath105 .
an approximation algorithm @xmath5 has _ relative performance guarantee @xmath102 _ , if @xmath106 holds for every instance @xmath13 of @xmath16 .
a _ polynomial - time approximation scheme ( ptas ) _ for @xmath16 is an algorithm @xmath5 , for which the input consists of an instance of @xmath16 and some @xmath107 , @xmath108 , such that for every fixed @xmath107 algorithm @xmath5 is a polynomial time approximation algorithm with relative performance guarantee @xmath109 .
efficient polynomial - time approximation scheme ( eptas ) _ is a ptas running in time @xmath110 , for some computable function @xmath27 and some constant @xmath23 .
a _ fully polynomial - time approximation scheme ( fptas ) _ is a ptas running in time @xmath111 , for two constants @xmath4 and @xmath112 .
obviously every fptas is an eptas and every eptas is a ptas .
next we recall relations between the existence of approximation schemes for optimization problems and fixed - parameterized algorithms . given some optimization problem @xmath16 the corresponding decision problem of @xmath16 is obtained by adding an integer @xmath12 to the input of @xmath16 and changing the task into the question , whether the size of an optimal solution is at least ( for maximization problems ) or at most ( for minimization problems ) @xmath12 . by choosing the threshold value @xmath12 as a parameter we obtain the so - called _ standard parameterization _
@xmath12-@xmath16 of the so - defined decision problem .
@xmath12-@xmath16 an instance @xmath13 of @xmath16 and an integer @xmath12 .
@xmath12 is there a solution such that @xmath113 ( for a maximization problem @xmath16 ) or @xmath114 ( for a minimization problem @xmath16 ) ? there are two useful connections between the existence of special ptas for some optimization problem @xmath16 and fpt - algorithms for @xmath12-@xmath16 .
[ theorem - fg ] if some optimization problem @xmath16 has an eptas with running time @xmath115 , then there is an fpt - algorithm that solves the standard parameterization @xmath12-@xmath16 of the corresponding decision problem in time @xmath116 .
[ theorem - cc97 ] if some optimization problem @xmath16 has an fptas with running time @xmath117 , then there is a polynomial fpt - algorithm that solves the standard parameterization @xmath12-@xmath16 of the corresponding decision problem in time @xmath118 .
the main idea in the proofs of theorem [ theorem - fg ] and theorem [ theorem - cc97 ] is that the given approximation scheme for @xmath119 for optimization problem @xmath16 leads to an fpt - algorithm that solves the standard parameterization of the corresponding decision problem @xmath12-@xmath16 with the given running time . for so - called scalable optimization problems ( cf .
@xcite ) the reverse direction of theorem [ theorem - cc97 ] also holds true . by the definition
, the existence of an approximation scheme for some optimization problem @xmath16 applies the fundamental parameter @xmath120 measuring the goodness of approximation .
every ptas provides for a fixed error @xmath107 a polynomial time algorithm .
since these algorithms are not very practical , the question arises whether @xmath121 can be taken out of the exponent of the input size .
this is the case if the ptas is even an eptas .
a formal method to combine the error bound @xmath107 and decision problems is the so - called gap version of an optimization problem , which was introduced by marx in @xcite .
let @xmath16 be some optimization or decision problem and @xmath17 the set of all instances of @xmath16 .
for some @xmath24 we denote by @xmath122 the value of the largest number occurring in @xmath13 .
an algorithm @xmath5 is _ pseudo - polynomial _
, if there is a polynomial @xmath123 such that for every instance @xmath13 the running time of @xmath5 on @xmath13 is at most @xmath124 , see @xcite .
a problem @xmath16 is _ pseudo - polynomial _ if it can be solved by a pseudo - polynomial algorithm .
[ def - str - np - h ] a problem is _ strongly np - hard _
, if it remains np - hard , if all of its numbers are bounded by a polynomial in the length of the input . [ th - strongnp - pseudo ] if some problem @xmath16 is strongly np - hard , then @xmath16 is not pseudo - polynomial .
the notation of pseudo - polynomial algorithms can be carried over to parameterized algorithms @xcite .
for an instance @xmath13 of some problem @xmath16 the function @xmath125 is defined by the maximum length of the binary encoding of all numbers in @xmath13 . since
the binary coding of an integer @xmath126 has length @xmath127 , the relation @xmath128 holds and function @xmath129 is a parameter for @xmath16 .
further , if there is a pseudo - polynomial algorithm @xmath5 for @xmath16 then there is a polynomial @xmath130 , such that for every instance @xmath13 of @xmath16 the running time of @xmath5 can be bounded by @xmath131 this implies that there are constants @xmath132 and @xmath133 such that the running time of @xmath5 can be bounded by @xmath134 and thus @xmath5 is an fpt - algorithm with respect to parameter @xmath125 .
[ th - val - para ] for every pseudo - polynomial problem @xmath16 there is an fpt - algorithm that solves @xmath129-@xmath16 in time @xmath135 .
the simplest of all knapsack problems is defined as follows .
max knapsack ( max kp ) a set @xmath136 of @xmath6 items , for every item @xmath137 , there is a size of @xmath3 and a profit of @xmath2 .
further there is a capacity @xmath4 for the knapsack .
find a subset @xmath138 such that the total profit of @xmath8 is maximized and the total size of @xmath8 is at most @xmath4 .
the max knapsack problem can be approximated very good , since it allows an fptas @xcite and thus can be regarded as one of the easiest hard problems . in this paper ,
the parameters @xmath6 , @xmath2 , @xmath3 , and @xmath4 are assumed to be positive integers , i.e. they belong to the set @xmath139 .
let @xmath140 and @xmath141 .
the same notations are also used for @xmath142 instead of @xmath143 . in order to avoid trivial solutions
we assume that @xmath144 and that @xmath145 . for some instance @xmath13 its size @xmath14
can be bounded by the number of items and the binary encoding of all numbers in @xmath13 ( cf .
@xcite ) . @xmath146
the size of the input is important for the analysis of running times . in order to show fpt - algorithms and kernelizations we frequently will use bounds on the size of the input and on the size of a solution of our problems .
let @xmath64 be a knapsack instance on item set @xmath147 .
instance @xmath148 on item set @xmath149 is a _
reduced instance _ for @xmath64 if @xmath150 and @xmath151 . for max
kp we can bound the number and sizes of the items of an instance as follows .
[ le - blb ] every instance of max kp can be transformed into a reduced instance , such that @xmath152 . in order to avoid trivial solutions we assume ,
that there is no item in @xmath5 , whose size is larger than the capacity @xmath4 , i.e. @xmath153 for every @xmath154 .
further for @xmath155 we can assume that there are at most @xmath156 items of size @xmath157 in @xmath5 . by the harmonic series we always can bound the number @xmath6 of items in @xmath5 by @xmath158 if we have given an instance @xmath64 for max kp with more than the mentioned number @xmath159 of items of size @xmath157 for some @xmath155
, we remove all of them except the @xmath159 items of the highest profit .
the new instance @xmath148 satisfies @xmath152 and is a reduced instance of @xmath64 .
the latter result is useful in remark [ rem - c - kern ] .
we have shown an alternative proof for lemma [ le - blb ] in @xcite . since the capacity @xmath4 and
the sizes of our items are positive integers , every solution @xmath8 of some instance of max kp even contains at most @xmath4 items . but this observation does not lead to a reduced instance . in
order so solve the problem using this bound one has to consider @xmath160 many possible subsets of @xmath5 which is much more inefficient than the dynamic programming approach mentioned in theorem [ t - dy2 ] .
integer programming is a powerful tool , studied for over 50 years , that can be used to define a lot of very important optimization problems @xcite .
max kp can be formulated using a boolean variable @xmath161 for every item @xmath162 , indicating whether or not @xmath137 is chosen into the solution @xmath8 , by a so - called binary integer program ( bip ) .
@xmath163\label{lp - kp3 } \end{aligned}\ ] ] for some positive integer @xmath6 , let @xmath164=\{1,\ldots , n\}$ ] be the set of all positive integers between @xmath165 and @xmath6 .
we apply bip versions for our knapsack problems to obtain parameterized algorithms ( theorem [ maintheorem ] ) .
dynamic programming solutions for max kp are well known .
the following two results can be found in the textbook @xcite .
[ t - dy2 ] max kp can be solved in time @xmath166 .
[ t - dy1 ] max kp can be solved in time @xmath167 , where @xmath168 is an upper bound on the value of an optimal solution . since for unary numbers the value of
the number is equal to the length of the number the running times of the two cited dynamic programming solutions is even polynomial .
thus max kp can be solved in polynomial time if all numbers are given in unary . in this paper
we assume that all numbers are encoded in binary .
although max kp is a well known example for a pseudo - polynomial problem we want to give this result for the sake of completeness .
[ th - si - bug - pseudo ] max kp is pseudo - polynomial .
we consider the running time of the algorithm which proves theorem [ t - dy1 ] . in the running time @xmath169
part @xmath170 is polynomial in the input size and part @xmath171 is polynomial in the value of the largest occurring number in every input .
( alternatively we can apply the running time of the algorithm cited in theorem [ t - dy2 ] . )
since max kp is an integer - valued problem defined on inputs of various informations , it makes sense to consider parameterized versions of the problem . by adding a threshold value @xmath12 for the profit to the instance and choosing a parameter @xmath15 from this instance @xmath13 , we define the following parameterized problem .
@xmath21-knapsack ( @xmath21-kp ) a set @xmath136 of @xmath6 items , for every item @xmath137 , there is a size of @xmath3 and a profit of @xmath2 .
further there is a capacity @xmath4 for the knapsack and a positive integer @xmath12 .
@xmath15 is there a subset @xmath138 such that the total profit of @xmath8 is at least @xmath12 and the total size of @xmath8 is at most @xmath4 . for some instance @xmath13 of @xmath21-kp its size @xmath14
can be bounded by @xmath172 next we give parameterized algorithms for the knapsack problem .
the parameter @xmath173 counts the number of distinct item sizes within knapsack instance @xmath13 .
.overview of parameterized algorithms for kp [ cols="<,<,<,^,<,<",options="header " , ] [ maintheorem2-mkp ] there exist kernelizations for the parameterized multiple knapsack problem such that the upper bounds for the sizes of a possible kernel in table [ tab - mkp - ker - survey ] hold true . for parameter @xmath174
we proceed as in the proof of theorem [ maintheorem2 ] which shows a kernel of size @xmath71 for @xmath6-kp . in the case of
@xmath6-mkp we have to scale @xmath1 inequalities of type ( [ eq - ker - n-1 ] ) on @xmath6 variables and one inequality of type ( [ eq - ker - n-2 ] ) on @xmath175 variables by theorem [ th - ft87 ] .
for the obtained instance @xmath53 we can bound @xmath176 by the number of items @xmath6 and @xmath177 numbers of value at most @xmath178 and @xmath179 numbers of value at most @xmath180 . thus @xmath6-mkp has a kernel of size @xmath181 we can assume @xmath182 ( cf .
beginning of section [ def - sec - mkp ] ) and since every item is assigned to at most one knapsack ( cf .
( [ lp - mkp3 ] ) ) we know @xmath183 .
thus we obtain a kernel of size @xmath184 for @xmath6-mkp . for parameter @xmath185
we obtain by theorem [ maintheorem - mkp ] and theorem [ th - me ] a kernel of constant size . for the remaining six parameters of table [ tab - mkp - ker - survey ] the upper bounds follow by theorem [ maintheorem - mkp ] and theorem [ th - ker ] .
we have considered the max knapsack problem and its two generalizations max multidimensional knapsack and max multiple knapsack .
the parameterized decision versions of all three problems allow several parameterized algorithms . from a practical point of view choosing
the standard parameterization @xmath12 is not very useful , since a large profit of the subset @xmath8 violates the aim that a good parameterization is small for every input .
so for kp we suggest it is better to choose the capacity as a parameter , i.e. @xmath32 , since common values of @xmath4 are low enough such that the polynomial fpt - algorithm is practical .
the same holds for d - kp and mkp .
further one has a good parameter , if it is smaller than the input size @xmath14 but measures the structure of the instance .
this is the case for the parameter number of items @xmath6 within all three considered knapsack problems . the special case of the max knapsack problem , where @xmath186 for all items @xmath154 is known as the subset sum problem .
for this case we know that @xmath187 and we conclude the existence of fpt - algorithms with respect to parameter @xmath6 , @xmath4 , and @xmath12 .
kernels for the subset sum problem w.r.t .
@xmath6 and the number of different sizes @xmath188 are examined in @xcite . the closely related minimization problem @xmath189 \label{lp}\end{aligned}\ ] ] is known as the change making problem , whose parameterized complexity is discussed in @xcite . in our future work , we want to find better fpt - algorithms , especially for d - kp and mkp .
we also want to consider the following additional parameters .
* @xmath190 for kp * @xmath190 , @xmath191 , @xmath192 , @xmath171 , @xmath193 , and @xmath129 for d - kp , and * @xmath173 , @xmath190 , @xmath129 , @xmath192 , @xmath171 , and @xmath193 for mkp . also from a theoretical point of view
it is interesting to increase the number of parameters for which the parameterized complexity of the considered problems is known .
for example if our problem is @xmath194$]-hard with respect to some parameter @xmath21 , then a natural question is to ask , whether it remains hard for the _ dual parameter _
that is , if @xmath21 measures the costs of a solution , then for some optimization problem the dual parameter @xmath196 measures the costs of the elements that are not in the solution @xcite . since @xmath12-d - kp is @xmath194$]-hard the question arises , whether d - kp becomes tractable w.r.t .
parameter @xmath197 .
more general , one also might consider more combined parameters , i.e.parameters that consists of two or more parts of the input .
for d - kp combined parameters including @xmath12 are of our interest .
the existence of polynomial kernels for knapsack problems seems to be nearly uninvestigated .
recently a polynomial kernel for kp using rational sizes and profits is constructed in @xcite by theorem [ th - ft87 ] .
this result also holds for integer sizes and profits ( cf .
theorem [ maintheorem2 ] ) . by considering polynomial fpt - algorithms we could show some lower bounds for kernels for kp ( cf .
table [ tab - kp - ker - survey ] ) .
we want to consider further kernels for d - kp and mkp , try to improve the sizes of known kernels , and give lower bounds for the sizes of kernels .
a further task is to extend the results to more knapsack problems , e.g. max - min knapsack problem and restricted versions of the presented problems , e.g. multiple knapsack with identical capacities ( mkp - i ) , see @xcite .
we also want to consider the existence of parameterized approximation algorithms for knapsack problems , see @xcite for a survey .
we would like to thank klaus jansen and steffen goebbels for useful discussions .
m. etscheid , s. kratsch , m. mnich , and rglin .
polynomial kernels for weighted problems . in _ proceedings of mathematical foundations of computer science _ ,
volume 9235 of _ lncs _ , pages 287298 .
springer - verlag , 2015 .
fellows , s. gaspers , and f.a .
parameterizing by the number of numbers . in _ proceedings of the symposium on parameterized and exact computation _ ,
volume 6478 of _ lecture notes in computer science _ , pages 123134 .
springer - verlag , 2010 .
f. gurski , j. rethmann , and e. yilmaz . computing partitions with applications to capital budgeting problems . in _
operations research proceedings ( or 2015 ) , selected papers_. springer - verlag , 2016 . to appear .
k. jansen . a fast approximation scheme for the multiple knapsack problem . in _ proceedings of the conference on current trends in theory and practice of computer science _ ,
volume 7147 of _ lncs _ , pages 313324 .
springer - verlag , 2012 .
j. nederlof , e. j. van leeuwen , and r. van der zwaan .
reducing a target interval to a few exact queries . in _ proceedings of mathematical foundations of computer science _ , volume 7464 of _ lncs _ , pages 718727 .
springer - verlag , 2012 .
r. niedermeier .
reflections on multivariate algorithmics and problem parameterization . in _ proceedings of the annual symposium of theoretical aspects of computer science _ ,
volume 5 of _ lipics _ , pages 1732 .
schloss dagstuhl - leibniz - zentrum fuer informatik , 2010 . | the knapsack problem ( kp ) is a very famous np - hard problem in combinatorial optimization .
also its generalization to multiple dimensions named d - dimensional knapsack problem ( d - kp ) and to multiple knapsacks named multiple knapsack problem ( mkp ) are well known problems .
since kp , d - kp , and mkp are integer - valued problems defined on inputs of various informations , we study the fixed - parameter tractability of these problems . the idea behind fixed - parameter tractability is to split the complexity into two parts - one part that depends purely on the size of the input , and one part that depends on some parameter of the problem that tends to be small in practice . further we consider the closely related question , whether the sizes and the values can be reduced , such that their bit - length is bounded polynomially or even constantly in a given parameter , i.e. the existence of kernelizations is studied .
we discuss the following parameters : the number of items , the threshold value for the profit , the sizes , the profits , the number @xmath0 of dimensions , and the number @xmath1 of knapsacks .
we also consider the connection of parameterized knapsack problems to linear programming , approximation , and pseudo - polynomial algorithms .
* keywords : * knapsack problem ; d - dimensional knapsack problem ; multiple knapsack problem ; parameterized complexity ; kernelization |
in recent years there have been a slew of textbooks @xcite , review articles @xcite , and lecture notes @xcite designed either as an introduction to numerical relativity , or as a convenient place to understand the state of the art of the field , not to mention classic texts on time evolution problems @xcite .
these resources already serve their purpose beautifully .
so , happy as i was to be asked to teach introductory material and provide written lecture notes , the obvious question is ; does the world need another set of introductory notes to hyperbolic systems ?
i ve tried to come to a solution that presents the heart - and - soul of the topic as concisely as possible . in this class
i will review concepts in the analysis of time - evolution partial differential equations , pdes , proving results only sparsely .
the main aim is to collect together , in the form of a tool - box , the necessary weapons for treating a given system of pdes .
i hope that where i have shamelessly copied , the authors of existing texts will accept my imitation as flattery . in the second lecture
i treat electromagnetism as a model for general relativity , and apply each of the tools to demonstrate how they are used in practice .
i expect that this application will be enlightening .
i highlight the effect of gauge freedom on the pdes analysis , for a large family of gauges . to my knowledge
such a treatment has not appeared elsewhere , although free - evolution formulations of electromagnetism have of course been studied in the literature @xcite . in physics and
applied mathematics we are frequently presented with systems of pdes .
well - posedness is a fundamental property of a pde problem .
it is the requirement that there be a unique solution that depends continuously , in some norm , on given data for the problem . without it ,
one has simply not built a reasonable mathematical abstraction of the physical problem at hand .
the model is without predictive power , because small changes in the given data might result in either arbitrarily large changes in the outcome or that there is no solution at all .
if we are given a complicated system , like the field equations of general relativity , we will probably have to find solutions numerically .
but if the formulation as a pde problem is ill - posed , _ no _ numerical approach can be successful !
afterall , how can one construct an approximation scheme that converges to the continuum solution if the solution does nt exist ?
therefore one might find it surprising that research in numerical relativity has been performed with problems that are ill - posed . spontaneously on hearing
that such versions of general relativity exist , you might think that this sounds a bit like a way of saying that the theory is broken , or somehow deficient .
that impression is wrong .
the answer is that for theories with gauge freedom , the precise formulation of the field equations as a system of pdes affects well - posedness . and it took time for this fact to be recognized in the context of numerical relativity . in the second lecture we will carefully investigate this for electromagnetism . the most crude way of classifying a pde is into one of the three classes , elliptic , parabolic or hyperbolic , names originally inspired by the conic - sections .
the class of a pde determines what type of data has any chance of producing a well - posed problem . from the intuitive point of view of the physicist
, one might summarize their character as follows : _ elliptic pdes _ have no intrinsic notion of time , and often arise as the steady - state , or end - state solutions of dynamical evolution , for example in electrostatics .
the solutions to well - posed elliptic problems are typically `` as smooth as the coefficients allow '' .
the prototype of a well - posed elliptic problem is the boundary value problem for the laplace equation .
_ parabolic pdes _ describe diffusive processes .
they have an intrinsic notion of time , but signals travel at infinite speed . even non - smooth initial data immediately become smooth as they evolve .
the prototype of a well - posed parabolic problem is the initial value problem for the heat equation .
_ hyperbolic pdes _ are the best .
they describe processes which are in some sense causal ; there is an intrinsic notion of time , and crucially signals travel with _
finite _ speed .
discontinuities in initial data for a hyperbolic pde will often be propagated , or may even form from smooth initial data .
the prototype of a well - posed hyperbolic problem is the initial value problem for the wave equation .
notice that all of the prototype well - posed problems specify both a simple pde , the type of data , and the domain that is appropriate .
one can also concoct pdes of mixed character , so this classification is certainly not complete .
numerical relativists have to face all three , and occasionally mixed classes .
but in this lecture we will focus exclusively on hyperbolic problems and well - posedness of the initial , and initial boundary value problems .
_ the initial value problem : _ consider a system of pdes , which can be written , @xmath0 with state vector @xmath1 . here
i employ the summation convention , denote @xmath2 , and assume that @xmath3 .
the highest derivative terms are called the principal part .
i will sometimes refer to @xmath4 as the principal matrix , although it is really a shorthand for three matrices , since i m assuming that we have three spatial dimensions .
the remaining terms on the right - hand - side of are called non - principal . in this lecture
we will assume that the matrices @xmath4 and @xmath5 are constant in both time and space .
we therefore call it a _ linear , constant coefficient system_. the initial value , or cauchy problem , is the following : specify data @xmath6 at time @xmath7 , with spatial coordinates @xmath8 . what is the solution @xmath9 at later times ?
in other words data is specified _ everywhere _ in space ; the domain of the solution is in this sense unbounded .
naturally many pdes of interest are not linear or do not have constant coefficients .
that said , local properties of more complicated systems are determined by the behavior of the system in linear approximation , which justifies the restriction .
_ well - posedness : _ if there exist constants @xmath10 and @xmath11 , such that for all initial data we have the estimate , @xmath12 norm , @xmath13 then the initial value problem for is called well - posed .
notice that we restrict to initial data that are bounded in @xmath14 .
_ strong hyperbolicity : _ given an arbitrary unit spatial vector @xmath15 , we say that the matrix @xmath16 is the principal symbol of the system .
the system is called weakly hyperbolic if for every unit spatial vector @xmath15 , the principal symbol has real eigenvalues .
if furthermore for every unit spatial vector @xmath15 , the principal symbol has a complete set of eigenvectors and there exists a constant @xmath10 , independent of @xmath15 , such that @xmath17 is formed with the eigenvectors of @xmath18 as columns , and we have the usual definition of the matrix norm @xmath19 , the system is called strongly hyperbolic . if a system is strongly hyperbolic and the multiplicity of the eigenvalues does not depend on @xmath15 we say that it is strongly hyperbolic of constant multiplicity .
notice that if the eigenvectors of the principal symbol depend continuously on @xmath15 then the second condition is automatically satisfied , because @xmath15 varies over a compact set . in most applications we will have continuous dependence , and
so checking for strong hyperbolicity amounts to doing a little linear algebra . _
characteristic variables : _ the components of the vector @xmath20 are called the characteristic variables in the @xmath15 direction . up to non - principal terms and derivatives
transverse to the @xmath15 direction the characteristic variables satisfy advection equations with speeds equal to the eigenvalues of the principal symbol . for this reason we will sometimes call the eigenvalues the speeds of the system . to see this
consider , @xmath21 here @xmath22 has the eigenvalues of the principal symbol on the diagonal , we denote longitudinal derivatives @xmath23 , and transverse derivatives @xmath24 by an upper case latin index @xmath25 .
_ well - posed @xmath26 strongly hyperbolic : _ the main result for the initial value problem for is that it is well - posed if and only if the system is strongly hyperbolic .
we will need to fourier transform in space , and use the convention , @xmath27 the time derivative of the variables after fourier transform is , @xmath28 where we write @xmath29 .
so in fourier space the general solution is @xmath30 assume that the system is strongly hyperbolic .
the key to the proof of well - posedness is the use of a symmetrizer .
we define the hermitian , positive definite symmetrizer @xmath31 by , @xmath32 which satisfies the crucial property , @xmath33 note that our choice for the definition of @xmath31 is not the most general that yields ; instead we could have chosen @xmath34 , with @xmath35 hermitian , positive definite , and commuting with @xmath22 .
we do not require the most general @xmath31 here , and so make do with this restriction .
define the norm @xmath36 by , @xmath37 computing the time derivative of the norm @xmath36 , a couple of lines gives the inequality , @xmath38 and integrating we have the well - posedness estimate , in the new norm @xmath36 , latexmath:[\ ] ] i leave it as an exercise for you to convince yourself that this is really a boundary condition on some combination of the electric and magnetic fields , actually @xmath262 in the terminology of teukolsky @xcite .
laplace - fourier transforming the boundary conditions and substituting the general solution into them , we can solve for the constants @xmath255 and obtain the solution @xmath263 at the boundary @xmath62 , where now i ve abandoned most of the shorthands so that you can see how it really looks .
all that remains is to show that each of the variables is bounded in @xmath264 and @xmath47 , which i leave as an exercise , but point you towards @xcite which shows the necessary estimates on all of the terms present here .
the essential point is that we do nt have to worry about the numerators in any of the fractions , because they are obviously bounded , and since the real part of @xmath132 is positive , terms like @xmath265 appearing in the denominators are bounded away from zero . recalling that by construction @xmath257 corresponds directly to constraint violation , and noting that , @xmath266 it is clear that the constraint preserving boundary conditions work when @xmath260 is chosen .
these calculations are largely performed in the mathematica notebook maxwell_lf.nb which accompanies the lecture . with the lorenz gauge , @xmath204 , well - posedness
can be shown for constraint preserving boundary conditions using the energy method @xcite with a special choice of symmetrizer , or alternatively with the kreiss - winicour cascade approach @xcite . to my knowledge
this is the first time that boundary stability has been demonstrated for arbitrary hyperbolic gauge conditions inside the family . in this lecture
we looked at different formulations of electromagnetism suitable for free - evolution .
we saw that for every strongly hyperbolic pure gauge , we could build a formulation which was itself strongly hyperbolic .
we saw furthermore that system is system is symmetric hyperbolic for all of these gauge choices .
working then in the high - frequency frozen coefficient approximation , we used the laplace - fourier method to investigate boundary stability with constraint preserving boundary conditions .
some parts of the calculations were not very explicitly presented . to understand the ins - and - outs i recommend that you study the mathematica notebooks in tandem with the lecture notes .
they are available at the school s website @xcite .
some of the calculations presented in this lecture used the mathematica package _ xtensor _
@xcite for abstract tensor calculations .
i would like to thank the organizers of the nrhep2 school for giving me the rewarding opportunity to present these lectures .
i have benefited greatly from discussions with bernd brgmann , ronny richter , milton ruiz and andreas weyhausen .
i am especially grateful to olivier sarbach for carefully reading the lecture notes and offering very helpful criticism .
i am supported by the dfg grant sfb / transregio @xmath267 .
frans pretorius .
binary black hole coalescence . in w.
b. burton , editor , _ physics of relativistic objects in compact binaries : from birth to coalescence _ ,
volume 359 , pages 305369 .
springer , netherlands , 2009 .
lee lindblom , mark a. scheel , lawrence e. kidder , harald p. pfeiffer , deirdre shoemaker , and saul a. teukolsky .
controlling the growth of constraints in hyperbolic evolution systems .
, 69:124025 , 2004 . | these lecture notes accompany two classes given at the nrhep2 school . in the first lecture
i introduce the basic concepts used for analyzing well - posedness , that is the existence of a unique solution depending continuously on given data , of evolution partial differential equations .
i show how strong hyperbolicity guarantees well - posedness of the initial value problem .
symmetric hyperbolic systems are shown to render the initial boundary value problem well - posed with maximally dissipative boundary conditions .
i discuss the laplace - fourier method for analyzing the initial boundary value problem .
finally i state how these notions extend to systems that are first order in time and second order in space . in the second lecture
i discuss the effect that the gauge freedom of electromagnetism has on the pde status of the initial value problem .
i focus on gauge choices , strong - hyperbolicity and the construction of constraint preserving boundary conditions .
i show that strongly hyperbolic pure gauges can be used to build strongly hyperbolic formulations .
i examine which of these formulations is additionally symmetric hyperbolic and finally demonstrate that the system can be made boundary stable . |
for several experimental setups , a description that completely specifies the nature of each device is unsatisfactory .
for example , in a realistic scenario the assumption that the provider of the devices is fully reliable is often overoptimistic : imperfections unavoidably affect the implementation , thus turning it away from its ideal description .
a device independent description of an experimental setup does not make any assumption on the involved devices , which are regarded as `` black boxes '' , while only the knowledge of the correlations between preparations , measurements and outcomes is considered . in this scenario ,
a natural question is whether it is possible to derive some properties of the non - characterized devices instead of assuming them , building only upon the knowledge of these correlations . in general
one could be interested in bounding the dimension of the systems prepared by a non - characterized device ; one could also ask whether a source is intrinsically quantum or can be described classically .
the framework of device independent dimension witnesses ( didws ) provides an effective answer to these questions , suitable for experimental implementation and for application in different contexts , such as quantum key distribution ( qkd ) or quantum random access codes ( qracs ) .
didws were first introduced in @xcite in the context of non - local correlations for multi - partite systems .
subsequently , the problem of didws was related to that of qracs in @xcite , and in @xcite it was reformulated from a dynamical viewpoint allowing one to obtain lower bounds on the dimensionality of the system from the evolution of expectation values . a general formalism for tackling the problem of didws in a prepare and measure scenario
was recently developed in @xcite .
the derived formalism allows one to establish lower bounds on the classical and quantum dimension necessary to reproduce the observed correlations .
shortly after , the photon experimental implementations followed , making use of polarization and orbital angular momentum degrees of freedom @xcite or polarization and spatial modes @xcite to generate ensembles of classical and quantum states , and certifying their dimensionality as well as their quantum nature .
didws also allow reformulating several applications in a device - independent framework .
for example , dimension witnesses can be used to share a secret key between two honest parties . in @xcite ,
the authors present a qkd protocol whose security against individual attacks in a semi - device independent scenario is based on didws .
the scenario is called semi - device independent because no assumption is made on the devices used by the honest parties , except that they prepare and measure systems of a given dimension .
another application is given by qracs , that make it possible to encode a sequence of qubits in a shorter one in such a way that the receiver of the message can guess any of the original qubits with maximum probability of success .
in @xcite qracs were considered in the semi - device independent scenario , with a view to their application in randomness expansion protocols . clearly
any experimental implementation of didws is unavoidably affected by losses - that can be modeled as a constraint on the measurements - and can reduce the value of the dimension witness , thus making it impossible to witness the dimension of a system .
based on these considerations , it is relevant to understand whether it is possible to perform reliable dimension witnessing in realistic scenarios and , in particular , with non - optimal detection efficiency .
we refer to this problem as the _ robustness of device independent dimension witnesses_. despite its relevance for experimental implementations and practical applications , this problem has not been addressed in previous literature .
the aim of this work is to fill this gap .
we consider the case where shared randomness between preparations and measurements is allowed .
our main result is to provide the threshold in the detection efficiency that can be tolerated in dimension witnessing , in the case where one is interested in the dimension of the system as well as in the case where one s concern is to discriminate between its quantum or classical nature .
the paper is structured as follows . in section [
sect : didw ] we introduce the sets of quantum and classical correlations and the concept of dimension witness as a tool to discriminate whether a given correlation matrix belongs to these sets .
section [ sect : sets ] discusses some properties of the sets of classical and quantum correlations . in section [
sect : robustness ] we provide a threshold in the detection efficiency that is allowed in witnessing the dimensionality of a system or in discriminating between its classical or quantum nature , as a function of the dimension of the system .
we summarize our results and discuss some further developments - such as dimension witnessing in the absence of correlations between preparations and measurements or entangled assisted dimension witnessing - in section [ sect : conclusion ] .
let us first fix the notation @xcite . given a hilbert space @xmath0 ,
we denote with @xmath1 the space of linear operators @xmath2 . a quantum state in @xmath0
is represented by a density matrix , namely a positive semi - definite matrix @xmath3 such that @xmath4 = 1 $ ] . given a pure state @xmath5 , we denote with @xmath6 the corresponding projector .
a set @xmath7 of states is said to be classical when the states commute pairwise , namely @xmath8 = 0 $ ] for any @xmath9 . here , the notion of classicality has to be understood in an operational sense : in our scenario , the observed correlations can be reproduced by a classical variable taking @xmath10 possible values if , and only if , they can be reproduced by measurements on pairwise commuting states acting on a hilbert space of dimension @xmath10 ( this will become clearer after lemma [ thm : classicalpovms ] below ) .
a general quantum measurement is represented by a positive - operator valued measure ( povm ) , namely a set of positive semi - definite hermitian matrices @xmath11 such that @xmath12 .
a povm @xmath13 is said to be classical when @xmath14 = 0 $ ] for any @xmath15 .
the joint probability of outcome @xmath16 given input state @xmath17 is given by the born rule , namely @xmath18 $ ] .
the general setup introduced in ref .
@xcite for performing device independent dimension witnessing is given by a preparing device ( let us say on alice s side ) and a measuring device ( on bob s side ) as in fig .
[ fig : setup ] . .
alice ( on the left hand side ) owns a preparing device which sends the state @xmath19 to bob whenever alice presses button @xmath20 $ ] .
bob owns a measuring device that performs measurement @xmath21 on the received state whenever bob presses button @xmath22 $ ] , giving the outcome @xmath23 $ ] . ] in the most general scenario , the devices may share a priori correlated information , classical and quantum .
however , in many realistic situations , one can assume that the preparing and measuring devices are uncorrelated and that all the correlations observed between the preparation and the measurement are due to the mediating particle connecting the two devices .
an intermediate and also valid possibility is to assume that the devices only share classical correlations . in this case , the value of a random variable @xmath24 distributed according to @xmath25 is accessible to preparing and measuring devices . in this work we focus on this last possibility .
alice chooses the value of index @xmath20 $ ] and sends a fixed state @xmath26 to bob .
bob chooses the value of index @xmath22 $ ] and performs a fixed povm @xmath21 on the received state , obtaining outcome @xmath27 $ ] . after repeating the experiment
several times ( we consider here the asymptotic case ) , they collect the statistics about indexes @xmath28 obtaining the conditional probabilities @xmath29 .
note that we also implicitly assume that we are dealing with independent and identically distributed events .
we now introduce the set @xmath30 ( the set @xmath31 ) of correlations achievable with quantum ( classical ) preparations . for any @xmath32
we define the _ set of quantum correlations _ @xmath33 as the set of correlations @xmath29 with @xmath20 $ ] , @xmath22 $ ] and @xmath34 $ ] such that there exist a hilbert space @xmath35 with @xmath36 , a quantum set @xmath37 of states and a set @xmath38 of povms @xmath39 for which @xmath40 $ ] , namely @xmath41\}. \end{aligned}\ ] ] for any @xmath32 we define the _ set of classical correlations _ @xmath42 as the set of correlations @xmath29 with @xmath20 $ ] , @xmath22 $ ] and @xmath34 $ ] such that there exist a hilbert space @xmath35 with @xmath36 , a classical set @xmath37 of states and a set @xmath38 of povms @xmath39 for which @xmath40 $ ] , namely @xmath43\}. \end{aligned}\ ] ] we write @xmath30 and @xmath31 omitting the parameters @xmath44 whenever they are clear from the context .
[ rmk : sr ] we notice that , when shared randomness is allowed between quantum ( classical ) preparations and measurements , the set of achievable correlations is given by @xmath45 ( @xmath46 ) , where for any set @xmath47 we denote with @xmath48 the convex hull of @xmath47 .
the following lemma shows that it is not restrictive to consider only classical povms , that is , measurements consisting of commuting operators , in the definitions of classical correlations .
[ thm : classicalpovms ] for any correlation @xmath49 there exist a classical set @xmath50 of states and a set @xmath51 of classical povms @xmath52 such that @xmath53 $ ] . by hypothesis there
exist a classical set @xmath50 of states and a set @xmath54 of povms @xmath55 such that @xmath56 $ ] for any @xmath28
. take @xmath57 where @xmath58 is an orthonormal basis with respect to which the @xmath17 s are diagonal ( it is straightforward to verify that @xmath59 for any @xmath60 and @xmath61 for any @xmath62 ) .
we have @xmath63 $ ] for any @xmath28 , which proves the statement .
lemma [ thm : classicalpovms ] thus proves that every set of probabilities obtained with commuting states can be performed with classical states and classical povms .
this clearly implies that commuting states may be equally regarded as classical variables , and commuting - element measurements as read - out of classical variables .
we can now introduce didws .
building only on the knowledge of @xmath29 , our task is to provide a lower bound on the dimension @xmath10 of @xmath0 for any set of correlations @xmath47 between @xmath64 preparations and @xmath65 measurements with @xmath66 outcomes , a _ device independent dimension witness _
@xmath67 is a function of the conditional probability distribution @xmath68 with @xmath69 $ ] , @xmath22 $ ] , and @xmath34 $ ] such that @xmath70 for some @xmath71 which depends on @xmath72 .
interestingly , in many situations the value of the bound @xmath71 in the definition of a dimension witness varies depending on whether one is interested in classical or quantum ensembles of states .
this gives a second application for dimension witnesses , namely quantum certification : if the system dimension is assumed , dimension witnesses allow certifying its quantum nature .
it is precisely this quantum certification what makes dimension witnesses useful for quantum information protocols @xcite
. motivated by remark [ rmk : sr ] , for any @xmath73 when @xmath74 [ when @xmath75 we say that @xmath67 is a classical ( quantum ) dimension witness for dimension @xmath10 in the presence of shared randomness . given a set @xmath76 of states and a set @xmath77 of povms @xmath78 , we define @xmath79 with @xmath68 and @xmath80 $ ] , and analogously for @xmath81 . in this work
we will consider only linear didws , namely inequalities of the form of eq .
such that @xmath82 where @xmath83 is a constant vector . notice that for any function @xmath84 and constant @xmath71 , the witness @xmath85 is only a representative of a class of equivalent witnesses such that if @xmath86 is a member of the class , then @xmath85 if and only if @xmath86 for any conditional distribution @xmath87 .
the following lemma provides a transformation that preserves this equivalence .
[ thm : norm ] given a function @xmath88 and a constant @xmath71 , take @xmath89 with @xmath90 and @xmath91 for any @xmath92 that does not depend on outcome @xmath16 .
then one has @xmath85 if and only if @xmath86 for any @xmath87 .
it follows immediately by direct computation . in the following our task
will be to find a set @xmath93 of quantum states and a set @xmath94 of povms such that a linear witness @xmath95 maximally violates inequality .
the following lemma allows us to simplify the optimization problem .
[ thm : pure ] the maximum of any linear dimension witness @xmath95 is achieved by an ensemble @xmath93 of pure states and without shared randomness .
the thesis follows immediately from linearity .
due to lemma [ thm : pure ] the maximization of eq .
is equivalent to the maximization of @xmath96 over the sets @xmath97 of pure states and the sets @xmath98 of povms @xmath55 .
before moving to the main results in this article , we discuss in this section several properties of the sets of classical and quantum correlations . in particular , we study whether the sets are convex and prove some inclusions among them .
these results allow gaining a better understanding of the geometry of these sets of correlations .
since classical correlations can always be reproduced by quantum ones , we immediately have @xmath99 and @xmath100 .
moreover , by definition we have @xmath101 and @xmath102 . here
we show an example where @xmath31 is non - convex ( namely @xmath103 ) and @xmath104 .
take @xmath105 , @xmath106 , @xmath107 , and @xmath108 in the setup of figure [ fig : setup ] .
consider the following conditional probability distribution @xmath29 of obtaining outcome @xmath16 on bob s device given input @xmath109 on alice s and @xmath62 on bob s @xmath110 where rows and columns are labeled by @xmath109 and @xmath16 , respectively .
first we show that @xmath111 .
indeed @xmath87 can be obtained when alice and bob share classical correlations represented by a uniformly distributed random variable @xmath24 taking values @xmath112 making use of the classical set @xmath113 of states and of the set @xmath114 of classical povms @xmath115 , with @xmath116 and @xmath117 which proves that @xmath118\ } \in { \operatorname{conv}}{\mathcal{c}}$ ] .
now we show that @xmath119 .
indeed @xmath87 can be obtained by alice and bob making use of the quantum set @xmath120 of states and of the set @xmath54 of quantum povms @xmath55 , with @xmath121 and @xmath122 which proves that @xmath119 .
finally , we verify that if alice and bob make use of classical sets of states and povms and do not have access to shared randomness there is no way to achieve the probability distribution @xmath87 given by eq . . indeed , to have perfect discrimination between @xmath123 and @xmath124 with povm @xmath125
] , one must take @xmath123 and @xmath124 orthogonal - let us say without loss of generality @xmath126 and @xmath127 , and @xmath128 and @xmath129 .
due to the hypothesis of classicality of the sets of states , @xmath130 must be a convex combination of @xmath123 and @xmath124 .
then , in order to have @xmath131 as in eq .
, one has to choose @xmath132 . finally , the only possible choice for @xmath133 is @xmath134 and @xmath135 , which is incompatible with the remaining entries of @xmath136 in eq .
this proves that @xmath137 . the relations between the sets of quantum and classical correlations are schematically depicted in fig .
[ fig : sets ] . of classical correlations without shared randomness ; the rectangle represents the set @xmath46 of classical correlations with shared randomness ; the ellipsoid represents the set @xmath45 of quantum correlations with shared randomness . ]
in practical applications , losses ( due to imperfections in the experimental implementations or artificially introduced by a malicious provider ) can noticeably affect the effectiveness of dimension witnessing .
the main result of this section is to provide a threshold value for the detection efficiency that allows one to witness the dimension of the systems prepared by a source or to discriminate between its quantum or classical nature .
the task is to determine whether a given conditional probability distribution belongs to a particular convex set , namely @xmath46 or @xmath45 ( see remark [ rmk : sr ] ) .
the situation is illustrated in figure [ fig : task ] . and
@xmath46 of the sets of quantum and classical correlations are represented as in figure [ fig : sets ] . in the presence of loss ,
only a subset of the possible correlations is attainable .
the subset , surrounded by bold line in the figure , is parametrized by detection efficiency @xmath138 .
the task is to find the threshold value in @xmath138 such that dimension witnessing is still possible .
for example , when the task is to discriminate between the quantum or classical nature of a source , one is interested in achieving correlations in the dark area of the figure , and our goal is to determine the values of @xmath138 such that this area is not null . ]
the experimental implementation is constrained to be lossy , namely it can be modeled considering an ideal preparing device followed by a measurement device with non - ideal detection efficiency .
this means that any povm @xmath21 on bob s side is replaced by a povm @xmath139 with detection efficiency @xmath138 , namely @xmath140 we notice that each lossy povm has one outcome more than the ideal one , corresponding to the no - click event . in a general model
, the detection efficiency @xmath138 may be different for any povm @xmath21 .
nevertheless , in the following we assume that they have the same detection efficiency , which is a reasonable assumption if the detectors have the same physical implementation @xcite .
analogously given a set @xmath77 of povms we will denote with @xmath141 the corresponding set of lossy povms . upon defining @xmath142 with @xmath143 $ ] , one clearly has @xmath144 to attain our task we maximize a given dimension witness over the set of lossy povms as given by eq . . due to the model of loss introduced in eq . and to the freedom in the normalization of dimension witnesses given by lemma [ thm : norm ] , in the following without loss of generality for any dimension witness @xmath145 as given in eq .
it is convenient to take @xmath146 then we have the following lemma .
[ thm : lossy ] given a set @xmath147 of states and a set @xmath148 of povms @xmath149 , for any linear dimension witness @xmath88 with @xmath20 $ ] , @xmath150 $ ] , and @xmath22 $ ] normalized as in eq .
one has @xmath151 one has @xmath152\\ & = \eta \ , w(r , p^{(1 ) } ) , \end{aligned}\ ] ] where the first equality follows from eq . and the second from the fact that @xmath153 due to the normalization given in eq . .
in particular from lemma [ thm : lossy ] it follows that for any linear dimension witness @xmath145 one has @xmath154 due to lemma [ thm : lossy ] , it is possible to recast the optimization of dimension witnesses in the presence of loss to the optimization in the ideal case . then due to lemma [ thm : pure ] it is not restrictive to carry out the optimization with pure states and no shared randomness .
consider the case where @xmath155 , @xmath156 , and @xmath157 .
using the technique discussed in appendix [ sect : optw ] one can verify that the witness given by eq . with the following coefficients
@xmath158 is the most robust to non - ideal detection efficiency .
this fact should not be surprising , as we notice that this witness relies on only @xmath159 out of @xmath160 outcomes . according to @xcite ,
we denote it @xmath161 . in @xcite
( see also @xcite ) it was conjectured that for any dimension @xmath10 the dimension witness @xmath161 is tight in the absence of loss .
now we provide upper and lower bounds for the maximal value @xmath162 where the maximization is over any set @xmath163 of states and any set @xmath54 of povms @xmath164 with @xmath165 .
[ thm : recbound ] for any dimension @xmath10 we have @xmath166 .
the statement follows from the recursive expression @xmath167 , where @xmath168 and noticing that @xmath169 and @xmath170 can be optimized independently .
a tight upper bound for @xmath171 was provided in @xcite . in the following lemma
we provide a constructive proof suitable for generalization to higher dimensions .
[ thm:2bound ] for dimension @xmath108 we have @xmath172 .
the statement follows from standard optimization with lagrange multipliers method and from the straightforward observation that given two normalized pure states @xmath173 and @xmath174 , if a pure state @xmath175 can be decomposed as follows @xmath176 then @xmath177 .
making use of lemmas [ thm : recbound ] and [ thm:2bound ] , we provide upper and lower bounds on @xmath178 as follows @xmath179 where the second inequality follows from the non discriminability of @xmath180 states in dimension @xmath10 ( see @xcite ) .
we now make use of these facts to provide our main result , namely a lower threshold for the detection efficiency required to reliably dimension witnessing .
we consider the problem of lower bounding the dimension of a system prepared by a non - characterized source in proposition [ thm : etac ] , as well as the problem of discriminating between the quantum or classical nature of a source in proposition [ thm : etad ] .
[ thm : etac ] for any @xmath10 there exists a dimension witnessing setup such that it is possible to discriminate between the quantum and classical nature of a @xmath10-dimensional system using povms with detection efficiency @xmath138 whenever @xmath181 furthermore one has @xmath182 we provide a constructive proof of the statement .
take @xmath155 , @xmath183 , and @xmath184 , and we show that @xmath161 satisfies the thesis .
we notice that the maximum value of @xmath161 attainable with classical states is given by @xmath185 @xcite
. then @xmath186 is the minimum value of the detection efficiency such that @xmath161 can discriminate a quantum system from a classical one .
due to lemma [ thm : lossy ] we have eq . . from eq .
the lower and upper bounds for @xmath186 given in eq . straightforwardly follow .
notice that @xmath161 in eq . can be numerically evaluated with the techniques discussed in appendix [ sect : optid ] .
figure [ fig : etac ] plots the value of @xmath186 for different values of the dimension @xmath10 of the hilbert space @xmath0 .
the threshold in the detection efficiency when @xmath108 is @xmath187 , going asymptotically to @xmath188 with @xmath10 as @xmath189 . as in eq . as a function of the dimension @xmath10 , obtained through numerical optimization of @xmath161 with algorithm [ thm : algo2 ] .
the lower bound ( lower line ) and upper bound ( upper line ) given by eq . are also plotted .
as expected , the upper bound is tight for @xmath108 .
the detection efficiency @xmath186 asymptotically goes to @xmath188 as @xmath190 since its upper and lower bound do the same . ]
[ thm : etad ] for any @xmath10 there exists a dimension witnessing setup such that it is possible to lower bound the dimension of a @xmath180-dimensional system using povms with detection efficiency @xmath138 whenever @xmath191 furthermore one has @xmath192 we provide a constructive proof of the statement .
take @xmath155 , @xmath183 , and @xmath184 , and we show that @xmath161 satisfies the thesis .
we notice that the maximum value of @xmath161 attainable in any dimension @xmath193 is given by @xmath10 @xcite
. then @xmath194 is the minimum value of the detection efficiency such that @xmath161 can lower bound the dimension of a @xmath180 dimensional system . due to lemma
[ thm : lossy ] we have eq . .
from eq . the lower bound to @xmath194 given by eq . straightforwardly follows .
notice that @xmath161 in eq . can be numerically evaluated with the techniques discussed in appendix [ sect : optid ] .
figure [ fig : etad ] plots the value of @xmath194 for different values of the dimension @xmath10 of the hilbert space @xmath0 .
the threshold in the detection efficiency when @xmath108 is @xmath187 , going asymptotically to @xmath188 with @xmath10 as @xmath189 .
we notice that @xmath194 grows faster than @xmath195 , thus showing that for fixed dimension , the discrimination between the quantum or classical nature of the source is more robust to loss than lower bounding the dimension of the prepared states . as in eq . as a function of the dimension @xmath10 , obtained through numerical optimization of @xmath161 with algorithm [ thm : algo2 ] .
the lower bound ( lower line ) given by eq . is also plotted .
as expected , the lower bound is tight for @xmath108 .
the detection efficiency @xmath194 asymptotically goes to @xmath188 as @xmath190 since its lower bound does the same ( and @xmath196 is a trivial upper bound ) . ]
in this work we addressed the problem whether a lossy setup can provide a reliable lower bound on the dimension of a classical or quantum system .
first we provided some relevant properties of the sets of classical and quantum correlations attainable in a dimension witnessing setup .
then we introduced analytical and numerical tools to address the problem of the robustness of didws , and we provided the amount of loss that can be tolerated in dimension witnessing .
the presented results are of relevance for experimental implementations of didws , and can be naturally applied to semi - device independent qkd and qracs .
we notice that , while we provided analytical proofs of our main results , i.e. propositions [ thm : etac ] and [ thm : etad ] , their optimality as a bound relies on numerical evidences . in particular , they are optimal if the dimension witness @xmath161 is indeed the most robust to loss for any @xmath10 , which is suggested by numerical evidence obtained with the techniques of appendix [ sect : optw ] and appendix [ sect : optid ] .
thus , a legitimate question is whether the bounds provided in propositions [ thm : etac ] and [ thm : etad ] are indeed optimal .
moreover , it is possible to consider models of loss more general than the one considered here , e.g. one in which a different detection efficiency is associated to any povm . a natural generalization of the problem of didws , in the ideal as well as in the lossy scenario , is that in the absence of correlations between the preparations and the measurements . in this case , as discussed in this work , the relevant sets of correlations are @xmath30 and @xmath31 , which are non - convex as shown in section [ sect : didw ] .
the non convexity of the relevant sets allows the exploitation of non - linear witnesses - as opposed to what we did in the present work .
an intriguing but still open question is whether there are situations in which this exploitation allows to dimension witness for any non - null value of the detection efficiency .
another natural generalization of the problem of didws is that of entangled assisted didws , namely when entanglement is allowed to be shared between the preparing device on alice s side and the measuring device on bob s side .
this problem is similar to that of super - dense coding @xcite .
consider again fig . [
fig : setup ] . in the simplest super - dense coding scenario ,
alice presses one button out of @xmath197 , while bob always performs the same povm ( @xmath198 ) obtaining one out of @xmath199 outcomes .
the dimension of the hilbert space @xmath0 is @xmath200 , but a pair of maximally entangled qubits is shared between the parties . in this case , the results of @xcite imply that a classical system of dimension @xmath201 ( quart ) can be sent from alice to bob by sending a qubit ( corresponding to half of the entangled pair ) .
consider the general scenario where now the two parties are allowed to share entangled particles .
the super - dense coding protocol automatically ensures that by sending a qubit alice and bob can always achieve the same value of any didw as attained by a classical quart .
remarkably , the super - dense coding protocol turns out not to be optimal , as we identified more complex protocols beating it .
in particular , we found a @xmath202 situation for which , upon performing unitary operations on her part of the entangled pair and subsequently sending it to bob , alice can achieve correlations that can not be reproduced upon sending a quart .
this thus proves the existence of communication contexts in which sending half of a maximally entangled pair is a more powerful resource than a classical quart .
this observation is analogous to that done in @xcite , where it was shown that entangled assisted qracs ( where an entangled pair of qubits is shared between the parties ) outperform the best of known qracs . for these reasons
we believe that the problem of entangled assisted didws deserves further investigation .
we are grateful to nicolas brunner , stefano facchini , and marcin pawowski for very useful discussions and suggestions .
m. d. thanks anne gstottner and the human resources staff at icfo for their invaluable support .
this work was funded by the spanish fis2010 - 14830 project and generalitat de catalunya , the european percent erc starting grant and q - essence project , and the japanese society for the promotion of science ( jsps ) .
given a linear dimension witness @xmath145 the following algorithm converges to a local maximum of @xmath95 .
[ thm : algo1 ] for any set @xmath203 of pure states and any set @xmath204 of povms @xmath205 , 1 .
let @xmath206 { |\psi_i^{(n)}\rangle}$ ] , 2 .
let @xmath207 \sqrt{\pi_k^{j,(n)}}\right\}^2 $ ] , 3 .
normalize @xmath208 , 4 .
normalize @xmath209 with @xmath210 .
as for all steepest - ascent algorithm , there is no protection against the possibility of convergence toward a local , rather than a global , maximum .
hence one should run the algorithm for different initial ensembles in order to get some confidence that the observed maximum is the global maximum ( although this can never be guaranteed with certainty ) .
any initial set of states and any initial set of povms can be used as a starting point , except for a subset corresponding to minima of @xmath95 .
these minima are unstable fix - points of the iteration , so even small perturbations let the iteration converge to some maxima .
the parameter @xmath211 controls the length of each iterative step , so for @xmath211 too large , an overshooting can occur .
this can be kept under control by evaluating @xmath95 at the end of each step : if it decreases instead of increasing , we are warned that we have taken @xmath211 too large . referring to fig .
[ fig : setup ] , the simplest non - trivial scenario one can consider is the one with @xmath105 preparations and @xmath212 povms each with @xmath213 outcomes , one of which corresponding to no - click event . in this case one has several tight classical didws . applying algorithm [ thm : algo1 ]
we verified that among them the most robust to loss is given by eq . with coefficients given by eq . .
the following lemma proves that the povms maximizing @xmath161 for any dimension @xmath10 are such that one of their elements is a projector on a pure state , thus generalizing a result from @xcite . for any fixed set @xmath218 of pure states
define @xmath219 , @xmath220 , and @xmath221
. then clearly @xmath222 , @xmath223 and @xmath224 for any @xmath62 . from eq .
it follows immediately that the optimal set @xmath225 of povms @xmath226 is such that @xmath227 $ ] .
the optimum of @xmath161 is achieved when @xmath216 is the sum of the eigenvectors of @xmath228 corresponding to positive eigenvalues . upon denoting with @xmath229 the eigenvalues of @xmath230 , the weyl inequality ( see for example @xcite ) @xmath231 holds for any @xmath232 . since @xmath233 and @xmath234 for any @xmath62 and for any @xmath235
, the thesis follows immediately .
the same remarks made about algorithm [ thm : algo1 ] hold true for algorithm [ thm : algo2 ] .
nevertheless , we verified that in practical applications algorithm [ thm : algo2 ] always seems to converge to a global , not a local maximum .
this can be explained considering that without loss of generality it optimizes over a smaller set of povms when compared to algorithm [ thm : algo1 ] .
moreover , we noticed that the optimal sets of states and povms are real , namely there exists a basis with respect to which states and povm elements have all real matrix entries .
a similar observation was done in @xcite in the context of bell s inequalities .
n. brunner , s. pironio , a. acn , n. gisin , a. a. mthot , and v. scarani , phys .
lett . * 100 * , 210503 ( 2008 ) .
s. wehner , m. christandl , and a. c. doherty , phys . rev .
a * 78 * , 062112 ( 2008 ) .
m. m. wolf and d. perez - garca , phys .
lett . * 102 * , 190504 ( 2009 ) .
r. gallego , n. brunner , c. hadley , and a. acn , phys .
lett . * 105 * , 230501 ( 2010 ) .
m. hendrych , r. gallego , m. miuda , n. brunner , a. acn , and j. p. torres , nature phys .
* 8 * , 588 - 591 ( 2012 ) .
h. ahrens , p. badziag , a. cabello , and m. bourennane , arxiv : quant - ph/1111.1277 . m. pawowski and n. brunner , phys .
a * 84 * , 010302 ( 2011 ) .
li , m. pawowski , z .- q .
yin , g .- c .
guo , and z .- f .
han , arxiv : quant - ph/1109.5259 . m. pawowski and m. ukowski , phys .
a * 81 * , 042326 ( 2010 ) .
i. l. chuang and m. a. nielsen , _ quantum information and communication _ ( cambridge , cambridge university press , 2000 ) .
this is not the case in the hybrid scenario where different types of detectors ( e.g. photodetectors and homodyne measurements ) are used .
a similar scenario was proposed for example in the context of bell inequalities @xcite .
d. cavalcanti , n. brunner , p. skrzypczyk , a. salles , v. scarani , phys .
a * 84 * , 022105 ( 2011 ) .
masanes , arxiv : quant - ph/0210073 c. h. bennett and s. j. wiesner , phys .
lett . * 69 * , 2881 ( 1992 ) .
masanes , arxiv : quant - ph/0512100 .
r. bhatia , _ positive definite matrices _ ( princeton university press , princeton , 2006 ) .
t. franz , f. furrer , and r. f. werner , phys .
* 106 * , 250502 ( 2011 ) . | device independent dimension witnesses provide a lower bound on the dimensionality of classical and quantum systems in a `` black box '' scenario where only correlations between preparations , measurements and outcomes are considered .
we address the problem of the robustness of dimension witnesses , namely that to witness the dimension of a system or to discriminate between its quantum or classical nature , even in the presence of loss .
we consider the case when shared randomness is allowed between preparations and measurements and provide a threshold in the detection efficiency such that dimension witnessing can still be performed . |
de sitter spacetimes are becoming of increasing interest in theoretical physics for a variety of reasons .
they provide interesting arenas in which to study the classical no - hair conjecture first proposed by ruffini and wheeler @xcite , which states that after a given distribution of matter undergoes complete gravitational collapse , the only long range information of the resultant black hole is its electromagnetic charge , mass and angular momentum .
the verification of this conjecture for a scalar field minimally coupled to gravity in asymptotically flat black - hole spacetimes has been extended to its de sitter counterpart @xcite,@xcite . while it is tempting to extend the no - hair theorem claim to all forms of matter , it is known that some long range yang - mills and/or quantum hair can be `` painted '' on a black hole @xcite .
explicit calculations have been carried out which verify the existence of a long range nielsen - olesen vortex solution as a single stable hair for a schwarzschild black hole in four dimensions @xcite , although it might be argued that this situation falls outside the scope of the classical no - hair theorem due to the non trivial topology of the string configuration .
another motivation for studying de sitter ( ds ) spacetimes is connected with the recently proposed holographic duality between quantum gravity on ds spacetime and a quantum field theory living on the past boundary of ds spacetime @xcite .
this proposed correspondence is undergoing extensive study in various directions @xcite-@xcite .
so far this work suggests that the conjectured ds / cft correspondence has a lot of similarity with the ads / cft correspondence , although some interpretive issues remain @xcite
. motivated by these considerations , we pursue in this paper a study of vortex solutions in de sitter spacetimes .
we have already shown that the @xmath4 higgs field equations have a vortex solution in both a four dimensional ads spacetime @xcite and in a four dimensional schwarzschild - ads background @xcite . employing the well known ads / cft correspondence conjecture
, the boundary conformal field theory can detect the presence of the vortex in the four dimensional ads spacetime : the mass density of the vortex solution is encoded in the discontinuity of the two - point correlation function of the dual conformal operator @xcite .
we have also shown that vortex solutions exist in the background of rotating kerr - ads and charged reissner - nordstrom - ads black holes @xcite , even in the extremal case.concurrently , it has recently been shown that in an asymptotically ads spacetime that a black hole can have scalar hair @xcite .
indeed , insofar as the no - hair theorem is concerned it has been shown that there exists a solution to the @xmath5 einstein - yang - mills equations which describes a stable yang - mills hairy black hole that is asymptotically ads @xcite . we therefore seek to learn if an analog of vortex holography discussed in @xciteholds true for ds spacetime . in this article
we take the first steps toward consideration of such a holographic phenomenon by searching for possible solutions of the abelian higgs field equations in a four dimensional ds background .
although an analytic solution to these equations appears to be intractable , we confirm by numerical calculation that vortex solutions do exist in ds spacetime .
we find both static and time - dependent vortex solutions .
the static solution corresponds to a core of vortex field energy located within the cosmological horizon in a static patch of ds spacetime . in the time - dependent solution
, the energy of the vortex core is diluted by the cosmological expansion in an inflationary patch . to our knowledge
this is the first construction of vortex solutions in asymptotically ds spacetimes .
we also consider the implications of our solutions for the recently conjectured ds / cft correspondence @xcite .
we compute the renormalization group flow associated with the time - dependent vortex solution in the inflationary patch and find that the generalized @xmath0-function monotonically increases , in accord with the generalized @xmath0-theorem @xcite . in section two
, we solve numerically the abelian - higgs equations in the static ds background for different values of the cosmological constant . in section three ,
we compute the effect of the vortex solution on the ds spacetime . in section four , we solve numerically the same equations in a big bang patch of ds background .
this is the first investigation of time dependent vortices in curved spacetime . in section five
, we obtain the behaviour of the ds @xmath0-function of this solution and find that it increases ( decreases ) in an expanding ( contracting ) ds patch .
we give some closing remarks in the final section .
in this section , we consider the abelian higgs lagrangian in the background of de sitter spacetime @xmath6 where @xmath7 is a complex scalar klein - gordon field , @xmath8 is the field strength of the electromagnetic field @xmath9 and @xmath10 in which @xmath11 is the covariant derivative .
we employ planck units @xmath12 which implies that the planck mass is equal to unity .
we use the following four dimensional static ds spacetime background @xmath13 where the cosmological constant @xmath14 is equal to @xmath15 the horizon is at @xmath16 and so the range of the coordinate @xmath17 is bounded to @xmath18 . in this coordinate system @xmath19 is a future - pointing timelike killing vector in only one diamond of the penrose diagram @xcite which generates the symmetry @xmath20 for any constant @xmath21 in other diamonds of the penrose diagram , this killing vector is spacelike or else past - pointing timelike . after redefining the new fields
@xmath22 by @xmath23 and employing a suitable gauge , the equations of motion for a string with winding number @xmath24 are @xmath25 @xmath26 where @xmath27 and @xmath28 note that by changing @xmath29 to @xmath30 , equations ( [ eqxds ] ) and ( [ eqpds ] ) change to the equations of motion in an ads background @xcite , or to the @xmath31 equations discussed in @xcite .
we have solved the above equations numerically for ds spacetimes with @xmath32 and unit winding number using the over - relaxation method @xcite .
figure [ fig12 ] shows the results for our calculation and also compares them with the vortex fields in the ads spacetime @xcite . for fixed @xmath33 , the @xmath34 field of the ds spacetime decreases with decreasing @xmath29 , in contrast to the ads spacetime which @xmath34 field increases with decreasing @xmath35 in general the value of the @xmath34 field for ads
spacetime is always greater than its value for any ds spacetime for a given @xmath36 . in other words , by increasing the cosmological constant @xmath14 from @xmath37 to @xmath38 the @xmath34 field decreases for fixed @xmath36 .
the @xmath34 field of flat spacetime is located between the ds and ads cases where @xmath39 . over this same range of @xmath14 the value of the @xmath40 field increases . *
* physically the negative cosmological constant exerts additional pressure on the vortex , causing it to become thinner ; a positive cosmological constant has the opposite effect , causing the core of the vortex to expand beyond the thickness it would have in flat spacetime .
moreover , we expect that the effect of the vortex on ds spacetime is to create a deficit angle in the metric ( [ dsmetric ] ) by replacing @xmath41 which @xmath42 is a constant . in the next section , we verify this point . for @xmath43 the coefficients of the @xmath44 and @xmath45 terms in equations ( [ eqxds ] ) and ( [ eqpds ] )
change sign and the numerical solution of the corresponding equations outside the cosmological horizon shows that the values of the @xmath34 and @xmath40 fields remain at their respective constant values of @xmath46 and @xmath47 .
consequently the vortex is confined totally behind the cosmological horizon . in the next section
we will find that in a different patch of ds spacetime the vortex solution exists everywhere in space and changes with time .
in this section , we consider the effect of the vortex on ds@xmath48 spacetime described by the static metric ( [ dsmetric ] ) .
this requires finding the solutions of the coupled einstein - abelian higgs differential equations in ds@xmath48 .
this is a formidable problem even for flat spacetime no exact solutions have yet been found . using a thin - core approximation and numerical methods
, we obtained in @xcite the effect of the vortex on ads@xmath48 . here , we use the same method to obtain the effect of the vortex on ds@xmath48 spacetime . to obtain physical results
, we make some approximations .
first , we again assume the thin - core approximation , namely that the thickness of the vortex is much smaller that all other relevant length scales .
second , we assume that the gravitational effects of the string are weak enough so that the linearized einstein - abelian higgs differential equations are applicable
. for convenience , in this section we use the following form of the metric of ds@xmath49 @xmath50 in the absence of the vortex , we must have @xmath51 @xmath52 , yielding the well known metric ( [ dsmetric ] ) of pure ds@xmath53 . employing the two assumptions concerning the thickness of the vortex core and its weak gravitational field , we solve numerically the einstein field equations , @xmath54 to first order in @xmath55 where @xmath56 is the energy - momentum tensor of the abelian higgs field in the ds background . by taking @xmath57 , where @xmath58 is the usual ds@xmath59 metric and @xmath60 is its first order correction , and writing @xmath61 we obtain first - order corrections to the three functions @xmath62 and @xmath63 in ( [ abcexpa ] ) . hence in the first approximation the equations ( [ einstein ] )
become @xmath64 where @xmath65 is the energy momentum tensor of the vortex field in the ds@xmath59background metric , and @xmath66 is the correction to the einstein tensor due to @xmath67 .
the rescaled components of the energy momentum tensor of string in the background of ds@xmath48 are given by @xmath68 where @xmath69 the einstein equations ( [ eineq ] ) are @xmath70 we consider the case @xmath71 for which the behaviour of @xmath34 and @xmath40 fields are given in figure [ fig12 ] . by using equations ( [ stresssph ] ) , we obtain figure [ fig125 ] showing the behaviour of the stress tensor components inside the cosmological horizon as a function of @xmath72 whose minimum value is @xmath73 * * we note that outside the cosmological horizon , the stress tensor vanishes due to the constant values of the vortex fields . with this knowledge ,
we solve the coupled differential equations ( [ eins ] ) , inside and outside the cosmological horizon which gives the behaviour of the functions @xmath74 and @xmath75 the results are plotted in figure [ fig126 ] .
these results emphasize that the functions @xmath76 and @xmath77 are constant over the entire range of @xmath36 to this order in @xmath78 . a very tiny deviation from the constant values in the curves @xmath79
are due to the numerical calculation .
hence by a redefinition of the time coordinate in ( [ abcmetric ] ) the metric can be rewritten as @xmath80 which is the metric of ds space with a deficit angle .
so , the effect of the vortex on ds@xmath48 spacetime is to create a deficit angle in the metric ( [ dsmetric ] ) by replacing @xmath81 where @xmath82 is a constant , since @xmath83 the above calculation for the effect of the vortex on ds spacetimes is not restricted to the special case @xmath84 , with other values of cosmological parameter @xmath29 , the final result is the metric ( [ dsdef ] ) , with some other deficit angle @xmath42 .
in this section , we consider the abelian higgs lagrangian ( [ lag ] ) in the following coordinate system @xmath85 where the coordinate @xmath86 after using the new fields @xmath87 given by ( [ new ] ) , and applying the radial variable @xmath88 the equations of motion for a string with winding number @xmath24 become @xmath89 @xmath90 by numerically solving equations ( [ eqxdscrunch],[eqpdscrunch ] ) , we are able to show that a vortex solution exists on a ds spacetime background for different values of the winding number @xmath24 and cosmological parameter @xmath29 . as in the pure ads case
@xcite and asymptotically ads spacetimes @xcite , @xcite the results indicate that increasing the winding number yields a greater vortex thickness .
furthermore , for a vortex with definite winding number , the string core increases with decreasing @xmath29 at constant positive time .
the @xmath34 and @xmath40 fields less rapidly approach their respective maximum and minimum values at larger distances as @xmath29 decreases . to obtain numerical solutions of ( [ eqxdscrunch ] ) and ( [ eqpdscrunch ] ) , we must first select appropriate boundary conditions . at large distances
physical considerations motivate a clear choice .
since in the limit @xmath91 , the results must be in agreement with the results of flat spacetime , we demand that the solutions approach the solutions of the vortex equations in the flat spacetime given in ref .
this means that we demand @xmath92 and @xmath93 as @xmath94 goes to infinity . on the symmetry axis of the string ,
we take @xmath95 and @xmath96 for all time .
finally we take @xmath97 and @xmath98 everywhere ( except on the symmetry axis ) at the initial time @xmath99 . in the following ,
we consider the case @xmath100 we can straightforwardly obtain similar results for other values of the winding number @xmath101 we then employ a polar grid of points @xmath102 where @xmath94 goes from @xmath47 to some large value @xmath103 which is much greater than @xmath29 and @xmath104 runs from a large negative number @xmath105 to a large positive number @xmath106 using finite difference methods , we rewrite the non linear partial differential equation ( [ eqxdscrunch ] ) and ( [ eqpdscrunch ] ) as @xmath107 @xmath108 where @xmath109 and @xmath110 for the interior grid points , the coefficients @xmath111 can be straightforwardly determined from the corresponding continued differential equations ( [ eqxdscrunch ] ) and ( [ eqpdscrunch ] ) . using the well known successive overrelaxation method @xcite for the above mentioned finite difference equations , we obtain the values of @xmath34 and @xmath40 fields inside the grid .
initially , inside the grid points , we set the value of @xmath34 and @xmath40 fields @xmath47 and @xmath46 respectively which we denote them by @xmath112 and @xmath113 .
then these values of @xmath34 and @xmath40 fields are used in the next step to obtain the values of @xmath34 and @xmath40 fields inside the grid which could be denoted by @xmath114 and @xmath115 . repeating this procedure ,
the value of the each field in the @xmath116-th iteration is related to the @xmath117-th iteration by @xmath118 @xmath119 where the residual matrices @xmath120 and @xmath121 are the differences between the left and right hand sides of the equations ( [ findiffx][findiffxx ] ) and ( [ findiffxp ] ) respectively , evaluated in the @xmath116-th iteration .
@xmath122 is the overrelaxation parameter .
the iteration is performed many times to some value @xmath123 such that @xmath124 and @xmath125 for a given error @xmath78 .
it is a matter of trial and error to find the value of @xmath122 that yields the most rapid convergence .
some typical results of this calculation are displayed in figures [ fig13 ] and [ fig14 ] for value of @xmath126 . in the figure [ fig13 ]
, time is running from @xmath105 to @xmath47whereas in figure [ fig14 ] the time runs from @xmath47 to @xmath106 we notice that by increasing the time from @xmath105 to @xmath127 the string core increases rapidly to a well defined value at time @xmath128 . then increasing the time to more positive values , the string core increases more and more such that at time @xmath129 the @xmath34 and @xmath40 fields approach the constant values @xmath47 and @xmath46 respectively .
so , we conclude that by increasing the time from @xmath130to @xmath131 for which the constant time slices of the spacetime ( [ bigbang ] ) become bigger and bigger , the string thickness becomes wider and wider .
moreover the energy density of the string spreads to bigger and bigger distances as time increases due to the larger size of the constant time slices .
the preceding calculation of vortex fields on ds spacetimes is not restricted to the special case of @xmath84 ; with other values of the cosmological parameter @xmath29 , we obtain similar results . for example , another typical results of our calculation are displayed in figures [ fig15 ] and [ fig16 ] for value of @xmath132 for time slices near @xmath133 the behaviour of the vortex fields do not change by changing the cosmological parameter @xmath134 the physical reason is that at this special time the spacetime is flat and independent of the cosmological constant , .
however , if we consider a constant positive time slice , then by increasing the cosmological parameter @xmath135 the string thickness decreases ; for a constant negative time slice , by increasing the cosmological parameter @xmath29 , it increases .
the asymmetric behaviour of string thickness for positive and negative times is due to the inflation function @xmath136 in ( [ bigbang ] ) .
hence by increasing @xmath29 , the density of vortex field time - constant contours concentrates mainly near the time constant @xmath128 slice . to obtain the effect of the vortex on the big bang patch of ds@xmath48spacetime
, we use the results of the preceding section , in which we found that vortex induces a deficit angle on the static spacetime metric given in ( [ dsdef ] ) . by using the following transformations , @xmath137 the following metric @xmath138 can be written in the well known static ds spacetime with deficit angle @xmath139 given by ( [ dsdef ] ) .
hence the effect of the vortex on a big bang patch of ds spacetime is to create a deficit angle in the x - y plane that is constant as the ( locally ) flat spatial slice evolves in time .
similar calculations for larger winding numbers show that increasing the winding number yields a greater vortex thickness in each constant timeslice relative to winding number @xmath100 this tendency runs counter to that of increasing @xmath29 , for which the vortex thickness decreases on a constant positive time slice .
as @xmath140 we can increase the winding number to larger and larger values , and we find that the resultant thickness increases .
hence the effect of increasing winding number to thicken the vortex dominates over the thinning of the vortex due to decreasing cosmological constant .
what we have found about the behaviour of the vortex in the big bang patch of ds spacetime can be straightforwardly generalized to the big crunch patch of the ds spacetime .
the big crunch patch is given by replacing @xmath141 in the metric ( [ bigbang ] ) .
increasing the time from @xmath130to @xmath142 for which the constant time slices of the big crunch spacetime become smaller and smaller , the string thickness becomes narrower and narrower .
in fact , increasing the time , the energy density of the string concentrates to smaller and smaller distances due to the smaller size of the constant time slices .
obtaining evidence in support of ( or against ) a conjectured ds / cft correspondence is somewhat harder to come by than its ads / cft counterpart .
although it is tempting to think of the former as a ` wick - rotation ' of the latter , a number of subtleties arise whose physical interpretation is not always straightforward @xcite .
one way of making progress in this area is via consideration of the uv / ir correspondence . in both the ads and ds cases
there is a natural correspondence between phenomena occurring near the boundary ( or in the deep interior ) of either spacetime and uv ( ir ) physics in the dual cft .
solutions that are asymptotically ds lead to an interpretation in terms of renormalization group flows and an associated generalized ds @xmath0-theorem .
this theorem states that in a contracting patch of ds spacetime , the renormalization group flows toward the infrared and in an expanding spacetime , it flows toward the ultraviolet @xcite .
more precisely , the @xmath143-function in @xmath144-dimensional inflationary ds is given by , @xmath145 where @xmath146 is the unit normal vector to a constant time slice and @xmath147 is the energy density on a constant time hypersurface .
it is natural to consider the time - dependent solution of the previous section in this context .
although we have not solved the einstein - abelian - higgs equations exactly , we have shown to leading order that the vortex solution we obtained induces a deficit angle into de sitter spacetime , and so to this order our solution is ( locally ) asymptotically de sitter .
furthermore , the gauge field values of our solution asymptotically approach the constant values expected in a de sitter vacuum , and so at large distances ( near the boundary ) we expect that our solutions will in general be asymptotically de sitter , at least locally .
so we want to examine the behaviour of the ( [ cfunction ] ) in the big - bang patch ( [ bigbang ] ) in the presence of a vortex .
the energy density of the vortex is given by : @xmath148 using the curves of the vortex fields we obtained for different time slices in figures ( [ fig13 ] ) and ( [ fig14 ] ) , we get figure ( [ fig17 ] ) for the @xmath0-function in terms of time at a fixed point @xmath149 of space .
we find that by increasing the time from @xmath105 to @xmath150 , the @xmath0-function monotonically increases as the universe expands .
we emphasize that this monotonic increase is not restricted to this special value of @xmath94 : by changing the value of @xmath94 we find the same increasing behaviour of the @xmath0-function as time evolves .
similar calculations show that in the big - crunch patch of ds spacetime , the @xmath151-function decreases monotonically as time evolves from @xmath105 to @xmath106
we have solved the nielsen - olesen equations in a static ds@xmath48 background , and found that the higgs and gauge fields are axially symmetric , with non - zero winding number .
our solution in the limit of large @xmath29 ( small cosmological constant ) reduces to the well known flat - space solution .
the solution ( to leading order in the gravitational coupling ) induces a deficit angle in ds@xmath48 .
we find in the static patch that an increasingly positive cosmological constant tends to make a thicker vortex solution due to cosmological expansion .
we solved the same equations in the big bang patch of the ds@xmath48 background , and found that by increasing the time from early times at @xmath152 to @xmath128 , the string thickness increases from an infinitesimally small value to a finite size . increasing the time to vary large values @xmath153 the string thickness grows more and more , so that at future infinity , the string fields approach constant values everywhere . at past infinity , which is in fact the ds@xmath48 horizon , although the magnitude of the magnetic field goes to infinity near the string axis , the spatial dimensions shrink relative to spatial dimensions at other times , yielding a constant value for the magnetic flux of the vortex .
in contrast to past infinity , at future infinity the magnetic field is very tiny over a big range of spatial coordinate , giving rise to the same value of the magnetic flux of the vortex at past infinity .
the presence of a vortex in asymptotically ds spacetime technically violates the cosmic no - hair theorem in that the spacetime is not pure de sitter ; rather it is only locally ds due to the deficit angle induced from the non trivial topology of the vortex solution .
violation of the cosmic no hair theorem has also been observed in einstein - maxwell - dilaton theory with a positive cosmological constant @xcite .
our results are in accord with the generalized ds @xmath0-theorem , providing further evidence in favor of a conjectured ds / cft correspondence .
an interesting future challenge is that of obtaining a holographic description of a vortex solution in ds spacetimes .
the boundary field theory will have to be on a sphere with the same deficit angle as that induced by the vortex . in the ads case ,
the conformal two - point correlation function is obtained by evaluating the bulk propagator of a scalar field between two points on the boundary by integrating over all spacelike geodesics paths between the points , and the presence of a vortex is signified by a discontinuity in this correlation function @xcite .
however in the ds case the geodesic paths will be timelike , and will necessarily have to penetrate the cosmological horizon to detect a vortex in the static patch .
not all of @xmath154 ( @xmath155 ) is causally connected to the vortex in the big bang ( big crunch ) patch , raising issues of causality reminiscent of those considered in the case of particles forming black holes in @xmath156 dimensions @xcite , and their resolution is far from clear .
we close by commenting on the physical relevance of our solutions .
since we employ planck units @xmath12 , the constant @xmath29 is in units of the planck length .
we have found that only for @xmath157 do our solutions appreciably differ from the flat space case . in this regime
the radius of curvature of the de sitter spacetime becomes comparable to the planck length , equalling it at @xmath158 , and one might be concerned that our results will be substantively modified due to quantum gravitational effects .
nevertheless we maintain that our results have both utility and validity for the following reasons .
first , the qualitative behaviour of our solutions for large @xmath29 is the same as for @xmath157 ; for example figure ( [ fig17 ] ) will retain the same qualitative features for all finite values of @xmath159 , and we have numerically checked this for @xmath160 .
however smaller values of @xmath29 more clearly highlight the significance of the features of these solutions , which is why we have presented them here .
second , it is at the very least useful to know what the classical physics is that underlies any quantum effects , and our solutions furnish will such knowledge when it becomes available . finally , quantum gravitational effects will typically induce quantitative corrections to our results of order @xmath161 , which is still only about @xmath162 even for @xmath84 .
hence we expect the qualitative features of our solutions to remain even when quantum gravity is taken into account .
inclusion of quantum gravitational effects is , of course , important .
other problems include the generalization of our solutions to asymptotically ds spacetimes with black holes and the possible ds / cft correspondence of these solutions .
work on these problems is in progress . | we investigate vortex solutions to the abelian higgs field equations in a four dimensional de sitter spacetime background . we obtain both static and dynamic solutions with axial symmetry that are generalizations of the nielsen - olesen gauge vortices in flat spacetime .
the static solution is located in the static patch of de sitter space .
we numerically solve the field equations in an inflationary ( big bang ) patch and find a time dependent vortex soution , whose effect to create a deficit angle in the spacetime . we show that this solution can be interpreted in terms of a renormalization group flow in accord with a generalized @xmath0-theorem , providing evidence in favour of a ds / cft correspondence .
hep - th/0203003 + watpphys - th02/02 * vortices in de sitter spacetimes * + a.m. ghezelbash@xmath1 and r. b. mann@xmath2 + @xmath3department of physics , university of waterloo , + waterloo , ontario n2l 3g1 , canada + @xmath1department of physics , alzahra university , + tehran 19834 , iran + + |
a variety of observational signatures point to the importance of outflowing gas within many types of active galactic nuclei ( agns ) . blueshifted absorption features ( in broad absorption line quasars , or balqsos ; see , e.g. , * ? ? ?
* ) are seen in approximately 15% @xcite of radio - quiet quasars , with velocities up to 0.1c .
in addition , radio - loud quasars display relativistic , collimated outflows .
more recently , both uv and x - ray absorbing gas have been observed in approximately 60% of seyfert 1 galaxies @xcite .
observational estimates hint that the mass outflow rate is nearly equal to the mass inflow rate ( for a review of mass outflow in agns , see * ? ? ?
the development of models to explain the mass outflow rates , geometry , and general kinematics of these winds has proven difficult , but progress on a number of possibilities has been encouraging . from very early on , researchers examined both radiative wind models ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and , to explain radio jets , hydromagnetic models ( e.g. , * ? ? ?
* hereafter bp82 ) .
later models of radiatively - driven winds were able to explain the balqso outflow velocities and population fraction ( * ? ? ?
* hereafter , mcgv95 ) as well as the single - peaked emission lines @xcite .
the line - driven models were also demonstrated in hydrodynamical simulations @xcite , which elucidated the density structure , geometry , and kinematics of the flow in two dimensions . in addition , models with combinations of continuum- and line - driving by x - rays were presented by @xcite .
one of the concerns with line - driving in agns has been the possible over - ionization of the gas by x - rays in the central continuum ( leaving the wind with too few lines to intercept flux in the lines ) : the need for `` shielding gas '' to prevent this overionization has been a persistent concern ( e.g. , mcgv95 , * ? ? ?
* ; * ? ? ?
in addition , hydromagnetic winds have also been developed to gain insight into the observations ; some of these models have included radiative driving . in these magnetocentrifugal winds ,
gas is loaded onto and tied to large - scale magnetic field lines ; those field lines then fling the matter centrifugally away from the disk , like beads along a wire .
magnetocentrifugal acceleration is commonly used to explain large - scale collimated outflows in radio galaxies ( bp82 ; for reviews of magnetocentrifugal driving , see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the context of the broad emission line region ( blr )
, magnetocentrifugal outflows made up of clouds were first called upon to explain the single - peaked broad emission lines , outflow velocities , and densities @xcite .
the `` torus '' ( obscuring gas that plays a central role in the unification model , yielding a dependence of observed properties with inclination angle ; see * ? ? ?
* ) has been explained as a dusty , continuous magnetic wind ( * ? ? ? *
hereafter kk94 ) where radiative acceleration on dust affects the wind geometry .
hydromagnetic disk winds with radiation pressure were also developed by @xcite as an alternative explanation for the population fraction of balqsos and to understand cloud confinement within the outflow .
in addition , another magnetocentrifugal wind model was developed to explain single peaked emission lines arising from a distribution of clouds @xcite , as well as the dynamics of the warm absorber in ngc 5548 @xcite .
finally , some effects of magnetic fields ( not including magnetocentrifugal driving as in bp82 ) have also been considered in two - dimensional hydrodynamic simulations @xcite . at the present time , both radiatively driven winds and magnetocentrifugally driven winds ( with radiative acceleration added in some models ) offer compelling but competing pictures of the key physics in the cores of agns .
no clear observational evidence yet discriminates the dominant physics of wind - launching in agns . for radiatively - driven winds , three main lines of evidence hint that radiative driving should be important in models of wind dynamics .
first , @xcite find a correlation between uv luminosity and the observed outflow velocity , which agrees with a basic prediction of line - driven wind models @xcite .
in addition , the `` ghost of ly-@xmath1 '' @xcite , as well as the realization that the radiative momentum removed from the continuum in blueshifted absorption lines in balqsos is a significant fraction of the total radiative momentum ( mcgv95 ) are both also important clues that radiative acceleration is an important component to any self - consistent model of agn winds .
there are still , however , concerns about how the `` shielding '' of the wind works @xcite .
in addition , in the case of stellar disk winds , observations of two nova - like variables seem to show that their winds are not dominated by radiative driving @xcite .
( this conclusion , however , rests on the prediction that line equivalent widths are direct measures of mass outflow rate , which may not be the case ; see * ? ? ?
* ) on the other hand , the leading model for collimated radio jets in agns ( bp82 ) already calls upon large scale , dynamically important magnetic fields , as do the hydromagnetic wind models ( mentioned above , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) that have also had success in explaining agn observations .
in addition , such a hydromagnetic wind would have no difficulty with overionization , and so might naturally serve as a `` shield '' for radiative acceleration further from the central source .
there are , however , currently no models which address the interplay between magnetocentrifugal driving and radiative acceleration ; if such wind models could be constrained , we may be able to observationally distinguish the physics of wind launching in the cores of agns , and gain insight into the role that outflows play both in accretion and in feedback of those winds in the galaxy and surrounding matter .
this paper develops a detailed photoionization and dynamical model for magnetocentrifugal winds in agns .
the model is designed to explore the radiative transfer within such magnetocentrifugal winds , but also to help understand how radiative driving impacts the kinematic structure of such winds .
constructing such a model builds the foundation for later work to determine absorption and emission line profiles in order to compare with observations and check for the presence of magnetocentrifugal winds within agns .
an overview of this model is presented in
[ modeloverview ] ; the model is then defined in detail in [ detailedmodel ] .
an examination of the structure of a particular `` fiducial '' magnetocentrifugal , radiatively - accelerated wind is described in [ fiducialmodel ] and then the dependences of radiative acceleration on some initial parameters are shown in [ modeldependences ] . conclusions and directions for future work
are summarized in [ results ] .
before examining the model in detail , it is instructive to present a summary of the basic design .
the semianalytic model developed here includes magnetic acceleration and radiative acceleration of a continuous wind launched from an accretion disk .
a detailed treatment of radiative transfer is included by using version 96.00 of the photoionization simulation program cloudy , last described by @xcite .
these elements are introduced in an approximate schematic of the wind model shown in figure [ thinwind ] , depicting a portion of the accretion disk and outflowing wind . in this figure ,
radiation ( entering from the left side of the schematic ) first encounters a purely magnetocentrifugally accelerated wind , which will be referred to as the `` shield '' , as it absorbs radiation from the central continuum .
the shield is introduced as a separate component in order to cleanly differentiate the effect of shielding from radiative acceleration ; radiative driving of the shield is therefore not considered in this work . beyond that shield is an optically thin , radiatively and magnetically accelerated wind streamline ( which we will henceforth refer to as the `` wind '' ) . in this portion of the model ,
both magnetocentrifugal and radiative forces help accelerate the flow off of the accretion disk ; the magnetic fieldlines are shown by the black lines bordering the outflow .
the included radiative acceleration is calculated by first simulating the photoionization within both the shield and the wind along radial paths such as the thick , black lines in the figure .
the separation of the wind into two components is , of course , artificial . in reality
, radiative acceleration would gradually increase in importance for portions of the wind that are increasingly shielded .
however , splitting the outflow into these two components allows a first - order , qualitative solution that can be used to gain some understanding of how magnetic and radiative forces might interact , and how a magnetic wind may be able to act as a radiative `` shield '' to allow more efficient radiative acceleration . while artificial , this method of using two wind components has already been used successfully to examine winds with magnetocentrifugal and radiative driving on dust ( e.g. , kk94 , * ? ? ?
figure [ flowchart ] presents a schematic flow chart of how model calculations proceed .
the wind starts as a self - similar magnetohydrodynamic model that yields the pure magnetocentrifugal wind solution ( covered in [ mhdwind ] ) .
next , simulations of the photoionization balance of that wind streamline ( [ photosims ] ) are run , and the resultant ionization balance and transmitted continuum are used to calculate the radiative acceleration behind the shield ( [ radacccalc ] ) .
next , the radiative acceleration is input ( as a function of polar angle , @xmath2 ) back into the self - similar magnetohydrodynamic model , modifying the structure of the wind streamline , while leaving the shield unaffected .
this process is then repeated , simulating the photoionization of that modified wind and recalculating the radiative acceleration terms .
we typically iterate five to eight times to converge to a final equilibrium solution . with the basic model now summarized , it may be instructive to compare and contrast it to the recent wind model examined in @xcite , where the combination of magnetic and radiative forces in disk winds is also investigated .
@xcite concentrates on numerical simulations of time - dependent winds with line driving and magnetic forces .
these numerical simulations allow large - scale models of outflows that are valuable in understanding global wind structures in many different astrophysical contexts .
in contrast , the radiatively - driven components of this model are more localized , since radiative acceleration is applied to the shielded streamline only .
this model setup has been chosen for its flexibility in radiative acceleration modeling ; using cloudy for such radiative acceleration models enables computations not only of radiative line driving but also of continuum driving , and allows us the freedom to include dust and easily vary the incident spectrum , for example . in addition
, the magnetic winds produced in @xcite are not magnetocentrifugal outflows ( as in bp82 ) , as are the semianalytic winds presented here .
further , these new models are steady - state , not time - dependent .
finally , semi - analytic , steady - state models allow an exploration of general behaviors through many parameter variations ; large - scale numerical simulations can usually vary only a few parameters . in summary , these models cover different facets of the disk - wind problem , yielding different perspectives on a complicated system .
in this section , we describe in detail the model s components , and derive key equations .
the magnetocentrifugal wind model is introduced first ( [ mhdwind ] ) , followed by the photoionization simulations ( [ photosims ] ) , radiative acceleration calculation ( [ radacccalc ] ) , and finally a discussion of the wind model equation of motion ( [ windeqmtn ] ) . to derive the equations governing the continuous magnetocentrifugal wind , we start with the equations of a stationary , axisymmetric magnetohydrodynamic ( mhd ) flow in cylindrical coordinates , make the assumption of self - similarity in the spherical radial coordinate , and utilize the continuity equation , conservation of angular momentum along the flow , and both the radial and vertical momentum equations , very much as in bp82 and kk94 ( see appendix [ selfsimwindeqns ] ) .
thermal effects are neglected in the wind , therefore effectively assuming that the wind starts out supersonic ( this assumption is checked later on , see [ assumptions ] ) . in deriving the wind equations
, we use the same simplifications as bp82 , except for the added complication that energy is not conserved in the radiatively - accelerated system due to the constant input of radiative energy into the wind . in the original formulation of bp82 , conservation of energy supplied an additional constraint which allowed a simplification of the equations of motion to two first - order differential equations . because energy is not conserved in this flow , the equivalent of three first - order differential equations must be integrated , solving for three parameters simultaneously instead of two as in bp82 . the detailed setup and derivation of this set of equations of motion
are given in appendix [ selfsimwindeqns ] .
the integration of the momentum equations starts by specifying the following initial parameters : the mass loading of the wind ( the ratio of mass flux to magnetic flux in the magnetocentrifugal wind , @xmath3 , where @xmath4 is the mass density of the wind , @xmath5 is the poloidal velocity of the wind ( @xmath6 ) , and @xmath7 is the poloidal magnetic field strength ) , the specific angular momentum of gas and field in the wind , and the power - law exponent ( @xmath8 ) that describes the change in density with spherical radius : @xmath9 . also input , as parameters , are the mass of the central black hole , @xmath10 , the wind s launch radius on the disk , @xmath11 , and the density at the base of the wind at the launch radius , @xmath12 .
the program employs a `` shooting '' algorithm ( using the slatec routine dnsq ; * ? ? ?
* ) to integrate from the singular point ( the alfvn point ) to the disk , solving for the height of the singular point above the disk ( @xmath13 ) and the slope of the streamline at both the disk and the alfvn point ( @xmath14 and @xmath15 ) by matching the integration results to boundary conditions on the disk .
after solving for the position of the alfvn point , the equations of motion are integrated from the disk to a user - specified height beyond the alfvn point ; along the streamline , the run of velocity , density , and magnetic field are calculated .
this code has been tested ( without radiative acceleration ) against the solution given in bp82 and have duplicated their results to within 8% .
this is close to the previously reported 4% variance in the recalculation of bp82 reported in @xcite .
the difference in these new results comes not only from using higher precision calculations compared to bp82 ( we use the same precision as * ? ? ?
* ) , but in addition , a more complex set of equations is being solved . next ,
version 96.00 of the photoionization code cloudy @xcite is used to simulate the absorption of the magnetocentrifugal shield and wind as well as the ionization state at the wind streamline .
photoionization simulations of the shield and wind are used to calculate the radiative acceleration and allow considerable flexibility in gas parameters ( such as gas abundances , dust , central continuum , etc ) .
this flexibility is gained at the cost of simulating the photoionization state of the wind as if it were a static medium , as cloudy assumes ; this is only true of the recombination timescale for the gas is much shorter than the transit timescale of gas in the region simulated ( @xmath16 ) .
we have however , verified , _ a posteriori _ , that @xmath17 for all of the radiatively accelerated ions at the base of the wind where radiative acceleration is important , using the recombination rate approximations in @xcite and @xcite to calculate the recombination times . even at high latitude , @xmath18 for all of the significant ions . at the highest latitudes , highly ionized o , which has the longest recombination time among the significantly radiatively accelerated ions , has a recombination timescale that is still a factor of two less than the transit time
meanwhile , highly ionized fe ions , which dominate the low level of radiative acceleration at high latitudes , have recombination timescales an order of magnitude less than the transit timescale .
using cloudy is therefore a reasonable approximation for the radiative equilibrium within our winds ( especially since , with our code , adiabatic and advective effects are added , see [ contwindphoto ] ) . as cloudy enables a flexible , self - consistent calculation of radiative acceleration , we accept this approximation to enable these calculations . of basic importance to the photoionization simulations is the illuminating spectral energy distribution ( sed ) .
for the purposes of these simulations , an sed adapted from @xcite is used ( an example of the sed is shown in figure [ transmittedcont ] ) .
this sed is input into cloudy via the generic `` agn '' continuum with @xmath19 k , @xmath20 @xcite , @xmath21 , and @xmath22 ( @xmath23 defines a single power - law that would describe the continuum between @xmath24 and 2 kev , @xmath25 is the slope of the low - energy component of the big blue bump , and @xmath26 is the x - ray power - law exponent ; our value of @xmath23 is taken from the middle of the range 0.8 to 1.0 given in @xcite ) .
spectral signatures and radiative acceleration also depend , of course , on the column density in the wind .
observations yield only rough constraints for this , so these columns as left as free parameters ; the effect of varying these columns will be investigated in this paper .
the columns throughout the shield and wind are set by the columns at the base of the shield and wind , denoted @xmath27 for the hydrogen column at the base of the wind . as the wind rises above the disk and accelerates , that column density ( @xmath28 ) drops as a function of height due to mass conservation ( an example of this
is shown later in fig .
[ fiducialwind ] ) . investigating
the shielding ability of such a dynamic shielding column is of central interest to this paper , and will be addressed in [ shieldcolumnvariations ] .
as depicted in figure [ thinwind ] , cloudy simulates the photoionization of the wind along radial sight lines through the shield and through the wind , ending at the site of radiative acceleration on the wind streamline .
the continuum incident on the wind streamline is also calculated .
the photoionization state and continuum at the end of the cloudy calculations are recorded , and then used to compute the radiative acceleration of that gas .
finally , the acceleration is tabulated and applied as a function of @xmath2 along the wind streamline by inputing the angle - dependent radiative acceleration into the gravitational term ( this is covered in more detail in appendix a , see eqn .
[ gammainselfsim ] ) . whereas cloudy is designed to simulate the photoionization balance of gases as in the shield and wind
, it can not easily incorporate adiabatic and advection effects : cloudy has no knowledge of the particular velocity profile of the wind in the overarching model , or the temperature difference between successive photoionization models as the wind climbs above the disk .
therefore , this model calculates both advective heating and adiabatic cooling in the wind , and adds those terms manually into cloudy s simulations .
both terms largely cancel in the wind , and have only a negligible effect on outflow dynamics , but they are included in all of the models for completeness .
the model then incorporates the above - mentioned results for the ionization structure and radiation field to calculate the radiative forces felt by the wind .
there are two different kinds of radiative acceleration to consider : continuum acceleration ( including radiative acceleration on dust ) and line acceleration .
it is convenient to express the radiative acceleration in terms of @xmath29 , @xmath30 where @xmath31 is the acceleration due to radiation , and @xmath32 is the local gravitational acceleration . in general , for continuum and line acceleration ,
the radiative acceleration is given by @xmath33 where @xmath34 is the local flux ( the flux transmitted through both the shield and wind column ) , @xmath35 is the electron density , @xmath4 is the gas density , @xmath36 is the speed of light , @xmath37 is the gravitational constant , @xmath10 is the mass of the central black hole , @xmath38 and @xmath39 are cylindrical coordinates centered on the black hole , and @xmath40 & @xmath41 are the `` force multipliers '' that relate how much the radiative forces on the gas ( on the line and continuum opacity , respectively ) exceed the radiative forces on electrons alone .
they are given below in terms of the continuum opacity , @xmath42 , and the line opacity , @xmath43 , for the continuum and lines , respectively : @xmath44 with @xmath45 where @xmath46 is the photon frequency , @xmath47 is the local ( transmitted ) flux in the line at the frequency of line number @xmath48 , @xmath49 is the sound speed in the gas , @xmath50 is the thermal line width , and @xmath51 compares the opacity of the line ( for a given ionization state of the gas ) to the electron opacity , representing all of the atomic physics in the radiative acceleration calculation .
the last remaining variable , @xmath52 , is often called the `` effective electron optical depth '' and encodes the dynamical information of the wind in the radiative acceleration calculation . this dynamical information is important because in an accelerating medium , one must also account for the doppler shift of the atomic line absorption energy in the accelerating gas relative to the emitted line photon s energy : beyond the sobolev length , @xmath53 , included in @xmath52 , a line photon will be doppler - shifted out of the thermal width of the absorption line and can escape the gas ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the above - mentioned force multipliers are calculated using the resonance line data of @xcite with solar abundances .
since @xmath40 , the continuum multiplier , depends only on the ionization state , it is tabulated solely as a function of height in the wind . in contrast to @xmath40 , the line multiplier ( @xmath54 ) is tabulated for a range of values of the parameter @xmath52 .
later , when calculating the equation of motion for the wind , the local velocity gradient is used to compute the actual value of @xmath52 , which is then used to linearly interpolate the table of @xmath55 and then evaluate the radiative acceleration . the force multiplier computation has been tested against @xcite , who also calculated radiative acceleration from photoionization simulations .
figure [ forcemultcompare ] compares these new calculations results against their fits ( noting that there is a typo in their eq . [ 2.9 ] ; z .- y .
li , personal communication ) , where the force multipliers as a function of the ionization parameter @xmath56 is presented ( @xmath56 is the ratio of hydrogen - ionizing photon density to hydrogen number density @xmath57 , given by @xmath58 , where @xmath59 is the number of incident hydrogen - ionizing photons per second , and @xmath60 is the distance from the continuum source ) .
overall , good agreement is found , especially considering that @xcite point out that their fit deviates from their calculations at low values of @xmath56 .
the increase in the newly - calculated continuum force multiplier over @xcite is most likely due to the different continuum opacity database included in cloudy 96 compared to the code ( mappings ) that was used in @xcite .
the multiplier values and trends with ionization parameter are still clearly very similar , however .
simply using the sobolev length , @xmath53 , in equation [ mlinesdef ] can be misleading . in early simulations
, we found that this sobolev length could , in regions where the gas is slowly accelerating , be much larger than the physical size of the shield and wind combined .
this is clearly not physical , so a simple non - sobolev method was employed to calculate the size of the combined shield and wind column .
first , for the wind , the length of the absorbing column is simply limited to the minimum of the wind s sobolev length and its true ( physical ) length .
second , for the shield , the length of the absorbing column is given by the minimum of the shield s sobolev length , its true physical length , and the average length of the column that absorbs photons at the wavelength of the dominant accelerating atomic lines .
this last length - scale requires more explanation . to calculate this length , the top 20 line force multipliers ( for individual atomic lines )
are found for each polar angle @xmath2 , and for each of those lines , the column of each particular ion in the shield is read from the cloudy simulations .
an average shielding column for all of the high - opacity lines is then calculated . to test this approximation ,
we have used lists of the top 10 , 20 , and 50 transitions in the wind and used them to calculate the limiting column in the shield . changing the number of lines included does not significantly change the final wind solutions , especially since this effect is only important at the extreme base of the wind . without considering all of these constraints on the size of the absorbing column , the sobolev length can significantly overestimate the optical depth in the shield ( even overestimating the physical length of the columns , for small accelerations ) , which results in the line driving acceleration dropping below the continuum acceleration .
other non - sobolev effects ( such as line blanketing within the shielding gas or the wind ) are not considered in this model ; for a consideration of these effects , see @xcite . with a length - scale found from the sum of the wind and shield lengths found above , that length - scale is then substituted for the sobolev length in the calculation of @xmath41 in eq . [ mlinesdef ] . including this physical length of the wind and shield will introduce dependences on the sizes of the wind and shield columns , which will be examined in [ modeldependences ] .
given the results from the ionization and radiative acceleration calculations , the next step is to solve for the effect of radiative acceleration on the wind , taking the magnetocentrifugal wind model already computed and augmenting its acceleration with radiative forces . to do this ,
the full equation of motion ( the euler equation ) is integrated along the streamline of the self - similar wind solution , recording @xmath29 , the radiative acceleration .
@xmath29 is then input into the self - similar model in the subsequent iteration ( see appendix a ; specifically , eqn .
[ gammainselfsim ] ) . in its simplest form ,
euler s equation is given by @xmath61 where the @xmath62 represent the various forces included .
as we are interested in steady - state winds , @xmath63 in eq .
[ simpeuler ] is set to zero .
as already mentioned , gravitational , radiation , and lorentz forces are included , while thermal terms are not .
this yields the expression below , with the magnetic force term is split into pressure and tension components .
m \rho}{r^2}\hat{r } - \frac{1}{8 \pi } \nabla b^2 + \frac{1}{4 \pi } ( \mathbf{b } \cdot \nabla)\mathbf{b}\end{aligned}\ ] ] for these calculations , it is more intuitive to integrate the equation of motion along the flow already given by the magnetocentrifugal wind solution .
therefore , taking the dot product of euler s equation with @xmath65 , which is defined as the direction along the flow , and expanding and simplifying the left - hand side of the equation , one finds : @xmath66 where @xmath67 in the same way , if @xmath2 is defined via @xmath68 the gravitational term can be written as @xmath69 \frac{g m}{r^2}\hat{r } \cdot \hat{s } & = & - [ 1 - \gamma(\theta ) ] \frac{g m}{r^2 + z^2 } \cos ( \theta - \theta_f).\end{aligned}\ ] ] next , we take the dot product of @xmath65 with the magnetic terms to find @xmath70 \cdot \hat{s } & = & -\frac{b_{\phi}}{4 \pi \rho r } \frac{\partial ( r b_{\phi})}{\partial s}.\end{aligned}\ ] ] combining all of those terms , the full euler equation is obtained : @xmath71 \frac{g m}{(r^2 + z^2 ) } \cos ( \theta - \theta_f ) \nonumber \\ & & - \frac{b_{\phi}}{4 \pi \rho r } \frac{\partial ( r b_{\phi})}{\partial s}.\label{fulleulereqn}\end{aligned}\ ] ] this equation is still dependent on @xmath72 , however , which can be eliminated by appealing to the induction equation and @xmath73 for such axisymmetric systems ( see , e.g. , * ? ? ?
* ) , to yield a relation between @xmath5 and @xmath72 : @xmath74 substituting this expression into the euler equation yields : @xmath75 \frac{g m}{(r^2 + z^2 ) } \cos ( \theta - \theta_f ) -\frac{b_{\phi}}{4
\pi \rho r } \frac{\partial ( r b_{\phi})}{\partial s}.\end{aligned}\ ] ] to evaluate the effective optical depth @xmath52 , an expression is required for @xmath76 , the spherical radial gradient of the spherical radial velocity . since the code integrates quantities only parallel to the flow , approximations to the perpendicular velocity gradients
must be used . assuming that the derivatives of @xmath2 and @xmath77 with distance along the streamline are small ( verified _ a posteriori _ to be true ) : @xmath78 , \\ & \approx & \cos ( \theta - \theta_f ) \left(\frac{dv_p}{dr } \sin \theta + \frac{dv_p}{dz } \cos \theta \right ) , \\ & \approx & \cos ( \theta - \theta_f ) \frac{dv_p}{ds } ( \sin \theta_f \sin \theta + \cos \theta_f \cos \theta ) , \\ & = & \cos^{2 } ( \theta - \theta_f ) \frac{dv_p}{ds}.\end{aligned}\ ] ] this integration procedure has been tested with radiative acceleration turned off , where it reproduces the original self - similar velocity profile to within one part in @xmath79 . with
the radiative acceleration turned on , the entire code has repeatedly converged within approximately eight iterations to an equilibrium magnetic wind structure ( see fig . [ winditerations ] ) .
these tests show that consistent solutions are found and that radiation pressure does indeed affect the shielded component of the wind . as with any steady - state wind dynamics problem , one must search for and consistently pass all critical points ( e.g. , * ? ? ?
critical points mark the location in the wind where the flow speed is equal to the speed of information propagation in the wind , and mark locations in the solutions where solutions branches , or roots , meet ; in the case of radiatively accelerated winds , the location of the critical point is set by the information propagation speed of radiative - acoustic wave , or abbott speed ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ) . in integrating the equation of motion for the radiatively - accelerated wind
, this code searches for critical points by looking for multiple roots in the solution to the equation of motion .
however , no critical points due to radiative acceleration are present in any of the wind solutions we have found ( the magnetocentrifugal wind does , however , always pass through its own alfvn critical point ) . to check this result
, we have duplicated the work of ( * ? ? ?
* hereafter , fs99 ) , verifying that for simple wind geometries and without magnetocentrifugal acceleration , the integration code does indeed encounter a radiative critical point as predicted and found by fs99 .
in particular , for the field geometries and forces used in fs99 , we have found identical solutions to both their analytical and numerical calculations .
we then gradually add , to the fs99 model , new components that are present in our new calculations .
when the centrifugal acceleration and the enforced corotation near the base of the wind are introduced into the framework of fs99 , the centrifugal acceleration overwhelms the acceleration of the second root that was present in fs99 .
therefore , only a single root is found , and with only a single root , a critical point can not be present in our solutions . another way to check
this is to examine the limit of winds launched at very large angles to the accretion disk .
indeed , for those large angles , the radiative acceleration begins to dominate the centrifugal acceleration , and the critical point reappears .
therefore , for the geometry of these magnetocentrifugal winds , where the angle of the outflow to the accretion disk surface is less than 60@xmath80 , no radiative critical point is expected within these solutions : a radiative critical point will not be present when centrifugal acceleration is dominant .
this model includes simplifying assumptions about the outflow in order to make these calculations possible . in this subsection ,
the assumptions and limitations of the model are summarized , as are the reasons for allowing those assumptions .
first and foremost among these assumptions is self - similarity . while enabling a relatively quick and flexible model that can be used to survey a wide variety of outflows ,
this assumption does impose constraints on the dynamics of the model .
however , the assumption of self - similarity is essential to the magnetocentrifugal wind solution , as it simplifies the complicated mhd equations .
simultaneously , the radial self - similarity accommodates , very easily , the radial geometry for the photoionization simulations .
the above - mentioned cloudy photoionization simulations assume a static medium , which is an approximation as well . as we show in [ photosims ] , however , this approximation is valid for the these wind models .
since the accretion disk is a boundary condition in these models , these winds are assumed to be loaded with matter from the disk .
accurate models of the accretion disk structure are beyond the scope of this paper , so it is assumed that the full mass outflow rate of the the wind is indeed input onto the magnetic fieldlines at the accretion disk surface .
in addition , the matter that is loaded onto those fieldlines is assumed to flow supersonically , i.e. , the gas has already passed the sonic point in the flow .
this is done to simplify the magnetocentrifugal wind equations and retain the basic model as outlined in bp82 ; as such , the same asymptotic expansions near the disk ( from bp82 ) are utilized here .
this treatment of the wind has been checked in several different ways .
first , the magnetic pressure in the wind is indeed greater than the thermal pressure throughout the entire wind .
also , the wind s final velocities are much greater than the sound speed at the base of the wind , showing that thermal effects are negligible in determining the final wind velocities .
finally , the lowest speeds found in the wind model are of order 20% of the sound speed , and such low mach numbers are found only very near the disk , at the base of the wind . given the above evidence of the dominance of magnetic fields and radiative acceleration , the `` cold - wind '' approximation is valid for these calculations . also , in the calculation of the radiative acceleration of the wind , an approximation to the velocity gradient along spherical rays is required .
this approximation is necessary because the euler integration for the wind yields velocity gradients only along the flow ( the poloidal velocity gradients ) , and thus the other components must be approximated geometrically ( see [ windeqmtn ] ) .
the approximations and limitations outlined above do constrain the use of this wind model , but in making these compromises , a versatile tool can be developed to study the chosen geometries and forces .
for definitiveness , this model is first employed to examine the radiative transfer within one fiducial irradiated magnetocentrifugal wind . for the purpose of this paper , ` fiducial ' is defined to indicate the parameters listed in table [ paramtable ] .
these parameters are not meant to represent a proposed model for any one particular agn , but to define a starting point from which to examine the structure of these outflows as well as the dependences of the outflows on the model parameters .
for instance , the shielding column of @xmath81 @xmath82 ( where @xmath83 represents the value of @xmath84 at the base of the wind , i.e. just above the accretion disk surface ) is chosen because it displays an amount of radiative acceleration between the extremes of the smaller and larger columns that will be tested .
similarly , the radiative wind column that is defined is again between the extremes of nearly optically thick @xmath85 @xmath82 and very optically thin @xmath86
@xmath82 . studying such
a model first will help bring into focus important issues concerning the interplay of dynamics and photoionization , and represents a foundation from which one can explore the parameter dependencies of the model .
this fiducial model was therefore run with the parameters given in table [ paramtable ] , and after eight iterations , converged to the final wind structure .
the results for the fiducial model are shown in figures [ windstreamline ] though [ winditerations ] . in these figures , an overview of the equilibrium state of this model
is presented .
the results displayed in these plots are discussed in detail below .
first , figure [ windstreamline ] shows the height of the poloidal streamline as a function of radius in units of the launching radius .
in addition , to illustrate the small difference in geometry between the final and initial wind models , figure [ windstreamlinedifference ] gives the fractional change in height as a function of radius .
both of these figures show that the wind still maintains a collimated state , achieving a height of @xmath87 ( where @xmath11 is the launching radius ) at a cylindrical radius of only @xmath88 .
so , in this fiducial model , despite the input from radiative acceleration , the wind maintains this streamline with only small changes in the structure of the wind throughout all iterations ( see fig .
[ windstreamlinedifference ] ) .
thus , for the case of the fiducial model with @xmath89 , this added acceleration does not significantly affect the structure of the magnetocentrifugal outflow .
these models do show changes in the velocity structure of the wind near the disk surface ( as will be shown in fig .
[ winditerations ] ) , but the poloidal wind structure does not change significantly : on the scale of figure [ windstreamline ] , the streamlines of the initial , purely magnetocentrifugal streamline would lie on top of the streamline shown .
lll @xmath12 & @xmath90 & initial density of the wind at the launch radius + @xmath91 & @xmath92 & gas shielding column at the base of the wind + @xmath93 & @xmath94 & gas column , behind the shield , that is radiatively accelerated + @xmath10 & @xmath95 & mass of the central black hole + incident spectrum & @xcite & spectrum for the central continuum + @xmath96 & @xmath97 & luminosity of the central continuum + @xmath11 & @xmath98 & launch radius of the wind + @xmath99 & 0.03 & dimensionless ratio of mass flux to magnetic flux in the wind + @xmath100 & 30.0 & normalized total specific angular momentum of the wind + @xmath8 & 1.5 & power - law describing variation of density with spherical radius in the wind : @xmath101 + dust in wind & no & presence of dust in the wind + .
note that the wind is still somewhat collimated , since the radiative acceleration input is fairly low for @xmath89 .
[ windstreamline],width=302 ] .
[ windstreamlinedifference],width=302 ] at logarithmically - spaced co - latitudinal angles along the streamline , cloudy photoionization simulations are run to determine the photoionization state of the gas , as well as the radiative transfer through the shield and wind .
changes in the continuum transmitted through the shield are shown in figure [ transmittedcont ] ; the various plots show the simulated continuum at various heights in the shield corresponding to the indicated columns .
as the shielding column decreases as a function of height above the disk , the shield transmits progressively more and more of the ionizing radiation .
this plot also displays how rapidly the column drops as a function of height above the disk : @xmath81 @xmath82 occurs at @xmath102 , @xmath103 @xmath82 at @xmath104 , @xmath105 @xmath82 at @xmath106 , and @xmath84 drops to @xmath107 @xmath82 at @xmath108 . the flux transmitted through the shield
then illuminates the radiatively accelerated wind ; results from the photoionization simulations for the wind are presented in figure [ fiducialwind ] , where the streamline , velocity , density , ionization parameter and temperature in the wind are plotted . in figure
[ fiducialwind]a , the height of a wind streamline as a function of distance along the streamline ( labeled @xmath109 , given in units of the initial radius , @xmath11 ) is shown ; this plot simply recasts the structure of the flowline shown in figure [ windstreamline ] in terms of @xmath109 for comparison with the remaining plots . in figure [ fiducialwind]b ,
velocities along the streamline are plotted , showing not only the rapid acceleration in the wind , but also comparing the components of the wind s velocities .
all velocities are given in units of the keplerian velocity at the base of the wind , @xmath110 ( @xmath111 km s@xmath112 for the parameter values in table [ paramtable ] ) .
this plot shows that the vertical velocity is dominant in these winds at large distances ( again showing the wind is somewhat collimated ) , with the radial and azimuthal velocities becoming approximately equal far from the launching radius of the outflow . near the very base of the disk , the radial velocity quickly dominates both the azimuthal and vertical velocities ( qualitatively similar to the velocity structure calculated for radiatively - dominated flows as in mcgv95 ) .
most importantly , though , we note the extraordinarily rapid acceleration of the gas from the disk ; such acceleration is a hallmark of both magnetocentrifugal as well as radiatively - dominated winds , which usually accelerate to their terminal velocities in a distance on the order of their launching radius .
due to mass conservation , the extremely rapid acceleration of the wind causes a sharp drop in both the number density and the column density with height above the disk : both the number density and column density immediately drop by three orders of magnitude as the wind rises above the disk .
this is displayed in figure [ fiducialwind]c .
this overall drop in density is extremely important for the ionization state of the wind , not only for the observational ramifications ( i.e. , what ions are present in various parts of the wind ) but for the acceleration of the wind as well , as will be shown very shortly .
figures [ fiducialwindveldiff ] and [ fiducialwinddensitydiff ] show in more detail how the radiative acceleration leads to a substantial change ( by approximately a factor of two ) in both velocity and density with height near the disk surface .
the changes in velocity for the pure magnetocentrifugal wind as compared to the magnetocentrifugal and radiatively - accelerated wind is shown in figure [ fiducialwindveldiff ] .
the difference in the density profile for the pure magnetocentrifugal wind as compared to the magnetocentrifugal and radiatively - accelerated wind is shown in figure [ fiducialwinddensitydiff ] .
( note that all densities are normalized to their value at the base of the wind . )
the corresponding ionization state of the wind and temperature are shown in figure [ fiducialwind]d .
most striking is the dramatic rise in the ionization parameter as the wind rises above the disk , which is simply due to the drop in density and in shielding already mentioned .
the ionization parameter is of prime interest , as the radiative acceleration in resonant lines is dependent on the number of atomic lines in the gas ; the rapid ionization of the gas prompts questions about how efficient line - driving will be within this magnetocentrifugal wind .
in addition , can a wind with such a dramatic drop in column density form an effective `` shield '' ? and for these models , how does the radiative acceleration then compare to magnetic acceleration ?
, width=302 ] these questions are addressed in figure [ lineovercont ] . from the cloudy simulations summarized in figure [ fiducialwind ] , both the bound - free and bound - bound radiative acceleration are calculated ; the resultant acceleration ( compared to the local gravity ) for the fiducial model is shown in figure [ lineovercont ] .
this model shows that both line - driving and continuum - driving have important roles to play , with the line - driving dominating the continuum driving in the high - density , low - ionization part of the wind , and continuum driving dominating line - driving at larger distances , when the density is much lower .
it is important to note that for the parameters of the fiducial model with @xmath89 , magnetic acceleration is still much greater than either line or continuum acceleration .
line - driving is greater than continuum - driving in only part of the outflow ( although this can change with the density at the base of the wind , as will be shown in [ ninitvariations ] ) , as the ionization parameter is low enough only at the base of the wind for significant numbers of atomic lines to exist .
in addition , radiative driving is not immediately important at the extreme base of the outflow , because of the low fluxes that penetrate the columns there .
meanwhile , the acceleration due to continuum - driving stays close to the eddington ratio , at @xmath113 .
this value is reasonable , as most of the continuum acceleration comes from electron scattering , so that @xmath114 would be expected .
minor increases above that value very near the disk surface are due to bound - free transitions in the portion of the wind closer to the disk , where @xmath115 rises to @xmath116 .
as has already been shown , at this low eddington ratio the structure of the wind does not change significantly .
however , the velocity at the base of the outflow is affected .
the change in velocity due to radiative acceleration is shown in figure [ winditerations ] .
this figure shows both the poloidal velocity as a function of distance along the flowline , and the variation in that velocity with iterations of the model , therefore showing the convergence in the model .
figure [ winditerations ] shows that line - driving near the disk does significantly accelerate the wind , but magnetic driving determines the terminal velocity at larger radii .
this figure also displays how the code converges ; the relatively slow convergence during the first four iterations is the result of the program slowly increasing the radiative acceleration to the computed value ( increased slowly in order to avoid severely overestimating the radiative acceleration in the lines and causing sudden deceleration in later iterations ) . in the later iterations , the calculation converges to a final velocity profile using the full radiative acceleration .
this profile shows the affect of line - driving near the base of the wind , where the velocity increases above that of the initial magnetocentrifugal wind .
however , as already mentioned , the magnetocentrifugal wind ( in this model , where @xmath89 ) still determines the velocity at large distances . at those distances ,
the gas is too ionized to be appreciably accelerated by line driving .
, width=302 ] it is important to note that considering non - sobolev effects leads to large changes in @xmath41 , the line force multiplier , found in the above calculations : `` capping '' the sobolev absorption length - scale by the actual absorbing column length leads to smaller optical depths and larger line accelerations .
this is illustrated in figure [ figsobolevcomp ] , where the fiducial model has been calculated with the sobolev approximation as well as our non - sobolev treatment ( see section [ nonsoboleveffects ] ) .
the difference in force multiplier is due to the strict sobolev treatment overestimating the column for the low - acceleration gas near the base of the wind .
this relatively straightforward modification is very important to correctly estimate the optical depth , as can be seen in figure [ figsobolevcomp ] .
overall , in this section , the fiducial model has shown how the wind velocity , number density , column density , and radiative acceleration all interact to determine the final state of a magnetocentrifugal wind .
these components have not previously been self - consistently combined in a magnetocentrifugal model , and so yield a new look at the state of these winds .
in addition , the importance of the number densities and column densities to the final result are most apparent , and clearly merit further investigation , which will be addressed in [ modeldependences ] .
having analyzed the fiducial model in detail , and observed how that model s properties change with height , how sensitive are the trends in [ fiducialmodel ] to those fiducial parameters ?
this is an important question , and one of the key attributes of this self - similar model is that it allows some flexibility in the selection of initial parameters . in this section ,
we test for variations in the wind by examining how the wind changes as parameters are modified .
one of the most difficult issues for radiative driving in agns is over - ionization of the wind . as shown in
[ fiducialmodel ] , as the wind accelerates and its density decreases , the magnetocentrifugal outflow can easily become too ionized to be efficiently accelerated to escape velocity solely by atomic lines .
this is the problem of the `` shielding gas '' that was mentioned in [ intro ] : for pure radiative line - driving , some shielding gas is required to intercept the x - ray ionizing radiation so that the remaining uv resonant line photons can be absorbed by the wind and radiatively accelerate it to the escape velocity .
some important papers have already been dedicated to examining the concept of shielding gas , such as ( * ? ? ? * considering very detailed photoionization simulations of gas shields with constant column density ) and ( * ? ? ?
* where multidimensional hydrodynamics simulations with approximate radiative effects are considered ) . in contrast , the models presented here include a shield where the column density varies with height in a shielding wind , and where detailed photoionization simulations can be employed .
the mhd wind model presented here already launches a wind magnetocentrifugally , so it is immune to concerns of overionization .
is it therefore possible for a magnetocentrifugally - driven wind , with its commensurate drop in column density with height above the disk , to act as a shield , allowing for more efficient radiative acceleration beyond it ? we can test this question by simply varying the shielding column in the fiducial model and checking the radiative acceleration seen by a wind launched behind the shield . as shown in figure
[ gamma_columnscan ] , as the shielding column is increased from @xmath117 to @xmath118 , line - driving in the wind increases from @xmath119 to @xmath120 .
( recall that for this model , continuum driving is @xmath121 . )
this shows that , for large shielding columns ( @xmath122 ) , line driving can be up to an order of magnitude more effective than continuum - driving at the base of the wind .
this increase in acceleration is due to the absorption of the ionizing radiation by the shield , which allows a lower ionization state in the wind , and more line - driving due to more atomic lines .
it is also apparent that , as the shielding is increased , the resultant lower total flux at the base of the wind means that the onset of significant radiative acceleration is delayed : this accounts for the offset of maximum radiative acceleration from the disk surface as the shielding column is increased .
further , in figure [ gamma_columnscan_mag ] , the ratio of radiative acceleration to magnetic acceleration along a streamline is shown . since the mhd effects are still supplying most of the acceleration ( with an acceleration roughly equal to and opposite that of gravity )
, the ratio of radiative to magnetic acceleration looks much like that in figure [ gamma_columnscan ] .
magnetic effects dominate in these wind models , even when large columns of shielding are included . .
) [ gamma_columnscan],width=302 ] , but comparing the strength of the total radiative acceleration ( both line and continuum acceleration ) to the acceleration due to magnetic fields , where both are normalized relative to gravity . to compare with the magnetic field acceleration ,
the radiative acceleration is scaled by a geometrical factor ( see eq .
[ fulleulereqn ] ) that yields the radiative acceleration along the streamline .
[ gamma_columnscan_mag],width=302 ] we have therefore shown that a magnetocentrifugal outflow can act as a shield and increase the efficiency of line - driving in the wind . however , it can also be seen that line - driving is important in these models only at the base of the wind .
this arises not only from the drop in the shield s column density with height above the disk , but the drop in the wind s density as well ( and the commensurate rise of the ionization parameter ) .
owing to the increase of line - driving with decreasing ionization parameter , higher accelerations would also be expected at higher densities .
thus , we investigate the effect of changes in the initial density in the wind in figures [ gamma_densityscan ] and [ vpvsdensity ] . displaying the effect of a range of initial densities ,
figure [ gamma_densityscan ] shows that line - driving is only effective in these magnetocentrifugal winds at relatively high densities . since continuum driving is approximately constant ( and relatively independent of density ) at @xmath123 for @xmath124 ,
any line - driving below that level is insignificant for these winds .
thus , for initial densities @xmath125 , line driving falls below the level of continuum driving and ceases to be important . for
the highest density tested , @xmath126 , line driving dominates continuum driving for all locations in the wind .
( the variations in each acceleration curve as a function of @xmath109 shown on this plot are chiefly due to variations in ionization parameter : line - driving is high near the disk due to shielding and the relatively high density , and rises towards the end of the streamline due to the dropping flux levels at large distances . ) . )
[ gamma_densityscan],width=302 ] since @xmath12 clearly has a great impact on the radiative acceleration , how do such changes affect observables , such as the velocity ?
figure [ vpvsdensity ] shows how the variation in radiative acceleration affects the poloidal velocity ( @xmath127 ) of the outflow . as the initial density and radiative acceleration increase , the wind s velocity shows substantial variations from the pure magnetocentrifugal model ( which dominates the @xmath128 and @xmath129 models ) . in the case of @xmath130 , where the greatest difference in @xmath127 is seen , the velocity increases by a factor of @xmath131 to 3 close to the disk . beyond the region close to the disk ( @xmath132 )
, however , magnetocentrifugal driving still dominates the final velocities for these winds .
but the velocity differences near the disk may be observationally important , especially if acceleration near the base of the wind is the source of single - peaked emission lines ( as in * ? ? ?
if true , such emission lines may be critical in testing the differences between wind models .
, except that the change in velocity along the streamline is shown for the variety of densities . for this eddington ratio , the magnetocentrifugal wind still dominates , but as the density increases , the importance radiative driving increases .
[ vpvsdensity],width=302 ] having already tested the more obvious parameters of the initial density and shield column density , we now turn to one of the most crucial parameters for the efficiency of line - driving : the optical depth in the lines . under normal circumstances , where the gas velocity is sufficiently low , or where the gas column is very low , the optical depths in the lines are simply governed by the ionic columns themselves . for large accelerations or large columns ,
however , the optical depths are dominated by the sobolev length , which is defined as the distance over which the relative velocity between atoms is equal to the thermal width , so that a photon emitted by one atom could be absorbed by another within that sobolev length ( see [ linecontaccel ] ) . with the eddington ratio in this magnetocentrifugal wind model ( @xmath124 ) , both regimes can be important .
depending on the initial parameters prescribed , the wind column can be small enough such that the column alone determines the opacity in the wind ( instead of the velocity ) and therefore the amount of observed acceleration , so that the sobolev length is not important
. this is demonstrated in figure [ gamma_radcolumnscan ] , where the variation in line - driving with the radiatively accelerated wind column is presented ( the shielding column is held constant ) . for the larger columns , the optical depth in the lines increases and the radiative acceleration decreases .
for the smallest columns , very large radiative acceleration is predicted due to the low opacity in the lines .
this is critical for these models , for at significantly high column , the wind would see no significant line driving ( this is true for magnetocentrifugal wind columns with @xmath133 ) . .
) [ gamma_radcolumnscan],width=302 ] of key importance to applications to agn is understanding the acceleration of outflows as a function of the eddington ratio , @xmath134 . to investigate the impact of varying eddington ratios in our model , we present figure [ eddratioscan ] , which displays the radiative acceleration ( both the combined line and continuum acceleration in panel _ a _ as well as the line driving in panel _ b _ ) in the fiducial model for three different eddington ratios : @xmath135 and @xmath136 .
the largest variation in figure [ eddratioscan]a is the continuum driving increasing linearly with the eddington ratio . as expected , the continuum acceleration , relative to gravity , is roughly equal to the eddington ratio .
the line driving , on the other hand , can be seen in both the deviations from the approximately constant continuum acceleration in figure [ eddratioscan]a and in figure [ eddratioscan]b .
we begin examining this figure by concentrating on the first three models in figure [ eddratioscan ] , which have @xmath137 @xmath82 . in these models in figure
[ eddratioscan]a , the increase in acceleration due to line - driving , relative to the continuum - driving , decreases as the eddington ratio increases .
this can also be seen in figure [ eddratioscan]b , where the line - driving peaks near the disk ( @xmath138 ) for @xmath89 , but decreases for eddington ratios an order of magnitude larger ( where the gas is overionized ) and an order of magnitude smaller ( where the radiation field does nt have the momentum to accelerate the wind as strongly as at @xmath89 ) .
now we turn to the fourth model in figure [ eddratioscan ] , where we keep @xmath139 but increase the shielding level to @xmath140@xmath82 .
this model shows the importance of shielding gas for @xmath139 .
the increase in @xmath141 and the increase in shielding relative to the fiducial model allow for increased line radiative acceleration that jumps almost two orders of magnitude in strength near the base of the wind .
if magnetocentrifugal winds power outflows in agns , they must certainly be affected by the intense radiation field they experience ; in turn , such winds will also influence the efficiency of radiative acceleration .
this paper has explored the radiative transfer through these magnetocentrifugal winds , and how they both are affected by and affect radiative acceleration of outflows from agns .
the model has been used to explore the detailed dynamics and ionization of a fiducial magnetocentrifugal disk wind , showing the inter - relation between shielding column , initial number density , outflow velocity , eddington ratio , and acceleration . as a result of this study ,
these models have shown : 1 . a magnetocentrifugal outflow , acting as a `` shield '' , can improve the efficiency of line - driving by factors of approximately two to three @xmath142@xmath82 ( [ shieldcolumnvariations ] ) and by up to almost two orders of magnitude for @xmath143@xmath82 ( [ eddratiovariations ] ) .
a magnetocentrifugal wind has the advantage that it can be accelerated without regard to the ionization state , whereas radiatively - driven winds must have a low ionization parameter in order for a critical abundance of atomic lines to be present .
therefore , magnetocentrifugal winds could play an important role in acting as a radiation shield and allow large radiative accelerations .
it may also be possible that pressure differences ( mcgv95 ) or disk photons @xcite may help `` lift '' the shield ; neither of those effects are considered here .
later work with this model will include the effect of disk - emitted photons .
the efficiency of line - driving is strongly dependent on the density at the base of the wind ( [ ninitvariations ] ) .
this is due to the very critical dependence of line acceleration on the ionization parameter .
the lower the ionization state , the more lines exist to aid in the momentum transfer from outward - streaming photons .
the density at the base of the disk is therefore crucial to setting to line - driving within these magnetocentrifugal models .
3 . small columns ( @xmath144 ) within magnetocentrifugal winds can be significantly accelerated by line - driving ( [ radcolumnvariations ] ) .
this point demonstrates the importance of `` non - sobolev '' effects ; i.e. , that at low columns , the optical depths in the lines drop below the opacity given by the sobolev length , and at such low columns , the radiative acceleration can be underestimated by the simple sobolev approximation .
in addition , by examining the fiducial model and the above cases where model parameters are varied , these solutions have displayed the importance of considering the detailed interaction between the dynamics and photoionization in agn outflows .
calculations of the ionization parameter along the flow and in the variation of shielding and optical depth along the flow are central issues to modeling these winds .
the issues outlined above are a few of the dependences that arise from such modeling , and may indeed ( by the variation in acceleration and therefore velocity along the streamlines ) lead to tests to observationally determine the physics of wind launching in agns .
thus , while this model has been developed to help address the above questions about shielding and the affects of radiative driving on magnetocentrifugal winds , the solutions available are not limited to the above , examined cases .
future papers will study further the variation of radiative acceleration on model parameters such as the sed , atomic line lists used to calculate the acceleration , initial densities , eddington ratios , and other parameters of the model .
the model can also be used to explore the absorption and emission features from such a wind , as well as , for instance , the possible role of `` clouds '' within a continuous wind @xcite .
in addition , this model is in no way constrained to only study agns .
the same basic physical framework could also be employed to study winds from accretion disks surrounding young stellar objects or cataclysmic variables .
i gratefully acknowledge my advisor , arieh knigl , for many useful conversations throughout the course of this project . in addition , many thanks to gary ferland and his collaborators for developing cloudy , making it freely available , and supporting it .
the referee s comments were very helpful , and those comments helped improve the paper .
thanks also to david ballantyne , pat hall , lewis hobbs , john kartje , ruben krasnopolsky , bob rosner , nektarios vlahakis , and don york for valuable comments .
this work would not have been possible without the generous support of nasa s atp program , in this case via grant nag5 - 9063 , and the support of the natural sciences and engineering research council of canada .
this research has made use of nasa s astrophysics data system .
in this appendix , a rederivation of the system of self - similar wind equations for the magnetocentrifugal wind is presented .
the equations utilized in this calculation advance upon those presented in bp82 and kk94 : the wind not only has an arbitrary density power - law index , @xmath8 , as in kk94 , but energy conservation is not required .
since the radiation field continually inputs energy into the outflow , this is an important modification that was not fully considered in the derivation presented in kk94 .
first , a stationary , axisymmetric , ideal , cold mhd flow in cylindrical coordinates @xmath145 is assumed .
the equations are based on both the radial and vertical momentum equations : @xmath146 where @xmath147 is the fluid velocity and @xmath148 is the magnetic field .
the thermal term is neglected in the limit that thermal affects are much less important than magnetocentrifugal and radiative - driving effects .
@xmath149 is the effective gravitational potential , defined as @xmath150 \frac{gm_{\bullet}}{(r^2 + z^2)^{1/2}}\label{gammainselfsim } , \end{aligned}\ ] ] where @xmath10 is the mass of the central black hole , and @xmath29 gives the local radiative - to - gravitational radial acceleration ( see eq . [ [ gammadef ] ] ) .
these equations are solved by first stipulating mass conservation , @xmath151 and then relating the flow velocity to the magnetic field via @xmath152 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , where @xmath153 is the ratio of mass flux to magnetic flux , and @xmath154 and @xmath155 are the field angular velocity and gas mass density of the flow , respectively .
both @xmath156 and @xmath157 are constant along magnetic fieldlines . in addition , while the specific energy is not constant , the total specific angular momentum @xmath158 is conserved .
self - similarity is then imposed on this system by specifying @xmath159 , \\ \mathbf{v } & = & [ \xi'(\chi ) f(\chi ) , g(\chi ) , f(\chi)]v_{k,0},\end{aligned}\ ] ] where @xmath110 is the keplerian speed at the base of the outflow , @xmath160 , and the prime indicates differentiation with respect to @xmath161 . at the same time , the above constants are re - expressed in dimensionless form : @xmath162 where @xmath163 is the poloidal magnetic field strength at the base of the wind . with this self - similar specification ,
the radial and vertical momentum equations become , after some simplification : @xmath165 s^3 - \frac{f}{\kappa \xi j^2 } \left ( \frac{-(1+\xi'^2)(b+1)}{2 } + \right .
\nonumber \\ & & \left .
\frac{(\chi + \xi \xi')\xi'}{\xi } - \frac{\xi''}{j s^2 } \right ) - \frac{\kappa f}{\xi } \frac{(\lambda - \xi^2)}{(m-1 ) } \nonumber \\ & & \left [ \frac{(\lambda - \xi^2)}{(m-1)}\frac{(-b+1)}{2 } + \right .
\nonumber \\ & & \left . \chi \left
( \frac{2 \xi \xi'}{(m-1 ) } + \frac{(\lambda - \xi^2)m ' } { ( m-1)^2 } \right ) \right ] , \label{radmomeq } \\
\frac{f}{\kappa \xi j } ( m ' - f \kappa \xi ' j + f \kappa \xi \chi \xi '' ) & = & -[1 - \gamma(\theta)]\chi s^3 + \frac{f \xi'}{\kappa \xi j^2 } \left ( \frac{-(1+\xi'^2)(b+1)}{2 } + \nonumber \right .
\\ & & \left
. \frac{(\chi + \xi \xi')\xi'}{\xi } - \frac{\xi''(\chi^2 + \xi^2)}{j } \right ) - \nonumber \\ & & \xi ' \kappa f ( \lambda - \xi^2 ) \left(\frac{(b+1)(\lambda-\xi^2 ) - 2(\lambda+\xi^2)}{2 \xi ( m-1)^2 } \right ) + \nonumber \\ & & \frac{(\lambda - \xi^2)^2 m ' \kappa f}{(m-1)^3 } , \label{vertmomeq}\end{aligned}\ ] ] where @xmath166 the two equations ( [ radmomeq ] ) and ( [ vertmomeq ] ) define the differential equations for @xmath167 and @xmath168 , which are , respectively , the spatial gradient in the poloidal alfvn mach number ( gradient with respect to height , @xmath161 ) and the ( cylindrical ) radial velocity gradient ( again with respect to @xmath161 ) .
one can see from close inspection of the above equations that many of the terms have a denominator of @xmath169 , showing that when the gas crosses the alfvn point ( where @xmath170 ) , the equations become singular . rewriting and solving the @xmath167 equation for the value of @xmath167 at the alfvn singular point : @xmath171 / \nonumber \\ & & \left [ 4 \xi j \left ( \frac{4 \kappa^2 \lambda \xi'^2 j^2}{s^2 } + m'^2 ( \chi + \xi \xi')^2 \right ) \right ] .
\label{mpaeqn}\end{aligned}\ ] ] this constraint is used to start the integral at the alfvn point with the value of @xmath172 given by equation ( [ mpaeqn ] ) .
as covered in the main text , these equations are solved using a `` shooting algorithm , '' integrating from both the alfvn point and the disk surface towards an intermediate point . matching the integrals of three first - order equations ( given by the first - order equation for @xmath167 and the second - order equation for @xmath168 in eqs .
[ radmomeq ] and [ vertmomeq ] ) at the common point allows us to solve for the three free parameters in the system : @xmath14 , @xmath15 , and @xmath13 .
abbott , d.c .
1980 , apj , 242 , 1183 antonucci , r. 1993 , ara&a , 31 , 473 arav , n. , li , z. , & begelman , m.c .
1994 , apj , 432 , 62 arav , n. , korista , k.t . ,
barlow , t.a . , begelman , m.c .
1995 , nature , 376 , 576 arav .
n. 1996 , apj , 465 , 617 arav , n. , barlow , t.a . , laor , a. , sargent , w.l.w , & blandford , r.d .
1998 , mnras , 297 , 990 arnaud , m. & raymond , j. 1992 , apj , 398 , 394 baldwin , j. , ferland , g.j . , martin , p.g . ,
corbin , m. , cota , s. , peterson , b.m . , & slettebak , a. 1991 , apj , 374 , 580 baldwin , j. , ferland , g.j . , korista , k. , verner , d. 1995 , apj , 455 , 119 balsara , d. & krolik , j.h .
1993 , apj , 402 , 109 blandford , r.d . , & knigl , a. 1979 , astrophys .
lett . , 20 , 15 blandford , r.d . , & payne , d.g .
1982 , mnras , 199 , 883 ( bp82 ) blandford , r.d .
2001 , in asp conf .
224 , probing the physics of active galactic nuclei by multiwavelength monitoring , ed .
b. m. peterson , r. s. polidan & r. w. pogge ( san francisco : asp ) , 499 bottorff , m. , korista , k.t .
, shlosman , i. , & blandford , r.d .
1997 , apj , 479 , 200 bottorff , m. , korista , k.t .
, & shlosman , i. 2000 , apj , 537 , 134 bottorff , m. , & ferland , g. 2001 , apj , 549 , 118 cassinelli , j. p. 1979 , ara&a , 17 , 275 castor , j.i . ,
abbott , d.c . , & klein , r.i .
1976 , apj , 195 , 157 chandrasekhar , s. 1956 , apj , 124 , 232 chelouche , d. , netzer , h. 2001 , mnras , 326 , 916 chelouche , d. , netzer , h. 2003 , mnras , 344 , 223 chelouche , d. , netzer , h. 2003 , mnras , 344 , 233 chiang , j. & blaes , o. 2003 , apj , 586 , 97 contopoulos , j. , & lovelace , r.v.e .
1994 , apj , 429 , 139 crenshaw , d.m . ,
kraemer , s.b .
, boggess , a. , maran , s.p . ,
mushotzky , r.f .
, wu , c. 1999 , apj , 516 , 750 crenshaw , d.m . , kraemer , s.b .
, george , i.m .
2003 , araa , 41 , 117 de kool , m. , & begelman , m.c .
1995 , apj , 455 , 448 deluit , s.j .
2004 , a&a , 415 , 39 draine b.t . , &
lee , h.m .
1984 , apj , 285 , 89 drew , j.e .
, & boksenberg , a. 1984 , mnras , 211 , 813 elvis , m. 2000 , apj , 545 , 63 elvis , m. , risaliti , g. , & zamorani , g. 2002 , apj , 565 , l75 emmering , r.t . ,
blandford , r.d . , & shlosman , i. 1992 , apj , 385 , 460 everett , j.e . ,
knigl , a. , arav , n. 2002 , apj , 569 , 671 fath , e.a .
1909 , lick obs .
, 149 , 71 feldmeier , a. & shlosman , i. 1999 , apj , 526 , 344 ferland , g.j . , korista , k.t . ,
verner , d.a . ,
ferguson , j.w .
, kingdon , j.b .
, & verner , e.m .
1998 , pasp , 110 , 761 ferreira , j. 2003 in star formation and the physics of young stars , eds . j. bouvier and j .- p .
zahn ( les ulis : edp sciences ) , 229 ( astro - ph/0311621 ) friend , d.b . &
macgregor , k.b .
1984 , apj , 282 , 591 fromerth , m.j & melia , f. 2001 , apj , 549 , 205 ganguly , r. , bond , n.a . ,
charlton , j.c . ,
eracleous , m. , brandt , w.n .
, & churchill , c.w .
2001 , apj , 549 , 133 gregori , g. , miniati , f. , ryu , d. , & jones , t.w .
2000 , apj , 543 , 775 hamann , f.w . , barlow , t.a . ,
chaffee , f.c .
, foltz , c.b .
, & weymann , r.j .
2001 , apj , 550 , 142 hartley , l.e . , drew , j.e .
, long , k.s . , knigge , c. , proga , d. 2002 , mnras , 332 , 127 kartje , j.f .
1995 , apj , 452 , 565 kartje , j.f . , knigl , a. , & elitzur , m. 1999 , apj , 513 , 180 krolik , j.h . 1999 , active galactic nuclei : from the central black hole to the galactic environment ( princeton : princeton university press ) krolik , j.h . & kriss , g.a .
2001 , apj , 561 , 684 knigl , a. , & kartje , j.f .
1994 , apj , 434 , 446 ( kk94 ) knigl , a. , & pudritz , r.e .
2000 in protostars & planets iv , ed . v. mannings , a. p. boss , s. s. russell ( tucson : university of arizona press ) , 759 kraemer , s.b . ,
crenshaw , d.m . ,
george , i.m . ,
netzer , h. , turner , t.j . , gabel , j.r .
2002 , apj , 577 , 98 krasnopolsky , r. , li , z .- y.,blandford , r. 1999 , apj , 526 , 631 kuncic , z. , celotti , a. , rees , m.j .
1997 , mnras , 284 , 717 laor , a. , & brandt , w.n .
2002 , apj , 569 , 641 martin , p.g . , &
rouleau , f. 1991 , in extreme ultraviolet astronomy , eds , malina , r.f . , bowyer s. , pergamon press , oxford , 341 mathews , w.g . , & ferland , g.j . 1987 , apj , 323 , 456 ( mf87 ) mathis , j.s . ,
rumpl , w. , & nordsieck , k.h .
1977 , apj , 217 , 425 meier , d.l . ,
edgington , s. , godon , p. , payne , d.g . , lind , k.r .
1997 , nature , 388 , 350 mestel , l. 1961 , mnras , 122 , 473 mihalas , d. & weibel - mihalas , b. 1999 , foundations of radiation hydrodynamics , ( mineola : dover publications , inc . )
murray , n. , & chiang , j. 1997 , apj , 474 , 91 murray , n. , chiang , j. , grossman , s.a . , & voit , g.m .
1995 , apj,451 , 498 ( mcgv95 ) netzer , h. 1990 , in active galactic nuclei , saas - fee advanced course 20 ( berlin : springer - verlag ) , 57 ouyed , r. , pudritz , r.e .
1997 , apj , 482 , 712 ouyed , r. , pudritz , r.e . , stone , j.m .
1997 , nature , 385 , 409 pereyra , n.a .
, owocki , s.p . ,
hillier , d.j .
, turnshek , d.a .
2004 , apj , 608 , 454 peterson , b.m .
1997 , an introduction to active galactic nuclei , ( new york : cambridge university press ) powell , m.j.d .
1970 , in numerical methods for nonlinear algebraic equations , ed , p. rabinowitz ( new york : gordon and breach ) .
proga , d. 1999 , mnras , 304 , 938 proga , d. 2000 , apj , 538 , 684 proga , d. 2003 , apj , 585 , 406 proga , d. , stone , j.m . , drew , j.e .
1998 , mnras , 295 , 595 proga , d. , stone , j.m . , & kallman , t.r .
2000 , apj , 543 , 686 , psk00 proga , d. , & kallman , t.r , 2004 , accepted to apj ( astro - ph/0408293 ) rees , m.j .
1987 , mnras , 228 , 47p reichard , t.a .
, richards , g , t . , hall , p.b . ,
schneider , d.p .
, vanden berk , d , e . , fan , x. , york , d.g .
; knapp , g.r . ; brinkmann , j. 2003 , aj , 126 , 2594 risaliti , g & elvis , m. 2004 , in supermassive black holes in the distant universe , ed . a. j. barger ( boston : kluwer ) romanova , m.m . , ustyugova , g.v . ,
koldoba , a.v . ,
chechetkin , v.m . , lovelace , r.v.e .
1997 , apj , 482 , 708 safier , p. 1993
, apj , 408 , 115 sobolev , v.v .
1958 in theoretical astrophysics , ed .
ambartsumian ( london : pergamon ) spruit , h.c .
1996 in physical processes in binary stars , eds .
wijers , m.b .
davies & c.a .
tout , ( dordrecht : kluwer ) ( astro - ph/9602022 ) steenbruge , k.c . ,
kaastra , j.s .
, de vries , c.p . ,
edelson , r. 2003 , a&a in press ( astro - ph/0302493 ) stevens , i.r , & kallman , t.r .
1990 , apj , 365 , 321 tielens , a.g.g.m . ,
mckee , c.f . ,
seab , c.g . ,
hollenbach , d.j .
1994 , apj , 431 , 321 tran , h.d .
2003 , apj , 583 , 632 urry , c.m . , & padovani , p. 1995 , pasp , 107 , 803 ustyugova , g.v . ,
koldoba , a.v . ,
romanova , m.m . ,
chechetkin , v.m . ,
& lovelace , r.v.e .
1995 , apj , 439 , l39 verner , d.a . , &
ferland , g.j .
1996 , apjs , 103 , 467 verner , d.a . , verner , e.m . , &
ferland , g.j .
1996 , atomic data nucl .
data tables , 64 , 1 vitello , p.a.j , & shlosman , i. 1988 , apj , 327 , 680 vlahakis , n. , tsinganos , k. , sauty , c. , trussoni , e. , 2000 , mnras , 318 , 417 wardle , m. , & knigl , a. 1993 , apj , 410 , 218 weber , e.j . & davis , l. jr .
1967 , apj , 148 , 217 weymann , r.j .
, morris , s.l . ,
foltz , c.b . , hewett , p.c .
1991 , apj , 373 , 23 | detailed photoionization and radiative acceleration of self - similar magnetocentrifugal accretion disk winds are explored .
first , a general - purpose hybrid magnetocentrifugal and radiatively - driven wind model is defined .
solutions are then examined to determine how radiative acceleration modifies magnetocentrifugal winds and how those winds can influence radiative driving in active galactic nuclei ( agns ) . for the models studied here ,
both radiative acceleration by bound - free ( `` continuum - driving '' ) and bound - bound ( `` line - driving '' ) processes are found to be important , although magnetic driving dominates the mass outflow rate for the eddington ratios studied ( @xmath0 ) .
the solutions show that shielding by a magnetocentrifugal wind can increase the efficiency of a radiatively - driven wind , and also that , within a magnetocentrifugal wind , radiative acceleration is sensitive to both the column in the shield , the column of the wind and the initial density at the base of the wind . |
supersymmetry ( susy ) is generally considered to be one of the most promising extensions of the standard model ( sm ) of particle physics .
its attractive features include the cancellation of quadratic divergences in the higgs sector , which implies that the soft susy breaking masses of the ( yet unobserved ) superpartners of the sm particles can not be much greater than the electroweak scale . if susy is indeed responsible for the stabilization of this scale against the planck scale , supersymmetric particles should therefore be discovered either at run ii of the fermilab tevatron @xcite or at the cern lhc @xcite .
in particular , the strongly coupling squarks and gluinos should be copiously produced at hadron colliders and lead to first measurements of their masses and production cross sections @xcite .
precision measurements of masses , mixings , quantum numbers , and couplings must , however , be performed in the clean environment of a future linear @xmath0 collider because of the large hadronic sm background and theoretical scale and parton density uncertainties at the fermilab tevatron and cern lhc .
for example , in @xmath0 annihilation the center - of - mass energy of the collision is exactly known , and threshold energy scans allow for a precise mass determination of pair - produced susy particles .
it will then be possible to establish whether the masses and couplings of the electroweak gauginos and of the gluino are indeed related , as expected .
a global analysis should ultimately lead to a reconstruction of the susy breaking model and its parameters . along these lines , detailed studies
have recently been performed for squarks , sleptons , charginos , and neutralinos @xcite , but not for gluinos , the reason being that gluino pairs are produced at @xmath0 colliders only at the one - loop level , while all other sparticles are produced at tree level . at tree level
, gluinos can be produced in pairs only in association with two quarks @xcite , or they are produced singly in association with a quark and a squark @xcite . both processes result in multi - jet final states , where phase space is limited and gluinos may be hard to isolate . in the minimal supersymmetric standard model ( mssm ) @xcite , the exclusive production of gluino pairs in @xmath0 annihilation
is mediated by @xmath3-channel photons and @xmath2-bosons , which couple to the gluinos via triangular quark and squark loops . in earlier studies of this process only the very low center - of - mass energy region ( @xmath4 gev ) with pure photon exchange and no squark mixing @xcite or
@xmath2-boson decays into light ( @xmath5 ) @xcite and very light ( @xmath6 gev ) @xcite gluinos have been considered .
some authors have presented results only for @xmath7 gev and outdated top quark masses of @xmath8 gev @xcite , while others have neglected the mixing of left- and right - handed squark interaction eigenstates into light and heavy mass eigenstates @xcite , which turns out to control the production cross section to a large extent .
it is the aim of this article to study the potential of high - energy linear @xmath0 colliders for the production of gluino pairs within the mssm .
taking into account realistic beam polarization effects , photon and @xmath2-boson exchange , and current mass exclusion limits , we scan the mssm parameter space for various @xmath0 center - of - mass energies to determine the regions , where gluino production should be visible . furthermore we clarify the theoretical questions of the relative sign between the two contributing triangular feynman diagrams , of the possible presence of an axial vector anomaly , and the conditions for vanishing cross sections three related issues , which have so far been under debate in the literature .
the remainder of this paper is organized as follows : in sec .
[ sec:2 ] , we present our analytical results and compare them with existing results in the literature . various numerical cross - sections for gluino pair production at future high - energy linear @xmath0 colliders are computed and discussed in sec .
[ sec:3 ] , and sec .
[ sec:4 ] contains our conclusions .
our conventions for squark mixing are defined in app .
[ sec : a ] , and a summary of all relevant feynman rules is given in app .
[ sec : b ] .
the scattering process e^-(p_1,_1)e^+(p_2,_2)(k_1)(k_2 ) with incoming electron / positron momenta @xmath9 and helicities @xmath10 and outgoing gluino momenta @xmath11 proceeds through the two feynman diagrams a and b in fig .
[ fig:1 ] with @xmath3-channel photon and @xmath2-boson exchange and triangular quark and squark loops .
higgs boson exchange is not considered due to the negligibly small electron yukawa coupling , but it could well be relevant at muon colliders .
the process occurs only at the one - loop level , since the gluino as the superpartner of the gauge boson of the strong interaction couples neither directly to leptons nor to electroweak gauge bosons .
taking into account chiral squark mixing ( see app .
[ sec : a ] ) and using the feynman rules in app .
[ sec : b ] , we decompose the corresponding scattering amplitude = _
i d_v^ g_^v into the lepton current l_^v = |v(p_2,_2 ) -ie_(v_e^v - a_e^v_5 ) u(p_1,_1 ) , the photon and @xmath2-boson propagators id^_v = -i g^ s - m_v^2+i , v = , z^0 , which depend on the squared center - of - mass energy @xmath12 , and the gluino current g_^v & = & -e |u(k_2 ) _ q _ i i^a_i,1 a ^i , v_i^b_i,2 + i^a_i,2 ^i , v_i^b_i,1 [ eq : gten1 ] + & & + _ i , j i^a_i,1 ^ij , v_q b ^ij , v_i^b_j,2 + i^a_i,2 ^ji , v_q ^ij , v_i^b_j,1 v(k_1 ) with squark mass eigenstates @xmath13 , a^i , v_= ^4-d ( /+/_2+m_q)_(v_q^v - a_q^v_5 ) ( /-/_1+m_q ) ( q^2-m_i^2+i)[(q - k_1)^2-m_q^2+i][(q+k_2)^2-m_q^2+i ] [ eq : loopa ] and b^ij , v_= ^4-d ( /-m_q)(2q - k_1 + k_2 ) _ ( q^2-m_q^2+i)[(q - k_1)^2-m_j^2+i][(q+k_2)^2-m_i^2+i ] [ eq : loopb ] and the quark flavor @xmath14 flowing both ways in the corresponding diagrams a and b , so that @xmath15 and @xmath16 @xcite .
( [ eq : gten1 ] ) can be simplified with the dirac equation and the anti - commutation relations for dirac matrices , and the tensor loop integrals in eqs .
( [ eq : loopa ] ) and ( [ eq : loopb ] ) can be expressed through the standard coefficient functions @xmath17 of the metric tensor @xmath18 and tensors constructed from the outgoing gluino momenta @xmath19 and @xmath20 @xcite .
the gluino current then reduces to g_^v = i e _ s 2 ^ab 2 |u(k_2 ) _ _ 5 v(k_1 ) _ q ( a^v_q + b^v_q ) with a^v_q & = & _ i c^qi_0 ( m^2_q a^-_qiv -m^2 _ a^+_qiv + 2 m_q m _ _ qiv ) + c^qi_1 4 m _ ( m_q _ qiv - m _
a^+_qiv ) + & & + c^qi_00 ( 2-d ) a^+_qiv - c^qi_11 2 m^2 _ a^+_qiv + c^qi_12 s-2 m^2_a^+_qiv , + b^v_q & = & _ i , j c^qij_00 2 b_qijv , where @xmath21 and @xmath22 are massive ( infrared - finite ) three - point functions and @xmath23 and @xmath24 in diagram a. a^_qiv & = & v^v_q ( s^q_i1s^q_i1-s^q_i2s^q_i2 ) a_q^v , + _ qiv & = & a^v_q ( s^q_i1s^q_i2+s^q_i2s^q_i1 ) , and [ eq : abcoup ] + b_qijv & = & s^q _ i1s^q_j1_q^ij , v - s^q_i2s^q _ j2_q^ji , v are combinations of vector ( @xmath25 ) , axial vector ( @xmath26 ) , and derivative couplings ( @xmath27 ) , and elements of the squark mixing matrix @xmath28 .
pairs of identical ( majorana ) gluinos are therefore produced by a parity violating axial vector coupling induced by mass differences between the chiral squarks and the axial vector coupling @xmath29 of the @xmath2-boson .
the ( mass - independent ) ultraviolet singularities contained in the @xmath30-functions cancel among @xmath31 and @xmath32 in dimensions .
as we have checked explicitly ( even for complex squark mixing matrices ) , adding the two amplitudes induces not only a cancellation of the ultraviolet singularities and of the logarithmic dependence on the scale parameter @xmath33 introduced in eqs .
( [ eq : loopa ] ) and ( [ eq : loopb ] ) in order to preserve the mass dimension of the loop integrals , but also a destructive interference of the finite remainders .
this happens separately for each weak isospin partner , as is to be expected for triangular loop diagrams involving one axial vector and two scalar ( not vector ) couplings and no closed fermion loop .
the ( finite ) total cross section for incoming electrons / positrons with helicities @xmath34 is then _ _
1_2 ( s ) = _ e^2_s^2 ( n_c^2 - 1 ) ^3 s 24 _ v_1,v_2 _ q ( a^v_1_q+b^v_1_q ) ( a^v_2_q+b^v_2_q)^ [ eq : polxsec ] with q^v_1v_2__1_2 = ( v_e^v_1v_e^v_2+a_e^v_1a_e^v2 ) ( 1 - 4_1_2)-(v_e^v_1a_e^v2+v_e^v_2a_e^v_1)(2_1 -2_2 ) , color factor @xmath35 , and gluino velocity @xmath36 , which contains the expected factors of @xmath37 and @xmath3 for @xmath38-wave production of two spin-1/2 majorana fermions .
the distribution in the center - of - mass scattering angle @xmath39 , ( s)=38 ( 1+^2)__1_2(s ) , is independent of the gluino mass and has to be integrated over just one hemisphere , since the two final state particles are identical @xcite . as a consequence ,
the forward - backward asymmetry vanishes for majorana fermions , but not for dirac fermions . our result for diagram a agrees with the unpolarized result ( s)=14__1,2=1/2 _ _ 1_2(s ) in eq .
( 4.5 ) of kileng and osland , if we identify @xcite c_0 ^qi & = & - f^00_qqi , + c_1 ^qi & = & + f^01_qqi , + c_00^qi & = & - g_qqi/2 , + c_11^qi & = & - f^02_qqi , + c_12^qi & = & - f^11_qqi , and reverse the sign of @xmath40 to account for opposite conventions of squark mass eigenstates .
however , our result for diagram b disagrees in sign with eq . ( 4.5 ) of kileng and osland , if we identify @xcite c_00^qij = - g_ijq/2 . if the sign of diagram b is reversed , the ultraviolet singularities cancel only after adding the contributions from the two weak isospin partners with opposite values of @xmath41 .
we trace this sign discrepancy to the feynman rules employed in ref.@xcite , which exhibit a relative minus sign to those in ref.@xcite for the @xmath2-boson coupling to quarks , but not to squarks , whereas our feynman rules ( see app . [ sec : b ] ) agree with those in ref .
@xcite in the limit of no squark mixing . except for the relative sign of diagrams
a and b and in the limit of vanishing gluino mass , we also find agreement with djouadi and drees @xcite .
we confirm , however , the relative sign for diagrams a and b of campbell , scott , and sundaresan @xcite , who found ( for non - mixing chiral squarks of different mass ) that the ultraviolet singularities cancel separately for each weak isospin partner , and that there is no anomaly .
this had also been claimed previously by kane and rolnick @xcite for chiral squarks of equal mass . in their limit , the cross section depends only on the weak isospin and not on the charge of the ( s)quarks @xcite , and the contribution of the photon vanishes @xcite . for the contribution of the @xmath2-boson to vanish , we must have @xcite 1 .
mass degeneracy in each quark isospin doublet , @xmath42 etc .
mass degeneracy in each squark isospin doublet , @xmath43 etc . , which contradicts the condition @xmath44 found by kane and rolnick @xcite .
condition ( 1 ) is violated most strongly for the third generation , as is condition ( 2 ) for most susy breaking models .
the analytical results presented in the previous section have been obtained in two independent analytical calculations .
they have been implemented in compact fortran computer codes , which depend on the looptools / ff library @xcite for the evaluation of the massive tensor three - point functions . as a third independent cross - check
, we have recalculated the production of gluino pairs in @xmath0 annihilation with the computer algebra program feynarts / formcalc @xcite and found numerical agreement up to 15 digits .
our calculations involve various masses and couplings of sm particles , for which we use the most up - to - date values from the 2002 review of the particle data group @xcite .
in particular , we evaluate the electromagnetic fine structure constant @xmath45 at the mass of the @xmath2-boson , @xmath46 gev , and calculate the weak mixing angle @xmath47 from the tree - level expression @xmath48 with @xmath49 gev . among the fermion masses ,
only the one of the top quark , @xmath50 gev , plays a significant role due to its large splitting from the bottom quark mass , @xmath51 gev , while the latter and the charm quark mass , @xmath52 gev , could have been neglected like those of the three light quarks and of the electron / positron .
the strong coupling constant is evaluated at the gluino mass scale from the one - loop expression with five active flavors and @xmath53 mev , corresponding to @xmath54 .
a variation of the renormalization scale by a factor of four about the gluino mass results in a cross section uncertainty of about @xmath55 % . like the heavy top quark
, all susy particles have been decoupled from the running of the strong coupling constant .
we work in the framework of the mssm with conserved @xmath56- ( matter- ) parity , which represents the simplest phenomenologically viable model , but which is still sufficiently general to not depend on a specific susy breaking mechanism .
models with broken @xmath56-parity are severely restricted by the non - observation of proton decay , which would violate both baryon and lepton number conservation .
we do not consider light gluino mass windows , on which the literature has focused so far and which may or may not be excluded from searches at fixed target and collider experiments @xcite .
instead , we adopt the current mass limit @xmath57 gev from the cdf @xcite and d0 @xcite searches in the jets with missing energy channel , relevant for non - mixing squark masses of @xmath58 gev and @xmath59 .
values for the ratio of the higgs vacuum expectation values , @xmath60 , below 2.4 are already excluded by the cern lep experiments @xcite . if not stated otherwise , we will present unpolarized cross sections for a @xmath61 gev linear @xmath0 collider like desy tesla , gluino masses of @xmath62 gev , and squark masses @xmath63 gev . we will consider two cases of large squark mass splittings : i. ) on the one hand , the masses of the superpartners of left- and right - handed quarks need not be equal to each other . in this scenario
we will vary the right - handed up - type squark mass parameters @xmath64 between 200 and 1500 gev .
ii . ) on the other hand , the superpartners of the heavy quarks can mix into light and heavy mass eigenstates ( see app .
[ sec : a ] ) .
this alternative is restricted by the cern lep limits on the light top and bottom squark masses , @xmath65 gev and @xmath66 gev @xcite , and on susy one - loop contributions @xcite to the @xmath67-parameter , @xmath68 @xcite . in this case
we assume the maximally allowed top squark mixing with @xmath69 , @xmath70 gev , and @xmath71 gev , which can be generated by choosing appropriate values for the higgs mass parameter , @xmath72 gev , and the trilinear top squark coupling , @xmath73 gev . for small values of @xmath60
, mixing in the bottom squark sector remains small , and we take @xmath74 . although the absolute magnitude of the cross section depends strongly on the gluino mass and collider energy , the relative importance of the different contributions is very similar also for higher gluino masses and collider energies .
first we examine the conditions found in section [ sec:2 ] for vanishing of the photon and @xmath2-boson contributions , restricting ourselves to the third generation .
since we expect the photon contribution to cancel for equal left- and right - handed squark masses , we vary the right - handed top squark mass parameter , @xmath75 , between 200 and 1500 gev , but keep @xmath76 gev fixed ( case
i ) , since top and bottom squarks generally interfere destructively due to their opposite charge and weak isospin quantum numbers . as can be seen from fig.[fig:4 ] , the photon contribution cancels indeed for @xmath77 .
this is due to the fact that for photons @xmath78 and @xmath79 in eq.([eq : abcoup ] ) , while unitarity of the squark mixing matrix leads to @xmath80 and @xmath81 .
therefore , the photon contributions cancels for all flavors @xmath14 with equal squark masses . due to their charge ,
top ( s)quarks contribute four times as much as bottom ( s)quarks , whose contribution is even more suppressed by the condition @xmath82 .
the @xmath2-boson contribution can never cancel , since @xmath83 , and therefore it depends only weakly on @xmath84 , but it can become minimal for @xmath85 . as @xmath84 gets significantly larger ( or smaller ) than @xmath86 , the photon contribution starts to dominate over the @xmath2-boson contribution .
if only @xmath84 differs from @xmath86 , the third generation contributes almost 100% to the total cross section .
however , if @xmath87 are varied simultaneously , all three generations contribute to the total cross section , which can therefore become significantly larger .
this is shown in fig.[fig:5 ] , where ( s)quark loop contributions from all three generations have been taken into account . when @xmath88 and large mass splittings are generated only by mixing in the top squark sector ( case ii ) , photon contributions are suppressed by more than two orders of magnitude .
[ fig:6 ] shows that the @xmath2-boson contributions from top and bottom squarks interfere destructively due to opposite values of their weak isospin quantum numbers , except for @xmath89 , where the imaginary parts of the amplitudes interfere constructively .
it is therefore advantageous to keep the bottom squark mass splitting small .
as is also evident from fig .
[ fig:6 ] , mixing in the bottom squark sector is of little importance .
note that the central region with maximal top / bottom squark mixing is excluded by the cern lep limits on @xmath90 , @xmath91 , and the @xmath67-parameter .
when @xmath86 and the diagonal elements of the squark mixing matrix ( see app.[sec : a ] ) become much larger than the quark masses and the off - diagonal elements of the matrix , the role of squark mixing is expected to be reduced .
this is confirmed numerically in fig .
[ fig:7 ] , where the dependence of the gluino production cross section on @xmath86 is shown for the cases of maximal and vanishing top squark mixing .
squarks from the first two generations contribute at most 10% at low @xmath86 and are otherwise strongly suppressed . at future linear @xmath0 colliders it will be possible to obtain relatively high degrees of polarization , _ i.e. _ about 80% for electrons and 60% for positrons
@xcite . in fig .
[ fig:8 ] we therefore investigate the effect of choosing different electron / positron polarizations on the gluino pair production process , including contributions from all ( s)quarks . since the @xmath92 and @xmath93 helicity amplitudes vanish for both photons and @xmath2-bosons , we only show the squares of the remaining @xmath94 and @xmath95 amplitudes , which coincide for photons , but not for @xmath2-bosons .
the unpolarized cross section falls short of the polarized ones , so that a high degree of polarization is clearly desirable . with the realistic degrees of polarization mentioned above , we show in fig .
[ fig:9 ] a scan in the center - of - mass energy of a future @xmath0 collider for various gluino masses and maximal top squark mixing ( case ii ) .
the cross section rises rather slowly due to the factor @xmath37 in eq.([eq : polxsec ] ) for @xmath38-wave production of the gluino pairs . for @xmath62 gev
we observe an interesting second maximum , which arises from the intermediate squark pair resonance at @xmath96 gev . at threshold ,
the cross section depends strongly on the gluino mass and is largest for @xmath97 gev , which we consider to be the lowest experimentally allowed value .
it drops fast with @xmath98 , so that for @xmath99 gev no events at colliders with luminosities of 1000 fb@xmath100 per year can be expected , irrespective of their energy .
smaller squark mixing ( cf .
[ fig:6 ] ) or larger values of @xmath86 ( cf .
[ fig:7 ] ) will reduce the cross section even further . far above threshold ,
it drops off like @xmath101 and becomes independent of the gluino mass .
the slow rise of the cross section can be observed even better in fig.[fig:11 ] , where the sensitivity of a @xmath102 gev collider like desy tesla to gluino masses around 200 gev has been plotted . for the cern lhc experiments , a precision of @xmath103 ( @xmath104 ) gev
is expected for gluino masses of 540 ( 1004 ) gev @xcite . if the masses and mixing angle(s ) of the top ( and bottom ) squarks
are known , a precision of @xmath105 gev can be achieved at desy tesla for @xmath62 gev and maximal top squark mixing with an integrated luminosity of 100 fb@xmath100 per center - of - mass energy point .
a center - of - mass energy scan for the scenario with no squark mixing , but large left-/right - handed squark splitting ( case i ) is shown in fig.[fig:10 ] for light ( full and dashed curves ) and heavy ( dotted and dot - dashed curves ) gluino masses .
since the photon contributes now significantly to the cross section , it proves to be advantageous to choose the lepton polarization such that the @xmath2-boson interferes constructively with the photon , even though it is by itself slightly smaller than for the opposite choice . since all three generations
add now to the cross section , it can become almost an order of magnitude larger than in the mixing scenario ( case ii ) , and even gluino masses of 1 tev may be observable at a multi - tev collider like cern clic .
however , also here the cross section drops sharply when the squark mass splitting is reduced from 1500 tev ( full and dotted curves ) to values close to @xmath86 ( dashed and dot - dashed curves ) .
the purpose of this paper has been two - fold : first , we have resolved a long - standing discrepancy in the literature about the relative sign of the quark and squark loop contributions to the production of gluino pairs in @xmath0 annihilation .
we confirm the result of two older papers that the divergence cancels for each squark flavor separately and not between weak isospin partners @xcite and trace the sign problem in one case to the feynman rules employed in the corresponding calculation @xcite .
our results rely on two completely independent analytical calculations and one computer algebra calculation .
second , we have investigated the prospects for precision measurements of gluino properties , such as its mass or its majorana fermion nature , at future linear @xmath0 colliders .
we have taken into account realistic beam polarization effects , photon and @xmath2-boson exchange , and current mass exclusion limits .
previously , only light gluinos at center - of - mass energies up to the @xmath2-boson mass had been investigated . within the general framework of the mssm ,
we have concentrated on two scenarios of large left-/right - handed up - type squark mass splitting and large top squark mixing , which produce promisingly large cross sections for gluino masses up to 500 gev or even 1 tev .
gluino masses of 200 gev can then be measured with a precision of about 5 gev in center - of - mass energy scans with luminosities of 100 fb@xmath100/point . however , when both the left-/right - handed squark mass splitting and the squark mixing remain small , gluino pair production in @xmath0 annihilation will be hard to observe , even with luminosities of 1000 fb@xmath100/year .
we thank b. kileng , g. kramer , and p. osland for useful discussions and a. brandenburg and b.a .
kniehl for a careful reading of the manuscript .
this work has been supported by deutsche forschungsgemeinschaft through grant no .
kl 1266/1 - 2 and through graduiertenkolleg _
zuknftige entwicklungen in der teilchenphysik_.
the ( generally complex ) soft susy - breaking terms @xmath106 of the trilinear higgs - squark - squark interaction and the ( also generally complex ) off - diagonal higgs mass parameter @xmath33 in the mssm lagrangian induce mixings of the left- and right - handed squark eigenstates @xmath107 of the electroweak interaction into mass eigenstates @xmath108 .
the squark mass matrix @xcite ^2 = cc m_ll^2+m_q^2 & m_q m_lr^ + m_q m_lr & m_rr^2+m_q^2 with m_ll^2&=&(t_q^3-e_q^2_w)m_z^22+m_^2 , + m_rr^2&=&e_q^2_w m_z^22+\ { l m_^2 , + m_^2 , .
+ m_lr & = & a_q-^\ { l + .
is diagonalized by a unitary matrix @xmath28 , @xmath109 , and has the squared mass eigenvalues m_1,2
^ 2=m_q^2 + 12m_ll^2+m_rr^2 . for real values of @xmath110 , the squark mixing angle @xmath111 , @xmath112 , in s = cc _ & _ + - _ & _ c _ 1 + _ 2 = s c _
l + _ r can be obtained from 2_=2m_qm_lrm_ll^2-m_rr^2 .
if @xmath110 is complex , one may first choose a suitable phase rotation @xmath113 to make the mass matrix real and then diagonalize it for @xmath114 and @xmath115 .
@xmath116 is the ( real ) ratio of the vacuum expectation values of the two higgs fields , which couple to the up- and down - type ( s)quarks . the weak isospin quantum numbers for left - handed up- and down - type ( s)quarks with hypercharge @xmath117 are @xmath118 , whereas @xmath119 and @xmath120 for right - handed ( s)quarks , and their fractional electromagnetic charges are @xmath121 . the soft susy - breaking mass terms for left- and right - handed squarks are @xmath122 and @xmath123 , @xmath124 , respectively , and @xmath125 is the mass of the neutral electroweak gauge boson @xmath2 .
denoting squark mass eigenstates by @xmath126 , lorentz indices by @xmath127 , and color indices of the fundamental ( adjoint ) representation of the color symmetry group su(3 ) by @xmath128 @xmath129 , we obtain the following propagators in feynman - gauge : & & -i g^ p^2-m_v^2+i , m_v^2 = \{0,m_z^2 } + & & i ( p/+m_q ) _ lm p^2-m_q^2+i + & & i _
p^2-m_^2+i dirac fermions carry an arrow , which indicates the fermion number flow , whereas majorana fermions , such as gluinos , do not carry arrows .
an additional arrow is depicted next to all fermion lines in order to obtain a unique orientation of the fermion flow , which is evaluated according to the rules in @xcite .
the interaction vertices are given by @xcite & & -i e _
( v_f^v - a_f^v _ 5 ) _ lm \ { .
i,1^a = -i _ s t^a_lm ( s_i1 p_l - s_i2 p_r ) + & & i _ i,2^a = -i _ s t^a_ml ( s_i1^p_r - s_i2^p_l ) the feynman rules in eq . ( [ eq : vert_yzqq ] ) apply to photon and @xmath2-boson interactions with quarks and charged leptons .
the latter carry electromagnetic charge @xmath130 and weak isospin @xmath131 ( left - handed ) and 0 ( right - handed ) , but no color ( @xmath132 ) .
the gauge couplings @xmath133 and @xmath134 of the weak isospin and hypercharge symmetries su(2)@xmath135 and u(1)@xmath136 have been expressed in terms of the electromagnetic coupling @xmath137 and the sine and cosine of the weak mixing angle @xmath138 .
the gluino - quark - squark vertices depend on the generators of the su(3 ) color symmetry group , @xmath139 , and on the yukawa coupling @xmath140 , which is identical to the strong gauge coupling @xmath141 in leading order , and on the squark mixing matrix @xmath28 , but not on the orientation of the fermion flow .
m. carena _ et al .
_ , arxiv : hep - ph/0010338 .
s. abel _ et al .
_ , arxiv : hep - ph/0003154 .
r. culbertson _ et al .
_ , arxiv : hep - ph/0008070 . s. ambrosanio _ et al .
_ , arxiv : hep - ph/0006162 . b. allanach _ et al .
_ , arxiv : hep - ph/9906224 .
a. airapetian _ et al . _ [ atlas collaboration ] , cern - lhcc-99 - 15 .
s. abdullin _ et al . _
[ cms collaboration ] , j. phys .
g * 28 * , 469 ( 2002 ) .
w. beenakker , r. hpker , m. spira and p. m. zerwas , nucl . phys .
b * 492 * , 51 ( 1997 ) .
j. a. aguilar - saavedra _ et al .
_ , arxiv : hep - ph/0106315 . b. a. campbell , j. r. ellis and s. rudaz , nucl .
b * 198 * , 1 ( 1982 ) .
d. h. schiller and d. whner , nucl .
b * 259 * , 597 ( 1985 ) .
a. brandenburg , m. maniatis and m. m. weber , arxiv : hep - ph/0207278 .
h. p. nilles , phys .
rept . * 110 * , 1 ( 1984 ) .
h. e. haber and g. l. kane , phys .
rept . * 117 * , 75 ( 1985 ) .
p. nelson and p. osland , phys .
b * 115 * , 407 ( 1982 ) .
g. l. kane and w. b. rolnick , nucl .
b * 217 * , 117 ( 1983 ) .
b. a. campbell , j. a. scott and m. k. sundaresan , phys .
b * 126 * , 376 ( 1983 ) .
b. kileng and p. osland , z. phys .
c * 66 * , 503 ( 1995 ) .
a. djouadi and m. drees , phys . rev .
d * 51 * , 4997 ( 1995 ) .
a. denner , fortsch .
phys . * 41 * , 307 ( 1993 ) .
g. passarino and m. j. veltman , nucl .
b * 160 * , 151 ( 1979 ) .
s. t. petcov , phys .
b * 139 * , 421 ( 1984 ) .
b. kileng and p. osland , private communication .
t. hahn and m. perez - victoria , comput .
commun .
* 118 * , 153 ( 1999 ) .
g. j. van oldenborgh , comput .
commun . *
66 * , 1 ( 1991 ) . t. hahn and c. schappacher , comput .
commun . * 143 * , 54 ( 2002 ) .
k. hagiwara _ et al .
_ [ particle data group collaboration ] , phys . rev .
d * 66 * , 010001 ( 2002 ) .
t. affolder _ et al . _
[ cdf collaboration ] , phys .
lett . * 88 * , 041801 ( 2002 ) .
s. abachi _ et al . _
[ d0 collaboration ] , phys .
lett . *
75 * , 618 ( 1995 ) .
m. felcini , c. tully _ et al . _
[ lep higgs working group collaboration ] , note 2001 - 04 , arxiv : hep - ex/0107030
. j. abdallah _ et al . _
[ lep susy working group collaboration ] , http://lepsusy.web . cern.ch/lepsusy .
r. barbieri and l. maiani , nucl .
b * 224 * , 32 ( 1983 ) .
m. drees and k. hagiwara , phys .
d * 42 * , 1709 ( 1990 ) .
h. chankowski , a. dabelstein , w. hollik , w. m. mosle , s. pokorski and j. rosiek , nucl .
b * 417 * , 101 ( 1994 ) .
j. f. gunion and h. e. haber , nucl .
b * 272 * , 1 ( 1986 ) [ erratum - ibid .
b * 402 * , 567 ( 1993 ) ] .
a. denner , h. eck , o. hahn and j. kblbeck , nucl .
b * 387 * , 467 ( 1992 ) . | we study the potential of high - energy linear @xmath0 colliders for the production of gluino pairs within the minimal supersymmetric standard model ( mssm ) . in this model , the process @xmath1 is mediated by quark / squark loops , dominantly of the third generation , where the mixing of left- and right - handed states can become large .
taking into account realistic beam polarization effects , photon and @xmath2-boson exchange , and current mass exclusion limits , we scan the mssm parameter space for various @xmath0 center - of - mass energies to determine the regions , where gluino production should be visible . |
the phenomenon of triaxiality has remained of central importance to our understanding of galaxy dynamics since the demonstration by schwarzschild ( 1979 ) of the existence of self - consistent triaxial equilibria .
schwarzschild s models had large , constant - density cores and the majority of the orbits were regular , respecting three integrals of the motion .
but the realization that the central densities of elliptical galaxies and bulges are very high ( @xcite ) , and that supermassive black holes are generic components of galaxies ( @xcite ) , has modified somewhat our ideas about the persistence of triaxiality . a central black hole can destroy triaxiality on large scales by rendering the center - filling box orbits stochastic ( @xcite ) . evolution to globally axisymmetric shapes can occur in just a few crossing times if the black hole contains of order @xmath5 of the galaxy s mass ( @xcite ; @xcite ) .
less is known about the possibility of maintaining triaxial shapes at the very centers of galaxies , where the gravitational force is contributed roughly equally by the stars and by the nuclear black hole .
regular orbits , both box - like and tube - like , exist in this region ( @xcite ; @xcite ; @xcite , paper i ) but the box - like orbits ( the `` pyramids '' ) have shapes that are generally contrary to the figure elongation , making them less useful than classical box orbits for maintaining a triaxial shape .
furthermore the pyramid orbits disappear within the `` zone of chaos , '' which begins at a radius where the enclosed stellar mass is a few times that of the black hole ( paper i ) , roughly @xmath6 pc in a typical bright elliptical galaxy . here
we demonstrate that triaxial equilibria do in fact exist in the vicinity of the black hole .
we first ( 2 ) use schwarzschild s technique to construct self - consistent superpositions of orbits computed in a fixed triaxial potential representing the stars and a central point mass .
we then ( 3 ) test the long - term stability of the models by using an @xmath4-body code to advance them forward in time ; we find that the @xmath4-body models maintain their triaxial shapes for many crossing times .
finally we present some of the observable properties of the models ( 4 ) and discuss the implications of triaxiality for observational and theoretical studies of galactic nuclei ( 5 ) .
we model the stellar nucleus as a triaxial spheroid with a power - law dependence of density on radius , @xmath7 inside of some surface @xmath8 discussed below .
we took @xmath9 or @xmath10 ; these values correspond roughly to the density profiles observed at the centers of bright and faint elliptical galaxies , respectively ( @xcite ) .
the triaxiality @xmath11 is defined as @xmath12 we chose @xmath13 for both models , corresponding to @xmath14 , `` maximal '' triaxiality .
the central black hole is represented by a point mass with @xmath15 .
expressions for the gravitational potential and forces corresponding to this mass model are given in paper i. although our mass model for the stars is scale - free , the presence of the black hole imposes a scale .
we identify three characteristic radii associated with the black hole . at a radius of @xmath16 , the enclosed mass in stars ( defined as the mass within an equidensity surface which intersects the @xmath17-axis at @xmath18 ) is equal to that of the black hole . in real galaxies , @xmath16 is typically a factor of @xmath19 a few greater than the `` sphere of influence '' @xmath20 , where @xmath21 depends on the stellar velocity dispersion @xmath22 the third fiducial radius is @xmath23 , the radius at which the character of the box - like orbits undergoes a sudden transition to chaos in triaxial potentials .
table 1 gives values of @xmath16 and @xmath23 for our two mass models ; values of @xmath23 are approximate and are taken from paper i. table 1 also gives values of the dynamical times at @xmath16 and @xmath23 , defined as the period of a circular orbit of the same energy in the equivalent spherical potential , defined to have a scale length @xmath24 .
the enclosed stellar mass within @xmath23 is about @xmath25 for @xmath9 and @xmath26 for @xmath27 . [
cols="^,^,^",options="header " , ] the orbital content of the latter solutions is described in figure 1 and table 2 .
we constructed two types of solution for each @xmath3 : solutions in which only regular orbits were used ; and solutions in which all orbits ( regular and chaotic ) were provided to the schwarzschild algorithm .
the dominant family of orbits in the models containing only regular trajectories was found to be the @xmath28-tubes , orbits which circulate about the short axis of the figure .
plots of the individual orbits used in the solutions confirm that many of the @xmath28-tubes have the correct shape for reproducing the triaxial figure , being more elongated in @xmath17 ( the long axis ) than in @xmath29 ( the intermediate axis ) .
most of the remaining contribution to the density of the regular nuclei was found to come from pyramid orbits and high - energy box orbits in the @xmath9 model , while for @xmath27 , roughly equal contributions came from pyramid orbits and from the @xmath17-tubes .
( see paper i , figure 1 for illustrations of the various orbit families . )
the most heavily occupied pyramid orbits had shapes similar to `` saucer '' orbits , @xmath28-tubes associated with a @xmath30 resonance ( paper i , figure 1c ) ; both models contained roughly equal numbers of saucers and pyramids .
high - energy box - like orbits are more important in the @xmath9 case than in the @xmath27 case , because : ( 1 ) high - energy tube orbits in the @xmath9 model are elongated opposite to the figure ; ( 2 ) most of the high - energy regular box orbits in the @xmath27 model are bananas , which are not very useful for filling the outermost shells . in the models
constructed using both regular and chaotic trajectories , the number of stars on chaotic orbits is surprisingly high : @xmath31 for @xmath9 and @xmath32 for @xmath27 at all radii .
plots of the most highly populated chaotic orbits suggest that they have reached a nearly steady state in a time - averaged sense , filling the volume defined by the equipotential surface corresponding to their energy .
stationary chaotic building blocks like these have been used before when constructing triaxial models ( e.g. @xcite ) but no published triaxial models have contained nearly so large fraction of stars on chaotic orbits .
we verified that our orbital superpositions correspond to true equilibria by realizing them as @xmath4-body models and integrating them forward in time in the gravitational potential computed from the @xmath4 bodies themselves and from the point mass representing the black hole .
initial conditions were prepared by re - integrating all orbits with nonzero occupation numbers .
the position and velocity of each trajectory were recorded at fixed time intervals ; the length of each integration was determined by the orbit s occupation number .
the sense of rotation of the tube orbits was chosen randomly such that the mean motion was everywhere zero .
the number of particles was @xmath33 for @xmath9 and @xmath34 for @xmath27 ; a smaller @xmath4 was chosen for @xmath27 since the @xmath4-body integrations were slower for the more condensed model .
the initial conditions were then advanced in time using the @xmath4-body code gadget ( @xcite ) , a parallel tree code with variable time steps .
the interparticle softening length was chosen to be @xmath35 ( @xmath36 ) for @xmath9 ( @xmath10 ) .
figure 2 shows the evolution of the axis ratios of the models , computed using the iterative procedure described by dubinski & carlberg ( 1991 ) .
the axis ratios fluctuate with time , but the models retain their triaxial figures very well over many dynamical times .
no signficant evolution was seen in the radial distribution of matter .
two observational signatures of triaxiality are isophote twists , and kinematic misalignments .
isophote twists require a variation with radius of the intrinsic axis ratios , which is not a feature of our models .
kinematic misalignments occur whenever the intrinsic angular momentum vector does not coincide with the minor axis of the figure , and are generic features of triaxial systems when both the short- and long - axis tube orbits are populated ( e.g. @xcite ) .
we illustrate the possibility of kinematic misalignments in our models in figure 3 . line - of - sight velocity contours are plotted for regular models after a fraction of the @xmath17- and @xmath28-tubes
were reversed , giving the models net rotation about both the long and short axes . for @xmath9 ,
all x - tubes in figure 3 rotate in the same sense while 60% of the @xmath28-tubes at each energy rotate in one sense and the remaining @xmath37 in the other . for @xmath27 ,
all @xmath17-tubes rotate in the same sense around the @xmath17 axis and all @xmath28-tubes rotate in the same sense around the @xmath28-axis .
rotation of the projected models along the apparent minor axis of the figure ( `` minor - axis rotation '' ) is apparent .
we have shown that long - lived triaxial configurations are possible for nuclei containing black holes , whether constructed purely from regular orbits , or from a combination of regular and chaotic orbits .
the latter solutions are remarkable given the large fraction of the mass on chaotic orbits , of order 50% or more ( table 2 ) .
such solutions violate jeans s theorem in its standard form ( e.g. @xcite ) but are consistent with a generalized jeans s theorem ( @xcite ) if we assume that the chaotic building blocks are `` fully mixed , '' that is , that they approximate a uniform population of the accessible phase space .
this appears to be the case based on visual inspection of the time - averaged chaotic trajectories .
while a sudden onset of chaos can effectively destroy triaxiality in models containing a large population of regular box orbits ( @xcite ; @xcite ) , our work shows that at least the central parts of galaxies containing black holes can remain triaxial even when dominated by chaotic orbits .
galaxy nuclei are typically modelled as oblate axisymmetric systems when deriving black hole masses from stellar - kinematical data ( e.g. @xcite ; @xcite ) .
modelling a triaxial nucleus as axisymmetric will generally lead to biassed estimates of @xmath38 ; for instance , if an elongated bar is viewed end - on , the velocity field can mimic that produced by a black hole even in the absence of a central mass concentration ( @xcite ) .
the high - resolution data on which such mass estimates are based are usually obtained from single slits , making it difficult to rule out triaxial shapes using information like that in figure 3 .
axisymmetry is typically also assumed when deriving rates of interaction of stars with nuclear black holes ( e.g. @xcite ; @xcite ) .
the triaxial models which we constructed from regular orbits were dominated by @xmath28-tubes ( table 2 ) , similar to the tube orbits that populate an oblate spheroid . the rate of capture or destruction of stars by the black hole in these models would be similar to the rates in an axisymmetric nucleus . however the dominance of chaotic orbits in our second set of solutions implies a qualitatively different sort of behavior : the `` loss cone '' of stars captured by the black hole would be refilled on a time scale determined by the frequency of near - center passages of the chaotic orbits , rather than by two - body scattering .
binney , j. & tremaine , s. 1987 , galactic dynamics ( princeton : princeton university press ) crane , p. et al .
1993 , aj , 106 , 1371 dejonghe , h. 1989 , apj , 343 , 113 dubinski , j. & carlberg , r. 1991 , apj , 378 , 496 gebhardt , k. et al .
1996 , aj , 112 , 105 gerhard , o. e. 1988 , mnras , 232 , 13p gerhard , o. & binney , j. j. 1985 , mnras , 216 , 467 joseph , c. l. et al .
2001 , apj , 550 , 668 kormendy , j. & richstone , d. o. 1995 , araa , 33 , 581 magorrian , j. & tremaine , s. 1999 , mnras , 309 , 447 franx , m. , illingworth , g. & de zeeuw , p. t. 1991 , apj , 383 , 112 merritt , d. 1999 , pasp , 111 , 129 merritt , d. & fridman , t. 1996 , apj , 460 , 136 merritt , d. & quinlan , g. d. 1998 , apj , 498 , 625 merritt , d. & valluri , m. 1999 , aj , 118 , 1177 poon , m.y . & merritt , d. 2001 , apj , 549 , 192 ( paper i ) sambhus , n. & sridhar , s. 2000 , apj , 542 , 143 schwarzchild , m. 1979 , apj , 232 , 236 schwarzschild , m. 1993 , apj , 409 , 563 sellwood , j. a. 2001 , astro - ph/0107353 springel , v. , yoshida , n. & white , s. d. m. 2001 , newa , 6 , 79 syer , d. & ulmer , a. 1999 , mnras , 306 , 35 van der marel , r. p. , cretton , n. , de zeeuw , p. t , rix , h .- w .
1998 , apj , 493 , 613 | we demonstrate that the nuclei of galaxies containing supermassive black holes can be triaxial in shape .
schwarzschild s method was first used to construct self - consistent orbital superpositions representing nuclei with axis ratios of @xmath0 and containing a central point mass representing a black hole .
two different density laws were considered , @xmath1 , @xmath2 .
we constructed two solutions for each @xmath3 : one containing only regular orbits and the other containing both regular and chaotic orbits .
monte - carlo realizations of the models were then advanced in time using an @xmath4-body code to verify their stability .
all four models were found to retain their triaxial shapes for many crossing times .
the possibility that galactic nuclei may be triaxial complicates the interpretation of stellar - kinematical data from the centers of galaxies and may alter the inferred interaction rates between stars and supermassive black holes . #
1#2#3#4#50.03#5pt#4pt # 1#23.6pt keywords : galaxies : elliptical and lenticular galaxies : structure galaxies : nuclei stellar dynamics |
the star @xmath0 scorpii ( hd 149438 ) is the mk standard for the b0 v spectral type , although walborn ( 1995 , private communication ) has classified it as b0.2 v on the basis of its stellar wind lines .
the star has served as a benchmark in studies of radiatively driven winds of hot stars .
this is because it lies in a pivotal region of the hr diagram where stellar winds are making a transition from the massive and fast winds of the o stars to the winds of b stars that can just marginally be driven by line forces . in the lamers & rogerson ( 1978 ) study of _ copernicus
_ satellite spectra of early - type stars , @xmath0 sco was the latest spectral type to show the superionization stage of o@xmath1vi .
stars with spectral types close to that of @xmath0 sco are also the latest to show the x - ray to bolometric luminosity relation @xmath12 that holds throughout the main sequence o spectral region . for stars of spectral type b1.5 v and later , this ratio of luminosities decreases by more than an order of magnitude , ( cassinelli et al .
it is generally thought that the x - rays from hot stars arise from shocks embedded in the stellar winds ( owocki , castor , & rybicki 1988 , hereafter ocr ; cooper 1994 ) , so the decrease in the x - ray luminosity may be an additional consequence of the slower and less massive winds of near main sequence b stars .
massa , savage , & cassinelli ( 1984 ) found from a study of the uv lines that the wind properties of stars of spectral types b0 v to b2 v are particularly sensitive to the stellar composition , with winds ranging from non - detectable to quite strong for stars with increasing metal abundances .
the stars near this spectral range are also commonly seen as emission line be stars , for which the stellar rotation may produce wind compressed equatorial disks as suggested by bjorkman & cassinelli ( 1993 ) and owocki , cranmer , & blondin ( 1994 ) or rotationally induced bistability as suggested by lamers & pauldrach ( 1991 ) and lamers et al .
( 1999 ) .
while @xmath0 sco is a standard star , its stellar wind is anything but typical .
the two most unusual properties of its wind , and those that we will focus on , are the peculiar red - shifted absorption seen in wind lines of high ionization species such as o@xmath1vi , which are not present in the photosphere of this star , and its excessively - hard x - ray spectrum .
the first of the peculiar wind properties of @xmath0 sco was discovered by lamers & rogerson ( 1978 ) in their analysis of _ copernicus _ observations .
they noted that the absorption produced by o@xmath1vi and n@xmath1v extended _ redward _ of line center by as much as + 250 km s@xmath3 . for stars with massive winds
, the strong uv resonance lines have p cygni - type profiles , which are noted for the absorption shortward of line center owing to scattering in the expanding wind .
the absorption longward of the line center in @xmath0 sco is unusual .
absorption redward of line center could occur if the observed resonance line is also present in the photosphere of the star
. however , the superionization stages of o@xmath1vi , n@xmath1v , and should not be present in the photosphere of @xmath0 sco , which has an effective temperature of @xmath13 k. two ways to explain the redshifted component are mentioned by lamers and rogerson ( 1978 ) : turbulence in the flow near the base of the wind , and infall of highly - ionized material .
both phenomena could produce a positive velocity component in these ionic lines .
the second peculiarity of @xmath0 sco is its very hard x - ray spectrum and large x - ray luminosity .
in their _ rosat _ survey of near - main sequence b stars , cassinelli et al .
( 1994 ) found that @xmath0 sco has a high x - ray to bolometric luminosity ratio : @xmath14 ; this is an order of magnitude larger than most o and early b stars .
furthermore , they found that the x - ray spectrum has an anomalously luminous hard component above 1 kev .
the hard x - rays from this wind are unusual in that they require plasma temperatures in excess of @xmath4 k. cohen , cassinelli , & waldron ( 1997 ; hereafter ccw ) have more recently presented observations of @xmath0 sco obtained with the _ asca _ satellite .
these observations confirm the earlier _ rosat_results requiring a very high temperature component to explain the observed x - ray spectrum .
the _ asca _
observations were fit with a multi - temperature equilibrium plasma model characterized by temperatures of @xmath15 k with emission measures of @xmath16 @xmath17 , respectively . the very high temperature gas inferred from the x - ray spectrum of @xmath0 sco is not explained in the standard picture of x - ray production in early type stars ( see discussion in ccw ) .
numerical simulations of radiative shocks formed in the unstable stellar winds of ob stars suggest that these shocks are unable to produce temperatures much greater than a few times 10@xmath18 k ( ocr ; cooper 1994 ) .
the shock - jump velocities in these simulations are limited to roughly half the local stellar wind velocity .
the presence of very hot gas in the wind of @xmath0 sco suggests that a different mechanism is responsible for at least some of the x - ray emission from this star .
though several groups have attempted to model the radiatively - driven outflow from @xmath0 sco , they were unable to reproduce the observed hardness of the x - ray spectrum ( e.g. , cooper 1994 ; macfarlane & cassinelli 1989 ; lucy & white 1980 ) . in this paper
we explore a phenomenological model to explain both the hard x - ray emission and the redshifted absorption of high ionization lines .
our model supposes density enhancements , or clumps , form in the smooth stellar wind and subsequently fall back towards the stellar surface , creating bow shocks at their leading edges .
we investigate the trajectories of such clumps in a stellar wind and find infall to be achievable assuming the optical depths of these dense parcels of material are large enough to choke off the radiative line driving force . using our calculated trajectories , we estimate the temperature of the shocked gas swept up by the clumps and comment on the ability of this model to provide the high - temperature gas implied by the observed x - ray emission ( ccw ; cassinelli et al .
while the numerical details presented here apply to the specific case of @xmath0 sco , the model developed in this paper is applicable to most clumped , radiatively - driven outflows .
the details and assumptions of our model are discussed in [
sec : model ] . in
[ sec : infall ] we derive the equation of motion for solitary clumps created within the wind and present the results of numerical integrations of clump trajectories .
we describe the temperature spectrum of shocked plasmas in our model and discuss the observed x - ray emission in the context of this model in [ sec : xrays ] . in [ sec : conclusions ] we briefly discuss why @xmath0 sco may be unusual among early - type stars , and we summarize our work in [ sec : summary ] .
table [ table : stardata ] lists the stellar parameters adopted throughout this work for @xmath0 sco .
empirical determinations of the mass loss rate of @xmath0 sco from uv lines gives discrepant results of @xmath19 ( lamers & rogerson 1978 ) , @xmath20 ( gathier et al .
1981 ) and @xmath21 m@xmath10 yr@xmath3 ( hamann 1981 ) .
these differences are due to differences in the adopted ionization fractions of the observed uv resonance lines of , , , n@xmath1v , o@xmath1vi , , and .
the terminal velocity of the wind of @xmath0 sco can not be derived from the observations , because the uv resonance lines are unsaturated .
lamers and rogerson ( 1978 ) derived a lower limit of 2000 km s@xmath3 . because of these uncertainties we adopt the values predicted by the modified line driven wind theory ( kudritzki et al .
1989 ) with the force multiplier parameters taken from abbott ( 1982 ) : @xmath22=0.156 , @xmath23=0.609 and @xmath24=0.12 .
this gives a mass loss rate of @xmath25 m@xmath10 yr@xmath3 and @xmath26 = 2400 km s@xmath3 , in agreement with the range of empirically derived parameters .
the winds of early type stars are driven by resonance line scattering of radiation in the stellar envelope .
this radiative driving mechanism is inherently unstable ( lucy and solomon 1970 ; lucy 1982 ; ocr ) .
small perturbations on an otherwise monotonic wind velocity structure tend to grow and steepen into shocks .
these instabilities have been well studied using one - dimensional hydrodynamical models ( see , e.g. , cooper 1994 , feldmeier 1995 ) , which show that the resulting wind is a mixture of rarified hot material and colder , dense regions .
these studies also suggest the shock velocity jumps have maximal amplitudes of order one half the local wind velocity .
the relationship between the velocity amplitude of a shock , @xmath27 , and the temperature behind the shock , @xmath28 , is given by @xmath29 in the case of most main sequence o stars and ob supergiants , the shock jump velocities are indeed strong enough to produce gas at the temperatures suggested by the observed x - ray emission ( @xmath30 k ; chlebowski , harnden , & sciortino 1989 ; cassinelli et al .
the x - ray spectrum of @xmath0 sco , however , implies gas with temperatures in excess of 10@xmath31 k ( ccw ) .
numerical models of b star winds can not produce shocks with such high temperatures .
velocity jumps of @xmath32 km s@xmath3 are typical in numerical simulations ( cooper 1994 ) , while velocity jumps of more than 700 to 800 km s@xmath3 are required to reproduce the hard x - ray emission from @xmath0 sco .
while the one - dimensional shock production problem has been well studied , little attention has yet been given to the production of shocks and the structure of the wind in two or three dimensions .
we assume that shocks and density enhancements in the stellar wind are relatively localized phenomena whose sizes are dictated by the local sound speed and instabilities that may tend to break up large sheets , such as the rayleigh - taylor instability .
an individual shock is assumed to cover a small solid angle when viewed from the star .
the one - dimensional simulations assume spherical symmetry ; our assumption is that the spherical shocks in these simulations would fragment relatively easily .
cassinelli and swank ( 1983 ) have suggested such a scenario to explain the lack of variability observed in their einstein solid state spectrometer observations of ob supergiants
. a stellar wind containing an ensemble of @xmath33 shocks that continuously form and dissipate can produce a non - detectable variability of order @xmath34 .
the time variability of single spherical shocks in the wind of @xmath0 sco is discussed by macfarlane & cassinelli ( 1989 ) .
a schematic of our model is shown in figure [ fig : cartoon ] .
the basic picture of the stellar wind that we adopt is one in which discrete density enhancements , or clumps , are embedded in an otherwise smooth stellar wind .
these dense clumps of material have shock interfaces with the faster moving ambient wind .
the most extreme of these clumps are dense enough to be optically thick to radiation in the strongest resonance lines , though still optically thin to uv continuum radiation .
in addition to their treatment of stellar winds with organized velocity gradients , the theory of radiative acceleration on material with no internal velocity gradient was developed by castor , abbott , & klein ( 1975 , hereafter cak ) , and we use their parameterization of the line - driving force . the equations governing this acceleration are given in
[ sec : infall ] . assuming a cak parameterization of the line driving force
, the net radiative acceleration on optically thick parcels of material with no internal velocity gradient can be relatively small ( see [ sec : infall ] ) .
gravity dominates the motion of these optically thick clumps , which fall inward , creating bow - shocks at their leading edges as they plow supersonically through the outflowing wind .
the velocity structure of the ambient wind is assumed to be relatively unaffected by such clumps .
they do not shadow a significant amount of the overlying wind , since they are optically thin to continuum radiation and the line opacity is significantly redshifted relative to the opacity driving the unperturbed wind .
as the clumps fall inward , they shock the out - flowing gas of the ambient ( inter - clump ) wind .
the temperature of gas the immediate post - shock region is proportional to the square of the shock jump velocity as in equation ( [ eqn : tshock ] ) .
since the relevant velocity for the calculation of the shock temperatures in these bow shocks is the _ relative _ velocity between the clumps and the ambient wind , the temperatures needed to produce the hard x - ray component observed by _ rosat _ and _ asca _ are readily obtained .
the x - ray producing shocks adjacent to the leading edge of the infalling clumps produce very highly - ionized material , thus accounting for the red - shifted high ions seen in absorption ( o@xmath1vi , n@xmath1v , etc . ) .
a related model was considered by lucy & white ( 1980 , hereafter lw ) , in the first paper attempting to explain the source of hot star x - rays as shock - heated gas .
they suggested that clumps are radiatively driven outward while the rest of the wind is pushed along by the drag of these objects . in our model
the clumps move outward ( or inward ) with a velocity less than that of the ambient wind , so the shocks occur on the inner face of the clump .
this placement of the shocks is more consistent with the results of ocr , who find that shocks tend to be inward facing with low - density , high - speed wind material being driven into slower and denser shocked regions . however , we retain the clump aspect of the lw model .
let us assume the clumps are self - contained parcels of gas confined by the ram pressure of the wind through which they fall .
following lw , we assume a balance of dynamical pressures and write the average density of a clump , @xmath35 , falling through the wind as @xmath36 in this expression @xmath37 is the wind density , a function of the distance @xmath38 from the star , @xmath39 and @xmath40 are the thermal velocities in the wind and clump , respectively , and @xmath41 is the relative velocity between the wind and the clump . we have ignored the effects of turbulence within the clumps but have included the constant @xmath42 as a scale factor to account for the range of angles at which the wind hits the bow - shock ( see lw ) .
we can see from this that once a clump begins to decelerate , thereby increasing @xmath41 , the oncoming wind confines it more effectively , making it denser as it falls towards the star .
this further decreases the ability of the radiation field to drive the clump outwards .
if the mass of material accumulated into a clump is specified , then the volume can be readily obtained from the density derived using equation ( [ eqn : dens ] ) .
this model has the desirable attribute of producing both the red - shifted high - ion absorption and the anomolously hard x - rays .
several independent lines of evidence suggest that clumping can be important in the winds of wolf - rayet ( e.g. , lpine , moffat , & henriksen 1996 ) and ob supergiant winds ( e.g. , sako et al .
1999 ; eversberg , lpine , & moffat 1998 ) .
it is within reason that near main sequence early type stars may also contain localized regions of significantly higher than average densities .
the exact mechanism for the formation of these dense clumps is not rigorously approached in this work , but it is worthwhile to explore the consequences of such a model before a more thorough hydrodynamical study is undertaken .
the main driving force for winds from massive stars is the radiative acceleration associated with atomic absorption line opacity .
the most important of the driving lines are the ultraviolet resonance lines of abundant ionic species .
thus , the driving force is not only dependent upon the elemental abundances in a wind , but also upon the wind s ionization balance ( macfarlane , cohen , & wang 1994 ; abbott 1982 ) .
the equation of motion describing the trajectory of a parcel of gas with respect to the star is @xmath43 where @xmath44 is the velocity of the clump of material , @xmath45 the mass of the star , @xmath38 the distance of a clump from its center , and @xmath46 is the force per unit mass the ambient wind exerts upon the clump in the form of a drag . "
the terms @xmath47 and @xmath48 are the ratios of the radiative to gravitational acceleration for electron scattering and resonance line scattering , respectively .
the drag force , @xmath46 , is zero if there is no relative motion between a clump and the ambient wind . in this treatment
we have not allowed for a thermal pressure gradient in the wind , which is negligible in the supersonic portion of such an outflow . for the ambient stellar wind , @xmath49 .
the requirement for the material to be accelerated away from the star is @xmath50 , implying @xmath51 , i.e. , that the radiative acceleration is greater than the gravitational acceleration .
a dense clump of material formed in the wind that has @xmath52 may decelerate and possibly fall back to the surface of the star , plowing its way through the out - flowing wind as it falls .
however , even for clumps with @xmath53 , the drag force , @xmath46 , may be great enough to push the clump outwards in the stellar wind . to determine if infall is physically reasonable
, we shall develop more fully each of the terms in equation ( [ eqn : motion1 ] ) .
the radiative driving force on a parcel of wind material is considered to have two components : radiative driving via electron scattering and resonance line opacities .
the radiative acceleration due to electron scattering of the stellar continuum is written @xmath54 here @xmath55 is the electron scattering opacity , which is @xmath56 @xmath57 g@xmath3 for @xmath0 sco , @xmath58 is the frequency - integrated stellar flux at the location of the material , and @xmath59 is the speed of light . a proper treatment of the radiative driving of stellar winds owing to absorption line opacity has been given by cak ( see also abbott 1982 ) .
their theory paramaterizes the line driving force on a wind having an organized velocity gradient @xmath60 . in cak theory , the radiative acceleration due to line scattering , @xmath61 ,
is written @xmath62 where @xmath63 is for a reference electron scattering opacity of 0.325 @xmath57 g@xmath3 .
so @xmath61 is simply the reference electron scattering acceleration modified by a multiplicative factor .
this factor , @xmath64 , the so - called `` force multiplier , '' is a function of @xmath65 , an optical depth parameter that is independent of line strength .
the force multiplier is defined as @xmath66 where @xmath67 is the electron density in units of @xmath68 @xmath17 , @xmath69 is the dilution factor and @xmath70 and @xmath24 are the cak parameters .
these parameters have been determined from studies employing extensive atomic line lists and depend upon the temperature of the absorbing material ( see , e.g. , abbott 1982 ) .
however , the driving force for the density enhancements we will consider here is slightly different from the typical cak driving force for the ambient wind . in the cak treatment of expanding atmospheres or winds
, @xmath65 is the electron scattering optical depth over the distance in which the wind velocity increases by the thermal velocity . hence @xmath65 is inversely proportional to the local velocity gradient for outflows with monotonically increasing velocity laws ( cak ) . in the case of parcels of gas with no organized velocity gradient , such as those being considered here
, the optical depth parameter @xmath65 is the integral of the electron scattering opacity over the path length @xmath71 through the region or clump [ equation ( 5 ) of cak ] : @xmath72 approximating the integral as @xmath73 , where @xmath74 is the volume of the clump , equation ( [ eqn : tdef2 ] ) may be rewritten , using equation ( [ eqn : dens ] ) , as @xmath75 assuming @xmath76 .
this latter assumption has the desirable consequence that the density of a clump will approach that of the ambient wind as the relative velocity , @xmath41 , approaches zero .
the ratio of the radiative to gravitational accelerations in cak formalism is written @xmath77 \left [ 1 + { \cal m}(t ) \right ] , \label{eqn : gamma}\ ] ] given @xmath78 , the luminosity of the star .
we will assume the values of the cak parameters tabulated by abbott ( 1982 ) for stellar winds at 30,000 k ( @xmath79 and @xmath80 ) are appropriate for use in the clumps considered here .
these are the same as those used for estimating the mass loss rate . given the high densities predicted for the clumps and the presence of a hard x - ray source at their leading edges ,
the ionization balance in the clumps could be different than that of the ambient wind .
thus while the cak parameters appropriate for the clump may be different than those used here , we adopt these parameters in order to proceed . in the end , the choice of cak parameters will likely not matter too much so long as the optical depth through a clump is very high .
equation ( [ eqn : gamma ] ) gives an upper limit to the radiative acceleration because it assumes that the flux from the star reaching a parcel of gas is unattenuated by the layers closer to the star . for a clump moving outwards at lower velocities than the ambient wind
, the photons producing the line radiation pressure on the dense clump may already have been absorbed or scattered by gas lower in the wind with the same outward velocity as the clump .
we will ignore this effect here , which is expected to be small .
the standard drag force per unit mass on a clump of area @xmath81 moving through an ambient medium with a relative velocity @xmath41 is given by @xmath82 in this treatment @xmath83 is the mass of the clump , and @xmath84 is the density of the ambient medium ( in this case the ambient stellar wind ) .
the drag coefficient , @xmath85 , dictates the efficiency of momentum transfer to the clump , and the factor of @xmath86 is conventionally included .
this expression describes the energy density impinging on the clump integrated over the area of the clump . for most of this work
we will assume @xmath87 , i.e. , perfect transfer of momentum from the ambient wind to the clump .
this is the most stringent assumption for determining if infall is viable .
the cross section of the clump may be approximated @xmath88 , giving @xmath89 using equation ( [ eqn : dens ] ) , and assuming @xmath90 and @xmath91 , we rewrite this as @xmath92 the density of the wind at a distance @xmath38 from the surface of the star is calculated assuming mass conservation : @xmath93 where m is the mass loss rate ( see table [ table : stardata ] ) .
the velocity of the wind @xmath94 is calculated using a @xmath95 velocity law with @xmath96 ( groenewegen & lamers 1989 ) : @xmath97 with the values of @xmath26 and @xmath98 given in table [ table : stardata ] .
the magnitude of the drag force can be relatively small .
we estimate its magnitude using equations ( [ eqn : drag ] ) and ( [ eqn : continuity ] ) with the data given in table [ table : stardata ] : @xmath99 where we have assumed @xmath100 , which in this equation is in units km s@xmath3 . if we define @xmath101 , where @xmath102 is the force of gravity per unit mass , we can rewrite equation ( [ eqn : motion1 ] ) as @xmath103 .
we find @xmath104 where @xmath105 is again in km s@xmath3 .
we can see that very near the star , the drag force is relatively small compared with the force of gravity because the velocity between the clump and the ambient wind is smaller than at larger radii .
this prediction is based on the assumption @xmath106 , but for clumps producing temperatures in excess of @xmath4 k , our numerical calculations discussed in [ subsec : traject ] suggest the drag force is much more significant ( relative to gravity ) at larger radii . for clumps with masses @xmath107 ( in grams )
the outward drag force is comparable to the force of gravity at large distances ( @xmath108 ) from the stellar surface . for lower mass clumps
, however , the drag force can play an important role in the dynamics at intermediate distances from the star ( @xmath109 ) .
we have derived the individual components of equation ( [ eqn : motion1 ] ) , and can hence obtain trajectories in position - velocity space for clumps in the wind of @xmath0 sco .
equation ( [ eqn : motion1 ] ) shows the dynamics of such clumps depend on the relative velocity , @xmath41 , between the clump and the ambient wind : @xmath110 , where @xmath94 and @xmath44 are the wind and clump velocities , respectively .
we have numerically integrated the equation of motion for parcels of gas with various initial conditions since it does not have a analytically tractable solution .
the initial conditions specified in our approach are the initial distance @xmath111 from the center of the star , the initial relative velocity @xmath112 , and the mass @xmath83 of the clump ( we assume mass conservation ) . by specifiying @xmath112 and @xmath111
, we are also then specifying the initial velocity of the clump , given equation ( [ eqn : windv ] ) .
we assume the value of @xmath112 is characteristic of the jump velocities typically seen in numerical simulations , i.e. , @xmath113 ( cooper 1994 ) .
the dependence on @xmath83 comes in through equations ( [ eqn : t2 ] ) and ( [ eqn : drag ] ) , the former affecting @xmath64 in the radiative acceleration .
our numerical integrations of the equation of motion show two types of clump trajectories : outflow and infall trajectories .
we term those trajectories in which the clumps are driven from the star , though they may initially decelerate , _ outflow _
though these clumps are driven outward , they always have velocities less than the terminal velocity of the wind .
those that decelerate and eventually descend towards the star follow _
infall _ trajectories .
the latter type is the most interesting , given the redshifted high ion absorption line observations .
it is not clear , however , that outflow and infall trajectories are mutually exclusive .
there may exist a mix of trajectories based upon a distribution of initial conditions for the clumps .
figure [ fig : trajectory ] displays the velocity of representative outflow and infall trajectories as a function of distance from the star .
also shown is the adopted @xmath95 velocity law of the wind . to demonstrate the sensitivity of these trajectories to initial conditions
, we have compiled a series of trajectories that are different in only one of the three possible input conditions .
figure [ fig : trajectory ] shows the trajectories of clumps with @xmath114 and @xmath115 km s@xmath3 ; the masses of the clumps in this figure are @xmath116 19.0 , 19.5 , and 20.0 ( @xmath83 in grams ) .
the low - mass clumps are driven outward through the wind while the high - mass objects relatively quickly stop and start to descend towards the star .
the low - mass clumps continue outwards because of their lower column density ( and correspondingly low @xmath117 ) , allowing for more effective radiation driving .
lower - mass clumps are also more effectively pushed outwards by the drag force provided by the on - rushing stellar wind , though this is usually minor compared with the force provided by resonance line scattering of radiation .
figure [ fig : trajectory2 ] shows trajectories for clumps having the same mass and initial velocity relative to the wind [ @xmath118 but with starting positions @xmath119 , 1.15 , 1.20 , and 1.25 @xmath120 ; and in figure [ fig : trajectory3 ] we show trajectories for clumps having a mass @xmath121 starting at @xmath122 with initial velocities relative to the wind of @xmath123 , 200 , 300 and 400 km s@xmath3 . clumps with larger @xmath124 tend to fall in more readily than those with low @xmath124 , as parcels of gas with smaller initial distances from the stellar surface , @xmath125 , are also more likely to fall inwards .
we note that the factor @xmath126 included in equation ( [ eqn : dens ] ) is not well known .
if this constant were larger , the clumps in our models would be more effectively confined and have higher densities .
this would make it easier for a clump of given initial conditions to fall inwards . in table
[ table : traj ] we give several properties for trajectories with a range of initial conditions . included in the table
are the initial conditions @xmath124 , @xmath125 , and @xmath127 .
also given is the lifetime @xmath128 of the clump , in hours , the final velocity of the clump in our calculations , the maximum relative velocity between the clump and the stellar wind , @xmath129 , and the corresponding maximum shock temperature . for the outflow trajectories ,
the final velocity is the velocity of the clump at a distance of @xmath130 from the surface of the star , and the lifetime is the time it takes to reach this point .
it is clear from figures [ fig : trajectory][fig : trajectory3 ] that conditions for infall can be met for a broad range of initial conditions .
our hypothesis that the infall of material may be responsible for the observed red - shifted component of the high - ionization lines is thus reasonable .
table [ table : traj ] shows that many combinations of initial conditions can yield infall trajectories .
infall velocities of @xmath32 km s@xmath3 ( as observed in the red wings of the o@xmath1vi lines ) are found for clumps with masses in excess of @xmath131 grams .
clumps originating at larger distances from the star reach higher infall velocities .
clumps following infall trajectories spend more time near their `` turning point , '' i.e. , where @xmath132 , than close to the photosphere where their infall velocity reaches its maximum .
the observed o@xmath1vi profiles require this , showing absorption consistent with a gaussian optical depth distribution centered at @xmath132 km s@xmath3 relative to the photosphere .
typical lifetimes for clumps following infall trajectories are @xmath133 hours , while outflow trajectories take @xmath134 hours to reach @xmath135 . in general clumps with initial distances
@xmath136 with @xmath137 require masses of order @xmath138 to achieve infall .
the highest - temperature gas is produced for clumps that reach the largest distances from the stellar surface .
thus lower mass clumps starting at relatively large distances from the star produce the hardest x - rays .
one can see from table [ table : traj ] that the trajectories followed by high - density clumps with @xmath139 tend to produce slightly harder x - rays than those with @xmath140 or 20.0 .
we believe the most likely clump masses for @xmath0 sco are in the range @xmath121 to 19.5 ( in grams ) given their ability to infall and produce very hot gas .
for infall trajectories @xmath129 occurs near the turn - around point in a cloud s trajectory .
specifically , the largest velocity jumps are usually found immediately after a clump of material has begun to fall inwards . at this point
the parcel of gas is at the maximum height above the star , and the ambient wind velocity is larger than at any other point in the clump s trajectory . though clumps may accelerate as they near the star , the decrease in the wind velocity with distance from the stellar surface offsets this increase in the infall velocity of the material
having shown that the interpretation of the red - shifted high - ion absorption as infall is a physically reasonable one , we now approach the anomolous x - ray spectrum for @xmath0 sco .
we have speculated that the observed x - rays are a by - product of the interaction of infalling clumps with the outgoing wind .
the temperature @xmath28 for gas in a section of a bow shock can be calculated by substituting @xmath141 for @xmath142 in equation ( [ eqn : tshock ] ) , where @xmath143 is the shock - jump velocity ( here the relative velocity between the wind and a clump , @xmath41 ) and @xmath144 is the angle at which the material enters the shock , relative to a ray normal to the shock s surface .
the temperature of gas is a function of position along the shock s leading edge , via the @xmath144 dependence .
calculating the true run of temperatures requires knowledge of the exact shape of the bow - shock .
this is an inherently difficult problem and is beyond the scope of this work . when discussing the temperatures produced by our infalling parcels of gas
, we will typically give results that assume @xmath145 , and , following lw , values corresponding to the shock at @xmath146 .
truth may be somewhere between these values for the majority of the shocked gas . the trajectories calculated in
[ subsec : traject ] permit us to estimate the bow - shock temperature at each point of a clump s history given its velocity relative to the ambient wind .
it is clear that the infall velocities of the trajectories displayed in figures [ fig : trajectory][fig : trajectory3 ] can exceed @xmath147 km s@xmath3 , sometimes approaching @xmath148 km s@xmath3 near the base of the wind . as mentioned in
[ subsec : traject ] , the relative velocity between the wind and the infalling clumps can be much higher than these values .
the maximum relative velocities in the trajectories for clouds having @xmath140 , 18.5 , 19.0 , and 19.5 in figure [ fig : trajectory ] are @xmath149 1660 , 1295 , and 1202 km s@xmath3 .
these velocities imply maximum shock temperatures of order @xmath150 k , respectively , well in excess of the temperatures required to match the ccw and cassinelli et al .
( 1994 ) fits .
even the outflow trajectories provide very large velocity jumps , and hence large shock - jump temperatures .
if we assume @xmath151 we find @xmath152 k. this is still quite close to the temperatures required to explain the _ asca _ observations of ccw . figures [ fig : tempdist1 ] and [ fig : tempdist2 ] show two trajectories as well as the model temperature distribution of the shock - heated gas .
the initial conditions for the trajectory shown in figure [ fig : tempdist1 ] are @xmath153 , while those for figure [ fig : tempdist2 ] are @xmath154 ; in each case @xmath155 . changing the initial height above the stellar surface between these two figures causes the clump to flow outwards in the wind .
the maximum shock temperatures expected from these two trajectories are near @xmath156 k. the insets show the expected _ relative _ temperature distribution over the lifetime of the clump .
these distributions have been weighted by @xmath157 for each point in the clumps trajectories to approximately account for the expected differences in the x - ray emission measure at different points in the wind .
we show both the distribution assuming @xmath158 ( solid histogram ) and @xmath146 ( dotted histogram ) in these figures .
the temperature distributions shown in figures [ fig : tempdist1 ] and [ fig : tempdist2 ] show that the trajectories of dense parcels of gas derived in [ sec : infall ] provide large enough shock - jump velocities to produce gas in excess of @xmath4 k. this treatment is very approximate .
for instance we have neglected the cooling of the gas in the bow shocks around the clumps .
the gas that is heated to @xmath159 will eventually cool through the lower x - ray emitting temperatures , providing for enhanced emission in the soft x - ray bands over that shown in figures [ fig : tempdist1 ] and [ fig : tempdist2 ] . though the temperature distributions peak at the same temperatures , the temperature distribution for the infall trajectory ( figure [ fig : tempdist1 ] ) is narrower than that of the outflow trajectory ( figure [ fig : tempdist2 ] ) , which shows a strong tail towards higher temperatures .
the ratio of the emission measures of x - ray emitting gas observed by ccw at ( 1.2 and @xmath160 k is @xmath161 .
both the outflow and infall trajectories have similar displayed in these figures produce x - ray emitting material at ( 1.2 and @xmath162 k in a ratio of 1.9:1 .
this value is consistent with the observational constraints , though the effects of radiative and adiabatic cooling will affect the final distribution of x - ray emission . in all likelihood a population of clumps in the wind of @xmath0 sco will have a range of initial conditions , providing a range of x - ray producing temperatures and x - ray luminosities per clump .
the distribution of intial conditions , the true shape of the bow shock at the leading edges of the infalling clumps , and the cooling of the post - shock gas will all determine the true temperature and luminosity distribution of x - ray emitting gas .
the derivation of these factors is beyond the scope of this paper .
our goal is to show that the production of very hot gas , in excess of @xmath4 k , is possible in the wind of @xmath0 sco through the infall of dense material .
our dynamical modelling of the dense clumps in [ sec : infall ] has shown that infall can occur .
the discussion presented in this section shows that the infall can create the requisite conditions for producing this very hot gas the x - ray luminosity predicted in our model is also reasonable .
the mechanical luminosity of the wind @xmath163 is a factor of @xmath164 greater than @xmath165 for this star . even if we assume the appropriate velocity for use is not the full terminal velocity , but rather @xmath166 km s@xmath3 , similar to the observed wind velocities at distances typical of our model clumps , we still have @xmath167 .
thus the ensemble of clumps need only convert @xmath168 of the wind mechanical luminosity to x - ray luminosity to match the observed @xmath165 .
we estimate the available luminosity , @xmath169 , for x - ray production from a single clump as @xmath170 .
the trajectory in table [ table : traj ] with initial conditions @xmath171 reaches a maximum height of @xmath172 with a relative velocity @xmath173 km s@xmath3 at this height .
this clump will have an area @xmath174 @xmath175 at its maximum height , and we find @xmath176 ergs s@xmath3 .
thus of order @xmath177 individual clumps , assuming perfect efficiency , are required to produce the observed x - ray luminosity of the star , @xmath178 ergs s@xmath3 ( cohen , cassinelli , & macfarlane 1997 ; hereafter ccm ) .
if we assume an efficiency of @xmath179 for conversion of kinetic energy to x - ray luminosity in the _ rosat _ bandpass ( see , e.g. , wilson & raymond 1999 ) , then the total number of clumps required by the x - ray observations is of order @xmath180 .
the observed luminosity of the hard component is significantly lower than the total ( @xmath181 ergs s@xmath3 ; ccw ) , in qualitative agreement with the temperature distributions shown in the insets of figures [ fig : tempdist1 ] and [ fig : tempdist2 ] .
we can estimate a lower limit to the number of clumps from the lack of ( short - term ) variability in the observed x - rays ( ccm ) .
the population of clumps that exist in the wind at any one time ( @xmath33 ) should be large enough to produce x - ray variations less than @xmath182 , or @xmath183 clumps .
assuming 600 clumps of mass @xmath184 with lifetimes similar to those in table [ table : traj ] , @xmath185 hours , this yields a creation rate of mass in clumps of @xmath186 m@xmath10 yr@xmath3 . assuming half of the clumps
actually escape in the wind ( follow outflow trajectories ) , the mass loss rate in clumps is @xmath187 of m. another constraint on the number of clumps residing in the stellar wind of @xmath0 sco comes from the o@xmath1vi absorption line profile .
the @xmath188 profile shows @xmath189 absorption near @xmath190 km s@xmath3 .
assuming the o@xmath1vi optical depth for an individual clump is much larger than unity and , as above , that the clumps are at a distance @xmath191 with mass @xmath192 , approximately @xmath193 clumps are required to cover 10% of the total solid angle of the star .
this corresponds to a mass loss rate of @xmath194 m@xmath10 yr@xmath3 for assumed lifetimes @xmath195 hours .
if the efficiency of mechanical to x - ray luminosity from the previous paragraph is @xmath189 , these two estimates of the number of clumps are in agreement .
it is worth noting that the infall trajectories are naturally weighted to provide more absorption near @xmath196 km s@xmath3 , i.e. , near the upper - most point in the trajectory , than at larger infall velocities .
this can be seen in figure [ fig : tempdist1 ] where we have plotted a dot every 1.25 hours along the clump s trajectory .
the high density of dots near the turning points implies that a random sample of clumps drawn from from this trajectory will be heavily weighted towards low velocities , in agreement with the gaussian shape of the o@xmath1vi absorption .
detailed line - profile calculations will be the subject of a future work .
our calculations suggest that the required number of clumps in the wind of @xmath0 sco is of order @xmath180 , with m@xmath197 approximately @xmath198 m@xmath10 yr@xmath3 , consistent with the adopted mass loss rate of this star .
these rough numbers can vary significantly depending on the assumptions made , however . at the present time there is no clear evidence for an inconsistency with the observational constraints .
further work will need to place more stringent constraints on the allowable parameters of this model .
though our simple model can explain several of the observational aspects of the stellar wind of @xmath0 sco , there are several questions and details that we have not approached .
principal among these is what makes @xmath0 sco so special ?
why do most hot stars not show the observed peculiarities of this star ?
this is may be due to the special location of @xmath0 sco on the main sequence where the mass loss rate steeply drops with @xmath199 .
main sequence stars hotter than b0 , i.e. , the main sequence o stars , have mass loss rates considerably higher than @xmath0 sco . for instance , the mass loss rate of o9 v stars with @xmath200 k , @xmath201 , and @xmath202 are of order @xmath203 m@xmath10 yr@xmath3 ( predicted by the cak theory ) , compared with less than @xmath204 or @xmath205 m@xmath10 yr@xmath3 for b1.5 and later stars .
this is observed in the sudden disappearance of the p cygni profiles of the uv resonance lines in stars between o9.5v and b0.5v ( see the atlas of uv line profiles of snow et al .
the mass loss rates of b1.5 v stars are typically a factor of @xmath206 below @xmath0 sco ( e.g. , the stars in ccm ) , with b2 star mass loss rates often being a factor of 100 lower than that adopted for @xmath0 sco .
we speculate that the prevailing conditions for stars near the position of @xmath0 sco in the hr diagram are just right not only for the formation of dense clumps of material within the wind , but also for their ability to infall .
stellar winds near the spectral type of @xmath0 sco are only marginally able to be driven by the radiation of the underlying stars .
clumps with no organized velocity gradient formed in such winds are less likely to be pushed outward than in the winds of earlier - type stars . to investigate the differences between @xmath0 sco and earlier - type o stars , we have calculated trajectories for high - density clumps in the wind of the o9 v star 10 lacertae .
we adopt stellar parameters given in brandt et al .
( 1998 ) and snow et al .
( 1994 ) in our calculations : @xmath207 , @xmath208 m@xmath10 , @xmath209 , @xmath210 m@xmath10 yr@xmath3 , @xmath211 k , and @xmath212 km s@xmath3 .
we note that like @xmath0 sco , 10 lac has a low projected rotation velocity ( @xmath213 km s@xmath3 ) .
we find no infall trajectories for initial distances from the stellar surface @xmath214 , initial relative velocities @xmath215 , and masses in the range @xmath216 .
this is primarily a function of the increased luminosity to mass ratio of the star , though the extra drag force plays a limited role .
this explains the lack of red - shifted absorption components of o@xmath1vi in the spectrum of 10 lac .
our calculations show that clumps in the wind of 10 lac will have shock jump velocities in excess of 1000 km s@xmath3 , producing temperatures in excess of @xmath4 k. this is very close to the value of @xmath217 found for shocks in `` normal '' stellar winds
. the emission from ( outflowing ) clumps in the wind of 10 lac will be indistinguishable from the normal x - ray flux due to shocks in unstable stellar winds of o - stars .
the lack of redshifted absorption profiles in the uv lines of main sequence stars of spectral type b0.5 and later is due to drastic decrease in mass loss .
a simple scaling of the optical depth of the o@xmath1vi lines by a factor 0.1 compared to @xmath0 sco , due to a ten times lower mass loss rate , makes these lines undetectable in the _ copernicus _ spectra , though few such stars were observed with _
the significantly lower mass loss rates of later - type stars may not provide the requisite conditions for formation of an ensemble of clumps like those thought to exist in the wind of @xmath0 sco . in general we would expect fewer , lower mass , clumps to form in the winds of later type stars .
such clumps would be less dense than their counterparts in the wind of @xmath0 sco , hence more likely to flow outward .
in the winds of later - type b stars , with significantly lower densities and terminal velocities , the clumps would produce lower temperature and lower luminosity x - rays , perhaps undetectable or indistinguishable from the expected shock emission .
we have outlined a model to explain the two peculiar observational properties of the stellar wind of @xmath0 sco : infall of high - density clumps surrounded by very highly - ionized gas that gives rise to the red - shifted absorption wings of the o@xmath1vi lines , and strong shocks that produce the hard x - ray spectrum .
our model hypothesizes that density enhancements in the stellar wind lead to the formation of dense clumps of material , confined by ram pressure , that decouple from the dynamical flow of the ambient stellar wind .
these clumps have very large optical depths in the uv resonance lines responsible for driving the stellar wind .
the resulting radiative acceleration of the clumps is insufficient to overcome the gravity , so they slow down and fall back towards the star . as they do so
, their interaction with the on - rushing stellar wind produces a bow shock , with shock temperatures reaching several times @xmath4 k. we have presented detailed dynamical models of density enhancements in the wind of @xmath0 sco , which show that infall can be achieved for a wide range of initial conditions . using these trajectories
we have shown that the shock - jump velocities can approach 2000 km s@xmath3 .
the anomolously hard x - rays observed from @xmath0 sco by cassinelli et al.(1994 ) and ccw are produced in these bow shocks at temperatures of @xmath218 k. the infalling high - ionization material observed in the o@xmath1vi doublet by lamers & rogerson ( 1978 ) is material in and around the infalling clumps that is ionized by the hard x - rays produced in the bow shocks at the clumps leading edges .
our calculations show that clumps with masses in the range of @xmath219 to @xmath220 g will fall back if they are formed at radii @xmath221 ( see table 2 ) .
only the clumps formed above 1.1 @xmath98 produce shocks with temperatures in excess of @xmath222 k. clumps formed this high have maximum infall velocities of order 500 km s@xmath3 .
this is higher than the observed maximum redshift of 250 km s@xmath3 , but the clumps spend very little time at such high velocities . even clumps with a large final infall velocity will give the maximum contribution to the the uv line profile at small velocities , because they spend much more time near their turning point , where @xmath223 , than at high velocity just before they disappear into the photosphere .
this explains why the o@xmath1vi absorption reaches a maximum near 0 km s@xmath3 .
we estimate that @xmath224 clumps of mass @xmath225 g are needed to explain the observed o@xmath1vi absorption lines and the hard x - ray flux .
this corresponds to a clump mass loss rate of m@xmath226 . in making these estimates
we have assumed several efficiency factors whose values are not well known at present and would require detailed modelling to be improved . for instance , ( a ) in comparing the predicted x - ray luminosities with observations we have assumed a perfect conversion of ram pressure heating to x - ray emission ; ( b ) we have neglected the radiative cooling of shocks which may result in a reduction of the predicted x - ray flux from an individual clump ; and ( c ) we have assumed that the clumps are optically thick in the o@xmath1vi transitions , and that the intrinsic line widths of the clumps is sufficient to account for the breadth of the observed absorption .
a reduction in these `` efficiencies '' may result in an increase in the required number of clumps .
this will be the subject of a subsequent study .
we explain the special situation of @xmath0 sco in terms of its location on the main sequence where the mass loss rate drops by about factor 100 from o9 v to b1 v. this can be tested and our model can be improved by high s / n observations of the far uv resonance lines of o@xmath1vi , , and of a number of o9.5 v to b1 v stars with the _ far ultraviolet spectroscopic explorer _ ( _ fuse _ ) .
this satellite will cover the wavelength range @xmath227 and will observe a large number of early - type stars .
our model would suggest that stars with spectral types similar to @xmath0 sco should also show infall in the o@xmath1vi doublet at @xmath228 and 1038 , which will be well observed with _
fuse_. the dynamical model developed in this work can be generally applied to clumps in line - driven outflows .
though it seems that the properties of stars near @xmath0 sco on the hr diagram may be the best for producing infalling clumps with recognizable observational consequences , clumping may occur in many types of early - type stellar winds , and may also be important in outflows from extragalactic sources ( e.g. , agns and qsos ) .
the dynamical model described here can be applied to all of these cases ( as discussed for the case of 10 lac above ) .
we thank d. cohen and a. cole for many conversations regarding this project .
jch recognizes support from a nasa graduate student researcher fellowship under grant number ngt-5 - 50121 .
jpc recognizes support from nasa under grant nag5 - 2854 .
lcl spectral type & b0.2 v & walborn(1994 ) distance ( pc ) & 132 & _ hipparcos _ ( perryman et al . 1998 ) @xmath229 ( mag ) & 2.82 & _ hipparcos _ ( perryman et al .
1998 ) @xmath230 ( @xmath231 ) & @xmath232 & diplas & savage ( 1994 ) @xmath199 ( k ) & 31,400 & kilian ( 1992 ) @xmath98 ( @xmath233 ) & 6.5 & snow et al . ( 1994 ) @xmath45 ( @xmath234 ) & 15 & snow et al .
( 1994 ) @xmath235 ( km s@xmath3 ) & 917 & snow et al . ( 1994 ) @xmath236 ( ergs s@xmath3 ) & @xmath237 & snow et al .
( 1994 ) @xmath165 @xmath238 ( ergs s@xmath3 ) & @xmath239 & cohen et al .
( 1997 ) @xmath240 ( km s@xmath3 ) & 20 & uesugi and fukuda ( 1982 ) m @xmath238 ( m@xmath10 yr@xmath3 ) & @xmath241 & this work @xmath242 @xmath238 ( km s@xmath3 ) & 2400 & this work cccccccc 18.0 & 1.10 & 176 & 352 & 6.3 & @xmath243 & 730 & 8 18.0 & 1.25 & 331 & 662 & 22.4 & @xmath244 & 1460 & 31 19.0 & 1.10 & 176 & 352 & 2.8 & @xmath245 & 580 & 5 19.0 & 1.15 & 235 & 470 & 4.5 & @xmath246 & 820 & 10 19.0 & 1.20 & 286 & 572 & 6.8 & @xmath247 & 1055 & 16 19.0 & 1.25 & 331 & 662 & 11.2 & @xmath248 & 1295 & 24 19.0 & 1.30 & 371 & 742 & 25.9 & @xmath249 & 1590 & 36 19.0 & 1.50 & 498 & 996 & 25.1 & @xmath250 & 1670 & 40 19.5 & 1.10 & 176 & 352 & 2.5 & @xmath251 & 560 & 5 19.5 & 1.15 & 235 & 470 & 3.8 & @xmath252 & 785 & 9 19.5 & 1.20 & 286 & 572 & 5.5 & @xmath253 & 1000 & 14 19.5 & 1.25 & 331 & 662 & 7.9 & @xmath254 & 1202 & 21 19.5 & 1.30 & 371 & 742 & 11.8 & @xmath255 & 1405 & 28 19.5 & 1.50 & 498 & 996 & 33.7 & @xmath256 & 1820 & 48 20.0 & 1.10 & 176 & 352 & 2.3 & @xmath257 & 555 & 4 20.0 & 1.20 & 286 & 572 & 4.8 & @xmath258 & 970 & 13 20.0 & 1.30 & 371 & 742 & 9.3 & @xmath259 & 1340 & 26 20.0 & 1.50 & 498 & 996 & 58.5 & @xmath260 & 1990 & 57 | observations of the b0.2 v star @xmath0 scorpii have revealed unusual stellar wind characteristics : red - shifted absorption in the far - ultraviolet o@xmath1vi resonance doublet up to @xmath2 km s@xmath3 , and extremely hard x - ray emission implying gas at temperatures in excess of @xmath4 k. we describe a phenomenological model to explain these properties .
we assume the wind of @xmath0 sco consists of two components : ambient gas in which denser clumps are embedded .
the clumps are optically thick in the uv resonance lines primarily responsible for accelerating the ambient wind .
the reduced acceleration causes the clumps to slow and even infall , all the while being confined by the ram pressure of the outflowing ambient wind .
we calculate detailed trajectories of the clumps in the ambient stellar wind , accounting for a line radiation driving force and the momentum deposited by the ambient wind in the form of drag .
we show these clumps will fall back towards the star with velocities of several hundred km s@xmath3 for a broad range of initial conditions .
the velocities of the clumps relative to the ambient stellar wind can approach 2000 km s@xmath3 , producing x - ray emitting plasmas with temperatures in excess of @xmath5 k in bow shocks at their leading edge .
the infalling material explains the peculiar red - shifted absorption wings seen in the o@xmath1vi doublet .
of order @xmath6 clumps with individual masses @xmath7 g are needed to explain the observed x - ray luminosity and also to explain the strength of the o@xmath1vi absorption lines .
these values correspond to a mass loss rate in clumps of @xmath8 to @xmath9 m@xmath10 yr@xmath3 , a small fraction of the total mass loss rate ( @xmath11 m@xmath10 yr@xmath3 ) .
we discuss the position of @xmath0 sco in the hr diagram , concluding that @xmath0 sco is in a crucial position on the main sequence .
hotter stars near the spectral type of @xmath0 sco have too powerful winds for clumps to fall back to the stars , and cooler stars have too low mass loss rates to produce observable effects . the model developed here
can be generally applied to line - driven outflows with clumps or density irregularities . |
next generation wireless communication networks are required to provide ubiquitous and high data rate communication with guaranteed quality of service ( qos ) .
these requirements have led to a tremendous need for energy in both transmitter(s ) and receiver(s ) . in practice ,
portable mobile devices are typically powered by capacity limited batteries which require frequent recharging . besides
, battery technology has developed very slowly over the past decades and the battery capacities available in the near future will be unable to improve this situation .
consequently , energy harvesting based mobile communication system design has become a prominent approach for addressing this issue . in particular , it enables self - sustainability for energy limited communication networks .
in addition to conventional energy harvesting sources such as solar , wind , and biomass , wireless power transfer has been proposed as an emerging alternative energy source , where the receivers scavenge energy from the ambient radio frequency ( rf ) signals @xcite@xcite .
in fact , wireless power transfer technology not only eliminates the need of power cords and chargers , but also facilitates one - to - many charging due to the broadcast nature of wireless channels .
more importantly , it enables the possibility of simultaneous wireless information and power transfer ( swipt ) leading to many interesting and challenging new research problems which have to be solved to bridge the gap between theory and practice . in @xcite ,
the authors investigated the fundamental trade - off between harvested energy and wireless channel capacity across a pair of coupled inductor circuit in the presence of additive white gaussian noise .
then , in @xcite , the study was extended to multiple antenna wireless broadcast systems . in @xcite ,
the energy efficiency of multi - carrier systems with swipt was revealed .
specifically , it was shown in @xcite that integrating an energy harvester into a conventional information receiver improves the energy efficiency of a communication network . in @xcite ,
robust beamforming design for swipt systems with physical layer security was investigated .
the results in @xcite@xcite indicate that both the information rate and the amount of harvested energy at the receivers can be significantly increased at the expense of an increase in the transmit power .
however , despite the promising results in the literature , the performance of wireless power / energy transfer systems is still limited by the distance between the transmitter and the receiver due to the high signal attenuation associated with path loss .
coordinated multipoint ( comp ) transmission is an important technique for extending service coverage , improving spectral efficiency , and mitigating interference @xcite@xcite .
a possible deployment scenario for comp networks is to split the functionalities of the base stations between a central processor ( cp ) and a set of remote radio heads ( rrhs ) . in particular
, the cp performs the power hungry and computationally intensive baseband signal processing while the rrhs are responsible for all radio frequency ( rf ) operations such as analog filtering and power amplification . besides
, the rrhs are distributed across the network and connected to the cp via backhaul links .
this system architecture is known as cloud computing network . as a result
, the comp systems architecture inherently provides spatial diversity for combating path loss and shadowing .
it has been shown that a significant system performance gain can be achieved when full cooperation is enabled in comp systems @xcite . however , in practice , the enormous signalling overhead incurred by the information exchange between the cp and the rrhs may be infeasible when the capacity of the backhaul link is limited .
hence , resource allocation for comp networks with finite backhaul capacity has attracted much attention in the research community @xcite@xcite . in @xcite
, the authors studied the energy efficiency of comp multi - cell networks with capacity constrained backhaul links . in @xcite and @xcite ,
iterative sparse beamforming algorithms were proposed to reduce the load of the backhaul links while providing reliable communication to the users .
however , the energy sources of the receivers in @xcite@xcite were assumed to be perpetual and this assumption may not be valid for power - constrained portable devices . on the other hand , the signals transmitted by the rrhs could be exploited for energy harvesting by the power - constrained receivers for extending their lifetimes . however , the resource allocation algorithm design for comp swipt systems has not been solved sofar , and will be tackled in this paper . motivated by the aforementioned observations , we formulate the resource allocation algorithm design for multiuser comp communication networks with swipt as a non - convex optimization problem .
we jointly minimize the total network transmit power and the maximum capacity consumption per backhaul link while ensuring quality of service ( qos ) for reliable communication and efficient wireless power transfer .
in particular , we propose an iterative algorithm which provides a local optimal solution for the considered optimization problem .
we use boldface capital and lower case letters to denote matrices and vectors , respectively .
@xmath0 , @xmath1 , and @xmath2 represent the hermitian transpose , trace , and rank of matrix @xmath3 ; @xmath4 and @xmath5 indicate that @xmath3 is a positive definite and a positive semidefinite matrix , respectively ; @xmath6 denotes the vectorization of matrix @xmath3 by stacking its columns from left to right to form a column vector ; @xmath7 is the @xmath8 identity matrix ; @xmath9 denotes the set of all @xmath10 matrices with complex entries ; @xmath11 denotes the set of all @xmath8 hermitian matrices ; @xmath12 denotes a diagonal matrix with the diagonal elements given by @xmath13 ; @xmath14 and @xmath15 denote the absolute value of a complex scalar and the @xmath16-norm of a vector , respectively .
in particular , @xmath17 is known as the @xmath18-norm of a vector and denotes the number of non - zero entries in the vector ; the circularly symmetric complex gaussian ( cscg ) distribution is denoted by @xmath19 with mean @xmath20 and variance @xmath21 ; @xmath22 stands for distributed as " ; @xmath23{x}$ ] is the ceiling function denoting the smallest integer not smaller than @xmath24 .
we consider a comp multiuser downlink communication network .
the system consists of a cp , @xmath25 rrhs , @xmath26 information receivers ( irs ) , and @xmath27 energy harvesting receivers ( ers ) , cf .
figure [ fig : system_model ] .
each rrh is equipped with @xmath28 transmit antennas .
the irs and ers are single antenna devices which exploit the received signal powers in the rf for information decoding and energy harvesting , respectively . in practice
, the ers may be idle irs which are scavenging energy from the rf for extending their lifetimes .
on the other hand , the cp is the core unit in the network . in particular
, it has the data of all information receivers .
besides , we assume that the global channel state information ( csi ) is perfectly known at the cp and all computations are performed in this unit . based on the available csi , the cp computes the resource allocation policy and broadcasts it to all rrhs .
specifically , each rrh receives the control signals for resource allocation and the data of the @xmath26 irs from the cp via a backhaul link .
furthermore , we assume that the cp supplies energy to the rrhs in the network via dedicated power lines to support the rrhs power consumption .
remote radio heads ( rrhs ) , @xmath29 information receivers ( irs ) , and @xmath30 energy harvesting receiver ( ers).,width=336 ] we focus on a frequency flat fading channel and a time division duplexing ( tdd ) system .
each rrh obtains the local csi of all receivers by exploiting channel reciprocity and handshaking signals .
subsequently , the rrhs feed their local csi to the cp for computation of the resource allocation policy .
the received signals at ir @xmath31 and er @xmath32 are given by @xmath33 where @xmath34 denotes the joint transmit signal vector of the @xmath25 rrhs to ir @xmath35 .
the channel between the @xmath25 rrhs and ir @xmath35 is denoted by @xmath36 , and we use @xmath37 to represent the channel between the @xmath25 rrhs and er @xmath38 . we note that the channel vector captures the joint effects of multipath fading and path loss . @xmath39 and @xmath40 are additive white gaussian noises ( awgn ) .
we assume that the noise variances , @xmath41 , are identical at all receivers . in each scheduling time slot , @xmath26
independent signal streams are transmitted simultaneously to the @xmath26 irs .
specifically , a dedicated beamforming vector , @xmath42 , is allocated to ir @xmath35 at rrh @xmath43 to facilitate information transmission . for the sake of presentation ,
we define a super - vector @xmath44 for ir @xmath35 as @xmath45\big).\end{aligned}\ ] ] @xmath46 represents a joint beamformer used by the @xmath25 rrhs for serving ir @xmath35 .
then , @xmath47 can be expressed as @xmath48 where @xmath49 is the data symbol for ir @xmath35 and @xmath50 , is assumed without loss of generality . on the other hand ,
the data of each ir is delivered from the cp to the rrhs via backhaul links . the backhaul capacity consumption for backhaul link @xmath43
is given by @xmath51\big\rvert}_0\,\ , r_k,\end{aligned}\ ] ] where @xmath52 is the required backhaul data rate for conveying the data of ir @xmath35 to a rrh .
we note that the backhaul links may be capacity - constrained and the cp may not be able to send the data to all rrhs as required for full cooperation .
thus , to reduce the load on the backhaul links , the cp can enable partial cooperation by sending the data of information receiver @xmath35 only to a subset of the rrhs .
in particular , by setting @xmath53 , rrh @xmath54 is not participating in the joint data transmission to ir @xmath35 .
thus , the cp is not required to send the data for ir @xmath35 to rrh @xmath54 via the backhaul link which leads to a lower backhaul link capacity consumption .
in the considered comp network , we assume that the cp transfers energy to the rrhs for supporting the power consumption at the rrhs and facilitating a more efficient network operation .
in particular , @xmath55 units of energy are transferred to rrh @xmath54 via a dedicated power line where @xmath56 , is the power supplied by the cp to rrh @xmath54 .
@xmath57 is the power loss in delivering the power from the cp rrh @xmath54 .
@xmath58 is a constant proportional to the ratio between the resistance of the adopted power line and the voltage of power transmission .
we note that @xmath59 always hold by the law of conservation of energy .
the achievable rate ( bit / s / hz ) between the @xmath25 rrhs and ir @xmath35 is given by @xmath60 is the receive signal - to - interference - plus - noise ratio ( sinr ) at ir @xmath35 . on the other hand ,
the information signal , @xmath61 , serves as a dual purpose carrier for conveying both information and energy concurrently in the considered system .
the total amount of energy harvested by er @xmath32 is given by @xmath62 where @xmath63 denotes the efficiency of the conversion of the received rf energy to electrical energy for storage .
we assume that @xmath20 is a constant and is identical for all ers .
besides , the contribution of the antenna noise power to the harvested energy is negligibly small compared to the harvested energy from the information signal , @xmath64 , and thus is neglected in ( [ eqn : er_power ] ) .
the system objective is to jointly minimize the weighted sum of the total network transmit power and the maximum capacity consumption per backhaul link while providing qos for reliable communication and power transfer .
the resource allocation algorithm design is formulated as the following optimization problem : @xmath65 where @xmath66 and @xmath67 in the objective function are constants which reflect the preference of the system operator for the capacity consumption of individual backhaul links and the total network transmit power consumption , respectively . besides , @xmath68 can also be interpreted as the energy / power cost in conveying information to the rrhs via backhaul .
@xmath69 in constraint c1 indicates the required minimum receive sinr at ir @xmath35 for information decoding . the corresponding data rate per backhaul link use for ir @xmath35
is given by @xmath70 .
in c2 , @xmath71 and @xmath72 are the hardware circuit power consumption and the maximum power available at the cp , respectively . in c3 , @xmath73 and @xmath74
are the hardware circuit power consumption and the maximum available power at rrh @xmath54 , respectively .
@xmath75 is a constant which accounts for the power inefficiency of the power amplifier .
@xmath76 in c4 is the maximum transmit power allowance for rrh @xmath54 , which can be used to limit out - of - cell interference .
constant @xmath77 in constraint c5 specifies the required minimum harvested energy at er @xmath38 .
c6 is the non - negativity constraint on the power optimization variables .
we note that the objective function considered in this paper is different from that in @xcite and @xcite .
in particular , we focus on the capacity consumption of individual backhaul links while @xcite and @xcite studied the total network backhaul capacity consumption .
although the considered problem formulation does not constrain the capacity consumption of the individual backhaul links , it provides a first - order measure of the backhaul loading in the considered comp network when enabling partial cooperation .
this information provides system design insight for the required backhaul deployment .
the optimization problem in ( [ eqn : cross - layer ] ) is a non - convex problem due to the non - convexity of the objective function , constraint c1 , and constraint c5 . in particular , the combinatorial nature of the objective function results in an np - hard optimization problem @xcite . to strike a balance between system performance and computational complexity
, we develop an iterative algorithm for obtaining a suboptimal solution . to this end
, we first reformulate the optimization problem by approximating the original non - convex objective function as a weighted sum of convex functions with different weight factors .
then , we recast the reformulated problem as a semidefinite programming ( sdp ) problem via sdp relaxation and solve it optimally .
subsequently , a suboptimal solution to the original optimization problem is obtained by updating the weight factors and solving the reformulated problem iteratively .
the non - convex weighted capacity consumption of backhaul link @xmath54 , @xmath78 , can be approximated as follows : @xmath79\big\rvert}_0\,\ , r_k\\ \hspace*{-2mm}&\stackrel{(b)}{\approx}&\hspace*{-2mm}\delta\sum_{k=1}^k{\big\lvert\big[\rho_k^l{\lvert\mathbf{w}_k^l\rvert}_2 ^ 2\big]\big\rvert}_1 r_k=\delta\sum_{k=1}^k \rho_k^l{\lvert\mathbf{w}_k^l\rvert}_2 ^ 2 r_k\notag\end{aligned}\ ] ] where @xmath80 in @xmath81 are given constant weight factors which can be used to achieve solution sparsity .
@xmath82 indicates that the value of the @xmath18-norm is invariant when the input arguments are squared .
@xmath81 is due to the fact that the @xmath18-norm can be approximated by its convex hull which is the @xmath83-norm .
this approximation is known as convex relaxation and is commonly used in the field of compressed sensing for handling @xmath18-norm optimization problems @xcite@xcite .
we substitute ( [ eqn : objective_approx ] ) into ( [ eqn : cross - layer ] ) and define @xmath84 , @xmath85 , and @xmath86 .
then , we recast the reformulated problem in its epigraph form @xcite which is given as follows : @xmath87 where @xmath88 is a block diagonal matrix with @xmath89 . @xmath90 in the objective function and constraint c7 is an auxiliary optimization variable .
constraints c8 , c9 , and @xmath91 , are imposed to guarantee that @xmath84 holds after optimization .
[ h ] [ 1 ] initialize the maximum number of iterations @xmath92 and a small constant @xmath93 set iteration index @xmath94 and @xmath95 solve ( [ eqn : sdp_relaxation ] ) for a given set of @xmath96 and obtain an intermediate beamforming vector @xmath97 update the weight factor as follows : @xmath98 then , we relax constraint @xmath99 by removing it from the problem formulation , such that the considered problem becomes a convex sdp given by @xmath100 we note that the relaxed problem in ( [ eqn : sdp_relaxation ] ) can be solved efficiently by numerical solvers such as cvx @xcite . if the solution @xmath101 of ( [ eqn : sdp_relaxation ] ) is a rank - one matrix , then the problems in ( [ eqn : rank_one ] ) and ( [ eqn : sdp_relaxation ] ) share the same optimal solution and the same optimal objective value .
otherwise , the optimal objective value of ( [ eqn : sdp_relaxation ] ) serves as a lower bound for the objective value of ( [ eqn : rank_one ] ) .
next , we reveal the tightness of the sdp relaxation adopted in ( [ eqn : sdp_relaxation ] ) in the following theorem .
[ thm : rankone_condition ] assuming the channel vectors of the irs , @xmath102 and the ers , @xmath103 can be modeled as statistically independent random variables then the solution of ( [ eqn : sdp_relaxation ] ) is rank - one , i.e. , @xmath104 , with probability one .
_ proof : _ please refer to the appendix . @xmath105 in other words , whenever the channels satisfy the condition stated in theorem 1 , the optimal beamformer @xmath106 of ( [ eqn : rank_one ] ) can be obtained with probability one by performing an eigenvalue decomposition of the solution @xmath101 of ( [ eqn : sdp_relaxation ] ) and selecting the principal eigenvector as the beamformer . in general , for a fixed weight factor , @xmath107 , the solution of ( [ eqn : rank_one ] ) does not necessarily provide sparsity and the approximation adopted in ( [ eqn : objective_approx ] ) may not be tight . for improving the obtained solution , we adopt the _
reweighted @xmath83-norm method _ which was originally designed to enhance the data acquisition in compressive sensing @xcite .
the overall resource allocation algorithm is summarized in table [ table : algorithm ] . in particular ,
the weight factor @xmath107 is updated as in line 5 of the iterative algorithm such that the magnitude of beamforming vectors @xmath108 with small values are further reduced in the next iteration . as a result ,
by iteratively updating @xmath107 and solving ( [ eqn : sdp_relaxation ] ) , a suboptimal beamforming solution with sparsity can be constructed .
we note that the iterative algorithm in table [ table : algorithm ] converges to a local optimal solution of the original problem formulation in ( [ eqn : cross - layer ] ) for @xmath93 and a sufficient number of iterations @xcite .
furthermore , when the primal - dual path - following method @xcite is used by the numerical solver for solving ( [ eqn : sdp_relaxation ] ) , the computational complexity of the proposed algorithm is @xmath109 for a given solution accuracy @xmath110 .
the computational complexity is significantly reduced compared to the computational complexity of an exhaustive search with respect to @xmath26 and @xmath25 , i.e. , @xmath111 .
rrhs , @xmath112 irs , and @xmath30 ers.,width=192 ] path loss exponent & 3.6 + multipath fading distribution & + total noise variance , @xmath41 & + minimum required sinr , @xmath113 & + circuit power consumption at cp , @xmath114 & @xmath115 + circuit power consumption at the @xmath54-th rrh , @xmath73 & @xmath116 + max .
power supply at the cp , @xmath117 & @xmath118 + power amplifier power efficiency & @xmath119 + max .
transmit power allowance , @xmath120 & @xmath121 dbm + min
. required power transfer , @xmath122 & @xmath123 dbm + rf to electrical energy conversion efficiency , @xmath20 & @xmath124 + power loss in transferring power from the cp to rrh , @xmath125 & @xmath126 +
in this section , we evaluate the network performance of the proposed resource allocation design via simulations .
there are @xmath127 rrhs , @xmath112 irs , and @xmath30 ers in the system .
we focus on the network topology shown in figure [ fig : simulation_model ] .
the distance between any two rrhs is @xmath128 meters .
the three rrhs construct an equilateral triangle while the irs and ers are uniformly distributed inside a disc with radius @xmath129 meters centered at the centroid of the triangle .
the simulation parameters can be found in table [ tab : feedback ] . in the iterative algorithm
, we set @xmath130 and @xmath92 to @xmath131 and @xmath132 , respectively .
the numerical results in this section were averaged over 1000 independent channel realizations for both path loss and multipath fading .
the performance of the proposed scheme is compared with the performances of a full cooperation scheme , an optimal exhaustive search scheme , and a traditional system with co - located transmit antennas . for the full cooperation scheme ,
the solution is obtained by setting @xmath133 , @xmath134 , and solving ( [ eqn : sdp_relaxation ] ) by sdp relaxation . for the exhaustive search
, it is expected that multiple optimal solutions for ( [ eqn : cross - layer ] ) may exist .
thus , for the set of optimal solutions , we further select the one having the minimal total system backhaul capacity consumption . if there are multiple optimal solutions with the same total system backhaul capacity consumption , then we select the one requiring the minimal total network transmit power . as for the co - located transmit antenna system , we assume that there is only one rrh located at the center of the system equipped with the same number of antennas as all rrhs combined in the distributed stetting .
besides , the cp is not at the same location as the rrh for the co - located transmit antenna system , i.e. , a backhaul is still needed .
furthermore , we set @xmath135 for the co - located transmit antenna system to study its power consumption .
in figures [ fig : backhaul_nt_subfig1 ] and [ fig : backhaul_nt_subfig2 ] , we study the average maximum backhaul capacity consumption per backhaul link and the average total system backhaul capacity consumption , respectively , versus the total number of transmit antennas in the network , for different resource allocation schemes .
we set @xmath136 and @xmath137 in ( [ eqn : sdp_relaxation ] ) for the proposed scheme to fully minimize the maximum capacity consumption per backhaul link .
the performance of the proposed iterative algorithm is shown for @xmath138 and @xmath132 iterations .
it can be seen from figure [ fig : backhaul_nt_subfig1 ] that the proposed iterative algorithm achieves a close - to - optimal backhaul capacity consumption in all considered scenarios even for the case of @xmath138 iterations .
we note that the gap between the proposed algorithm and the exhaustive search in figure [ fig : backhaul_nt_subfig1 ] is caused by the sub - optimality of the objective function approximation in ( [ eqn : objective_approx ] ) and insufficient numbers of iterations .
in fact , the superior average maximum system backhaul capacity consumption of the optimal exhaustive scheme in figure [ fig : backhaul_nt_subfig1 ] compared to the proposed scheme comes at the expense of an exponential computational complexity with respect to the number of irs and rrhs . on the other hand ,
the performance gap between the proposed iterative resource allocation algorithm and the full cooperation scheme(/co - located antennas system ) increases as the total number of transmit antennas . in figures [ fig : backhaul_nt_subfig1 ] and [ fig : backhaul_nt_subfig2 ] , we observe that the average backhaul capacity consumption of the proposed algorithm decreases monotonically with an increasing number of antennas and converges to constant values close to the lower bounds , respectively .
the lower bounds in figures [ fig : backhaul_nt_subfig1 ] and [ fig : backhaul_nt_subfig2 ] are given by @xmath139{\frac{k}{l}}\log_2(1+\gamma_{\mathrm{req}})$ ] and @xmath140 , respectively .
indeed , when both the power budget and the number of antennas at the rrhs are sufficiently large , full cooperation may not be beneficial . in this case , conveying the data of each ir to one rrh may be sufficient for providing the qos requirements for reliable communication and efficient power transfer . hence , backaul system resources can be saved . besides , it can be seen from figure [ fig : backhaul_nt_subfig2 ] that the system with co - located antennas requires the smallest amount of total system backhaul capacity since the data of each ir is conveyed only to a single rrh . however , the superior performance of the co - located antenna system in terms of total network backhaul capacity consumption incurs the highest capacity consumption per backhaul link among all the schemes , cf .
figure [ fig : backhaul_nt_subfig1 ] . in figure
[ fig : pt_nt ] , we study the average total transmit power versus total number of transmit antennas for different resource allocation schemes . it can be observed that the total transmit power decreases monotonically with increasing number of transmit antennas .
this is due to the fact that the degrees of freedom for resource allocation increase with the number of transmit antennas , which enables a more power efficient resource allocation .
besides , the proposed algorithm consumes a lower transmit power compared to the optimal exhaustive search scheme .
this is because the exhaustive search scheme consumes a smaller backhaul capacity at the expense of a higher transmit power .
furthermore , the system with co - located antennas consumes a higher transmit power than the proposed scheme and the full cooperation scheme in all considered scenarios which reveals the power saving potential of comp due to its inherent spatial diversity . on the other hand
, it is expected that the full cooperation scheme is able to achieve the lowest average total transmit power at the expense of an exceedingly large backhaul capacity consumption , cf .
figure [ fig : backhaul_nt ] . in figure
[ fig : hp_nt ] , we study the average total harvested power versus the total number of transmit antennas in the network for different resource allocation schemes . we compare the average total harvested power of all resource allocation schemes with a lower bound which is computed by assuming that constraint c5 is satisfied with equality for all ers .
as can be observed , the total average harvested powers in all considered scenarios are monotonically non - increasing with respect to the number of transmit antennas .
this is because the extra degrees of freedom offered by the increasing number of antennas improve the efficiency of resource allocation . in particular ,
the direction of beamforming matrix @xmath101 can be more accurately steered towards the irs which reduces the power allocation to @xmath101 and the leakage of power to the ers .
this also explains the lower harvested power for the full cooperation scheme and the system with co - located antennas since they both exploit all transmit antennas in the network for joint transmission . on the other hand ,
the highest amount of radiated power can be harvested for the exhaustive search scheme at the expense of a higher total transmit power .
in this paper , we studied the resource allocation algorithm design for comp multiuser communication systems with swipt .
the algorithm design was formulated as a non - convex combinatorial optimization problem with the objective to jointly minimize the total network transmit power and the maximum capacity consumption of the backhaul links .
the proposed problem formulation took into account qos requirements for communication reliability and power transfer .
a suboptimal iterative resource allocation algorithm was proposed for obtaining a locally optimal solution of the considered problem .
simulation results showed that the proposed suboptimal iterative resource allocation scheme performs close to the optimal exhaustive search scheme and provides a substantial reduction in backhaul capacity consumption compared to full cooperation .
besides , our results unveiled the potential power savings enabled by comp networks compared to centralized systems with multiple antennas co - located for swipt .
it can be verified that ( [ eqn : sdp_relaxation ] ) satisfies slater s constraint qualification and is jointly convex with respect to the optimization variables .
thus , strong duality holds and solving the dual problem is equivalent to solving the primal problem @xcite . for the dual problem ,
we need the lagrangian function of the primal problem in ( [ eqn : sdp_relaxation ] ) which is given by @xmath141 where@xmath142.\notag\end{aligned}\ ] ] here , @xmath143 denotes the collection of terms that only involve variables that are independent of @xmath101 .
@xmath144 is the dual variable matrix for constraint c8 .
@xmath145 , @xmath146 , @xmath147 , @xmath148 , @xmath149 , @xmath150 , and @xmath151 are the scalar dual variables for constraints c1c7 , respectively .
then , the dual problem of ( [ eqn : sdp_relaxation ] ) is given by@xmath152subject to @xmath153 .
for the sake of notational simplicity , we define @xmath154 and @xmath155 as the set of optimal primal and dual variables of ( [ eqn : sdp_relaxation ] ) , respectively .
now , we consider the following karush - kuhn - tucker ( kkt ) conditions which are useful in the proof : @xmath156 where @xmath157 is obtained by substituting the optimal dual variables @xmath158 into ( [ eqn : a_k ] ) .
@xmath159 in ( [ eqn : kkt - complementarity ] ) indicates that for @xmath160 , the columns of @xmath161 are in the null space of @xmath162 .
therefore , if @xmath163 , then the optimal beamforming matrix @xmath164 must be a rank - one matrix .
we now show by contradiction that @xmath165 is a positive definite matrix with probability one in order to reveal the structure of @xmath162 .
let us focus on the dual problem in ( [ eqn : dual ] ) .
for a given set of optimal dual variables , @xmath158 , power supply variables , @xmath166 , and auxiliary variable @xmath167 , the dual problem in ( [ eqn : dual ] ) can be written as @xmath168 suppose @xmath165 is not positive definite , then we can choose @xmath169 as one of the optimal solutions of ( [ eqn : dual2 ] ) , where @xmath170 is a scaling parameter and @xmath46 is the eigenvector corresponding to one of the non - positive eigenvalues of @xmath165 .
we substitute @xmath169 into ( [ eqn : dual2 ] ) which leads to @xmath171 on the other hand , since the channel vectors of @xmath172 and @xmath173 are assumed to be statistically independent , it follows that by setting @xmath174 , the term @xmath175 and the dual optimal value becomes unbounded from below . besides , the optimal value of the primal problem is non - negative for @xmath69
thus , strong duality does not hold which leads to a contradiction .
therefore , @xmath165 is a positive definite matrix with probability one , i.e. , @xmath176 . by exploiting ( [ eqn : lagrangian_gradient ] ) and a basic inequality for the rank of matrices , we have @xmath177 thus , @xmath178 is either @xmath179 or @xmath180 .
furthermore , @xmath181 is required to satisfy the minimum sinr requirement of ir @xmath35 in c1 for @xmath69 .
hence , @xmath163 and @xmath182 hold with probability one .
in other words , the optimal joint beamformer @xmath106 can be obtained by performing eigenvalue decomposition of @xmath161 and selecting the principal eigenvector as the beamformer .
d. w. k. ng , e. s. lo , and r. schober , `` wireless information and power transfer : energy efficiency optimization in ofdma systems , '' _ ieee trans .
wireless commun .
_ , vol . 12 ,
6352 6370 , dec .
d. w. k. ng , e. s. lo , and r. schober , `` robust beamforming for secure communication in systems with wireless information and power transfer , '' _ ieee trans .
wireless commun .
_ , vol .
pp , apr . 2014 .
d. lee , h. seo , b. clerckx , e. hardouin , d. mazzarese , s. nagata , and k. sayana , `` transmission and reception in lte - advanced : deployment scenarios and operational challenges , '' _ ieee commun . magazine _ , vol .
50 , pp . 148155 , feb .
r. irmer , h. droste , p. marsch , m. grieger , g. fettweis , s. brueck , h. p. mayer , l. thiele , and v. jungnickel , `` coordinated multipoint : concepts , performance , and field trial results , '' _ ieee commun . magazine _ , vol .
102111 , feb . 2011 .
d. w. k. ng , e. s. lo , and r. schober , `` energy - efficient resource allocation in multi - cell ofdma systems with limited backhaul capacity , '' _ ieee trans .
wireless commun .
_ , vol . 11 , pp
36183631 , oct . | this paper studies the resource allocation algorithm design for multiuser coordinated multipoint ( comp ) networks with simultaneous wireless information and power transfer ( swipt ) . in particular ,
remote radio heads ( rrhs ) are connected to a central processor ( cp ) via capacity - limited backhaul links to facilitate comp joint transmission . besides , the cp transfers energy to the rrhs for more efficient network operation .
the considered resource allocation algorithm design is formulated as a non - convex optimization problem with a minimum required signal - to - interference - plus - noise ratio ( sinr ) constraint at multiple information receivers and a minimum required power transfer constraint at the energy harvesting receivers . by optimizing the transmit beamforming vectors at the cp and energy sharing between the cp and the rrhs
, we aim at jointly minimizing the total network transmit power and the maximum capacity consumption per backhaul link .
the resulting non - convex optimization problem is np - hard . in light of the intractability of the problem
, we reformulate it by replacing the non - convex objective function with its convex hull , which enables the derivation of an efficient iterative resource allocation algorithm . in each iteration
, a non - convex optimization problem is solved by semi - definite programming ( sdp ) relaxation and the proposed iterative algorithm converges to a local optimal solution of the original problem .
simulation results illustrate that our proposed algorithm achieves a close - to - optimal performance and provides a significant reduction in backhaul capacity consumption compared to full cooperation . besides
, the considered comp network is shown to provide superior system performance as far as power consumption is concerned compared to a traditional system with multiple antennas co - located . |
in the present article the results of microscopic calculations for the high spin properties of the superdeformed ( sd ) rotational bands in the nucleus @xmath0 tb are presented .
this nucleus is a direct neighbor of the `` doubly magic '' superdeformed nucleus @xmath2dy which is characterized by particularly large gaps in the neutron and proton deformed single - particle spectra .
we investigate the rotational bands together with their microscopic structure in terms of the single - particle orbitals .
the theoretical description of such properties , most often applied in the literature , is based on the average field approximation ( see ref.@xcite for a recent review ) .
most of the present knowledge about the nucleonic structure underlying the properties of the sd rotational bands comes from analyzing the single - particle routhians generated by making use of the rotating nilsson or deformed woods - saxon ( ws ) hamiltonians .
the total nuclear energy as a function of spin is then obtained by using the strutinsky total energy expression ( see for example ref.@xcite and references therein ) .
similar analyses based on the hartree - fock ( hf)@xcite , or relativistic mean field ( rmf)@xcite approximations are more difficult and have not been , up to now , applied for global calculations , with the notable exception of ref.@xcite . however , they have an advantage of directly providing the total energy as a function of spin which can be compared with the experimental data . in the present work we apply and compare the ws and hf cranking approximations
. uncertainties in predictions coming from various models lead to ambiguities in the interpretation of the experimental results .
the problem of a ( non-)uniqueness in the interpretation of the microscopic structure of sd bands is strongly aggravated by the fact that the spins of nearly all measured bands in question are not known .
moreover , the nuclear interactions responsible for the single - particle level ordering in the strongly deformed nuclei are obtained mainly by extrapolating the properties of the interactions known from spherical and moderately deformed nuclei .
direct tests of the structure of wave functions of the single - nucleonic orbitals ( via e.g. magnetic moments ) are considered extremely difficult from the experimental point of view due to the very short life - times of the sd high - spin states .
in such a situation we are confronted not only with the problem of a microscopic interpretation of the quickly growing body of the experimental information but also with the problem of a cross - check analysis of theoretical predictions coming from distinct nuclear interactions used .
it will be one of our objectives to provide such a test by comparing the microscopic properties calculated within the strutinsky and hf methods . for a realization of our objective , the experimental knowledge of eight bands in one single sd nucleus , @xmath0 tb ,
is extremely useful . at present
there are only very few nuclei in which the number of known sd bands is comparable to that in @xmath0 tb .
moreover , the transition energies of those bands behave regularly as functions of the angular momentum @xmath3 .
such a regularity provides the optimal conditions for theoretical verifications of single - nucleonic configurations via the dynamical moments defined by @xmath4 this is so , because the abrupt band crossings ( relatively abundant in other nuclei ) may otherwise cause strong variations of @xmath1-moments as a function of @xmath5 , thus rendering the comparisons of the typical behavior ( and values ) of the @xmath1-moments with experiment less direct . in some cases
the band crossings may play an important role in helping to identify some orbitals via the characteristic crossing frequencies .
however , we found it preferable to first discuss the structural elements , such as the global behavior of the @xmath1-moments and their values , in relation with some characteristic nucleonic configurations .
another group of problems related to the band crossings , crossing frequency systematics and alignments is left for future investigations .
an advantage of basing the verification of the theoretical results with experiment on the @xmath1-moments is that this quantity does not depend on the exit spin ( the spin corresponding to the lowest observed transition in an sd band ) while many other quantities do .
for this reason we intend to focus on the rotational frequency rather than spin as an independent variable in function of which several microscopically calculated quantities will be displayed .
however , the question of the relative spins in the multi - band problem , or equivalently , of the relations between the exit spins in the bands observed , is one of the most urgent at present . for this reason
a special section will be devoted to discussing the total energy vs. spin dependence , @xmath6 .
let us mention here that , similarly to numerous other studies devoted to the interpretation of the intrinsic configurations of the sd bands , our work will not be able to give definite answers in several points .
we feel , nevertheless , that our method allows to pin down relatively reliably at least the possible alternative interpretations of the bands in question . while the lack of knowledge of the exit spins remains one of the most serious experimental problem ,
many analyses of the sd nuclei have benefited from the strong similarities between various bands in various sd nuclei ( the `` identical band '' correlations , cf .
ref.@xcite for a recent review ) . by comparing the theoretically predicted structures of the bands related by the `` identicality '' ( or `` sameness '' ) relations
, it was possible to confirm or eliminate certain hypotheses related to the underlying single - nucleonic structure .
this type of correlations are abundant in the @xmath0 tb nucleus and will be used to further test the microscopic results and predictions .
it is expected that some amount of pairing correlations remains in the sd bands up to the highest spins measured and that the pairing correlations decrease with increasing angular momentum .
although , as we will demonstrate it , in the @xmath0 tb nucleus the most important properties of the sd bands such as , e.g. , the behavior of the dynamical moments in function of the rotational frequency can be described reasonably well without taking into account the pairing , the systematic small discrepancies which exist , are increasing when the pairing correlations are taken into account .
moreover , the arguments will be given demonstrating that at least in the case of some parametrizations of the nuclear interactions another source of small systematic deviations between theory and experiment may be attributed to slightly exaggerated nuclear volumes as calculated within the mean field approaches ( cf .
sec.[sec5 ] for details ) .
a number of structural features related to the high - spin behavior of the sd nuclei in the a@xmath7150 region have been discussed in the literature .
early studies were based on the nilsson or ws cranking approaches , and performed without pairing@xcite or with the dynamic pairing correlation@xcite taken into account by using either the random phase approximation ( rpa ) or the particle number projection ( pnp ) together with the monopole pairing interaction . in most of those early studies the experimental information has been strongly limited , as compared to the amount of data available today . in more recent studies , the ws@xcite or the hartree - fock - bogolyubov (
hfb)@xcite methods have been applied to a few low - lying bands in a number of selected a@xmath7150 nuclei with pairing correlations treated in terms of the lipkin - nogami method .
the present study differs from the previously published ones in the following : ( i ) due to a rich experimental information available now , we can concentrate on systematic discrepancies between the theoretical and experimental data , ( ii ) we discuss the analogies and differences between the self - consistent ( hf ) and the non - self - consistent ( ws ) cranking approaches within the same analysis ( applied simultaneously on a common set of experimental data ) , ( iii ) we analyze simultaneously all available data in the nucleus in question , ( iv ) by applying the particle - number - projected pairing approach we are able to show that the systematic discrepancies can not be attributed to the pairing correlations , and consequently ( v ) we arrive at a suggestion that a modification of the underlying interactions will be needed in order to improve the agreement with data .
as mentioned earlier , in the @xmath0 tb nucleus eight sd bands have been observed .
in addition to the two sd bands discovered relatively early@xcite , six new sd bands have been found and assigned to this nucleus recently@xcite using the eurogam spectrometer .
those recent data were obtained from experiments at the nuclear structure facility , daresbury laboratory , using the @xmath8te(@xmath9al,6n)@xmath0 tb reaction at a bombarding energy of 154mev .
the details of the experimental procedure can be found in@xcite ; here we only recall that the @xmath10-rays were detected by the eurogam spectrometer@xcite which consisted , for this experiment , of 42 operational ge detectors .
the data rates in the individual ge detectors were 7 - 9 khz .
the trigger condition used to select unsuppressed events was that at least 7 ge detectors were involved in a coincidence . after the suppression requirement was included
, the average number of coincident detectors dropped to @xmath73.9 .
a total data set of 5.5@xmath11 suppressed events with ge fold @xmath12 3 was recorded and after unpacking the high - fold - multiplicity events led to 3.1 @xmath13 @xmath14 coincidences .
the final spectra were obtained by summing combinations of 3-dimensional gates set on the sd transitions in a 4-dimensional analysis@xcite .
the assignment of the new 6 excited sd bands to @xmath0 tb was based on an unambiguous evidence of known @xmath10-ray transitions of the normal deformed nucleus following the de - excitation from the sd band members .
the sd bands have been labeled 1,2,3,@xmath15,8 .
the previously reported sd bands have been labeled 1 ( called the yrast band , although its position in energy with respect to other bands is not experimentally established ) and 2 ( the band which shows a remarkable similarity in terms of the @xmath16-ray energies with the sd band 1 of @xmath2dy ) .
the other excited sd bands have been arbitrarily attributed labels ranging from 3 to 8 . employing the same reaction
, the @xmath0 tb nucleus has been recently reinvestigated using the eurogam ii array installed at the vivitron accelerator of the _ centre de recherches nuclaires _ in strasbourg .
this array consisted of 54 high - efficiency ge detectors ( 30 tapered coaxial detectors and 24 clover detectors ) providing a photopeak efficiency of 7.4% at 1.33mev . with the large statistics and high peak - to - background ratio , it has been possible to obtain a higher precision of the sd @xmath10-ray transition energies which in the present study are used to extract accurate values of the corresponding dynamic moments .
the experimental results for bands 1 and 2 are presented in table [ tab1 ] .
the intensities ( with uncertainties ) of the different bands relative to the yrast band are : 50(5)% ( band 2 ) , 35(5)% ( band 3 ) , 6(2)% ( band 4 ) , 10(2)% ( band 5 ) , 9(2)% ( band 6 ) , 11(2)% ( band 7 ) , 7(3)% ( band 8) .
let us also mention that in the determination of the experimental frequencies and dynamic moments we use the expressions [ eq300]@xmath17 where @xmath18=@xmath19@xmath20@xmath21 .
in this section we summarize the main points of the calculations based on the average nuclear field with the deformed ws potential . the following three subsections
present : ( i ) the structure of the hamiltonian used , sec.[sec3a ] , ( ii ) some remarks related to the self - consistency in the case of _ a priori _ non - self - consistent formalism with the ws potential , sec.[sec3b ] , and ( iii ) a short description of the actual variant of the strutinsky method applied , sec.[sec3c ] .
the details can be found in the literature cited below and will not be repeated here . even though most of the mathematical expressions are available in the literature
, there exist differences in notation between various authors , and we found important to recall the definitions of some basic quantities .
the single - nucleonic energies and wave functions for nucleons moving in the deformed potential of the ws type , @xmath22 } { \{1+\exp [ \mbox{dist}_{\sigma_{0 } } ( \bbox{r};r_{0 } ) /a]\ } } , \ ] ] are found by solving the corresponding schrdinger equation .
the spin - orbit term is included in the usual deformation - dependent form : @xmath23 } { \{1+\exp [ \mbox{dist}_{\sigma_{\text{so } } } ( \bbox{r};r_{\text{so } } ) /a_{\text{so}}]\ } } \right]\times{\bbox{p}}\right\}\cdot{\bbox{s } } , \end{aligned}\ ] ] and the coulomb potential is added for protons . in the above relations the `` + '' sign applies for protons and the `` @xmath20 '' sign for neutrons , while @xmath24 and @xmath25 denote the linear momentum and the spin operators , respectively . for reasons of compatibility with previous studies
, we keep the factor @xmath26}^{2}$ ] in the definition of the spin - orbit potential ( @xmath27 is the nucleonic mass ) .
then the spin - orbit strength parameter , @xmath28 , is a dimensionless quantity .
we use @xmath29 for neutrons and @xmath30 for protons .
the nuclear surfaces @xmath31 and @xmath32 , defining the central and spin - orbit fields , respectively , are expressed in terms of the spherical harmonics as @xmath33.\ ] ] in the above relation , @xmath34 is a function of all the deformation parameters that guarantees the constant volume for any deformation , and with the central - potential radius parameter @xmath35fm for neutrons and @xmath36fm for protons ; similarly @xmath37fm for neutrons and @xmath38fm for protons .
the central - potential depth - parameters are : @xmath39mev and @xmath40 while the diffuseness parameters are @xmath41fm for the central part and @xmath42fm for the spin - obit part of the potential , independently of the isospin ( for details see refs.@xcite and@xcite ) .
finally , the geometrical distance of a point @xmath43 from the auxiliary surfaces @xmath44 defined above is denoted by @xmath45 ; the latter quantity is negative for @xmath43 in the nuclear interior and positive otherwise . in our calculations
we allow for the total routhian ( for definition of the routhians see the following sections ) minimization with the deformation - mesh defined by @xmath46 and @xmath47 with @xmath48 corresponding to 5 `` planes '' for fixed @xmath10 deformations [ @xmath49 = @xmath50@xmath51(@xmath10 ) and @xmath48 = @xmath52 @xmath50 @xmath53(@xmath10 ) ] as follows @xmath54 we solve the cranking equations of the standard form @xmath55 for @xmath56 with @xmath57 where the ws potential @xmath58 and the related spin - orbit term @xmath59 are defined in eqs.([eq31 ] ) and ( [ eq32 ] ) while the coulomb potential @xmath60 is that of the classical uniform sharp - edge charge distribution corresponding to the actual nuclear shape . as always in numerical calculations of the matrix elements used for constructing the hamiltonian matrix , both the number of basis states and the number of points in the gauss - hermite integration formulae
have been selected in such a way that the final results do not depend in any significant way of that choice ( for more details see ref.@xcite ) . in our case
the number of the deformed harmonic - oscillator states used in each of the two parities was about 300 . in the following ,
the results of the cranking calculations performed within the hf approach will be compared with those obtained for the deformed ws potential .
although it is not possible to formulate the self - consistency concept in the two approaches in the same way , and the ws calculations are _ a priori _ considered non - self - consistent , it will be of some importance to recall that our way of defining the deformed ws potential imposes a correlation between the equipotential surfaces @xmath61 and the surfaces of the constant density defined by @xmath62 as discussed in detail in@xcite .
the second - order thomas - fermi expression@xcite for the density functional reads @xmath63}^{3/2 } \\ & \times & \left\{1-\frac{1}{8}\frac{{\hbar}^{2}}{2 m } \left[\frac{\bbox{\nabla } v(\bbox{r } ) \cdot \bbox{\nabla } v(\bbox{r } ) } { 4{[\lambda - v(\bbox{r})]}^{3}}\right ] + \frac{\delta v(\bbox{r } ) } { { [ \lambda - v(\bbox{r})]}^{2 } } \right\ } , \nonumber \end{aligned}\ ] ] where the terms of the order @xmath64 have been neglected . requiring that the potential is equal to a constant , cf .
eq.([eq39 ] ) , implies by virtue of eq.([eq31 ] ) that @xmath65 using the fact that @xmath66 and that @xmath67 for @xmath43 belonging to the equipotential surface , one obtains after elementary transformations , cf .
ref.@xcite , that also @xmath68 as a result , at least within the accuracy of @xmath64 in the thomas - fermi approximation , condition ( [ eq39 ] ) implies condition ( [ eq39a ] ) . such a correlation between the geometry of the potential and that of the density is not valid generally
neither the nilsson model nor a deformed ws potential with another type of argument in the exponent of eq.([eq31 ] ) , as used sometimes by other authors , gives such a correlation .
obviously , the spin - orbit term may modify the above result ( cf .
e.g. ref.@xcite ) , but since this term has , on the average , matrix elements which are nearly an order of magnitude weaker as compared to those of the central part , its effect in the discussed context is also expected to be relatively weak . anticipating some of our results presented below , and in particular surprising similarities between the high - spin behavior properties obtained by using the self - consistent hf method and a non - self - consistent approach with the deformed ws hamiltonian , one should bear in mind the particular property of definition ( [ eq31 ] ) used in this study .
there are numerous presentations of the strutinsky method , as given for example in the original articles@xcite , and then applied to finite potentials ( as e.g. in ref.@xcite ) . here
we closely follow the presentation given in ref.@xcite , and we only briefly discuss the total - routhian - surface calculations , for which several authors use prescriptions slightly differing in details .
our calculations of the total nuclear routhians are based on the expression @xmath69 where the summation extends over all the occupied nucleonic states ( i.e. , in particular takes into account the particle - hole excitations with respect to the reference configuration introduced in sec.[sec5 ] ) .
the microscopic ( or shell ) energy is defined as usual by @xmath70 where the summation over @xmath71 extends now over all the single - nucleonic states of the reference configuration .
the last term denotes the results of the strutinsky averaging procedure .
this averaging depends on two standard parameters : the smoothing - polynomial order @xmath72 and the averaging width @xmath73 . in our applications we have @xmath74 and @xmath75 with @xmath76mev .
arguments related to the stability of the final result with respect to ( small ) uncertainties in these parameters can be found in several publications , see e.g. ref.@xcite and references therein .
in the following sections , some details of a particular realization of the hf approach in our study will be described ; most of the formal development can be found in the literature and are also summarized in ref.@xcite .
section [ sec4a ] presents some technical aspects related to the harmonic oscillator basis employed , the numerical calculations of the @xmath1-moments are discussed in sec.[sec4b ] , and the use of the quadrupole constraint to stabilize the iteration process is briefly commented in sec.[sec4c ] . in the present study we have performed the hf calculations of sd rotational bands using the skyrme effective interaction . for this purpose
a new numerical code hfodd has been constructed@xcite which uses the three - dimensional cartesian deformed harmonic oscillator ( ho ) basis .
the ho basis has up to now been used in solving the hf equations with the gogny interaction@xcite . on the other hand , for the skyrme interaction
this method has been used at early stages of investigations@xcite and then abandoned in favor of the calculations in spatial coordinates@xcite .
the main advantage of using the spatial coordinates consists in the fact that one may use the same spatial grid of points to treat nuclei of different deformations , while in the former method the basis parameters may need to be adjusted with varying deformation .
( the sensitivity of the results with respect to this kind of a dependence decreases with an increasing harmonic - oscillator basis cut - off ; in principle the problem disappears in the limit of an infinite basis ) .
however , the sd states have very similar deformations in many neighboring nuclei , and in such a case this particular deficiency disappears .
we may then profit from the fact that the calculations which use the ho basis are in general much more stable , have better convergence properties , and hence require less computational effort .
the details concerning the hfodd code will be presented elsewhere@xcite ; here we only give a few of its basic characteristics pertaining to the present application .
the calculations have been performed using a fixed basis given by the ho frequencies @xmath77=11.200mev and @xmath78=6.246mev in the directions perpendicular and parallel to the harmonic - oscillator symmetry - axis , respectively .
these values have been obtained by standard prescriptions developed for diagonalizing the deformed ws hamiltonian@xcite in the ho basis , and correspond to the ws potential with the deformations @xmath79=0.61 and @xmath80=0.10 .
the basis has been restricted to the fixed number @xmath81 of ho states , i.e. , keeping the lowest @xmath81 single - particle energies @xmath82=@xmath83+@xmath84+@xmath85+@xmath86)@xmath78 .
the actual calculations have been performed with @xmath81=306 .
this corresponds to the maximal numbers of oscillator quanta equal to 8 and 15 in the perpendicular and parallel directions , respectively .
the stability of results with respect to increasing the size of the ho basis is the main concern in all analyses using this technique , and have been studied in detail in ref.@xcite .
in fact it is well known that the total energies are very slowly convergent as functions of the basis size .
indeed , in the present calculations a similar effect is seen when we increase the basis from @xmath81=306 to @xmath81=604 , which corresponds to reaching the maximal numbers of 11 and 20 ho quanta in the two main directions .
then the total energy of the yrast configuration of @xmath0 tb at the rotational frequency of @xmath87=0.6mev decreases from @xmath201199.322 to @xmath201201.889mev . a similar mechanism is well known also from other type of studies , with the schrdinger equation diagonalized in a given basis ( as for instance with the ws or the folded yukawa - type potentials ) .
fortunately the relative ( excitation ) energies become stable for smaller bases than the absolute values@xcite .
none of the interesting rotational properties depends in a significant way on an increase of the basis from @xmath81=306 to @xmath81=604 .
for example , we obtain the changes of the average angular momentum from 59.95 to 59.99@xmath88 , of the mass quadrupole moment from 39.82 to 39.90b , and of the mass hexadecapole moment@xcite from 4.21 to 4.24b@xmath89 .
the relative changes as function of @xmath87 are even smaller .
most interestingly , the @xmath1-moment changes at the same time only from 84.71 to 84.84@xmath90mev@xmath91 . in view of the fact that the calculations for @xmath81=604 are a factor of five slower than those for @xmath81=306 , we consider it reasonable to keep the latter size of the basis in all calculations presented .
an important advantage of using the ho basis consists in the fact that a rather small number of iterations is sufficient to obtain a high level of convergence .
for example , in the case of the yrast configuration in @xmath0 tb at @xmath87=0.6mev , starting from the rotating ws state and performing 50 iterations is enough to obtain the total energy accurate up to 0.5kev .
for all values of rotational frequency and with the same number of iterations performed , the relative precision remains comparable to the one cited above .
this is very important for the calculation of the second moment of inertia @xmath1 .
this quantity has to be calculated by a numerical differentiation , either of the average angular momentum @xmath3 @xcite , @xmath92 with respect to the rotational frequency @xmath5 , eq.([eq11 ] ) , or of the total energy @xmath93 , @xmath94 with respect to @xmath5 : @xmath95 where the cranking wave functions @xmath96 are obtained from the hf variational equation @xmath97 in order to obtain accurate values for @xmath1 , the relative errors of @xmath3 and @xmath93 as functions of @xmath5 should be kept as small as possible .
the difference of values of @xmath1 calculated from eqs.([eq11 ] ) and ( [ req12 ] ) is a measure of self - consistency of solutions . in our example
, the previously quoted value of @xmath98=84.71@xmath90mev@xmath91 has been obtained from eq.([eq11 ] ) , while at the same time eq.([req12 ] ) gives @xmath1=84.66@xmath90mev@xmath91 .
these results have been obtained from the first - order finite - difference approximation of derivatives between @xmath87=0.60 and 0.61mev .
together with the discussed uncertainties related to the restricted size of the basis , these values indicate that our calculated dynamical moments are accurate to about 0.1@xmath90mev@xmath91 , which is generally a higher precision than that presently available for the experimental values of @xmath1 .
the sd states are localized in the pocket of the potential energy surface which is several mev above the main minimum .
when the hf equations are solved by the standard iteration procedure@xcite , which we also use in our calculations , the wave functions may become gradually polluted by components of the solutions residing principally in the main well , and this may slow - down or even preclude any convergence to the secondary minimum . in order to prevent this kind of effects we have included in the minimized functional a quadratic constraint on the quadrupole moment@xcite , i.e. , we minimize @xmath99 with a small stiffness parameter @xmath100=0.01mev / b@xmath89 and the constraint value of @xmath101=42b .
this is a very weak constraint which has no visible effect on the self - consistent values of the quadrupole moment .
it shifts , however , the normal - deformed - minimum energy sufficiently high , so that it does not present any undesired competition with the sd one .
the quadrupole constraint is at the origin of the second term in eq.([req12 ] ) , which compensates for the fact that the quadratic constraint adds to the functional a term which depends on the rotational frequency .
the magnitude of this second term depends on the variations of the average quadrupole moment @xmath102 along the rotational band ( see sec.[sec5 ] ) and on the difference between @xmath103 and the target value @xmath101 . in the present calculations this term can be for some bands as large as 0.5@xmath90mev@xmath91 , and , therefore ,
can not be neglected when comparing the values of @xmath1 calculated from eqs.([eq11 ] ) and ( [ req12 ] ) .
in this section we present the results of the hf calculations of the multi - band excitation structure of the @xmath0 tb nucleus .
in several cases the cross - check results based on the deformed potential of the ws form will also be presented .
we divide this discussion into several steps : secs.[sec5a ] and [ sec5c ] are devoted to the analysis of the rotational bands in terms of the single - particle routhians , total energies , and dynamical moments @xmath1 , while the next ones present problems related to exit spins , shape changes along the rotational bands , and pairing . most probably , only the lowest energy sd bands are populated in the present - day experiments on the sd nuclei .
more precisely , the bands lying not too high in the energy scale , at spins between , roughly , 60@xmath88 and 80@xmath88 , can actually be detected .
unfortunately not much more can be said about the population selectivity and no rules for , for instance , the population probability as a function of the actual configuration have been established . under these circumstances
the only way of performing a systematic theoretical analysis consists in calculating numerous bands in excess of the number of those actually measured and in proceeding with a more detailed and critical selection analysis next . as the first step the single - particle spectra of the deformed ws cranking hamiltonian , cf .
sec.[sec3 ] , have been calculated as functions of the cranking frequency for @xmath87 ranging from 0 to 1mev for a few points corresponding to the typical quadrupole @xmath49 @xmath70.58 and the hexadecapole @xmath105 @xmath70.10 deformations known to represent well , on the average , the sd nuclei in the @xmath106@xmath7150 mass range .
knowing those levels , we define the reference configuration that corresponds to the lowest - level occupation scheme for possibly largest @xmath87 range .
the symbols of the particle - hole configurations used below refer to such a specific reference configuration as the underlying structure .
it is worth mentioning that often more than one configuration could quite well serve as the reference and that the actual choice of such a configuration is a matter of commodity .
in fact , an inconvenient choice of the reference would merely result in a more complicated labeling of the excited bands : a 1-particle 1-hole excitation with respect to a given reference configuration may become 2-particles 2-holes , 3-particles 3-holes etc .
excitation with respect to the inconveniently chosen reference . in the present analysis
we consider hamiltonians having two discrete symmetries : the parity @xmath107 and the @xmath108-signature @xmath109 @xmath110 which has the eigenvalues @xmath111=@xmath112 or @xmath111=@xmath113 .
the reference of our choice is defined in terms of four types of the occupation numbers : @xmath114 , @xmath115 , @xmath116 and @xmath117 referring to positive - parity and signature @xmath111=@xmath112 , positive - parity and signature @xmath111=@xmath113 , etc .
, respectively .
we have two sets of such numbers for protons and neutrons each . in @xmath0
tb the self - consistent solution for the reference configuration is defined by occupying the single nucleonic orbitals according to the following scheme @xmath118 in fig.[fig5.1 ] we show the single - particle routhians for protons and neutrons obtained with the skm*@xcite parametrization of the skyrme interaction for the reference configuration given in eq.([eq41 ] ) .
all hf calculations are in the present study performed using the complete skyrme functionals , cf
. discussion in ref.@xcite .
let us remark that the interpretation of the experimental sd @xmath0tb(1 ) yrast band in terms of the configuration implied by relations ( [ eq41 ] ) corresponds to the occupation of all the lowest - lying proton routhians with the last orbital occupied @xmath119[6,5,1]3/2(@xmath111=@xmath113 ) ( at large @xmath5 taking another label : @xmath119[6,4,2]5/2(@xmath111=@xmath113 ) , cf .
fig.[fig5.1 ] ) .
this reference configuration has the lowest calculated total energy at a fixed spin , and therefore , from the theoretical point of view , we can call it the yrast sd configuration in this nucleus .
configuration assignment ( [ eq41 ] ) is confirmed by experimentally observed band similarities between @xmath0tb(1 ) and @xmath120gd(2 ) , whose several transitions correspond as being nearly identical .
indeed , this observation can be viewed as a direct consequence of exciting the last occupied proton orbital , @xmath119[3,0,1]1/2(@xmath111=@xmath112 ) in @xmath120gd(2 ) , into @xmath119[6,5,1]3/2(@xmath111=@xmath113 ) , the last occupied proton level in @xmath0tb(1 ) .
since the alignment properties and the @xmath98-contributions originating from the @xmath119[3,0,1]1/2(@xmath111=@xmath112 ) orbital are known to be negligible , refs.@xcite , the assignments of both configurations in the two nuclei can be viewed as confirmed . in order to establish the uncertainty margin due to various parametrizations of the interactions that have been used in the literature , the results analogous to those in fig.[fig5.1 ] but for the siii parametrization of ref.@xcite
are presented in fig.[fig5.2 ] .
one can see that the characteristic gap structures for the protons ( in the form of the sequence of gaps at @xmath121 , @xmath122 , and @xmath123 visible in figs.[fig5.1 ] and [ fig5.2 ] ) are very similar in both parametrizations .
these gap structures are defined by very similar positions of the @xmath119@xmath1243/2 $ ] and @xmath119@xmath1251/2 $ ] orbitals .
other proton orbitals , lying above and below those gaps , differ only slightly in the level ordering .
the differences at @xmath126 in the level positions do not exceed typically @xmath7500 kev
. it will be instructive at this point to compare the single - particle spectra obtained by using the hf method with the analogous results obtained by making use of the deformed ws cranking hamiltonian .
for this purpose the three - dimensional total - routhian surfaces for the @xmath0 tb have been calculated in function of the quadrupole ( @xmath50,@xmath10 ) and hexadecapole ( @xmath127 ) deformations , cf .
sec.[sec3a ] .
the total - routhian surfaces have been calculated in function of the rotational frequency and the minimum - deformation path composed of sets of ( @xmath50,@xmath10,@xmath127)-deformation parameters that vary with increasing @xmath128 has been found . the deformed ws cranking single - particle spectra obtained in this way
are presented in fig.[fig5.3 ] .
let us observe by comparing figs.[fig5.1 ] , [ fig5.2 ] , and [ fig5.3 ] , that the global features of the single - nucleon spectra obtained by making use of the deformed average field model with the ws hamiltonian and those of the hf with the skm * and siii interactions are quite similar .
what is perhaps most striking is the fact that the ws hamiltonian and the hf method with the siii parametrization give very similar single - nucleonic levels indeed .
the similarities are particularly well visible in the neutron spectra in figs.[fig5.1 ] and [ fig5.3 ] , and only to a slightly lesser extent in the proton spectra displayed in the figures .
this observation can be interpreted in favor of both techniques that employ the two different average fields thus increasing our confidence in the implied results of the microscopic calculations .
since the exit spins of most sd bands are not known , at present the only possible way of testing the calculated @xmath93 versus @xmath3 relation by experiment consists in comparing the resulting @xmath1-moments . before proceeding systematically with such comparisons
some additional comments will be useful .
we begin by discussing the ws calculations which have been performed here using the so - called `` universal '' parameters .
the corresponding parameter set has been introduced in ref.@xcite , and later tested in many detailed calculations of the high - spin properties in nuclei .
the corresponding parametrization of the central part of the ws potential employs the radius - parameter values of @xmath129(@xmath71 ) = @xmath130fm and @xmath129(@xmath119 ) = @xmath131fm for the neutrons and protons , respectively .
historically , these parameter values have been fitted to improve the quality of the single - particle level spectra rather than to reproduce well the geometrical properties of nuclei , and it happens that they exceed considerably the more commonly accepted `` geometrical '' value of @xmath129@xmath132@xmath133fm .
one of the direct implications of the above difference is that the kinematical moments of inertia @xmath134 , @xmath135 calculated from , e.g. , the fermi - gas model with the geometry defined by using the above radius - parameter values systematically exceed the rigid - body moments of inertia calculated from the classical formula with the standard nuclear matter density . within a classical but also within a semiclassical model , the moments of inertia simply scale with the square of the radius parameter , @xmath136 . as a result one should expect that the moments of inertia in various bands , calculated by using the `` universal '' parametrization of the ws hamiltonian will be , on the average , too large .
we can see that such an expectation , valid for the kinematical moments will apply directly to the dynamical moments ( [ eq11 ] ) , @xmath137 and a scaling of the @xmath134-moments implies a similar scaling of the @xmath1 ones . by scaling the calculated @xmath134 and @xmath98-moments we should , as a consequence , also scale the related physical quantities : rotational frequency , energies and routhians .
one can see this after noticing that the scaling of the @xmath134 moment with a factor @xmath138 , that satisfies relation ( [ eq42 ] ) , for the quantized angular momenta @xmath139 ( which remain independent of any scaling ) , implies the necessity of scaling the rotational ( cranking ) frequency by a factor of @xmath140 .
it then follows that the energies and routhians also scale as @xmath140 , so as to consistently satisfy the canonical relation @xmath141 .
the results of the calculations of the @xmath1 moment for the reference ( in this case coinciding with the yrast ) configuration of @xmath0 tb are given in fig.[fig5.4 ] , where the ws and hf curves are compared with experimental data .
top part of the figure gives results with no scaling applied .
one can see , that the calculated values systematically exceed the experimental ones .
in the ws case we know that , as mentioned above , the numerical values of the radius - parameters are too large and we can clearly consider at least a part of the discrepancy as related to `` too large a volume of the nucleus '' .
therefore , in the bottom part of fig.[fig5.4 ] we show the same @xmath1-moments scaled by the factor @xmath138=0.9 . in the hf case
there is no obvious discrepancy between the calculated and experimental nuclear radii .
nevertheless , the calculated @xmath1-moments as well as quadrupole moments are systematically slightly too large ( see refs.@xcite for recent measurements of the quadrupole moments and ref.@xcite for the results of systematic hf calculations of the quadrupole moments and comparison with experiment ) .
such an overestimation is consistently obtained for many different skyrme force parameters@xcite . at present
, no explanation of this systematic effect can be found . for the sake of the presentation we have therefore decided to use the same phenomenological scaling factor @xmath138=0.9 to present the results of hf calculations .
this scaling procedure , as we have checked by an explicit comparison with the experimental results , works well for all the bands in the @xmath0 tb nucleus simultaneously .
moreover , we have checked that by using the same scaling factor @xmath138=0.9 for many bands and for several nuclei simultaneously one reproduces well the experimental @xmath1 moments which otherwise remain too large .
this conclusion is based on comparisons for the yrast bands in @xmath0dy , @xmath2dy , @xmath142dy , @xmath143 tb , and @xmath120 tb , and also for a few dozens of excited sd bands in the mass @xmath106@xmath7150 region .
light gadolinum nuclei ( @xmath144gd and @xmath145gd ) have strong alignment effects in their yrast sd bands and are not well suited for the tests discussed here .
slopes of the @xmath1 moments in @xmath146gd , @xmath143gd , and @xmath120gd are overestimated in the calculations ; for the discussion of the latter nucleus see sec[sec5b ] .
conversely , not introducing the aforementioned scaling we find the calculated values systematically too high , by 8 to 10% .
after applying the scaling procedure according to the discussion above , ( cf .
fig.[fig5.4 ] for illustration ) , the @xmath98-moments are multiplied by the factor of @xmath138 and the angular frequencies by a factor of @xmath147 . then , a very good correspondence with the @xmath5 dependence of the experimental data is worth noticing .
below we present all results scaled according to the prescription and parameters defined above .
let us mention that no physical conclusion of this paper depends on using or not using the scaling procedure described above ; the scaling can be viewed as merely a `` graphical '' trick to facilitate a comparison with the experimental data .
moreover , we have not attempted any detailed fit of the scaling factor @xmath138=0.9 which can still be varied within about @xmath148 without any obvious general deterioration of the agreement with data .
the problem of the definition of the yrast configuration discussed in sec.[sec5a1 ] did not pose any serious difficulty .
firstly , in nuclei with the nucleon numbers in the vicinity of large sd gaps ( as e.g. @xmath149=@xmath150 and @xmath151=@xmath152 ) the yrast configurations corresponding to the occupation of the lowest - lying levels are relatively easily and unambiguously determined over a relatively long @xmath5-range since no , or very few level crossings occur at the fermi level ( cf .
fig.[fig5.1 ] ) .
secondly , the experimentally dominating sd band ( i.e. , the one with the highest population intensity ) is also easily distinguished . in the case of the other sd bands neither of the observations applies . in experiment
several bands are populated with comparable ( usually weak ) intensities and their relative excitation energies are totally unknown . in theory ,
numerous close lying bands remain _ a priori _ the candidates for a much smaller number of the experimental bands to be associated with .
moreover , the uncertainties in the relative positions of the theoretically calculated levels which sometimes influence the level ordering may reach 500kev or more .
this estimate corresponds to a typical discrepancy between the results obtained within commonly used average field models ( like ws potential or skyrme interaction parametrizations , skm * and siii as discussed above and illustrated in figs.[fig5.1 ] , [ fig5.2 ] and [ fig5.3 ] ) and may , occasionally , become markedly larger . consequently , in analyzing the structure of the nucleus in question we have applied the following approach : \(i ) our primary goal in this section is to recognize within the applied formalism the most probable single - nucleonic ( particle - hole ) structure that underlies the behavior of the experimentally found bands .
\(ii ) we are using , for the sake of _ qualitative _ argumentation , the calculated ordering of various sd bands : the lower the band energy at the _ high - spin limit _ , the higher the chance to find such a band among those populated in experiment .
\(iii ) we use the experimental @xmath1-moments and verify the possible single - nucleonic configurations by directly comparing @xmath1(@xmath5 ) with experiment . another important piece of information suitable for identifying single - particle configurations is the sameness of bands in different nuclei , i.e. , the equality of the corresponding dynamical moments and specific relations which occur for the relative alignments , see the review in ref.@xcite . in the present work we mainly use and analyze the properties of the dynamical moments , because those of the relative alignments are very sensitive to the details of the effective interactions@xcite . a readjustment of the interaction parameters , especially those related to time - odd components of the mean field , seems to be necessary for a precise theoretical determination of the relative alignments . in order to apply the guidelines ( ii ) and ( iii ) above
, we first present the results for the lowest - lying one - particle one - hole excited bands . in figs.[fig5.5 ] and
[ fig5.6 ] we show the total energies for the proton and neutron excitations , respectively . here and in the following we label the configurations by the corresponding nilsson labels of hole and particle states , as well as by the total parity and signature quantum numbers supplemented by ( arbitrary ) consecutive number attached to each configuration .
since in @xmath0 tb the size of the proton shell gap is smaller than that for the neutrons , there are more low - lying proton than neutron excited bands .
since we know that there exist uncertainties in the single - particle level positions that may influence the relative positions of some bands in a non - negligible manner , the results in figs.[fig5.5 ] and [ fig5.6 ] should be treated as a semi - quantitative guide .
the figures show 10 proton particle - hole and 12 neutron particle - hole excited bands , respectively , that appear to be the lowest at @xmath153 , i.e. , in the spin range where the superdeformed bands are populated . we have placed some of the curves in the upper parts , and some others in the lower parts of the figures trying to illustrate an existence of `` families '' of bands , each family characterized by some specific features like slopes or band crossings . in the case of neutrons ,
the separation into two families is particularly well visible .
the differences in this case ( cf . also figs .
[ fig5.7 ] and [ fig5.8 ] ) correspond to the structure differences in the two characteristic hole - orbitals ( @xmath1543/2(r$]=@xmath155 or @xmath1543/2(r$]=@xmath156 , top parts , and @xmath1573/2(r$]=@xmath156 bottom parts ) .
figures [ fig5.7 ] and [ fig5.8 ] present the theoretical predictions for the dynamical moments corresponding to the bands shown in figs .
[ fig5.5 ] and [ fig5.6 ] , respectively .
one can see that the theoretically calculated moments form `` families '' characterized by a similar behavior also in the proton case .
in particular the bands built on the configurations of the type @xmath158(r$]=@xmath159 @xmath160 @xmath161(r$]=@xmath155 and @xmath1623/2(r$]=@xmath156 @xmath160 @xmath161(r$]=@xmath155 are characterized by similar slopes , in terms of the @xmath1 moments ; these slopes are different from that of the reference band . on the neutron side ,
the similarities between the bands related to exciting the @xmath1573/2(r$]=@xmath156 orbital into the orbitals @xmath1639/2(r$]=@xmath159 , @xmath1643/2(r$]=@xmath159 , and @xmath1655/2(r$]=@xmath159 deserve noticing . distinguishing among these properties will be particularly useful when attributing the configurations to the experimental bands since , as mentioned before , several theoretical configurations provide similar @xmath1 moment dependence on the rotational frequency , while their energies are also quite close .
we proceed with the interpretation of the experimental results for bands 2 , 3 , and 4 .
although the most likely structures of these bands have already been discussed before in the literature@xcite , we found it instructive to illustrate the procedure of attributing the theoretical bands to the experimental ones , the procedure which will be of more importance for other bands for which an adequate interpretation remains less easy . by eliminating the bands
whose @xmath1-moments differ relatively strongly with respect to the experimental result for bands 2 - 4 , we arrive at a comparison presented in fig.[fig5.9 ] . here and in the following we denote the hf states by their dominant nilsson labels at _ high frequencies _ ,
fig.[fig5.1 ] .
for example , the above mentioned @xmath1611/2(r$]=@xmath155 routhian originates at low frequencies from the @xmath1623/2(r$]=@xmath155 state .
experimental results for bands 2 and 4 agree rather well with calculations for configurations based on exciting either of the signature partners of the @xmath1581/2(r$]=@xmath159 orbitals into the @xmath1611/2(r$]=@xmath155 state . as seen in fig.[fig5.5 ]
, the configuration based on the @xmath1581/2(r$]=@xmath112 ) routhian has the energy lower by about 1mev than that based on the corresponding signature partner , and therefore the former should be retained as the most likely interpretation of band 2 .
similarly , the dynamical moment of the @xmath1623/2(r$]=@xmath156 @xmath160 @xmath1611/2(r$]=@xmath155 configuration , fig.[fig5.9 ] , middle part , reproduces very well the data for band 3 , and this supports the corresponding interpretation ( proposed also in ref.@xcite ) .
the suggestion of this configuration is based simultaneously on the fact that it corresponds to a low - energy excitation , as it is seen from fig.[fig5.5 ] , and that it provides a good agreement of the @xmath1 versus @xmath5 dependence with the experimental one . on the other hand , as seen in figs.[fig5.1 ] , [ fig5.2 ] , and [ fig5.3 ] , all calculations give at high frequencies the low - lying routhian @xmath1691/2(r$]=@xmath155 that might produce configurations which are lower in energy , cf .
fig.[fig5.5 ] , bottom part , than those based on excitations to the @xmath1611/2(r$]=@xmath155 state ( cf .
also fig.[fig5.7 ] , bottom part ) .
the experimental results for bands @xmath0tb(5 ) and @xmath0tb(6 ) indicate a correlation with the yrast sequence of the isotope @xmath120tb@xcite .
this suggests that the neutron routhian @xmath71[7,7,0]1/2(@xmath111=@xmath113 ) , cf .
figs.[fig5.1 ] and [ fig5.3 ] , is not occupied in the related configurations .
[ due to the level interactions at large @xmath5-values the label of this level changes into @xmath71[7,6,1]3/2(@xmath111=@xmath113 ) ] .
indeed , calculations indicate that several neutron particle - hole excitations give the required dependence of @xmath1 versus @xmath5 , as seen in fig.[fig5.8 ] .
these are excitations from the @xmath71[7,6,1]3/2(@xmath111=@xmath113 ) orbital to either of the @xmath71[4,0,2]5/2(@xmath111=@xmath172 ) , @xmath71[5,2,1]3/2(@xmath111=@xmath172 ) , or @xmath71[5,1,4]9/2(@xmath111=@xmath172 ) orbitals .
a comparison of the dynamical moments corresponding to the configuration @xmath71[7,6,1]3/2(@xmath111=@xmath113 ) @xmath173 @xmath71[5,2,1]3/2(@xmath111=@xmath172 ) is given in fig.[fig5.10 ] as an example , while the excitation energies are presented in fig.[fig5.6 ] .
one can see that the energies of the latter configurations are @xmath7(600 to 800 ) kev higher than those of the former ones , thus favoring the involvement of the neutron @xmath71[4,0,2]5/2 and @xmath71[5,2,1]3/2 orbitals as a more likely explanation .
the experimental results for bands @xmath0tb(7 ) and @xmath0tb(8 ) indicate a correlation with the yrast band in the @xmath143gd nucleus@xcite .
such a correlation suggests that the last proton intruder level , @xmath119[6,4,2]5/2(@xmath111=@xmath113 ) , remains unoccupied with the corresponding proton on a low lying orbital with a small @xmath1-contribution .
in fact , the low - lying @xmath119[5,3,0]1/2(@xmath111=@xmath112 ) orbital visible , e.g. , in fig.[fig5.1 ] provides an excellent result , cf .
fig.[fig5.11 ] .
however , since a pair of degenerate bands is seen in the experiment , one should also consider some other possibilities .
in fact , several neutron configurations mentioned in the preceding section remain equally good candidates , and in fig.[fig5.11 ] we also present a comparison with the experiment of the results obtained for the @xmath71[7,6,1]3/2(@xmath111=@xmath113 ) @xmath160 @xmath71[4,0,2]5/2(@xmath111=@xmath172 ) excitations .
it is important to observe that the similarity between the @xmath98-moments in @xmath143gd(1 ) , and those in the @xmath0tb(7,8 ) bands is accompanied by noticeable differences in both the @xmath176 and @xmath177 moments calculated for the bands with the configurations mentioned .
for instance the proton excited configuration carries about 0.5b more than the neutron one : @xmath178 $ ] @xmath179 @xmath180 $ ] , as it is seen in fig.[fig5.11 ] . for the bands @xmath0tb(7 ) and @xmath0tb(8 ) we thus have three possible configurations which give almost indistinguishable dynamical moments .
the importance of the above example lies in the fact that the configurations for rather different quadrupole moments and/or deformations may still lead to very similar dynamical moments .
therefore , a direct association of `` identical '' quadrupole moments and `` identical '' dynamical moments is not supported by microscopic calculations . the bands having identical dynamical moments
may still carry quite different deformations and/or quadrupole moments .
the problem of the exit spins in sd bands is one of the most actual in the present - day studies of the nuclear superdeformation .
many questions remain half - answered since the connections between the sd bands and the rest of the nuclear decay schemes are not known in the @xmath106@xmath7150 nuclei from the experimental point of view .
this makes it impossible to measure , e.g. , the absolute alignments or @xmath134-moments , and to estimate the degree to which the rapid nuclear collective rotation resembles that of a rigid body .
such questions are being posed also in relation to the pairing behavior at high spins and in many other contexts .
our calculations allow , in principle , to predict the absolute ( and thus also the relative ) exit spins of the sd bands studied . in practice , however , there are some deficiencies of ( all ) the methods used which impose limitations on the predictive power related to these particular results of calculations . first of all , in our hf calculations the pairing correlations have been neglected and the obtained results concerning the alignment effect may introduce at least small uncertainties . although it is often believed that the spins calculated at a given cranking frequency @xmath5 must be lower
when the pairing correlations are included , the actual tendency will depend on the position of the fermi level relative to the occupation of the large-@xmath181 shells : the down - sloping ( beginning of the shell ) and the up - sloping ( upper part of the shell ) orbitals contribute to the total alignment in opposite ways .
several aspects related to pairing will be presented in sec.[sec5b ] .
we have analyzed the calculated spin versus frequency plots for all the bands studied in this article .
the experimental spins of the last levels seen experimentally are not known .
therefore , we can only compare the theoretical and experimental _ differences _ of spin along the calculated and observed bands .
to this end , we define the theoretical difference of spin as @xmath182 where @xmath183 is calculated from the two lowest experimental transitions according to the standard expression ( [ eq300a ] ) .
this theoretical difference can be compared to the experimental difference of spin @xmath184 ( as a function of @xmath5 ) relative to the ( unknown ) spin @xmath185 of the state fed by the lowest transition seen in experiment .
this experimental difference of spin simply equals twice the number @xmath186 of the gamma transition counted from the bottom of the band .
a shift of 2@xmath88 is added to both definitions ( [ eq402a ] ) and ( [ eq402b ] ) to account for the fact that the lowest experimental value of the frequency ( [ eq300a ] ) ( @xmath183 ) is known at @xmath185 + 2@xmath88 .
for all the sd bands in @xmath0 tb , calculated within the hf method using the skm * force and with the scaling factor @xmath138=0.9 discussed in sec.[sec5a2 ] , the differences between @xmath187 and @xmath188 do not exceed @xmath189 , for all the frequencies studied , i.e. , from 0.3 to 0.9mev/@xmath88 .
the same is also true ( up to 0.5@xmath88 ) for the yrast band in @xmath2dy . on the other hand ,
similar comparison gives a disagreement for the yrast band in @xmath120gd , which indicates a necessity to consider the pairing correlations in this nucleus , see sec.[sec5b ] .
the spins calculated at @xmath183 give predictions for the exit spins @xmath190 in the sd bands , @xmath191 - 2\hbar , \ ] ] where the brackets @xmath192 $ ] denote the rounding to the nearest integer or half - integer compatible with the total signature and the parity of the number of particles .
the obtained values of @xmath190 are collected in table [ tab5 ] together with the summary of the theoretical hf configurations used . in some cases
where clearly more than one configuration may be considered , two of them are included in the table to illustrate the possible uncertainty of the configuration .
the results for the @xmath190-values have been given in the table as differences with respect to the @xmath193=69/2@xmath88 value obtained for band 1 which is used as the reference .
the representation in terms of differences is used to stress that the absolute value of @xmath193 may be considered less certain ( given the uncertainties of the scaling procedure , neglect of pairing , etc . ) as compared to relative alignments between different bands , which are expected to be much less model dependent .
for example , the exit spin obtained in @xmath2dy(1 ) from the hf calculations without pairing is @xmath190=26@xmath88 ( for the state fed by the 602.4kev transition ) .
this is fully compatible with the exit spin of 57/2@xmath88 obtained in band @xmath0tb(2 ) ( for the next transition of 646.5kev ) and giving the relative alignment of + 1/2@xmath88 , cf .
@xcite .
the effects of dynamical correlations caused by pairing on the alignment mechanism in sd bands of the neighboring nucleus @xmath2dy have been studied , e.g. , in ref.@xcite .
these effects are expected to be small , of the order of a few @xmath88 at the most , and to be rather regular functions of the frequency .
the corresponding alignment corrections were calculated to be negative for several bands in the @xmath106@xmath7150 nuclei .
we have repeated the analysis of these correlations for the even - even neighbors of @xmath0 tb , i.e. , for @xmath2dy and @xmath120gd , for which the pairing effects are not influenced by the blocking mechanism , and are easier to extract in a relatively pure form ( cf .
sec.[sec5b ] ) . without pairing ,
the same value of @xmath190=26@xmath88 is for @xmath2dy(1 ) obtained both within the hf and ws calculations . as shown in sec.[sec5b ] , the inclusion of pairing in the ws method brings this value down to 24@xmath88 . because the pairing influence increases from @xmath2dy towards @xmath120gd , we can very tentatively estimate that the hf absolute exit spins in terbium obtained without pairing overestimate the values with pairing by about 4@xmath88
. hence the inclusion of pairing would give the exit spin of @xmath193=61/2@xmath88 for the state fed by the 768.4kev transition in the @xmath0tb(1 ) band .
it is important to observe that the obtained exit - spin values vary rather markedly from band to band . in particular , in @xmath0tb(1 ) the value of @xmath190 is significantly larger ( by 6@xmath88 ) than that predicted for the first excited band @xmath0tb(2 ) .
the shape evolution for various configurations studied in this article has been determined by calculating the expectation values of the quadrupole moments , @xmath176 and @xmath194 , and the hexadecapole moments @xmath177 , @xmath195 , and @xmath196@xcite . the calculated axial quadrupole and hexadecapole moments , @xmath176 and @xmath177 , decrease with increasing rotational frequency for all the bands studied . the corresponding results for the most important proton moments , i.e. , @xmath176 and @xmath177 are presented in fig.[fig5.12 ] . in fig.[fig5.13 ] we show for each band the proton quadrupole moments in function of the dynamical moments
. one can see that these relations are nearly linear , which indicates that the changes in @xmath1 are correlated with changes of @xmath176 .
one can interpret this characteristic result as a dynamical - moment evolution accompanying the nuclear - shape polarization due to the rotation .
it is also worth observing that the proton excited configurations have noticeably higher quadrupole moments and that their slopes are slightly higher than those of the neutron excited bands .
slight but systematic deviations from the axiality are predicted with increasing rotational frequency .
it is worth mentioning that the deviations from axiality for the signature - partner configurations , @xmath119[3,0,1]1/2(@xmath111=@xmath112 ) @xmath160 @xmath119[6,6,0]1/2(@xmath111=@xmath112 ) and @xmath119[3,0,1]1/2(@xmath111=@xmath113 ) @xmath160 @xmath119[6,6,0]1/2(@xmath111=@xmath112 ) , attributed to bands 2 and 4 , respectively , have opposite signs , although they both remain small .
the axial hexadecapole moments @xmath177 are relatively diversified in that their variations in function of the configuration and rotational frequency may reach as much as 25% .
the more exotic hexadecapole moments with the non - zero @xmath197-components , i.e. , @xmath195 and @xmath196 , have also been calculated .
the interest in those moments has grown recently in relation with the so - called @xmath198=2 staggering observed in transition energies of some rare - earth sd nuclei .
the staggering could manifest a presence of those exotic multipoles in the nuclear shapes and in particular , the non - zero @xmath196-moments could manifest the nuclear @xmath199-symmetry , as suggested in ref.@xcite and discussed in@xcite . the effect in question may be very small , as shown by microscopic calculations@xcite . experimentally , the staggering phenomenon has not been reported to exist in @xmath0 tb nucleus .
nevertheless , we found it of interest to estimate the order of magnitude of the relevant hexadecapole moments .
the calculations indicate that the @xmath195-moments , although typically two orders of magnitude smaller than the @xmath177-moments , remain non - zero , while the @xmath196-moments are still at least by a factor of a few smaller .
the main purpose of the scaling procedure introduced in sec.[sec5a2 ] was , on the one hand , to quantify the systematic deviations from the experiment of the obtained results for the @xmath1-moments .
on the other hand this procedure proves indeed that the obtained discrepancies are systematic since the same scaling factor has been successfully used in various nuclei ( and for all the bands known in each of them ) .
the comparisons were made for several dozens of bands ; presentation of these results was limited so far in this paper to 8 bands in @xmath0 tb only .
discrepancies of similar order of magnitude were also present in calculations of other authors , but were not discussed in any major detail . only as a result of an extended comparison with the experimental data we were lead to a suggestion that the discrepancies are systematic and
thus that the underlying theories ( or parametrizations of the forces ) might be , and most likely should be improved .
as it is well known , the pairing correlations influence the moments of inertia in a rather characteristic manner .
typically ( although not generally ) , the stronger the pairing correlations , the smaller the kinematical moments @xmath134 . the purpose of the present section is to estimate the effect of pairing on the discussed moments , thus putting more restriction on the uncertainties produced by the applied nuclear effective forces .
a technique used here differs from those of refs.@xcite and@xcite in that we perform the particle number projection ( pnp ) of the rotating bcs type wave functions _ before _ variation ( in the former publication the rpa technique within the nilsson model has been used while in the latter , an approximate projection in terms of the lipkin - nogami method together with the hfb approximation ) . here
we apply the pairing - self - consistent bogolyubov method and the pnp before variation , using the standard monopole pairing hamiltonian with the constant average matrix elements , different for the protons and the neutrons ( see below ) , and the ws cranking approach .
calculations@xcite of the sd bands in @xmath106@xmath7150 nuclei have shown that the inclusion of pairing through the density - dependent zero - range interactions improves the agreement with data in @xmath120gd , but at the same time the agreement in @xmath2dy is destroyed .
it seems therefore , that at present one needs some more effort , also in the case of the hf approach , to clarify the problem of the effective interactions used , and although here we will not be able to carry through such a complex project , we would nevertheless like to quantify the difficulty and prepare the starting point for a more advanced discussion .
below we are going to concentrate on the two neighbor - isotones of the @xmath0tb@xmath200 nucleus , i.e. , @xmath120gd@xmath200 and @xmath2dy@xmath200 .
we believe that studying the two even - even nuclei has , in the present context , some important advantages . as our calculations , as well as those of other authors indicate , in the isotones in question the main pairing contributions come from the @xmath151=86 neutron system .
since the relative proton / neutron pairing strengths are uncertain , in studying the weak pairing limit , the two even - even nuclei mentioned offer less uncertainties as compared to odd-@xmath106 ones in which the blocking in the case of the odd proton may cause an additional difficulty . in other words : it is not known at present whether at the weak pairing limit that we are confronted with at high spins the usual blocking procedure is a sufficient an approximation . in order to estimate the influence of the pairing on the alignment process
, we have calculated the kinematical and dynamical moments in function of rotational frequency at fixed deformations representing the average equilibrium - deformation values for the two nuclei in question ( for details see captions to figs.[fig5.14 ] and [ fig5.15 ] ) .
we have taken @xmath151 and @xmath149 single - particle states into account when constructing the bogolyubov transformation for neutrons and protons , respectively . when parametrizing the pairing constants ( average matrix elements ) we took as a starting point the pairing force strength parameters from ref.@xcite @xmath201 , \\
\label{eq45 } { g}_{p}&=&\frac{1}{a}\left[17.90 + 0.176(n - z)\right ] , \end{aligned}\ ] ] where they have been adjusted to experimental data on normal nuclei by applying the pairing - self - consistent bogolyubov formalism without the pnp .
when used within the particle projection formalism , these values have to be decreased , by about 15% ( several aspects related to the pnp technique can be found in ref.@xcite ( solvable model ) and , e.g. , in refs.@xcite ( cranking with the ws hamiltonian ) ) . in the present application we use the reduction factors of @xmath202 for both the neutrons and the protons , i.e. , we apply @xmath203 and @xmath204 in the monopole pairing hamiltonian . in this respect
we in fact apply the parametrization studied earlier in the normal nuclei thus not introducing new parameters for those nuclei studied in this paper .
the results corresponding to our pnp calculations are presented in figs.[fig5.14 ] and [ fig5.15 ] for the @xmath2dy and @xmath120gd nuclei , respectively .
let us observe that the high spin results for the dynamical moments in @xmath120gd agree very well with experiment .
note that the scaling factor used here , @xmath138=0.9 , is the same as that applied in the calculations without pairing presented previously . in other words , in this nucleus the calculations without pairing would not agree with the data . altogether , similarly as in ref.@xcite the results with pairing correspond to an increase in @xmath1 moments as compared to those without pairing . at lower spins the crossing observed in the data through the rise in the @xmath1 moments appears at a similar frequency also in the calculations , although the theoretical crossing frequency is slightly too large . in any case
such a crossing is to a far an extent a matter of individual level positions rather than the collective pairing phase , and as such is of less importance for the discussion in this section . the results for the dynamical moments in @xmath2dy present similar defect at small frequency limit where the calculation provide a crossing - like structure at frequencies which are slightly too high .
again , for most of the spin values , the dynamical moments obtained with pairing are larger than those without pairing .
consequently , the pairing can not be a remedy for the too high moments as calculated ( without scaling ) both within the ws and hf methods . concerning the high spin behavior of the dynamical moments ( slightly different slopes in theory as compared to the data )
: we have checked that a shape evolution ( ignored in the calculations presented in figs.[fig5.14 ] and [ fig5.15 ] ) brings in the effects of broad proton crossings whose onsets are already visible in the routhians from fig.[fig5.3 ] .
these effects are present in a form of a slight modification of the @xmath1 vs. @xmath5 dependence in the experimental frequency range at highest spins only when the proton pairing is taken into account .
the aforementioned modification is sufficient to provide the correct asymptotic of the @xmath98 at high spins in @xmath2dy .
having obtained an overall agreement with the data on the dynamical moments we have also compared the @xmath205-moments as well .
the agreement of the kinematical moments can be considered very good for both nuclei when the experimental exit - spin values ( so far not measured ) are assumed to be equal to those calculated according to the method described in sec[sec5d ] , i.e. , @xmath190=24 and 32@xmath88 for @xmath2dy and @xmath120gd , respectively . incidentally , without pairing , the same method gives the value of @xmath190=38@xmath88 in @xmath120gd , and an incorrect behavior for the spin difference @xmath187 .
let us note that by taking into account the pairing correlations a good description of the kinematical moments at the high - spin limit is obtained for the same values of the scaling factor and pairing constants common for both nuclei , although the effects of pairing , especially in the neutrons , are calculated to be markedly different in the two isotones discussed .
in all compared situations a scaling factor was necessary in order to bring the too high no - pairing results on the @xmath134 and @xmath1-moments towards the experimental data simultaneously for both nuclei studied .
for all the three nuclei , i.e. , @xmath120gd@xmath200 , @xmath0tb@xmath200 , and in @xmath2dy@xmath200 , the inertia contribution coming from the neutrons is expected to dominate that of the protons by about a factor of two ; also the pairing effects are calculated to have about a factor of two more effect in the neutron case .
comparison of the results in the top and the bottom parts of figs.[fig5.13 ] and [ fig5.14 ] indicates that at the highest frequency range the differences between the proton contributions for the case with pairing as compared to the no - pairing case amount to not more than about 1@xmath90mev@xmath91 ( about 1% ) in @xmath2dy nucleus and about 0.5@xmath90mev@xmath91 ( about 0.5% ) in the @xmath120gd nucleus . in the case of the neutrons the corresponding differences are slightly larger but also very small and thus the pairing is calculated to influence the kinematical moments at the highest frequency limit by about @xmath206 . at the low frequency range
the pairing correlations are expected to be systematically more important .
the effect of pairing at @xmath207 on the moment of inertia is about @xmath208 in @xmath2dy and about @xmath209 in @xmath120gd .
the main difference between the calculations for @xmath2dy and @xmath120gd is the smaller deformation in the case of @xmath120gd ; the deformations for most of the bands in the @xmath0 tb nucleus are in between those for the discussed even - even isotones .
consequently we should expect that at the low frequency range in the @xmath0 tb nucleus , the pairing correlations should still amount to @xmath210 on the average , those of the protons being negligible for two reasons : firstly they are very small in the two neighboring even - even nuclei and , secondly , the @xmath211 odd proton in the @xmath0 tb is expected to diminish the proton pairing effect even more due to the blocking arguments .
this is also why the results for the terbium nucleus are expected to lie in between those for the neighboring even - even nuclei studied here .
the detailed analysis of this nucleus calls first , however , for a microscopic justification of the scaling factor introduced in the present study , and for an improved treatment of pairing in an odd nucleus . there exists of course an uncertainty margin in the above estimates .
the dynamical moments are sensitive to the detailed variation of the single - nucleonic routhians with the rotational frequency and a slow alignment of one single orbital may modify the @xmath1 moment considerably thus increasing the risk of estimating the collective pairing effect from the influence of one level . secondly , it is impossible to experimentally separate the proton and neutron contributions and thus in the calculations one may obtain in principle similar results by increasing , e.g. , the proton pairing strength while decreasing that of the neutrons . fortunately ,
in the two nuclei discussed the protons contribute significantly less to the total inertia as compared to the neutrons and , in addition , the neutron pairing effects at @xmath212 in the dy and gd isotones differ considerably .
consequently the second uncertainty mentioned has a rather small influence on the final result .
in this article the two most efficient microscopic approaches in the high - spin nuclear structure calculations , i.e. , those based on the average nuclear field parametrization in terms of the deformed ws potential and the self - consistent hf approach together with the cranking method have been applied .
the results of the calculations employing these two different methods and also the results of the calculations with two different parametrizations of the skyrme interactions have been compared . despite of the fact that the formalisms used differ strongly , the efforts put earlier ( by other authors ) in optimizing the parametrizations of the interactions seem to converge : the single - nucleon spectra generated by the hf and the ws hamiltonians are very similar .
in particular , we have demonstrated that at least for the nucleus studied the self - consistent hf solutions and the ws - strutinsky energy minimized results lead to similar super - deformed shell closures . in protons
there is a series of shell gaps corresponding to @xmath149=62 , 64 and 66 with a characteristic high level - density areas closing those gaps from below and from above .
the @xmath119[3,0,1]1/2 orbital separates the former two gaps and the @xmath119[6,5,1]3/2 orbital the latter two gaps in both skm * and siii parametrizations of the skyrme interaction and in the ws case .
of course in the high - level density areas the relative positions of the single - particle states differ slightly but practically all the same nilsson orbitals are represented there . in the neutron
spectra the @xmath151=86 sd gap is cut into two ( @xmath151=86 and @xmath151=85 gaps ) by one of the signature partners of the @xmath71[7,7,0]1/2 orbital .
the neutron spectra manifest two systematic differences ( when comparing hf and ws ) : ( i ) the signature splitting of the @xmath71[7,7,0]1/2 orbital is much stronger in the hf approach independently of the force , and ( ii ) below the big gaps we find , in the decreasing order , the @xmath71[6,4,2]5/2 , @xmath71[6,5,1]1/2 , and @xmath71[4,1,1]1/2 orbitals ; in the ws case these n=6 orbitals are inverted .
the inversion , however , did not cause any significant problem in interpretations of the spectra in the @xmath0 tb nucleus .
theoretical calculations of the dynamical moments indicate a fair or good agreement for eight bands known experimentally in @xmath0 tb .
this is possible , however , only after a phenomenological scaling factor , common for all bands , is introduced .
generally , for higher - lying sd bands a unique interpretation of the single - nucleonic structure in terms of the nilsson labels does not seem possible ( at least at present ) for several reasons .
firstly , the single - nucleonic routhian spectra contain levels of different intrinsic characteristics which remain nearly degenerate over the long ranges of rotational frequency .
they lead to numerous configurations with very similar alignment and dynamical - moment properties and of very close total energies .
secondly , the band degeneracy ( twinned band effect ) does not provide major help in such cases since several among excited bands may produce nearly identical band spectra .
the deformation changes in function of the band - configurations and rotational frequency are very characteristic : in all bands discussed not only the quadrupole ( @xmath176 ) and the hexadecapole ( @xmath177 ) moments decrease with increasing frequency , as observed also by other authors , but also their variation seems clearly correlated with that of the dynamical and kinematical moments . with progressing alignment small triaxialities ( equivalent to the @xmath10-deformations of the order of a few degrees ) develop as well . the possibility of the existence of some special symmetries ( like , e.g. , an occurrence of the shapes with the @xmath199-symmetry ) has been studied .
no indication for the @xmath196-moments has been found ( an upper limit is predicted at the level of at most 10@xmath213@xmath214@xmath177 ) .
the calculations allow to predict the values of the exit spins , i.e. , the spins of the lowest observed sd transitions .
these predictions are given by the values of spin calculated at the rotation frequency corresponding to the lowest observed transition , and rounded to the nearest half - integer value compatible with the total signature .
the differences between the exit - spin values in different bands are expected to be more reliable than the corresponding absolute values .
we found that the systematic discrepancies between the experimental and calculated results exist in both ws and hf approaches , even if the pairing correlations are taken into account .
this suggests a need for more effort in determining the optimal parametrizations of the effective interactions used for a description of the high - spin properties of the strongly - deformed nuclei .
we would like to express our thanks to the _ institut du dveloppement et de ressources en informatique scientifique _ ( idris ) of cnrs , france , which provided us with the computing facilities under project no .
this research was supported in part by the polish committee for scientific research under contract no . 2 p03b 034 08 , and by the computational grant from the interdisciplinary centre for mathematical and computational modeling ( icm ) of the warsaw university .
p. bonche , h. flocard and p .- h .
heenen , nucl . phys .
* a523 * , 300 ( 1991 ) .
b.q . chen , p .- h .
heenen , p. bonche , m.s .
weiss and h. flocard , phys .
rev . * c46 * , 1582 ( 1992 ) .
egido and l.m .
robledo , phys .
70 * , 2876 ( 1993 ) .
m. girod , j.p .
delaroche , j.f .
berger , and j. libert , phys .
b325 * , 1 ( 1994 ) . j.dobaczewski and j. dudek , phys
* c52 * , 1827 ( 1995 ) .
p. bonche , h. flocard , and p .- h .
heenen , nucl . phys . *
a598 * , 169 ( 1996 ) .
w. koepf and p. ring , nucl . phys .
* a493 * , 61 ( 1989 ) ; * a511 * , 279 ( 1990 ) . j. knig and p. ring , phys .
* 71 * , 3079 ( 1993 ) .
afanasjev , j. knig , and p. ring , nucl . phys .
* a608 * , 107 ( 1996 ) .
c. baktash , b. haas , and w. nazarewicz , annu .
nucl . part .
sci . * 45 * , 485 ( 1995 ) .
t. bengtsson , i. ragnarsson , and s. berg , phys .
b * 208 * , 39 ( 1988 ) .
y. r. shimitzu , e. vigezzi , r. a. broglia phys . lett . *
198b * , 33 ( 1987 ) ; y. r. shimitzu , e. vigezzi , r. a. broglia nucl . phys . * a509 * , 80 ( 1990 ) . j. dudek , b. herskind , w. nazarewicz , z. szymaski and t. werner , phys .
rev . * c38 * , 940 ( 1988 ) . w. nazarewicz , r. wyss and a. johnson , phys . lett . * 225b * , 208 ( 1989 ) .
w. nazarewicz , r. wyss and a. johnson , nucl . phys .
* a503 * , 285 ( 1989 ) . w. satua , r. wyss , and p. magierski , nucl .
a578 * , 45 ( 1994 ) .
p. fallon , a. alderson , m.a .
bentley , a.m. bruce , p.d .
forsyth , d. howe , j.w .
roberts , j.f .
scharpey - schafer , p.j .
twin , f.a .
beck , th .
byrski , d. curien and c. schuck , phys .
b218 * , 137 ( 1989 ) .
byrski , f.a .
beck , d. curien , c. schuck , p. fallon , a. alderson , i .ali
bentley , a.m. bruce , p.d .
forsyth , d. howe , j.w .roberts , j.f .
scharpey - schafer , g. smith and p.j .
twin , phys .
lett . * 64 * , 1650 ( 1990 ) .
g. defrance , b. haas , i. ragnarsson , p.j .
twin , c.w .
beausang , f.a .
beck , th .
byrski , s. clarck , p.j .
dagnall , d. curien , g. duchne , p. fallon , s. flibotte , s. frobes , p.d .
forsyth , b. kharraja , j.c .
lisle , j.c .
merdinger , c.m .
petrache , d. prvost , j.f .
scharpey - schafer , j. simpson , j.p .
vivien and k. zuber , phys .
b331 * , 290 ( 1994 ) .
b. kharraja , th .
byrski , f.a .
beck , c.w .
beausang , m.a .
bentley , d. curien , p.j .
dagnall , g. duchne , s. flibotte , g. de france , zs .
fulop , b. haas , a.z .
kiss , j.c .
lisle , c.m .
petrache , ch .
theisen , p.j .
twin , j.p .
vivien and k. zuber , phys .
b341 * , 268 ( 1995 ) . for technical reasons
explained in ref.@xcite we use the @xmath108 axis as the cranking axis . in the present study
we identify the projection of the angular momentum on the cranking axis with the total angular momentum .
h. savajols , a. korichi , d. ward , d. appelbe , g.c . ball , c. beausang , f.a .
beck , t. byrski , d. curien , p. dagnall , g. de - france , d. disdier , g. duchene , s. erturk , c. finck , s. flibotte , b. gall , a. galindo - uribarri , b. haas , g. hackman , v.p .
janzen , b. kharraja , j.c .
lisle , j.c .
merdinger , s.m .
mullins , s. pilotte , d. prevost , d.c .
radford , v. rauch , c. rigollet , d. smalley , m.b .
smith , o. stezowski , j. styczen , c. theisen , p.j .
twin , j.p .
vivien , j.c .
waddington , k. zuber , i. ragnarsson , phys .
lett . * 76 * , 4480 ( 1996 ) .
s. flibotte , h.r .
andrews , g.c .
ball , c.w .
beausang , f.a .
beck , g. belier , t. byrski , d. curien , p.j .
dagnall , g. de france , d. disdier , g. duchne , ch .
finck , b. haas , g. hackman , d.s .
haslip , v.p .
janzen , b. kharraja , j.c .
lisle , j.c .
merdinger , s.m .
mullins , w. nazarewicz , d.c .
radford , v. rauch , h. savajols , j. styczen , ch .
theisen , p.j .
twin , j.p .
vivien , j.c .
waddington , d. ward , k. zuber , and s. berg , phys .
lett . * 71 * , 4299 ( 1993 ) .
@xmath2231/2(r\mbox{=$+i$})\rightarrow \pi[6,6,0]1/2(r\mbox{=$+i$})~ & -6 & 646.5 \\ 3 & ~(+,+i)~ & ~\pi[6,5,1]3/2(r\mbox{=$-i$})\rightarrow \pi[6,6,0]1/2(r\mbox{=$+i$})~ & -3 & 727.1 \\ 4 & ~(-,+i)~ & ~\pi[3,0,1]1/2(r\mbox{=$-i$})\rightarrow \pi[6,6,0]1/2(r\mbox{=$+i$})~ & + 3 & 865.8 \\ 5 & ~(+,-i)~ & ~\nu[7,6,1]3/2(r\mbox{=$-i$})\rightarrow \nu[5,2,1]3/2(r\mbox{=$-i$})~ & -2 & 710.1 \\ 5 & ~(-,-i)~ & ~\nu[7,6,1]3/2(r\mbox{=$-i$})\rightarrow \nu[4,0,2]5/2(r\mbox{=$-i$})~ & -2 & 710.1 \\ 6 & ~(+,+i)~ & ~\nu[7,6,1]3/2(r\mbox{=$-i$})\rightarrow \nu[5,2,1]3/2(r\mbox{=$+i$})~ & + 3 & 838.2 \\ 6 & ~(-,+i)~ & ~\nu[7,6,1]3/2(r\mbox{=$-i$})\rightarrow \nu[4,0,2]5/2(r\mbox{=$+i$})~ & + 3 & 838.2 \\ 7 & ~(-,+i)~ & ~\pi[6,5,1]3/2(r\mbox{=$-i$})\rightarrow \pi[5,3,0]1/2(r\mbox{=$+i$})~ & + 1 & 753.9 \\ 7 & ~(-,+i)~ & ~\nu[7,6,1]3/2(r\mbox{=$-i$})\rightarrow \nu[4,0,2]5/2(r\mbox{=$+i$})~ & -1 & 753.9 \\ 8 & ~(-,-i)~ & ~\nu[7,6,1]3/2(r\mbox{=$-i$})\rightarrow \nu[4,0,2]5/2(r\mbox{=$-i$})~ & ~~0 & 785.0 \\ \hline \end{array}\ ] ] | structure of eight superdeformed bands in the nucleus @xmath0 tb is analyzed using the results of the hartree - fock and woods - saxon cranking approaches .
it is demonstrated that far going similarities between the two approaches exist and predictions related to the structure of rotational bands calculated within the two models are nearly parallel .
an interpretation scenario for the structure of the superdeformed bands is presented and predictions related to the exit spins are made .
small but systematic discrepancies between experiment and theory , analyzed in terms of the dynamical moments , @xmath1 , are shown to exist .
the pairing correlations taken into account by using the particle - number - projection technique are shown to _ increase _ the disagreement .
sources of these systematic discrepancies are discussed they are most likely related to the yet not optimal parametrization of the nuclear interactions used . |
the motion of an elastic medium across a quenched disordered background presents a simple paradigm for the understanding of several experiments@xcite .
these include studies of the depinning of charge - density waves@xcite , transport measurements in the mixed phase of type - ii superconductors@xcite as well as measurements of the flow of colloidal particles across rough substrates@xcite .
the issue of universality at continuous depinning transitions has traditionally dominated much of this literature , especially in the cdw context@xcite . however , the behaviour of non - universal quantities in the vicinity of the depinning transition is often of more interest to the experimenter@xcite .
the nature of order , correlations and response within the moving phase are also questions which underly many recent investigations of the physics of non - equilibrium steady states , motivated in large part by the considerable experimental literature on dynamical states of flux lines in the mixed phase of driven , disordered type - ii superconductors@xcite .
the canonical example of a remarkable _ non - universal _ feature of the depinning transition is the peak effect often seen in the mixed phase in the vicinity of the upper critical field @xmath0@xcite .
the peak effect , a generic property of weakly disordered type - ii superconductors , describes the non - monotonic behaviour of the critical current @xmath1 as a function of the temperature @xmath2 or applied magnetic field @xmath3@xcite .
this critical current measures the ( depinning ) force required to induce observable motion of the vortex line array@xcite .
the peak effect is an often spectacular phenomenon , with @xmath1 rising sharply in a narrow region whose width is comparable to that of the zero - field superconducting transition@xcite .
investigations of the peak effect describe a host of unusual phenomena associated with this narrow regime@xcite .
these include `` finger print phenomena '' , slow voltage oscillations , history dependent dynamic response , enhanced low - frequency noise with a @xmath4 spectrum and many other remarkable features .
these are often collectively referred to as `` peak effect anomalies '' @xcite .
approaches to understanding the peak effect have typically followed two distinct paths .
the first views the peak effect as arising solely from the softening of the flux - lattice close to h@xmath5 , as reflected in the vanishing of the shear elastic constant @xmath6 , while pinning strengths soften more gradually , as @xmath7@xcite . as suggested initially by pippard@xcite
, softer lattices should be able to adapt better to random pinning@xcite . in the second class of theories ,
the peak effect is a reflection of the underlying phase diagram of a weakly pinned , flux - line array in the @xmath8 plane@xcite .
it has been argued that such phase diagrams should generically accomodate intermediate glassy phases close to the melting transition@xcite .
the peak effect is then proposed to be associated with abrupt changes in transport across such phase boundaries , with the anomalies rationalized in terms of the glassy nature of such intermediate states@xcite .
a third , as yet unexplored , alternative to these approaches which addresses the origin of the anomalies directly , combines the scenario of an underlying static phase transition with the possibility that driving such a system induces dynamical steady states with no static counterpart . with this motivation in mind ,
the central questions addressed in this paper are the following : consider an underlying static phase transition in a pure system as modified or broadened by weak quenched disorder .
how are signals of this transition manifest in dynamical measurements ?
further , can novel dynamical states with no analog in either the pure or the disordered _ undriven _ system be obtained once the system is driven ?
finally , could some of the remarkable phenomenology of the peak effect anomalies possibly originate in the properties of such states ?
we recently proposed a suitable model system capable of addressing some of these issues@xcite .
our model uses interacting particles in two dimensions close to zero temperature .
these particles form a crystal in the absence of disorder .
the interaction potential contains a simple two - body repulsive power - law interaction as well as a short - range three - body interaction .
the two - body interaction favours a triangular lattice .
the three - body term favours a square lattice .
we tune between a square and a triangular ground state by varying the strength of a single parameter , the coefficient of the three - body interaction @xmath9 .
we place @xmath10 such interacting particles in a quenched disordered background , modeled numerically in terms of a gaussian random field with specified strength and two - point correlations . after finding the ground state of the interacting particles in the disordered background using simulated annealing techniques
, we apply a uniform driving force to the particles .
this results , first , in a depinning transition and then a sequence of partially ordered states with varying degrees of spatial correlations as the force is increased .
the advantages to such a formulation are several .
signatures of phase transitions in dynamical measurements are typically overwhelmed by thermal fluctuations for purely temperature driven transitions , thereby obscuring the very effects we wish to characterize .
our model of a @xmath11 transition between square and triangular phases surmounts this problem while also mirroring similar structural phase transitions in vortex lattices , typically between triangular and distorted rectangular phases , across which broadened peak effects are seen @xcite .
the sequence of steady states obtained in our model as a function of the uniform external driving force @xmath12 acting on the particles , and for various values of the three - body interaction strength @xmath9 , is summarized in the dynamical `` phase '' diagram of fig .
[ phasedia ] .
this phase diagram extends a similar phase diagram proposed earlier to much larger values of @xmath9@xcite .
the phase diagram shows a variety of phases : pinned states which may have dominantly triangular or square correlations , a plastically flowing `` liquid - like '' state , a moving anisotropic hexatic phase , flowing triangular and square states ordered over the size of our simulation cell and a dynamic coexistence regime .
: schematic dynamical phase diagram of our model system , plotted as a function of the three - body interaction strength @xmath9 and of driving force @xmath12 ( see text for a more detailed discussion ) .
the phases are ( a ) pinned triangular ( b ) pinned square ( c ) plastic flow ( d ) anisotropic hexatic ( e ) flowing triangular ( g ) coexistence regime and ( h ) flowing square .
the precise location of the lines separating these phases is disorder - dependent , although the general topology of the phase diagram is not sensitive to disorder .
the boundary separating plastic flow from the flowing square phase is a very strong function of the driving force @xmath12 at large @xmath9 .
, width=302 ] this paper presents a detailed characterization of these states .
we define and calculate appropriate correlation functions for @xmath13atic ( where @xmath14 , mostly ) orientational order , in addition to structure factors measuring the distribution of particles .
we capture local bond - orientational order in terms of distribution functions of a complex number characterizing the local orientation .
this representation is used to understand the action of an external force in biasing the axes of orientational order .
we demonstrate that the hexatic order we measure arises as an artefact of averaging an anisotropic quantity , which decays either exponentially to zero or to a constant value along two principal directions . in
the `` ordered '' moving phases , square and triangular , shown in fig .
[ phasedia ] , orientational order appears to be established over length scales much large than our system sizes . in the coexistence regime ( labeled ( g ) in the phase diagram of fig .
[ phasedia ] ) , we investigate the nucleation and growth of ordered domains of one type ( square or triangular ) in another .
such nuclei are , in general , anisotropic , forming along the principal crystalline directions of each phase .
the interface region of different domains is remarkably dynamic .
we assign the substantial increase in noise we see within the coexistence regime to the unusual properties of this interface , quantifying this proposal by linking measures of fluctuation magnitude to coordinations intermediate between square and triangular .
we measure several quantities on the dynamical side , including the basic current - force relations , current statistics and current noise at many different points in the phase diagram .
we calculate the koshelev - vinokur `` shaking temperature''@xcite , defined in detail below , to understand how disorder - induced fluctuations in the flow might provide an effective , pure - system temperature in terms of which phase behaviour can be discussed .
this shaking temperature , while a monotonically decreasing function of the applied force over much of the @xmath15 phase diagram , is strikingly _ non - monotonic _ in @xmath12 within the coexistence regime .
we correlate local orientational and density fluctuations within the coexistence state to understand the origins of the anomalous noise in the coexistence regime .
we compare the phenomenology of the simulations , specifically relating to the coexistence regime , to what is seen in experiments on peak effect anomalies in the mixed state .
we argue that this comparison , supported by general phenomenological arguments , suggests strongly that there may be a generic explanation for peak effect anomalies .
the outline of this paper is the following . in section
ii , we describe the model system we use , explaining our methodology and discussing relevant features of the simulations and what we calculate . in section iii
, we provide an overview of the phase diagram of the driven system , discussing , in particular detail , the hexatic vortex glass and the coexistence phase .
we discuss growth and fluctuations of one phase within another , the behaviour upon quenching and the origins of noise in this regime . in section iv
, we discuss some aspects of peak effect anomalies seen in the experiments , pointing out the close relationship between the diversity we see in our simulations with the experimental data .
we then conjecture that behaviour analogous to what we see in the coexistence regime may be generic to all driven disordered systems in the vicinity of an underlying static first - order phase transition in the pure limit . finally , in our concluding section , section
v , we summarize our results briefly and suggest further lines of research .
our model system is two - dimensional and consists of particles with two and three - body interactions@xcite .
the three - body interaction , parametrized through a single parameter @xmath9 , tunes the system across a square - triangular phase transition .
the total interaction energy for particles confined to two - dimensions and labelled by their position vectors @xmath16 is thus @xmath17 where @xmath18 .
we take the two - body interaction to be of the power - law form @xmath19 while the three - body term is @xmath20.\ ] ] the function @xmath21 for @xmath22 and @xmath23 otherwise and @xmath24 is the angle between @xmath25 and @xmath26 .
the two - body interaction favors a triangular ground state while the three - body term favors @xmath27 and @xmath28 bonds and hence a square structure . energy and length scales are set using @xmath29 and @xmath30 .
the zero - temperature phase diagram for particles interacting with this potential has been calculated in ref . @xcite . as a function of the parameter @xmath9 , which measures the strength of the three - body term
, a discontinuous transition between a triangular lattice , obtained for @xmath31 , and a square lattice , obtained for @xmath32 , is seen .
a similar potential was used by stillinger and weber in an early study of melting of a square solid@xcite . particles also interact with background quenched disorder in the form of a one - body gaussian random potential field v@xmath33 with zero mean and exponentially decaying ( short - range ) correlations .
this potential field is defined on a fine grid following a methodology due to chudnovsky and dickman@xcite , and interpolation is used to find the value of the potential at intermediate points@xcite .
the disorder variance is set to @xmath34 and its spatial correlation length is @xmath35 .
larkin length estimates@xcite yield @xmath36 , with @xmath37 the lattice parameter , somewhat larger than our system size .
the system evolves through standard langevin dynamics ; @xmath38 here @xmath39 is the velocity , @xmath40 the total interaction force , and @xmath41 the random force acting on particle @xmath42 , simulating thermal fluctuations at temperature @xmath2 . a constant force @xmath43 drives the system .
the zero - mean thermal noise @xmath44 is specified by @xmath45 with @xmath46 , well below the equilibrium melting temperature of the system .
the unit of time @xmath47 , with @xmath48 the viscosity .
our system consists of @xmath10 particles , with @xmath10 between @xmath49 and @xmath50 , in a square box at number density @xmath51 .
configurations obtained through a simulated annealing procedure are the initial inputs to our langevin simulations .
this annealing procedure involves equilibration using a nvt monte carlo scheme at fixed density and within a fixed background potential of a tunable amplitude .
the strength of the disorder is then increased in steps to the working disorder strength , with the system equilibrated at each step for around @xmath52 monte carlo steps .
varying the strength of disorder in this fashion enables the system to converge to its true minimum energy state more efficiently than methods which employ a temperature annealing schedule .
we evolve the system using a time step of @xmath53 .
the external force @xmath54 is ramped up from a starting value of 0 , with the system maintained at upto @xmath55 steps at each @xmath54 .
given the local instantaneous particle density @xmath56 we calculate a variety of structural observables at equal time , such as the static structure factor @xmath57 defined by @xmath58 where @xmath59 .
delaunay triangulations yield the probability distributions @xmath60 of @xmath61 and @xmath62 coordinated particles ( @xmath63)@xcite . we define order parameters @xmath64 to distinguish between square and triangular phases , @xmath65 to distinguish between liquid ( disordered ) and triangular crystals and @xmath66 to distinguish between liquid and square crystals .
in addition , we compute the hexatic order parameter @xmath67 and its correlations , defined via @xmath68 where @xmath69 defines the bond angle associated with the vector connecting neighbouring particles , as measured with respect to an arbitrary external axis . the second moment of the distribution of @xmath70 is the bond - orientational susceptibility .
below , we describe alternative distribution functions which quantify the extent to which the axes of the crystal align along the driving force direction .
these include `` argand plots '' of the distribution of a complex quantity which characterizes local orientational order .
the dynamical variables we study include the center of mass velocity @xmath71 , and the particle flux and its statistics .
the centre - of - mass velocity is defined via @xmath72 where the brackets @xmath73 denote an average in steady state and @xmath74 is the velocity of particle @xmath42 at time @xmath75 .
we measure the particle flux by counting the number of particles which cross an imaginary line crossing the @xmath76 axis in a single time step and then averaging this result over time .
this flux , essentially a current @xmath77 integrated over all @xmath78 at fixed x , which must be independent of @xmath79 in steady state , has fluctuations about a constant value .
we measure and discuss the power spectrum of these fluctuations .
koshelev and vinokur ( kv)@xcite have suggested that the combination of the drive and the disorder should yield an effective `` shaking '' temperature in the moving phase .
such an effective temperature manifests itself in transverse and longitudinal fluctuations of the velocity .
we calculate the kv shaking temperature @xmath80@xcite appropriate to the drive and transverse directions , obtaining it from @xmath81 we measure the variation of the @xmath82 s as a function both of force and @xmath9 at various points in our phase diagram .
the transition to the coexistence phase is marked by the growth of highly anisotropic , obliquely inclined square nuclei , in a background of approximately triangularly coordinated solid .
to understand their role in the nucleation kinetics , we define and compute quantities which measure the anisotropy of such clusters . to determine the source of excess noise in the coexistence phase
, we measure the probability distribution of the instantaneous velocity excess above the mean , as a function of the local coordination as well as of the local density . as we show below
, these measurements indicate that the substantial portion of the noise originates from regions which are neither wholly square or triangular but to be found at the interface between such locally ordered states .
a qualitative understanding of the basic structure of the phase diagram of fig .
[ phasedia ] can be obtained from snapshots of instantaneous configurations in the steady state .
such snapshots are shown in fig .
[ recryst ] , obtained from simulations at @xmath46 with @xmath83 .
the configurations labelled ( a)-(f ) are for driving forces @xmath84 and @xmath85 respectively .
particles are colored according to their coordination number @xmath86 = 4 ( magenta ) , 5 ( green ) , 6 ( blue ) , 7 ( orange ) and 8 ( grey ) , obtained from delaunay triangulations and a voronoi analysis@xcite .
disorder - induced inhomogeneities spawn local defects and dislocations ( @xmath87 and @xmath62 coordinated particles ) in the system at low @xmath12 .
boosting the external force transforms the disordered pinned phase , into a plastically flowing disordered fluid - like `` plastic - flow '' phase .
this then turns into a coherently moving triangular phase at still larger @xmath12 .
we show the evolution of the coordination number histograms of the system for the corresponding forces in fig .
[ coordno ] . in the disordered state ,
5-fold and 7-fold coordinated particles are roughly similar in number as expected . the number of particles with 5 and 7 fold coordination decrease with the drive ( fig .
[ coordno ] ( a),(b ) ) .
there appears to be a small intermediate regime where we see some clustering of these defects , as in ( c ) and ( d ) . finally , at much larger @xmath12 , the number of non - hexagonal coordinated particles declines abruptly ( fig .
[ coordno ] ( e),(f ) ) and the system freezes into a triangular latttice .
similar observations hold for the case of scans in @xmath12 at larger @xmath9 , with the difference that the ultimate large - force state is the flowing square lattice .
configurations in the `` coexistence regime '' obtained at much larger forces ( @xmath88 for @xmath89 ) are discussed separately in the sections which follow . for small @xmath12 the solid
is pinned . in this regime , for @xmath90 , there is no centre of mass motion at the longest times we simulate .
we see transient motion in the initial stages of the application of the force , which then dies down once the system optimizes its location within the background of pinning sites . as @xmath12
is increased across the depinning threshold , there are long transient time - scales for motion to set in@xcite .
these time - scales appear to diverge as the system approaches the transition , in agreement with expections concerning continuous depinning transitions .
however , depinning appears to be hysteretic , on the length and time scales we consider _ i.e. _ the reverse path , from depinned to pinned states yields an abrupt transition between these states@xcite .
the depinned state just above the transition is inhomogeneous and undergoes plastic flow@xcite consistent with earlier numerical work . for larger @xmath54
the velocity approaches the asymptotic behaviour @xmath91 .
the pinned phase is a phase with short - range order in both translational and orientational correlation functions .
correlations typically extend to about 4 - 8 interparticle spacings at the levels of disorder we consider . at non - zero noise strength , strictly speaking , we would expect an activated creep component to the motion .
this presumably lies beyond the time scales of our simulation , given the relatively low temperatures ( @xmath92 ) we work at . upon increasing the force
, we enter a regime of substantial plastic flow@xcite .
this regime is one in which particle motion is extremely inhomogeneous .
similar plastic flow is seen in a large number of simulations of particle motion in a random pinning background@xcite .
what is unusual here , however , is the strongly non - monotonic character of the plastic flow boundary , as illustrated in the phase diagram of fig .
[ phasedia ] .
note that the plastic flow phase boundary in the @xmath15 plane is largely independent of @xmath12 for small @xmath9 .
however , once the critical value of @xmath9 for the square - triangular transition is crossed , this boundary becomes a strong function of @xmath12 , with the plastic flow region expanding considerably before it collapses again .
the structure factor @xmath93 of the plastically moving phase ( c ) obtained in a narrow region just above the depinning transition consists of liquid - like isotropic rings , as shown in fig .
[ sqfigs](a ) . for much larger values of @xmath9 , the disordered
, pinned phase appears to depin directly into the highly ordered moving square lattice phase , with no trace of a plastic flow regime .
we have searched for tetratic phases , with algebraically decaying tetratic correlations in the vicinity of this depinning transition . however
, no such phase is apparent in our numerics . in the plastic flow regime ,
transport properties are noisy , reflecting the underlying highly disordered nature of the phase , in agreement with previous work on such plastic flow states in systems without competing phases@xcite .
we see no evidence for a two - peak structure in the velocity distribution function , unlike some previous work , provided we average enough .
thus no particle is stationary over the full time - scales of our simulations .
as @xmath12 is increased further , we encounter a narrow regime in which translational correlations are short - ranged while angle - averaged orientational correlations appear to decay as power laws . within this phase
the circular ring in @xmath93 concentrates into six smeared peaks , as shown in fig .
[ sqfigs](b ) .
the presence of 6-fold order in the absence of the sharp bragg - peaks associated with crystalline ordering suggests that this phase may have hexatic orientational order .
we thus tentatively identify this phase as a driven hexatic , as shown in ( d)@xcite ; the terminology `` anisotropic '' is justified in what follows
. however , to strengthen this assignment , other possible assignments , such as to a coexistence regime known to plague similar analyses of hexatics in two dimensional systems , must be ruled out@xcite .
we compute the correlations of the hexatic order parameter for a very large system at varying values of @xmath12 , as shown in fig .
[ hexcorr ] , where @xmath94 .
this figure displays the evolution of @xmath95 , as @xmath12 is varied across the phases @xmath96 at fixed @xmath89 .
we observe a sharp exponential decay of hexatic correlations in ( c ) and find that at the liquid to hexatic transition for @xmath97 , the decay fits the universal behavior @xmath98 expected at the fluid hexatic transition in two - dimensional non - disordered fluids .
this exponent is obtained in a relatively narrow range of forces .
our system sizes are comparable to typical sizes employed to observe a metastable hexatic phase in two dimensional melting of pure solids . :
the hexatic correlation function @xmath95 for @xmath99 and @xmath100 , each averaged over 50 independent configurations of a @xmath101 particle system at @xmath102 .
the solid line indicates the universal behavior @xmath103 at the liquid to hexatic transition .
, width=302 ] in fig .
[ combinedscaling](a ) , we compare the distribution of @xmath104 over the full system , for the plastically flowing , moving triangular and the intermediate hexatic - glass regimes .
the figure illustrates that in the liquid / plastic regime ( @xmath105 ) , the distribution is primarily governed by non six - fold coordinated particles . in the high driving force regime ( @xmath106 ,
six - fold coordinated particles contribute in main .
this raises the question of whether the intervening ` hexatic ' regime we observe in terms of @xmath57 and the correlation of the hexatic order parameter @xmath95 at @xmath107 , is truly hexatic or defined by a coexistence of the ` solid ' and ` disordered ' phases . if configurations resemble those at solid - fluid coexistence , the distribution would be expected to be a sum of the disordered , relatively ordered and interface distributions ( weighted with their relative areas ) . to settle this issue
we investigate the dependence of the distribution of @xmath104 on the size of the system .
the distribution is obtained by dividing the system into a number of blocks and computing the distribution of @xmath104 within each block for every configuration . in this way
, information over many length scales can be obtained from the same set of configurations .
this provides us with the distribution of @xmath104 for various fractions of the total system .
we find , as shown in fig .
[ combinedscaling](b ) , that apart from finite size effects , there is not much difference between the distributions . also , the distributions for the two largest systems , _ viz . _ @xmath108 and @xmath109 ( @xmath110 being the size of the simulation box ) , coincide to within statistical errors . in the case of
two - phase coexistence one would expect a strong size dependence when the size of the blocks is comparable to the size of the phase separating clusters .
these observations rule out two - phase coexistence , at least of the conventional kind , as an explanation for the long orientational correlations obtained on circular averaging . as we show below , a more detailed characterization of this phase yields the following results : long - range order in orientation is always present in the direction of drive , while remaining short - range in the perpendicular direction .
when averaged over all orientations , the resulting function appears power - law - like over 2 - 3 decades _
i.e. _ over length scales addressed in virtually all simulations so far .
thus , while we call the intervening phase as the `` anisotropic hexatic '' , we diverge sharply from previous work in our claim that the true phase is never a true hexatic in the real sense , since it always has long - range order in the drive direction . we conclude that the apparent hexatic correlation arises , in fact , as an artefact of circularly averaging a very anisotropic correlation function , with qualitatively different decays in the longitudinal and transverse directions . the power - law decay of orientational correlations shown in fig .
[ hexcorr ] , taken together with the scaling of @xmath104 suggests that the driven system in the vicinity of @xmath111 might best be described as a hexatic , with power - law correlations in the local orientation but short - range translational order .
such a state is analogous to the low - temperature phase of the 2-dimensional xy model , where the mermin - wagner theorem rules out long - range order at any finite temperature but vortex - like excitations responsible for the transition to the disordered phase exist chiefly as bound vortex - anti - vortex pairs .
however , the presence of the drive direction introduces an anisotropy into the system .
the consequences of this anisotropy have not , to the best of our knowledge , been explored in previous studies of putative hexatic phases in driven disordered solids@xcite . in our studies
, we have been motivated by an analogy to the physics of the xy model in an applied field , specifically by the intriguing possibility that the drive might play a role equivalent to that of the magnetic field in the xy model .
renormalization group studies @xcite of the 2-dimensional xy model in an external magnetic field with hamiltonian , @xmath112 where the external field @xmath113 points along the positive @xmath79-axis , indicate three distinct phases .
these are a linearly confined phase , a logarithmically confined phase and a free vortex phase obtained as the temperature is gradually increased .
vortex - antivortex pairs at low temperatures are linearly confined by a string of overturned spins . with increasing temperature these strings
participate in a proliferation transition , but vortices still remain confined due to a residual logarithmic attraction . as the temperature is further increased , the vortices overcome this attraction and are finally deconfined .
since the external field breaks rotational symmetry , the magnetization is non - zero in all the three phases .
further , the free energy is argued to be smooth ; thus the distinction between phases can only be found in the structure and distribution of topological defects .
the correlation function @xmath114 is then predicted to saturate to a constant value ( @xmath115 , say ) for @xmath116 .
this constant @xmath117 since the order parameter ( @xmath118 ) couples linearly to the field @xmath119 . on the other hand , the correlation function @xmath120 which corresponds to spin fluctuations in the direction transverse to the applied field ( see eq .
( 6.6 ) ) , is predicted to show an exponential decay in all the three phases .
to test this possibility , we must map local orientational order to an appropriate xy - like two - dimensional vector from which defect structures can be extracted .
this is done through the local hexatic order parameter , @xmath121 , defined earlier .
it is convenient to identify this local quantity ( a phasor ) with a ` hexatic - spin ' with components @xmath122 $ ] and @xmath123 $ ] , thus mapping local geometrical order to a soft - spin xy degree of freedom .
we illustrate this mapping from the local orientations of particles in real space to xy spins in fig .
[ mapspin ] . unlike in the xy model ,
such hexatic spins are not attached to a fixed lattice but associated with the moving particles .
they thus encode important information concerning the equal - time orientational correlations in the driven system .
[ hexspin ] shows real space configurations of the particle system with associated hexatic spins at various values of @xmath12 , illustrating the variety of associated spin configurations present in this mapping .
these configuration maps enable the identification of topological defects in the ordering in the associated xy model .
note the presence of locally aligned regions as well as vortex - like excitations of strength 1 and 1/2 .
qualitatively , we find that in both the disordered liquid phase and the anisotropic hexatic phases , vortex configurations in such spin configurations appear to have little correlation with each other , suggesting that they may be either unbound or relatively weakly bound at best ( see fig .
[ hexspin ] ( a ) and ( b ) ) ; the boxed regions of these figures as indicated are expanded in fig .
[ hexspin](i ) and ( ii ) for ( a ) and in ( iii ) for ( b ) .
however , we have not been able to establish a quantitative distinction between the disordered liquid phase and the anisotropic hexatic phase using our simulation data .
some quantification is , however , possible for larger forces , in the flowing triangular phase , ( fig .
[ hexspin ] ( c ) ) , where defects in the ordering appear to be associated with strips , or strings , of defect . in this case , the defect is locally a solid with square symmetry .
( an expanded plot of the boxed region in fig .
[ hexspin](c ) is shown in ( iv ) , illustrating the one - dimensional character of the defect ) .
we expect that such defects should be _ linearly _ bound , with an energy proportional to the length of the strip
. the binding energy in the latter case should be proportional to a non - equilibrium analog of a surface tension between the square and the triangular crystal .
this argument is supported by calculations of the moment of inertia tensor of the set of particles which belong to such a defect , averaged over configurations which contain such defects .
the largest eigenvalue ( @xmath124 ) of this tensor measures the length of these extended string - like defects . in fig .
[ string ] we show the probability distribution @xmath125 at a few different values of the external drive .
it is clear that within the intermediate force regime , the distribution is exponential , implying that the energy for these excitations scale linearly with their size .
these string like excitations align preferentially along the crystallographic axes of the surrounding triangular lattice . as the force
is increased , these defects offer nucleation sites for square crystals which are less anisotropic .
the probability distribution then ceases to be linear in @xmath126 , leading to the long tail in the data shown in fig .
[ string ] at @xmath127 .
: probability distribution of the largest eigenvalue @xmath124 in a semi - logarithmic scale for external forces @xmath128 and @xmath129 .
it is clear that for all but the largest force , the probability distribution is exponential showing that the energy of string - like excitations within the triangular phase is linear with size . for the largest force which is inside the coexistence region ( see text ) , the probability distribution decays slower than an exponential.,width=302 ] fig .
[ argands ] shows the probability distribution of the ` hexatic - spin ' phasors for forces @xmath130 ( a ) , @xmath131 ( b ) , @xmath132 ( c ) and @xmath85 ( d ) . while fig .
[ argands](a ) , obtained within the plastic flow phase , appears to have a uniform distribution of hexatic spins with angle , figs .
[ argands](b ) , ( c ) and ( d ) display substantial non - uniformity in this distribution .
note the following feature of fig .
[ argands](c ) : the mapped spins tend to overwhelmingly point along the drive direction in the flowing triangular state .
this should be contrasted with the fact that the corresponding distribution function for the x - y model in zero external field is isotropic across the disordered to quasi - long - range - ordered transition , peaking below it at @xmath133 , a value independent of the phase angle , in a finite system .
this indicates that orientational order in our problem is strongly biased by the drive @xmath12 , if @xmath12 is sufficiently large .
the probability distribution of the hexatic - spins in the complex argand plane formed by the re@xmath134 $ ] and im@xmath134 $ ] axes clearly describes how the external symmetry - breaking field ( @xmath54 ) builds up anisotropy in the our system , thus ordering the spin orientations . at low drive ,
the anisotropy is masked by disorder - induced fluctuations at the scale of our simulation box .
as such fluctuations are increasingly suppressed at higher drive values , the system appears to organize into a coherently moving lattice structure whose principal axes are biased by the force .
the constant part in the angular correlation function along the drive direction goes quadratically with the driving field strength above @xmath135 as predicted by fertig _ et al._@xcite .
the data points ( circles ) are for @xmath83 and the dashed line is a guide to the eye . , width=302 ]
the spin - spin correlation functions in our model , fig .
[ combined](a ) and fig .
[ combined](b ) , obtained via the correlation functions @xmath136 ( longitudinal ) and @xmath137 ( transverse ) , correspond to the local quantities re@xmath134 $ ] and im@xmath134 $ ] .
they are @xmath138 re[\psi_{6}(r ) ] \rangle\ ] ] @xmath139 im[\psi_{6}(r ) ] \rangle\ ] ] in all the phases ( c ) , ( d ) and ( e ) of fig
. [ phasedia ] , the orientational correlation function @xmath136 saturates asymptotically , as expected , to a constant . in phase ( d ) , such saturation is obtained only in the drive direction . the transition from phase ( d ) to phase ( e ) , when triangular _ translational _ order increases continuously with @xmath54 , appears to be smooth .
the correlation function @xmath137 decays exponentially in all the three phases .
note that our mapping for orientational order and its correlations in the moving state is free from any underlying lattice effects , in contrast to earlier studies of the xy model in a field .
therefore , any anisotropy reflected in the spin distribution ( see fig . [ argands ] ) or their correlation functions reflects an intrinsic property of the particular driven phase .
the quadratic dependence of the asymptotic saturation value of @xmath136 on the applied field @xmath54 within the fluid ( plastic flow ) regime , as predicted by this intuitive mapping , is shown in fig .
[ specop ] .
the results presented here address one fundamental issue in the literature on orientational order in driven disordered states , in particular the question of whether a quasi - long - range ordered phase , the `` hexatic '' can exist .
we show here conclusively that it can not .
what exists is an unusual intermediate state which possesses long - range orientational order in the direction singled out by the drive whereas orientational order decays exponentially in the transverse direction .
what happens at still larger force values depends on the value of @xmath9 . for small @xmath9 , the system transits directly from anisotopic hexatic glass to triangular moving crystal .
the bragg peaks sharpen into sharp bragg spots with six - fold symmetry . at intermediate values of @xmath9 ,
the system appears to undergo an unusual transition into what we term a `` coexistence phase '' , discussed in more detail below . in this phase ,
the system has both triangular and square domains and interconverts between them over a broad distribution of time - scales .
inspection of configurations suggests an analogy to equilibrium phase coexistence with a large , heterogeneous distribution of domain sizes of triangular and square regions , although this is an explicitly non - equilibrium system . at larger values of @xmath9
, the coexistence regime appears bounded .
however , the disordered , plastic flow regime is observed to expand . at these values of @xmath9
the system undergoes a direct transition into the square phase . for much larger @xmath9 ,
the phase boundary between plastic and square phases collapses again with increasing @xmath9 , reducing the extent of the plastic regime . in this large @xmath9 regime ,
the system depins discontinuously and elastically from a pinned to moving square crystal with no intervening plastic flow phase that our numerics can resolve .
the plastic flow regime ( c ) , as well as that of the hexatic glass ( d ) expands at larger @xmath9 due to the frustration of local triangular translational order by three - body interactions . on further increasing @xmath54 ,
the structure obtained depends on the value of @xmath9 : for low @xmath9 the final crystal is triangular ( e ) whereas for large @xmath9 it is square ( f ) .
we assign these states through a study of the structure factor @xmath93 , as well as the coordination number probability distributions shown in fig .
[ coordno ] , observing that the slighly smeared six - fold coordination of the hexatic glass consolidates into sharp bragg - peaks across the transition into the ordered states , as shown in fig .
[ sqfigs ] . for intermediate @xmath9 and @xmath12 ,
the system exhibits a remarkable `` coexistence '' regime ( g ) best described as a mosaic of dynamically fluctuating square and triangular regions . from a direct calculation of the structure factor we see the simultaneous appearance of peaks corresponding to hexagonal and square order@xcite .
the intensity of the peaks from the hexagonal and square phases are comparable . as @xmath12
is increased , clusters of @xmath140-coordinated particles grow in size .
the evolution of the configuration with increasing force is shown in fig .
[ coex ] , which illustrates the square - triangle domain mosaic characteristic of the coexistence regime .
the @xmath141-coordinated regions decrease in size and the dynamics of the interfacial region - with predominantly @xmath87-coordinated particles and with a few isolated @xmath62-coordinated particles - shows enhanced and co - ordinated fluctuations . real space configurations ( fig .
[ coex ] ) exhibit islands of square and triangular coordination connected by interfacial regions with predominately 5 coordinated particles .
the configuration , as viewed in the co - moving frame , is extremely dynamic , with islands rapidly interconverting between square and triangle .
this interconversion has complex temporal attributes . :
the shaking temperatures in the transverse ( top ) and drive ( bottom ) directions ( @xmath142 and @xmath143 ) for different @xmath9 as a function of the centre of mass velocity ( @xmath144 ) , showing the quantity is almost independent of @xmath9 except in and around the coexistence regime , due to effects of fluctuation . , title="fig:",width=321 ] : the shaking temperatures in the transverse ( top ) and drive ( bottom ) directions ( @xmath142 and @xmath143 ) for different @xmath9 as a function of the centre of mass velocity ( @xmath144 ) , showing the quantity is almost independent of @xmath9 except in and around the coexistence regime , due to effects of fluctuation .
, title="fig:",width=321 ] as substrate randomness is averaged out due to the motion of the particles , the shaking temperature of the system can reasonably be expected to decrease .
this is reflected in the decrease of the width in velocity component distributions .
the `` shaking temperature '' predictions of koshelev and vinokur , would indicate a @xmath145 and @xmath146 nature of the fall in the transverse and drive directions respectively .
our results for @xmath80 are in agreement with this prediction _ outside _ the coexistence regime .
we find , as shown in fig .
[ tempsxy ] , that @xmath80 is nearly independent of @xmath9
. however , within the coexistence regime , @xmath80 behaves non - monotonically .
typically , for a particular disorder configuration and for @xmath147 , @xmath80 appears to increase sharply at a well defined @xmath54 , signifying the start of coexistence . within g ,
@xmath80 remains high but drops sharply at the upper limit of g , to _ continue to follow the interrupted kv behavior_. this anomalous enhancement of fluctuation magnitudes provides strong evidence for a genuine coexistence phase , since increasing the driving force would be expected to _ reduce _ current noise monotonically once the system depins , as observed in all previous simulation work on related models @xcite .
the limits of the coexistence region , though sharp for any typical disorder realization , vary considerably _ between _ realizations .
clusters within the coexistence state appear through a nucleation and growth mechanism , as modified by the anisotropy induced by the presence of the drive .
[ nucl ] shows that for a particular disorder realization at a force @xmath148 , a nucleus of square region appears which grows with time . in all the disorder realizations we studied , the square nucleus first formed in the hexagonal phase is anisotropic and elongated along the transverse direction ( fig .
[ nucl ] ) , a consequence of the fact that the drive introduces a preferred direction into the system .
square clusters in the coexistence regime tend to be less anisotropic compared to the elongated string - like defects of square within a triangular background obtained at smaller force values .
this can be seen from fig .
[ quenchnucl ] which plots the distribution of angles made , with respect to the drive direction , by the principal direction of the square nucleus , corresponding to the larger eigenvalue of the moment of inertia tensor . for small @xmath12 , the distribution peaks around @xmath149 , indicating that the square nucleus forms preferentially at a @xmath150 angle with respect to the drive . for forces within the coexistence regime , however , the distribution of this angle is smooth and has no sharp peaks , concomitant with the shape of the nucleus becoming more isotropic .
the interconversion between square and triangular regions leads to complex spatio - temporal behavior . to demonstrate this , we compute the power spectrum of fluctuations of the particle current . we first obtain the statistics of the number of particles crossing an imaginary line parallel to the transverse ( @xmath78 ) direction per unit time , per unit length of the line for a particular force ( @xmath54 ) value .
the power spectrum @xmath151 is the fourier transform of the auto - correlation function of this time series of particle flux , averaged over different choices of the position of the imaginary line and over different choices of the time slots of observation .
considerable statistics were taken to ensure that all quantities were well averaged .
about @xmath87 different choices of the position of the transverse imaginary line were taken , in addition to @xmath152 different choices of time slots .
averaging is performed over a large span of simulation time steps ( @xmath153 ) .
the power spectrum @xmath151 of current fluctuations in the moving triangle phase ( open triangles @xmath154 ) and in the coexistence region ( red circles @xmath155 ) , logarithmically binned and plotted for a range of frequencies below the washboard frequency .
current fluctuations in the coexistence region are _ enhanced _ , also showing a @xmath156 decay in the low frequency range .
, width=302 ] our results are summarized in fig .
[ autocor ] for two values of @xmath54 . for @xmath157 , when the system is in the moving triangular phase , @xmath151 is , to a large extent , featureless and flat , except for large @xmath158 where effects due to the streaming of the entire system become important .
in contrast , for a high force @xmath159 , when the system is within the coexistence phase , we obtain a @xmath4 regime , with @xmath160 in @xmath151 over about a decade .
in addition the particle current fluctuations remarkably _ enhanced _ by 3 - 4 orders of magnitude in the coexistence phase as compared to the triangular regime . to investigate the origins of the noise in the coexistence regime
, we define the local quantity @xmath161 where the summation is over a defined region ( @xmath162 ) surrounding a particle i. here @xmath24 is the bond angle between particle @xmath42 and @xmath163 and @xmath164 .
we choose particles @xmath163 and @xmath164 such that they are all within a specified cutoff radial distance from particle i. fig .
[ omegas](a ) displays the probability distribution of @xmath165 for a state in the coexistence region .
there are two prominent peaks .
the peak for small values of @xmath166 corresponds to the square lattice ( @xmath167 for the ideal square crystal ) , whereas the peak at higher values of @xmath166 corresponds to the triangular structure .
intermediate values of @xmath166 are obtained for particles at the interface between square and triangle .
[ omegas](b ) shows a snapshot of the system in the coexistence region with particles color coded according to their value of @xmath166 .
we next examine the fluctuations of the velocity about the average value @xmath168 as a function of the local coordination . in fig .
[ omegas](c ) this is plotted as a function of @xmath166 .
we see that velocity fluctuations are larger in the interfacial region .
further , fig .
[ omegas](d ) shows a plot of the local volume per particle as a function of the velocity fluctuations , showing that the relatively lower density of the interfacial region ( due to the presence of a large concentration of defects ) causes it to fluctuate more than the rest of the solid .
the rapid fluctuations of the interface also results in rapid interconversion between square and triangular coordinated particles and leads also to enhanced fluctuations in the coordination number .
since the driven solid is elastically constrained to move as a whole in the direction of the drive without gaps or cracks , this sets up strong correlations among the particle trajectories .
such correlations are ignored in the theory of the shaking temperature of koshelev and vinokur .
in this section , we turn to a possible application of this model , in the context of experiments on transport anomalies associated with peak effect phenomena in the superconducting mixed phase . to recapitulate , the peak effect refers to the sharp increase in the critical current @xmath1 in the mixed phase of a disordered type - ii superconductor close to h@xmath169 .
this critical current is thus a non - monotonic function of @xmath2 ( or @xmath3 , depending on which is varied experimentally ) , since it decreases steadily from its low temperature value till the onset temperature @xmath170 of the peak effect , from whence it increases sharply over a small temperature range to its maximum value , obtained at temperature @xmath171 .
as @xmath2 is raised still further , the critical current collapses again , to reach zero in the normal state .
the temperature interval [ @xmath172 in which the critical current rises anomalously is the peak effect regime .
this regime is dynamically anomalous , displaying : ( i ) large current noise amplification at low frequency , ( ii ) a @xmath173 spectrum of current fluctuations , which is very non - gaussian , ( iii ) a `` fingerprint '' effect in which apparently random spikes in the differential resistivity as a function of drive are retraced as the drive is decreased , ( iv ) a history - dependent dynamic response ( v ) a memory of direction , amplitude and frequency of applied currents , ( vi ) a strong suppression of ac response by a dc bias as well as a variety of other behaviour@xcite . a very large number of experiments probing such anomalous behavior , including all those referenced above , are transport - based , thus serving as probes of the dynamics of vortices within this narrow region of parameter space .
interpretations of these phenomena are largely phenomenological .
one particularly influential proposal considers an underlying order - disorder transition `` contaminated '' by sample surfaces or `` edges '' .
such surfaces , with associated surface barriers for vortex entry , provide an entry point for vortices driven into flow@xcite .
the surface should provide an intrinsically more disordered environment for vortices than the bulk , particularly in fairly pure samples where @xmath1 is low .
thus , vortices might be expected to enter through the boundaries in a highly disordered state , only to anneal in the nearly pure bulk , when a current is applied across the sample .
this spatial separation of disordered and ordered states and the slow annealing of one into the other is argued to be the central feature underlying the anomalous behavior seen in the peak effect regime .
magneto - optic imaging via hall bar arrays support the surface contamination scenario .
however , such methods do not access the dynamics of annealing and phase transformations directly .
much recent work appears consistent with a bulk coexistence of disordered and ordered phase@xcite , while decoration experiments see a `` multi - domain '' structure in the peak effect regime@xcite , as proposed in refs .
@xcite and accessed indirectly in refs .
@xcite . for related simulations ,
see ref .
@xcite .
the edge contamination scenario implicitly assumes that the underlying order - disorder transition is unaffected by the drive , serving only to provide a background to the annealing process .
however , in a generic driven system , the possibility that the drive has more non - trivial effects must be expected . in particular , the drive may alter the very nature of the driven bulk , stabilizing dynamical states that are truly non - equilibrium in character , as illustrated in the simulations discussed above . a beautiful recent experiment ( ref .
@xcite ) performs a variant of scanning probe microscopy , using a mounted local hall probe .
the probe is sensitive to variations in the local magnetic induction averaged across a mesoscopic scale of around a micron .
the system is tuned across the peak effect regime and then perturbed weakly through a low - amplitude ac field applied from below the sample .
the hall probe , placed above the sample and linked , though lock - in techniques to the frequency of the ac perturbation , records a local susceptibility , indicative of pinning response , as a function of space and integrated over the thickness of the sample .
the spatial resolution is limited by the size of the hall probe , typically of the order of a micron or so in size .
as parameters are varied across the peak effect , these experiments see a remarkable coexistence between a strong pinning regime and a weak pinning regime .
a complex interface is seen between these coexisting states with a dynamics which is exquisitely sensitive to the field and the disorder , such coexistence is also a feature of other phenomenological approaches to this problem , which address transport measurements .
locally more disordered regions of the sample appear to nucleate more stable regions of strong pinning whereas small variations of the applied field cause large changes in the inhomogeneous pinning pattern .
however , the complex geometry of the coexisting regimes appears largely stable if the temperature and field are fixed , suggesting that thermal fluctuations are not dominant . while vortices entering from the sample boundaries
do appear to contribute to this dynamics in no small measure , there is significant evidence for non - trivial dynamics in the bulk , with regions of strongly pinned phase being nucleated far from any boundary .
thus , these experiments point to a more active role for the bulk than envisaged in the boundary injection scenario . in this context , the authors of ref . @xcite
have specifically argued that the complex topology of the two - phase interface should be largely responsible for the history dependence seen in the experiments .
how are these remarkable observations related to the model we study here ?
we suggest that the link is the _ emergence of a self - organized , disorder - stabilized , dynamically sustained drive - induced coexistence phase _ seen both in the experiments and in simulations of our model system .
the similarities between the two are striking : first , the coexistence itself . both the experiments and our simulations here provide incontrovertible evidence for dynamical states in driven disordered systems which resemble phase coexistence at equilibrum phase transitions , with the added complication of spatial inhomogeneities due to quenched disorder@xcite .
reasoning from the experiments , the complex interconversion of one phase into the other and the spatially inhomogeneous character of the dynamics is the hallmark of vortex dynamics within the peak effect regime@xcite .
this is precisely the situation which obtains in the simulations . as pointed out in the previous section ,
what is unusual about the coexistence regime is that the strong disorder - induced fluctuations seen and manifest in all the dynamic properties we measure are obtained _ above _ the depinning transition , surviving even at large values of the applied force .
second , the coexistence seen in the experiments is very disorder - sensitive@xcite . in experiments ,
this is manifest in terms of the complex structure of differentially pinned regions in the sample , presumably reflecting a non - trivial pinning landscape .
we see similar dynamical behavior within the coexistence regime , with the structure at fixed drive influenced by the underlying microscopic disorder and sensitive to even marginal changes in the pinning .
thus , the nature of inhomogeneities connected to dynamical phase coexistence in this model as well as in the experiments appears to be dictated primarily by the underlying disorder .
third , the slow dynamics and long relaxation times seen in the experiments , reflecting the complex dynamics of the interface separating the regimes of different pinning strength , is seen in our simulations as well@xcite . in our simulations , if the drive is switched off and the system allowed to anneal , the nuclei of one phase within the other live anomalously long , suggesting that the dynamics has stabilized long - lived metastable states , precisely as seen in the experiments . in fig .
[ quenchnucl2 ] we show the dynamics of a square nucleus within a triangular background , extracted from a typical configuration within the coexistence regime , when quenched to zero drive .
far from vanishing over a short time scale , the nucleus appears to be remarkably stable out to the longest time scales accessed in our simulations , providing evidence that the metastability exhibited in the coexistence regime survives even after the drive is turned off .
these are reflected in the memory experiments , in which the removal of the driving transport current essentially appears to freeze the system into a metastable state from which it only recovers upon a reapplication of the drive .
fourth and finally , the noise spectra within the coexistence regime , including the anomalously large noise and the power - law fall - off with a @xmath4 spectrum is a prominent feature of the peak effect regime .
experiments see a power - law falloff , with @xmath174 , as well as a substantial enhancement in noise power of a few orders of magnitude@xcite .
the substantial non - gaussian features in the noise , as obtained in ref .
@xcite , indicate that a small number of fluctuators contribute ; it is tempting to assign these to a few large domains which fluctuate collectively within the coexistence regime , as suggested by the simulations .
the edge contamination scenario _ assumes _ that the drive in the bulk anneals the disordered vortices injected across the boundaries into the smoothly flowing state .
in contrast , we show that in the vicinity of an underlying transition , the bulk flowing state is generically inhomogeneous and dynamically non - trivial , thereby questioning this basic assumption .
we believe that the edge contamination scenario should apply , more immediately , close to but away from the peak effect regime , where the drive in the bulk acts solely to smoothen flow , precisely as seen in a recent experiment@xcite .
we stress that the system we simulate is far from a literal translation of the vortex system .
it lacks the @xmath175 , ( with @xmath176 the penetration depth ) , interactions of vortex lines , which can become fairly long - range if @xmath176 is large . in our simulations
we drive transitions between the two different pure system phases by varying a parameter @xmath9 , whereas these transitions are density or temperature driven in the classic scenario of the peak effect in vortex systems .
( the underlying order - disorder transition in the vortex system is likely related to a melting - like transition , at least for small @xmath3 and for temperatures close to @xmath177 and not a structural transition between two crystalline phases as in our model@xcite . )
the experiments are in three dimensions , whereas our simulations are two - dimensional .
our system shows very little variation in the depinning force close to the transition , and thus no apparent peak effect .
( it is known , however , that factoring in the temperature dependence of the coherence length @xmath178 and @xmath176 is essential to obtain a peak effect in two - dimensional simulations@xcite . ) for these , as well as other reasons , our proposal for the origin of peak effect anomalies should be thought of as indicating a generic scenario within which much of the observed physics finds a common explanation
. clearly more work , including simulations of the three - dimensional case capable of describing vortex entanglement , is required to further illuminate the relationship conjectured here .
this paper has studied the complex dynamical behavior of a two dimensional driven disordered solid which undergoes a square to triangular structural transition as a single parameter @xmath9 is tuned .
our interest in this model has several origins : the problem of dynamical probes of an underlying static phase transition in a weakly disordered system is interesting in itself .
the possibility that such behavior might underlie a classic and ill - understood problem in the literature on type - ii superconductivity , the problem of the origin of peak - effect anomalies , provides further motivation for this study . specifically , using the initial study of ref .
@xcite and this paper , we have demonstrated the following : first , the existence of a _ complex phase diagram containing a large number of states induced purely by the drive _ , such as the anisotropic hexatic state and the coexistence state .
second , the demonstration of _ dynamically unusual behaviour in the coexistence region _ , as measured through several dynamical quantities , including the observation of @xmath4 correlations in the current noise , the existence of highly metastable states and the observation of a very substantial noise enhancement . third , the _ breakdown of single - particle , effective temperature descriptions in the coexistence regime _ , where measures of a flow and disorder induced temperature indicate substantially non - monotonic variation within the coexistence regime .
fourth , the demonstration that the _ putative driven hexatic glass phase can not have algebraically decaying orientational correlations _ , since correlations along the drive direction are always long - ranged while those along the transverse direction are generically short - ranged .
fifth , the intuitively unexpected _ expansion of the plastic flow regime at large @xmath9 and its subsequent collapse_. finally , we have suggested a _
possible connection to the classic problem of the origin of peak effect anomalies _ , conjecturing that the explanation might be found in the very nature of the unusual dynamical states obtained when a system close to a first - order ( structural or melting ) phase transition is driven across a quenched - disordered background .
there are several realizations of non - equilbrium states of complex fluids , such as sheared lamellar phases , worm - like micelles , and driven colloidal suspensions , which exhibit remarkably non - trivial behaviour as a consequence of dynamical phase transitions@xcite .
possible relations to equilibrium descriptions of systems at phase coexistence have also been outlined@xcite .
many features , including profoundly non - linear response , complex spatio - temporal behaviour and large noise signals appear to be common to these systems when driven away from equilibrium@xcite . what singles out the category of systems we study here is the additional complication of quenched disorder .
clearly these are suggestive links at the intersection of these various fields .
further exploration of these connections would appear to be fruitful .
the authors thank s. bhattacharya , d. dhar and c. dasgupta for discussions .
in addition , gim is grateful to s. bhattacharya for many interactions over the years concerning the phenomenology of the peak effect .
the work of all three authors was supported in large part by the dst ( india ) . as thanks s. n. bose national centre for basic sciences for computational facilities and financial support , and gratefully acknowledges ongoing support from sfb tr6 ( dfg ) . | we study a simple model for the depinning and driven steady state phases of a solid tuned across a polymorphic phase transition between ground states of triangular and square symmetry .
the competition between the underlying structural phase transition in the pure system and the effects of the underlying disorder , as modified by the drive , stabilizes a variety of unusual dynamical phases .
these include pinned states which may have dominantly triangular or square correlations , a plastically flowing liquid - like phase , a moving phase with hexatic correlations , flowing triangular and square states and a dynamic coexistence regime characterized by the complex interconversion of locally square and triangular regions .
we locate these phases in a dynamical phase diagram and study them by defining and measuring appropriate order - parameters and their correlations .
we demonstrate that the apparent power - law orientational correlations we obtain in our moving hexatic phase arise from circularly averaging an orientational correlation function which exhibits long - range order in the ( longitudinal ) drive direction and short - range order in the transverse direction .
this calls previous simulation - based assignments of the driven hexatic glass into question .
the intermediate coexistence regime exhibits several novel properties , including substantial enhancement in the current noise , an unusual power - law spectrum of current fluctuations and striking metastability effects .
we show that this noise arises from the fluctuations of the interface separating locally square and triangular ordered regions by demonstrating a correlation between enhanced velocity fluctuations and local coordinations intermediate between the square and triangular .
we demonstrate the breakdown of effective `` shaking temperature '' treatments in the coexistence regime by showing that such shaking temperatures are non - monotonic functions of the drive in this regime .
finally we discuss the relevance of these simulations to the anomalous behaviour seen in the peak effect regime of vortex lines in the disordered mixed phase of type - ii superconductors .
we propose that this anomalous behavior is directly linked to the behavior exhibited in our simulations in the dynamical coexistence regime , thus suggesting a possible solution to the problem of the origin of peak effect anomalies . |
although a scalar field theory is often used only as a first approximation towards a more precise description of a physical system , it can nevertheless reveal significant properties of the system . among many features of scalar field theories that have been extensively studied ,
those describing special kinds of solutions solitons , instantons and bounces are particularly interesting .
bounces , for example , are related to stability properties of classical and quantum configurations in such theories .
this relation exists because a stable classical state may be only metastable quantum - mechanically .
the instability is realized by allowing a metastable state to tunnel to a stable state via a quantum barrier penetration or to thermally climb over the barrier in order to arrive at the stable state .
these processes are widely used in describing various physical systems ranging from phase transitions in solids to bubble formation in cosmological inflaton fields @xcite . due to its prevalence in theoretical models , the tunneling in scalar field theories should be thoroughly understood and accurate approximation methods for its equations should be developed . toward the latter goal ,
we specifically focus on approximation methods for the bounce in euclidean field theories .
the bounce itself and the first approximation for it ( called the thin - wall approximation ) was introduced by coleman @xcite .
callan and coleman @xcite developed the first quantum corrections for the bounce .
coleman , glaser and martin @xcite proved that , for a wide class of potentials , spherically - symmetric solutions to equations of motion are the solutions with the lowest action .
coleman and de luccia @xcite considered modification to the thin - wall approximation due to gravitational effects and showed that increasing the effects of gravity can render the false vacuum stable . since these seminal works on the bounce , significant progress has been made in improving accuracy and generality of the thin - wall approximation , for example , for a restricted class of polynomial potentials @xcite and non - polynomial potentials @xcite .
quantum corrections for the bounce were also enhanced and the decay rate of the false vacuum was obtained , for example , in the one loop effective action calculations @xcite , @xcite .
further studies showed intricate properties of gravitational bounces , @xcite , @xcite . here
we propose a new method of approximate solutions for the multidimensional spherical bounce .
the method starts with the thin - wall approximation as its first step and proceeds to higher orders iteratively with fast convergence and high accuracy .
analysis of the problem from a new perspective demonstrates some universal properties of the bounce .
the method is not restricted to only certain types of potentials or dimensions of space , and we demonstrate its computational power with the general fourth - order polynomial potential .
we find that the approximation works well beyond its intended range of applicability .
consider a scalar field theory defined by the action @xmath0 , \label{action}\end{aligned}\ ] ] where @xmath1 is a scalar field , @xmath2 is a potential function ( which we assume to be continuously differentiable ) , @xmath3 is the gradient operator in @xmath4 , and @xmath5 is the norm in @xmath4 .
the corresponding euler - lagrange equation is @xmath6 where @xmath7 is the laplace operator in @xmath4 .
let @xmath8 be a minimum of @xmath9 .
for the theory defined by the action @xmath10 , the solution @xmath8 is classically stable , but its quantum stability depends on the type of the minimum . if the minimum is absolute , the solution is stable and is called a true vacuum ; if the minimum is relative , the solution is unstable and is called a false vacuum . at zero temperature , which we assume throughout our analysis , a false vacuum decays into a true vacuum by the process of barrier tunneling . to study the simplest example of such tunneling , we choose @xmath9 with two minima and one maximum , set the absolute minimum at @xmath11 , the relative minimum at @xmath12 and the relative maximum at @xmath13 . for computational convenience , and without any loss of generality ,
certain conditions can be imposed on the function @xmath9 .
to derive them , we start with the analog of for the variables @xmath14 , @xmath15 and transform these to the variables @xmath16 according to @xmath17 where @xmath18 and @xmath19 are constants , and @xmath20 , @xmath21 and @xmath22 are nonzero constants . as a result ,
becomes @xmath23 which coincides with if the constraint @xmath24 holds .
as we have freedom to choose five coefficients @xmath20 , @xmath18 , @xmath21 , @xmath19 and @xmath22 subject to this constraint , this is equivalent to having freedom to impose four independent conditions on the function @xmath9 .
for the first two conditions , we set @xmath25 and @xmath26 to take particular values , and it is convenient to choose @xmath27 and @xmath28 . for the third condition , we set @xmath29 . finally , for the fourth condition , we set @xmath30 the reason for the form of which will become clear in sec .
[ subsection_orders ] .
we denote @xmath31 for some @xmath32 and write as @xmath33 if the condition is satisfied without the fourth restriction on the function @xmath9 ( as is the case for the general fourth - order polynomial potential considered in sec .
[ section_polynomial ] ) , we can impose one additional condition on @xmath9 .
see fig .
[ figure_u_phi ] for examples of @xmath9 . of different degrees of asymmetry with two minima and one maximum . to simplify comparison of the resulting solutions , we choose , without any loss of generality ( see sec . [ section_bounces ] ) , the absolute minimum at @xmath11 , the relative minimum at @xmath12 and @xmath29 for all potentials .
we set the relative maximum at @xmath13 and the turning point at @xmath34 , where @xmath35 ; for @xmath36 , we have @xmath37 .
the energy - density difference between the true and false vacua @xmath38 characterizes the degree of asymmetry of @xmath9 .
the quantities @xmath39 , @xmath40 and @xmath41 are labeled only for one curve .
the potential functions shown here are precisely those for which the corresponding solutions and their approximations are shown in the right part of this figure and in figs .
[ figure_u_tilde_0 ] and [ figure_phi_0 ] .
right : the spherical half - bounce solutions for @xmath42 and the potentials in the left part of this figure .
the solid curves are for the exact numerical solutions and the dashed curves are for the approximate analytic solutions given by , , and .
the quantities @xmath39 , @xmath40 and @xmath43 are labeled only for one curve , which corresponds to the curve labeled in the left part of this figure . each color of curves in this figure and in figs .
[ figure_u_tilde_0 ] and [ figure_phi_0 ] represents the same value of @xmath44 . ]
a solution @xmath45 of is called a bounce if it satisfies the boundary conditions @xmath46 where @xmath47 .
these conditions mean that the field starts at the false vacuum @xmath12 at @xmath48 , reaches the turning point @xmath34 at @xmath49 , bounces back and finally reaches the false vacuum at @xmath50 .
we are interested in a bounce for which the action is finite .
in addition to @xmath29 that we have set above , finiteness of the action also requires @xmath51 for a large class of potentials @xmath9 , spherically symmetric solutions of are the solutions with the lowest action @xcite and we will be concerned here only with these .
a function @xmath52 of the euclidean distance @xmath53 satisfies the equation @xmath54 and the boundary conditions @xmath55 the solution of , and is a spherical half - bounce since the field starts at the turning point @xmath34 at @xmath56 and reaches the false vacuum @xmath12 at @xmath57 . upon carrying out the angular integration
, the action becomes @xmath58 , \
a=\frac{2\pi^{(n+1)/2}}{\gamma((n+1)/2 ) } , \label{action_r}\end{aligned}\ ] ] where @xmath59 is the area of an @xmath60-dimensional unit sphere .
the classical analog of is a particle moving in the potential @xmath61 and subject to a viscous damping force ; see e.g. @xcite .
the viscous damping always dissipates energy and so for the bounce the field at the turning point @xmath34 still has lower potential energy than the final field @xmath26 , i.e. , @xmath62 , but also @xmath63 and @xmath64 ; see fig .
[ figure_u_phi ] .
in this section we derive the exact solution of the field equation for @xmath36 , @xmath66 although elementary , it serves as a starting point for our approximation solution of for @xmath67 for which no exact solution is known .
we first note that according to , the particle moves without viscous damping , so that its energy is conserved . multiplying both sides of by @xmath68 , integrating over @xmath69 and using the boundary condition together with @xmath31
, we find @xmath70 this implies @xmath71 where we have chosen the positive sign of the square root for the positive half - bounce for which @xmath72 . the integration of with the boundary condition @xmath73 ( because @xmath37 for @xmath36 ) now gives @xmath74 since @xmath75 is a monotonically increasing function , we can define the radius @xmath76 ( so that @xmath77 ) and write in the form which is more convenient for subsequent calculations , @xmath78 equation is the one - dimensional instanton centered at @xmath79 , for which the action becomes @xmath80 where we used twice .
while there is no exact solution of the field equation for @xmath81 , one approximate solution is well - known @xcite .
this so - called thin - wall approximation applies when the potential function @xmath9 is nearly symmetric and consequently the energy - density difference between the false and true vacua @xmath38 is small .
the small asymmetry of the function @xmath9 implies that viscous damping for the particle from the mechanical analogy is small and that the turning point for the bounce is near the absolute minimum , @xmath82 .
it turns out that @xmath83 is exponentially small in @xmath44 .
the particle moving in such a potential spends a long time in the neighborhood of @xmath34 before it crosses the potential valley .
the crossing happens somewhere between @xmath34 and @xmath12 and it is convenient to take @xmath84 as the center of the valley ; see the left part of fig .
[ figure_u_phi ] .
it is also said that the wall separating the regions of the false and true vacua is located at @xmath79 ; see the right part of fig .
[ figure_u_phi ] .
since it takes a long time for the particle to reach the wall , it follows that @xmath43 is large for small @xmath44 ; indeed , we will find that @xmath85 .
( note that for @xmath86 the particle does not have enough energy to overcome the friction and to reach the local minimum at @xmath12 ; it oscillates around @xmath13 and finally stops there . for @xmath87 ,
the particle arrives to @xmath12 with positive energy at a finite @xmath69 , at which point it starts accelerating towards @xmath88 . ) finally , after crossing the potential valley , the particle spends a long time approaching @xmath12 as @xmath89 . to prove the above qualitative statements , in the rest of this subsection we solve separately for small @xmath69 and large @xmath69 and
match the two solutions at @xmath79 . for small @xmath69 we have @xmath90 and
so we expand @xmath91 for this potential the solution of satisfying the boundary condition @xmath92 is @xmath93 where @xmath94 and @xmath95 is the modified bessel function of the first kind .
since @xmath83 is exponentially small in @xmath44 ( which we prove below ) , the function changes very slowly near @xmath56 .
in other words , the particle spends a long time in the neighborhood of @xmath34 before it crosses the potential valley . for large @xmath69
we have @xmath96 and so we expand @xmath97 for this potential the solution of satisfying the boundary condition @xmath98 is @xmath99 where @xmath100 is a constant and @xmath101 is the modified bessel function of the second kind . as a result ,
@xmath45 approaches its asymptotic value @xmath26 exponentially slowly . matching the solutions and and their derivatives at @xmath79
, we find @xmath102},\label{phi_0_constant}\\ b&\approx\frac{(\phi_{+}-\phi_{-})k_{-}(k_{+}r)^\nu i_{\nu+1}(k_{-}r)}{k_{-}i_{\nu+1}(k_{-}r)k_\nu(k_{+}r)+k_{+}i_{\nu}(k_{-}r)k_{\nu+1}(k_{+}r)}.\label{b_constant}\end{aligned}\ ] ] since @xmath43 is large for small @xmath44 ( which we prove in secs .
[ subsection_thin_wall_approximation ] and [ section_method ] ) , we need the expansions for large @xmath103 and @xmath104 , @xmath105 as stated above , the difference @xmath83 is indeed exponentially small for large @xmath104 . to estimate the time it takes for the particle to cross the potential valley
, we compute @xmath106 and see that the transition mostly occurs over the interval @xmath107 which is much smaller than the length of the interval , @xmath108 , over which the particle moves between the two vacua .
in other words , relative to the whole transition , the passage through the potential valley is very fast .
the approximate solution given by , , and is a very poor approximation except for very small @xmath44 ; see the right part of fig .
[ figure_u_phi ] for examples .
one direct method to improve it is to include in and terms of higher order in @xmath109 and @xmath110 , respectively , but , unfortunately , no exact solutions of are known for such potentials .
it appears that an alternative method of treating higher order terms in @xmath9 as small perturbations leads to an approximation which is less accurate and more complicated than the solution we derive in the following sections .
one reason for this is that we can view the above approximations as local ( due to their reliance on series expansions around either @xmath11 or @xmath12 ) , while the thin - wall approximation and our generalization of it have features of global solutions which do not give preference to any particular value of @xmath45 . to continue with the thin - wall approximation
, we now choose radii @xmath111 and @xmath112 which are close to @xmath43 and satisfy @xmath113 , and proceed with solving in three separate regions : @xmath114 , @xmath115 and @xmath116 .
for @xmath114 we have @xmath117 . for @xmath115
we ignore the friction term @xmath118 in since viscous damping is small .
furthermore , since asymmetry of @xmath119 is small , we replace @xmath119 with its even part @xmath120 and find @xmath121 according to , the particle s motion is approximately the motion in the potential @xmath122 without viscous damping , so that its energy is approximately conserved .
we can now proceed as in sec .
[ section_exact ] with only small changes due to different boundary conditions .
since @xmath123 , instead of we have @xmath124 integrating with the boundary condition @xmath125 , we find @xmath126 for @xmath116
we have @xmath127 .
to compute the action , we consider contributions to the integral in from the three regions used above , @xmath128 .
\label{action_approximation_sum}\end{aligned}\ ] ] for @xmath114 we have @xmath129 and @xmath130 and find that this region contributes approximately @xmath131 to , where we set @xmath132 .
for @xmath115 we use @xmath133 and @xmath134 and find that this region contributes approximately @xmath135 to .
now setting @xmath136 in the integrand , changing to the integration over @xmath45 , using @xmath137 together with @xmath138 and @xmath139 , we find that the region @xmath115 contributes approximately @xmath140 to .
the contribution of the region @xmath116 to the action can be ignored since @xmath129 and @xmath141 there . combining the above results , we obtain @xmath142 the wall location @xmath79 can be determined from if we know @xmath39 .
alternatively , we can find it by maximizing @xmath10 in with respect to @xmath43 .
solving @xmath143 for @xmath43 , we find @xmath144 substitution of which into finally gives @xmath145^{n+1}. \label{action_approximation}\end{aligned}\ ] ] equations and represent the thin - wall approximation for the bounce @xcite .
when compared with standard perturbation methods for differential equations , the thin - wall approximation is rather irregular in its derivation . despite the presence of a small parameter in the problem , it is not immediately clear how to proceed with the derivation of higher order corrections . in the following section we develop a systematic approximation scheme which includes the thin - wall approximation as its first iteration .
as we saw in secs .
[ section_exact ] and [ section_thin_wall_approximation ] , the thin - wall approximation for the solution of for @xmath81 can be obtained from the exact solution of for @xmath36 with minimal changes . to develop an iterative method for generating approximate solutions of for @xmath81 , we effectively reduce the problem to the case @xmath36 by rewriting in the form @xmath146 since @xmath45 in the spherical half - bounce solution is restricted to the interval @xmath147 $ ] , the effective potential @xmath148 introduced via is defined on the same interval . adding an arbitrary constant to @xmath148 does not change , and we conveniently choose the constant such that @xmath149 similarly to @xmath150 that we set in sec .
[ section_bounces ] .
since the particle moves in the potential @xmath151 without viscous damping , its energy is conserved ; this implies @xmath152 since @xmath153 .
equations and lead to @xmath154 since @xmath155 and @xmath68 approach zero and @xmath69 approaches infinity when @xmath45 goes to @xmath26 , implies that @xmath156 .
it also follows that @xmath157 for @xmath158 since we consider only positive half - bounces for which @xmath159 . from @xmath160
we now conclude that @xmath161 for @xmath162 .
since the exact solution of is available , the similarity between and leads directly to the iterative solution of .
we proceed as in sec .
[ section_exact ] with only small changes due to different boundary conditions for and . using the boundary condition together with @xmath152 ,
we find @xmath163 instead of and .
the equations and would completely solve the problem of finding @xmath164 for a given @xmath119 if it were not for the need to determine @xmath165 without knowing @xmath164 .
we set out towards an eventual resolution of this difficulty by first examining the relationship between @xmath9 and @xmath148 more closely . integrating over @xmath45 , substituting @xmath166 and using the boundary values @xmath29 and @xmath153 , we find @xmath167 now leads to @xmath168^{-1 } , \label{u_u_tilde}\end{aligned}\ ] ] which together with gives @xmath9 directly in terms of @xmath148 ; unfortunately , we need to reverse this procedure and find @xmath148 in terms of @xmath9 .
one way to arrive at a formula expressing @xmath148 in terms of @xmath9 is to use @xmath166 in to rewrite it in the form @xmath169 equation is a first - order linear differential equation for @xmath170 with the general solution @xmath171 , \label{u_tilde_constant}\end{aligned}\ ] ] where the integration constant @xmath172 can be found as follows .
equations and give @xmath173 substitution of which into leads to @xmath174 , \
r\to 0 .
\label{}\end{aligned}\ ] ] as a result , @xmath175 now requires @xmath176 .
we finally arrive at @xmath177 where the second form , obtained from the first form by integration by parts , might be more convenient for calculations .
although appears to express @xmath148 in terms of @xmath9 , unfortunately , it also requires the function @xmath164 , which itself can be found only when @xmath148 is known ; to avoid circular reasoning here , we can not solve for @xmath164 since @xmath148 in is an instrument towards @xmath164 via .
returning now to , we first notice that it together with directly gives @xmath9 in terms of @xmath148 , but since our goal is to find the inverse operation , we face a non - linear integral equation for @xmath148 . despite its complexity , is particularly suitable for developing an iterative method for finding approximations of @xmath148 in terms of @xmath9
. a naive method of solving by iterations does not work .
indeed , if we ignore the second term on the right - hand side of , we find the zeroth - order approximation @xmath178 . now substituting this approximation into the right - hand side of
, we find the first - order approximation @xmath179 while gives in the zeroth - order approximation @xmath180 since the energy of the analogous classical particle moving in the potential @xmath61 is dissipated for the motion from @xmath39 to @xmath26 , it follows that @xmath181 .
continuity now implies that @xmath182 for some @xmath183 , and for such @xmath45 the square roots in and will be complex - valued , which is not allowed .
comparing this situation with the thin - wall approximation in sec .
[ section_thin_wall_approximation ] , where the even part of @xmath119 appeared , we see the need to introduce the even and odd parts of the potential functions , @xmath184 and rewrite as @xmath185^{-1}. \label{u_plus_minus}\end{aligned}\ ] ] since the function @xmath148 is defined via only on the interval @xmath147 $ ] , it follows from @xmath186 and @xmath27 that the functions @xmath187 are defined via only on the interval @xmath188 $ ] for @xmath189 and can not be defined at all for @xmath190 . despite this
, we will extend the domain of the function @xmath148 to the whole real line by considering the integral equation for @xmath187 as the definition of @xmath187 .
we will see in the rest of this section that this extension does not lead to problems when finding @xmath45 through @xmath148 by inverting since we restrict the domain of the function @xmath75 to @xmath147 $ ] to obtain only physically meaningful half - bounce solutions . in what follows we will separate even and odd functions in with the help of the identities @xmath191 which hold for any even function @xmath192 and any odd function @xmath193 .
we choose the energy - density difference between the false and true vacua @xmath38 as a small positive parameter and set @xmath194 to develop a perturbation theory for which the thin - wall approximation is the first term in the expansion in terms of @xmath44 , we consider power series expansions @xmath195 the @xmath44-dependence of @xmath196 is in accordance with the relation @xmath85 in .
we note that we will find @xmath43 iteratively directly from ; in contrast , the thin - wall approximation in sec .
[ section_thin_wall_approximation ] relied on maximizing the action with respect to @xmath43 .
once the iterative expansions for @xmath197 and @xmath43 are found , we proceed to the corresponding expansions for the bounce solution @xmath164 . to this end
, we first expand the function @xmath198 into a power series , @xmath199 and find each term @xmath200 recursively from .
we can stop here if the bounce solution in terms of the inverse function @xmath198 is sufficient for our purposes , but we can also proceed to finding the direct function @xmath201 ( which is the inverse of the function @xmath198 ) iteratively at the cost of slightly reducing the accuracy .
namely , we seek @xmath201 as a power series @xmath202 here @xmath203 is the inverse of the function @xmath204 and @xmath205 for @xmath206 are found recursively from the identity @xmath207 .
since this step depends on finding an analytic expression for @xmath208 , it can not be done for an arbitrary @xmath9 and we will perform it only for the specific potential function considered in the example in sec .
[ section_polynomial ] .
we make a note on the notation in the proceeding paragraph .
we distinguish the inverse function in the half - bounce solution @xmath198 from the generic variable @xmath209 appearing in the derivation of the half - bounce solution ; similarly , @xmath201 is the direct function in the half - bounce solution and @xmath210 is the generic variable .
the specific functions @xmath211 and @xmath212 will appear again ( through their expansions in terms of @xmath213 and @xmath214 ) only in the end of this section and in secs .
[ section_polynomial ] and [ section_discussion ] .
let us return to the iterative solution of .
since @xmath9 and @xmath148 are nearly equal and the asymmetry of @xmath9 is small , we need to set @xmath148 to be an even function in the zeroth order to start the iteration of the approximating sequence , @xmath215 to extract @xmath216 and @xmath217 terms from , it is enough to keep only the term @xmath218 in the brackets there .
although the resulting factor @xmath219 makes any terms in @xmath220 smaller than @xmath216 irrelevant for this approximation order , we nevertheless keep the term @xmath217 in @xmath220 to avoid the problem of complex - valued square roots as in and . with these steps ,
becomes @xmath221 where we have used and .
the terms @xmath216 and @xmath217 in give @xmath222 respectively .
the integral equation can be trivially solved for @xmath223 . indeed ,
since the right - hand side of does not depend on @xmath45 , it means that @xmath224 for some infinitesimal constant @xmath225 , substitution of which into gives @xmath226 the term @xmath217 in gives @xmath227 the @xmath217 constant @xmath225 is arbitrary and a choice for its value effects terms of all orders in our expansions .
we require @xmath228 so that the square roots in and are real - valued .
we find from that @xmath229 , which implies that @xmath230 holds for any value of @xmath225 since @xmath231 as we have seen in sec .
[ section_bounces ] , no generality is lost upon choosing @xmath9 to satisfy , so that becomes @xmath232 . from now on we set the value of @xmath225 to its lower bound , @xmath233 one of the reasons for this choice is that now @xmath234 , which is analogous to @xmath235 ( although , more generally , @xmath236 for any @xmath45 )
. equations , , , , and give the zeroth - order and first - order approximations , which coincide with the thin - wall approximation . to proceed to terms @xmath237 in , we expand @xmath238 to terms @xmath239 , @xmath43 to terms @xmath240 and @xmath241 to terms @xmath240 .
let us work through the cases with @xmath242 , for which we need the expansion @xmath243{\nonumber}\\ & \times\frac{1}{r_0}\biggl[1-\frac{1}{r_0}\biggl(r_1+\int_{\phi_*}^{\phi_1}\frac{d\phi_2}{p(\phi_2)}\biggr)+\frac{1}{r_0 ^ 2}\biggl(r_1+\int_{\phi_*}^{\phi_1}\frac{d\phi_2}{p(\phi_2)}\biggr)^2{\nonumber}\\ & -\frac{1}{r_0}\biggl(r_2-\int_{\phi_*}^{\phi_1 } d\phi_2\,\frac{\tilde{u}_{-,1}(\phi_2)}{p(\phi_2)^3}\biggr)-\frac{1}{r_0 ^ 3}\biggl(r_1+\int_{\phi_*}^{\phi_1}\frac{d\phi_2}{p(\phi_2)}\biggr)^3{\nonumber}\\ & + \frac{2}{r_0 ^ 2}\biggl(r_1+\int_{\phi_*}^{\phi_1}\frac{d\phi_2}{p(\phi_2)}\biggr)\biggl(r_2-\int_{\phi_*}^{\phi_1 } d\phi_2\,\frac{\tilde{u}_{-,1}(\phi_2)}{p(\phi_2)^3}\biggr){\nonumber}\\ & -\frac{1}{r_0}\biggl(r_3+\int_{\phi_*}^{\phi_1 } d\phi_2\biggl(\frac{3\tilde{u}_{-,1}(\phi_2)^2}{2p(\phi_2)^5}-\frac{\tilde{u}_{+,2}(\phi_2)+\tilde{u}_{-,2}(\phi_2)}{p(\phi_2)^3}\biggr)\biggr)+o(\epsilon^3)\biggr ] , \label{u_plus_minus_order_2_and_3}\end{aligned}\ ] ] where @xmath244 using , , , , , , , , and for terms @xmath245 in , we find @xmath246 note that @xmath247 holds for any value of @xmath248 due to and @xmath249 the quantity @xmath248 is undetermined . however , if we follow the previously derived @xmath235 and @xmath234 with the analogous @xmath250 , we find @xmath251 which we set from now on . as a result ,
@xmath252 note that @xmath253 is analogous to @xmath236 found earlier .
equations , and give the second - order approximation . using , , , , , , , , , , , and for terms @xmath254 in , we find @xmath255 , \label{u_tilde_plus_3_with_delta_3}\\ & \tilde{u}_{-,3}(\phi)=-\frac{n}{2r_0}\int_{-\phi}^{\phi } d\phi_1\,p(\phi_1)\biggl[-\frac{r_2}{r_0}+\frac{1}{r_0 ^ 2}\biggl(\int_{0}^{\phi_1 } \frac{d\phi_2}{p(\phi_2)}\biggr)^2-\frac{\tilde{u}_{-,1}(\phi_1)^2}{2p(\phi_1)^4}+\frac{\tilde{u}_{+,2}(\phi_1)}{p(\phi_1)^2}{\nonumber}\\ & -\frac{\tilde{u}_{-,1}(\phi_1)}{r_0 p(\phi_1)^2}\int_0^{\phi_1 } \frac{d\phi_2}{p(\phi_2)}+\frac{1}{r_0}\int_{\phi_*}^{\phi_1 } d\phi_2\,\frac{\tilde{u}_{-,1}(\phi_2)}{p(\phi_2)^3}\biggr ] .
\label{u_tilde_minus_3}\end{aligned}\ ] ] since the right - hand side of does not depend on @xmath45 , it follows that @xmath256 for some infinitesimal constant @xmath257 .
the quantities @xmath258 and @xmath257 are undetermined . similar to @xmath259 for @xmath260
, we set @xmath261 , and solve , and for @xmath258 and @xmath257 to find @xmath262,\label{r_2}\\ & \delta_3=0 .
\label{delta_3}\end{aligned}\ ] ] equations , , and give the third - order approximation . using , , , , , , , , , , , , , , , and for terms @xmath263 in , we find @xmath264+\frac{n}{2r_0 ^ 4}\int_\phi^{\phi_{+}}d\phi_1\,p(\phi_1){\nonumber}\\ & \times\biggl[r_0 ^ 3\biggl(\frac{\tilde{u}_{-,1}(\phi_{1})^3}{2p(\phi_1)^6}-\frac{\tilde{u}_{-,1}(\phi_1)\tilde{u}_{+,2}(\phi_1)}{p(\phi_1)^4}+\frac{\tilde{u}_{-,3}(\phi_1)}{p(\phi_1)^2}\biggr)-\biggl(\int_0^{\phi_1}\frac{d\phi_2}{p(\phi_2)}\biggr)^3{\nonumber}\\ & + r_0\frac{\tilde{u}_{-,1}(\phi_1)}{p(\phi_1)^2}\biggl(\int_0^{\phi_1}\frac{d\phi_2}{p(\phi_2)}\biggr)^2+r_0 ^ 2\biggl(\frac{\tilde{u}_{-,1}(\phi_1)^2}{2p(\phi_1)^4}-\frac{\tilde{u}_{+,2}(\phi_1)}{p(\phi_1)^2}\biggr)\int_0^{\phi_1}\frac{d\phi_2}{p(\phi_2)}{\nonumber}\\ & -r_0 ^ 2\int_0^{\phi_1}d\phi_2\biggl(\frac{3\tilde{u}_{-,1}(\phi_2)^2}{2p(\phi_2)^5}-\frac{\tilde{u}_{+,2}(\phi_2)}{p(\phi_2)^3}\biggr)+r_0\biggl(2\int_0^{\phi_1}\frac{d\phi_2}{p(\phi_2)}-r_0\frac{\tilde{u}_{-,1}(\phi_1)}{p(\phi_1)^2}\biggr){\nonumber}\\ & \times\biggl(r_2-\int_{\phi_{*}}^{\phi_1}d\phi_2\,\frac{\tilde{u}_{-,1}(\phi_2)}{p(\phi_2)^3}\biggr)\biggr],\label{u_tilde_plus_4_with_r_3}\\ & \tilde{u}_{-,4}(\phi)=\frac{n}{2r_0 ^ 2}\int_{-\phi}^\phi d\phi_1\,p(\phi_1)\biggl[r_3+\int_0^{\phi_*}d\phi_2\biggl(-\frac{3\tilde{u}_{-,1}(\phi_2)^2}{2p(\phi_2)^5}+\frac{\tilde{u}_{+,2}(\phi_2)}{p(\phi_2)^3}\biggr)\biggr].\label{u_tilde_minus_4_with_r_3}\end{aligned}\ ] ] the quantity
@xmath265 is undetermined .
similar to @xmath259 for @xmath266 , we set @xmath267 and solve for @xmath265 to find @xmath268 which leads to @xmath269,\label{u_tilde_plus_4}\\ & \tilde{u}_{-,4}(\phi)=0.\label{u_tilde_minus_4}\end{aligned}\ ] ] note that @xmath270 is analogous to @xmath236 and @xmath253 found earlier .
equations , and give the fourth - order approximation . to obtain approximations for the half - bounce solution
, we expand the right - hand side of to terms @xmath254 and find the first few terms in , @xmath271 the first few terms in are found similarly . after solving @xmath272 for the function @xmath273 ,
we iterate and find @xmath274.\label{psi_2}\end{aligned}\ ] ] it is clear how to proceed to higher orders .
although the approximation method developed in sec . [ section_method ] works for an arbitrary continuously differential function @xmath9 with two minima and one maximum , it may appear that the formulas derived for various orders of approximation are difficult to use in practice .
in particular , one may expect the need for many terms in the approximation in order to achieve an acceptable accuracy for a half - bounce with the thick wall , or that the higher order approximations will become too complicated to be useful . to demonstrate applicability of the above results , we consider in this section an example of the general fourth - order polynomial potential @xmath275 we now proceed to deduce the values of the coefficients @xmath276 by imposing the conditions specified in sec . [ section_bounces ] on the function @xmath9 , which have been shown to lead to no loss of generality .
we first choose @xmath277 and @xmath278 , which give @xmath279 and @xmath280 .
the requirement @xmath281 now leads to @xmath282 , after which @xmath283 yields @xmath284 , which result in @xmath285 for the function @xmath9 to have two minima and one maximum , we need @xmath286 , which implies @xmath287 and @xmath288 .
we compute @xmath289 and find that is satisfied .
we thus have freedom to impose one additional condition on the function @xmath9 , for which we choose @xmath290 without any loss of generality . as a result ,
@xmath291 we find @xmath292 and see that we need @xmath293 in order for @xmath294 to be the relative minimum .
using equations from sec .
[ section_method ] , we find all the results necessary to obtain the approximations for the half - bounce for this example through the fourth order . in the remainder of this section , all functions of @xmath45 will be restricted to the domain @xmath295 .
we start with @xmath296 and use , , , , and to find the zeroth - order and first - order quantities @xmath297 up to this order , our approximation coincides with the thin - wall approximation . going beyond it , , and give the second - order quantities @xmath298 now , , and lead to the third - order quantities @xmath299,\\ & r_2=\frac{(n-1)(6-\pi^2)\epsilon}{16n},\end{aligned}\ ] ] where @xmath300 is the dilogarithm function .
we give the expression for only one fourth - order quantity , @xmath301{\nonumber}\\ & = \frac{9(4\ln{2}-1)\epsilon^3}{256n}+o(\epsilon^5),\label{r_3_polynomial}\end{aligned}\ ] ] which we calculate from .
we note that @xmath302 in and @xmath303 in contradict @xmath304 in .
our choice @xmath290 explains this discrepancy because it leads to @xmath305 in in contrast to @xmath306 that was assumed in the derivation of approximations in sec .
[ subsection_orders ] .
we made such a choice for @xmath307 ( which is the only one possible to get @xmath305 ) specifically to test accuracy of the iterative method in the worst possible case ( for any fourth - order polynomial potential ) when the orders of some terms in expansions have slightly different @xmath44-dependencies . despite these discrepancies ,
the agreement between exact numerical solutions and approximate analytic solutions is outstanding ( see below ) , and it should be clear that the agreement will only improve for any other choice of @xmath307 satisfying @xmath287 ( as required by ) . with these preliminary results ,
we now obtain from , and @xmath308,\end{aligned}\ ] ] which via , and finally leads to approximations for the half - bounce solution @xmath309.\end{aligned}\ ] ] we note that despite complicated intermidiate results leading to the bounce solution @xmath310 , the expressions for the first approximation @xmath311 and even the second approximation @xmath312 to some extend are rather simple . together with the high numerical accuracy shown in the following section , we view this simplicity as a demonstration of the strength of our approximation method .
the accuracy of our successive approximations can be seen in figs .
[ figure_u_tilde_0 ] , [ figure_phi_0 ] and [ figure_r_error_phi_error ] . in fig .
[ figure_u_tilde_0 ] we show the exact numerical solution @xmath148 and the approximate analytic solutions @xmath313 for @xmath314 . in fig .
[ figure_phi_0 ] we compare the exact numerical solution @xmath310 with the approximate analytic solutions @xmath315 for @xmath316 .
finally , fig .
[ figure_r_error_phi_error ] gives the accuracy for the wall radius @xmath43 in terms of the @xmath44-dependence of the relative errors @xmath317 for @xmath318 and the @xmath44-dependence of the integrated deviation of the exact numerical solution @xmath310 from the approximate analytic solutions @xmath315 for @xmath316 .
( the solid curve ) and the approximate analytic solutions @xmath313 for @xmath319 ( the dotted curve ) , @xmath320 ( the dot - dashed curve ) and @xmath321 ( the dashed curve ) for @xmath42 and the potential with different @xmath44 .
( some curves nearly coincide . )
each color of curves in this figure and in figs .
[ figure_u_phi ] and [ figure_phi_0 ] represents the same value of @xmath44 . ]
( the solid curve ) and the approximate analytic solutions @xmath315 for @xmath322 ( the dotted curve ) , @xmath319 ( the dot - dashed curve ) and @xmath320 ( the dashed curve ) for @xmath42 and the potential with different @xmath44 .
( some curves nearly coincide . )
each color of curves in this figure and in figs .
[ figure_u_phi ] and [ figure_u_tilde_0 ] represents the same value of @xmath44 . ]
-dependence of @xmath317 for @xmath322 ( the dotted curve ) , @xmath319 ( the dot - dashed curve ) , @xmath320 ( the dashed curve ) and @xmath321 ( the solid curve ) for @xmath42 and the potential .
right : the @xmath44-dependence of @xmath323 for @xmath322 ( the dotted curve ) , @xmath319 ( the dot - dashed curve ) and @xmath320 ( the dashed curve ) in for @xmath42 and the potential . ]
we have proposed a new method of iterative approximate solutions for the spherical bounce which works for any continuously differentiable potentials in any number of dimensions .
the zeroth - order and first - order approximations coincide with the thin - wall approximation of coleman @xcite and all higher - order approximations are derived iteratively .
the iterative approximations have global features which distinguish them from more straightforward local approximations obtained via standard series expansions .
a local approximation typically works best near the center of the expansion , but its accuracy rapidly decreases far from the center .
the situation is slightly better with matched series approximations , where several expansions centered at different points are glued at points between the centers by matching the first few derivatives of the solution .
we gave an example of this matched series expansion in sec .
[ subsection_power_series_approximation ] , where we saw that its accuracy is not great for large asymmetry in the potential @xmath9 .
on the other hand , having smaller errors @xmath324 our iterative approximations better represent the exact solutions for a broad range of values of @xmath69 , especially for @xmath325 .
we also note that even if the quantities @xmath315 differ significantly from @xmath310 for small @xmath69 , the presence of the jacobian factor @xmath326 in the numerator in makes these differences for @xmath67 much less important than the corresponding differences for large @xmath69 .
compare figs .
[ figure_phi_0 ] and [ figure_r_error_phi_error ] in this regard .
our method proceeds to higher orders iteratively with fast convergence and high accuracy .
analysis of the problem from a new perspective demonstrates some universal properties of the bounce .
the method is not restricted to only certain types of potentials or dimensions of space .
for example , there is nothing special about the potential being a fourth - order polynomial or the space being three - dimensional for the successful application of the method to the example we investigated in sec .
[ section_polynomial ] .
we also note that the approximation works well beyond its intended range of applicability of small asymmetry of the potential .
compare figs .
[ figure_u_phi ] , [ figure_phi_0 ] , and [ figure_r_error_phi_error ] in this regard .
the potential functions shown in the left part of fig . [ figure_u_phi ] are precisely those for which the corresponding solutions and their approximations are shown in the right part of the fig .
[ figure_u_phi ] and in fig .
[ figure_phi_0 ] .
although hardly any of these potential functions can be considered as having small asymmetry , the approximations in fig .
[ figure_phi_0 ] are quite accurate .
once approximations for the classical bounce solution are known in the analytic form , the next obvious step is to compute the decay rate of the false vacuum .
the rate is the product of the exponential term given by the classical action of the bounce and the pre - exponential factor expressed in terms of functional determinants . with the iterative method for spherical bounces developed in this paper , deriving corresponding approximations for
the pre - exponential factor should be a relatively straightforward procedure .
another promising direction is to develop a similar approximation method for the gravitational bounce ; it would be interesting to see how the required modifications agree , in particular , with the findings of refs .
@xcite , @xcite and @xcite .
i am grateful to thomas w. kephart for his participation in the early stages of this project and for useful discussions .
p. h. frampton , phys .
lett . * 37 * , 1378 ( 1976 ) [ erratum - ibid . * 37 * , 1716 ( 1976 ) ] .
p. h. frampton , phys .
d * 15 * , 2922 ( 1977 ) .
s. r. coleman , phys .
d * 15 * , 2929 ( 1977 ) [ erratum - ibid .
d * 16 * , 1248 ( 1977 ) ] . c. g. callan and s. r. coleman , phys .
d * 16 * , 1762 ( 1977 ) .
s. r. coleman , v. glaser and a. martin , commun .
phys . * 58 * , 211 ( 1978 ) .
s. r. coleman and f. de luccia , phys . rev .
d * 21 * , 3305 ( 1980 ) .
t. c. shen , phys . rev .
d * 37 * ( 1988 ) 3537 .
d. a. samuel and w. a. hiscock , phys .
b * 261 * ( 1991 ) 251 .
j. baacke and g. lavrelashvili , phys .
d * 69 * , 025009 ( 2004 ) [ hep - th/0307202 ] .
g. v. dunne and h. min , phys .
d * 72 * , 125004 ( 2005 ) [ hep - th/0511156 ] .
d. a. samuel and w. a. hiscock , phys .
d * 44 * , 3052 ( 1991 ) .
a. masoumi , s. paban and e. j. weinberg , phys .
d * 94 * , no . 2 , 025023 ( 2016 ) [ arxiv:1603.07679 [ hep - th ] ] . | we develop a new iterative method for finding approximate solutions for spherical bounces associated with the decay of the false vacuum in scalar field theories .
the method works for any generic potential in any number of dimensions , contains coleman s thin - wall approximation as its first iteration , and greatly improves its accuracy by including higher order terms . |
in the last 25 years more than 40 accretion - powered x - ray pulsars have been discovered .
at least half of these are transient , of which 12 have known be star companions .
neutron stars with be companions accrete material from the slow , dense , stellar outflow thought to be confined to the equatorial plane of the be star .
cep x4 has recently been associated with a be star of magnitude v=14.2 ( @xcite ; @xcite ) that lies within the _ rosat _ error circle ( @xcite ) .
this association is based upon positional coincidence , strong h-@xmath6 and h-@xmath7 emission lines ( typical of be / x - ray binaries ) , and measurement of the optical absorption column density which is in good agreement with _ rosat _ measurements .
the companion is most likely a b1v - b2ve star at a distance of [email protected] kpc ( @xcite ) .
cep x4 was discovered with _
oso-7 _ in 1972
june - july , but no pulsations were detected ( @xcite ) .
ginga _ all - sky monitor first detected a new transient source , gs 2138 + 56 , on 1988 march 19 at a 50 mcrab level .
gs 2138 + 56 was later identified with cep x4 because it fell in the 0.4 error box and had a similar spectrum .
pulsations at a frequency of 15.09457 @xmath8 0.00002 mhz ( mjd 47,263.5 ) were detected during a month long outburst which peaked at about 100 mcrab ( 1 - 20 kev ) in early april 1988 .
the source apparently did not appear again until 1993 june when it was detected with _
rosat _ ( @xcite ) and the burst and transient source experiment ( batse ) on the _ compton gamma - ray observatory ( cgro)_. near the outburst peak , pulsations at a frequency of 15.0941 @xmath8 0.0002 mhz ( mjd 49,163.0 ) were detected with batse .
the outburst lasted about two weeks and had a peak pulsed flux of 15 - 20 mcrab ( 20 - 50 kev ) . in 1997
july - august , batse and the all - sky monitor ( asm ) on the _ rossi x - ray timing explorer ( rxte ) _ observed an outburst from cep x4 .
batse detected pulsations at a frequency of 15.0882 @xmath8 0.0002 mhz ( mjd 50,641 ) near the outburst peak .
this outburst lasted about 2 weeks and peaked at a pulsed flux of about 10 - 15 mcrab ( 20 - 50 kev ) .
the asm observed a longer ( 4 week ) outburst with a peak flux of about 40 mcrab ( 2 - 12 kev ) .
the best fit average spin - down rate is @xmath9 hz s@xmath3 between the 1988 and 1993 outbursts . between the 1993 and 1997 outbursts , the average spin - down rate , @xmath10 hz s@xmath3 ,
is 10 times larger than the spin - down rate between the first two outbursts .
apparent spin - up observed during the first two outbursts and spin - down in the recent outburst may be intrinsic or due to binary orbital motion . in this paper
we present observations of two outbursts from cep x4 in 1993 and 1997 .
our observations with batse include histories of pulse frequency , pulsed flux , and average 20 - 50 kev pulse profiles .
we obtained public _
rxte _ proportional counter array ( pca ) data from 1997 july 18 & 25 , after batse stopped detecting cep x4 .
we present pulse frequency measurements , pulse profiles , and power spectra from these data .
three different instruments spanning energies from 2 - 1800 kev were used to study cep x4 : batse on _ cgro _ , and the pca and asm both on _ rxte_. batse consists of eight identical uncollimated detector modules positioned on the corners of the _ cgro _ spacecraft such that the normal vectors of the detectors are perpendicular to the faces of a regular octahedron , providing all - sky coverage .
the batse data presented here are taken with the large - area detectors ( lads ) , which are nai(tl ) scintillation crystals with a geometric area of 2025 @xmath11 and a thickness of 1.27 cm .
the lads are sensitive to photons from 20 to 1800 kev .
two batse data types were used in this analysis , the cont ( 2.048 s , 16 energy channel ) data and the discla ( 1.024 s , 4 energy channel ) data .
a more complete description of the instrument and data types can be found in @xcite .
the pca on _ rxte _ consists of five proportional counters sensitive to photons from 2 - 60 kev and has a total collecting area of 6500 @xmath11 .
the pca is a collimated instrument with an approximately circular field - of - view with a fwhm of about 1 ( @xcite ) .
three pca data types were used in this analysis , standard 1 ( 125 ms , no energy resolution ) data , good xenon data ( 1 @xmath12s , 256 energy channel ) data , and standard 2 ( 16 s , 128 channel ) data .
the _ rxte _
asm consists of three wide - angle shadow cameras equipped with xenon proportional counters with a total collecting area of 90 @xmath11 .
the asm provides 90 second images of most of the sky every 96 minutes in three energy channels from 2 - 12 kev ( @xcite ) . for this analysis we used single day
averaged data from the full 2 - 12 kev band .
histories of pulse frequency and pulsed flux ( 20 - 50 kev ) were determined on 2-day intervals of batse discla data .
pulse frequencies were estimated from fits to data at a range of trial frequency offsets from a pulse phase model .
the pulse phase model used a constant frequency , @xmath13 , obtained from daily power spectra of the batse discla data .
each 2-day interval of data were fit with a background model plus a 6 harmonic fourier expansion in the pulse phase model to generate a pulse profile at the model frequency .
the background was modeled as a quadratic spline , with segments every 300 s and with value and slope continuous between adjacent segments .
( see @xcite for a detailed description of pulsed flux and pulse frequency estimation techniques . ) for each 2-day span a range of trial frequency offsets , @xmath8 3.5 cycles day@xmath3 , from the model frequency was searched . for each trial frequency
offset , @xmath14 , the pulse profile at the model frequency was shifted in phase by multiplying the fourier harmonic coefficients by a phase factor @xmath15 , where @xmath16 is the harmonic number .
the phase offset , @xmath17 , is given by @xmath18 where @xmath19 is a time within to a 2-day interval and @xmath20 is an epoch near the center of the 2-day interval .
this method is equivalent to refitting the background model plus fourier expansion for each frequency @xmath21 .
the best fit frequency for each 2-day interval was determined by comparing pulse profiles for different frequency offsets using the z@xmath22 test ( @xcite ) which measured the significance of the first 6 fourier amplitudes .
the root - mean - square ( rms ) pulsed flux was estimated from the best fit pulse profile as @xmath23^{1/2}\ ] ] where @xmath24 is the pulse profile at phase @xmath25 , @xmath26 , and @xmath27 is the average flux .
the pulsed fluxes were determined assuming an exponential energy spectrum , @xmath28 with a normalization @xmath29 and e - folding energy @xmath30 .
pulse frequencies were estimated from _ rxte _
pca data using a similar method .
since both data sets were quite short , 2000 - 2500 s , a model consisting of a constant background plus an @xmath31 harmonic fourier expansion in the pulsed phase model was fit to the data . in both observations ,
the pulsed phase model used a constant frequency , based upon batse measurements .
this model was fit to the data for a grid of trial frequencies and numbers of fourier coefficients @xmath31 .
the best - fit frequency was determined by the h - test ( @xcite ) .
background variations observed in these intervals were at low frequencies relative to the cep x4 s pulse frequency .
deviations from the constant background model resulted in low frequency noise well outside the search range used which did not affect pulse frequency measurements .
batse data from 1991 april to 1997 december were searched for outbursts from cep x4 using the pulse frequency and flux estimation method described in section [ sec : tech ] .
two outbursts , each lasting about 2 weeks were detected in the batse data : the first from 1993 june 20 - july 6 ( mjd
49,158 - 49,174 ) and the second from 1997 july 3 - 15 ( mjd 50,632 - 50,644 ) .
figure [ fig : cepx4_93 ] shows the history of pulse frequency and pulsed flux for the 1993 outburst .
_ rosat _ also observed this outburst and measured a pulse frequency of @xmath32 hz s@xmath3 ( @xcite ) which agrees well with the batse measurements . from figure
[ fig : cepx4_93 ] one observes that cep x4 s frequency increased significantly from 1993 june 24 - 28 ( mjd 49,162 - 49,166 ) .
the frequency rate is @xmath33 hz s@xmath3 , which may be intrinsic or due to orbital motion . in 1997
july , cep x4 underwent another outburst which was detected with batse and _ rxte_. figure [ fig : cepx4_97 ] shows the history of pulse frequency as measured with batse and the _ rxte _ pca .
the average frequency rate is @xmath34 hz s@xmath3 , which is comparable in magnitude to the brief spin - up episode seen in the 1993 outburst .
this spin - down may be intrinsic or due to orbital motion .
also shown is the pulsed flux history from 20 - 50 kev batse data and the phase averaged flux from _ rxte _ asm data . to determine the pulse frequencies from the pca data , standard 1 ( 125 ms , no energy resolution ) data from 1997 july 18 ( mjd 50,647 ) and 1997 july 25 ( mjd 50,654 )
were used with techniques described in section [ sec : tech ] .
we found that the scatter in frequency measurements for different values of the number of fourier coefficients , @xmath31 ( see section [ sec : tech ] ) , was much larger than expected for poisson noise .
the best fit frequencies , with errors corrected for aperiodic noise ( described below ) , were [email protected] mhz ( @xmath35 ) on july 18 ( mjd 50,647 ) and [email protected] mhz ( @xmath35 ) on july 25 ( mjd 50,654 ) .
excess aperiodic noise from cep x4 was likely the cause of the observed scatter in our frequency measurements , so we performed power spectral analyses on about 2000 s of good xenon data from 1997 july 18 and about 2400 s from july 25 to search for aperiodic variability .
these data were binned into 125 ms time bins containing no gaps .
the background in these intervals was modeled using pcabackest , which estimates the background from standard 2 data at 16 second intervals .
this model was then interpolated to 125 ms resolution by polynomial fitting .
power due to the 66 s pulsations was removed by subtracting a model consisting of a 14th degree fourier expansion in the best fit pulse frequency .
next a fourier transform with no oversampling was performed on the background subtracted data .
the fourier coefficients @xmath36 are given by @xmath37 where @xmath38 and @xmath39 are the total counts and background counts in bin @xmath40 , @xmath41 the number of measurement bins , and @xmath42 . equation ( 1 ) in morgan & remillard ( 1997 ) was used to calculate the dead - time corrected poisson level with a time bin size of @xmath43 ms , a dead - time per event of @xmath44s , and a dead - time window per very large event ( vle ) of @xmath45s . in the normalization of leahy et al . ( 1983 ) , given by @xmath46 where @xmath47 is the total number of photons , the poisson level is @xmath48 for purely poisson noise .
these calculations yielded average dead - time corrected poisson levels of @xmath49 for july 18 and @xmath50 for july 25 .
the aperiodic noise at the pulse frequency and its harmonics was estimated by averaging the source power , @xmath51 , in the region @xmath52 , where @xmath53 is the pulse frequency ( n=1 ) and @xmath54 ( n=2,3 , ... ) are the harmonics of the pulse frequency .
this power was then compared to the average dead - time corrected poisson level , @xmath55 for each region .
the errors on the frequency measurements , @xmath56,listed above and shown in figure [ fig : cepx4_97 ] include a correction due to this excess aperiodic noise , given by @xmath57 , where @xmath58 is the error expected for poisson noise only .
next , we wished to measure the rms variation of the aperiodic noise component .
we renormalized the power such that the fractional rms variation in source intensity over a given frequency range is related to the integral of the power in that range , i.e. @xmath59 where @xmath60 is the variance on the source count rate due to the frequency range @xmath61 $ ] and @xmath62 is the mean source count rate .
the power spectrum was then calculated by @xmath63 where @xmath64 is the duration of the data and @xmath65 is the total source counts in bin @xmath40 .
figure [ fig : pow ] shows the power spectra , plotted as fractional power per logarithmic frequency interval , from 1997 july 18 in the top panel and july 25 in the bottom panel . in these units , a @xmath66 power law is a horizonal line .
aperiodic variability is present in both time intervals , with an rms variation of 17 - 19% ( @xmath670.01 - 1 hz ) for both days . in the 1988 outburst , _ ginga _ observed aperiodic noise with an rms variation of about 6% for the same frequency interval .
the source was apparently noisier in _ rxte _ observations .
clearly evident in figure [ fig : pow ] is the presence of considerable low frequency noise on 1997 july 25 ( bottom panel ) which is not present in the top panel .
we compared variations in the pulsed flux , mean background subtracted flux , and the background model ( generated using pcabackest ) to determine if these features were intrinsic to cep x4 .
the unpulsed flux and pulsed flux for july 25 were measured by fitting a constant plus a 14 harmonic fourier expansion in the best - fit pulse frequency to intervals exactly 2 pulse periods long .
the background model was binned at the same resolution and subtracted from the unpulsed flux .
the variations observed in the unpulsed flux prior to background subtraction were about 3 - 4 times the background model variations .
figure [ fig:2period ] shows the background model , the background subtracted unpulsed flux , and the pulsed flux .
the background subtracted unpulsed flux ( denoted by the solid histogram ) and pulsed flux ( denoted by the dashed histogram ) increased slowly to a broad peak , followed by a faster decrease .
the background model ( denoted by the dotted line ) also slowly rose to a flat peak and then decreased , but it peaked later .
the pulse fraction remained essentially constant throughout this interval , so it was unlikely that the variations observed were due to background fluctuations . on july 18 , when cep x4 was about 2 times brighter in the asm , the unpulsed and pulsed flux remained approximately constant during our observations .
apparently cep x4 was less stable at lower intensities .
average pulse profiles were generated by epoch - folding @xmath5 10 days of 20 - 50 kev batse cont ( 2.048 s resolution data ) for both the 1993 and 1997 outbursts .
a quadratic frequency model fit to the observed frequencies in figures [ fig : cepx4_93 ] & [ fig : cepx4_97 ] was used in the epoch - folding .
pulse profiles from 1993 june 20 - 29 ( mjd 49,158 - 49,167 ) and 1997 july 4 - 14 ( mjd 50,633 - 50,643 ) are shown in the top and bottom panels respectively in figure [ fig : batse_profiles ] .
both profiles show a large main peak ( phase @xmath68 ) followed by smaller second peak atop a broad shoulder ( phase @xmath69 ) and a valley ( phase @xmath70 ) .
pulse profiles from _ rxte _ observations were generated by epoch - folding good xenon data from 1997 july 18 and july 25 .
good xenon data were binned into 125 ms bins and divided into 8 energy bands .
the background in each band was estimated at 16 s intervals using pcabackest and interpolated to 125 ms resolution by polynomial fits .
the background subtracted data were folded at the best fit frequencies from the fits to the standard 1 data shown in figure [ fig : cepx4_97 ] .
figure [ fig : profiles ] shows the pulse profiles from july 18 ( left hand panel ) and july 25 ( right hand panel ) .
the pulse profile shows energy dependent features and considerable structure with a large main peak ( phase @xmath50.7 - 1.0 ) followed by a deep notch at low energies ( phase @xmath51.0 - 1.1 ) .
the notch is followed by a second peak ( phase @xmath51.1 - 1.4 ) and a valley ( phase @xmath51.4 - 1.7 ) .
the profiles evolve from double peaked , with a narrow notch between the peaks at low energy , to a dominant first peak ( phase @xmath71 ) followed by a broad shoulder .
the second peak ( phase @xmath72 ) appears to be weaker relative to the first peak on july 25 than on july 18 .
figure [ fig : hardness ] is a hardness ratio for july 18 given by @xmath73 where @xmath74 is the average background subtracted count rate for 9 - 20 kev , @xmath75 is the mean subtracted pulse profile count rate for 9 - 20 kev , @xmath76 is the average background subtracted count rate for 2 - 6 kev , and @xmath77 is the mean subtracted profile for 2 - 6 kev .
the errors shown are the errors on the difference @xmath78 , since we are interested in where the hardness ratio deviated from the mean value , [email protected] .
hardness ratios show that the notch is the hardest feature and that it is surrounded by a softer double lobed feature ( phase @xmath5 0.7 - 1.4 ) .
similar features were also seen in hardness ratios from july 25 .
however , the significance of these features was reduced by the lower intensity and smaller pulse fractions in these data .
the root mean squared ( rms ) intensities for each energy band were calculated as @xmath79 , where @xmath80 is the background subtracted source count rate in phase bin @xmath40 , @xmath81 is the number of phase bins , and @xmath82 is the mean source count rate .
rms pulse fractions , @xmath83 are shown in figure [ fig : frac ] .
the pulse fraction has considerable energy dependence and appears to be decreasing as the mean source intensity decreases . the average 20 - 50 kev batse pulse profile from 1997 july 4 - 14 ( fig .
[ fig : batse_profiles ] , bottom panel ) is very similar to the 1997 july 18 pca profile in the 20 - 30 kev energy band ( fig . [ fig : profiles ] bottom left ) .
the source intensity was too low to allow detection of significant profile variations at energies @xmath84 20 kev with either instrument .
_ ginga _ observed pulse shape variations with energy during the 1988 outburst .
the second peak was brighter than the first at low energies ( 1 - 7 kev ) and roughly equal at higher energies ( 13 - 37 kev ) ( @xcite ) .
the optically identified companions for transient pulsars with periods longer than one second are all be or oe stars .
recent long - term studies with batse revealed that be / x - ray binaries exhibit series of often periodic outbursts and most appear to have a long - term spin - up trend ( @xcite ) .
one other system , a 1118 - 616 , may be experiencing a spin - down trend , but the duty cycle for this source , like cep x4 , is quite low .
long - term spin - down in x - ray pulsars is thought to be caused by the propeller effect ( @xcite ) .
simple accretion theory assumes that material from the companion star is flowing onto a rotating neutron star with a magnetic field .
the magnetic field is so strong that it determines the motion of the material in a region of space surrounding the neutron star called the magnetosphere .
the magnetospheric radius @xmath85 , is defined to be the distance from the neutron star at which all magnetic field lines are closed loops . assuming a dipole field at large distances from the neutron star
, the magnetospheric radius can be estimated by equating the magnetic energy density to the total kinetic energy density of the accreting gas .
therefore from shapiro & teukolsky ( 1983 ) , @xmath86 where @xmath87 is the magnetic moment of the dipole field in units of @xmath88 g @xmath89 , @xmath90 is the neutron star mass in units of @xmath91 , @xmath92 is the mass accretion rate in units of @xmath93 g s@xmath3 , and @xmath94 for disk accretion ( @xcite ; @xcite ; @xcite ; @xcite ) and @xmath95 for wind accretion ( @xcite ) .
when @xmath85 exceeds the corotation radius , @xmath96 , the distance from the neutron star where centrifugal forces just balance local gravity , the propeller effect is believed to occur .
matter will be ejected from the magnetic field lines when the centrifugal forces exceed gravitational force .
this ejected matter gains angular momentum from the pulsar , causing it to spin down .
if we assume the long - term spin - down observed in this system is due to the propeller effect , then by equating the magnetospheric and co - rotation radii , we can estimate the critical mass accretion rate where the propeller effect begins @xmath97 where @xmath98 is the pulsar spin period in seconds . at this critical mass transfer rate , corbet ( 1996 )
predicts a transition from neutron star accretion to magnetospheric accretion , accompanied by a large drop in luminosity . using a magnetic field estimate of @xmath99 g ( @xcite ) , a distance estimate @xmath100 kpc ( @xcite ) , a spin period of @xmath101 s , along with typical neutron star parameters @xmath102 and @xmath103 cm
, we estimate the minimum x - ray flux for neutron star accretion , @xmath104 erg @xmath105 s@xmath3 , and the maximum for magnetospheric accretion , @xmath106 erg @xmath105 s@xmath3 . in 1993
january _ rosat _ observed a quiescent 0.1 - 2.5 kev flux level of @xmath107 erg @xmath105 s@xmath3 from cep x4 ( @xcite ) .
this flux level fell in the gap between @xmath108 and @xmath109 .
schulz et al . ( 1995 ) estimated that the _ rosat _ flux in the outburst in 1993 june was about 10% of the bolometric flux .
assuming that the cep x4 spectrum was unchanged during the quiescent _ rosat _ observations , we estimate a quiescent bolometric flux level of @xmath110 erg @xmath105 s@xmath3 , which still falls in the gap , if we assume wind accretion ( @xmath95 ) . clearly the observed flux was near the expected value for the onset of the propeller regime .
the estimated bolometric flux from the _ rosat _ observations was a factor of @xmath5 3 below the expected value for wind accretion , which implies that the spectrum was much harder at low @xmath111 , similar to a 0535 + 262 ( @xcite ) .
future quiescent observations at higher energies are needed to confirm this spectral hardening . for wind accretion ,
additional uncertainties in the predicted flux arise because when the propeller effect sets in at the neutron star s equator , accretion still occurs at higher latitudes . comparing the estimated bolometric flux with the predicted flux
suggests that we may be observing disk accretion .
however , additional uncertainties in the predicted flux arise from uncertainty in dimensionless parameter @xmath112 . in both wind and disk accretion , the validity of the assumptions used to derive the predicted flux is uncertain because the physics at the onset of the propeller effect is poorly understood .
long - term studies show that be / x - ray transients exhibit two types of outbursts , giant outbursts characterized by high luminosities and high spin - up rates , and normal outbursts which typically are fainter , exhibit less intrinsic spin - up , and are spaced by some multiple of the orbital period .
normal outbursts appear to be associated with peristron passage in be / x - ray transients with known orbits ( @xcite ) .
the outbursts observed from cep x4 are most likely normal outbursts .
therefore the spacing of the 4 outbursts constrains the orbital period .
we assume that the two previous outbursts in 1972 june - july ( @xcite ) and 1988 march - april ( @xcite ) occurred at the same orbital phase ( likely near periastron ) as the two outbursts observed by batse .
outburst mid - times were calculated for each observation .
errors were assigned to each mid - time according to how well the mid - point of the outburst was known .
an evenly spaced grid of orbital periods ranging from 25 - 10@xmath113 days was searched , minimizing the following @xmath114 : @xmath115 where @xmath116 is the mid - time of outburst @xmath16 , @xmath117 is the estimated error on @xmath116 , @xmath118 is an epoch , and @xmath119 is an orbital period
. a number of acceptable orbital periods , all @xmath120 147.3 days , were found . combining this result with koyama et al .
( 1991 ) , implies 23 days @xmath121 days .
the energy dependent pulse profile of cep x4 in figure [ fig : profiles ] consists of two peaks separated by a narrow valley at low energies ( @xmath122 kev ) which becomes filled in as the energy increases , until the leading peak ( phase @xmath5 0.7 - 1.0 ) becomes dominant at energies above roughly 9 kev .
this energy dependent behavior can be interpreted as the sum of two pulse components : a softer double lobed component that occupies pulse phase @xmath5 0.7 to 1.4 in figure [ fig : profiles ] and a narrower harder peak centered at phase @xmath5 1.05 that is offset towards the leading peak of the softer double lobed component .
the relationship between the two components is well illustrated by the hardness ratio shown in figure [ fig : hardness ] .
this energy dependent behavior is quite similar to that observed in her x-1 ( @xcite ) .
a possible interpretation of this energy dependent behavior can be made using a cyclotron backscattering scenario . in this scenario
the outward bound x - rays from a given magnetic pole are divided into two main components : a hard pencil beam directed along the accretion column and an antipodally directed fan beam of backscattered radiation ( @xcite ) .
the fan beam is produced by outward bound photons that travel up through the accretion column until their energy equals the local cyclotron resonance energy whence they are backscattered .
the backscattered photons will either strike the neutron star surface or be focused around the neutron star forming an antipodal fan beam .
softer photons scatter from higher altitudes in the accretion column and hence a smaller fraction of the backscattered photons are intercepted by the neutron star surface than is the case for the harder photons that scatter at lower altitudes .
therefore , the antipodal fan beam lobes produced by the softer photons will be wider , i.e. more filled in " than the lobes produced by harder photons . at even higher energies , near the surface cyclotron energy and above , most photons
will not be backscattered and will escape forming a pencil beam .
for an appropriate angle of the magnetic axis to the spin axis and an observer positioned with a line - of - sight to the neutron star near the spin latitude of the magnetic dipole axis , a pulse profile consisting of a single hard pencil beam filling in the center of a softer fan beam from the antipodal pole will be observed .
a simple beam model consisting of a pencil beam surrounded by a backscattered fan beam was constructed to explore possible pulse profiles illustrative of a cyclotron scattering model .
the pencil beam is represented by an emitting point on the neutron star surface and has a gaussian beam profile .
similarly , the fan beam is represented by an emitting point above the neutron star surface that is at the apex of a conical beam pointed toward the neutron star and has a gaussian profile for the conical beam walls " .
gravitational light bending effects are taken into account by calculating photon trajectories from that height . in figure
[ fig : pulsemod ] we show an example profile that is reminiscent of the cep x4 profile .
the five panels on the left represent the pulse profile in five different energy bands .
the panels on the right show the apparent locations of the emitting point when the beams are visible .
the height @xmath123 of the fan beam emitting point decreases with increasing energy as @xmath124 kev where @xmath125 is the cyclotron scattering energy , @xmath126 is the surface magnetic field strength in units of @xmath127 gauss and @xmath128 is the radius of the neutron star .
sharp drops in the fan beam emission occur when the beam intersects the neutron star surface .
use of a more realistic extended emission region rather than a point would smooth out these sharp drops.the pencil beam increases in strength relative to the fan beam with increasing energy and contributes the shaded area to the pulse profile . the observer is positioned such that only a single pencil beam is visible from one magnetic pole while only the fan beam from the opposite pole is visible .
the pole emitting the pencil beam has been offset slightly to move the pulse phase centroid of the resulting pulse profile with respect to the pulse profile contributed by the fan beam .
this is done to simulate the asymmetry observed in the cep x4 profile .
jahoda , k. , swank , j. h. , giles , a. b. , stark , m. j. , strohmayer , t. e. , zhang , w. & morgan , e. h. 1996 , in spie 2808 , euv , x - ray , and gamma - ray instrumentation for space astronomy vii , ed . o. h. w. siegmund & m. a. grummin,(bellingham : spie ) , 59 | we report on x - ray observations of the 66 s period transient x - ray pulsar cep x4 ( gs 2138 + 56 ) with the burst and transient source experiment ( batse ) on the _ compton gamma - ray observatory ( cgro ) _ and with the rossi x - ray timing explorer ( _ rxte _ ) . two outbursts from cep x4 were observed with batse in 1993 june - july and 1997 july .
pulse frequencies of @xmath0 mhz on 1993 june 25 ( mjd 49,163 ) and @xmath1 mhz on 1997 july 12 ( mjd 50,641 ) were each measured from 2 day spans of batse data near each outburst s peak .
cep x4 showed an average spin down rate of @xmath2 hz s@xmath3 between the 1993 and 1997 outbursts .
after batse could no longer detect cep x4 , public observations were performed on 1997 july 18 & 25 with the proportional counter array ( pca ) on _ rxte_. a pulse frequency of @xmath4 mhz was measured from observations on 1997 july 18 ( mjd 50,647 ) .
significant aperiodic noise , with an rms variance of @xmath518% in the frequency range 0.01 - 1.0 hz was observed on both days .
energy and intensity dependent pulse shape variations were also seen in these data .
recently published optical observations associate cep x4 with a be companion star . if all 4 outbursts observed from cep x4 are assumed to occur at the same orbital phase , we find that the orbital period is between 23 days and 147.3 days . |
wolf - rayet ( wr ) stars are the evolutionary descendents of massive o - type stars . with mass loss rates in the interval ( 1 - 5)10@xmath3 and terminal velocities of 1000 - 3000 ( @xcite ; @xcite ) , these hot and luminous stars are one of the most powerful stellar wind sources in our galaxy
. of - type stars are also characterized by high mass loss rates and terminal velocities ( lamers & leitherer 1993 ; prinja , barlow & howarth 1990 ) .
the mechanical energy released to the interstellar medium ( ism ) during the wr phase only ( @xmath4 710@xmath5 yr , @xcite ) is in the range ( 1 - 30)10@xmath6 erg , comparable to the mechanical energy injected during a supernova explosion .
both the mass flow from the wr star and the previous o - type star phases strongly modify the energetics , the morphology and the chemical abundances of the ism in the environs of the star .
the interaction of the strong stellar winds with the surrounding interestellar matter has been analyzed by several authors , taking into account different enviroments ( e.g. @xcite ) .
the stellar flow sweeps up the surrounding material creating _ interstellar bubbles _ , which are detected at different wavelengths from the uv to the radio range . in the optical regime
, these structures are generally observed as filamentary ring nebulae in the light of h@xmath7 and [ oiii ] ( e.g. @xcite ; @xcite )
. they are related to many wr stars and to a relatively large number of o and of stars .
shell - shaped structures created by stellar winds from massive stars are also identified in the far infrared and in the thermal radio continuum emission ( e.g. @xcite ; cappa , goss & pineault 2002 ) .
interstellar bubbles appear as cavities and expanding shells in the neutral hydrogen 21 cm line emission distribution ( e.g. @xcite and references therein ) .
bubbles are external to their optical and radio continuum counterparts , and expand at relatively low velocities ( @xmath8 10 ) . in most of the cases the derived dynamical ages are larger than the duration of the wr phase , suggesting that the stellar winds of the massive o - type star progenitor has also contributed to creating the structures . as part of a systematic search for neutral gas bubbles around massive stars
, we present here a study of the ism surrounding wr85 (= hd155603b = lss3982 ) based mainly on 21cm - line data belonging to the southern galactic plane survey ( sgps ) and additional molecular and radio continuum data .
wr85 is classified as wn6h+ob ? .
different distance estimates have been published for this star .
wr85 belongs to the hd155603 group , for which @xcite find a distance of 1.8 kpc . @xcite and @xcite locate the star at d = [email protected] and [email protected] kpc , respectively , while hipparcos measurements indicate 1.7 kpc .
the x - ray point like source 1wgaj1714.4 - 3949 coincides in position with the wr star @xcite .
hd155603b is associated with the optical ring nebula rcw118 , of about 25@xmath1 in diameter ( @xcite ; heckathorn , bruhweiler & gull 1982 , @xcite ; @xciteb ) .
@xcite observed rcw118 with the curtis - schmidt-0.6 m telescope at ctio in h@xmath7 , [ oiii]@xmath95007 and [ sii]@xmath96730 .
these authors found that the ionized gas in the nebula has a lsr velocity of 15 , and derived an expansion velocity @xmath10 10 and a dynamical age of 6.710@xmath5 yr , compatible with the duration of the wr phase .
based on the velocity of the ionized gas and on standard galactic rotation models , they estimated a kinematical distance d@xmath11 2.3 kpc .
@xcite reobserved the nebula using the same telescope .
their h@xmath7 image reveals the presence of two semi - circular rings .
the inner ring has a diameter of 93 and is centered on the wr star , whereas the outer one , 25 in diameter , is centered slightly to the south and south - east of the star .
the two concentric structures can be clearly identified in the h@xmath7 image displayed in figure [ optica ] .
rcw118 is in the same line of sight to the snr g347.30.5 , discovered in the rosat all sky survey ( rass ) by @xcite .
this snr , 1@xmath12 in diameter , is located at @xmath13 6 kpc ( see @xcite ) .
image in ( @xmath7 , @xmath14 ) coordinates showing the two concentic rings around wr85 ( taken from @xcite ) .
north is up and east is to the left.,width=317,height=283 ]
the 21cm - line emission data used in this paper belong to the southern galactic plane survey ( sgps ) ( @xcite ) obtained with the autralia telescope compact array ( atca ) and the parkes radiotelescope ( short spacing information )
. the data cube is centered at ( @xmath15 ) = ( 347@xmath1214,0@xmath1229,40 ) , covers a region of about 2@xmath122@xmath12 around the wr star , and has a synthesized beam of 2621 . to improve the s / n ratio we applied a hanning smoothing to the individual line images
consequently , the original rms noise level of 2.4 k was lowered to 1.3 k and the channel velocity resolution was doubled .
the main observational parameters of the final data cube are summarized in table [ tabla1 ] . .
data : main observational parameters [ cols="<,^ " , ]
we have analyzed 21 cm line data belonging to the southern galactic plane survey searching for an bubble related to wr85 and its ring nebula rcw118 .
the main results of our search can be summarized as follows : 1 .
the data allowed the identification of a cavity in the neutral gas distribution surrounded by a slowly expanding shell .
the structure was detected at velocities spanning the interval 28.0 to 16.5 , and has a systemic velocity of 21@xmath05 .
the cavity can also be identified in the co emission distribution at similar velocities .
2 . the ring nebula rcw118 appears projected onto the inner border of the shell .
the systemic velocity of the structure is similar to the velocity of the ionized gas in rcw118 .
wr85 is seen in projection onto the central part of the cavity and close to its geometrical center .
the coincidence , within errors , of the kinematical distance to the structure and the stellar distance strongly suggests that the wr star is placed within the cavity .
the morphological coincidence between the inner optical nebula and the emission at about 28 suggests that the 93 nebula is located on the approaching part of the expanding shell .
the wr star , which appears closely related to the inner nebula , is also probably located near the approaching part of the shell .
the co emission distribution reveals the presence of a molecular shell almost encircling the hi envelope .
the position of the iras protostellar candidates projected onto the molecular shell , suggests that star formation might be occuring in the region .
however , additional studies should be performed to investigate this point .
the stellar wind of wr85 is capable of blowing the observed ionized and neutral structure .
the bulk of evidence ( i ) to ( ix ) strongly indicates that the structure is the neutral gas counterpart of the optical bubble and shows the action of the stellar winds on the surrounding material . adopting a distance of [email protected] kpc ,
the neutral interstelar bubble has a linear radius of @xmath1621 pc .
taking into account an expansion velocity of @xmath172 , its dynamical age is @xmath18 yr , suggesting that both the present wr star and its massive stellar progenitor have contributed in shaping the bubble .
the associated neutral and ionized masses are 3000 m@xmath19 , which indicates that the bubble evolved in a medium with an average ambient density of 3 @xmath20 .
we acknowedge the referee dr a. p. marston for many suggestions that improved the final presentation of this paper .
we thank dr .
t. dame for making available to us the co data , and dr .
r. gamen for allowing us to use information on wr85 in advance of publication .
this project was partially financed by the consejo nacional de investigaciones cientficas y tcnicas ( conicet ) of argentina under project pip 607/98 and fcag , unlp under projects 11/g049 and 11/g072 , and agencia nacional
de promocin cienffica y tecnolgica ( anpcyt ) under project pict 14018/03 .
the digitized sky survey ( dss ) was produced at the space telescope science institute under us government grant nagw-2166 .
cappa c. , goss w. m. , van der hucht k. a. , 2004 , aj , 127 , 2885 cappa c. , arnal m. , cichowolski s. , goss w.m . , pineault s. 2003 , iau symp .
212 , a massive star odyssey , from main sequence to supernova , p. 596 prinja r.k . , barlow m.j .
, howarth i.d . , 1990,apj , 361 , 607 prinja r.k .
, fullerton a.w . ,
crowther p.a . , 1996 ,
a@xmath21a , 331 , 264 rochowicz k. , nieldzieski a. , 1995 , aca , 45 , 307 slane p. , gaensler b. m. , dame t. m.,hughes j. p. , plucinsky p. p. , green a. , 1999 , apj , 525 , 357 weaver r. , mccray r. , castor j. , shapiro , p. , moore r. , 1977 , apj , 218 , 377 | we analyze the distribution of the interstellar matter in the environs of the wolf - rayet star lss3982 (= wr85 , wn6+ob ? ) linked to the optical ring nebula rcw118 .
our study is based on neutral hydrogen 21cm - line data belonging to the southern galactic plane survey ( sgps ) .
the analysis of the data allowed the identification of a neutral hydrogen interstellar bubble related to wr85 and the 25-diameter ring nebula rcw118 .
the bubble was detected at a systemic velocity of 21.5 , corresponding to a kinematical distance of [email protected] kpc , compatible with the stellar distance .
the neutral stucture is about 25@xmath1 in radius or 21@xmath08 pc , and is expanding at 9@xmath02 .
the associated ionized and neutral masses amount to 3000 @xmath2 .
the co emission distribution depicts a region lacking co coincident in position and velocity with the structure .
the 93-diameter inner optical nebula appears to be related to the approaching part of the neutral atomic shell .
the void and shell are the neutral gas counterparts of the optical bubble and have very probably originated in the action of the strong stellar wind of the central star during the o - type and wr phases on the surrounding interstellar medium .
the bubble appears to be in the momentun conserving stage .
[ firstpage ] ism : bubbles stars : wolf - rayet ism : regions |
fluid and diffusion limits for queuing systems have been applied successfully toward performance analysis and optimization of various queuing systems .
we are concerned here with performance analysis in steady - state and , more specifically , with brownian steady - state approximations for continuous time markov chains ( ctmcs ) .
the framework of diffusion limits begins with a sequence of ctmcs @xmath2 , and properly scaled and centered versions @xmath3 for some sequence @xmath4 that arises from the specific structure of the model . with appropriate assumptions on the parameters of the ctmc , and on the sequence of initial conditions @xmath5 , one typically proceeds to establish process convergence @xmath6 in the appropriate function space where @xmath7 is a diffusion process .
if each of the @xmath8 as well as @xmath7 are ergodic , and @xmath9 is a continuous function such that @xmath10 is uniformly integrable , one can subsequently conclude that @xmath11\rightarrow{\mathbb{e}}\bigl[f\bigl({\widehat{x}}(\infty ) \bigr)\bigr ] \qquad\mbox{as } n\rightarrow\infty , \ ] ] where @xmath12 and @xmath13 have , respectively , the steady - state distributions of @xmath14 and @xmath7 . a relatively general framework toward proving the required uniform integrability has been developed in @xcite and applied there to generalized jackson networks ;
see also @xcite .
it was subsequently applied successfully to other queueing systems .
this so - called _ interchange of limits _ establishes that @xmath11={\mathbb{e}}\bigl[f\bigl ( { \widehat{x}}(\infty ) \bigr)\bigr]+o(1)\label{eq : gapintro},\ ] ] and supports using @xmath15 $ ] as an approximation for @xmath16 $ ] .
a central benefit of the limit approach to approximations is the relative tractability of the diffusion @xmath7 relative to the original ctmc .
the convergence rate embedded in the @xmath17 term is not , however , precisely captured by these convergence arguments . in this paper , we prove that an appropriately defined sequence of diffusion models , that are as tractable as the diffusion limit , provides accurate approximations for the steady - state of the ctmcs with an approximation gap that shrinks at a rate of @xmath1 . our approach does not require process convergence as in ( [ eq : fclt ] ) .
we proceed to an informal exposition of the results and key ideas .
the markov chains that we consider have a semi - martingale representation @xmath18 where @xmath19 is a local martingale with respect to a properly defined filtration .
we define a fluid model by ( heuristically ) removing the martingale term , that is , @xmath20 if the fm has a unique stationary point @xmath21 satisfying @xmath22 , it subsequently makes sense to center @xmath23 around @xmath24 and consider the centered and scaled process @xmath25 .
the process @xmath14 satisfies the equation @xmath26 where @xmath27 . under appropriate conditions , a strong approximation for @xmath28
is given by the diffusion process @xmath29 where @xmath30 is a standard brownian motion and @xmath31 arises naturally from the markov - chain transition functions and is intimately related to the predictable quadratic variation of the martingale @xmath19 .
strong - approximations theory predicts an approximation gap that is logarithmic in @xmath32 where @xmath33 is the time horizon ; see remark [ rem : strong ] .
a cruder approximation is obtained by replacing the ( state dependent ) diffusion coefficient with its value at the stationary point of the fm , @xmath21 , to obtain the diffusion process specified by the equation @xmath34 our main finding is that this straightforward heuristic derivation of the dm building on a stationary point of the fluid model to construct a simplified diffusion model may provide , insofar as steady - state analysis is concerned , an impressively accurate approximation .
more precisely , but still proceeding informally at this stage , we prove the following .
let @xmath35 be the generator of the diffusion @xmath36 . if there exists a function @xmath37 together with finite positive constants @xmath38 and a compact set @xmath30 ( all not depending on @xmath0 ) such that @xmath39 then @xmath40-{\mathbb{e}}\bigl[f\bigl ( { \widehat{x}}^n(\infty)\bigr)\bigr]={\mathcal{o}}(1/\sqrt{n})\ ] ] for all functions @xmath9 with @xmath41 . the uniform lyapunov requirement ul must be proved on a case - by - case basis , and we illustrate this via two examples in section [ sec : examples ] .
the requirement ul restricts the scope of our results to ( sequences of ) chains in which the corresponding dm is exponentially ergodic .
the sequence of poisson equations ( associated with the sequence of dms ) is central to our proofs .
let @xmath42 be the steady - state distribution of the diffusion model and @xmath43 be that of the scaled ctmc .
let @xmath9 be such that @xmath44 .
[ the requirement that @xmath44 is not necessary and is imposed in this discussion for expositional purposes . ]
we will show that a solution @xmath45 exists for the dm s poisson equation @xmath46 based on it s rule one expects that @xmath47={\mathbb{e}}_{\pi^n } \bigl[u^n_f \bigl({\widehat{y}}^n(0)\bigr ) \bigr]+{\mathbb{e}}_{\pi^n } \biggl[\int _ 0^t { \mathcal{a}}^n u^n_f \bigl({\widehat{y}}^n(s)\bigr)\,ds \biggr].\ ] ] since the dm has , by construction , a diffusion coefficient that does not depend on the state , the poisson equation is ( for each @xmath0 ) a linear pde , and we are able to build on existing theory to identify gradient estimates that are uniform in the index @xmath0 .
these gradient estimates facilitate proving that @xmath48={\mathbb{e}}_{\nu^n } \bigl[u^n_f\bigl ( { \widehat{x}}^n(0)\bigr ) \bigr]+{\mathbb{e}}_{\nu^n } \biggl[\int _ 0^t { \mathcal{a}}^n u^n_f \bigl({\widehat{x}}^n(s)\bigr)\,ds \biggr]+t{\mathcal{o}}(1/\sqrt{n}).\ ] ] informally speaking , this shows that @xmath49 `` almost solves '' the poisson equation for the ctmc .
stationarity then allows us to conclude that @xmath50=-t{\mathbb{e}}_{\nu ^n } \biggl[\int_0^t f\bigl({\widehat{x}}^n(s)\bigr)\,ds \biggr]=t{\mathcal{o}}(1/\sqrt{n}),\ ] ] and , in particular , that @xmath51 recalling that @xmath44 , it then follows that @xmath52 in the process of proving these results , we explore connections between the stability of the ctmc and that of the corresponding fm and dm .
refined properties of the poisson equation in the context of diffusion approximations for diffusions with a fast component are used in @xcite . in the spirit of this paper , derivative bounds for certain dirichlet problems
are used in @xcite to study _ universal _
approximations for the birth - and - death process underlying the so - called erlang - a queue .
the proofs there are based on the study of excursions but are closely related to ours ; we revisit the erlang - a queue in section [ sec : examples ] .
the use of gradient estimates in conjunction with martingale arguments is also the theme in @xcite where these are used to study optimality gaps in the control of a multi - class queue .
the poisson equation is replaced there with the pde associated with the hjb equation . _
notation_. unless stated otherwise , all convergence statements are for @xmath53 .
we use @xmath54 to denote the euclidean norm of @xmath55 in @xmath56 ( the dimension @xmath57 will be clear form the context ) . for two nonnegative sequences
@xmath58 and @xmath59 we write @xmath60 if @xmath61 . throughout we adopt the convention that @xmath62 .
we let @xmath63 and denote its closure by @xmath64 . following standard notation ,
we let @xmath65 be the space of @xmath66-times continuously differentiable functions from @xmath56 to @xmath67 , and for @xmath68 we let @xmath69 and @xmath70 denote the gradient and the hessian of @xmath71 , respectively .
given a markov process @xmath72 on a complete and separable metric space @xmath73 , we let @xmath74 be the probability distribution under which @xmath75 for @xmath76 and @xmath77 = { \mathbb{e}}[\cdot|\xi(0 ) = x]$ ] be the expectation operator w.r.t . the probability distribution @xmath74 .
let @xmath78 denote the probability distribution under which @xmath79 is distributed according to @xmath80 and put @xmath81 $ ] to be the expectation operator w.r.t .
this distribution . a probability distribution @xmath80 defined on @xmath73
is said to be a stationary distribution if for every bounded continuous function @xmath9 @xmath82 = { \mathbb{e}}_{\pi}\bigl[f \bigl(\xi(0)\bigr)\bigr]\qquad \mbox{for all } t\geq0.\ ] ] it is said to be the steady - state distribution if for every such function and all @xmath76 , @xmath83 \rightarrow{\mathbb{e}}_{\pi } \bigl[f\bigl(\xi(0)\bigr)\bigr ] \qquad\mbox{as } t\rightarrow\infty.\ ] ] given a probability distribution @xmath84 and a nonnegative function @xmath9 , we define @xmath85 ( which can be infinite ) . for a general ( not necessarily nonnegative ) function , we define @xmath86 as above whenever @xmath87 . finally , whereas our results are not concerned with process - convergence , we will be making connections to the functional central limit theorem .
all the processes that we study are assumed to be right continuous with left limits ( rcll ) , and @xmath88 will be used for convergence in the space @xmath89 of such functions unless otherwise stated . for rcll processes we use @xmath90 and let @xmath91 .
we consider a sequence @xmath92 of continuous time markov chains ( ctmcs ) .
the chain @xmath23 moves on a countable state space @xmath93 according to transition rates @xmath94 for @xmath95 .
given a nonrandom initial condition @xmath96 , the dynamics of @xmath23 are constructed as follows : @xmath97 where @xmath98 and @xmath99 are independent unit - rate poisson processes ; see @xcite , section 6.4 . letting @xmath100 , we rewrite @xmath101 where @xmath102 provided that @xmath23 is nonexplosive , @xmath103 is a local martingale with respect to the filtration @xmath104 see @xcite , theorem 6.4.1 .
the local ( predictable ) quadratic variation of @xmath19 is given by @xmath105 where @xmath106 in essence , @xmath107 and @xmath108 are defined only for values in @xmath109 .
we henceforth assume that they are extended to @xmath56 and , with some abuse of notation , denote by @xmath107 and @xmath108 these extensions .
the requirements that we impose on these extensions will be clear in what follows .
_ fluid models_. given @xmath55 , we define the @xmath0th fluid model by @xmath110 or , in differential form , @xmath111 if @xmath107 is lipschitz continuous , the fluid model has a solution .
we will assume that there exists a unique @xmath21 satisfying @xmath112 this requirement is intimately linked to our lyapunov requirement ; see lemma [ asum : l ] . _ centered and scaled process .
_ define the processes @xmath113 and denote by @xmath114 the state space of @xmath115 .
letting @xmath116 we have @xmath117 the martingale @xmath118 has the local predictable quadratic variation process @xmath119 where @xmath120 _ assumptions .
_ we assume that the jump sizes are bounded uniformly in @xmath0 : @xmath121 and that @xmath0 is sufficiently large so that @xmath122 . the sequence @xmath123 is assumed to be uniformly lipschitz , and @xmath124 is assumed to have linear growth around @xmath125 .
formally , there exist constants @xmath126 , @xmath127 such that , for all @xmath0 , @xmath128 and @xmath129 the requirements ( [ eq : flip ] ) and @xmath130 guarantee , in particular , that @xmath131 . condition ( [ eq : alip ] ) is equivalently stated in terms of the ( unscaled ) @xmath108 as @xmath132 we further assume that _
@xmath133 is positive definite for each @xmath0 _ and that @xmath134 where @xmath135 is itself positive definite . the matrix @xmath135 is not used in specifying the diffusion model in section [ sec : main ] , but the assumption of convergence is used in our proofs , most notably in that of theorem [ thmm : expo_ergo ] . in various settings ,
including our own examples in section [ sec : examples ] , @xmath136 in which case the convergence requirement is trivially satisfied .
the requirement that the continuous extension @xmath137 satisfies the uniform lipschitz requirement ( [ eq : flip ] ) is a restriction .
it excludes , for example , single - server queueing systems ; we revisit this point in section [ sec : conclusions ] . for each @xmath138 ,
@xmath23 is nonexplosive , irreducible , positive recurrent and satisfies ( [ eq : boundedjumps])([eq : quadraticvar_fluid ] ) .
[ asum : base ] positive recurrence and irreducibility imply ergodicity of @xmath23 and , in particular , the existence of a steady - state distribution ( which is also the unique stationary distribution ) . in certain cases ,
positive recurrence of @xmath23 need not be a priori assumed ; see theorem [ thmm : ctmc_lyap ] and remark [ rem : allinone ] .
assumption [ asum : base ] is imposed for the remainder of this paper .
recall that @xmath139 is a stationary point for the _ fluid model _
@xmath140 and that @xmath141 .
fix a probability space and a @xmath57-dimensional brownian motion , and let @xmath36 be the strong solution to the sde @xmath142 the existence and uniqueness of a strong solution follow from the lipschitz continuity and linear growth of @xmath137 and the constant diffusion coefficient ; see , for example , @xcite , theorems 5.2.5 and 5.2.9 .
[ rem : strong]the strong approximation for @xmath14 is a diffusion obtained ( heuristically at first ) by taking the `` density '' @xmath143 of the quadratic variation in ( [ eq : m_quad_var ] ) as the diffusion coefficient , to define the process @xmath144 the process @xmath145 provides a `` good '' approximation for the dynamics of the ctmc in the sense that @xmath146 where @xmath147 are random variables with exponential tails ( uniformly in @xmath0 ) ; see , for example , @xcite , chapters 7.5 and 11.3 . given the cruder ( state independent ) diffusion coefficient , the dm @xmath36 is not likely to be as precise , over finite horizons , as the strong approximation . in terms of tractability , however , the analysis of steady - state is simpler for the dm , insofar as its steady - state distribution ( when it exists ) involves linear pdes ; see , for example , @xcite , chapter 4.9 .
our main result , theorem [ thmm : main ] , shows that this increased tractability co - exists with an impressive steady - state - approximation accuracy .
suppose that , in addition , assumption [ asum : base ] @xmath148 uniformly on compact subsets of @xmath56 . if @xmath149 , then @xmath150 where @xmath151 is the strong solution to the sde @xmath152 with @xmath153 and @xmath154 is as in ( [ eq : quadraticvar_fluid ] ) ; see @xcite , theorem 6.5.4 . given ( [ eq : betaconv ] ) , requirements ( 5.9 ) and ( 5.14 ) of that theorem are trivially satisfied here due to the bounded jumps . the final requirement in @xcite , theorem 6.5.4 , that @xmath155 has @xmath156 almost surely , follows immediately from the fact that @xmath151 is a strong solution .
further , it is easily proved that @xmath157 .
thus , within a diffusion - limit framework , the dm is consistent with the diffusion limit in the sense that @xmath36 and @xmath14 converge to the same limit . for functions @xmath158 ,
the generator of @xmath36 coincides with the second order differential operator @xmath35 defined , for such functions , by @xmath159 see , for example , @xcite , proposition 5.4.2 .
we next state the uniform lyapunov assumption .
we say that @xmath160 is a _ norm - like _ function if @xmath161 as @xmath162 .
a function @xmath163 is said to be sub - exponential if @xmath164 and there exist constants @xmath165 and @xmath166 such that @xmath167 and @xmath168 [ asum : l ] there exist a sub - exponential norm - like function @xmath169 and finite positive constants @xmath170 ( not depending on @xmath0 ) such that @xmath171 and , for each @xmath0 and all @xmath172 , @xmath173 < \infty , \qquad t \geq0 .
\label{eq : finite_integral}\ ] ] assumption [ asum : l ] is imposed for the remainder of this paper .
the requirement that @xmath164 is made without loss of generality .
if a norm - like function @xmath37 satisfies ul , there exists re - defined constants @xmath38 and @xmath174 such that @xmath175 satisfies ul .
all polynomials @xmath176 satisfy ( [ eq : expobound1 ] ) and ( [ eq : expobound])the former is used only in the proof of lemma [ lem : fromdisctocont ] , and the latter is used in the derivations of gradient bounds following the statement of theorem [ thmm : pde ] .
requirement ( [ eq : finite_integral ] ) is relatively unrestrictive as it is imposed on each individual @xmath0 ( rather than uniformly in @xmath0 ) .
lyapunov conditions are frequently used in the context of stability of continuous time markov processes ( corresponding to fixed @xmath0 here ) ; see @xcite .
the requirement of a uniform lyapunov condition imposed on a family of markov processes is less common ( see @xcite for a related example ) . in section [ sec : examples ]
we study two examples for which all the requirements of assumption [ asum : l ] are met . with assumption [ asum : l ] , the existence and uniqueness of a steady - state distribution , @xmath42 , for @xmath36 follows from @xcite , sections 4 and 6 , as does the fact that @xmath36 is exponentially ergodic and that , for each @xmath0 , @xmath177 for all functions @xmath9 with @xmath41 ; see @xcite , theorem 4.2 . for @xmath37 that satisfies ( [ eq : expobound1 ] ) we have , for all @xmath178 and @xmath179 , that @xmath180=v(x)+ { \mathbb{e}}_x \biggl[\int_0^{t } \mathcal{a}^n v\bigl({\widehat{y}}^n(s)\bigr)\,ds \biggr ] ;
\label{eq : dynkin}\ ] ] see , for example , @xcite , theorem 6.3 .
ul then guarantees that @xmath180\leq v(x)+ { \mathbb{e}}_x \biggl[\int_0^t \bigl(- \delta v\bigl({\widehat{y}}^n(s)\bigr)+b\bigr)\,ds \biggr]\label{eq : tbound}\ ] ] for all @xmath178 and @xmath179 and , consequently , that @xmath181 for all functions @xmath9 with @xmath41 ; see also @xcite , corollary 2 .
important for our analysis is the following consequence of assumption [ asum : l ] .
[ thmm : expo_ergo ] let @xmath42 be the steady - state distribution of @xmath36 . then there exist finite positive constants @xmath182 and @xmath183 such that @xmath184-\pi^n(f ) \bigr|\leq\mathcal{m}e^{-\mu
t},\qquad t\geq 0.\ ] ] bounds on the convergence rate of exponentially ergodic markov processes to their steady - state distribution have been studied extensively in recent literature .
our proof builds specifically on @xcite .
the constants @xmath182 and @xmath183 are related to a minorization condition for the discrete - time process @xmath185 . in the standard application , these constants may depend on @xmath0 . to obtain constants that can be used for all @xmath138
we must argue that a minorization condition is satisfied uniformly in @xmath0 ; the proof of theorem [ thmm : expo_ergo ] is postponed to section [ sec : expo_ergo ] .
theorem [ thmm : expo_ergo ] has the following important implication : fixing a function @xmath9 with @xmath41 and @xmath44 , we have for all @xmath179 , that @xmath186\bigr{\vert}\leq\mathcal{m } v(x)e^{-\mu t},\qquad t\geq0,\ ] ] so that @xmath187\bigr { \vert}\,ds\leq \mathcal{m } v(x)\int_0^{\infty}e^{-\mu s } \,ds = c v(x)<\infty\ ] ] for all @xmath179 , where the constant @xmath188 does not depend on @xmath0 or @xmath55 .
we conclude that @xmath189\,ds\ ] ] is a well - defined function of @xmath179 and that , for all @xmath0 , @xmath190 also , for any fixed @xmath191 and @xmath138 , @xmath192\,ds-\int_0^{\infty } { \mathbb{e}}_x\bigl[f\bigl({\widehat{y}}^n(s)\bigr)\bigr]\,ds\biggr{\vert}=0.\label { eq : poisson_convergence}\ ] ] define @xmath193 and @xmath194 the introduction of @xmath195 is motivated by the analysis of the ( sequence of ) poisson equations , specifically by the gradient estimates that require bounds on local fluctuations of @xmath9 ; see the derivations following theorem [ thmm : pde ] .
our main result , stated next , establishes that the steady - state distribution of the markov chain and the dm are suitably close provided that moments of the former are uniformly bounded .
[ thmm : main ] fix @xmath37 that satisfies assumption [ asum : l ] and a function @xmath9 such that @xmath44 and @xmath196 .
let @xmath43 and @xmath42 be , respectively , the steady - state distributions of @xmath115 and @xmath36 . if @xmath197 then @xmath198 theorem [ thmm : main ] and the remaining results of this section are proved in section [ sec : ito ] . if @xmath9 satisfies @xmath196 but @xmath199 , consider instead the function @xmath200
. then @xmath201 .
by ( [ eq : fbound ] ) , @xmath202 and , in turn , @xmath203 .
further , @xmath204 satisfies that @xmath205 . finally , if @xmath37 satisfies assumption [ asum : l ] , so does the function @xmath206 .
thus the results that follow hold for functions @xmath9 with @xmath196 regardless of whether @xmath44 or not . in general , proving requirement ( [ eq : requirement ] ) ( which implies , in particular , tightness of the sequence @xmath207 of steady - state distributions ) is far from trivial .
as we show next ( [ eq : requirement ] ) can be argued in advance in our setting .
one expects that , as @xmath0 grows , the property ( [ eq : tbound ] ) of the dm will be approximately valid for the ctmc allowing to draw an implication similar to ( [ eq : fbound ] ) with @xmath36 there replaced by @xmath14 .
the next theorem shows that this intuition is valid provided that @xmath37 satisfies additional simple properties .
given a function @xmath208 , define for @xmath179 , @xmath209_{2,1,b_x ( \bar{\ell}/\sqrt{n } ) } = \sup_{y , z\in b_x ( \bar{\ell}/\sqrt{n } ) } \frac{|d^2\psi(y)-d^2\psi ( z)|}{|y - z|},\label{eq : square_bracket}\vadjust{\goodbreak}\ ] ] where the right - hand side may be infinite .
[ thmm : ctmc_lyap ] let @xmath37 be as in assumption [ asum : l ] .
suppose , in addition , that there exists a finite positive constant @xmath188 such that , for each @xmath0 , and all @xmath179 , @xmath210_{2,1,b_x(\bar{\ell}/\sqrt { n})}\bigr ) \bigl(1+|x|\bigr)\leq cv(x).\label{eq : dmtoctmc2}\ ] ] then , for all sufficiently large @xmath0 , and all @xmath172 , @xmath211\leq v(x)+ { \mathbb{e}}_x \biggl[\int_0^t \biggl(- \frac { \delta } { 2}v\bigl(\widehat{x}^n(s)\bigr)+b \biggr ) \,ds
\biggr],\qquad t\geq0 , \label{eq : lyap_ctmc}\ ] ] where @xmath212 is as in assumption [ asum : l ] . consequently ,
@xmath28 is ergodic for all such @xmath0 and , furthermore , @xmath213 if @xmath214 , condition ( [ eq : dmtoctmc2 ] ) can be replaced with @xmath215 using taylor s theorem we have , for all @xmath179 , that @xmath216_{2,1,b_x(\bar{\ell}/\sqrt { n})}&\leq & \sup_{\eta\in b_x(\bar{\ell}/\sqrt{n})}2\bigl(1+| \eta|\bigr)\bigl|d^3v(\eta)\bigr| \\ & \leq & 2 c \bigl(\sup_{\eta\in b_x(\bar{\ell}/\sqrt{n})}v(\eta ) \bigr)\leq 2c_3cv(x),\end{aligned}\ ] ] where the last inequality follows from the sub - exponential property ( [ eq : expobound ] ) of @xmath37 and @xmath217 . note that ( [ eq:3diff1 ] ) is satisfied by any polynomial @xmath164 .
fix @xmath37 that satisfies assumption [ asum : l ] .
suppose that there exists @xmath218 that , itself , satisfies assumption [ asum : l ] as well as ( [ eq : dmtoctmc2 ] ) and @xmath219 then , @xmath220 and , in particular , ( [ eq : requirement ] ) holds for @xmath37.[cor : tightness ] suppose that @xmath214 and satisfies assumption [ asum : l ] and ( [ eq:3diff1 ] ) .
if there exists @xmath221 such that @xmath222 and @xmath223 satisfies ( [ eq : finite_integral ] ) , then we can take @xmath224 in corollary [ cor : tightness ] .
indeed , for an integer @xmath225 , @xmath226 with @xmath227 and @xmath212 as in assumption [ asum : l ] and @xmath188 as in ( [ eq:3diff1 ] )
. thus if @xmath214 is sub - exponential and satisfies ul and ( [ eq:3diff1 ] ) , so does @xmath223 .
[ rem : simplecase ] combined , theorem [ thmm : main ] and corollary [ cor : tightness ] establish the following : if there exist functions @xmath37 and @xmath218 both satisfying assumption [ asum : l ] such that ( [ eq : dmtoctmc2 ] ) holds for @xmath218 and @xmath228 , then we _ simultaneously _ have : ( i ) the positive recurrence of @xmath14 for sufficiently large @xmath0 , ( ii ) the moment bound in ( [ eq : requirement ] ) ( which implies , in particular , the tightness of @xmath43 ) and ( iii ) the @xmath229 approximation gap . with the exception of the simple requirement ( [ eq : finite_integral ] ) ,
this reduces the requirements to properties of the dm .
[ rem : allinone ] we conclude this section with an observation pertaining to the connection between the stability of the fm and the dm .
suppose that there exist a norm - like function @xmath37 and a constant @xmath230 such that @xmath231 letting @xmath232 we have @xmath233 so that the fm is stable in the sense that , for each @xmath0 and any initial condition @xmath234 , @xmath235 as @xmath236 .
moreover , @xmath237 \\[-8pt ] \nonumber & \leq&-\eta\bigl(v(y)-v(0)\bigr)+\bigl|\bar { a}^n(0)\bigr|\bigl|d^2v(y)\bigr|.\end{aligned}\ ] ] the following is an immediate consequence .
let @xmath163 be a sub - exponential norm - like function satisfying ( [ eq : finite_integral ] ) and ( [ eq : fmlyap ] ) . if @xmath238 then @xmath37 satisfies ul and , in turn , assumption [ asum : l ] .
[ lem : fmtodm ]
in what follows , fixing a set @xmath239 , @xmath240 denotes the space of twice continuously differentiable functions from @xmath241 to @xmath242 . for @xmath243 , recall that @xmath69 and @xmath70 denote the gradient and the hessian of @xmath71 , respectively .
the space @xmath244 is then the subspace of @xmath245 members of which have second derivatives that are lipschitz continuous on @xmath241 .
that is , a twice continuously differentiable function @xmath246 is in @xmath244 if @xmath247_{2,1,{\mathcal{b}}}=\sup_{x , y\in{\mathcal{b } } , x\neq y } \frac { |d^2u(x)-d^2u(y)|}{|x - y|}<\infty.\ ] ] [ in equation ( [ eq : square_bracket ] ) the set @xmath241 is taken to be @xmath248 . ]
we define @xmath249 where @xmath250 stands for the boundary of @xmath241 , and we let @xmath251 . we define
@xmath252_{j,{\mathcal{b}}}^*+\sup_{x , y\in{\mathcal{b}},x\neq y}\,d_{x , y}^{3 } \frac{|d^2u(x)-d^2 u(y)|}{|x - y| } , \label{eq : ustardefin}\ ] ] where @xmath253_{j,{\mathcal{b}}}^*=\sup_{x\in { \mathcal{b}}}d_x^j |d^j u(x)|$ ] for @xmath254 .
above @xmath255 ( resp . , @xmath256 ) denotes the @xmath66th power of @xmath257 ( resp .
, of @xmath258 ) .
we let @xmath259_{0,{\mathcal{b}}}^*=\sup_{x\in{\mathcal{b}}}|u(x)|$ ] , and @xmath260 we say that the function is locally lipschitz if @xmath261 for all @xmath179 , where @xmath262 is as in ( [ eq : mbxdefin ] ) .
[ thmm : pde ] fix @xmath37 that satisfies assumption [ asum : l ] and a locally lipschitz function @xmath9 with @xmath41 and @xmath44 .
then , for each @xmath0 , the poisson equation @xmath263 has a unique solution @xmath264 given by @xmath265\,dt.\label { eq : poissonsolution}\ ] ] moreover , there exist a finite positive constant @xmath266 ( not depending on @xmath0 ) such that @xmath267 consequently , for all @xmath0 and @xmath179 , @xmath268 and @xmath269_{2,1,{\mathcal{b}}_x}\leq8\theta \bigl(\bigl|u_f^n\bigr|_{0,{\mathcal{b}}_x}+|f|^{(2)}_{0,1,\mathcal{b}_x } \bigr)\bigl ( 1+|x|\bigr)^3.\label{eq : pde2}\ ] ] several observations are useful for what follows : recall ( [ eq : ubound ] ) that @xmath270 for some constant @xmath188 . by the assumed sub - exponentiality of @xmath37 @xmath271 for all @xmath272 , where @xmath166 is as in ( [ eq : expobound ] ) . in turn , @xmath273 for a function @xmath9 with @xmath196 [ see ( [ eq : barf ] ) ] and for all @xmath272 , @xmath274 so that @xmath275 for all @xmath179 .
defining @xmath276 we have , by theorem [ thmm : pde ] ( and assuming , without loss of generality that @xmath277 ) , that for all @xmath138 and @xmath179 , @xmath278_{2,1,b_x(\bar{\ell}/\sqrt { n})}&\leq & c_v(x).\nonumber\end{aligned}\ ] ] proof of theorem [ thmm : pde ] we first prove that @xmath49 in ( [ eq : poissonsolution ] ) solves the poisson equation ( [ eq : poisson ] ) .
since @xmath9 is fixed throughout we omit it from the notation .
fixing @xmath279 , let @xmath280 be the solution to dirichlet problem @xmath281 in the boundary condition , @xmath282 is as in ( [ eq : poissonsolution ] ) .
the existence and uniqueness of a solution @xmath283 follows directly from @xcite , theorem 6.13 , recalling that @xmath137 is lipschitz continuous and @xmath133 is a constant matrix and hence trivially lipschitz .
theorem 6.13 of @xcite requires that @xmath280 is continuous in @xmath55 on @xmath284 .
this follows exactly as in part ( c ) of @xcite , theorem 1 , using ( [ eq : poisson_convergence ] ) .
we omit the detailed argument .
it follows that @xmath285,\ ] ] where @xmath286 ; see @xcite , proposition 5.7.2 and lemma 5.7.4 .
we have that @xmath287 with @xmath288 as in ( [ eq : poissonsolution ] ) .
this assertion is proved as in @xcite , theorem 1 , part ( d ) . since @xmath279 is arbitrary
we conclude that , @xmath288 solves the poisson equation ( [ eq : poisson ] ) . to establish the gradient estimates observe that , since @xmath133 is bounded in @xmath0 , there exists a constant @xmath289 ( not depending on @xmath0 ) such that ( with the notation in @xcite , theorem 6.2 ) @xmath290 . from the positive definiteness of @xmath133 , and since @xmath291 for a positive definite @xmath135
, it follows that there exists a constant @xmath292 such that @xmath293 for all @xmath0 and all @xmath294 . finally , following the notation in @xcite , theorem 6.2 , @xmath295_{0,1,{\mathcal{b}}_x}^{(1 ) } \\ & = & \bigl[{\widehat{f}}^n\bigr]_{0,{\mathcal{b}}_x}^{(1)}+\sup _ { y , z\in{\mathcal{b}}_x } d_{y , z}^2 \frac { |{\widehat{f}}^n(y)-{\widehat{f}}^n(z)|}{|y - z| } \\ & = & \sup_{y\in{\mathcal{b}}_x } d_y\bigl|{\widehat{f}}^n(y)\bigr|+\sup _ { y , z } d_{y , z}^2 \frac { |{\widehat{f}}^n(y)-{\widehat{f}}^n(z)|}{|y - z| } \\ & \leq & 2 k_f,\end{aligned}\ ] ] where @xmath126 is as in ( [ eq : flip ] ) . in turn , by @xcite , theorem 6.2 , that @xmath296 where @xmath266 depends only on @xmath297 and the constant @xmath298 in ( [ eq : lambda ] ) ( for @xmath299 there , we take @xmath300 ) .
bounds ( [ eq : pde0])([eq : pde2 ] ) now follow from the definition of @xmath301 applied to points in the subset @xmath302 of @xmath262 .
specifically , for each @xmath303 , @xmath304_{1,{\mathcal{b}}_x}^ * \leq\bigl|u_f^n\bigr|_{2,1,\mathcal{b}_x}^*.\ ] ] noting that @xmath305 for all @xmath306 we have , for all such @xmath307 ( in particular for @xmath55 itself ) , that @xmath308 equations ( [ eq : pde1 ] ) and ( [ eq : pde2 ] ) are argued similarly .
the following simple lemma is proved in the . given a function @xmath309
we write @xmath310 for the @xmath311th coordinate of @xmath312 and @xmath313 for the @xmath314th coordinate of @xmath315 .
[ lem : ito ] let @xmath309 be such that , for all @xmath316 and @xmath178 , @xmath317 \\[-8pt ] \nonumber & & \hspace*{80pt}{}+ [ \psi ] _ { 2,1,b_{\widehat{x}^n(s)}(\bar{\ell}/\sqrt{n } ) } \bigr ) \bigl(1+\bigl | \widehat{x}^n ( s)\bigr| \bigr ) \,ds \biggr]<\infty.\end{aligned}\ ] ] then , for all @xmath318 and @xmath178 , @xmath319= \psi(x)+{\mathbb{e}}_x \biggl[\int_0^t { \mathcal{a}}^n \psi \bigl(\widehat{x}^n(s)\bigr)\,ds \biggr]+a_{\psi}^{n , x}(t)+d_{\psi } ^{n , x}(t),\label{eq : almostito}\ ] ] where @xmath320 is as in ( [ eq : diff_gen ] ) and , for all @xmath321 and @xmath178 , @xmath322_{2,1,b_{\widehat{x}^n(s)}(\bar{\ell}/\sqrt{n})}\bigl|\bar { a}^n\bigl(\widehat{x}^n(s)\bigr)\bigr| \,ds \biggr ] , \\
d_{\psi}^{n , x}(t)&=&\frac{1}{2}{\mathbb{e}}_x \biggl[\sum_{i , j}^d \int _ 0^t \psi _ { ij}\bigl ( \widehat{x}^n(s)\bigr ) \bigl(\bar{a}_{ij}^n \bigl(\widehat{x}^n(s)\bigr)-\bar { a}_{ij}^n(0 ) \bigr)\,ds \biggr].\end{aligned}\ ] ] below @xmath195 is as in ( [ eq : barf ] ) and @xmath323 as in ( [ eq : cfdefin ] ) .
[ cor : interim ] fix @xmath37 that satisfies assumption [ asum : l ] and a function @xmath9 such that @xmath196 .
then there exists a finite positive constant @xmath188 ( not depending on @xmath0 ) , such that , for all @xmath316 and @xmath178 , @xmath324- u_f^n(x)-{\mathbb{e}}_x \biggl [ \int_0^t { \mathcal{a}}^n u_f^n\bigl(\widehat{x}^n(s ) \bigr)\,ds \biggr ] \biggr| \\ & & \qquad\leq c \biggl({\mathbb{e}}_x \biggl[\int_0^t \frac{c_v(\widehat{x}^n(s))}{\sqrt{n } } \biggl(1+\frac{|\widehat { x}^n(s)|}{\sqrt { n } } \biggr)\,ds \biggr ] \biggr).\end{aligned}\ ] ] by ( [ eq : gradient_est ] ) we have , for @xmath179 , that @xmath325_{2,1,b_x ( { \bar{\ell}}/{\sqrt{n } } ) } \bigr ) \bigl(1+|x|\bigr ) & \leq&3 c_v(x ) \bigl(1+|x|\bigr ) \\ & \leq&\varepsilon\bigl(1+|x|\bigr)^4v(x)\end{aligned}\ ] ] for some finite positive constant . by assumption [ asum : l ] , specifically ( [ eq : finite_integral ] ) , @xmath326<\infty,\ ] ] so that @xmath37 satisfies the requirements of lemma [ lem : ito ] , and we have that @xmath327\nonumber \\ & \leq&\frac{k_a}{2\sqrt{n}}{\mathbb{e}}_x \biggl [ \int _ 0^t\bigl |d^2 u^n_f \bigl(\widehat{x}^n(s)\bigr)\bigr|\bigl|\widehat{x}^n(s)\bigr|\,ds \biggr ] \\ &
\leq & \frac{k_a}{2\sqrt{n}}{\mathbb{e}}_x \biggl[\int_0^t c_v\bigl(\widehat { x}^n(s)\bigr)\,ds \biggr ] .
\nonumber\end{aligned}\ ] ] the second inequality follows from ( [ eq : alip ] ) .
the last inequality follows from ( [ eq : gradient_est ] ) .
next , @xmath328_{2,1,b_{\widehat { x}^n(s)}(\bar { \ell } /\sqrt{n})}\bigl|\bar{a}^n
\bigl(\widehat{x}^n(s)\bigr)\bigr|\,ds \biggr ] \nonumber\\ & \leq & \frac{\bar{\ell}}{2\sqrt{n}}{\mathbb{e}}_x \biggl[\int _
0^t \bigl[u^n_f \bigr]_{2,1,b_{\widehat{x}^n ( s)}(\bar{\ell}/\sqrt{n})}\bigl|\bar{a}^n(0)\bigr|\,ds \biggr ] \\ & & { } + \frac{\bar{\ell}}{2\sqrt{n}}{\mathbb{e}}_x \biggl[\int_0^t \bigl[u^n_f\bigr]_{2,1,b_{\widehat{x}^n(s)}(\bar{\ell}/\sqrt{n})}\bigl|\bar { a}^n \bigl(\widehat{x}^n ( s)\bigr)-\bar { a}^n(0)\bigr|\,ds \biggr ] .
\nonumber\end{aligned}\ ] ] using ( [ eq : alip ] ) , ( [ eq : quadraticvar_fluid ] ) and ( [ eq : gradient_est ] ) we conclude that @xmath329,\ ] ] which completes the proof .
we are ready to prove theorem [ thmm : main ] .
proof of theorem [ thmm : main ] as @xmath43 is a stationary distribution we have , by ( [ eq : ubound ] ) and ( [ eq : requirement ] ) , that @xmath330={\mathbb{e}}_{\nu^n}\bigl[u_f^n \bigl(\widehat { x}^n(0)\bigr)\bigr]\leq c\nu ^n(v)<\infty\ ] ] for all sufficiently large @xmath0 and all @xmath178 . recalling that @xmath331 , corollary [ cor : interim ] guarantees the existence of a finite positive constant @xmath332 ( not depending on @xmath0 ) such that @xmath333 \biggr{\vert}&\leq&\vartheta { \mathbb{e}}_{\nu^n } \biggl[\int_0^t \frac{c_v(\widehat{x}^n ( s))}{\sqrt { n } } \biggl(1+\frac{|\widehat{x}^n(s)|}{\sqrt{n } } \biggr)\,ds \biggr ] \nonumber \\[-8pt ] \\[-8pt ] \nonumber & = & \vartheta t{\mathbb{e}}_{\nu^n } \biggl[\frac { c_v(\widehat{x}^n(0))}{\sqrt{n } } \biggl(1 + \frac{|\widehat { x}^n(0)|}{\sqrt{n } } \biggr ) \biggr]\end{aligned}\ ] ] for all @xmath178 , where the interchange of integral and expectation is justified by the nonnegativity of the integrands . using again ( [ eq : requirement ] ) and the nonnegativity of @xmath37 we have , for all @xmath334 , that @xmath335\leq{\mathbb{e}}_{\nu^n } \biggl [ \int _ 0^t v\bigl(\widehat{x}^n(s ) \bigr)\,ds \biggr]= t\nu^n(v)<\infty.\ ] ] this justifies replacing integral and expectation in ( [ eq : inter1 ] ) to conclude that , with @xmath336 , @xmath337\biggr{\vert}\leq\vartheta{\mathbb{e}}_{\nu^n } \biggl [ \frac { c_v(\widehat{x}^n ( 0))}{\sqrt{n } } \biggl(1+\frac{|\widehat{x}^n(0)|}{\sqrt{n } } \biggr ) \biggr ] \\ & = & { \mathcal{o}}(1/\sqrt{n})\end{aligned}\ ] ] for a ( re - defined ) constant @xmath332 as required , where the last equality follows from ( [ eq : requirement ] ) recalling the definition of @xmath323 in ( [ eq : cfdefin ] ) . proof of theorem [ thmm : ctmc_lyap ] let @xmath37 be as in assumption [ asum : l ] . applying lemma [ lem : ito ] as in the proof of corollary [ cor : interim ] we have that @xmath338_{2,1,b_{\widehat{x}^n ( s ) } ( { \bar{\ell}}/{\sqrt{n } } ) } \bigl| \bar{a}^n(0)\bigr|\,ds \biggr ] \\ & & { } + \frac{\bar{\ell}}{2\sqrt{n}}{\mathbb{e}}_x \biggl[\int_0^t [ v]_{2,1,b_{\widehat{x}^n(s ) } ( { \bar{\ell}}/{\sqrt { n } } ) } \bigl|\bar { a}^n\bigl(\widehat{x}^n(s ) \bigr)-\bar{a}^n(0)\bigr|\,ds \biggr ] \\ & \leq&{\mathbb{e}}_x \biggl[\int_0^t \frac{\delta}{4 } v\bigl(\widehat { x}^n(s)\bigr)\,ds \biggr]\end{aligned}\ ] ] for all sufficiently large @xmath0 .
the last inequality follows noting that , by ( [ eq : alip ] ) , ( [ eq : quadraticvar_fluid ] ) and ( [ eq : dmtoctmc2 ] ) , there exists a finite positive constant @xmath188 such that @xmath339_{2,1,b_{\widehat{x}^n(s ) } ( { \bar{\ell}}/{\sqrt { n } } ) } \bigl|\bar { a}^n(0)\bigr|\leq c v\bigl ( \widehat{x}^n(s)\bigr)\ ] ] and @xmath339_{2,1,b_{\widehat{x}^n(s ) } ( { \bar{\ell}}/{\sqrt { n } } ) } \bigl|\bar { a}^n\bigl(\widehat{x}^n(s ) \bigr)-\bar{a}^n(0)\bigr|\leq\frac{ck_a}{\sqrt { n}}v\bigl(\widehat{x}^n(s ) \bigr),\ ] ] where @xmath127 is as in ( [ eq : alip ] ) .
similarly one argues , using ( [ eq : alip ] ) and ( [ eq : dmtoctmc2 ] ) , that for all sufficiently large @xmath0 , @xmath340 \\ & \leq&{\mathbb{e}}_x \biggl[\int_0^t \frac{\delta}{4 } v\bigl(\widehat { x}^n(s)\bigr)\,ds \biggr],\end{aligned}\ ] ] to conclude from assumption [ asum : l ] and lemma [ lem : ito ] that @xmath341\leq v(x)+{\mathbb{e}}_x \biggl[\int_0^t \biggl(-\frac{\delta } { 2}v\bigl(\widehat{x}^n(s)\bigr)+b \biggr)\,ds \biggr].\ ] ] in turn , ( [ eq : lyap_ctmc ] ) holds for all sufficiently large @xmath0 .
this guarantees that @xmath14 is ergodic for all such @xmath0 ; see , for example , @xcite , theorem 8.13
. using ( [ eq : lyap_ctmc ] ) and the nonnegativity of @xmath37 , we have for all sufficiently large @xmath0 and all @xmath336 that @xmath342\leq\frac{1}{t}2\delta ^{-1}\bigl(v(x)+b t \bigr).\ ] ] letting @xmath43 be the steady - state distribution of @xmath14 we have , for each @xmath279 , that @xmath343= \lim_{t\rightarrow\infty } \frac{1}{t}{\mathbb{e}}_x \biggl[\int _ 0^t
v\bigl(\widehat{x}^n(s)\bigr ) \wedge m \,ds \biggr]\leq2\delta^{-1}b.\ ] ] the result now follows from the nonnegativity of @xmath37 and the monotone convergence theorem .
lyapunov functions that satisfy assumption [ asum : l ] must be identified on a case - by - case basis .
for the first example the erlang - a queue this is a straightforward task . for the second example a queue with many servers and phase - type service time distribution this task is substantially more difficult , but recent work @xcite provides us with the required function .
we consider a sequence of queues with a single pool of i.i.d .
servers that serve one class of impatient i.i.d . customers .
arrivals follow a poisson process ( with rate @xmath0 in the @xmath0th queue ) , service times are exponentially distributed with rate @xmath183 and customers patience times are exponentially distributed with rate @xmath344 . in the @xmath0th queue , there are @xmath345 servers in the server pool .
let @xmath346 be the total number of jobs in the @xmath0th queue ( waiting or in service ) at time @xmath347 .
then @xmath348 is a birth and death process with state space @xmath349 , birth rate @xmath0 in all states and death rate @xmath350 in state @xmath55 where , for the remainder of the paper , we use @xmath351 , @xmath352 .
_ we assume that @xmath353 _ so that positive recurrence of @xmath23 follows easily .
the drift @xmath107 is then specified here by @xmath354 and is trivially extended here to the real line by allowing @xmath55 to take real values ( including negative values ) .
the fm is then given by @xmath355 there exists a unique point @xmath21 in which @xmath356 . at this point @xmath357 so that @xmath358 the dm for the erlang - a queue is subsequently given by @xmath359 where @xmath360 and @xmath361 .
it is easily verified that there exists @xmath362 such that @xmath363 when @xmath364 and @xmath365 if @xmath366 .
fixing @xmath367 and taking @xmath368 we have that @xmath369 for all @xmath370 and @xmath371 note that @xmath223 is trivially sub - exponential .
further , for all sufficiently large @xmath54 , @xmath372 so that the conditions of lemma [ lem : fmtodm ] are satisfied and , in turn , ul holds for the dm .
further , for each @xmath178 , @xmath373 where @xmath374 is the number of arrivals by time @xmath347 . condition ( [ eq : finite_integral ] ) then follows from basic properties of the poisson process .
we have the following consequence .
fix @xmath367 and positive @xmath221 .
then , @xmath375 satisfies assumption [ asum : l ] for the dm of the erlang - a queue .
fixing @xmath221 and choosing sufficiently large @xmath376 , we can take @xmath377 in corollary [ cor : tightness ] ; see remark [ rem : simplecase ] .
the following is now a direct consequence of theorem [ thmm : main ] and corollary [ cor : tightness ] .
[ thmm : erlanga ] consider a sequence of erlang - a queues as above and let @xmath9 be such that @xmath378 for some @xmath221 .
then @xmath379 above , we did not impose any restrictions on the way in which the number of servers , @xmath345 , scales with @xmath0 so that one may interpret our dm as a _ universal _
approximation for the erlang - a queue .
universality for this queue ( and its contrast with the assumption of a so - called operational regime ) are discussed at length in @xcite ; see also the references therein .
a similar result is proved there for the erlang - a queue using an approach that , while having important similarities to the approach we take here , is based on approximating the excursions of the process @xmath23 above and below @xmath380 . in this one - dimensional markov chain ,
the poisson equation we use here is ( informally ) a `` pasting '' of the dirichlet problems studied in @xcite . in their greatest generality ,
the results of @xcite are not a special case of theorem [ thmm : erlanga ] above . in @xcite
the authors allow the service rate @xmath183 to vary with @xmath0 .
this is facilitated by the excursion approach taken there but violates the assumptions required to apply our results , particulary , the uniform lipschitz continuity of @xmath137 .
moreover , the approach in @xcite seems to be easily extendable to the case with @xmath381 in which case the dm is not exponentially ergodic and assumption [ asum : l ] is not satisfied .
we next consider the single class @xmath382 queue .
this is a generalization of the erlang - a queue where the exponential service time is replaced by a phase - type service - time ; see @xcite for a detailed construction .
we repeat here only the essential details .
let @xmath383 be the number of service phases , and let @xmath384 be the average length of phase @xmath385 .
we assume that @xmath386 , corresponding to all customers commencing their service at phase 1 ( the diffusion limits in @xcite cover the general case where @xmath387 is an arbitrary probability vector ) . having completed phase @xmath311 a job transitions into phase @xmath66 with probability @xmath388 .
the triplet @xmath389 defines the phase - type service - time distribution .
let @xmath390 note that @xmath391 .
as before , the patience rate is @xmath353 .
we consider a sequence of such queues indexed by the _ arrival rate @xmath392_. let @xmath393 let @xmath394 be the number of customers in the first phase of their service and waiting in the queue at time @xmath347 . for @xmath395 ,
let @xmath396 be the number of customers in phase @xmath311 of service at time @xmath347 .
the process @xmath397 is then a ctmc . for simplicity of exposition
we assume here that @xmath398 is integer valued for each @xmath0 and that the number of servers @xmath345 satisfies @xmath399 .
this implies , trivially , that @xmath400 which corresponds to the so - called halfin
whitt many - server regime and allows us subsequently to build on the results of @xcite and @xcite that study diffusion limits in this regime .
the analysis below is easily extended to the case @xmath401 for some @xmath402 .
define @xmath403 and the scaled and centered process @xmath14 as in ( [ eq : hxdefin ] ) .
then , @xmath404 this is written , in matrix notation , as @xmath405 and , for @xmath406 , @xmath407 the functions @xmath137 and @xmath408 satisfy ( [ eq : flip ] ) and ( [ eq : alip ] ) . assumption
[ asum : base ] holds in this example as the chain is trivially nonexplosive and irreducible .
the positive recurrence follows immediately from the fact that @xmath353 .
the diffusion model is given by @xmath409 with @xmath137 as in ( [ eq : ph_drift_matrix ] ) and diffusion coefficient @xmath408 as in ( [ eq : ph_diff1])([eq : ph_diff2 ] ) .
note ( [ eq : ph_drift_matrix])([eq : ph_diff2 ] ) that @xmath137 and @xmath410 do not , in fact , depend here on @xmath0 .
the existence of a quadratic lyapunov function , @xmath37 , for @xmath36 then follows from @xcite , theorem 3this function is specified in equation ( 5.24 ) there .
( to extend this argument to the general case with @xmath411 , note that @xmath37 in @xcite is still a lyapunov function for each @xmath0 if we perturb @xmath137 by a constant and @xmath133 by a term that shrinks proportional to @xmath412 . ) with a careful choice of the smoothing function @xmath413 there , the function @xmath414 ( for any constant @xmath367 ) is also sub - exponential .
finally , ( [ eq : finite_integral ] ) is argued as in the erlang - a case using crude bounds on the poisson arrivals .
the function @xmath414 thus satisfies assumption [ asum : l ] .
it is easily verified that @xmath415 ) and satisfies ( [ eq:3diff1 ] ) so that , as in remark [ rem : simplecase ] , @xmath416 satisfies assumption [ asum : l ] with re - defined constants @xmath417 and @xmath174 . choosing sufficiently large @xmath376 guarantees that @xmath418 .
the following is then an immediate consequence of theorem [ thmm : main ] and corollary [ cor : tightness ] .
consider the sequence of phase - type queues as above , and let @xmath9 be such that @xmath419 for some @xmath221 .
then @xmath420
thus , as in remark [ rem : allinone ] , we have a lyapunov function that allows us to establish simultaneously the stability of the markov chain for each sufficiently large @xmath0 , the uniform integrability of moments and the approximation gap .
it is worth noting that the fact that @xmath421 was already established , by alternative means and for more general ( multiclass ) phase - type queues , in @xcite .
the main step in this proof is a uniform minorization condition for a time - discretized version of @xmath36 .
once this is established ( see lemma [ lem : minorization ] below ) , we build on @xcite to complete the argument .
the proofs of the lemmas that are stated in this section appear in the .
we first consider a linear transformation of @xmath36 .
specifically , let @xmath422 be the unique square root of the matrix @xmath133 ; see @xcite , theorem 7.2.6 .
in particular , @xmath423 .
the matrix @xmath422 is itself invertible and its inverse is the square root of the inverse of @xmath133 ; see @xcite , page 406 .
let @xmath424 and define @xmath425 then @xmath426 is a @xmath57-dimensional brownian motion with drift @xmath137 , that is , @xmath427 where @xmath428 .
we next consider the discrete - time analogues of both @xmath426 and @xmath36 .
let @xmath429 let @xmath430 and @xmath431 be the corresponding one - step transition functions . below @xmath432 is the family of borel sets in @xmath56 .
[ lem : minorization ] fixing @xmath433 , there exist a probability measure @xmath434 with @xmath435 and a constant @xmath436 ( both not depending on @xmath0 ) such that @xmath437 there consequently exists a constant @xmath438 ( not depending on @xmath0 ) such that @xmath439 the following translates the lyapunov property ul into the discrete time setting .
[ lem : fromdisctocont ] let @xmath37 be as in assumption [ asum : l ] . then there exist finite positive constants @xmath440 and @xmath441 ( not depending on @xmath0 ) such that for all @xmath138 and all @xmath442 , @xmath443\leq(1- \gamma)v(x)+\bar{b}\mathbh{1}_{\overline { b}_0(k)}(x).\label{eq : discretedrift}\ ] ] using the fact that @xmath444 as @xmath445 , ( [ eq : discretedrift ] ) implies that there exist finite positive constants @xmath174 , @xmath446 and @xmath279 such that @xmath447 \leq \cases { \lambda v(x),&\quad $ \mbox{if } x\notin \overline { b}_0(k),$ \vspace*{2pt}\cr m , & \quad$\mbox{if } x\in\overline{b}_0(k).$}\ ] ] the following is then a direct consequence of @xcite , theorem 1.1 .
assumptions ( a1)(a3 ) there hold by lemmas [ lem : minorization ] , [ lem : fromdisctocont ] and by ( [ eq : baxendalcond ] ) .
there exist constants @xmath448 and @xmath183 ( not depending on @xmath0 ) such that for each @xmath221 , @xmath449-\pi^n(f ) \bigr|\leq\mathbb{m}e^{-\mu m}.\ ] ] with these we are ready for the proof of theorem [ thmm : expo_ergo ] .
proof of theorem [ thmm : expo_ergo ] the proof of the theorem now follows as in @xcite , page 536 .
specifically , let @xmath450 @xmath451-\pi^n(f)\bigr| & = & \sup_{|f|\leq v}\bigl| { \mathbb{e}}_x\bigl[f\bigl({\widehat{y}}^n\bigl(\lfloor t\rfloor+s\bigr ) \bigr)\bigr]-\pi^n(f)\bigr| \\ & = & \sup_{|f|\leq v}\bigl|{\mathbb{p}}_{{\widehat{y}}^n}^s(x , dy ) \bigl ( { \mathbb{e}}_x\bigl[f\bigl({\widehat{y}}^n\bigl(\lfloor t\rfloor \bigr ) \bigr)\bigr]-\pi^n(f)\bigr)\bigr| \\ & \leq&\int_{y}{\mathbb{p}}_{{\widehat{y}}^n}^s(x , dy)\sup _ { |f|\leq v}\bigl|{\mathbb{e}}_y\bigl[f\bigl({\widehat{y}}^n\bigl ( \lfloor t\rfloor\bigr)\bigr)\bigr]-\pi^n ( f)\bigr| \\ & \leq & \mathbb{m}e^{-\mu\lfloor t\rfloor } { \mathbb{e}}_x\bigl[v\bigl ( { \widehat{y}}^n(s)\bigr)\bigr ] \\ & \leq & \mathbb{m } e^{\mu } e^{-\mu t } \bigl(v(x)+b\bigr),\end{aligned}\ ] ] where @xmath452 is the transition probability function of @xmath36 in @xmath453 time units .
in the last inequality we used ( [ eq : tbound ] ) and the fact that @xmath454 . finally , since @xmath164 , the theorem holds with the constants @xmath455 and @xmath183
diffusion models are useful in the approximation of markov chains . we proved that , under a uniform lyapunov condition , the steady - state of some multidimensional ctmcs can be approximated with impressive accuracy by the steady - state of a relatively tractable diffusion _ model_. the existence of a diffusion limit that satisfies the lyapunov requirement as is the case for the phase - type queue considered in section [ sec : ph]can facilitate the application of our results .
the distinction between the diffusion model and diffusion limit is , however , important . a central motivation behind this work is to bypass the need for diffusion limits with the objective of providing steady - state diffusion approximation whose precision does not depend on assumption with regards to limiting values of underlying parameters .
that is , we ultimately seek to provide `` limit - free '' ( or _ universal _ ) approximations . a uniform lyapunov condition , as we require in assumption [ asum : l ] , need not hold in general .
informally , one expects such a condition to hold if the scale parameter @xmath0 has limited effect on the drift of the process around the fms stationary point .
many - server queues with abandonment , as those we use to illustrate our results , seem to satisfy this characterizations : diffusion limits ( regardless of the parameter regime , determining how the number of servers @xmath345 scales with @xmath0 ) are generalizations of the ou process .
it remains to identify the broadest characterization of markov chains for which a uniform lyapunov condition can be expected to hold .
in addition , the following extensions seem important : _ state - space collapse_. a fundamental phenomenon in diffusion limits for multi - class queueing systems is that of state - space collapse ( ssc ) . with ssc , the diffusion limit `` lives '' on a state - space that is of lower dimension relative to the ctmc : some coordinates of the ctmc become , asymptotically , deterministic functions of others . for example ,
if one allows for arbitrary initial - phase vectors @xmath387 in the example of section [ sec : ph ] , the number of customers in queue with initial phase @xmath456 is asymptotically equal to @xmath457 ; see @xcite . to exploit state - space collapse within the diffusion - model framework used in this paper , one must develop bounds ( rather than convergence results ) for state - space collapse . _ single server queues and reflection .
_ a key challenge with single - server queueing systems is that of reflection . such reflection may violate our assumptions on @xmath137 .
consider , for example , the @xmath458 queue this is a single - server version of the erlang - a queue discussed in section [ sec : examples ] .
suppose that the arrival rate and service rate in the @xmath0th queue satisfy @xmath459 , and @xmath460 ( for @xmath461 ) .
let @xmath353 be the patience parameter .
then @xmath462 so that @xmath463 .
also , @xmath464 and , in particular @xmath465 as @xmath53 .
clearly , ( [ eq : flip ] ) is violated .
it is fair to conjecture that similar results as ours can be proved in such settings provided that the reflection is explicitly captured in the dm .
extending our results to dms with reflection seems to present a challenge insofar as the theory of pdes that arise from the poisson equation for such networks is less developed and poses a challenge in terms of the gradient bounds that are central to our analysis here ; see , for example , @xcite , where the poisson equation for constrained diffusion is discussed as well as , in the context of ergodic control , @xcite .
proof of lemma [ lem : ito ] fix @xmath318 . by it
s rule applied to the pure jump process @xmath466 we have that @xmath467 \\[-8pt ] \nonumber & & { } + \sum_{s\leq t } \biggl[\psi\bigl ( \widehat{x}^n(s)\bigr)-\psi
\bigl(\widehat{x}^n ( s- ) \bigr)-\sum_{i=1}^d \psi_i \bigl(\widehat{x}^n(s-)\bigr)\delta\widehat{x}^n_i(s ) \biggr ] .
\label{eq : itojump}\end{aligned}\ ] ] from the linear growth of @xmath137 and from ( [ eq : integrability_cond ] ) , it then follows that @xmath468<\infty.\ ] ] we can then apply lvy s formula for ctmcs ( see , e.g. , @xcite , exercise i.2.e2 ) to get that @xmath469 is a martingale with respect to the filtration in ( [ eq : filtration ] ) and , in turn , for all @xmath178 , @xmath470={\mathbb{e}}_x \biggl[\sum_{i=1}^d \int_0^t \psi_i\bigl(\widehat { x}^n(s)\bigr){\widehat{f}}_i^n\bigl ( \widehat{x}^n(s)\bigr)\,ds \biggr].\ ] ] to treat the second line of ( [ eq : itojump ] ) , we decompose it into @xmath471 and @xmath472 \\[-8pt ] \nonumber & & \hspace*{60pt}\qquad { } -\frac{1}{2}\sum_{i , j}^d \psi_{ij}\bigl(\widehat{x}^n ( s-)\bigr)\delta \widehat{x}^n_i(s)\delta\widehat{x}^n_j(s ) \biggr].\end{aligned}\ ] ] we treat ( [ eqd ] ) first . by ( [ eq : alip ] ) ,
@xmath473 so that , by ( [ eq : integrability_cond ] ) , @xmath474 < \infty , \qquad t \geq0 , x\in\widehat{e}^n,\ ] ] and applying lvy s formula once again , we obtain @xmath475 \\ & & \qquad = \frac{1}{2}{\mathbb{e}}_x \biggl[\sum _ { i , j}^d\sum_{\ell } \int _ 0^t \psi_{ij}\bigl ( \widehat{x}^n(s)\bigr)\ell_i\ell_j \frac{1}{n}\beta _ { \ell } ^n\bigl(\sqrt { n } \widehat{x}^n(s)+\bar{x}^n_{\infty}\bigr)\,ds \biggr ] \\ & & \qquad = \frac{1}{2 } { \mathbb{e}}_x \biggl[\sum _ { i , j}^d \int_0^t \psi_{ij}\bigl(\widehat{x}^n(s)\bigr)\bar{a}_{ij}^n \bigl(\widehat { x}^n(s)\bigr)\,ds \biggr ] \\ & & \qquad = \frac{1}{2}{\mathbb{e}}_x \biggl[\sum _ { i , j}^d \int_0^t \psi_{ij}\bigl(\widehat{x}^n ( s)\bigr)\bar { a}_{ij}^n(0)\,ds \biggr ] \\ & & \qquad\quad { } + \frac{1}{2}{\mathbb{e}}_x \biggl[\sum _ { i ,
j}^d \int_0^t \psi_{ij}\bigl(\widehat{x}^n(s)\bigr ) \bigl ( \bar{a}^n\bigl(\widehat { x}^n(s)\bigr)-\bar { a}_{ij}^n(0)\bigr)\,ds \biggr].\end{aligned}\ ] ] the second item in the last line is @xmath476 in the statement of the lemma .
we have proven thus far that @xmath477 \\ & & \qquad = \psi(x)+{\mathbb{e}}_x \biggl[\sum_{i=1}^d \int_0^t \psi_i\bigl(\widehat { x}^n(s)\bigr){\widehat{f}}_i^n\bigl ( \widehat{x}^n ( s)\bigr)\,ds \biggr ] \\ & & \quad\qquad{}+ \frac{1}{2 } { \mathbb{e}}_x \biggl[\sum_{i , j}^d \int _ 0^t \psi_{ij}\bigl ( \widehat{x}^n ( s)\bigr)\bar { a}_{ij}^n(0)\,ds \biggr ] + d_{\psi}^{n , x}(t)+a_{\psi } ^{n , x}(t ) \\ & & \qquad = \psi(x)+{\mathbb{e}}_x \biggl[\int_0^t \mathcal{a}^n\psi\bigl(\widehat{x}^n ( s)\bigr)\,ds \biggr]+d_{\psi}^{n , x}(t)+a_{\psi}^{n , x}(t),\end{aligned}\ ] ] where @xmath478 is as in the statement of the lemma and @xmath479 $ ] ( we will prove below that this expectation is well defined ) . to bound @xmath480 note that , by taylor s theorem , @xmath481 where @xmath482 $ ] .
thus @xmath483 here note that @xmath484 .
let @xmath485 .
note that @xmath486_{2,1,b_{x}(\bar{\ell}/\sqrt{n})}$ ] for @xmath487 with @xmath488 . since @xmath489 , we have that @xmath490 \\ & & \qquad\leq \frac{\bar{\ell}}{\sqrt{n}}\frac{1}{2 n}{\mathbb{e}}_x \biggl[\sum _ { i , j}^d \int_0^t [ \psi]_{2,1,b_{\widehat{x}^n(s)}(\bar{\ell}/\sqrt{n})}\sum_{\ell } |\ell _ i||\ell _
j|\beta_{\ell}^n \bigl(x^n(s)\bigr)\,ds \biggr ] \\ & & \qquad \leq \frac{\bar{\ell}}{\sqrt{n}}\frac{1}{2 n}{\mathbb{e}}_x \biggl[\int _ 0^t [ \psi ] _ { 2,1,b_{\widehat{x}^n(s)}(\bar{\ell}/\sqrt { n})}\bigl|a^n \bigl(x^n(s)\bigr)\bigr|\,ds \biggr ] \\ & & \qquad = \frac{\bar{\ell}}{2\sqrt{n}}{\mathbb{e}}_x \biggl[\int_0^t [ \psi ] _ { 2,1,b_{\widehat{x}^n(s)}(\bar{\ell}/\sqrt{n})}\bigl|\bar { a}^n\bigl(\widehat{x}^n ( s ) \bigr)\bigr|\,ds \biggr ] < \infty,\end{aligned}\ ] ] where the finiteness follows from ( [ eq : alip ] ) and condition ( [ eq : integrability_cond ] )
. we can apply lvy s formula one final time to conclude that @xmath491\bigr|&= & \biggl{\vert}\frac{1}{2 n}\sum _ { i , j}^d { \mathbb{e}}_x \biggl [ \int _
0^t \sum_{\ell } \widetilde{\psi}_{ij}\bigl(\widehat{x}^n(s),\widehat { x}^n(s)+\ell/\sqrt { n}\bigr)\ell _ i\ell_j \beta_{\ell}^n\bigl(x^n(s)\bigr)\,ds\biggr]\biggr { \vert}\\ & \leq&\frac{\bar{\ell}}{2\sqrt{n}}{\mathbb{e}}_x \biggl[\int_0^t [ \psi ] _ { 2,1,b_{\widehat{x}^n(s)}(\bar{\ell}/\sqrt{n})}\bigl|\bar { a}^n\bigl(\widehat{x}^n ( s ) \bigr)\bigr|\,ds \biggr]\end{aligned}\ ] ] as required . toward the proof of lemma [ lem : minorization ]
we first prove that @xmath492 inherits the lipschitz continuity of @xmath137 .
[ lem : psd ] there exists a finite positive constant @xmath174 ( not depending on @xmath0 ) such that @xmath493 since , for each @xmath0 , @xmath133 is symmetric positive definite as is @xmath135 , these matrices have strictly positive eigenvalues ; see , for example , @xcite , theorem 7.2.1 . also , the eigenvalues of the square - root matrix @xmath422 are the square roots of the eigenvalues of @xmath133 . since @xmath291 , the eigenvalues of @xmath422 , @xmath494 , converge to those of @xmath495 , @xmath496 .
the eigenvalues of the inverses @xmath497 and @xmath498 are given by the reciprocals and , in turn , satisfy @xmath499 .
in particular @xmath500 and @xmath501 ( where , following common notation , @xmath502 is the spectral norm of @xmath503 ; see @xcite , section 5.1 . since the matrices are symmetric this norm is equal to the spectral radius of the matrix , that is , to its maximal eigenvalue ) . by definition of the matrix norm
it then holds that @xmath504 for some finite positive constant @xmath505 where the last inequality follows from the fact @xmath506 argued above .
similarly , @xmath507 for a finite positive constant @xmath508 . finally , using ( [ eq : flip ] ) we have that @xmath509 which completes the proof .
proof of lemma [ lem : minorization ] we consider first the chain @xmath510 .
fix @xmath174 and let @xmath511 .
let @xmath512 .
by ( [ eq : interim21 ] ) , there exists a constant @xmath513 not depending on @xmath0 such that @xmath514 by lemma [ lem : psd ] there exist @xmath515 and @xmath227 not depending on @xmath0 such that @xmath516 for all @xmath179 and @xmath517 . also , since @xmath518 it satisfies also a linear growth condition uniformly in @xmath0 . using @xcite , theorem 3.1 and ( [ eq : bark1 ] ) we have that @xmath519 for some @xmath520 where @xmath521 is the transition density of @xmath426 from @xmath55 to @xmath307 in time @xmath347 .
in particular , @xmath522 where @xmath298 is here the lebesgue measure and @xmath523 using the invariance of lebesgue measure under invertible linear transformations we have for any @xmath524 that @xmath525 where @xmath526 is here the determinant of the positive definite matrix @xmath497 , and we use the simple fact that @xmath527 .
since @xmath528 , it also holds that @xmath529 so that there exists @xmath530 ( not depending on @xmath0 ) such that @xmath531 let @xmath532 . defining the measure
@xmath533 we conclude that @xmath534 the result for @xmath535 follows immediately from the above .
indeed , @xmath536 which completes the proof
. proof of lemma [ lem : fromdisctocont ] this argument is almost identical to the proof in @xcite , page 27 . under condition ( [ eq :
expobound1 ] ) , dynkin s formula holds up to @xmath347 , that is , @xmath537= v(y)+ { \mathbb{e}}_y \biggl[\int_0^t \mathcal { a}^n v\bigl({\widehat{y}}^n(s)\bigr)\,ds \biggr];\ ] ] see , for example , @xcite , theorem 6.3 . setting @xmath538\quad\mbox{and}\quad h(t)={\mathbb{e}}_y\bigl[\mathcal{a}^nv\bigl({\widehat{y}}^n(t)\bigr)\bigr]+\delta g(t),\ ] ] we have that @xmath539 ( @xmath212 and @xmath227 as in assumption [ asum : l ] ) and @xmath540 solving this differential equation we get @xmath541 setting @xmath542 and @xmath543 we have the statement of the lemma .
the author is grateful to junfei huang and to an anonymous referee for their careful reading of this paper and for their numerous insightful comments . | motivated by queues with many servers , we study brownian steady - state approximations for continuous time markov chains ( ctmcs ) .
our approximations are based on _ diffusion models _ ( rather than a diffusion limit ) whose steady - state , we prove , approximates that of the markov chain with notable precision .
strong approximations provide such `` limitless '' approximations for process dynamics .
our focus here is on steady - state distributions , and the diffusion model that we propose is tractable relative to strong approximations . within an asymptotic framework , in which a scale parameter @xmath0 is taken large , a uniform ( in the scale parameter ) lyapunov condition imposed on the sequence of diffusion models guarantees that the gap between the steady - state moments of the diffusion and those of the properly centered and scaled ctmcs shrinks at a rate of @xmath1 .
our proofs build on gradient estimates for solutions of the poisson equations associated with the ( sequence of ) diffusion models and on elementary martingale arguments . as a by - product of our analysis
, we explore connections between lyapunov functions for the fluid model , the diffusion model and the ctmc . |
lr date & 1 september 2006 + number of unique fields & 2950 + number of sources with @xmath0 & 947638 + number of uv sources with @xmath0 & 508415 + number of sources in om unique filters with @xmath0 & 364741 + typical positional uncertainty & @xmath1 + the optical monitor ( om ) catalogue ( omcat ) contains entries for every point - like source detected in imaging or fast mode om data .
the omcat was constructed from a complete reprocessing of the om data using the standard _
omichain_/_omfchain _ pipelines in sas 6.5.0 . for each observation ( obsid )
the reprocessing created a source list ; the omcat is a concatenation of these source lists .
thus , if the same region of sky was observed by multiple obsids , then some sources will be listed multiple times .
each listing should have the same coordinates ( to the limit of the astrometric accuracy ) and thus it should be reasonably obvious which listings refer to the same source .
we have opted to retain multiple listings ( rather than to combine them into a `` mean '' entry ) to retain any useful information of temporal variability in an easily accessible manner .
an overview of the catalogue statistics is given in table [ tab : over ] . in this paper
2 provides a brief description of the om and its primary observation modes , 3 describes the standard pipeline processing , the further processing done to produce the source lists , and the output products that are unique to our processing .
4 contains some useful statistics describing the catalogue . in 5 we demonstrate the extent to which the om filter set allows photometric classification of point - like sources , and in 6 we explore some of the scientific uses of the om - specific bands .
* the telescope and detector : * the om @xcite is a 30 cm f/12.7 ritchey chretien telescope coaligned with the x - ray telescopes and operating simultaneously with them . the detector is a micro - channel plate intensified charge - coupled device ( ccd ) .
photons striking a photocathode produce electrons that are amplified by two successive micro - channel plates .
the electron clouds then strike a phosphor , and the resulting photon splashes are recorded by a ccd ; the location of the photon splash is centroided on board .
the centroids are stored in units of 1/8 of a ccd pixel .
since it is these photon splashes that are recorded by the ccd , rather than individual photons , the ccd is read out very rapidly ( every 11@xmath2s ) , and the centroids of the photon splashes determined and stored .
thus , the ccd is used more like a photon - counting device than an accumulator , although it is an image , rather than an event list , that is produced .
the photocathode is optimized for the blue and ultraviolet .
the `` native '' pixel size is @xmath3 and the point spread function ( psf ) fwhm is @xmath4-@xmath5 depending upon filter .
the largest possible field of view ( fov ) is roughly @xmath6 .
lrrrrr v & 5407 & 5230 & 684 & 1.35 & 19.0 + b & 4334 & 3980 & 976 & 1.39 & 19.7 + u & 3472 & 3270 & 810 & 1.55 & 19.5 + uvw1 & 2905 & 2680 & 620 & 2.0 & 19.3 + uvm2 & 2298 & 2210 & 439 & 1.8 & 18.3 + uvw2 & 2070 & 2000 & 500 & 1.98 & 17.6 + white & & & & & 22.2 + the om has a smaller fov than _
galex_(a 1.2@xmath7 circle ) but better angular resolution ( _ galex _ has a @xmath8 fwhm psf in its fuv filter ( 1350 - 1750 , and a @xmath9 fwhm psf in its nuv filter ( 1750 - 2800 ) * ? ? ? * ) .
the effective areas of the om and _ galex _ filters are shown in figure [ fig : filts ] while the om filter particulars are given in table [ tab : filters ] .
thus , the omcat data in the uvw2 and uvm2 filters provide an excellent higher resolution complement to the _ galex _ nuv data , while uvw1 data is somewhat redder than the _ galex _ band .
the _ swift _ uvot is , essentially , an improved om , with similar filters , so comparison of data in this catalogue with uvot data should be straightforward . * the observation modes : * due to the onboard centroiding , memory limitations , and telemetry limitations , setting the om observation mode has to be a balance of temporal resolution and spatial coverage ; the higher the temporal resolution the lower the spatial coverage . as a result
, the om allows a large number of observing modes that place different emphases on temporal and spatial optimization .
these modes define different `` science windows '' covering only portions of the entire fov ; events falling outside of those windows are discarded .
there are two primary observation modes at the extremes : the default `` imaging '' mode and the default `` fast '' mode .
the default imaging mode consists of five consecutive sub - exposures , each of which employs two science windows ; one high - resolution window and one low - resolution window .
note that resolution , in this case , refers to the degree to which the image is sampled , not to an intrinsic change in the psf size .
the fov of the default imaging mode is shown in the left - hand panel of figure [ fig : fov ] . the high - resolution window ( roughly @xmath10 ,
marked `` hr '' in the figure ) is always located at the center of the fov .
the five low - resolution windows ( marked `` sw '' in the figure ) cumulatively cover the entire fov ( roughly @xmath6 ) with a center square surrounded by five rectangular regions . for any number of reasons ,
not all of the sub - exposures of a default image may actually be taken , but , for the default imaging mode , there will always be a high - resolution sub - exposure for each low - resolution sub - exposure .
it should be noted that multiple observing modes may be used during the course of a single observation , so not all exposures of a single observation need cover the same region .
the use of five different science windows to cover the fov , with some overlap between the windows , means that the exposure is not uniform across the fov .
there are two other common full - field low - resolution modes , `` eng-2 '' and `` eng-4 '' , which are also included in our processing .
the fov for these modes are shown in the right - hand panel of figure [ fig : fov ] ; the two modes cover the same area but with different binning , the `` eng-2 '' mode having twice the pixel size as the `` eng-4 '' mode .
there is no repeated high - resolution window for these modes .
the default fast mode uses the same windows as the default imaging mode with the addition of a third science window ( roughly @xmath11 ) at an observer - defined location ( typically the center of the fov ) .
* the default modes : * if the observer did not specify an om mode , and there was no bright source in the fov , the om took exposures in the default imaging mode . for the first two years of the mission the default filters were b , uvw2 , u , and uvw1 , in order of priority .
the filter priority was then changed to uvm2 , uvw1 , and u , in order to optimize the use of the unique capabilities of the om . *
magnitude system : * the standard om processing produces _ instrumental _ magnitudes .
we have opted to continue to work in the instrumental magnitude system .
the definition of the instrumental magnitudes and the current conversion to ab magnitudes is given in the _ xmm - newton _ user s handbook in rough terms , the v and b the instrumental magnitudes are similar to ab magnitudes , but the uv the instrumental magnitudes are typically smaller than the ab magnitudes by a magnitude to a magnitude and a half .
* image mode processing : * for the most part , we have used the standard _
omichain _ processing with the default settings ; exceptions are detailed below .
the standard _
omichain _ processing ( processing for images ) handles the images produced by each science window separately . for each science window image
_ omichain _ applies a flatfield .
the photon splash centroiding algorithm calculates the centroid to 1/8 of a pixel , but due to the algorithm , not all values are equally likely .
this problem results in `` modulo-8 '' fixed pattern noise .
the _ omichain _
processing applies a redistribution to correct for this effect . for every science window image _ omichain _ runs a source detection algorithm , measures the count rates for the sources , and applies a calibration to convert to instrumental magnitudes .
once all of the science windows are processed , _
omichain _ produces a `` master '' source list by combining the source list for each science window , matching sources in common between the lists , and determining the mean @xmath12 for each source .
the standard processing has the option to use an external catalogue to correct the coordinates of the master source list ; we have used this option with the usno - b1 catalogue @xcite .
it should be noted that the coordinate correction using the usno - b1 catalogue will fail if there are too few matching sources , in which case no significant solution can be found .
the _ omichain _
algorithm requires at least ten matches in order to produce a significant coordinate correction .
if there are too many sources in the field , coordinate correction will also fail , presumably because some fraction of matches are spurious and the solution will not converge .
we have found that the coordinate correction _ can _ fail for almost any density of sources , though we did not determine the cause of that failure .
in addition to combining the source lists for all the filters , _
omichain _ mosaics the low - resolution science window images ( but not the high - resolution science window images ) for each filter .
the world coordinate system ( wcs ) of the first science window for a given filter sets the coordinate system of the entire image mosaic .
note that the standard _
omichain _ processing can correct the master source list , but not the images .
further , the correction using an external catalogue will be applied only if there are at least ten sources .
one does have the option of doing the same correction to the individual science windows , but there are often not enough sources in a single science window to perform such a correction .
we have applied the correction derived from the external catalogue to the low - resolution mosaics .
since the mosaic images use the wcs from the first science window in the mosaic , we compared the @xmath12 in the source list for the first science window to the @xmath12 in the usno corrected master list .
we then determined the @xmath13 which must be added to the @xmath12 of the first source list in order to obtain the @xmath12 in the usno corrected master list .
we then applied that correction to the header keywords of the mosaicked image .
it must be noted that since the telescope can drift by 1 - 2 arcseconds between exposures , this correction is done separately for each filter .
the standard _
omichain _ processing does not combine the images from the high - resolution science windows .
however , we have done so , though not with _
ommosaic _ , the standard sas tool .
the _ ommosaic _
program uses the wcs keywords to determine the offsets needed to align the wcs frames of the individual science windows before summing .
we allow the wcs of the summed image to be set by the first science window . for each successive high resolution image
, we compare the source list to the source list from the first image , determine the @xmath13 required to match the source lists , and apply that offset to the image before adding it to the mosaic .
not all sources are used for determining the offsets , only sources appearing in at least half of the images ; this selection removes sources with poorly determined positions .
the offsets are rounded to the nearest integer pixel ; subpixelization did not seem to produce a significant improvement in the resultant psf , and so was not used for this processing . compared to the direct sum of the images made by _
, our processing does improve the psf of the summed image , sometimes improving the fwhm by as much as a pixel ( see figure [ fig : hires_demo ] ) .
we compare the source list from the first image with the master source list to correct the summed image in the same manner used for the low - resolution mosaics .
* further coordinate correction : * after our initial processing of the public archive we found that the pipeline coordinate correction ( done by _ omichain _ ) using the usno catalogue failed for @xmath14% of the fields .
further , the failure was not limited to extremely high or extremely low source densities ( see figure [ fig : bad_corr ] ) .
we thus found it worthwhile to create our own coordinate correction routine ( the `` post - pipeline '' correction ) using the usno catalogue .
although the bulk of fields need only a small correction , some fields need substantial corrections ( @xmath15 ) .
thus , although we attempt to find high precision corrections for all fields , it is still worthwhile to find lower accuracy corrections for those fields that do not have a large enough number of matches with the usno catalogue to attempt a high precision solution .
we used a fairly simple and robust algorithm for matching the om sources to the usno sources . if there were @xmath16 matches we iteratively solved for the offset in @xmath12 that minimized the offset between the om source list with the usno source list . by iterating the solution we could eliminate some portion of the false matches .
we have not solved for a rotation for two reasons : 1 ) adding a rotation to the fit did not significantly improve the solution , and 2 ) given that there are systematic offsets from one science window to another , the rotation could be strongly biased by the offset of a single science window .
if there were @xmath17 matches we merely calculated the mean offset between the om and usno sources , and used that offset as the coordinate correction .
if there were @xmath18 sources we did not attempt a correction .
we applied the same correction to the individual images that we applied to the source lists . for each om source in the source list with the _ omichain_-calculated maximum - liklihood significance in any filter @xmath19
, the matching algorithm finds the closest usno source .
it then creates the distribution of the distances between the om sources and their closest usno counterparts .
if all of the om sources had usno counterparts , then this distribution would be a gaussian whose width is the coordinate uncertainties of the two catalogues and whose peak is the offset between the two catalogues .
if there were no true matches between the om sources and the usno catalogue , then the distribution would be given roughly by the probability distribution for the minimum distance between a given point and a uniform distribution of sources : @xmath20 } \rho\pi[(r+\delta r)^2-r^2 ] e^{\left[-\rho\pi [ ( r+\delta r)^2-r^2]\right]}\ ] ] where @xmath21 is the surface density of sources and @xmath22 is the binsize of one s histogram of distances . for this distribution both the peak of the distribution and the width of the distribution scale as @xmath23 .
we expect that some fraction of the om sources have true usno matches and that the remainder will not . as a result
, the observed distribution of sources has a sharp peak with a width of @xmath24 due to matches and a low , broad distribution for the spurious matches .
this algorithm has problems with high density regions ; for source densities of 5000 sources / image ( 0.005 sources arcsec@xmath25 ) the distribution of spurious matches peaks at @xmath26 with the lower half - maximum at @xmath27 .
although the peak of the matching sources typically has @xmath28 , the true match rate is likely to be small compared to the spurious match rate , and so it is difficult , if not impossible , to find the true match peak in this distribution .
however , at these source densities , source confusion is a serious problem as well , so even if the matching algorithm worked , the coordinate solution would remain problematic . for the matching algorithm , we simply fit the distribution with a gaussian .
if the width of the gaussian is smaller than @xmath29 , then the algorithm takes all of the sources within @xmath30 of the peak of the distribution as real matches .
an initial solution is determined from those matches , and a fit is made in the image coordinate frame to find the @xmath12 offset that minimizes the distance between the om sources and their usno matches .
the source matching is redone with the new offset , and the process is iterated until it converges . after application of our coordinate correction routines ,
only @xmath31% of the fields remained without any coordinate correction .
besides fields with very few objects , the fields without coordinate corrections were characterized by very broad distribution of the matches suggesting a combination of large pointing error and large source density , and thus a large number of spurious identifications . for fields where the pipeline processing found a good coordinate solution using the usno catalogue our coordinate correction was not significantly different .
however , our coordinate correction did improve the mean distance between om and usno sources for @xmath32% of fields , and provided the only coordinate corrections for @xmath33% of fields .
( figure [ fig : bad_corr ] ) the rms residual between the om sources and the matching usno sources was calculated for every field .
a value of zero indicates that there was no coordinate solution .
coordinate solutions with rms residuals @xmath34 should be considered to be poor .
* fast mode processing : * the bulk of the fast mode processing is concerned with the production of light - curves of the source .
the fast mode images consist of @xmath11 regions containing , typically , a single source .
we combine all of the images for each filter using the same method applied to the high - resolution images .
* further processing : * further processing is required to provide a more useful source list to be incorporated into the omcat . to the standard image catalogue ( a binary fits table ) we add images of each source from each filter .
each `` postage stamp '' image is @xmath35 pixels in size , extracted from the low - resolution image mosaics ( the pixel size is @xmath36 and the image is @xmath37 in size ) .
since the sources were derived from all of the science windows , some sources can fall in high - resolution science windows without low - resolution counterparts . in that case
the postage stamp is extracted from the high - resolution image and binned to the same resolution and size as the other postage stamps .
sources that appear only in `` fast '' science windows are treated similarly .
note that postage stamps are extracted from all of the available filters , not just the filters for which the source was detected ; many postage stamps may thus appear to be empty .
* processing summary : * for each obsid our processing produces a coordinate corrected source list , a coordinate corrected low - resolution mosaicked image for each filter , a coordinate corrected high - resolution mosaicked image for each filter ( if possible ) , or a summed fast mode image .
* caveats : * 1 ) individual science windows may be significantly offset ( 1 - 2@xmath38 ) from the remainder of the mosaic . in this case
the correction by use of the usno - b1 catalogue will not be wholly satisfactory , and sources will appear to be offset in the postage stamps . the extent of this problem for any individual source can determined by looking at the `` rms_resid '' column which contains the rms residual from the fit of source list to the usno - b1 catalogue .
\2 ) the source lists will contain spurious sources , sources due to ghost images , diffraction spikes , readout streaks , saturation around bright sources , and other effects .
some of these sources can be removed by consulting the data quality flag ( q_flag parameter ) similarly , confused sources are flagged by the c_flag parameter .
however , we have found that filtering out sources with signal - to - noise ratios of less than three was a more efficient means of removing spurious sources than reference to the quality flags .
figure [ fig : demo_imag ] and figure [ fig : demo_source ] demonstrate the extent to which the data quality , confusion , and extension flags can be used .
figure [ fig : demo_imag ] shows an image with typical difficulties : ghosts , readout streaks , and diffraction spikes .
figure [ fig : demo_source ] shows images of some of the individual sources from the image in figure [ fig : demo_imag ] .
the first row of images contains uvw1 band images of the first 36 sources in the source list which have q_flag@xmath39 , or good quality sources .
the second row of images contains uvw1 band images of the first 36 sources in the source list which have q_flag@xmath40 , or poor quality sources .
it is clear that some of the `` good quality sources '' are unreliable , while some of the `` poor quality sources '' are reliable .
the third and fourth rows of images contain , respectively , sources with c_flag@xmath39 ( unconfused ) and c_flag@xmath40 ( confused ) ; again the sources are taken in the order of the source list ( which happens to be from low r.a . to high r.a . ) and all images are in the uvw1 band . as can be seen , the confusion flag is not reliable .
the fifth and sixth rows of images contain , respectively , sources with e_flag@xmath39 ( non - extended ) and e_flag@xmath40 ( extended ) ; this flag does seem to be robust .
figure [ fig : demo_signif ] shows images of some of the individual sources from the image in figure [ fig : demo_imag ] , selected to demonstrate the effect of the signif parameter .
the top row of images shows the effect of increasing significance , as characterized by the signif parameter in the uvw1 band . for this column of images
, we sorted the sources by signif , and sampled the sequence uniformly to represent the entire range of the signif parameter .
the value of the signif parameter is given directly below the images , and the source signal - to - noise ratio directly below that .
similarly , the bottom row of images samples sources of increasing value of the source signal - to - noise ratio ; the values of the signif parameter and the signal - to - noise ratio are shown below the images .
neither the signif parameter nor the signal - to - noise ratio provides a completely reliable estimator of the reality of a particular source .
\3 ) although we provide the mosaicked low - resolution and mosaicked high - resolution images through the archive , these images , according to the sas documentation , should not be used for photometry .
there are a number of corrections in the photometric reduction which can not be made from the mosaicked images .
however , since coincidence - loss and dead - time corrections are small for faint sources and faint extended emission , photometry of faint sources using standard non - sas tools is possible from the mosaics , as has been demonstrated by d. hammer ( private communication ) .
\4 ) although there may be substantial exposure for a given mosaic , the detection limit in the current catalogue is not that of the mosaic since the source detection is done on the individual science windows rather than on the mosaicked images .
since the individual science windows have a median exposure of @xmath41 s , and the bulk of the individual science windows have exposures of 1000 s , the detection limit of the omcat is more uniform but somewhat lower than one would expect from the mosaic exposure times . *
availability : * the bulk of the data in the omcat can be accessed either through the browse facility at the high energy astrophysics science archive ( heasarc ) or through the multimission archive at stsci ( mast ) .
slightly more information for each source ( shape , confusion flags , and quality flags , as a function of filter ) are retained in the source lists for individual obsids , and can be downloaded from the heasarc with the rest of the om data .
_ observation statistics : _ of the 4373 observations that were public by 1 september 2006 , 2950 observations had om imaging mode data and 202 had om fast mode data .
about 25% of the fields imaged were observed more than once , allowing some measure of temporal variability .
however , the number of fields with multiple observations is a very strongly declining function of the number of repetitions ( see figure [ fig : dup_stat ] ) .
as can be seen in figure [ fig : sky_map ] , the galactic plane has a high density of observations ( particularly towards the galactic center ) , the region with @xmath42 has a lower density of observations , and the region with @xmath43 is relatively uniform , though coma and the magellanic clouds have visible concentrations of observations .
the imaging mode observations cover a cumulative @xmath44% of the sky .
figure [ fig : expo_stat ] shows the distribution of total exposure time per field and the exposure time per science window for the most commonly used filter ( uvw1 ) .
most science windows have exposures of a kilosecond , while the total exposure per field is significantly higher .
thus , since the point source detection is done in the individual science windows , the omcat could be made significantly deeper were point source detection to be executed on mosaicked images .
as noted above , since the bulk of individual science window exposures are @xmath45 seconds , the detection depth of the catalogue is relatively uniform .
however , given the multiple coverage of sources by individual science windows , the actual measurment accuracy is significantly better than that obtained from a single 1000 second exposure . figure [ fig : stat_mags ] shows the distribution of magnitudes for each filter .
_ coordinate statistics : _ figure [ fig : stat_resid ] shows the distribution of the residuals between the corrected source catalogues and the usno - b1 catalogue for those fields for which a coordinate correction was successful .
the residuals before the post - pipeline correction are peaked around @xmath46 with a secondary peak around @xmath29 .
after correction , the distribution is more symmetrically distributed around @xmath46 . it should be noted that a portion of the residual for any given image can be due to the offset of individual science windows within the mosaic . the distribution of the calculated offsets between coordinates before and after the post - pipeline correction by the usno catalogue is shown in figure [ fig : stat_offs ] .
the distribution is a broad skewed gaussian peaking at @xmath47 with a significant tail extending to @xmath48 .
note that this distribution does not reflect the pointing ability of the telescope , but rather the performance of the tracking corrections in the om pipeline .
_ source statistics : _ the omcat contains roughly 3.7@xmath49 entries , of which 82% have detection significance ( a maximum likelihood measure calculated within _ omichain _ )
greater than three in at least one band and 72% have a significance greater than three in at least one band .
roughly 71% of all entries have had successful coordinate corrections .
only @xmath50 of sources are classified as extended ; the rest are considered point - like .
approximately 60% of the sources are flagged for data quality in at least one band , and approximately 74% of the sources are flagged for confusion in at least one band . however ,
if one looks only at individual measurements for which the signal - to - noise ratio is @xmath19 , 49% are flagged for data quality , 3% are flagged for confusion , and 4% are flagged as extended . given the discussion in the previous section , it is clear that the data quality flag in the current processing is a very poor measure of the true data quality , so these statistics , though they give one pause , are not cause for alarm .
lccccccc v & 1028785 & 233685 & 207608 & 165249 & 79448 & 56517 & 1867 + b & 233685 & 548128 & 225180 & 133163 & 60190 & 66840 & 4600 + u & 207608 & 225180 & 651438 & 247353 & 84993 & 62469 & 3546 + uvw1 & 165249 & 133163 & 247353 & 768349 & 144107 & 76204 & 2807 + uvm2 & 79448 & 60190 & 84993 & 144107 & 256142 & 79766 & 950 + uvw2 & 56517 & 66840 & 62469 & 76204 & 79766 & 191067 & 359 + white & 1867 & 4600 & 3546 & 2807 & 950 & 359 & 11095 + v & 8536 & 2585 & 2732 & 2720 & 644 & 491 & 56 + b & 2585 & 9694 & 4978 & 3964 & 803 & 685 & 403 + u & 2732 & 4978 & 15278 & 7016 & 1451 & 864 & 375 + uvw1 & 2720 & 3964 & 7016 & 25477 & 2829 & 1852 & 334 + uvm2 & 644 & 803 & 1451 & 2829 & 6459 & 1467 & 143 + uvw2 & 491 & 685 & 864 & 1852 & 1467 & 7521 & 70 + white & 56 & 403 & 375 & 334 & 70 & 143 & 812 + due to the interests of observers and the changing default filter priorities , the distribution of exposures among the various filters is uneven .
when measured in terms of the number of sources with @xmath30 detections in each filter , the uvw1@xmath51u , b@xmath51v , and u@xmath51bcolors have the best statistics .
table [ tab : color_stats ] shows the number of sources with @xmath30 detections for each filter and filter combination .
as one might expect , the distribution for x - ray selected sources ( that is , x - ray point sources in the field with omcat counterparts rather than just sources observed because of their x - ray properties ) is somewhat different , with uvw1@xmath51u , u@xmath51b , uvw1@xmath51b , and uvw2@xmath51uvw1 having the best statistics .
lccccccc 14 & 0.96 & 0.81 & 0.96 & 0.94 & 0.89 & 0.93 & 1.00 + 15 & 0.96 & 0.95 & 0.94 & 0.93 & 0.93 & 0.90 & 0.87 + 16 & 0.92 & 0.95 & 0.92 & 0.89 & 0.85 & 0.75 & 0.93 + 17 & 0.91 & 0.91 & 0.91 & 0.86 & 0.79 & 0.58 & 0.95 + 18 & 0.85 & 0.88 & 0.89 & 0.80 & 0.63 & 0.58 & 0.87 + 19 & 0.74 & 0.84 & 0.82 & 0.73 & 0.60 & ... & 0.88 + 20 & 0.61 & 0.73 & 0.69 & 0.65 & ... & ... & ... + 21 & 0.54 & 0.66 & 0.66 & 0.72 & ... & ... & ... + 14 & 0.94 & 0.79 & 0.92 & 0.91 & 0.82 & 0.88 & 1.00 + 15 & 0.92 & 0.89 & 0.89 & 0.88 & 0.84 & 0.75 & 0.85 + 16 & 0.87 & 0.88 & 0.86 & 0.82 & 0.66 & 0.53 & 0.91 + 17 & 0.78 & 0.83 & 0.82 & 0.70 & 0.55 & 0.36 & 0.87 + 18 & 0.63 & 0.75 & 0.69 & 0.57 & 0.37 & 0.33 & 0.84 + 19 & 0.45 & 0.61 & 0.55 & 0.47 & 0.32 & ... & 0.79 + 20 & 0.34 & 0.44 & 0.41 & 0.37 & ... & ... & ... + 21 & 0.38 & 0.38 & 0.39 & 0.39 & ... & ... & ... + 14 & 0.80 & 0.73 & 0.82 & 0.81 & 0.61 & 0.70 & 0.00 + 15 & 0.81 & 0.74 & 0.77 & 0.74 & 0.64 & 0.47 & 0.79 + 16 & 0.70 & 0.72 & 0.71 & 0.65 & 0.40 & 0.30 & 0.88 + 17 & 0.53 & 0.65 & 0.62 & 0.46 & 0.31 & 0.19 & 0.80 + 18 & 0.37 & 0.53 & 0.42 & 0.33 & 0.20 & 0.17 & 0.78 + 19 & 0.24 & 0.35 & 0.31 & 0.25 & 0.16 & ... & 0.58 + 20 & 0.18 & 0.23 & 0.22 & 0.19 & ... & ... & ... + 21 & 0.12 & 0.20 & 0.21 & 0.27 & ... & ... & ... + _ photometric statistics : _ from the omcat one can empirically determine the repeatability of flux measurements as a function of source flux as , for example , by @xcite to study the sources of systematic errors .
if the same source is measured twice for a given filter and produces measurements of @xmath52 and @xmath53 , then figure [ fig : phot_stab ] ( left ) shows the distribution of the magnitude difference @xmath54 as a function of the mean magnitude @xmath55 .
one can see that the bulk of the 199000 pairs measured have @xmath56 magnitudes , though there is a long tail to much higher values , in part due to intrinsic variability and ( especially in more crowded fields ) source matching errors .
figure [ fig : phot_stab ] ( right ) shows the distribution of @xmath57 as a function of @xmath55 . if the difference in magnitudes were strictly statistical , the contours in this diagram would be strictly horizontal with the 68% contour at the y - axis value of unity .
table [ tab : phot_stab ] shows the fraction of pairs with @xmath58 0.2 , 0.1 , and 0.05 magnitudes as a function of magnitude for each of the filters .
as can be seen in figure [ fig : phot_stab ] , excluding the very brightest and very faintest sources , the @xmath54 is roughly what one would expect from the quoted uncertainties .
_ comparison to the galex ais : _ perhaps the most comparable uv survey is the _ galex _ ais , an `` all sky survey '' which , when completed , will cover @xmath59% of the sky in its nuv band to a limiting ab magnitude of 20.8 for a @xmath60 detection @xcite
. the closest om equivalent to the _ galex _ nuv filter is the uvm2 filter .
the magnitude distribution of _ galex _ nuv sources with signal - to - noise ratios @xmath61 is sharply peaked at 20.8 and drops to 10% of the peak value by a magnitude of 21.4 .
the magnitude distribution of om sources in the uvm2 filter with signal - to - noise ratios @xmath61 is more broadly peaked ( due to the range of exposure times ) , with a peak at an instrumental magnitude of @xmath62 and declines to 10% of the peak value by an instrumental magnitude of @xmath63 .
the equivalent ab magnitudes are @xmath64 and @xmath65 respectively .
thus , the current omcat has a limiting magnitude in uvm2 between 1.5 and 2.5 magnitudes brighter than the limiting magnitude in the _ galex _ nuv filter .
( however , it should be noted that because of the smaller fov and different brightness limits , the omcat has good coverage of the galactic plane , which _
galex _ is not able to observe . )
_ comparison to the sdss : _ the sdss u filter , like the omcat u filter , is similar to the johnson u. the magnitude distribution of the sdss u band sources with signal - to - noise ratios @xmath61 is sharply peaked at 21.5 and drops to 10% of the peak value by a magnitude of 22.4 .
the magnitude distribution of om sources in the u filter with signal - to - noise ratios @xmath61 is more broadly peaked with a peak at an instrumental magnitude of @xmath66 and declines to 10% of the peak value by an instrumental magnitude of @xmath67 .
the equivalent ab magnitudes are 19.7 and 21.4 .
thus the current omcat has a limiting magnitude in u about 2 magnitudes brighter than the sdss .
the om unique filters are uvw1 , uvm2 , and uvw2 , in order of decreasing throughput ( see figure [ fig : filts ] ) . in order to explore the abilities of the om filters
, we created uvw1@xmath51u _ vs. _ uvm2@xmath51uvw1 and uvw1@xmath51u _ vs. _ uvw2@xmath51uvw1 color - color diagrams of the sources in the omcat ( figure [ fig : color_color ] ) . on those diagrams
we have plotted the expected locus for main - sequence stars ( solar metallicity stars from the 1993 kurucz atlas ) , as well as points representative of galaxies ( taken from the kinney & calzetti atlas at stsci see * ? ? ?
* ) and agn ( taken from stsci agn atlas collection of spectra see * ? ? ?
the conversion from spectra to photometric colors was made using the om spectral response matrices available from the _ xmm - newton _ soc .
it is immediately apparent that the om colors are a good match to those expected from the kurucz atlas , except for the stars later than about g5 .
this problem appears most clearly in the uvw1@xmath51u _ vs. _ uvw2@xmath51uvw1 diagram , but appears as well in the b@xmath51v _ vs. _ u - b diagram .
the problem region is off the bottom of the uvw1@xmath51u _ vs. _ uvm2@xmath51uvw1 diagram .
the true nature of the problem is not yet clear .
for much of the stellar tracks the reddening vectors are roughly parallel to the stellar track .
there is a strong overlap between the early dwarf stars ( down to a5 ) and agn , as well as somewhat later dwarf stars ( a5 to f5 ) and typical galaxies .
there is no way to distinguish early stars from agn using the uv colors alone , and the uv colors of qso do not change much with redshift .
the agn seem to be more offset from the early - type stars in the optical color - color diagram .
given the extinction in the uv , it would be unreasonable to expect to find agn in regions with early type stars , though the converse is not necessarily true , and indeed , finding high latitude o and b stars would be interesting .
division of sources into low ( @xmath42 ) and high ( @xmath68 ) galactic latitude samples shows that in the uv color - color diagram the early type star region is populated somewhat more strongly at higher galactic latitudes while the later type star region is populated somewhat more strongly at lower galactic latitudes .
we have cross - correlated the omcat with the 12@xmath69 veron qso catalogue @xcite and a sdss agn catalogue @xcite and found that the objects in common lie in the expected locations in the color - color diagram ( figure [ fig : qso_color ] ) . as can be seen in figure [ fig : cross ] ,
the cross - correlation of the cfa redshift survey ( the 1995 version as held by heasarc , * ? ? ? * ) with the omcat suggests that the sources falling within the matching radius will be strongly contaminated by false - matches ( at the @xmath70% level ) , meaning that many omcat stellar sources will be falsely matched to the cfa survey .
this high false - matching rate is presumably due to the fact that either the cfa redshift survey has a very high density of objects that are much fainter than those detected in the omcat , or that the cfa redshift survey , being primarily a catalogue of extended objects , have positions that are inherently less precise than those in catalogues of true point sources .
however , the matching sources from the cfa redshift survey fall at the expected location in the color - color diagrams , though their scatter may be a bit larger than the other surveys .
rrrcccc usno b-1 & 1045913669 & 455481 & @xmath71 & 0.303 & 0.199 & 0.702 + 2mass & 470992970 & 265991 & @xmath72 & 0.255 & 0.162 & 0.580 + palomar - green & 1878 & & @xmath73 & & & + tycho-2 & 2539913 & 5322 & @xmath74 & 0.271 & 0.189 & 0.650 + ppm & 468861 & 1177 & @xmath75 & 0.469 & 0.309 & 1.087 + sao & 258944 & 578 & @xmath15 & 0.920 & 0.478 & 1.876 + cfa z cat & 58738 & 1366 & @xmath76 & 1.470 & 0.574 & 2.619 + veron & 108080 & 1099 & @xmath76 & 0.706 & 0.341 & 1.388 + sdss nbc qso & 100563 & 395 & @xmath77mas & 0.323 & 0.210 & 0.742 + galex & 110236958 & 42528 & @xmath76 & 0.647 & 0.349 & 1.346 + fuse & 4037 & 376 & @xmath78 & 1.225 & 0.442 & 2.109 + td1 & 31215 & 82 & @xmath71 & 2.036 & 1.089 & 4.215 + the uses of the omcat detailed here involve the cross - correlation of the omcat with other catalogues .
all the catalogues were extracted from the heasarc catalogue resources using the heasarc browse tool .
the cross - correlation for a given catalogue was done by determining the closest entry to each omcat source .
we then plotted the distribution of the distances between the omcat source and the closest catalogue source .
cross - correlation of the omcat with catalogues with a high density of sources typically produced a distribution of distances with a strong peak at @xmath79 arcseconds and a fwhm of @xmath800.3 arcseconds , and a tail due to serendipitous matches ; the higher the density of catalogue sources , the higher the serendipitous match rate at large distances .
an omcat source was generally considered to be matched if the distance between it and the catalogue source was less than the peak of the distribution @xmath81 .
the catalogues and match criteria used are shown in table [ tab : cross ] while the distributions are shown in figure [ fig : cross ] .
most of the sources in the omcat have usno counterparts .
this suggests that the omcat is shallower in the uv than the usno is in the b and r bands .
the omcat is therefore not dominated by sources previously undetected at other wavelengths . of interest to studies of the halo of the galaxy
are uv - bright stars and agn that can be used as background sources for measuring the column density , velocity , and metallicity of halo gas .
stars are of particular use in determining the distance to high and intermediate velocity gas ( e.g. , * ? ? ?
* ) , while agn provide measures of the column density through the entire halo .
targets are usually found by combing catalogues of sources for objects with optical colors suggesting high uv fluxes .
since the fields in the omcat were ( usually ) chosen for their x - ray sources , rather than the uv sources in the same field , comparing the uv sources in the omcat with the types of catalogues that have been used in the past to find uv - bright objects provides an indication of how many uv - bright sources may have been missed .
we have cross - correlated the omcat with several uv catalogues ; the _ galex _ catalogue , the far ultraviolet explorer ( _ fuse _ ) observation log ( not technically a _ source _ catalogue ) , and the td1 catalogue ; the offset distributions are shown in figure [ fig : cross ] .
we also cross - correlated with the far - uv space telescope ( _ faust _ ) far - uv point source catalogue , but did not find any matches , likely due to the small number of sources in that catalogue .
there are a number of other euv catalogues from _ euve _ and the _ rosat _ wfc , but the position uncertainties are @xmath82 , so they are not useful for this study . at the time that the first omcat was constructed , _ fuse _ was still operational and could have been used for follow - up spectroscopic studies .
even with the demise of _ fuse _ , using the _ fuse _ criteria for finding uv sources of interest to be studied with uv spectroscopy has application for future missions / instruments such as the hst cosmic origins spectrograph .
if searching the omcat for _ fuse _ observable sources produces a trivial result , then the omcat is not likely to be of interest for future missions ; a bountiful return would suggest that the omcat will be useful .
the top panels of figure [ fig : fuse ] show the om color - color diagram with the _ fuse _ matches marked in red and coded by source type .
agn found from cross - correlating the omcat with a number of agn catalogues have also been plotted .
the bulk of the _ fuse _ observations with matching om sources are agn .
the difficulty is that the agn , early type stars , extragalactic star - forming regions , and some galaxies overlap in these color - color diagrams .
we can define a long rectangular region along the early type star track as being contaminated by stars .
above and to the left is a large irregular region that has a significant population of agn .
the lower panels of figure [ fig : fuse ] show the om uvw1 and uvm2 ( or uvw2 ) magnitudes of the sources .
again , the om sources matching _ fuse _ observations have been marked in red and coded by _
fuse _ type .
_ fuse _ observed sources down to uvw1 magnitudes @xmath83 , which suggests sources that are well detected by the om should have been observable by _
we have selected potential spectroscopic targets with @xmath84 , detection significance @xmath19 , @xmath85 magnitudes , @xmath86 magnitudes , and @xmath87 magnitudes or @xmath88 magnitudes , which yielded @xmath801500 uvm2 candidates and @xmath801000 uvw2 candidates .
( even with identical coverage , one would still expect fewer uvw2 candidates simply because of the lower response in the uvw2 filter . )
we then required the sources to fall within the boxes shown in the color - color diagrams of figure [ fig : fuse ] , reducing the number of candidates to @xmath80600 .
we kept the candidates in the `` stellar '' box separate from those in the more extended `` agn '' box . by visual inspection
, we then removed all sources that were actually optical defects ( bad pixels , ghosts , diffraction spikes , etc . ) or were emission regions in nearby galaxies , or were diffuse with no point - like source .
this selection reduced the candidate list to @xmath80250 .
since we are interested in uv sources not in previous catalogues we then removed sources that had been observed by _ fuse_. combining the candidates obtained from the uvw1@xmath51u / uvm2@xmath51uvw1 color - color diagram with those obtained from the uvw1@xmath51u / uvw2@xmath51uvw1 color - color diagram , we found a total of 54 unique sources in the `` agn '' box and a total of 97 unique sources in the `` stellar '' box .
the magnitude distribution is similar to that of the survey as a whole ; of the `` agn '' box sources , 5 are fainter than [email protected] and of the `` stellar '' box sources , 10 are fainter than [email protected] .
the brightest 10% of the sources in both boxes are previously known agn which were the explicit subject of study for the corresponding x - ray observations . combining the two boxes
, we obtained @xmath80100 sources with [email protected] that have not been previously studied .
there is still the matter of source classification and the contamination of the source candidates ( hoped to be stars and agn ) by galaxies . of the candidate sources , somewhat over a third have sdss counterparts with spectroscopic source types . of the sources in the `` agn '' box , @xmath8080%
are categorized as qso and the remainder are categorized as galaxies .
of the sources in the `` stellar '' box , @xmath8040% are categorized as qso , @xmath8040% are categorized as galaxies , and the remainder are categorized as stars .
thus , while the `` agn '' box is a much cleaner sample of agn , the `` stellar '' box still produces a similar number of agn . of the candidates , only 9 ( 3 from the `` agn '' box , 6 from the `` stellar '' box ) fall on or near high velocity clouds .
the remainder of the sources are uniformly distributed across the high galactic latitude sky and will be useful for the study of the structure of the galactic halo in the manner of @xcite .
we matched the _ xmm - newton _ serendipitous source catalogue ( ssc , the second xmm - newton serendipitous source pre - release catalogue , _ xmm - newton _ survey science centre , 2006 ) against the omcat .
we first filtered out all of the _ xmm - newton _ serendipitous sources that did not fall within the field of view of the corresponding om observation . because we did not know the relative positional uncertainty to expect when comparing the x - ray detections with the om detections , we matched the ssc to the omcat in the same way that we matched the usno catalogue to the omcat .
the distribution of the distance between the x - ray sources and the closest om source is shown in figure [ fig : xo_dist_stat ] ; the sharp peak at @xmath91 is presumably due to true matches between x - ray and om sources while the broader peak is due to uncorrelated sources . in order to determine the extent to which our matches are contaminated by serendipitous alignments , we calculated the probability distribution of equation 1 for each source in the ssc that fell within the om fov using the density of om sources in the fov for the observation containing the ssc source .
because some of the ssc sources do have real matches , this calculation overestimates the probability distribution of serendipitous matches . in order to correct for this overestimation
, we assumed that sources with distances of @xmath92 had a non - negligible probability of being real matches .
we then summed over the probability distribution for @xmath93 and normalized that to the total number of matches found with @xmath93 .
we then used this normalized model distribution to determine the fraction of matches at distance @xmath94 which are due to serendipitous coincidences of uncorrelated sources .
we have taken as `` real '' x - ray - omcat matches all those with distances @xmath95 . at @xmath96
there are still four times as many real matches as there are random coincidences but the contamination rate for @xmath95 is @xmath97% . of the 53848 sources from the xmm serendipitous source catalogue falling within the fov of the corresponding om observations and having coordinate corrections from the usno
, 12986 have om source counterparts , of which 1092 are expected to be spurious .
it should be noted that in matching the x - ray sources to the om sources , the usno corrected coordinates provide a much closer match than do the original om coordinates .
the bulk of the x - ray sources are expected to be background agn and galactic stars .
given the observational interest in galaxies , there will also be a small , probably negligible , contribution from extragalactic x - ray binaries and star - forming regions .
the question of interest when comparing catalogues at different energies is whether the x - ray sources detected in the uv fall in distinctive portions of x - ray / uv color / hardness diagrams .
figure [ fig : color_hard ] compares the uvw1@xmath51u color ( the uv color for which we have the greatest number of sources ) with the 2.0 - 12.0/0.5 - 2.0 kev x - ray hardness . from figure
[ fig : color_color ] , we expect agn to have uvw1@xmath51u@xmath98 and stars to have uvw1@xmath51u@xmath99 . the same behavior is observed in the color - hardness diagram , with an additional separation in x - ray hardness ; the x - ray sources with uv colors of agn have log(hard)-log(soft ) peaked around 0.2 and the x - ray sources with uv colors of stars have a broad distribution peaking at log(hard)-log(soft)@xmath99 .
this distribution reflects the well understood difference between the soft thermal x - ray spectra of stars and the harder power - law x - ray spectra of agn .
the combination of x - ray and uv colors are probably a more powerful star - agn discriminator than either alone .
the diagonal line in figure [ fig : color_hard ] marks a reasonable separation between these two categories .
to test this separation we have cross - correlated the sources with good x - ray and uv data with the sdss dr5 catalogue of sources with classifications made from the sdss spectra .
indeed , the agn ( blue ) are clustered as expected .
the bulk of the galaxies ( green ) have stellar colors , but many have agn - like colors .
since the sdss sources chosen for spectroscopy were selected to have a low probability of being stars , based on their colors and being unresolved , the low number of stars ( red ) , and that fact that they have agn - like colors , should not be surprising .
the contours in figure [ fig : hard_uv ] show the density of ssc catalogue sources in the hard ( 2.0 - 12.0 kev ) _ versus _ soft ( 0.5 - 2.0 kev ) band space ; the points show the locations of sources with uvw1@xmath51u colors .
not surprisingly , many of the sources with only soft x - ray detections have uv detections while there are only a few sources that have only hard x - ray detections that also have uv detections .
the hard x - ray sources with uv detections tend to have agn - like colors . to state it in the converse , very few hard x - ray sources ( i.e. , strongly absorbed sources ) have uv counterparts .
the ssc sources with uv counterparts that also have qso counterparts in the sdss spectroscopic survey are shown in figure [ fig : sdss_agn ] . the sources cluster along the tracks expected if the x - ray and uv flux are related by @xmath100 as found by @xcite , where the monochromatic luminosities are in erg s@xmath101 hz@xmath101 .
we assumed the qso template spectrum from @xcite for calculating the uv flux and magnitudes .
we have assumed the photon index of the x - ray spectrum to be @xmath102 , and ( for the plotted tracks ) no internal absorption .
assuming internal absorption moves the tracks by @xmath103 , so the effect is not strong for this x - ray energy band .
although there is significantly more scatter than one would expect from the uncertainties , the sources do lie roughly where expected .
it should be noted that this small sample of sources ( _ xmm - newton _ x - ray sources with om counterparts and sdss spectroscopic survey counterparts ) seems to be a relatively unbiased sample of the sdss spectral catalogue ; the distribution of spectroscopic classification and redshift in the sample matches the distribution of all of the sources in the @xmath104 radius sdss spectroscopic catalogue extracts centered on the om pointing directions . with the larger sample of om sources with x - ray counterparts one will be able to begin to address the source of this scatter .
the omcat provides a quick source of photometric data of point - like sources in the optical and near ultraviolet over an increasing fraction of the sky .
the omcat will continue to be augmented at the heasarc as the _ xmm - newton _ data becomes public .
although the short average exposure places a relatively high detection limit compared to optical catalogues such as usno - b , the current detection limit provides suitably bright targets for current uv spectrometers , and provides high angular resolution data suitable for meaningful comparison with _ galex
_ images .
a first glance at the agn counterparts in the omcat suggests their uv properties are not particularly unexpected .
no doubt more targeted querying of the omcat will produce interesting science in several different fields .
it is expected that the current version of the omcat will be superseded by a sigificantly improved processing produced by the _ xmm - newton _ project . until that catalogue is produced
, the heasarc will continue to augment the current version .
we would like to thank the referee , marcel ageros , for many useful comments .
we would like to thank martin still for many useful discussions about the current and future omcats .
we would like to thank mike arida for his help in making the om data available through the heasarc .
we would like to thank karen levay , randy thompson , and rick white for their help in making this data mast accessible .
we would like to thank luciana bianchi for many useful discussions , as well as b.g .
anderson for his help with _
fuse_. we would also like to thank lorenzo principe for his help with this work .
this research used data obtained through the browse facility of the high energy astrophysics science archive research center ( heasarc ) .
the om object catalogue is now available through browse and in a static ascii version .
several databases accessed through browse are actually held by the _ vizier _
service at http://vizier.u-strasbg.fr . @xcite .
this research has made use of the _ galex _
gr2/gr3 database / archive at the multimission archive at stsci ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
this research has made use of data obtained from data release 5 of the sloan digital sky survey ( sdss ) .
funding for the sloan digital sky survey has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington . | the optical monitor catalogue of serendipitous sources ( omcat ) contains entries for every source detected in the publicly available _ xmm - newton _ optical monitor ( om ) images taken in either the imaging or `` fast '' modes . since the om is coaligned and records data simultaneously with the x - ray telescopes on
_ xmm - newton _ , it typically produces images in one or more near - uv / optical bands for every pointing of the observatory . as of the beginning of 2006 ,
the public archive had covered roughly 0.5% of the sky in 2950 fields .
the omcat is not dominated by sources previously undetected at other wavelengths ; the bulk of objects have optical counterparts .
however , the omcat can be used to extend optical or x - ray spectral energy distributions for known objects into the ultraviolet , to study at higher angular resolution objects detected with _ galex _ , or to find high - galactic - latitude objects of interest for uv spectroscopy . |
in the strongly - interacting regime dephasing is seen to modify the current - driving profile from an ndc curve to being linear in @xmath6 . at @xmath45
this is also coincident with the breakdown of the particle domain pinned to the left boundary giving way to a linear density profile @xmath43 .
both these features suggest that dephasing alters the strongly - interacting maximally biased regime from being insulating to diffusive . here
we present scaling evidence to prove this claim . at @xmath45 , divided by the resulting density difference @xmath138 between the ends of the system ( excluding the boundary ) , is plotted against the system size @xmath139 up to @xmath140 sites .
the data is for dephasing rates @xmath141 ( @xmath142 ) and @xmath143 ( @xmath144 ) .
the solid lines show the fitting to @xmath145 for each @xmath4 . these yield @xmath146 and @xmath147 for @xmath141 , and @xmath148 and @xmath149 for @xmath143 . ]
a diffusive current fulfills the usual diffusion equation @xmath150 , where @xmath151 is the conductivity .
given that when @xmath137 we observe that @xmath43 is linear in @xmath10 everywhere but the boundary sites , its gradient simplifies to @xmath152 where @xmath138 is the density difference between opposite ends of the system ( after discarding the boundary sites ) .
this indicates that a signature of diffusive transport will be a scaling @xmath153 in fig .
[ fig0 ] results are presented for the @xmath154 at @xmath45 with @xmath44 as a function of the system size up to @xmath140 sites with dephasing rates @xmath141 and @xmath143 . also plotted are power - law fittings for each @xmath4 which are seen to accurately model the data .
these show that the decay of current with @xmath12 is sub - diffusive , decaying slower than @xmath155 , for @xmath141 but has approached diffusive behaviour once @xmath156 .
formally a dark - state @xmath157 of an open quantum system is a pure state which is simultaneously an eigenstate of the hamiltonian @xmath158 and a zero - eigenvalue eigenstate of all the jump operators comprising the dissipator @xmath159 describing noise acting on the system .
here we show that deep in the strongly - interacting limit @xmath77 , with fully polarised driving @xmath45 as shown in fig .
[ fig1](a ) , there exists a state @xmath98 which , as the system size @xmath12 grows , becomes exponentially close to satisfying these requirements and is thus an approximate dark state . in the extreme @xmath160 limit configuration states such as @xmath161 , where the particle occupancy on each site of the chain is explicitly specified ,
are exact eigenstates of @xmath79 .
a key property of boundary driving is that for any value of @xmath6 , it only incoherently connects configurations within a quadraplet of states @xmath162 , where @xmath163 is any length @xmath164 occupancy bit string .
thus , if @xmath163 has @xmath165 1 s then the driving couples states within the total number sectors @xmath166 and @xmath167 .
this structure constrains the evolution caused by the driving processes to shuffling population between states in these isolated quadraplets . as depicted in fig .
[ fig1](b ) , at @xmath45 within each quadruplet , defined by @xmath163 , there is one configuration @xmath168 which , owing to the occupancy on the left and empty site on the right , is entirely decoupled from the driving , while also being the sink for all driving transitions . as a result
the effect of this incoherent evolution alone is to eventually drive all the population among the quadraplet of states into this _ dark - configuration_. of particular relevance here are the dark - configurations possessing an @xmath80 particle domain pinned to the left boundary as @xmath169 for each particle number sector @xmath80 the configuration @xmath93 is separated from other relevant configurations by an energy gap @xmath170 , akin to a domain binding energy .
as we move to the limit @xmath77 , hopping between configuration states occurs resulting in @xmath93 giving rise to an eigenstate @xmath78 . in fig .
[ fig2](a ) the exact spectrum for @xmath171 chain with @xmath172 is shown .
for each number sector @xmath80 the spectrum displays a sequence of narrow bands , split by the small finite hopping , and separated by gaps of order @xmath3 .
the highest lying eigenstate @xmath78 for each sector is also highlighted .
its properties are readily determined by treating hopping as a perturbation . specifically , to lowest - order in @xmath173 hopping hybridizes , across an energy gap of @xmath3 , the state @xmath93 with the configuration @xmath174 , where the outermost particle of the domain has broken away . in fig .
[ fig1](c ) the pattern of incoherent driving transitions and coherent hopping for @xmath175 is illustrated .
we note that in this case @xmath176 has no direct coherent transition to any configurations which couples to the @xmath45 driving .
more generally , so long as @xmath177 and @xmath178 we observe that the break - away configuration for a domain state @xmath93 is also a dark configuration of its own quadraplet .
this approach can be taken further by computing higher - order corrections due to hopping . since @xmath77
it is instructive to do this approximately by focusing on states of the form @xmath179 where @xmath180 and @xmath181 , which describe a single break - away particle or hole propagating from the domain wall .
since the repeated action of hopping on @xmath93 originates around the domain wall , and is also detuned by the gap @xmath3 , we find that hopping mixes in configurations where the particle and/or hole have hopped @xmath182 times in total , with an amplitude scaling as @xmath183 .
this indicates that the particle / hole propagation is suppressed with its distance from the domain wall through the unit - filled / empty regions . for each @xmath80 , the eigenstate @xmath78 is therefore predicted by the particle - hole picture to have a domain wall that remains exponentially localized at site @xmath80 , within a length scale @xmath184 @xcite , and a deviation @xmath185 from the perfect domain configuration @xmath93 given by @xmath186 the validity of this picture is established by comparing this to the actual density deviation @xmath185 of the exact eigenstate @xmath78 . in fig .
[ fig2](b ) this is done for @xmath187 with @xmath171 , and the agreement between @xmath185 and @xmath188 is seen to be excellent everywhere but the boundaries . and @xmath172 .
energies @xmath189 have been shifted by @xmath190 so that the state @xmath191 has zero energy .
the spectrum includes contributions from all particle number sectors @xmath192 with each composed of sequences of narrow bands , split by the small finite hopping , and separated by gaps of order @xmath3 ( since the spectra arising from @xmath193 are identical to those of @xmath194 , only the former are shown in addition to @xmath195 ) .
the highest lying eigenstate for each sector @xmath78 is highlighted and seen to be isolated by a gap @xmath3 from eigenstates composed of breakaway configurations @xmath196 .
( b ) for the eigenstate @xmath197 the exact density deviation @xmath185 from the corresponding boundary domain configuration @xmath93 is shown @xmath198 , along with @xmath188 predicted by considering single particle / hole propagation ( solid line ) .
the inset shows a schematic of the expected exponential localization of the domain wall . ]
since hopping predominately connects @xmath93 with dark - configurations of the form @xmath168 , the leading order contribution to the amplitude in @xmath78 for a configuration which is not dark is @xmath199 .
this corresponds to whether the hole or particle has the shortest path to the left or right boundary , respectively . in the @xmath77 limit
we therefore conclude that the eigenstates @xmath78 , with @xmath200 , form a hierarchy of states with @xmath80 , characterised by a decreasing amplitude for any configuration coupling to the @xmath45 boundary driving .
eigenstates @xmath78 with a domain size @xmath80 scaling with @xmath12 thus become exponentially close to being zero eigenstates of the @xmath45 driving with increasing system size . of these states the one with @xmath84 , where the domain spans exactly half the chain , has
the most suppressed coupling @xmath201 to the driving and is thus the closest approximation of all of them to an exact dark state .
next , we examine the expected structure of the non - equilibrium stationary state ( ness ) @xmath19 in the @xmath77 limit .
the open dynamics of particle number conserving systems like that considered here have been studied extensively @xcite with regard to their ergodic and mixing properties establishing that a unique ness exists for any @xmath6 . the key property ensuring this is that for a finite hopping amplitude every configuration within each particle number sector can be reached from any other , while the incoherent ejection / injection of particles by the driving connects neighbouring sectors . as a result the complete state space of the system can be accessed .
furthermore , the ness for this open system will be block diagonal in the number sectors . at @xmath45 the approximate dark states
@xmath78 for each sector @xmath80 are expected to play a prominent role due to their ability to trap population .
this can be better understood by approximating the ness as a statistical mixture with probabilities @xmath83 of these eigenstates in each sector as @xmath202 the purity of the ness in this approximation , @xmath203 , is reduced only through mixing between sectors .
the probabilities @xmath83 are then determined by demanding that at stationarity there is detailed balance condition between the incoherent transition rates connecting neighbouring number sectors ( see inset of fig .
[ fig3 ] ) . for sector @xmath80 , assumed to be frozen in the state @xmath78 , the output transition rates scale with the probability of a hole being at the left boundary as @xmath204 , and the probability of a particle being at the right boundary as @xmath205 .
considering sectors @xmath206 , @xmath80 and @xmath167 we then have the equality of incoming and outgoing transitions in @xmath80 as @xmath207 = \quad\quad\quad\quad\quad\quad\quad\quad \nonumber \\ \quad\quad p_{n-1}\left(\frac{1}{|2\delta|}\right)^{2(n-1 ) } + p_{n+1}\left(\frac{1}{|2\delta|}\right)^{2(n - n-1)}.\end{aligned}\ ] ] these equations are solved inwards from the extremal @xmath208 and @xmath209 sectors , where @xmath210 and @xmath211 , to give @xmath212 for @xmath213 and where @xmath214 is fixed by the normalization condition @xmath215 . in fig .
[ fig3](a ) these predicted probabilities of occupation for each sector @xmath80 are plotted against the exact values for the ness with @xmath216 and @xmath172 , and found to yield excellent agreement aside from the extremal sectors .
owing to the hierarchy of approximate dark states this indicates that the ness will be predominately a mixture of @xmath78 peaked around the
best " dark state with @xmath217 .
this result also predicts that the purity of @xmath19 scales as @xmath218 independent of @xmath12 . in fig .
[ fig3](b ) this prediction is plotted against the exact value of @xmath219 of the ness for @xmath216 as a function of @xmath3 , again showing excellent agreement even as @xmath220 . on a logarithmic scale of the ness @xmath19 occupying the @xmath80 particle sector
is shown for an exact calculation ( @xmath221 ) for @xmath216 and the predicted form ( solid line ) given in eq .
( [ eq : p_predict ] ) .
( b ) the purity @xmath219 against @xmath3 of the ness @xmath19 on a logarithmic scale for an exact calculation ( @xmath142 ) for @xmath216 and the predicted form ( solid line ) given in eq .
( [ eq : purity ] ) .
the inset shows a schematic of the detailed balance approximation applied to compute these predictions . ]
while the discussion of the driven interacting system considered has revolved mainly around its ness properties , the existence of approximate dark states at @xmath45 will also heavily influence its dynamical behaviour far from stationarity .
in particular , if we envisage starting at time @xmath222 in with an empty chain , since the growth of a domain requires the propagation of holes toward the left boundaries , the increasing suppression of this process in each sector @xmath80 implies that the convergence to the ness will be exponentially slow .
this is the physical reason why our numerical calculation of the ness for @xmath45 is limited to @xmath41 .
indeed effective suppression of the current sets in once a small domain , @xmath223 has formed @xcite . as a result dephasing enhancement , identical to that seen here in the ness , equally applies to transient dynamical states where the domain has not yet saturated at @xmath84 .
the analysis so far has focused on @xmath77 and @xmath45 . in this case
we saw that numerous approximate dark states , whose occupation is favoured by @xmath45 boundary driving , cause the stationary state to become insulating .
remaining in the strongly interacting limit we now wish to isolate the effects on the transport for all drivings @xmath6 , i.e. the current driving profile , due to the existence of such bound states .
to do so we construct a simple toy model capturing the essential physics .
the structure of the model is motivated by considering the half - filled domain state @xmath224 and its corresponding break - away configurations @xmath225 which eventually connect it to the boundary driving .
as such the toy model is composed of @xmath92 configuration states @xmath226 for some size @xmath227 . to mimic the interaction binding energy of @xmath224 , we distinguish the configuration @xmath89 by elevating it in energy by @xmath3 above the set of otherwise degenerate configurations @xmath228 which model the break - away states @xmath225 .
in addition to these configuration energies , the state @xmath226 are also coherently coupled to their neighbours via hopping " processes given by @xmath229 the current operator for the model then follows as @xmath230 which measures the flow within the coherently connected configurations @xmath226 . to model the driving in the full system , which incoherently connects one particle number sector to another
, we introduce an auxiliary state @xmath87 whose function is simply to be an intermediary .
the jump operators describing the driving then take the form @xmath231 where @xmath232 and @xmath233 , with @xmath234 and @xmath235 .
thus , via @xmath87 , the driving incoherently induces transitions between the boundary configurations @xmath97 and @xmath89 with a bias @xmath6 . at @xmath31 driving in both directions
is equal and it is easily confirmed that the ness is @xmath236 , yielding @xmath237 as in the full spinless fermion chain . at the opposite limit , @xmath45 , population is asymmetrically driven from @xmath238 . to complete the analogy with the full system the toy model also includes dephasing , at a rate @xmath4 , via the jump operator @xmath239 where @xmath240 whose action is to scramble the phase of any superpositions between @xmath89 and the other configurations .
a schematic of the toy model illustrating all the coherent and incoherent contributions is shown in fig .
[ fig4](a ) . , an energy offset @xmath3 for state @xmath89 , incoherent transitions @xmath241 and @xmath242 between the boundary states and the auxiliary state @xmath87 , and dephasing at a rate @xmath4 .
the rates for the incoherent driving transitions @xmath243 are also listed .
( b ) the ness current @xmath62 , rescaled by @xmath244 , for @xmath90 , as a function of the driving @xmath6 and dephasing @xmath4 , described by the approximation in eq .
( [ eq : toy_model ] ) .
the behaviour at @xmath45 with @xmath4 is emphasised by the additional ( red ) line . ]
the current @xmath245 for the ness @xmath19 can be solved analytically for the toy model as a function of @xmath246 and @xmath3 , although the complete expression is lengthy . since the model was motivated by the perturbative limit @xmath77 of the full system the physically relevant part of this result
is found by keeping only the lowest order terms in @xmath247 .
this gives @xmath248 in fig .
[ fig4](b ) the current - driving profile @xmath62 , rescaled by @xmath244 , is plotted for @xmath90 .
two key features emerge from this result .
for @xmath63 the @xmath249 in the numerator of eq .
( [ eq : toy_model ] ) , which enforces zero current at the @xmath31 and @xmath45 , causes @xmath62 to display negative differential conductivity ( ndc ) .
moreover , as the expansion involves only even powers of @xmath247 this behaviour is independent on the sign of @xmath3 .
dephasing enhancement of the @xmath62 is evident from the linear @xmath4 term in eq .
( [ eq : toy_model ] ) which eventually destroys the ndc effect .
however , since @xmath4 and @xmath5 appear only linearly it indicates that this lowest order expression is also only valid for weak dephasing and driving rates .
in particular the dephasing degrading behaviour expected for large @xmath4 , due to the zeno effect , is not described by eq .
( [ eq : toy_model ] ) .
the expression is also only valid for @xmath250 because ndc is not seen for @xmath251 .
this is similar to how ndc is only seen for @xmath177 in the full system . just like in fig .
[ fig1](c ) for the full system , having no direct coherent coupling between the boundary configurations where driving occurs , i.e. @xmath97 and @xmath252 for @xmath253 , is essential for ndc to emerge . in fig .
3(b ) of the main article the exact current - driving profile for the toy model is shown for @xmath44 as a function of moderate @xmath4 s , beyond the applicability of eq .
( [ eq : toy_model ] ) .
this confirms the wider similarity of the response of the toy model to that observed in the full spinless - fermion system . of the hamiltonian for the toy model for @xmath90 and @xmath44 .
eigenstate @xmath254 is the highlighted auxiliary state @xmath87 whose energy has been set arbitrarily .
the highest - lying eigenstate @xmath255 is also highlighted and is the approximate dark - state of the model @xmath98 . beneath this state , separated by a gap @xmath170 ,
is a band of eigenstates @xmath256 split by the finite hopping .
( b ) the probability distributions @xmath257 of the eigenstates over the configurations @xmath258 are shown .
the band of eigenstates are seen to be delocalized over the configurations @xmath259 and expunged from the boundary configuration @xmath89 , while characteristic of a bound state @xmath98 is predominately peaked at @xmath89 with exponential tail into bulk . ]
given their similar behaviour , the toy model provides a tractable means of unravelling the origins of ndc and dephasing enhancement in the full many - body system . in fig .
[ fig5](a ) we plot the energy eigen - spectrum of the toy model . by construction
we see that it mimics some of the features seen in fig .
[ fig2](a ) for a single particle number sector of the spinless fermion system . specifically
, there is a high - lying eigenstate @xmath98 , separated by a gap @xmath170 from a dense band of eigenstates . in fig .
[ fig5](b ) this band of eigenstates are seen to be delocalized over the bulk of the system excluding the boundary configuration @xmath89 . for a given size @xmath92 the eigenstate
@xmath98 has the form @xmath96 , to within @xmath260 , as seen in fig .
[ fig5](b ) .
as @xmath261 such an exponential wave function is simply the discrete analogue of the well known bound state of a 1d @xmath262-potential .
given that the amplitude for the left boundary configuration @xmath97 scales as @xmath99 , we see that @xmath98 becomes exponentially close , with increasing @xmath92 , to being a dark state of the driving when @xmath45 .
the driving at @xmath45 exclusively pumps into the configuration @xmath89 whose dominant overlap is with @xmath98 .
consequently so long as @xmath63 population gets progressively trapped in this dark state giving rise to @xmath263 , characteristic of an insulating ness .
remaining at @xmath45 and switching on a non - zero dephasing directly decoheres the exponentially decaying superposition within @xmath98 .
this is equivalent to the coherent trapping of population , caused by the energetic gap , being bypassed by dephasing induced incoherent transitions connecting @xmath98 directly to the delocalized band of eigenstates .
current flow in the system is thus made possible via the ensuing non - stationary mixture of these eigenstates .
further increases in dephasing eventually degrades the current once the monotonically decreasing mobility of the delocalized eigenstates , caused by the zeno effect , outweighs the flux of population escaping from @xmath98 . since the toy model only has a single approximate dark state @xmath98 it confirms that its existence , at one isolated point @xmath45 and @xmath100 , is alone enough to make the current - driving profile for @xmath264 and @xmath265 exhibit both ndc and dephasing enhancement .
( b ) the ness current @xmath62 , rescaled by @xmath244 , for @xmath90 is shown as a function of the driving @xmath6 and dephasing @xmath4 zoomed in around the limit @xmath45 , @xmath100 insulating point .
the behaviour at @xmath45 with @xmath4 is emphasised by the additional ( red ) line .
( b ) in the strong driving limit @xmath68 , the @xmath266 process @xmath267 occurs at a rate @xmath268 making it slow ( as indicated by being faded out ) , while the @xmath269 process @xmath103 occurs at a rate @xmath270 and so is rapid .
the dominant driving process is therefore @xmath271 , which has the effect of destroying any coherence @xmath89 has with other states , such as @xmath272 . ]
another key insight from the toy model is that the emergence of a non - zero current from the insulating point @xmath45 and @xmath100 , involves identical physics either when @xmath4 is increased slightly from zero , or when @xmath6 is reduced slightly from unity .
examining eq .
( [ eq : toy_model ] ) at @xmath45 shows that @xmath273 , while for @xmath100 an expansion about @xmath45 the current is @xmath274 to lowest order .
this suggests a correspondence @xmath275 which is confirmed in fig .
[ fig6](a ) in an exact calculation for @xmath90 .
the equivalence is understood by considering the effect of the driving slightly below @xmath45 where a backward flow process is introduced . in this limit
the rate of driving from @xmath267 taking the state out of @xmath98 is given by @xmath276 and therefore slow .
in contrast the rate of driving @xmath103 back is @xmath277 , and therefore rapid .
focusing on the dynamics of these two driving processes alone , as in fig . [ fig6](b ) , we suppose that the initial state of the system over the configurations @xmath278 and @xmath89 has the form @xmath279 where there is some coherence between @xmath272 and @xmath89 , but no population initially in @xmath87 . evolving this state according only to these driving processes yields solutions
@xmath280 we therefore find that for the populations the stationary @xmath281 limit is approached at a rate @xmath5 to give @xmath282 and @xmath283 , while the coherence @xmath284 decays to zero at a rate @xmath285 .
now , by considering only the dephasing process we instead find that the populations are unchanged while the coherence decays as @xmath286 matching of these decoherence rates fits with eq .
( [ eq : gamma_f ] ) .
we therefore conclude that the emergence of a non - zero current when reducing @xmath6 from unity is , to leading order , caused by the resulting decoherence of the dark state @xmath98 , identical to the effect of dephasing alone .
this behaviour with @xmath6 around @xmath45 , combined with @xmath263 at @xmath31 and the continuity of @xmath62 with @xmath6 , is already enough to imply that ndc behaviour will be observed in the current - driving profile .
thus ndc and dephasing enhancement are underpinned by the same mechanism . across the chain as a function of the fraction distance @xmath287 about the centre for @xmath44 and @xmath71 .
in ( a ) a moderate dephasing @xmath288 was included .
the correlations are long - ranged and extend over a larger fraction with increasing @xmath12 . this evidenced by the kink , coming from @xmath289 changing sign , moving to larger @xmath287 . in
( b ) a strong dephasing @xmath290 was present and the correlations are short - ranged as shown by their diminishing size and extent with increasing @xmath12 . ]
returning to the full spinless fermion system we now consider the effect of dephasing on two - point correlations across the chain .
we find that in addition to modifying the ness current and density profiles , dephasing also leaves a signature on the correlations which distinguishes between weakly- and strongly - interacting regimes .
specifically we compute the ness density - density correlations @xmath291 , where sites @xmath292 and @xmath10 are positioned symmetrically around the centre of the chain .
this is conveniently represented as @xmath289 where @xmath293 is the fractional separation of the points for the system size @xmath12 , so for example @xmath294 gives the correlations between the boundaries . in fig .
[ fig7](a ) @xmath289 is plotted for different sizes @xmath12 in the strongly interacting regime with @xmath44 , @xmath71 and a moderate dephasing rate @xmath288 .
finite correlations are seen to exist even for a large fractional separation @xmath287 , and although they decay with @xmath287 importantly these correlations are retained for a growing fraction of the chain as @xmath12 increases .
this property has been argued , in the @xmath63 case , to be evidence that the ness possess genuine long - range order @xcite . here
our results show that this long - range order persists even in the presence of moderate dephasing , and so correlations similar to that at @xmath63 exist in the dephasing enhanced ness .
the situation for stronger dephasing @xmath290 , shown in fig .
[ fig7](b ) , is markedly different .
now correlations are smaller and diminish faster with @xmath287 than at weaker dephasing rates . also the correlations which are present
cover a smaller fraction @xmath287 ( implying a fixed absolute size ) as the system size @xmath12 grows .
the dephasing has therefore degraded the correlations to being short - ranged . moving to the weakly interaction regime with @xmath56
it has previously been found @xcite that correlations are short - ranged , similar in form to fig .
[ fig7](b ) , even at @xmath63 .
indeed a cross - over for @xmath289 from short - ranged to long - ranged correlations at @xmath63 was found at @xmath295 @xcite . like the current and density profiles , we see that also the correlations in the strongly interacting regime with large dephasing rates become increasingly similar to those of the weakly interacting regime , washing out this cross - over . | a key insight from recent studies is that noise , such as dephasing , can improve the efficiency of quantum transport by suppressing coherent single - particle interference effects .
however , it is not yet clear whether dephasing can enhance transport in an interacting many - body system .
here we address this question by analysing the transport properties of a boundary driven spinless fermion chain with nearest - neighbour interactions subject to bulk dephasing .
the many - body non - equilibrium stationary state is determined using large scale matrix product simulations of the corresponding quantum master equation .
we find dephasing enhanced transport only in the strongly interacting regime , where it is shown to induce incoherent transitions bridging the gap between bound dark - states and bands of mobile eigenstates .
the generic nature of the effect is illustrated and shown not to depend on the integrability of the model considered . as a result dephasing enhanced transport
is expected to persist in more realistic driven systems of strongly - correlated particles .
_ introduction _ @xmath0 recently the effects of noise on the efficiency of quantum transport phenomena have been scrutinised intensely by the scientific community .
this has been motivated in part by a series of ground - breaking nonlinear spectroscopic experiments on light - harvesting complexes demonstrating surprisingly long - lived quantum coherence during exciton transport , even in a warm and wet environment @xcite . yet transport for purely coherent exciton dynamics in such protein pigment networks is highly suppressed due to destructive interference between different propagation pathways .
instead studies revealed that the remarkably high transport efficiency observed ( above 95% ) in fact emerges in combination with local noise , such as dephasing , which disrupts this interference opening up previously inhibited pathways for transmission @xcite .
transport properties in such open systems thus not only defy the traditional understanding of when quantum effects should play a significant role , but also challenge the notion that couplings to the environment unconditionally degrade quantum phenomena .
different effects of dephasing have been studied in networks populated by single particles , in scenarios such as transport through quantum optical systems @xcite , information transmission @xcite and quantum information processing @xcite .
however , the interplay between noise and strong correlations induced by interactions in a many - body setting is not yet fully understood .
in particular a recent experiment @xcite with cold - atoms exploring these effects during time - of - flight expansion raises an important question as to whether and how dephasing can enhance transport in an interacting system .
here we answer this question in the affirmative by considering a concrete example composed of spinless fermions with nearest - neighbour interactions hopping through a tight - binding chain , as depicted in fig .
[ fig1 ] .
this model makes an ideal testbed for several reasons .
first , it is equivalent to the well studied @xmath1 spin-@xmath2 chain @xcite , representing one of the simplest models of strongly - correlated electron systems , and its physics is still not fully understood despite being bethe ansatz solvable @xcite .
second , the transport properties of such low - dimensional interacting quantum systems remains an important open problem , exhibiting anomalous features such as ballistic spin transport and unusually high thermal conductivity , reported experimentally in so - called spin - chain materials @xcite .
in addition to solid state systems like chains of coupled quantum dots @xcite or molecular wires embedded between electrodes @xcite , understanding this model is directly relevant to ion - trap @xcite , coupled - cavity array @xcite , and cold - atom @xcite quantum systems .
of particular relevance are recent seminal experiments that revealed contact and bulk resistivity of cold fermionic atoms flowing within a narrow mesoscopic channel between a pair of reservoirs @xcite . across a chain ,
subject to a nearest - neighbour density - density interaction of strength @xmath3 , local dephasing at a rate @xmath4 , and boundary driving that injects / ejects fermions at a rate proportional to @xmath5 and driving bias @xmath6 . ]
very similar to this cold - atom experiment we consider a chain attached to two unequal markovian reservoirs at its boundaries providing continuous incoherent driving , along with local dephasing noise along its extension ( see fig . [ fig1 ] ) @xcite .
the transport properties of the chain are then found by computing the current - driving characteristics of its non - equilibrium stationary state ( ness ) .
being a one - dimensional and homogeneous chain , previously studied single - particle interference effects originating from geometry or disorder are absent @xcite .
surprisingly , we show that dephasing - enhanced transport nonetheless emerges so long as the interactions are strong enough . in this regime
the ness forms a cooperative many - body quantum state with a long - ranged domain of particles pinned to one boundary strongly suppressing the current , analogous to a coulomb or pauli blockade insulator @xcite .
even a small dephasing rate induces incoherent transitions between this bound state , preferentially occupied by the driving , and bands of mobile scattering states .
this effect significantly enhances transport , most especially for regimes where the current is otherwise vanishingly small .
we isolate the essential conditions for enhancement and demonstrate its generality beyond the integrable system considered .
furthermore , the mechanism described helps explain recent experimental results with cold - atoms where the expansion of a strongly - interacting atomic gas was found to be slow in the absence of noise and substantially increased once noise was added @xcite . _ model _
@xmath0 we study the interacting spinless fermion chain described by the hamiltonian ( taking @xmath7 throughout ) @xmath8,\ ] ] where @xmath9 are standard fermionic creation / annihilation operators for site @xmath10 , @xmath11 is the associated number operator , and @xmath12 is the number of sites .
in addition to the tight - binding hopping terms , with amplitude @xmath13 , this hamiltonian has a nearest - neighbour density - density interaction @xmath14 which , depending on the sign of its strength @xmath3 , either penalizes ( @xmath15 ) or promotes ( @xmath16 ) the adjacency of fermions .
we take @xmath17 to set the energy scale to which all other quantities are measured .
the dynamics of the system is described by a lindblad quantum master equation @xcite @xmath18 + \mathcal{l}(\rho),\ ] ] where @xmath19 is the density matrix of the chain , and @xmath20 is the dissipator describing the coupling to the markovian reservoirs . in lindblad form
the dissipator can be expressed as @xmath21 , where @xmath22 is the anti - commutator and the sum is over a set of jump operators @xmath23 .
we consider a dissipator that splits into three parts @xmath24 . here
@xmath25 and @xmath26 describe the coupling to external particle reservoirs at the left and right boundaries , respectively , each with two jump operators @xmath27 and @xmath28 , where @xmath5 is the coupling strength , identical for both reservoirs , and @xmath29 is the driving bias @xcite . we consider moderate coupling @xmath30 throughout this paper @xcite .
the effect of the driving is to force the system far from equilibrium thereby giving a quantum analogue of the well studied classical exclusion stochastic model @xcite . as depicted in fig .
[ fig1 ] it is composed of two processes : _ forward pumping _ creating particles at the left edge and removing them at the right , and _ backwards pumping _ doing the opposite .
when @xmath31 these processes have equal strength and so cancel exactly , resulting in the stationary solution @xmath32 to eq .
( [ master_eq ] ) with no net current , irrespective of @xmath3 @xcite .
for @xmath33 the bias favours forward pumping raising the possibility of a ness possessing a finite current . the remaining contribution
@xmath34 accounts for bulk dephasing in the chain and is described by a jump operator @xmath35 for each site @xmath36 with a uniform dephasing rate @xmath4 . by directly simulating eq .
( [ master_eq ] ) and taking the long time limit , the state @xmath37 converges to the time - independent ness of the system .
a solution can be computed efficiently in a controlled way and accounting for significant many - body correlations by applying the time evolving block decimation ( tebd ) algorithm @xcite to a matrix product operator description of @xmath37 .
this highly compact representation enables the efficient evaluation of relevant expectation values and makes accessible much larger system sizes than exact diagonalization or monte carlo approaches @xcite .
moreover tebd can be applied effectively over a large parameter range allowing us to examine the properties of the system as a function of @xmath6 beyond the @xmath38 linear response regime @xcite .
the transport properties are found by computing the current crossing site @xmath10 from the operator @xmath39 . in the ness
the current @xmath40 is homogeneous throughout the system . for increasing ( as indicated by the arrow ) dephasing rates @xmath4 and @xmath41 .
( b ) identical plot to ( a ) but in the strongly - interacting regime @xmath42 .
( c ) the density @xmath43 of the system when @xmath44 , @xmath45 , and @xmath41 , corresponding to @xmath46 .
( d ) the current as a function of @xmath4 for driving biases @xmath47 and @xmath48 ( from bottom to top curve ) with @xmath42 , and @xmath49 .
the dashed lines correspond to @xmath50 the non - interacting analytic result @xcite . ]
_ dephasing enhanced transport _
@xmath0 it is known that in the weakly interacting regime @xmath51 , in the absence of dephasing , the system is an ideal ballistic conductor for any driving @xmath6 , with a nearly flat density @xmath43 and a current @xmath52 which is independent of @xmath12 @xcite .
the introduction of dephasing has been shown to induce diffusive transport , where @xmath43 instead features a constant gradient and @xmath53 , characteristic of an ohmic conductor @xcite . in either case
the maximum current through the chain occurs at maximal bias @xmath45 , where only forward pumping is present , as might be intuitively expected . in the non - interacting limit @xmath54
it has been proven rigorously that a homogeneous chain can not exhibit any dephasing enhanced transport @xcite ; this behaviour was also suggested for weakly - interacting systems where @xmath55 @xcite . in fig .
[ fig2](a ) a complete picture of the latter result is presented .
specifically , for @xmath56 the reported current - driving profiles show that dephasing monotonically degrades the current for any driving , confirming that this behaviour persists even in the presence of weak interactions .
thus , the present work focuses on the strongly interacting regime @xmath57 . in the absence of dephasing , and for weak driving @xmath38
, transport was found @xcite to be diffusive when @xmath58 , a controversial finding given that the integrability of the system was initially conjectured to lead to ballistic transport @xcite .
however , for strong driving @xmath59 it was discovered @xcite that the ness exhibits a particle domain at the left edge of the chain , e.g. dominated by the configuration @xmath60 , irrespective of the sign of @xmath3 .
this strongly suppresses the current as @xmath61 , characteristic of an insulator .
consequently , the current @xmath62 at @xmath63 exhibits non - linear behaviour with the driving @xmath6 , leading to an effect known as negative differential conductivity ( ndc ) where increasing the driving eventually decreases the current @xcite . in fig .
[ fig2](b ) the @xmath63 curve shows that this causes a near complete suppression at @xmath45 for @xmath42 .
consequently , in the strongly interacting regime the system presents the intriguing property that more current flows at an intermediate bias @xmath64 where some backward pumping is present .
the main result of this work is that for @xmath65 the presence of a small bulk dephasing can significantly enhance the current .
this striking behaviour is illustrated for @xmath44 in fig .
[ fig2](b ) where dephasing up to a moderate rate @xmath66 is seen to increase the current .
this elevation in the current is shown to occur _ for any _ @xmath33 , however , it is not uniform in @xmath6 resulting in the current - driving profile changing with @xmath4 . specifically , around @xmath67
the ndc effect is lost and further increases in @xmath4 yield a linear profile in @xmath6 , resembling a diffusive conductor ( see supplementary material ) .
this indicates that the most dramatic enhancement of the current occurs for strong driving biases @xmath68 , where the system changes from being insulating at @xmath63 to yielding the maximal current once @xmath69 . in fig .
[ fig2](c ) this change at @xmath45 is shown to be coincident with the breakdown of the particle domain at the left boundary into a near linear density profile even for small dephasing rates .
in this regime dephasing therefore induces not just quantitative increases in the current , but rather causes a major qualitative change in the behaviour of the system . for a given @xmath6
there is an optimal dephasing rate @xmath70 , beyond which the current is degraded by dephasing .
this is illustrated in fig .
[ fig2](d ) , where it is shown that both the maximum enhancement of the current and @xmath70 increase with @xmath6 . while for a weak driving @xmath71 the current enhancement at its @xmath70 is @xmath72 relative to @xmath63 , for @xmath45 the enhancement is @xmath73 .
for @xmath74 the current is reduced because the @xmath34 contribution to eq .
( [ master_eq ] ) dominates over the coherent hopping terms and freezes out the dynamics by the zeno effect @xcite .
in fact for increasingly large dephasing rates the interaction becomes increasingly irrelevant , as demonstrated in fig .
[ fig2](d ) by the convergence of the ness current to the exact @xmath75 solution with dephasing @xmath76 $ ] @xcite . _ enhancement mechanism _
@xmath0 the origin of the dephasing enhanced transport in the strongly - interacting regime is intimately connected to the appearance of an insulating ness at @xmath45 .
this is most easily revealed by examining the very strongly - interacting limit @xmath77 with totally biased driving @xmath45 .
consistent with the numerical findings above , the ness in this extreme limit is insulating for two reasons : ( i ) when @xmath77 there exists an eigenstate @xmath78 of @xmath79 for each number sector @xmath80 , which is a bound state energetically separated from other eigenstates ; as @xmath81 it becomes exponentially close , with increasing @xmath12 , to being a dark state of @xmath45 driving ; ( ii ) the ergodic nature of the driving and coherent dynamics results in the unique ness being well approximated by a statistical mixture of @xmath82 , with the probability @xmath83 exponentially peaked at @xmath84 half - filled domain .
the formation of an insulating ness thus arises due to an interplay between the nature of the boundary driving and the eigenstructure of the chain hamiltonian . in the supplementary material
we show ( i ) and ( ii ) by examining the structure of the driving and coherent hybridisation of the @xmath80 particle boundary configuration @xmath85 that forms @xmath78 .
crucially , bulk dephasing induces energy dissipation at a rate @xmath86 proportional to the kinetic energy of the state . as a result dephasing incoherently bridges the spectral gap between @xmath78 and mobile bands of scattering states , thereby dramatically enhancing the current as population escapes the dark state .
coupled coherently to their neighbours and with incoherent driving , via an auxiliary state @xmath87 , between states @xmath88 .
crucially state @xmath89 is elevated in energy by @xmath3 .
see supplementary material for precise definitions of the model .
( b ) the current - driving profiles of the toy - model as a function of @xmath4 with @xmath42 and @xmath90 . ] the effect outlined is captured concretely by a simple toy - model , depicted in fig . [ fig3](a ) .
the model comprises of states @xmath91 , for some size @xmath92 , with coherent coupling between neighbouring states .
the state @xmath89 is elevated in energy by @xmath3 above the rest , analogous to the configuration @xmath93 ; the other states correspond to breakaway configurations like @xmath94 .
this yields a mobile band of @xmath95 eigenstates and a single bound state of the form @xmath96 , irrespective of the sign of @xmath3 . via an auxiliary state @xmath87 , modelling another particle number sector , incoherent driving is included between states @xmath97 and @xmath89 with an identical @xmath6 dependence as the full system ( see supplementary material for details ) .
dephasing is also introduced which scrambles the phase between @xmath89 and all other states .
now since the amplitude for @xmath97 in @xmath98 scales as @xmath99 it is exponentially close to being a dark state of @xmath45 driving .
moreover , as @xmath45 driving pumps into the majority component @xmath89 of @xmath98 , it leads to an insulating ness at the extremal point @xmath45 and @xmath100 .
the current - driving profile is shown as a function of @xmath6 and @xmath4 in fig .
[ fig3](b ) , and is seen to reproduce both ndc and dephasing enhancement of the strongly - interacting regime of the full system .
the toy - model thus reveals that this behaviour is entirely a consequence of the insulator at @xmath45 and @xmath100 .
moreover , it also shows that same physics underpins the ndc effect as dephasing enhancement . when @xmath45 and @xmath101 dephasing induces incoherent transitions to the mobile band .
similarly for @xmath100 and @xmath64 the leading order effect of the weak backward pumping process @xmath102 , followed rapidly by the stronger forward pumping process @xmath103 , is to decohere @xmath98 . and @xmath49 : ( a ) the half - chain entropy @xmath104 as a function of @xmath3 for different @xmath4 s , and ( b ) the optimal dephasing rate @xmath105 as a function of @xmath3 .
the circles indicate tebd results , and the solid line corresponds to the fitted function @xmath106 , with @xmath107 and @xmath108 .
the inset plots show the generic behaviour of @xmath62 with @xmath4 above and below @xmath109 .
( c ) the scaling of @xmath109 with @xmath12 along with the fitted power - law shown where @xmath110 , @xmath111 and @xmath112 .
( d ) the current for several staggered potential strengths @xmath113 with @xmath114 ( upper panel ) and @xmath115 ( lower panel ) . ] _ signature of a non - equilibrium phase transition _ @xmath0 the toy - model is only valid for @xmath77 and predicts ndc for any @xmath116 . in the non - interacting @xmath75 limit the exact solution
@xmath50 indicates that dephasing acts only to degrade the current .
however , numerical results presented in fig . [ fig2](a ) already indicate that dephasing - degrading behaviour persists with weak interactions @xcite .
this suggests that between the regimes @xmath117 and @xmath77 a cross - over in the response of the system to dephasing emerges .
the point @xmath118 is of particular significance since at zero - temperature it separates the equilibrium phases of the underlying spinless fermion hamiltonian , marking the opening of a gap above the ground state , and more generally divides a continuous eigen - spectrum for @xmath119 from one with numerous gaps for @xmath65 @xcite . in the driven system it was previously observed that @xmath120 also separates ballistic and diffusive transport regimes for weak driving @xcite , however directly linking these equilibrium and non - equilibrium properties can be misleading @xcite . for @xmath100
, studies have attempted to identify the non - equilibrium regimes by examining two - point correlations , such as @xmath121 .
the strongly - interacting regime was found to retain correlations extending over a fraction of the entire system as its size increases , while the weakly - interacting regime does not , with a sharp cross - over found at @xmath122 @xcite .
here we refine this signature by adopting a more general measure of correlations . specifically , we compute the entropy @xmath123 of the schmidt coefficients @xmath124 arising when the full ness density operator @xmath19 is factorized into two half - chains as @xmath125 @xcite .
the entropy @xmath104 quantifies both the quantum and classical correlations between the two halves of the chain and is readily accessible within the matrix product operator description used in tebd numerics . for zero dephasing fig .
[ fig4](a ) shows that @xmath104 features a peak at @xmath126 , indicating that a significant elevation in correlations occurs as the ness reorganises itself across the expected non - equilibrium phase transition between ballistic and diffusive transport there . for finite dephasing the entropy @xmath104 monotonically decreases with @xmath4 , but interestingly , the peak is maintained for moderate rates shifting slightly to larger @xmath3 .
similar to @xmath127 this distinguishing feature in @xmath104 is progressively washed out as the system becomes diffusive for all @xmath3 with dephasing @xcite ( see also supplementary material ) . despite this , dephasing in fact offers an alternative and clear signature of this underlying non - equilibrium phase transition . in fig .
[ fig4](b ) we show the optimal dephasing rate @xmath70 for weak driving as a function of @xmath3 , for @xmath49 .
a threshold of @xmath128 is apparent where for @xmath129 , @xmath130 , indicating dephasing - degrading transport , while for @xmath131 , @xmath105 is non - zero and increases monotonically with @xmath3 , indicating dephasing - enhanced transport .
the value of @xmath109 is size - dependent , and a scaling analysis shown in fig . [
fig4](c ) demonstrates that , to good approximation , the isotropic point @xmath132 separates these regimes in the thermodynamic limit .
the abrupt on - set of dephasing enhancement therefore occurs when the many - body strong correlations are largest .
_ conclusions _
@xmath0 we have presented a study of the effects of dephasing on the transport properties in a boundary driven one - dimensional interacting spinless fermion chain .
the appearance of a cooperative many - body ness exhibiting ndc provides a previously unexplored form of dephasing enhanced transport , distinct in origin from earlier examples in non - interacting systems . using a toy model we isolated the minimum requirements for observing ndc and dephasing enhanced transport in a strongly - interacting system .
the effect is also unrelated to integrability @xcite , suggesting that our findings it will also apply in related , more realistic models of strongly - correlated electrons such as the @xmath133 or hubbard models , and in heat transport @xcite .
moreover , tantalising signs of dephasing energy dissipation from bound states predicted here have already been observed in a recent cold - atom experiment @xcite .
finally , the prospects of verifying more directly both ndc and dephasing enhancement with cold - atoms seems promising @xcite .
j. j. m .- a . acknowledges departamento administrativo de ciencia , tecnologa e innovacin colciencias for economic support .
d.j . and s.r.c .
thank the national research foundation and the ministry of education of singapore for support .
99 g. s. engel , t. r. calhoun , e. l. read , t .- k .
ahn , t. mancal , y .- c .
cheng , r. e. blankenship and g. r. fleming , nature * 446 * , 782 ( 2007 ) .
h. lee , y .- c .
cheng , and g. r. fleming , science * 316 * , 1462 ( 2007 ) .
e. collini , c. y. wong , k. e. wilk , p. m. g. curmi , p. brumer , and g. d. scholes , nature * 463 * , 644 ( 2010 ) .
m. mohseni , p. rebentrost , s. lloyd and a. aspuru - guzik , j. chem . phys . * 129 * , 174106 ( 2008 ) . f. caruso , a. w. chin , a. datta , s. f. huelga and m. b. plenio , j. chem . phys . * 131 * , 105106 ( 2009 ) .
p. rebentrost , m. mohseni , i. kassal , s. lloyd and a. aspuru - guzik , new j. phys .
* 11 * , 033003 ( 2009 ) .
a. w. chin , a. datta , f. caruso , s. f. huelga and m. b. plenio , new . j. phys .
* 12 * , 065002 ( 2010 ) .
f. caruso , n. spagnolo , c. vitelli , f. sciarrino and m. b. plenio , phys .
rev .
a * 83 * , 013811 ( 2011 ) .
f. caruso , s. f. huelga and m. b. plenio , phys .
rev .
lett . * 105 * , 190501 ( 2010 ) .
a. bermudez , t. schaetz and m. b. plenio , arxiv : 1210.2860v2 ( 2012 ) . c. derrico , m. moratti , e. lucioni , l. tanzi , b. deissler , m. inguscio , g. modugno , m.b .
plenio and f. caruso , arxiv:1204.1313 ( 2012 ) .
d. gobert , c. kollath , u. schollwck and g. schtz , phys .
rev .
e * 71 * , 036102 ( 2005 ) .
s. langer , f. heidrich - meisner , j. gemmer , i. p. mcculloch and u. schollwck , phys .
rev .
b * 79 * , 214409 ( 2009 ) .
s. keler , a. holzner , i. p. mcculloch , j. von delft and f. marquardt , phys .
rev .
a * 85 * , 011605(r ) ( 2012 ) .
h. bethe , z. phys .
a * 71 * , 205 ( 1931 ) .
a. v. sologubenko , t. lorenz , h. r. ott and a. freimuth , j. low temp . phys . *
147 * , 387 ( 2007 ) .
f. heidrich - meisner , a. honecker and w. brenig , eur .
phys .
j. special topics * 151 * , 135 ( 2007 ) .
n. hlubek , p. ribeiro , r. saint - martin , a. revcolevschi , g. roth , g. behr , g. bchner and c. hess , phys .
rev .
b * 81 * , 020405(r ) ( 2010 ) . o. janson , a. a. tsirlin and h. rosner , phys . rev .
b * 82 * , 184410 ( 2010 ) .
r. hanson , l. p. kouwenhoven , j. r. petta , s. tarucha and l. m. k. vandersypen , rev .
mod .
phys . * 79 * , 1217 ( 2007 ) .
j. c. cuevas and e. scheer , _ molecular electronics : an introduction to theory and experiment _ ( world scientific , 2010 ) .
s.trotzky , p. cheinet , s. flling , m. feld , u. schnorrberger , a. m. rey , a. polkovnikov , e. a. demler , m. d. lukin , and i. bloch , science * 319 * , 295 ( 2008 ) .
j. simon , w. s. bakr , r. ma , m. e. tai , p. m. preiss , and m. greiner , nature * 472 * , 307 ( 2011 ) .
j. t. barreiro , m. mller , p. schindler , d. nigg , t. monz , m. chwalla , m. hennrich , c. f. roos , p. zoller , and r. blatt , nature * 470 * , 486 ( 2011 ) .
a. kay and d. g. angelakis , epl ( europhysics letters ) * 84 * , 20001 ( 2008 ) .
j .-
p .
brantut , j. meineke , d. stadler , s. krinner , and t. esslinger , science * 337 * , 1069 ( 2012 ) .
d. stadler , s. krinner , j. meineke , j .-
p .
brantut , and t. esslinger , nature * 491 * , 736 ( 2012 ) .
g. benenti , g. casati , t. prosen and d. rossini , epl ( europhysics letters ) * 85 * , 37001 ( 2009 ) .
g. benenti , g. casati , t. prosen , d. rossini and m. nidari , phys .
rev .
b * 80 * , 35110 ( 2009 ) .
t. prosen , phys .
rev .
lett . * 106 * , 217206 ( 2011 ) .
m. nidari , phys .
rev .
lett . * 106 * , 220601 ( 2011 ) .
m. nidari , new j. phys .
* 12 * , 043001 ( 2010 ) .
h .-
p .
breuer and p. petruccione , _ the theory of open quantum systems _
( oxford university press , 2002 ) .
the validity of the master equation description in this regime was established in earlier works @xcite .
b. derrida , phys .
rep . * 301 * , 65 ( 1998 ) .
o. golinelli and k. mallick , j. phys .
a * 39 * , 12679 ( 2006 ) .
d. burgarth and v. giovannetti , phys .
rev .
lett . * 99 * , 100501 ( 2007 ) .
m. zwolak and g. vidal , phys .
rev .
lett . * 93 * , 207205 ( 2004 ) .
s. al - assam , s. r. clark , d. jaksch and tnt development team , tnt library alpha version ( 2012 ) , http://www.tensornetworktheory.org .
m. michel , o. hess , h. wichterich and j. gemmer , phys .
rev .
b * 77 * , 104303 ( 2008 ) . j. wu and m. berciu , phys .
rev .
b * 83 * , 214416 ( 2011 ) .
v. popkov , m. salerno , and g. m. schtz , phys . rev .
e * 85 * , 031137 ( 2012 ) .
simulations at strongly biased drivings @xmath59 and small dephasing rates is extremely demanding due to the formation of an insulating ness central to this work .
this limits the system sizes accessible .
m. nidari , j. stat .
mech .
l05002 ( 2010 ) .
m. nidari , phys .
rev .
e * 83 * , 011108 ( 2011 ) .
t. prosen and m. nidari , j. stat .
mech .
p02035 ( 2009 ) .
m. b. plenio and s. f. huelga , new j. phys .
* 10 * , 113019 ( 2008 ) .
x. zotos , f. naef , and p. prelovek , phys .
rev .
b * 55 * , 11029 ( 1997 ) . t. prosen , phys . rev . lett . * 107 * , 137201 ( 2011 ) .
b. sutherland , _ beautiful models .
70 years of exactly solved quantum many - body problems _ ( world scientific , 2005 ) .
m. takahashi , _ thermodynamics of one - dimensional solvable models _ ( cambridge university press , 1999 ) .
t. prosen and m. nidari , phys .
rev .
lett . * 105 * , 060603 ( 2010 ) .
the transport phenomena outlined here occur irrespective of the sign of @xmath3 , meaning that it can involve highly excited eigenstates of @xmath79 , and are also insensitive to the presence of a uniform potential @xmath134 @xcite , in contrast to the equilibrium transition .
j. j. mendoza - arenas , s. al - assam , s. r. clark , and d. jaksch , _ in preparation_. m. nidari , t. prosen and i. piorn , phys . rev .
a * 78 * , 022103 ( 2008 ) .
integrability of the hamiltonian in eq . can be broken by adding a staggered potential @xmath135 . for @xmath63
this has the effect of turning the system into a diffusive conductor in the gapless regime @xmath55 for any driving , while not affecting the existence of ndc at large drivings for @xmath65 @xcite . in fig .
[ fig4](d ) we confirm , for several field strengths @xmath113 , that dephasing - enhanced transport occurs for @xmath65 only , while for @xmath55 dephasing monotonically decreases the current . * supplementary material : + `` dephasing enhanced transport in non - equilibrium strongly - correlated quantum systems '' * in these appendices we consider in more detail a number of points discussed in the main text .
specifically , in sec .
[ sec : diffusive ] we provide evidence that for a strongly interacting regime @xmath77 at maximal bias @xmath45 the transport becomes diffusive in the large dephasing limit @xmath136 .
this is followed in sec .
[ sec : dark_state ] with an analysis of the structure of driving at @xmath45 between configuration states . using this
we demonstrate that a dark state exists in the limit @xmath77 and we characterise it and the properties of the non - equilibrium stationary state ( ness ) using perturbative arguments . in sec .
[ sec : toy_model ] the essential features of the strongly interacting limit are reduced down into a toy - model which is seen to exhibit a current - driving profile very similar to the full spinless fermion system . the simplicity of the model allows an analytic study of its properties and complements the results given in the main text . finally , in sec .
[ sec : correlations ] the effects of dephasing on the two - point correlations in the full system are outlined . |
since the discovery of the first optical afterglow @xcite associated with a gamma - ray burst ( grb ) , several robotic telescopes have been developed to perform rapid follow - up observations of grb afterglows .
fast robotic follow - ups are now possible because of the improved localization of grbs provided by current gamma - ray satellites ( _ hete - ii _ ,
_ integral _ , _ swift _ , _ interplanetary network _ ) ; real - time grb alerts are received from these satellites through the grb coordinates network ( gcn ) @xcite within a few seconds of the grb onset and provide positions with typical uncertainties of a few arcmin .
observations within minutes of the grb can be crucial to determine the properties of the close circumburst environment , potentially providing clues to the nature of the progenitor . to date
, only a few grbs have measurements of their optical afterglow starting as early as a few minutes after the grb : 990123 , 021004 , 021211 , 030418 , 041006 , 041218 , 041219a , 050319 , 050401 , 050502a , 050525 , 050712 , 050713a , 050730 , 050801 , 050802 , 050820a , 050824 , 050904 , 050908 , 050922c , 051021a ( october 2005 ) .
figure [ fig : lc ] shows the early afterglow light curves for all of them within the first day after the burst .
for each grb we collected unfiltered , @xmath5 , @xmath6 and @xmath7 observations .
the liverpool telescope ( lt ) , owned and operated by liverpool john moores university ( ljmu ) , has a 2-m diameter primary mirror , altitude - azimuth design , final focal ratio f/10 . it was designed and built by telescope technologies limited , with the robotic control system and instrumentation being provided by the ljmu astrophysics research institute @xcite .
it is situated at the observatorio del roque de los muchachos , on the canary island la palma . due to the fully opening enclosure and to the fast slew rate of 2@xmath8/s ,
the telescope can start observations within 13 min of the receipt of the gcn alert notice .
lt has five instrument ports : four folded and one straight - through , selected by a deployable rotating mirror in the ag box within 30 s. the telescope is equipped with the _ ratcam _ optical ccd camera , and with the _ supircam _ 12.5 @xmath9 m camera . a _
prototype spectrograph _ and an optical polarimeter ( _ ringo _ ) based on a design by @xcite are expected to be operating later in 2006 .
a higher resolution , higher sensitivity and generally more capable spectrograph ( _ frodospec _ ) is being developed for deployment in 2007 .
table [ tab : instr ] reports the main characteristics of each instrument .
in addition to the grb follow - up program , which has high priority on the lt , the robotic control @xcite and automated scheduler @xcite allow the lt to conduct an optimized science program of observations of time variable sources over a wide range of timescales from target of opportunity ( e.g. novae , supernovae , anomalies in gravitational lenses ) to longer - term monitoring ( e.g. agn , variable stars ) , when no grb observations are required .
ll optical ccd camera - & 2048@xmath102048 pixels , 0.135``/pixel , fov [email protected] , + & 8 filter selections ( u , g , r , i , z , b , v , h@xmath11 , nd2.0 ) + & - from lt first light , july 2003 + 1 - 2.5 micron camera - & 256@xmath10256 pixels , 0.4''/pixel , fov [email protected] , + ( with imperial college ) & z , j , h , k filters - from late 2005 + - & 49 , 1.7 " fibres , 512@xmath10512 pixels , r=1000 ; + ( with university of manchester ) & 3500 @xmath4 @xmath12 @xmath4 7000 - from 2006 + optical polarimeter & ring polarimeter based on the design of @xcite + & - from 2006 + integral field & r=4000 , 8000 ; + double beam spectrograph - & 4000 @xmath4 @xmath12 @xmath4 9500 - from 2007 + ( with university of southampton ) & +
when a grb fulfills all the requirements to be observed with the lt , the grb over - ride mode is triggered . the grb program takes control of the telescope and triggers the so - called `` detection mode '' ( sec .
[ sec : dm ] ) , during which a small number of sloan @xmath0-band short - exposure images are taken .
these images are then transferred to the lt proxy machine , a network - attached on - site computer . the automatic lt grb pipeline , called _ liverpool telescope transient rapid analysis pipeline _
( hereafter _ lt - trap _ ) , is then invoked to process , analyze and interpret these images and returns a value to indicate whether any variable optical transient ( ot ) candidate has been identified .
the result of this analysis drives the automatic selection of the subsequent observing mode as follows : * identification of a fading ot with @xmath13 @xmath14 multi - color imaging mode ( sec . [ sec : mcim ] ) ; * identification of a fading ot with @xmath15 @xmath14 polarimetry mode ( 2006)/spectroscopy mode ( 2007 ) ( sec .
[ sec : sm ] ) ; * no identification of an ot @xmath14 riz mode ( sec . [ sec : riz ] ) . figure [ fig : strategy ] shows a diagram of the overall observing strategy when the grb over - ride mode is triggered .
the grb target of opportunity control agent ( grb toca ) is a shell script that is invoked in response to a grb alert being received at lt via a socket connection .
this script uses the toca of the lt robotic control system ( rcs ) to take control of the telescope , provided that the grb position is observable and other requirements are fulfilled , e.g. : * the positional error circle is sufficiently small ( error radius @xmath16 arcmin ) ; * altitude above horizon is @xmath17 ; * the position does not match any known x - ray catalog source ; * in the case of _ swift _ alert , the image significance should not be low ( to - date , no real grb was found to have the low image significance flag raised ) ; * none of the five known soft - gamma ray repeaters ( sgrs ) , sgr 1900 + 14 , sgr 1806 - 20 , sgr 1627 - 41 , sgr 0526 - 66 and sgr 1801 - 23 , lies in the error circle . as noted above , the average delay between the arrival time of the gcn alert and the start time of observations ranges between 1 and 3 min . in the case of an sgr lying in the error circle ,
the grb toca triggers a sgr follow - up ir observation with the _ supircam _ according to the strategy defined in an sgr follow - up approved proposal led by the lt grb group . currently , the automatic sgr follow - up program is still under test , so the presence of an sgr in the error circle merely invalidates the grb follow - up observation .
when all of the above conditions are fulfilled , the grb toca repoints the telescope , then triggers the acquisition of the first set of images scheduled in the detection mode ( following commissioning of ringo ( 2006 ) , the detection mode exposures will be preceded by an initial short polarimetric exposure ) .
the _ lt - trap _ is then invoked to analyze these images and consequently choose the most suitable observing strategy ( see sec .
[ sec : dm ] for a detailed description ) .
the ringo exposure will be analyzed offline at a later time and plays no part in real - time decision making ; the motivations of this choice are explained in section [ sec : pm ] .
the grb toca and the _ lt - trap _ communicate through the noticeboard , which is a plain text file continuously updated until the end of observations .
this not only reports in real time a detailed chronological sequence of all the actions and results acquired through , following technical conventions , but also represents a useful real - time human - readable interface for the users responsible for monitoring the on - going observation sequence .
the dm consists of a sequence of two sets of three 10-s exposure images taken in @xmath0 .
these are the first images of the grb field acquired by the telescope .
once each set has been acquired and copied to the proxy machine , the grb toca invokes the grb astrometric fit script on each image .
after object extraction and comparison with standard catalogs , an ot candidates list is extracted for each set of three images .
these ot candidates are then processed by the grb variability script , which aims to identify sources common to each image set and evaluate possible variability .
finally the dm reports the best ot candidate ( if any ) out of the two merged lists .
this shell is responsible for fitting the astrometry and identifying possible ot candidates .
below we report the detailed sequence of operations performed on each single image by this script : 1 . source extraction using a customized version of _ sextractor 2.3.2 _
@xcite is performed ; if the number of sources extracted is lower than 3 , it does not attempt to fit the astrometry .
in contrast , if more than 20 sources are extracted , only the 20 brightest ones are used in the astrometric fit procedure .
a @xmath18 portion of the usno b1.0 astrometric catalog @xcite , centered on the centroid position of the grb , is retrieved from a complete copy stored locally on the lt proxy machine . in case of problems accessing the local catalog ,
the script attempts to retrieve it through the on - line interface of the usno b1.0 .
3 . the retrieved catalog image is then cropped to match exactly the field of view of the lt .
optical extinction for the field is determined from the dust maps by @xcite .
an astrometric fit is performed using an adapted version of the wcs @xcite tool _
imwcs 3.5.3 _ @xcite .
the fit is considered to have failed if either the number of matched sources is lower than 3 or if the fraction of matched sources is lower than 50% .
if the astrometric fit is successful , the pixel - to - sky coordinates transformation is performed through the wcs tool _
xy2sky 3.5.7 _ @xcite , based on the result of the astrometric fit . a typical value for the fit residual is @xmath190.3 arcsec . 7 .
in case of astrometric fit failure , all the image sources are considered in principle as potential ot candidates and they are reported in the ot candidates files , although with a lower confidence level ( see below ) .
the pixel - to - sky coordinate transformation is performed assuming the nominal pointing ( typical accuracy currently of @xmath191020 arcsec , with a target on completion of commissioning of the telescope of 2 arcsec ) . in this case
, the script ends at this point .
selection of the ot candidates : if the astrometric fit is successful , all the image sources are cross - checked with the catalog stars and classified according to a set of criteria .
a 17-bit variable is assigned to each source and describes its properties .
each image source not associated with any usno
b1.0 star is assigned 50 points . each bit of the 17-bit variable
is assigned a given number of points to be added to the initial 50 points credit ( see table [ tab : grbpipe ] ) .
9 . selected classes of non - associated image sources
are a priori excluded from the candidate search . among them , the objects with clearly non - stellar psfs are probably the most important ( e.g. cosmic ray events , reflections , failed deblending ) . a mean stellar psf is defined on each image based on real - time associated objects median statistics .
each selected ot candidate is assigned a confidence level directly related to the total number of points accrued . 11 .
each selected ot candidate is cross - checked with the on - line gsc2.3 and 2mass catalogs and its confidence level is adjusted accordingly ( see table [ tab : grbpipe ] ) . when the astrometric fit is successful , magnitudes are estimated from the usno
b1.0 stars that have been matched .
we use as zero point the median value of the offset needed to have a match between the instrumental magnitude and the @xmath5 magnitude reported in the usno
b1.0 catalog .
the error associated with the zero - point is then evaluated in terms of the clipped ( @xmath20 ) root mean square deviation with respect to the usno b1.0 .
rrllr 1 & 1 & the extracted magnitude is biased by bright neighbours & astrometric & 0 + 2 & 2 & the object has been deblended & astrometric & -10 + 3 & 4 & at least one pixel is saturated ( @xmath21 60000 counts ) & astrometric & 0 + 4 & 8 & object truncated ( close to the image border ) & astrometric & -50 + 5 & 16 & aperture data incomplete or corrupted & astrometric & -10 + 6 & 32 & isophotal data incomplete or corrupted & astrometric & -10 + 7 & 64 & memory overflow occurred during deblending & astrometric & -10 + 8 & 128 & memory overflow during extraction & astrometric & -20 + 9 & 256 & was nt possible to check gsc and/or 2mass & astrometric & -20 + 10 & 512 & a gsc2.3 object lies close ( @xmath4 2 arcsec ) & astrometric & -50 + 11 & 1024 & a 2mass object lies close ( @xmath4 2 arcsec ) & astrometric & -30 + 12 & 2048 & a bright usno star ( r@xmath413 ) is close ( @xmath450px ) & astrometric & -20 + 13 & 4096 & astrometric fit failed assuming nominal pointing & astrometric & -50 + 14 & 8192 & object detected in at least 2 ot lists & variability & + 30 + 15 & 16384 & object not detected in at least 1 ot list & variability & -10 + 16 & 32768 & variability detected over 23 images & variability & + 50 + 17 & 65536 & astrometric fit warning & variability & 0 + after the _ lt - trap _ script described above has run on the three images , the three resulting ot candidates files , each reporting a list of potential ots found in each single image , are cross - checked through the so - called `` variability script '' .
this shell script is invoked by the grb toca and aims to recognize objects common to each of the ot candidates files , as well as to evaluate possible significant magnitude variations between the three images .
this variation is evaluated by performing a @xmath22 test on the zero hypothesis that there is no significant variation . if the three values of magnitude are found to be consistent with a constant value with a significance lower than 1% , the variability bit is raised ( table [ tab : grbpipe ] ) .
each ot candidate confidence level is finally adjusted on the basis of the variability results ( bit number 16 , see table [ tab : grbpipe ] ) .
table [ tab : grbpipe ] reports the meaning of each bit of the 17-bit variable assigned to each ot candidate and the points assigned to each flag .
the confidence level is evaluated by dividing the total points by 100 : when this ratio is @xmath23 , the confidence level is set to zero ; when the ratio is @xmath24 , the level is assigned 1.0 ( highest confidence level possible ) . the final threshold on the confidence level for an ot candidate to trigger the multi - color imaging mode ( sec . [ sec : mcim ] )
is currently set at 0.7 . in case of more than one ot candidate with confidence level above the threshold
, the grb toca accepts that with the highest value .
if the ambiguity is still unresolved , it assumes the first on the list .
however , all the information regarding the other candidates is reported in the noticeboard file as well as in the product files .
the meaning of each single bit described in table [ tab : grbpipe ] is self - explanatory , apart from the last bit , called `` astrometric fit warning '' , which requires an explanation .
this bit is raised whenever at least one of the three images has a failure in the astrometric fit and at least another one has a successful astrometric fit . in this case , the identification of the same object in different images is performed on the basis of pixel coordinates , while the sky coordinates are taken from the image with the correct astrometry .
however , so far this case has turned out to be very rare : either all or none of them are successfully astrometrically fitted .
for this reason , the current points assigned to this bit are zero . to test the sensitivity of the script
, we simulated on real images an ot fading according to a power law , @xmath25 ( where @xmath26 is the grb onset time ) .
the script detected significant variability for @xmath2712 , for objects at least as bright as @xmath281617 , provided that the dm is triggered at @xmath29 s. also , in the case of grb 050502a , where the dm was triggered at @xmath30 s , the _ lt - trap _ detected significant variability : the magnitudes evaluated by the _ lt - trap _ were for the three dm images : @xmath31,@xmath32 and @xmath33 .
the corresponding @xmath22 test gave the following result : @xmath34 , with a significance of @xmath35 .
it must be pointed out that the systematic error affecting the usno
b1.0 magnitudes is around 0.3 mag ; furthermore , it is not possible to correct for color terms at this stage .
however , this is not relevant , as the purpose of the script is to establish relative variations , regardless of the absolute values . to account for possible variations in the zero point
, the script adopts the following approach : it calculates the root mean square deviation of the three zero points and adds it in quadrature to the statistical error of each of the three magnitudes .
after the dm has completed its imaging and analysis sequence , the appropriate observation mode is triggered automatically on the basis of the results of the analysis performed by the _ lt - trap _ on the dm images ( see fig .
[ fig : strategy ] and table [ tab : om ] ) .
this mode is triggered by the grb toca when the dm finds an ot candidate with the required confidence level .
it consists of a multi - color imaging sequence cycling through three sloan @xmath36 , @xmath3 and @xmath0 filters until the end of the first hour .
the ordered sequence , @xmath36 , @xmath3 and @xmath0 , is repeated with different exposure times : the first two times with 30-s exposure , then it continues with 60-s exposure cycles up to 18 times . before may 16 , 2005 , this mode operated in the same way but used a slightly different set of filters , @xmath37 , as used for grb 050502a @xcite .
the motivation for changing and optimizing the filter choice ( post may 16 ) was threefold : 1 ) the greater sensitivity at @xmath36 band , 2 ) the small separation between @xmath38 and @xmath7 filter central wavelengths ( @xmath39 and @xmath40 , respectively ) and 3 ) the recognized need for denser sampling of the light curves at early times . a key unsolved problem in the astrophysics of grbs is the geometry of the ejecta and the role of magnetic fields .
direct measurements of polarization at _ early times _ provide stringent constraints on current jet and internal shock models .
as described in section [ sec : lt ] , we will have unique access to a simple optical polarimeter , ringo , on the lt from the end of 2006 .
polarimetry mode ( pm ) will be triggered following the identification by _ lt - trap _ of a candidate ot with @xmath41 mag and high significance , and will consist of an automatic series of polarimetric exposures taken with ringo , with increasing exposure times : 30s , 60s , 120s , 240s , etc .
these exposures will not be reduced and analyzed in real - time , but post - processing will extract the @xmath0 band light curve and any polarization evolution of the ot .
although less sensitive than ratcam , ringo will detect polarization to a few % in @xmath42 15 mag ots in a 30-s exposure . in combination with the initial polarimetric exposure that will be obtained immediately
before the dm begins , this mode aims to measure the polarization properties from the first minutes after the burst . following commissioning of frodospec ( 2007 )
, the identification of a candidate ot with @xmath41 mag and high significance will result in the triggering of an automatic spectroscopic observation .
the goal of the initial spectroscopic observations will be to obtain an early redshift estimate , whilst subsequent spectra taken during the first hour will provide valuable information on the time evolution of the grb environment and light curve .
short exposure @xmath0 will be interspersed with spectroscopic observations to ensure the ot has not faded below the sensitivity limit of the spectrograph . in the case of a rapidly fading ot , the observing mode will revert to mcim after the initial spectrum has been obtained .
lll dm & 2 x ( 3x10 s in @xmath0 ) & ( 3x10 s in @xmath5 ) + ( 1x10 s in @xmath43 ) + rizm & ( @xmath44 x 2 min ) in ( @xmath0 , @xmath3 , @xmath2 ) , @xmath45 & + mcim & 2 x ( 30 s in @xmath46 ) + @xmath44 x ( 1 min in @xmath46 ) & cycles of @xmath47 : 1x30s , 1x1min , @xmath44x ( 2 min , 3 min ) + sm & for ots with @xmath41 mag ( expected in 2007 ) & @xmath48 mag + pm & 30s , 60s , 120s ... for ots with @xmath41 mag ( expected in 2006 ) &
+ when the dm does not identify an ot candidate , the grb toca invokes the `` riz mode '' .
this consists of a sequence of 2-min exposure @xmath0 , @xmath3 and @xmath2 images : first , one @xmath0 , one @xmath3 and one @xmath2 , then two @xmath0 followed by two @xmath3 and by two @xmath2 , then three @xmath0 followed by three @xmath3 and three @xmath2 and so on . prior to november 2005 , this mode consisted of @xmath49 cycles only ; we have subsequently adapted this mode to include additional @xmath2 exposures that are inserted at the end of each @xmath49 cycle to enable identification of possible higher redshift bursts ( z @xmath50 6 ) that may be faint in @xmath0 but brighter in @xmath2 .
this mode is designed to be sensitive to possible faint optical afterglows ( whether intrinsically faint or at higher redshift ) , allowing @xmath51 band light curves to be constructed when the ot is detectable in longer exposures but too faint to have been identified automatically by the detection mode , or to provide deep upper limits at early time when no ot is present . under average conditions , i.e. seeing of @xmath52 arcsec , 2-min exposure limiting magnitudes are around @xmath532122 , @xmath542021 and @xmath5520 .
the lt is the prime member of the global 2-m robotic telescopes network called `` _
robonet-1.0_. '' this network includes two other facilities : the faulkes telescope north ( ftn ) located in maui ( hawaii ) and the faulkes telescope south ( fts ) in siding spring ( australia ) , both mainly supported by the dill faulkes educational trust and intended for use by uk schools and various scientific observing programs .
all the three telescopes are operated remotely through the operations management centre in liverpool jmu and , when fully commissioned , will all operate in fully robotic mode . under the _ robonet-1.0 _
project , comprising a consortium of ten uk universities funded by the uk pparc , a fraction of time on the two fts is reserved for post first hour grb follow - up by consortium members ( see @xcite for details ) . following the successful operation of _ lt - trap _ on the lt
, it was deployed on the faulkes telescopes , with a slightly - modified observing mode choice following the ot identification stage ( see fig . [
fig : strategy ] ) .
fts is currently undergoing commissioning so is not yet available robotically ; the ftn is fully robotic and has performed a number of follow - up observations minutes after a grb ; for this reason , hereafter we will limit our discussion to ftn .
the rapid ( @xmath56 hr ) robotic follow - up observations of grbs on ftn is a collaborative project between liverpool john moores university and the university of leicester .
the optical filters available on the ft are different from the lt : bessell @xmath57 , sloan @xmath3 , neutral @xmath58 , @xmath59iii and @xmath60 .
the dm consists of only one set of three 10-s images in @xmath5 .
regardless of the results of the dm , three more 10-s images are acquired in the following filters , respectively : @xmath38 , @xmath7 , @xmath3 . in the case of identification of an ot candidate with @xmath48 , the grb toca will triggers a spectroscopic observation . as in the case of lt ,
the threshold on the confidence level for an ot candidate to be considered as a good candidate is currently set to 0.7 .
the logic of the selection of the ot candidate is exactly the same as for the lt ( sec .
[ sec : grb_pipe_var ] ) . presently , regardless of the possible identification of an ot candidate with @xmath61 and confidence level @xmath62 from the dm , the grb toca triggers a sequence of multi - color observations according to the following scheme : cycles of @xmath63 with increasing exposure times : 30 s ( once ) , 60 s ( once ) , then it goes on continuously switching between 120 s and 180 s exposures in order to limit the effects of tracking problems .
although the number of grbs so far observed is still small , we have attempted to estimate the reliability of the ot candidates automatically identified by the _ lt - trap _ as follows . for each of the grbs followed up robotically within minutes , we counted the ot candidates identified as a function of their confidence level .
when the confidence level is above the threshold ( currently set to 0.7 ) , the ot is taken to be a true ot candidate ( toc ) , otherwise it is a fake ot candidate ( foc ) .
when a true afterglow is present , we call it a true ot ( to ) ; similarly , if any non - grb source that could mimic a grb afterglow is present , we call it a fake ot ( fo ) .
let @xmath64 be the probability that a to is automatically identified as a toc .
the closer @xmath64 to 1 , the better the _ lt - trap _ capability of identifying genuine grb afterglows .
likewise , let @xmath65 be the probability that a fo is automatically identified as a foc .
the closer @xmath65 to 1 , the better the _ lt - trap _ capability of rejecting fake grb afterglows . in practice
we do know that @xmath64 and @xmath65 are somehow correlated : a low threshold on the confidence level would increase @xmath64 but also decrease @xmath65 .
the opposite is true for a high value of the same threshold .
so far , the current value ( 0.7 ) has turned out to be a good trade - off . however , as a first approximation we treat them as unrelated and eventually we discuss the opposite case of total correlation .
let @xmath66 and @xmath67 the total number of to observed and the total number of to identified as toc , respectively .
likewise , let @xmath68 and @xmath69 the total number of fo observed and the total number of fo identified as foc , respectively .
the probability @xmath70 ( @xmath71 ) to identify @xmath67 ( @xmath69 ) toc ( foc ) out of @xmath66 to ( @xmath68 fo ) , given @xmath64 ( @xmath65 ) , is given by the binomial distribution ( under the sensible assumption that different grbs are independent from each other ) : @xmath72 we do not know either @xmath64 or @xmath65 , but so far we counted the following : @xmath73 , @xmath74 ; @xmath75 , @xmath76 ( in fact _ lt - trap _ so far correctly identified both @xmath77 ots and failed to reject 2 out of 15 fake ots ) . from the bayes theorem
, we derive the probability @xmath78 to have @xmath79 , given @xmath80 and @xmath81 : @xmath82 we then apply the maximum likelihood method and find the value for @xmath79 which maximizes @xmath78 . to this end ,
the denominator of eq .
[ eq : bayes ] can be ignored . concerning the knowledge of the prior @xmath83 , in both cases as a first guess we may assume the most simple case of a uniform distribution
. then we refine it thanks to the information derived from our tests , according to which a sensible approximation is given by a truncated normal distribution centered on 0.8 and sigma of 0.3 .
however , it must be pointed out that the prior strongly depends on the kind of afterglow population one considers . for this reason
, we consider these two opposite cases of uniform and truncated normal distributions .
when a uniform prior is assumed , the best value is straightforwardly given by @xmath84 : @xmath85 and @xmath86 .
when we assume a more refined prior , i.e. @xmath87 , the best values are @xmath88 ( fig . [
fig : pp ] , solid line ) and @xmath89 .
eventually , we can provide confidence intervals on both @xmath64 and @xmath65 at given levels from the distributions obtained with the truncated normal priors : @xmath90 and @xmath91 ( 50% cl ) ; @xmath92 and @xmath93 ( 90% cl ) . if we take into account the correlation between @xmath64 and @xmath65 , the total probability of eq .
[ eq : bayes ] is replaced by the more general one : @xmath94 at this stage , it is difficult to model the anticorrelation between @xmath64 and @xmath65 in the bivariate prior @xmath95 , particularly for the small sample of grbs so far collected .
we then studied the opposite case of a total anticorrelation between the two variables , in which @xmath65 is completely determined by the value of @xmath64 , by solving eq .
[ eq : anticorr ] : @xmath96 where @xmath97 is the same expression derived in eq .
[ eq : bayes ] , assuming truncated normal priors .
the meaning of eq .
[ eq : anticorr ] is that , for each value @xmath98 , @xmath99 is determined so that the probability that @xmath100 , is equal to the probability that @xmath101 ; it is straightforward to verify that the greater @xmath99 , the smaller @xmath98 and vice versa .
this is just one sensible assumption of total anticorrelation among other possible alternative assumptions .
figure [ fig : pp ] shows the probability density for @xmath64 derived in this case of total anticorrelation ( dashed line ) .
the best values for @xmath64 and @xmath65 and their confidence intervals derived in this case ( truncated normal priors ) are the following : @xmath102 and @xmath103 ( 50% cl ) ; @xmath104 and @xmath105 ( 90% cl ) .
remarkably , in spite of the opposite cases of variables @xmath64 and @xmath65 either totally unrelated or totally anticorrelated , the results seem to give consistent confidence intervals , as shown by fig . [
fig : pp ] .
the wide confidence interval of @xmath64 is understandably due to the low number of to observed so far , rather than the limits of the _ lt - trap_. we expect to give more precise estimates of both probabilities as soon as the sample of grb afterglows detected is more numerous .
to - date ( october 2005 ) , over a period of two months , lt followed up in a fully automated fashion using the real - time pipeline six grbs and ftn followed up five more .
table [ tab : results ] reports for each of these eleven follow - up observations the start time from the grb trigger , the filters used , the number of frames acquired and the gcn circulars issued by the lt / ftn collaboration reporting on the corresponding lt / ftn observations . the most successful case so far is represented by the automatic detection of the afterglow of grb 050502a @xcite followed by the acquisition of the first early ( @xmath56hr ) multi - color light curve @xcite .
this latter case is also notable for longer term follow - up using the _ robonet-1.0 _ network .
the other case of prompt detection is grb 050713a @xcite , although in this case the afterglow was not identified automatically because of the poor quality of the images affected by the presence of a @xmath106 star in the field , but recognized afterward by visual inspection .
llllrlrc ftn & 050412 & _ swift _ & @xmath107 & 2.5 & @xmath108 & 23 & none + lt & 050502a & _ integral _ & @xmath37 & 3.1 & @xmath109 & 25 & 3325 + ftn & 050504 & _ integral _ & @xmath107 & 3.7 & @xmath110 & 38 & 3351 + lt & 050520 & _ integral _ & @xmath49 & 4.5 & @xmath111 & 18 & 3437 + lt & 050528 & _ swift _ & @xmath49 & 2.5 & @xmath112 & 36 & 3497 + lt & 050713a & _ swift _ & @xmath0 & 2.4 & @xmath113 & 3 & 3588 + ftn & 050713b & _ swift _ & @xmath5 & 3.3 & @xmath114 & 6 & 3592 + ftn & 050716 & _ swift _ & @xmath107 & 3.8 & @xmath115 & 20 & 3625 + lt & 050730 & _ swift _ & @xmath49 & 50 & @xmath116 & 36 & 3706 + lt & 050904 & _ swift _ & @xmath49 & 3.8 & & 31 & none + ftn & 050925 & _ swift _ & @xmath107 & 3.3 & @xmath117 & 12 & 4035 + the power and robustness of the _ lt - trap _ is illustrated in the case of grb 050730 @xcite : this grb occurred before twilight at the lt site , so was triggered manually and , although the ot was not recognized during the dm because of the high sky background due to the sky not yet being completely dark , it was automatically identified in the subsequent rim as an unknown fading source of @xmath118 at @xmath119 min after the grb . in the other cases , the combination of sensitivity and rapid response made it possible to derive deep upper limits , as in the cases of grb 050504 with @xmath110 @xcite and grb 050716 with @xmath115 @xcite 34 min after the grb trigger time . in the future ,
we plan to refine the strategy when an ot candidate with @xmath61 and confidence level @xmath62 is found from the dm ; in particular , exposure times will be tuned dynamically on the basis of the estimated magnitude of the best ot candidate .
automatic polarimetry , near - infrared and spectroscopy follow - up modes will be implemented on the lt following completion of instrument commissioning .
cg and ag acknowledge their marie curie fellowships from the european commission .
cgm acknowledges financial support from the royal society .
am acknowledges financial support from the uk pparc .
mfb is supported by a pparc senior fellowship .
the liverpool telescope is operated on the island of la palma by liverpool john moores university at the observatorio del roque de los muchachos of the instituto de astrofisica de canarias .
the faulkes telescope north is operated with support from the dill faulkes educational trust .
gomboc , a. , et al . , 2005a , proc .
of the 4th workshop gamma
ray bursts in the afterglow era , rome , 2004 , eds : piro , l. , amati , l. , covino , s. , & gendre , b. , il nuovo cimento , in press ( preprint : astro - ph/0502505 ) | the 2-m liverpool telescope ( lt ) , owned by liverpool john moores university , is located in la palma ( canary islands ) and operates in fully robotic mode . in 2005 ,
the lt began conducting an automatic grb follow - up program .
on receiving an automatic grb alert from a gamma - ray observatory ( _ swift _ , _ integral _ , _ hete - ii _ , _ ipn _ )
the lt initiates a special override mode that conducts follow - up observations within 23 min of the grb onset .
this follow - up procedure begins with an initial sequence of short ( 10-s ) exposures acquired through an @xmath0 band filter .
these images are reduced , analyzed and interpreted automatically using pipeline software developed by our team called `` _ lt - trap _ '' ( _ liverpool telescope transient rapid analysis pipeline _ ) ; the automatic detection and successful identification of an unknown and potentially fading optical transient triggers a subsequent multi - color imaging sequence . in the case of a candidate brighter than @xmath1 , either a polarimetric ( from 2006 ) or a spectroscopic observation ( from 2007 ) will be triggered on the lt . if no candidate is identified , the telescope continues to obtain @xmath2 , @xmath0 and @xmath3 band imaging with increasingly longer exposure times .
here we present a detailed description of the _ lt - trap _ and briefly discuss the illustrative case of the afterglow of grb 050502a , whose automatic identification by the lt just 3 min after the grb , led to the acquisition of the first early - time ( @xmath41 hr ) multi - color light curve of a grb afterglow . |
in discussions of f - term @xmath1 supersymmetry ( susy ) breaking , r - symmetry plays a crucial role .
importance of a u(1 ) r - symmetry in spontaneous susy breaking was clearly addressed in ref .
it was shown that , `` a continuous r - symmetry is a necessary condition for spontaneous supersymmetry breaking and a spontaneously broken r - symmetry is a sufficient condition , in models where the gauge dynamics can be integrated out and in which the effective superpotential is a generic function consistent with the symmetries of the theory . ''
however , in this paper , we are going to show that there exist a large number of generic and renormalizable wess - zumino(wz ) models where existence of a @xmath0 r - symmetry is sufficient to break susy spontaneously .
it is well known that if there is no r - symmetry , continuous or discrete , in wz models with canonical khler potentials , then the global minima preserve susy @xcite . with the help of the above two results we can conclude that existence of a @xmath0 r - symmetry is a necessary condition for breaking in a global minimum for generic and renormalizable wz models .
models of f - term susy breaking with a @xmath0 r - symmetry can be obtained most simply from the models of f - term susy breaking with u(1 ) r - symmetry by adding a completely different and decoupled sector to break u(1 ) r - symmetry explicitly down to a @xmath0 r - symmetry . in sec [
s : deformor ] , we illustrate this idea by adding some more terms to the famous oraifeartaigh @xcite ( or ) model
. however these models are trivial . in sec .
[ s : nontrivial ] , we explicitly discuss a non - trivial model where the superpotential can not be broken into such two non - interacting parts . we have also shown that this model has a vacuum where susy as well as discrete r - symmetry is spontaneously broken for large regions of parameter space .
we then give some variations of this model by adding more fields . in sec .
[ s : serises ] , we identify the form of these models , and then , three series of models are given where each series contains a large number of such models .
a note about notations . for u(1 )
r - symmetry we use @xmath2 to denote the u(1 ) r - charge of any chiral scalar superfield @xmath3 in the normalization where the r - charge of @xmath4 is 1 .
we use @xmath5 to denote @xmath0 r - charge of the superspace co - ordinate @xmath4
. discrete r - charge of pseudo - moduli superfield @xmath6 is @xmath7 and that of any other superfield is just the subscript of that field .
adding a completely different and decoupled sector to any one of f - term susy breaking models with u(1 ) r - symmetry ( we will call it as old sector ) , we can break u(1 ) r - symmetry to a @xmath0 r - symmetry explicitly .
value of @xmath8 is controlled by the new sector .
if we find a new sector ( or in other words , a @xmath0 r - symmetry ) for which no new term to the old sector is allowed even though r - symmetry becomes weaker , then susy breaking conditions coming from the old sector will not alter . in this way , we get models of f - term susy breaking with a discrete r - symmetry in generic theories . to illustrate our idea , we consider the famous or model as an example of the old sector : @xmath9 the above superpotential is generic with a u(1 ) r - symmetry where @xmath10 and a @xmath11 internal symmetry under which @xmath6 transforms trivially whereas @xmath3 s transform non - trivially .
let s now add five new fields @xmath12 to the old sector and get the following superpotential .
@xmath13 this superpotential does not have a u(1 ) r - symmetry , as can be easily checked .
however , spontaneous susy breaking still occurs because the f - terms , @xmath14 and @xmath15 , are not changed due to the inclusion of the new terms .
the above superpotential is generic with the following three symmetries . 1 . a @xmath16 r - symmetry with @xmath17 .
2 . a @xmath11 internal symmetry under which all the @xmath18 s transform non - trivially whereas @xmath6 and @xmath19 s transform trivially .
3 . a @xmath20 internal symmetry under which all the @xmath19 s transform as @xmath21 whereas the remaining fields are invariant . we can get different variations of the above model easily , for a @xmath0 r - symmetry with @xmath22 and @xmath17 as follows .
@xmath23 where @xmath24 when @xmath25 .
thus we have proved that there exists a large number of generic ( trivial ) models where existence of a @xmath0 r - symmetry is sufficient to break susy via f - terms .
one can also use this technique to other susy breaking models @xcite with @xmath26 r - symmetry .
one side comment about these models .
old and new sectors of any one of these models can communicate to each other through gauge interactions if we gauge some of the internal symmetries .
for example , we can easily promote the fields @xmath3 and @xmath27 to transform under adjoint representation and promote the field @xmath6 to remain invariant under a gauge group , without forbidding any term of the old and new sectors .
however , these models will then no longer be wz models
. now , we can ask whether it is possible to construct generic f - term susy breaking models with the following charateristics : ( @xmath28 ) there is no u(1 ) r - symmetry in the superpotential ; ( @xmath29 ) the superpotential can not be subdivided into two disjoint sectors / parts . in the rest of the paper , we show examples of models with all these characteristics .
we consider a renormalizable wz model with @xmath30 r - symmetry and @xmath31 .
we also consider that other than @xmath6 , there are @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 and @xmath38 fields in the theory .
so we will have the following generic superpotential .
@xmath39 where , without loss of generality we can take all the parameters , except @xmath40 , to be real and positive .
the above superpotential does not have a u(1 ) r - symmetry . with @xmath41 ,
there is a u(1 ) r - symmetry with the following r - charge assignments .
@xmath42 but then the superpotential is not generic . for @xmath43 , this u(1 )
r - symmetry is explicitly broken .
there is spontaneous susy breaking .
this can be easily realized by observing the following f - terms , @xmath44 notice that vacuum expectation values ( vevs ) of @xmath14 and @xmath45 terms can not be simultaneously zero .
we can vanish all other f - terms for any value of @xmath46 ( vev of @xmath38 ) by choosing appropriate vevs of other fields .
now minimum of scalar potential depends on @xmath46 . like the oraifeartaigh model @xcite we have two cases , ( a ) for @xmath47 ,
minimum is at @xmath48 , whereas ( b ) for @xmath49 , minimum is at @xmath50 .
hence we have a vacuum where supersymmetry as well as discrete r - symmetry get spontaneously broken .
tree level scalar potentials of susy breaking often have flat directions @xcite .
for example , for the case of @xmath51 , minimum of tree level potential is independent of @xmath52 and @xmath53 .
so , it is necessary to calculate 1-loop correction to check whether these flat directions are lifted or not .
one loop correction is given by coleman - weinberg ( cw ) @xcite potential , @xmath54 where @xmath55 and @xmath56 are mass matrices for scalar and fermion fields .
non - zero eigenvalues @xmath57 and @xmath58 of @xmath59 and @xmath60 respectively for @xmath51 are follows .
@xmath61 where @xmath62 , @xmath63 and @xmath64 denote @xmath65 .
putting these eigenvalues to and expanding @xmath66 about @xmath67 , we find @xmath68 constants @xmath69 and @xmath70 are positive and hence after addition of 1-loop correction , total scalar potential have a local minimum at @xmath71 .
we can get different variations of the above model by adding more @xmath3 fields in the theory .
for example , we can add any number of fields from the list @xmath72 . in this way
we get 31 more models .
if we add all the fields from the list , then the superpotential takes the following form @xmath73 note that addition of these fields do not change @xmath74 and @xmath75 and hence there is susy breaking .
in this section we are going to show that there exists a large number of non - trivial models where existence of a discrete r - symmetry is sufficient to break supersymmetry .
we consider the superpotentials are of the following form with a @xmath76 r - symmetry and @xmath77 .
@xmath78 where @xmath79 are natural numbers , @xmath80 contains cubic terms which are independent of @xmath81 , and @xmath82 denotes quadratic terms for @xmath3 fields other than @xmath83 fields .
note that the superpotentials of the previous section are of the above form with @xmath84 and @xmath85 . due to this form of superpotentials ,
susy gets spontaneously broken .
one can show this by observing the following f - terms @xmath86 above equations are the same as for @xmath84 and @xmath85 . in this series of models
@xmath87 is multiple of four and @xmath85 i.e. superpotentials have a @xmath88 r - symmetry with @xmath89 .
field content of this series for any @xmath87 is given below .
@xmath90 note that the model with @xmath84 have the same r - symmetry and @xmath5 as the models given in the previous section .
but this model is different from those models because it has different field content .
to show that there is no u(1 ) r - symmetry in this series of models , we use method of contradiction .
if there were a u(1 ) r - symmetry in any model , existence of terms @xmath91 , @xmath92 , and @xmath93 would imply that @xmath94 , @xmath95 and @xmath96 . from the terms
@xmath97 , @xmath98 and @xmath99 , we could conclude @xmath100 , @xmath101 and @xmath102 .
similarly we could construct a r - charge assignment chain for other fields as shown in fig .
[ f:6kplus2 ] .
now , for @xmath84 we have @xmath103 . but according to the chain ( at the point a ) @xmath104 . hence the superpotential for @xmath84
do not have a u(1 ) r - symmetry . moving down the chain
, one can easily show that there is no u(1 ) r - symmetry in any model of this series .
( 200,50)(120,-15 ) ( 21,0)[l]@xmath105 ( 21,-8)[lt](@xmath106 ) ( 48,0)(68,0 ) ( 58,0)(58,15 ) ( 58,16)[b]@xmath105 ( 69,0)[l]@xmath107 ( 69,-8)[lt](@xmath108 ) * a * ( 98,0)(118,0 ) ( 108,0)(108,15 ) ( 108,16)[b]@xmath109 ( 119,0)[l]@xmath110 ( 119,-8)[lt](@xmath111 ) ( 148,0)(168,0 ) ( 158,0)(158,15 ) ( 158,16)[b]@xmath110 ( 169,0)[l]@xmath112 ( 169,-8)[lt](@xmath113 ) * b * ( 198,0)(218,0 ) ( 208,0)(208,15 ) ( 208,16)[b]@xmath109 ( 219,-8)[lt](@xmath114 ) ( 219,0)[l]@xmath115 ( 248,0)(268,0 ) ( 258,0)(258,15 ) ( 258,16)[b]@xmath110 ( 269,0)[l]@xmath116 ( 269,-8)[lt](@xmath117 ) * c * ( 298,0)(318,0 ) ( 308,0)(308,15 ) ( 308,16)[b]@xmath109 ( 319,-8)[lt](@xmath118 ) ( 319,0)[l]@xmath119 ( 348,0)(368,0 ) ( 358,0)(358,15 ) ( 358,16)[b]@xmath110 ( 369,0)[l]@xmath120 ( 369,-8)[lt](@xmath121 ) * d * ( 403,0)[l]@xmath122 we are now going to prove that all the generic and renormalizable superpotentials of this series of models are of the form as given in . there is no term quadratic in @xmath6 in the superpotentials because field content for any @xmath87 does not contain the field @xmath123 .
also , cubic term in @xmath6 is not allowed by discrete r - symmetries .
similarly , one can show that the terms @xmath124 and @xmath125 are also not allowed . for a cubic term of the form @xmath126 to exist , we need @xmath127 . without loss of generality
, we can take @xmath128 . from field content given in , we find @xmath129 and hence @xmath130
. also one of them must be odd since their sum is odd .
the only field having odd discrete r - charge in this range is @xmath131 .
but field content of any model does not contain the field @xmath132 and so , @xmath133-terms are independent of the field @xmath134 .
there is only one cubic term containing @xmath6 , @xmath135 , because discrete r - charges of the @xmath3 fields lie between @xmath87 and @xmath136 .
thus superpotentials of this series are of form as given in and hence there is f - term susy breaking .
let s now give superpotential for @xmath84 .
@xmath137 one can explicitly verify that the above superpotential does not have a u(1 ) r - symmetry yet there is f - term susy breaking .
superpotentials of this series have @xmath138 r - symmetry where @xmath87 is multiple of 6 with starting value 12
. discrete r - charge of superspace co - ordinate @xmath4 is @xmath139 or @xmath140 .
field content for any @xmath87 is given below .
@xmath141 to show that there is no u(1 ) r - symmetry , we have taken same the strategy as of the earlier case .
let s first tabulate u(1 ) r - charges of first twelve fields from the above list .
@xmath142 these fields are common to all models of this series .
u(1 ) r - charges for the fields @xmath143 and @xmath144 can be derived easily .
we obtained r - charges for @xmath145 from the terms @xmath146 , @xmath147 , @xmath148 , @xmath149 and @xmath150 respectively .
u(1 ) r - charge assignment chain for rest of the fields is given in the fig .
[ f:6kplus4 ] . from
these given information , one can easily show that there is no u(1 ) r - symmetry in any model of this series .
( 200,50)(180,-15 ) ( 119,0)[l]@xmath151 ( 119,-8)[lt](@xmath152 ) ( 148,0)(168,0 ) ( 158,0)(158,15 ) ( 158,16)[b]@xmath151 ( 169,0)[l]@xmath116 ( 169,-8)[lt](@xmath153 ) * a * ( 198,0)(218,0 ) ( 208,0)(208,15 ) ( 208,16)[b]@xmath109 ( 219,-8)[lt](@xmath154 ) ( 219,0)[l]@xmath119 ( 248,0)(268,0 ) ( 258,0)(258,15 )
( 258,16)[b]@xmath151 ( 269,0)[l]@xmath155 ( 269,-8)[lt](@xmath156 ) * b * ( 298,0)(318,0 ) ( 308,0)(308,15 ) ( 308,16)[b]@xmath109 ( 319,-8)[lt](@xmath157 ) ( 319,0)[l]@xmath158 ( 348,0)(368,0 ) ( 358,0)(358,15 ) ( 358,16)[b]@xmath151 ( 369,0)[l]@xmath159 ( 369,-8)[lt](@xmath117 ) * c * ( 403,0)[l]@xmath122 there will be a @xmath160-term only if @xmath161 . without loss of generality , we can take @xmath162 .
minimum value for @xmath163 is @xmath87 . as there is no field with r - charge @xmath164 , @xmath163
can not be equals to @xmath87 . next higher value of discrete r - charge is @xmath165 and hence @xmath166 . in this range
discrete r - charges of the fields are multiple of 6 .
as @xmath167 is not multiple of 6 , there can not a @xmath133-term for @xmath168 .
so , superpotentials of this series also of the form as given in and which in turn guarantee spontaneous breakdown of susy .
let s give first model of this series .
@xmath169 one can explicitly verify that there is no u(1 ) r - symmetry and susy is spontaneously broken .
superpotentials of this series have @xmath170 r - symmetry with @xmath171 and @xmath172 .
field content for any @xmath87 is given below .
@xmath173 if we demand that the superpotentials have an u(1 ) r - symmetry , then we will have a table ( ) and a chain ( fig .
( [ f:6kplus6 ] ) ) of r - charge assignments . from these inputs
one can conclude that there is no u(1 ) r - symmetry in any model of this series .
@xmath174 ( 200,85)(120,-15 ) ( 48,50)(66,50 ) ( 47,50)[r]@xmath175 ( 58,30)(58,50 ) ( 67,50)[l]@xmath176 ( 47,0)[r]@xmath105 ( 21,-8)[lt](0 ) ( 48,0)(68,0 ) ( 58,0)(58,20 ) ( 63,20)[b]@xmath105 ( 69,0)[l]@xmath107 ( 69,-8)[lt](2 ) ( 98,0)(118,0 ) ( 108,0)(108,20 ) ( 108,20)[b]@xmath107 ( 119,0)[l]@xmath115 ( 119,-8)[lt](@xmath177)*a * ( 148,0)(168,0 ) ( 158,0)(158,20 ) ( 158,20)[b]@xmath105 ( 169,0)[l]@xmath178 ( 169,-8)[lt](4 ) * b * ( 198,0)(218,0 ) ( 208,0)(208,20 ) ( 208,20)[b]@xmath109 ( 219,-8)[lt](@xmath179 ) ( 219,0)[l]@xmath180 ( 248,0)(268,0 ) ( 258,0)(258,20 ) ( 258,20)[b]@xmath105 ( 269,0)[l]@xmath116 ( 269,-8)[lt](@xmath108 ) * c * ( 298,0)(318,0 ) ( 308,0)(308,20 ) ( 308,20)[b]@xmath109 ( 319,-8)[lt](@xmath111 ) ( 319,0)[l]@xmath119 ( 348,0)(368,0 ) ( 358,0)(358,20 ) ( 358,20)[b]@xmath105 ( 369,0)[l]@xmath181 ( 369,-8)[lt](@xmath182 ) * d * ( 403,0)[l]@xmath122 ( 95,50)(115,50 ) ( 115,50)[l]@xmath183 ( 145,50)(165,50 ) ( 165,50)[l]@xmath184 ( 155,50)(155,65 ) ( 155,65)[b]@xmath185 ( 195,50)(215,50 ) ( 215,50)[l]@xmath186 ( 205,50)(205,65 ) ( 205,65)[b]@xmath187 there will be a @xmath133-term for @xmath188 only if @xmath189 where @xmath190 are discrete r - charges of fields coupled to it .
thus @xmath191 for first case @xmath192 . as there is no field with odd discrete r - charge in this range ,
this possibility is ruled out .
one can check that second possibility is also ruled out .
hence superpotentials of this series are also of the form given in and there are f - term susy breaking .
let s give first model of this series , i.e. for @xmath193 , so that one can verify non - existence of a u(1 ) r - symmetry and spontaneous breakdown of susy .
@xmath194 in the above we have given only three series of models .
however one can construct many series of such models .
we have shown that there exists a large number of generic and renormalizable wess - zumino models where existence of a @xmath0 r - symmetry is sufficient to break susy spontaneously . and it is well known that if there is no r - symmetry in wz models with canonical khler potential , then global minima always preserve susy .
so , existence of a @xmath0 r - symmetry in a generic and renormalizable wess - zumino model is a necessary condition for f - term susy breaking .
however , for even @xmath8 with @xmath195 , one can not have models of susy breaking because for these cases superpotentials as a whole transform trivially and terms which are allowed or forbidden by these r - symmetries can always be reproduced by some internal symmetries . | we have shown that in a large number of generic and renormalizable wess - zumino models existence of a @xmath0 r - symmetry is sufficient to break supersymmetry spontaneously .
this implies that existence of a @xmath0 r - symmetry is a necessary condition for supersymmetry breaking in generic and renormalizable wess - zumino models . |
the consensus of the astrophysics community is that a large fraction of the universe s mass consists of non - luminous , non - baryonic material , known as dark matter @xcite .
although the nature of this substance or substances remains unknown , weakly interacting massive particles ( wimps ) represent a particularly attractive and well motivated class of possibilities .
although the most studied wimp candidate is the lightest neutralino @xcite in supersymmetric models , many other possibilities have also been proposed , including kaluza - klein states in models with universal @xcite or warped @xcite extra dimensions , stable states in little higgs theories @xcite , and many others . in each of the above mentioned cases
, many new particle species , in addition to the wimp itself , are expected to lie within the discovery reach of the large hadron collider ( lhc ) , making the task of deducing the nature of the wimp immeasurably simpler . in supersymmetry , for example , gluinos and squarks
are expected to be produced prolifically . by studying the cascades produced in the decays of such particles , the masses of several superparticle masses , including the lightest neutralino ,
are likely to be determined
. if squarks , gluinos , and other additional superpartners are too heavy to be produced , however , the lightest neutralino will also be very difficult to study at the lhc , even if rather light itself .
more generally speaking , in the absence of heavier particles with shared quantum numbers , wimps will not be easily detected or studied at the lhc .
although an electroweak scale , cold thermal relic particle , if it exists , would almost certainly be produced at the lhc , identifying and characterizing the nature of the wimp simply from missing energy studies is a daunting , perhaps impossible , task @xcite .
although the usual list of prospective wimps mentioned above contains some very attractive and well - motivated candidates for dark matter , there are certainly many possible forms of dark matter that have not yet been considered .
as the first observations of particle dark matter might well come from direct and/or indirect detection experiments , it is possible that these results may be misinterpreted as a result of theoretical bias , anticipating dark matter to have the properties of a neutralino or other often studied candidates . to avoid such confusion ,
model - independent studies of dark matter phenomenology can play an important role ( for previous work in this direction , see refs.@xcite ) . in this article , rather than consider a wimp candidate from a specific theoretical model , we study model - independent wimps with different combinations of spins and interaction forms with standard model particles .
these interactions are limited only by the requirements of lorentz invariance and a consequent wimp abundance consistent with cosmological observations . for each spin and interaction form
, we evaluate the constraints from and prospects for direct and indirect detection of wimps in current and future experiments .
although we will be forced to adopt some assumptions in order to make the problems at hand tractable , we attempt to be as general as possible throughout our study . beyond the starting point
that the dark matter is a wimp in the form of a single species of a cold thermal relic , we adopt only two assumptions : 1 .
any new particle species in addition to the wimp has a mass much larger than the wimp .
the wimp interactions with standard model particles are dominated by those of one form ( scalar , vector , _ etc . _ ) .
an implication of the first assumption is that the wimp s thermal abundance is not affected by resonances or coannihilations . at a later stage of this paper
, we will discuss the impact of relaxing these assumptions .
the remainder of this paper is structured as follows . in sec.[fermion ] we explore the phenomenology of a generic fermionic wimp , including its annihilation cross section and relic abundance , elastic scattering cross section and direct detection prospects , and indirect detection prospects in the form of a neutrino flux from the sun and gamma rays and charged particles produced in galactic annihilations . in sec .
[ scalar ] , we repeat this exercise for the case of a scalar wimp . in each of these two sections ,
we also consider dark matter candidates from specific particle physics frameworks and discuss how they fit into our model - independent analysis . in sec .
[ conclusion ] we summarize our results and present our conclusions .
we begin with the case of a fermionic wimp , and study five types of interactions consistent with the requirement of lorentz invariance . as mentioned in the introduction , we assume that the wimp is the only new particle at the electroweak scale .
this enables us to describe the interaction between wimps and standard model fermions in terms of an effective field theory , in which we keep only the first term in the @xmath0 expansion of the heavy propagator term ( here , @xmath1 and @xmath2 are the momentum and mass of the propagator , respectively ) .
we note that the effective interaction lagrangians are not invariant under the standard model @xmath3 gauge symmetry ; however , this is acceptable as our theory need only be valid at energy scales below the scale of electroweak symmetry breaking . to begin , we only consider wimp annihilations to fermion - antifermion pairs , neglecting for the moment the possibility of final states that include gauge or higgs bosons .
in particular , we consider the following interaction lagrangians between two fermionic wimps ( @xmath4 ) and two standard model fermions ( @xmath5 ) : @xmath6 we will now proceed to calculate the annihilation cross section , relic density , and elastic scattering cross sections for a fermionic wimp . in each of the cases listed above
, we are interested in determining the cosmological density of wimps produced in the early universe .
the first step is to calculate the annihilation cross sections to fermion - antifermion pairs as a function of the mandelstam variable @xmath7 for each of the five cases .
the result is @xmath8 \label{sigmas}\\ \sigma_p & = & \frac{1}{32\pi } \
\sum_f g_{p , f}^2 \ , c_f \sqrt{\frac{s-4 m_f^2}{s-4 m_\chi^2 } } \ s \\ \sigma_v & = & \frac{1}{32\pi } \
\sum_f g_{v , f}^2 \ , c_f \sqrt{\frac{s-4 m_f^2}{s-4 m_\chi^2 } } \ \left[s+4m_\chi^2 + \frac{(s-4 m_\chi^2)\ ( s-4m^2_f)}{3\ , s } + 4m^2_f \right ] \\ \sigma_a & = & \frac{1}{32\pi } \
\sum_f g_{a , f}^2 \ , c_f \sqrt{\frac{s-4 m_f^2}{s-4 m_\chi^2 } } \ \left[s-4m_\chi^2 + \frac{(s-4 m_\chi^2)\ ( s-4m^2_f)}{3 \ , s } + 4m^2_f \right ] \\
\sigma_t & = & \frac{1}{32\pi } \ \sum_f g_{t , f}^2 \ , c_f \sqrt{\frac{s-4m_f^2}{s-4m_{\chi}^2 } } \ \left[6s + 4m_{\chi}^2 + \frac{8(s-4m_{\chi}^2)(s-4m_f^2)}{3\ , s } + 4m_f^2 \right ] \label{sigmaaa } \end{aligned}\ ] ] where the sum is over the final state fermion species and @xmath9 are the color factors , equal to 3 for quarks and 1 for leptons . to determine the density of relic wimps , we solve the boltzmann equation : @xmath10,\ ] ] where @xmath11 is the hubble rate and @xmath12 is the thermally averaged wimp annihilation cross section @xcite .
in thermal equilibrium , the number density of wimps is given by : @xmath13 where @xmath14 is the number of degrees of freedom of a fermionic wimp . at @xmath15 ,
the number density of wimps was very close to its equilibrium value and nearly as abundant as any other particle .
as the temperature dropped below @xmath16 , however , the number density was exponentially suppressed until , eventually , the annihilation and production rates became much smaller than the expansion rate , and the species froze out of equilibrium . since we are considering cold thermal relics ,
freeze - out occurred when wimps were non - relativistic and had velocities much smaller than unity . substituting @xmath17 to eqs .
( [ sigmas]-[sigmaaa ] ) , and expanding in powers of the relative velocity between two annihilating wimps up to order @xmath18 , we find @xmath19 \label{svs } \\
\sigma_p \vert v \vert & \approx & \frac{1}{4\pi } \sum_f g_{p , f}^2 \ , c_f \ m_\chi^2 \
\sqrt{1-m_f^2/m_\chi^2 } \ \\
\sigma_v \vert v \vert & \approx & \frac{1}{4\pi } \sum_f g_{v , f}^2 \ , c_f \
\sqrt{1-m_f^2/m_\chi^2 } \
\left [ \left ( 2 + \frac{m_f^2}{m_\chi^2}\right ) + \frac{1}{12}\left ( 1 - \frac{m_f^2}{m_\chi^2}\right ) v^2 \right ] \\
\sigma_a \vert v \vert & \approx & \frac{1}{4\pi } \sum_f g_{a , f}^2 \ , c_f \ m_\chi^2 \
\sqrt{1-m_f^2/m_\chi^2 } \
\left [ \frac{m_f^2}{m_\chi^2 } + \frac{1}{12 } \left(2-\frac{m_f^2}{m_\chi^2 } \right)v^2 \right ] \\
\sigma_t \vert v \vert & \approx & \frac{1}{4\pi } \sum_f g_{t , f}^2 \
c_f \ m_{\chi}^2 \
\sqrt{1-m_f^2/m_\chi^2 } \
\left [ \left ( 7 + \frac{m_f^2}{m_\chi^2}\right ) + \frac{2}{3}\left ( 1 - \frac{m_f^2}{m_\chi^2}\right ) v^2 \right ] \label{svt } .
\end{aligned}\ ] ] numerical solutions of the boltzmann equation yield a relic density of @xcite : @xmath20 where @xmath21 , @xmath22 is the temperature at freeze - out , @xmath23 is the number of relativistic degrees of freedom available at freeze - out ( @xmath24 for a freeze - out temperature between the bottom quark and @xmath25 boson masses ) , and @xmath26 and @xmath27 are terms in the partial wave expansion of the wimp annihilation cross section , @xmath28 , \label{xf}\ ] ] where @xmath29 is an order unity parameter determined numerically .
wimps with electroweak - scale masses and couplings generically freeze out at temperatures in the range of approximately @xmath30 20 to 30 .
, @xmath31 , @xmath32 , @xmath33 , and @xmath34gev@xmath35 . in the remaining four frames ,
results are shown for @xmath36 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 gev@xmath35 . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] , @xmath31 , @xmath32 , @xmath33 , and @xmath34gev@xmath35 . in the remaining four frames , results are shown for @xmath36 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 gev@xmath35 . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] , @xmath31 , @xmath32 , @xmath33 , and @xmath34gev@xmath35 . in the remaining four frames ,
results are shown for @xmath36 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 gev@xmath35 . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] , @xmath31 , @xmath32 , @xmath33 , and @xmath34gev@xmath35 . in the remaining four frames ,
results are shown for @xmath36 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 gev@xmath35 . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] , @xmath31 , @xmath32 , @xmath33 , and @xmath34gev@xmath35 . in the remaining four frames , results are shown for @xmath36 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 gev@xmath35 . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here . also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] , @xmath31 , @xmath32 , @xmath33 , and @xmath34gev@xmath35 . in the remaining four frames , results are shown for @xmath36 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 gev@xmath35 . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here . also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] in the absence of resonances and coannihilations @xcite , an annihilation cross section of @xmath37 @xmath38/s @xmath39 pb is required ( at the temperature of freeze - out , @xmath40 ) to obtain a relic abundance in agreement with the dark matter abundance measured by wmap , @xmath41 @xcite . although the annihilation cross section in the low - velocity limit ( relevant to indirect dark matter searches ) is not much lower than this value in many models , it can be considerably suppressed at low velocities if terms in the annihilation cross section proportional to @xmath18 dominate the cross section ( _ i.e. , _ if @xmath42 ) . furthermore , if the depletion of wimps in the early universe occurs through resonance channels or via coannihilations with other states , the low velocity annihilation cross section can be considerably lower than the value at freeze - out . for more details regarding the relic density calculation ,
see refs .
@xcite . in fig .
[ fermionrelic ] , we show the thermal relic density of a fermionic dark matter candidate with scalar , pseudoscalar , vector , and axial interactions .
we do not include a separate figure for the tensor case , since its annihilation cross section is nearly identical to the vector case .
as discussed in the introduction , these results were found under the assumptions that a given wimp s interactions are dominated by those of only one form ( scalar , vector , _ etc . _ ) , that the wimp s interactions are mediated by particles much heavier than the wimp mass ( thus avoiding the possibility of resonance effects ) , and that the wimp is considerably lighter than any other new particles ( thus making coannihilations unimportant ) .
also , we include only annihilations to fermion - antifermion pairs ( neglecting the possibility of final states including gauge or higgs bosons ) .
in each frame of fig .
[ fermionrelic ] , we show the relic density for various values of the effective couplings . in the upper left and upper right frames ,
we show results for couplings of @xmath43 gev@xmath35 .
this proportionality of the couplings to the fermion mass is predicted for yukawa couplings of a higgs mediated interaction , for example . in the remaining four frames ,
we show results for the case of universal couplings , @xmath44 gev@xmath35 .
if any of our assumptions are broken , the resulting thermal relic abundance will be altered as well .
in particular , resonances ( or more generally , a departure from @xmath45 ) or coannihilations could potentially reduce the abundances shown in fig.[fermionrelic ] considerably . additionally , annihilations to final states such as gauge or higgs bosons , if significant , could also reduce the relic density . the effective couplings described in fig .
[ fermionrelic ] that lead to the correct relic abundance , therefore , can be thought of as approximate maximal values allowed for a thermal wimp .
smaller couplings are possible if appropriate departures are made from our set of assumptions .
although only weakly coupled to baryons , wimps can occasionally scatter elastically with atomic nuclei , providing the potential for detection .
direct detection experiments attempt to measure the recoil energies of nuclei resulting from such interactions .
the interactions leading to the elastic scattering of wimps with nuclei can be classified as either spin - independent or spin - dependent . in the former case ,
wimps scatter coherently with an entire nucleus , leading to a cross section that scales with the square of the atomic number of the target nuclei . in the later case , the wimp couples to the spin of the target nucleus .
in the relevant non - relativistic limit , scalar , vector , and tensor couplings result in a spin - independent interaction , whereas axial couplings lead to a spin - dependent interaction @xcite . in this subsection ,
we focus on the spin - independent elastic scattering of wimps with nuclei , as the direct detection constraints for this class of interactions are considerably more stringent . in the next subsection
, we will return to spin - dependent scattering within the context of wimp capture in the sun . the wimp - nucleus cross section for spin - independent elastic scattering
is given by @xmath46 ^ 2\ ] ] where @xmath47 and @xmath48 are the atomic mass and atomic number of the target nuclei .
the effective couplings to protons and neutrons , @xmath49 , can be written in terms of the wimp s couplings to quarks . in the case of a scalar interaction @xmath50 where @xmath51 denotes the wimp s
effective fermi coupling for a given quark species .
the first term reflects scattering with light quarks , while the second term accounts for interactions with gluons through a heavy quark loop .
the values of @xmath52 are proportional to the matrix element , @xmath53 , of quarks in a nucleon and have been measured to be @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 @xcite .
the value of @xmath60 is given by @xmath61 and is approximately @xmath62 and @xmath63 for protons and neutrons , respectively .
in the case of a yukawa - like scalar interaction ( @xmath64 ) , there are significant contributions from both light and heavy quarks . in the case
in which the ratio of the effective scalar coupling to the quark mass , @xmath65 , is the same for each quark species , we arrive at a cross section per nucleon of @xmath66 ^ 2.\ ] ] in contrast , if we consider the case in which the scalar couplings to all quarks types are equal ( universal couplings ) , the resulting cross section is much larger : @xmath67 gev@xmath35 and for effective vector couplings to each quark of @xmath68 gev@xmath35 . also shown as solid curves are the current upper limits from the cdms @xcite and xenon @xcite experiments .
we do not show the case in which the scalar couplings are equal for each quark species , as its leads to much larger cross sections and are strongly excluded.,title="fig:",width=336 ] gev@xmath35 and for effective vector couplings to each quark of @xmath68 gev@xmath35 .
also shown as solid curves are the current upper limits from the cdms @xcite and xenon @xcite experiments .
we do not show the case in which the scalar couplings are equal for each quark species , as its leads to much larger cross sections and are strongly excluded.,title="fig:",width=336 ] the cross section for the scalar case of @xmath69 is shown in the left frame of fig .
[ fermionsi ] , and compared to the current upper limits from the cdms @xcite and xenon @xcite experiments . for fermionic wimps heavier than about @xmath70 gev , scalar couplings are constrained to be smaller than @xmath71 . comparing this result to those shown in fig.[fermionrelic ]
, we find that fermionic wimps must either be heavier than the top quark threshold to avoid being overproduced in the early universe _ and _ avoid direct detection constraints _ or _ some combination of resonance annihilation , coannihilation , or annihilations to final states other than quarks must dominate the thermal freeze - out process . if instead we were to consider the case of universal scalar couplings to all quark types , as applied in eq .
( [ universalelastic ] ) , we find an even more stringent constraint . in particular , the entire range of couplings that could potentially lead to an acceptable relic density is excluded by current direct detection constraints by multiple orders of magnitude .
we therefore conclude that if a wimp is to annihilate largely through scalar interactions , its couplings to light quarks must be considerably suppressed ( such as in the case of yukawa - like couplings , @xmath72 ) if it is to avoid being excluded by current direct detection constraints .
in contrast to scalar interactions , pseudoscalar interactions do not lead to a significant elastic scattering cross section between wimps and nucleons in the low velocity limit .
the reason for this can be seen if one explicitly computes the quark contribution of the vertex , @xmath73 , which goes to zero in the limit of zero momentum @xcite . the same conclusion is reached in ref.@xcite , in which the relevant nuclear matrix elements are calculated . in the case of a dirac ( non - majorana )
fermion , a vector coupling can also generate a spin - independent elastic scattering cross section .
in contrast to the scalar case , a vector interaction will be dominated by couplings to the up and down quarks in the nucleon : @xmath74 if we assume @xmath75 , this leads to a spin - independent elastic scattering cross section ( per nucleon ) of @xmath76 . from the right frame of fig.[fermionsi ] , we see that this cross section is in excess of current experimental limits @xcite unless @xmath77 gev@xmath35 . comparing this to fig .
[ fermionrelic ] , however , we find that in order for a dirac fermionic wimp with a mass in the range 10 to 1000 gev to annihilate largely through a vector interaction , it must be depleted in the early universe by some combination of resonance annihilation , coannihilation , or annihilations to final states other than quarks if it is to avoid direct detection constraint without being overproduced in the early universe .
this conclusion also holds for a fermionic wimp with a tensor interaction .
we would like to emphasize that the elastic scattering cross sections we have calculated here should be thought of as approximate upper limits ( again , assuming no late time decays or other non - thermal mechanisms are responsible for the dark matter density ) .
if coannihilations , resonances , or annihilations to leptons , gauge or higgs bosons dominated the freeze - out process , then the effective couplings required to generate the observed relic abundance may be considerably smaller , leading to reduced elastic scattering cross sections with nuclei .
if wimps accumulate in the core of the sun in sufficient numbers , their annihilations can potentially produce an observable flux of high - energy neutrinos @xcite .
wimps in the solar system elastically scatter with nuclei in the sun and become gravitationally bound at the rate approximately given by @xcite @xmath78 where @xmath79 , @xmath80 and @xmath81 are the spin - dependent ( sd ) and spin - independent ( si ) elastic scattering cross sections of wimps with hydrogen ( protons ) and helium nuclei , respectively .
the factor of @xmath82 reflects the solar abundance of helium relative to hydrogen and well as dynamical factors and form factor suppression .
the number of wimp in the sun , @xmath83 , evolves as @xmath84 where @xmath85 is the wimp s annihilation cross section times the relative velocity divided by the effective volume of the sun s core .
the present annihilation rate in the sun is given by @xmath86 where @xmath87 billion years is the age of the solar system .
the annihilation rate is maximized when it reaches equilibrium with the capture rate ( _ i.e. , _ when @xmath88 ) .
these wimp annihilations lead to a flux of neutrinos at earth given by @xmath89 where @xmath90 is the capture rate of wimps in the sun , @xmath91 is the non - equilibrium suppression factor ( approximately @xmath92 for capture - annihilation equilibrium ) , @xmath93 is the earth - sun distance and @xmath94 is the neutrino spectrum from the sun per wimp annihilating , which depends on the mass of the wimp and its dominant annihilation modes . due to @xmath95 vacuum oscillations ,
the muon neutrino flux observed at earth is the average of the @xmath96 and @xmath97 components .
muon neutrinos produce muons in charged current interactions with nuclei in the material inside or near the detector volume of a high - energy neutrino telescope .
the rate of neutrino - induced muons observed in a high - energy neutrino telescope is given by @xmath98 where @xmath99 is the neutrino - nucleon charged current interaction cross section , @xmath100 is the fraction of neutrino energy that goes into the muon and @xmath101 is the effective area of the detector .
the factor @xmath102 is either the distance a muon of energy @xmath103 travels before falling below the muon energy threshold of the experiment , called the muon range , or the width of the detector , whichever is larger .
the spectrum and flux of neutrinos generated in wimp annihilations is determined by the wimp s mass and leading annihilation modes .
if the rate at which wimps are captured in the sun is dominated by spin - independent scattering , one can translate the bounds from cdms @xcite and xenon @xcite into an upper limit on the neutrino flux .
in fact , even for the maximum elastic scattering cross section allowed by these experiments , no more than a few neutrino - induced muons will be generated per year in a kilometer - scale detector @xcite .
this is well below the sensitivity of next generation neutrino telescopes such as icecube @xcite .
thus , if we are to detect wimp annihilations using neutrino telescopes , the capture rate must be dominated by spin - dependent scattering , which is far less constrained by direct detection experiments .
the wimp - nucleus spin - dependent elastic scattering cross section is approximately given by @xcite @xmath104 where @xmath105.\ ] ] in these expressions , @xmath106 is the nuclear spin , and @xmath107 are the expectation values of the spin content of protons or neutrons in the target nucleus .
the quantities @xmath108 are coefficients of the matrix element of the axial current in a nucleon , with values given by @xmath109 , @xmath110 , and @xmath111 . inserting these values into the above equations , the wimp - proton , spin - dependent cross section reduces to @xmath112 ^ 2 , \label{cancel}\ ] ] which , for approximately universal couplings , yields ) .
departures from the universality of @xmath113 , @xmath114 and @xmath115 , however , could lead to larger cross sections than those estimated here . considering the axial couplings of the @xmath48 boson to fermions , for example , the opposite signs of the couplings to up and
down - type fermions leads to an elastic scattering cross section about @xmath116 times larger than estimated in eq .
( [ sigmaa ] ) . ]
@xmath117 currently , the strongest constraints on spin - dependent wimp - proton scattering come from the coupp @xcite and kims @xcite collaborations , which exclude cross sections larger than @xmath118 pb . this limit , however , is well beyond the range anticipated for a thermal wimp . in fig .
[ fermionneutrino ] , we plot the annihilation rate of wimps in the sun for the case of a fermionic wimp with axial couplings to quarks . to be detected over the atmospheric neutrino background
, the annihilating wimps must generate tens of neutrino - induced muons per year in a kilometer - scale , high - energy neutrino telescope , such as icecube . in fig .
[ fermionneutrino ] we also plot the approximate annihilation rate required to generate 20 events ( above a muon threshold energy of 50 gev ) per year at icecube .
this reach is shown as solid lines for the case of wimp annihilations to bottom quarks or gauge bosons .
gev@xmath35 . also shown as solid lines are the approximate rates needed to be detected by an experiment such as icecube ( 20 events per square kilometer , year with a 50 gev muon energy threshold ) .
the solid lines denote the reach for wimps annihilating to bottom quarks ( top ) or gauge bosons ( bottom).,width=336 ] in addition to neutrinos , products of wimp annihilations including gamma rays @xcite , electrons , positrons @xcite , and antiprotons @xcite could also potentially provide detectable signals .
the reach of these efforts depend on a combination of astrophysical inputs , such as the distribution of dark matter in the galaxy and the properties of the galactic magnetic field , and on the wimp s properties , namely its mass , annihilation cross section , and dominant annihilation modes .
although we will not , in this article , discuss the astrophysical inputs impacting such searches , we will briefly comment on the wimp s annihilation cross section as it appears in our model - independent analysis .
if we fix the wimp s effective couplings such that its annihilation cross section at the temperature of freeze - out is equal to the value required to yield the observed dark matter abundance , then we can proceed to estimate its annihilation cross section in the low - velocity limit ( the relevant limit for indirect searches ) . from eqs .
( [ svs]-[svt ] ) , we see that fermionic wimps annihilating through pseudoscalar , vector , and tensor interactions do so largely through terms for which @xmath119 is constant , rather than @xmath120 .
this leads to a low - velocity annihilation cross section of approximately @xmath121 @xmath38/s in these cases .
scalar or axial interaction forms , in contrast , lead to an annihilation cross section that scales as @xmath122 , and thus imply rates suppressed by a factor of about @xmath32 for wimp annihilations in the galactic halo .
our model - independent results for a fermionic wimp are summarized in fig .
[ fermionsummary ] . in each frame
, the solid dark ( black ) line denotes the combinations of wimp masses and couplings that lead to a thermal abundance of dark matter equal to the value measured by wmap @xcite .
as we have pointed out , however , these calculations were performed under the assumption that resonances , coannihilations , and annihilations to gauge and higgs bosons do not play a significant role in the thermal freeze - out process .
if any of these processes have significant effects , the wimp couplings could be considerably smaller while still producing a dark matter abundance consistent with wmap .
although eqs .
( [ sigmas]-[sigmaaa ] ) do not form a complete set of lorentz - invariant interaction lagrangians , they are representative of a larger set of combinations of interactions : @xmath123 up to factors of @xmath124 , wimps with interactions of these types will have annihilation cross sections and relic densities determined by the wimp interaction form ( the left side of the lagrangian ) , and elastic scattering cross sections determined by the fermion interaction form ( the right side of the lagrangian ) .
consequently , interaction lagrangians of these forms will provide results that are redundant with what we have obtained here
. we will consider interaction lagrangians of this form for scalar wimps , where we obtain non - redundant results . to roughly estimate the effect of a potential resonance ,
consider a wimp annihilating through the @xmath7-channel exchange of a mediator , @xmath125 , to fermion - antifermion pairs .
if @xmath126 then we can write an effective fermi coupling , @xmath127 , where @xmath128 and @xmath129 are the mediator s couplings to the wimps and final state fermions , respectively
. if the mediator s mass is not much greater than twice the wimp mass , however , the effects of the resonance on the annihilation cross section can be significant .
in particular , we can roughly estimate an effective @xmath130 for calculating the wimp s annihilation cross section : @xmath131^{1/2}},\ ] ] where @xmath132 is the width of the mediating particle .
for a 450 gev mediator with a narrow width and a 200 gev wimp , the effective value of @xmath133 is a factor of about 5 larger than is found neglecting the effects of the resonance , which enables the measured dark matter abundance to be generated with a product of couplings ( _ i.e. _ , @xmath134 ) that is smaller by a factor of 5 than those shown to be required in fig.[fermionsummary ] . in other words ,
the effective value of @xmath130 for the purposes of calculating the wimp annihilation cross section ( but not for calculating elastic scattering cross sections ) is increased by a factor of 5 in this case .
for the same 450 gev mediator and a 220 gev wimp , the resonance enhances the effective value of @xmath130 by a factor approximately 20 .
although one should integrate the cross section over the thermal distribution to accurately account for the effect of a resonance on the relic abundance @xcite , we provide this estimate to illustrate how such a feature is qualitatively expected to impact the resulting dark matter density .
this effect is important in interpreting the constraints from direct detection experiments and the reach of neutrino telescopes shown in fig .
[ fermionsummary ] .
consider , for example , the case of scalar interactions with @xmath72 shown in the upper left frame .
although by simply comparing the dark solid line to the lighter ( blue ) solid line , the constraints from cdms and xenon appear to rule out a wimp with the measured relic abundance unless it is heavier than about @xmath135 gev , this conclusion can be relaxed considerably if the wimp annihilates through a resonance .
similarly , if coannihilations or annihilations to gauge / higgs bosons play an important role in the freeze - out process , the required effective couplings will be considerably reduced as well .
furthermore , departures from the universality of the wimp s couplings to fermions can also alter the results summarized here .
wimps that couple preferentially to light ( heavy ) quarks will be more ( less ) significantly constrained by direct detection experiments and neutrino telescopes . in an extreme case , we can imagine a wimp that annihilates almost entirely through couplings to gauge boson final states rather than fermions , which in turn would lead to highly suppressed elastic scattering cross sections . to summarize our results for the case of a fermionic wimp
, we find : * fermionic wimps with scalar interactions are required by direct detection constraints to either 1 ) be heavier than about @xmath135 gev , 2 ) annihilate in the early universe through a resonance or coannihilations , or 3 ) couple preferentially to leptons , heavy quarks , or gauge / higgs bosons .
the case of universal couplings is very strongly disfavored by current direct detection constraints ( see the middle - left frame of fig . [ fermionsummary ] ) . *
the conclusions described for a fermionic wimp with scalar interactions also apply to the case of a dirac fermionic wimp with vector interactions and fermionic wimps with tensor interactions .
this is the reason why a heavy 4th generation dirac neutrino , for example , is no longer an acceptable candidate for dark matter . * fermionic wimps with uniquely pseudoscalar or axial interactions
are not constrained by direct detection experiments at this time . *
next generation , kilometer - scale neutrino telescopes will be capable of constraining the case of fermionic wimps with axial interactions . departing for a moment from our model - independent analysis , we would like to comment on our results as interpreted within the context of neutralinos , which are attractive dark matter candidates in supersymmetric models @xcite .
neutralinos are majorana fermions , and undergo scalar , psuedoscalar , and axial interactions . roughly speaking
, neutralinos will be overproduced in the early universe unless at least one of the following conditions are met : 1 ) they are able to coannihilate efficiently with the lightest stau or other superpartners ( the coannihilation region ) , 2 ) they are able to annihilate efficiently through the cp - odd higgs boson resonance ( the @xmath47-funnel region ) , 3 ) they are a strongly mixed gaugino - higgsino , leading to large couplings ( the focus point region ) , or 4 ) much of the supersymmetric spectrum is relatively light , making efficient annihilations possible ( the bulk region ) . in the @xmath47-funnel region ,
neutralinos annihilate near resonance via pseudoscalar interactions , but also elastically scatter through scalar interactions associated with cp - even higgs exchange ( and squark exchange ) , leading to a constraint similar in form to that shown in the upper left frame of fig .
[ fermionsummary ] , but with the solid dark relic abundance contour reduced by at least one order of magnitude or more . both the @xmath47-funnel and bulk regions are beginning to be significantly explored by direct detection experiments and , in the absence of a positive detection , will be highly constrained in the coming years . in the focus point region ,
the neutralino s couplings are enhanced , leading to scalar elastic scattering cross sections near the current constraints from cdms and xenon .
although the cdms / xenon constraint shown in fig .
[ fermionsummary ] is somewhat weakened by the the fact that neutralino annihilations in the focus point region proceed largely to gauge boson final states , direct detection experiments will essentially close the focus point region if no detection is made in the next couple of years .
furthermore , focus point neutralinos have sizable couplings to the @xmath48 boson , leading to large spin - dependent elastic scattering cross sections through axial interactions .
as mentioned before , the opposite sign of the @xmath48 s couplings to up and down type fermions leads to a much greater reach for icecube than is shown in the lower right frame of fig .
[ fermionsummary ] .
hundreds or thousands of events per year at icecube are predicted throughout much of the focus point region .
finally , neutralinos in the stau coannihilation region are the least constrained by direct and indirect searches , as their couplings can be very small without leading to their overproduction in the early universe .
in this section , we consider the case of a scalar wimp , with scalar and vector , as well as combinations of scalar / pseudoscalar and vector / axial vector , interaction forms . in analogy with eqs.([lint][fermionlag ] ) , we write @xmath136 where @xmath137 denotes the scalar wimp .
note that in @xmath138 has mass dimension of @xmath139 for the scalar and scalar - pseudoscalar interactions , and @xmath140 for vector and vector - axial vector interactions . for scalar wimps ,
the annihilation cross sections to fermion - antifermion pairs are given by @xmath141 \\
\sigma_v & = & \frac{1}{16\pi } \sum_f f_{v , f}^2 \ , c_f \sqrt{\frac{s-4 m_f^2}{s-4 m_\phi^2 } } \ , \left [ \frac{2\ , ( s-4m^2_{\phi})(s+2m^2_f)}{3\,s}\right ] \\ \sigma_{sp } & = & \frac{1}{16\pi } \sum_f f_{sp , f}^2 \ , c_f \sqrt{\frac{s-4 m_f^2}{s-4 m_\phi^2 } } \\ \sigma_{va } & = & \frac{1}{16\pi } \sum_f f_{va , f}^2 \ , c_f \sqrt{\frac{s-4 m_f^2}{s-4 m_\phi^2 } } \ , \left [ \frac{2\ , ( s-4m^2_{\phi})(s-4m^2_f)}{3\,s}\right ] \end{aligned}\ ] ] expanding in powers of relative velocity , we arrive at @xmath142 \\ \sigma_v \vert v\vert & \approx & \frac{1}{4 \pi}\sum_f f_{v , f}^2 \ , c_f \
m_\phi^2 \ \sqrt{1 - m_f^2/m^2_{\phi } } \ , \left[\frac{1}{3}\left(2+\frac{m^2_f}{m_{\phi^2}}\right ) \ , v^2 \right ] \\
\sigma_{sp } \vert v\vert & \approx & \frac{1}{4\pi}\sum_f f_{sp , f}^2 \ , c_f \
\sqrt{1- m_f^2/m^2_{\phi } } \ , \left[\frac{1}{4}\right]\\ \sigma_{va } \vert v\vert & \approx & \frac{1}{4 \pi}\sum_f f_{va , f}^2 \ , c_f \
m_\phi^2 \ \sqrt{1 - m_f^2/m^2_{\phi } } \ , \left[\frac{1}{3}\left(2-\frac{m^2_f}{m_{\phi^2}}\right ) \ , v^2 \right ] .
\label{vascalar}\end{aligned}\ ] ] gev@xmath35 . in the other four frames , results for @xmath143 gev@xmath35 are given .
if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here . also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] gev@xmath35 . in the other four frames , results for @xmath143 gev@xmath35 are given . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] gev@xmath35 . in the other four frames ,
results for @xmath143 gev@xmath35 are given . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] gev@xmath35 . in the other four frames , results for @xmath143 gev@xmath35 are given .
if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here . also shown as horizontal lines
is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] gev@xmath35 . in the other four frames , results for @xmath143 gev@xmath35 are given . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] gev@xmath35 . in the other four frames , results for @xmath143 gev@xmath35 are given . if resonances , coannihilations , or annihilations to final states other than fermion - antifermion pairs are significant , the relic abundance is expected to be significantly lower than shown here .
also shown as horizontal lines is the range of the cold dark matter density measured by wmap @xcite.,title="fig:",width=336 ] to calculate the thermal relic abundance of a scalar wimp , we follow the same procedure as described in sec .
[ fermion ] .
we show the results of this calculation in fig .
[ scalarrelic ] .
the calculation of the elastic scattering cross section for a scalar wimp with nuclei is similar to that described for a fermionic wimp in sec .
[ directfermion ] .
although we will not repeat the details of this calculation here , we will comment on the most important differences . in the case of a scalar wimp with a scalar interaction with quarks
, the effective coupling @xmath144 possesses a different dimensionality than @xmath51 .
this , in turn , leads to a stronger dependence on the wimp mass .
in particular , heavier wimps have a somewhat smaller elastic scattering cross section and thus are less constrained by direct detection experiments .
the elastic scattering cross sections for a scalar wimp are shown in fig .
[ scalarsi ] . by comparing figs .
[ scalarrelic ] and [ scalarsi ] , we can see that scalar interactions of the form @xmath145 that lead to an acceptable relic density also exceed direct detection constraints if @xmath146 . for wimps heavier than the top quark
, smaller couplings allow a wimp to evade current direct detection constraints while also yielding an acceptable dark matter abundance .
as in the fermionic case , we find that a scalar wimp that annihilates largely through universal scalar couplings ( equal for all fermion species ) will be essentially excluded by existing direct detection constraints .
scalar wimps with vector interactions are also severely constrained by present direct detection experiments . by comparing the lower frames of figs .
[ scalarrelic ] and [ scalarsi ] , we find that scalar wimps with vector interactions with fermions must be heavier than several tev to evade current elastic scattering bounds , unless resonances , coannihilations or annihilations to gauge / higgs bosons play an important role , in which case lighter wimps may also be allowed .
spin - dependent scattering between scalar wimps and nuclei occurs only in the case of vector - axial vector interactions .
although the capture rate of scalar wimps in the sun may potentially be large in this case , the annihilation cross section scales with @xmath18 ( see eq . [ vascalar ] ) , thus suppressing the annihilation rate in the sun s core , and along with it the resulting high - energy neutrino flux .
scalar wimps are , therefore , not expected to be within the reach of icecube or other planned high - energy neutrino telescopes .
the prospects for the indirect detection of scalar wimps using gamma - rays or charged particles ( @xmath147 , @xmath148 ) once again depend on the relationship between the wimp s annihilation cross section and relative velocity . in the case of scalar couplings ,
the annihilation cross section , @xmath119 , is nearly independent of of the wimps relative velocity , whereas vector interactions yield @xmath149 . as a result
, the indirect detection prospects for a scalar wimp with vector interactions are expected to be highly suppressed .
gev@xmath35 and for effective vector couplings to each quark of @xmath150 gev@xmath35 . also shown as solid curves
are the current upper limits from the cdms @xcite and xenon @xcite experiments .
we do not show the case in which the scalar couplings are equal for each quark species , as its leads to much larger cross sections and are strongly excluded.,title="fig:",width=336 ] gev@xmath35 and for effective vector couplings to each quark of @xmath150 gev@xmath35 .
also shown as solid curves are the current upper limits from the cdms @xcite and xenon @xcite experiments .
we do not show the case in which the scalar couplings are equal for each quark species , as its leads to much larger cross sections and are strongly excluded.,title="fig:",width=336 ] our model - independent results for a scalar wimp are summarized in fig .
[ scalarsummary ] . in each frame
, the solid dark ( black ) line denotes the combinations of wimp mass and couplings that lead to a thermal abundance equal to the measured dark matter density , again in the absence of significant effects of resonances , coannihilations , or annihilations to gauge / higgs bosons .
the lighter ( blue ) curve in each frame denotes the current constraints from the direct detection experiments cdms and xenon . to summarize our results for the case of a scalar wimp
, we find : * scalar wimps with scalar interactions with standard model fermions are required by direct detection constraints to 1 ) be heavier than about @xmath151 gev , 2 ) annihilate in the early universe through a resonance or coannihilations , or 3 ) couple preferentially to leptons , heavy quarks , or gauge / higgs bosons .
the case of universal couplings is very strongly disfavored by current direct detection constraints ( see the upper - right frame of fig .
[ scalarsummary ] ) .
* the conditions described for a scalar wimp with scalar interactions also apply to the case of a scalar wimp with vector interactions . in the absence of resonances , coannihilations , and/or annihilations to gauge / higgs bosons , current direct detection constraints exclude such a dark matter candidate by multiple orders of magnitude .
* neutrino telescopes are not likely to constrain scalar wimps beyond the level already achieved by direct detection experiments .
even if the large hadron collider does not reveal physics beyond the standard model , a dark matter candidate in the form of a weakly interacting massive particle may still exist . in this article , we have studied how the nature of such a wimp could be deduced by its signatures in astrophysical experiments . in our analysis , we have taken a general and model - independent approach , considering fermionic or scalar wimps with a variety of interaction forms . in table
[ t1 ] , we summarize our findings . for each combination of spin and interaction form , we indicate the constraints placed by and the prospects for direct detection experiments , high - energy neutrino telescopes , and indirect detection experiments using gamma - rays or charged particles . under the column of direct detection
, we use the phrases `` strongly excluded , '' `` weakly excluded , '' or `` within near future reach , '' to denote the sensitivity or prospects for each case . by
`` strongly excluded , '' we indicate instances in which the effective couplings to quarks , as relevant to elastic scattering with nuclei , must be suppressed by more than a factor of ten relative to the value required to generate thermally the observed dark matter abundance .
as we have discussed , such a suppression could result from resonant annihilations , coannihilations , or annihilations to gauge / higgs bosons .
the label `` weakly excluded , '' in contrast , indicates only that the case is excluded if the effective couplings to quarks are not suppressed by such effects .
lastly , the label `` within near - future reach '' indicates an elastic scattering cross section ( without suppression ) that is within approximately two order of magnitudes of current direct detection limits . under the column of neutrino telescopes
, we classify each case as either not sensitive or sensitive over a range of wimp masses ( for next generation experiments , such as icecube ) .
this evaluation depends on the annihilation products of the wimp , however , and thus are highly approximate . under the column of gamma - rays and
charged cosmic rays , we simply indicate whether the wimp s annihilations are or are not suppressed by the square of the wimp s velocity . if such velocity suppression is present , it is highly unlikely that glast , pamela or other planned indirect detection experiments will be capable of detecting dark matter .
this leads us to the most obvious and important question : will the information provided by direct and indirect detection experiments be able to be used to infer the particle nature of the dark matter ?
although there are certainly cases in which measurements by these experiments will not lead to an unambiguous identification , there are many scenarios in which a great deal could be learned . for example , if icecube or another high - energy neutrino telescope were to observe neutrinos from wimp annihilations in the sun , we would be able to conclude that the wimp is likely fermionic , and that it possesses an axial interaction with light quarks . by studying
the precise rate observed , one could also potentially determine whether the wimp s axial interaction played a dominant or only subdominant role in the process of thermal freeze - out .
this could be combined with observations from direct detection experiments to further constrain the possible interactions possessed by the wimp .
as a second possible scenario , imagine that near future direct detection experiments observe a wimp with a mass of a few hundred gev and that , shortly afterward , glast observes a corresponding gamma - ray signal from wimps annihilating in the halo . from table
[ t1 ] , we can see that this leads us to only three likely possibilities : the wimp is either a fermion with vector interactions , a fermion with pseudoscalar - scalar interactions , or a scalar with yukawa - like scalar interactions .
although previous studies have shown that dark - matter experiments have the potential to constrain the parameters of supersymmetry @xcite or even to help identify the theoretical framework from which the dark matter arises @xcite , here we have demonstrated that a far more model - independent approach can also be fruitful . in particular , without assuming any particular theoretical framework or model , we have shown that direct and indirect dark matter experiments can be used to considerably constrain the spin and interactions of the dark matter , even in the absence of any discoveries at the lhc .
c c c c c c c c interaction & & direct detection & & neutrino telescopes & & @xmath152-rays , @xmath147 , @xmath148 & + scalar & & strongly excluded @xmath153 gev & & not sensitive & & suppressed by @xmath18 & + ( @xmath72 ) & & weakly excluded @xmath154 gev & & & & & + & & within near future reach @xmath155 gev & & & & & + scalar & & strongly excluded @xmath156 gev@xmath157 tev & & na & & suppressed by @xmath18 & + ( @xmath130 universal ) & & & & & & & + pseudoscalar & & not sensitive & & not sensitive & & unsuppressed & + vector / tensor & & strongly excluded @xmath158 gev & & not sensitive & & unsuppressed & + & & weakly excluded @xmath159 gev@xmath140 tev & & & & & + axial & & not sensitive & & sensitive @xmath160 gev & & suppressed by @xmath18 & + scalar - pseudoscalar & & not sensitive & & not sensitive & & suppressed by @xmath18 & + pseudoscalar - scalar & & weakly excluded @xmath161 gev & & not sensitive & & unsuppressed & + ( @xmath72 ) & & within near future reach @xmath162 gev & & & & & + vector - axial & & not sensitive & & not sensitive & & unsuppressed & + axial - vector & & strongly excluded @xmath156 gev@xmath140 tev & & not sensitive & & unsuppressed & + & & weakly excluded @xmath163 tev & & & & & + c c c c c c c c interaction & & direct detection & & neutrino telescopes & & @xmath152-rays , @xmath147 , @xmath148 & + scalar & & weakly excluded @xmath164 gev & & not sensitive & & unsuppressed & + ( @xmath145 ) & & within near future reach @xmath165 gev & & & & & + scalar & & strongly excluded @xmath166 gev@xmath157 tev & & na & & unsuppressed & + ( @xmath138 universal ) & & & & & & & + vector & & strongly excluded @xmath167 gev@xmath139 tev & & not sensitive & & suppressed by @xmath18 & + & & weakly excluded @xmath168 tev & & & & & + scalar - pseudoscalar & & not sensitive & & not sensitive & & unsuppressed & + vector - axial & & not sensitive & & not sensitive & & suppressed by @xmath18 & + the results presented in table [ t1 ] rely upon the set of assumptions we have adopted . it must be noted that if dark matter consists of non - thermally produced wimps , or of multiple species of particles , our conclusions could be altered considerably .
furthermore , one might worry that the effects of resonances , coannihilations , or annihilations to gauge / higgs bosons , which we have largely neglected in our analysis , might dramatically change our conclusions . to some degree , however , the impact of such processes are encapsulated in our definitions of `` strongly excluded '' and `` weakly excluded '' , as used in table [ t1 ] . for example , if a wimp annihilates largely through a narrow resonance such that twice the mass of the wimp lies within approximately 5% of the exchanged particle , then the effective couplings relevant for elastic scattering can be reduced by a factor of ten without the wimp being overproduced in the early universe ( see sec . [ resonance ] ) .
this mildly ( 5% or less ) fine - tuned resonance corresponds to the `` weakly excluded '' label used in the table .
anything labeled `` strongly excluded '' would require the masses to be tuned even more precisely to the resonance value to remain viable .
similarly , if a significant fraction of wimp annihilations in the early universe proceeded to a combination of gauge or higgs bosons , or occurred through coannihilations with another species of particle , the elastic scattering cross section could be suppressed . for the scenarios we have labeled
as `` strongly excluded '' to have remained hidden from direct detection experiments , however , about @xmath169% or more of the annihilations / coannihilations of wimps in the early universe must have occurred through such processes .
so although the conclusions we have reached here are not entirely immune to the inclusion of such effects , they are quite robust in all but the most extreme cases . in conclusion , we find that in the case that the large hadron collider does not discover physics beyond the standard model , astrophysical experiments may still be able to constrain the nature of the dark matter , even without assuming supersymmetry or any other specific particle physics framework .
in particular , the spin and interaction forms of dark matter can potentially be identified by combining results from direct detection experiments , neutrino telescopes , and indirect detection experiments using gamma - rays or charged cosmic rays .
h. goldberg , phys .
* 50 * , 1419 ( 1983 ) ; j. r. ellis , j. s. hagelin , d. v. nanopoulos , k. a. olive and m. srednicki , nucl .
b * 238 * , 453 ( 1984 ) ; g. l. kane , c. f. kolda , l. roszkowski and j. d. wells , phys .
d * 49 * , 6173 ( 1994 ) [ arxiv : hep - ph/9312272 ] .
e. w. kolb and r. slansky , phys .
b * 135 * , 378 ( 1984 ) ; h. c. cheng , j. l. feng and k. t. matchev , phys .
89 * , 211301 ( 2002 ) [ arxiv : hep - ph/0207125 ] ; d. hooper and s. profumo , phys . rept . * 453 * , 29 ( 2007 ) [ arxiv : hep - ph/0701197 ] . g. servant and t. m. p. tait , nucl .
b * 650 * , 391 ( 2003 ) [ arxiv : hep - ph/0206071 ] ; k. agashe and g. servant , phys .
* 93 * , 231805 ( 2004 ) [ arxiv : hep - ph/0403143 ] ; jcap * 0502 * , 002 ( 2005 ) [ arxiv : hep - ph/0411254 ] ; k. agashe , a. falkowski , i. low and g. servant , arxiv:0712.2455 [ hep - ph ] .
h. c. cheng and i. low , jhep * 0408 * , 061 ( 2004 ) [ arxiv : hep - ph/0405243 ] ; i. low , jhep * 0410 * , 067 ( 2004 ) [ arxiv : hep - ph/0409025 ] ; a. birkedal , a. noble , m. perelstein and a. spray , phys .
d * 74 * , 035002 ( 2006 ) [ arxiv : hep - ph/0603077 ] .
f. giuliani , phys .
lett . * 93 * , 161301 ( 2004 ) [ arxiv : hep - ph/0404010 ] .
a. kurylov and m. kamionkowski , phys .
d * 69 * , 063503 ( 2004 ) [ arxiv : hep - ph/0307185 ] .
p. gondolo and g. gelmini , nucl .
b * 360 * , 145 ( 1991 ) .
z. ahmed _ et al . _ [ cdms collaboration ] , arxiv:0802.3530 [ astro - ph ] .
j. angle _ et al .
_ [ xenon collaboration ] , phys .
100 * , 021303 ( 2008 ) [ arxiv:0706.0039 [ astro - ph ] ] .
f. halzen and a. d. martin , _ new york , usa : wiley ( 1984 ) 396p_. l. bergstrom , j. edsjo and p. gondolo , phys .
d * 55 * , 1765 ( 1997 ) [ arxiv : hep - ph/9607237 ] ; phys .
d * 58 * , 103519 ( 1998 ) [ arxiv : hep - ph/9806293 ] ; v. d. barger , f. halzen , d. hooper and c. kao , phys .
d * 65 * , 075022 ( 2002 ) .
a. gould , astrophys .
j. * 388 * , 338 ( 1991 ) .
f. halzen and d. hooper , phys .
d * 73 * , 123507 ( 2006 ) [ arxiv : hep - ph/0510048 ] .
t. deyoung [ icecube collaboration ] , int .
a * 20 * , 3160 ( 2005 ) ; j. ahrens _ et al . _ [ the icecube collaboration ] , nucl .
. suppl . * 118 * , 388 ( 2003 ) [ arxiv : astro - ph/0209556 ] .
e. behnke _
[ coupp collaboration ] , science * 319 * , 933 ( 2008 ) [ arxiv:0804.2886 [ astro - ph ] ] .
h. s. lee .
_ [ kims collaboration ] , phys . rev . lett . * 99 * , 091301 ( 2007 ) [ arxiv:0704.0423 [ astro - ph ] ] . l. bergstrom , p. ullio and j. h. buckley , astropart . phys .
* 9 * , 137 ( 1998 ) [ arxiv : astro - ph/9712318 ] ; p. ullio , l. bergstrom , j. edsjo and c. g. lacey , phys . rev .
d * 66 * , 123502 ( 2002 ) [ arxiv : astro - ph/0207125 ] ; s. dodelson , d. hooper and p. d. serpico , phys .
d * 77 * , 063512 ( 2008 ) [ arxiv:0711.4621 [ astro - ph ] ] ; e. a. baltz _ et al .
_ , jcap * 0807 * , 013 ( 2008 ) [ arxiv:0806.2911 [ astro - ph ] ] .
e. a. baltz and j. edsjo , phys .
d * 59 * ( 1999 ) 023511 [ arxiv : astro - ph/9808243 ] ; d. hooper and j. silk , phys .
d * 71 * , 083503 ( 2005 ) [ arxiv : hep - ph/0409104 ] . s. profumo and p. ullio , jcap * 0407 * , 006 ( 2004 ) [ arxiv : hep - ph/0406018 ] ; l. bergstrom , j. edsjo and p. ullio , astrophys . j. * 526 * , 215 ( 1999 ) [ arxiv : astro - ph/9902012 ] ; a. bottino , f. donato , n. fornengo and p. salati , phys .
d * 58 * , 123503 ( 1998 ) [ arxiv : astro - ph/9804137 ] ; t. bringmann and p. salati , phys .
d * 75 * , 083006 ( 2007 ) [ arxiv : astro - ph/0612514 ] . c. arina and n. fornengo , jhep
* 0711 * , 029 ( 2007 ) [ arxiv:0709.4477 [ hep - ph ] ] ; t. falk , k. a. olive and m. srednicki , phys .
b * 339 * , 248 ( 1994 ) [ arxiv : hep - ph/9409270 ] .
b. a. dobrescu , d. hooper , k. kong and r. mahbubani , jcap * 0710 * , 012 ( 2007 ) [ arxiv:0706.3409 [ hep - ph ] ] .
e. a. baltz , m. battaglia , m. e. peskin and t. wizansky , phys .
d * 74 * , 103521 ( 2006 ) [ arxiv : hep - ph/0602187 ] ; d. hooper and a. m. taylor , jcap * 0703 * , 017 ( 2007 ) [ arxiv : hep - ph/0607086 ] ; b. altunkaynak , m. holmes and b. d. nelson , arxiv:0804.2899 [ hep - ph ] ; d. hooper and l. t. wang , phys .
d * 69 * , 035001 ( 2004 ) [ arxiv : hep - ph/0309036 ] .
d. hooper and g. zaharijas , phys .
d * 75 * , 035010 ( 2007 ) [ arxiv : hep - ph/0612137 ] ; g. bertone , d. g. cerdeno , j. i. collar and b. c. odom , phys .
lett . * 99 * , 151301 ( 2007 ) [ arxiv:0705.2502 [ astro - ph ] ] . | despite compelling arguments that significant discoveries of physics beyond the standard model are likely to be made at the large hadron collider , it remains possible that this machine will make no such discoveries , or will make no discoveries directly relevant to the dark matter problem . in this article
, we study the ability of astrophysical experiments to deduce the nature of dark matter in such a scenario . in most dark matter studies ,
the relic abundance and detection prospects are evaluated within the context of some specific particle physics model or models ( _ e_.g . , supersymmetry ) . here
, assuming a single weakly interacting massive particle constitutes the universe s dark matter , we attempt to develop a model - independent approach toward the phenomenology of such particles in the absence of any discoveries at the large hadron collider .
in particular , we consider generic fermionic or scalar dark matter particles with a variety of interaction forms , and calculate the corresponding constraints from and sensitivity of direct and indirect detection experiments .
the results may provide some guidance in disentangling information from future direct and indirect detection experiments . |
it is difficult to count the incidence of crime at the local level .
typically , the only systematically collected crime data are those compiled by the local police department for submission to the federal bureau of investigation s ( fbi ) uniform crime reporting ( ucr ) program .
these numbers are best suited for tracking the amount of crime that is reported to the police and how often these reported crimes are cleared by arrest or exceptional means ( fbi , 2009 ) .
however , they often carry considerable weight in assessing how well a police department is performing , how safe a community is , and whether a police department needs more or different kinds of resources ( maltz , 1999:2 ) .
news media and law enforcement agencies routinely report levels and trends in crimes known to the police as `` crimes '' and `` criminal behavior . '' in prepared testimony to the u.s .
house of representatives , carbon ( 2012:16 ) writes that `` [ t]he ucr is the national ` report card ' on serious crime ; what gets reported through the ucr is how we , collectively view crime in this country . '' sometimes , these assessments veer into explicit crime rate rankings of cities ( rosenfeld and lauritsen [ 2008 ] discusses the scope of this problem ) a practice that has been condemned by the american society of criminology ( 2007 ) .
criminologists sometimes use and compare point estimates of crime rates for different jurisdications at various levels of aggregation , and they report relationships between crime and various social and economic indicators as if there was no uncertainty in those estimates beyond sampling variation . in our view
, there is no inherent problem with considering whether a crime rate is higher in one place than another at the same time or whether a crime rate is higher at one time than another in the same place . to be absolutely clear
, the problem arises when ambiguities in the statistics undermine the validity of the comparison .
for example , a major obstacle to using police - based crime statistics to infer within - community changes in crime over time or between - community crime differences arises from the well - known fact that many crimes are not reported to the police ( baumer and lauritsen , 2010 ; james and council , 2008 ) .
therefore , when crime rates vary across space or time , it is hard to know how much of that change is caused by shifts in real criminal behavior or changes in the reporting behavior of victims ( biderman and reiss , 1967 ; maltz , 1975 ; eck and riccio , 1979 ; blumstein et al . , 1991 , 1992 ; also for a similar idea in state sat rankings see wainer , 1986 ) . consider a simple anecdote that illustrates our concerns .
a recent newspaper article in the _ charlotte observer _ reported that `` the number of crimes dropped 7.1 percent last year , a development that charlotte police chief rodney monroe credited largely to officers keeping a close eye on potential criminals before they struck '' ( lyttle , 2012 ) .
the comparison expressed in this news coverage implicitly makes the strong and untestable assumption that the reporting behavior of victims stayed the same and all of the change in the number of crimes known to the police from one year to the next is due to changes in criminal behavior ( eck and riccio , 1979 ; blumstein et al . , 1991 ; brier and fienberg , 1980 ; nelson , 1979 ; skogan , 1974 ; rosenfeld and lauritsen , 2008 ; bialik , 2010 ) .
analytically , the same problems exist when criminologists try to explain variation in crime rates across different cities with identified explanatory variables like police patrol practices , dropout rates , home foreclosures , and unemployment ; they also arise when researchers try to measure and explain short - term changes in crime rates within the same jurisdiction . whether the analysis involves simple year - over - year percent change comparisons for different cities or
more complex statistical models for cross - sectional and panel data sets , the analytical ambiguities are the same .
in fact , variation in crime reporting patterns injects considerable ambiguity into the interpretation of police - based crime statistics ( eck and riccio , 1979 ; blumstein et al . , 1991 ;
levitt , 1998 ) .
recent work by baumer and lauritsen ( 2010 ) building on a long series of detailed crime reporting statistics from the national crime survey ( ncs ) and its successor , the national crime victimization survey ( ncvs ) makes the compelling point that there may be a causal relationship between the mobilization of the police and the likelihood that a community s citizens will report victimization experiences to the police .
police departments that cultivate strong working community partnerships may actually increase reporting of certain crimes simply because people believe the police will take useful actions when those crimes are reported : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ police notification rates are indicators of public confidence in the police and the legitimacy of the criminal justice system , and increasing police - public communication is a key goal of community - oriented policing strategies to reduce crime and the fear of crime ( baumer and lauritsen , 2010:132 ) . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ even changes in the number of police in a particular area may affect crime reporting practices of the citizenry ( levitt , 1998 ) .
variation in reporting rates can create the illusion of a shift in crime even if real crime levels are perfectly stable ( eck and riccio , 1979 ; maltz , 1975 ) .
in fact , residential burglary reporting rates do exhibit year - to - year volatility . from 2010 to 2011 , the rate at which residential burglary victimizations were reported to the police dropped from 59% to 52% ( truman , 2011:10 ; truman and planty , 2012:9 ) .
if these kinds of changes occur at the local level as well , they could easily explain a good deal of the variation we typically see from one year to the next in local , police - based robbery and burglary statistics . in this paper
, we conduct a case study of local level crime measurement while trying to pay close attention to some important sources of ambiguity . specifically , our objective is to estimate the incidence of residential burglary for each of the 10 most populous cities in north carolina in 2009 , 2010 , and 2011 .
the analysis is informed by data measuring the likelihood that residential burglary victimizations are reported to the police .
we focus on residential burglaries in the 10 largest north carolina cities because : ( 1 ) residential burglary is a clear , well - defined crime about which the public expresses considerable fear and concern ( blumstein and rosenfeld , 2008:18 - 20 ) ; ( 2 ) unlike most other states , north carolina law enforcement agencies publicly report residential burglaries known to the police separately from non - residential burglary ; ( 3 ) residential burglaries are household - level victimizations which correspond closely to the household - level structure of the ncvs ( the ncvs does not measure reporting behaviors for commercial burglaries ) ; and ( 4 ) conducting the analysis across cities and over time allows us to comment directly on the kinds of comparisons that are often conducted with police - based crime statistics .
we are not the first to consider this issue ( see , for example , maltz , 1975 ; eck and riccio , 1979 ; blumstein et al . , 1991 ; levitt , 1998 ; lohr and prasad , 2003 ; westat , 2010 ; raghunathan et al . , 2007 ) .
nonetheless , the interval estimates or bounds we propose in this paper consider several key sources of uncertainty and stand in contrast to the excessively definitive point estimates that are often the subject of public discourse about crime .
the first issue to address is why would we expect different rates of reporting crimes to the police in different jurisdictions ( cities , counties , states ) ?
since most police work is reactive rather than proactive , questions about the relationship between public perceptions of the police and the propensity of citizens to report crime victimizations to the police loom large .
if the public perceives the police as indifferent and unlikely to do anything to help , the likelihood of crimes being reported to the police could be affected ( baumer and lauritsen , 2010 ) .
concerns that the police are unresponsive to the needs of the community can lead to a phenomenon called `` legal cynicism . ''
kirk and colleagues ( kirk and matsuda 2011:444 ; kirk and papachristos 2011 ) have argued that legal cynicism is a `` cultural frame in which the law and the agents of its enforcement are viewed as illegitimate , unresponsive , and ill equipped to ensure public safety . '' in addition , legal cynicism is understood to be `` an emergent property of neighborhoods in contrast to a property solely of individuals '' in part because it is formed not only in reaction to one s own personal experiences with legal actors and institutions but through interaction with others in the community ( kirk and matsuda 2011:448 ) . according to this view , culture , and legal cynicism as part of it ,
is not perceived as a set of values , goals , or `` things worth striving for '' ( merton 1968:187 ) but rather as a repertoire or toolkit to use in understanding the world ( swidler 1986 ) .
there are two consequences that follow from the level of legal cynicism in a community .
first , if legal institutions like the police are perceived as illegitimate then citizens are less willing to comply with laws with the result that there is going to be more actual crime .
for example , fagan and tyler ( 2005 ) found that adolescents who perceived a lack of procedural justice among authorities also exhibited higher levels of legal cynicism .
adolescents rated higher in legal cynicism ( i.e. , expressing agreement with statements like `` laws are made to be broken '' ) were also higher in self - reported delinquency than those less cynical . in a survey of adult respondents , reisig , wolfe , and holtfreter ( 2011 ) reported that self - reported criminal offending was significantly related to their measure of legal cynicism net of other controls including self - control .
finally , kirk and papachristos ( 2011 ) found that legal cynicism in chicago neighborhoods explained why they had persistently high levels of homicide in spite of declines in both poverty and general violence .
in addition to these studies of legal cynicism , there are numerous studies which have shown a link between measures of legitimacy of legal institutions such as the courts and police and a higher probability of violating the law ( paternoster et al .
1997 ; tyler 2006 ; papachristos , meares , and fagan 2011 ) .
a second consequence of legal cynicism of central concern in this paper is that citizens are not likely to cooperate with the police , including reporting a crime when it occurs .
when citizens believe that the police are not likely to be responsive or will do little to help people like them , then we would expect more crimes to go unreported and offenders to go unarrested . the perception that it would do no good to cooperate with the police is an integral part of the cultural system described by anderson ( 1999:323 ) as the `` code of the street '' : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ t]he most public manifestation of this alienation is the code of the street , a kind of adaptation to a lost sense of security of the local inner - city neighborhood and , by extension , a profound lack of faith in the police and judicial system .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ several studies have found that community members both adults and juveniles are unlikely to cooperate with the police , including reporting crime and providing information , when law enforcement is seen as illegitimate ( sunshine and tyler 2003 ; tyler and fagan 2008 ; slocum et al .
kirk and matsuda ( 2011 ) found a lower risk of arrest in neighborhoods with high levels of legal cynicism . in sum , an increasing array of conceptual and empirical work
has linked a perceived lack of responsiveness on the part of legal actors to both more crime and less reporting of crime .
communities characterized by high levels of legal cynicism or a sudden change in the level of legal cynicism ( because of perceived mishandling of an event ) may exhibit not only a higher level of crime but also a greater unwillingness of citizens to report a crime to the police .
given the reactive nature of most police work , there are sound reasons for believing that citizens lack of cooperation and faith in the police are reflected in a lower rate of official police - based crime statistics though the actual rate may be higher .
although legal cynicism accounts for some of the variation in the rate at which citizens report a crime to the police , other factors are also involved .
the point here is not to offer legal cynicism as the only hypothesis or even test this conjecture as a hypotheseis .
rather , it is to provide some justification that there are credible _ a priori _ reasons to believe there is systematic variation in the reporting of crimes across jurisdictions ( and over time within the same jurisdiction ) and that such variation is one source of the ambiguity in police statistics . in the following sections , we give formal expression to this ambiguity using an approach which brings the fragile nature of police - based crime statistics to center stage .
we begin our analysis by examining the number of residential burglaries reported by the police to the north carolina state bureau of investigation s ( sbi ) 2010 uniform crime reports for the 10 most populous city - level jurisdictions in north carolina during the 2009 - 2011 period ( state bureau of investigation , 2012 ) .
the sbi statistics count both attempted and completed residential burglaries .
we verified that each of these 10 cities participated in the sbi s ucr program for each month of the 2009 - 2011 calendar years .
table 1 identifies the 10 cities included in our study ( column 1 ) along with the frequency of residential burglaries reported by each city s police department in 2009 ( column 2 ) , 2010 ( column 3 ) , and 2011 ( column 4 ) .
we denote residential burglaries reported by the police to the sbi - ucr program as @xmath0 .
.residential burglaries counted in the ucr @xmath1 [ cols="<,>,>,>",options="header " , ] a prominent aspect of the evidence in table 6 is that there is real variation in the number of persons per household in large north carolina cities . on average ,
asheville and wilmington have smaller household sizes ( @xmath2 persons per household ) while cary and high point have the highest density households ( @xmath3 persons per household ) .
it is possible , then , that two cities could have an identical rate of residential burglary when that rate is expressed in terms of population size but have different residential burglary rates when expressed in terms of the number of households in the city .
we now consider how much our inferences about residential burglary rates depend upon the issues we have considered in this paper .
we do nt expect much sensitivity to the uncertainty of the hierarchy rule since the adjustments are small .
it is less clear how sensitive our results will be to uncertainty about the size of the population ( including whether we scale by the number of persons or the number of households ) and uncertainty about the fraction of residential burglaries reported to the police . in order to estimate the actual rate of residential burglaries per 100,000 persons ( @xmath4 )
, we define its lower bound as : @xmath5 while the upper bound is : @xmath6 this interval estimate identifies the outer limits of what is possible in terms of the residential burglary rate per 100,000 persons assuming that @xmath7 and @xmath8 form the proper bounds on the size of the population , that each city s reporting rate falls within the 95% confidence interval of the ncvs - estimated reporting rate , and that our adjustments for the fbi s hierarchy rule are accurate .
we combine these estimates with the standard residential burglary rates per 100,000 population based on the information from tables 1 and 5 to produce the point and interval estimates in figure 1 .
for each of the 10 cities in figure 1 there are 2 sets of estimates : ( 1 ) the upper and lower bound ( interval ) estimates of the actual residential burglary rate ( @xmath4 ) for 2009 - 2011 ; and ( 2 ) the `` standard '' ( point ) estimates of the residential burglary rate per 100,000 population ( based on tables 1 and 5 ) for 2009 - 2011 .
comparisons between the point estimates over time within the same city implicitly assume that the reporting rate , @xmath9 is the same across the years while comparisons between point estimates of different cities implicitly assume that the reporting rate is constant between the cities being compared ( at the time they are compared ) .
while these assumptions seem implausibly strong , they are commonly invoked for both journalistic and research purposes .
we also consider the impact of adjusting for the number of households instead of the number of people and then placing residential burglaries on a scale per 1,000 households in each city during each year ( figure 2 ) .
this analysis relies on the information presented in tables 4 and 6 . to estimate the lower bound on the residential burglary rate per 1,000 households we obtain : @xmath10 and
the upper bound is given by : @xmath11 broadly speaking , the two sets of rate comparisons in figures 1 and 2 seem to tell similar stories . from this analysis ,
our major conclusion is that the major source of uncertainty in estimating residential burglary rates in these 10 north carolina cities over the 2009 - 2011 time frame is the reporting rate , and to a lesser extent , the size of the population .
it is worth considering a couple of example implications of our results .
suppose we set out to compare the burglary rates between charlotte and wilmington in 2009 .
using the standard approach for comparing the two cities , we would find that charlotte had a residential burglary rate of 1,051 per 100,000 persons while wilmington s rate is 1,184 ( nearly 13% higher ) .
some might use this evidence to say that wilmington had a higher residential burglary rate than charlotte in 2009 .
but upon further analysis , we find that the actual rate of residential burglaries per 100,000 population could plausibly lie in the @xmath12 $ ] interval in charlotte and the @xmath13 $ ] interval in wilmington . since these intervals overlap ( see figure 1 ) , the sign of the difference between charlotte and wilmington is not identified .
this lack of identifiability becomes even more prominent when we focus on household incidence of burglary .
for charlotte , the bounds on the residential burglary rate per 1,000 households in 2009 are @xmath14 $ ] ; for wilmington , the bounds are @xmath15 $ ] .
this is a marked increase in the degree of overlap between the charlotte and wilmington interval estimates .
we can attribute most of this difference to the fact that charlotte households had an average of 2.46 persons in 2006 - 2010 while wilmington households were smaller on average ( 2.19 persons ) over the same time period . in short , to speak about a clear difference in these two cities is an example of unwarranted certitude .
it is possible that charlotte s burglary rate is higher , lower , or the same as wilmington s .
considering plausible sources of uncertainty in the comparison , the data are simply not strong enough to tell us .
a comparison of charlotte and raleigh on the other hand leads us to a stronger set of conclusions . in 2011 , for example , charlotte s estimated residential burglary rate per 100,000 population based exclusively on the information in tables 1 and 5 was 818 while raleigh s rate was 597 ( a @xmath16 difference ) . we conclude that the difference between the burglary rates in charlotte and raleigh can not be explained by the uncertainties considered in this paper .
the residential burglary rate interval for charlotte in 2011 is @xmath17 $ ] while the interval for raleigh in the same year is @xmath18 $ ] .
as figure 2 shows , there is a similar pattern for burglary incidence scaled by the number of households . since these intervals
do not overlap , it seems credible to argue that charlotte s rate is higher than raleigh s rate in 2011 .
figures 1 and 2 are also helpful for displaying the over - time change within cities .
using the standard crime rate estimator , we can see that the point estimates of charlotte s residential burglary rate reveal what appears to be a meaningful decline from 2009 ( 1,051 ) to 2011 ( 818 ) ( a drop of about @xmath19 ) .
the problem is that there is some evidence that reporting rates could have also changed a good deal over the same time period .
what this means is that the burglary rate interval for charlotte ( accounting for the hierarchy rule , reporting rate uncertainty , and population size uncertainty ) is @xmath20 $ ] in 2009 and @xmath17 $ ] in 2011 .
since these intervals overlap we can not discern whether the charlotte burglary rate increased , decreased , or stayed the same over this time period .
a counterexample is provided by durham . in 2009 ,
the standard burglary rate estimate was 1,281 ; in 2011 that rate estimate increased to 1,472 an increase of @xmath21 .
our interval estimates suggest that this increase was real ; in 2009 , the interval was @xmath22 $ ] while in 2011 , the entire interval shifted up to @xmath23 $ ] . in the case of durham , we can confidently conclude that residential burglaries increased
why that increase occurred , of course , is a different question .
a good deal of contemporary discussion about local crime patterns in the u.s . is marred by unwarranted certitude about the numbers and rates underlying that discussion .
criminal justice officials , journalists , and even academic criminologists count crimes known to the police while ignoring key sources of uncertainty about those numbers . since the late 1960 s and early 1970 s , for example , it has been common criminological knowledge that many crimes are not reported to the police but somehow that knowledge ends up playing only a tangential role ( if any role at all ) in our public discourse about crime patterns at the local level .
part of the problem is that there has been little methodological attention to the task of expressing and transmitting uncertainty about crime patterns to policy and lay audiences ( manski , 2003:21 ) .
based on manski s work on bounds and partial identification , however , we think it will be useful for criminologists to begin reporting crime patterns in terms of a range of uncertainty that expresses both what is known and unknown about the numbers that are used to measure those patterns .
a key feature of the methods used here is that they explicitly abandon the goal of obtaining point estimates in favor of a more realistic and reasonable goal of obtaining interval estimates .
our approach provides one path by which criminologists can _ begin _ to reasonably express both what is known and unknown with current publicly available datasets .
another feature of our approach is that we move away from the `` incredible certitude '' problem described by charles manski ( 2011 ) the practice of developing unqualified and unjustifiably `` certain '' inferences based on weak data .
criminologists are often asked by the media to comment on small year - to - year movements in police - based crime statistics . in our conversations with other criminologists ,
we have noted that many feel quite uncomfortable characterizing this or that small movement in the crime rate .
the analysis in this paper illustrates why these feelings of apprehension are justified .
as eck and riccio ( 1979 ) observed over 30 years ago , a movement of a few percentage points in the police statistics may or may not reflect real changes in crime .
we think our approach to this problem is useful because it allows us to transmit our uncertainty especially to lay audiences in systematic ways that have not been obvious in the past .
still , there are limitations .
first and foremost , we believe our bounds on the probability that a residential burglary is reported to the police ( @xmath9 ) are a reasonable starting point but improving our understanding of this interval would be constructive .
this highlights an important direction for future research : achieving disciplinary consensus on the likely bounds for crime reporting probabilities should be a high priority .
one reviewer of a previous version of this manuscript criticized our reporting rate intervals as being too narrow .
that reviewer found it inconceivable that the local and national estimates would exhibit any particular comparability .
most of the data that can be used to check on this were collected in the 1970 s in a series of city crime surveys conducted by the national criminal justice information and statistics service ( 1974 , 1975 , 1976 ; see also levitt , 1998 ) and a research report by lauritsen and schaum ( 2005 ) .
while there is not much local data to go on , it appears from the weight of this evidence that most cities have residential burglary reporting rates that are within a reasonably proximate range of the national estimates of their time .
the reviewer s comment nonetheless highlights the need for greater understanding of how closely local reporting rates track what is observed nationally . and if the reviewer turns out to be correct then the burglary rate intervals estimated in our work will be too narrow ; a result which amplifies rather than diminishes our arguments .
we have considered several examples where point estimates based on conventional methods prove to be highly misleading . using those methods one would draw the conclusion that one city had a higher rate than another city or that a city s rate changed in a meaningful way from one year compared to another .
our analysis shows that in some of these comparisons , a plausible rival hypothesis can not be excluded : it is possible that the burglary rates are the same while only the reporting rate differs .
since the reporting rate , @xmath9 , is not identified we can only make assumptions about its value the data can not be used to resolve this ambiguity . only information that reduces our uncertainty about the rate at which residential burglaries are reported to the police in the two cities will resolve it .
the good news is that the development of this kind of information is feasible .
the national research council along with the bjs has recently considered a range of possibilities for improving on the small - area estimation capabilities of the ncvs ( groves and cork , 2008 ; bureau of justice statistics , 2010 ) .
most of the attention has focused on small - area estimation of victimization rates but further refinement of reporting rate estimates should also be a priority .
this is not a new idea eck and riccio ( 1979 ) emphasized the possibilities of this approach decades ago yet combining this emphasis with a focus on interval estimation of crime rates may prove to be a viable way forward .
a key benefit of this kind of information would be a substantial reduction of the uncertainty that is evident in our figures 1 - 2 .
it it noteworthy that we are able to make useful statements about residential burglary rates for north carolina cities because the state reporting program clearly identifies residential burglaries known to the police .
this is not done in the fbi s uniform crime report which presents counts of all burglaries both residential and commercial known to the police in a single number . and
we encounter difficulty using the fbi s burglary numbers since the ncvs only measures reporting behaviors for residential burglaries .
expansion of our approach to other crimes will require careful consideration of how the crimes described in the ucr relate to the victimization incidents counted in the ncvs .
there is a well - developed literature on this topic ( see , for example , blumstein et al . , 1991 , 1992 ; lynch and addington , 2007 ) but there will be some difficulties in ensuring that the reporting probability gleaned from the ncvs maps onto ucr crime categories in a meaningful way . in our view , the field will be well served by taking on these challenges .
we encountered a few other ambiguities in addition to the reporting probability ; namely , uncertainty due to the ucr s hierarchy rule , the size of the population and the question of whether to scale by the number of households or by the number of person ( gibbs and erickson , 1976 ) .
it is surprising how large some of the differences in population estimates were and this uncertainty should be considered in more detail .
we verified that each of the jurisdictions we studied participated in the state uniform crime reporting program each month of each year during the 2009 - 2011 calendar years ( but greensboro did not participate in the federal program in 2011 ) ; still we clearly have no way to verify the accuracy of the numbers reported by the police departments ( westat , 2010:vii - viii ) .
this issue is always a threat to analyses that rely on police - based crime statistics and our study is no exception . in our view
, it will be useful for criminologists to : ( 1 ) be aware of the kinds of uncertainties discussed in this paper ; ( 2 ) develop better information about uncertain parameters such as the probability of victimizations being reported to the police at the local level ; ( 3 ) create analytic methods that will formally incorporate and transmit key sources of uncertainty in the measurement of crime rates ; and ( 4 ) explore ways of conducting sensitivity analysis to assess the fragility of our results .
a fifth priority should be a program of research to consider how identification problems such as those discussed in this paper can be addressed within the framework of statistical models commonly used to estimate effects of social and economic changes on crime rates . logically , there is no difference between a comparison of burglary rates in charlotte and raleigh from 2009 to 2010 and the kinds of panel regresssion , difference - in - difference , and pooled - cross - sectional time series estimators commonly used to identify causal relationships in crime data .
all of the uncertainties discussed here are present in space - and - time crime regressions commonly estimated by criminologists . yet the issues discussed in this paper loom as major sources of uncertainty for these models .
we view our approach as an initial , constructive , and necessary step in the direction of a more balanced and informative use of aggregate crime statistics .
= 0.15 in addington , lynn a. ( 2007).using nibrs to study methodological sources of divergence between the ucr and ncvs . in _ understanding crime statistics : revisiting the divergence of the ncvs and ucr _ , pp .
225 - 250 .
new york : cambridge university press .
american society of criminology ( 2007).http://www.asc41.com / policies / policypositions.html[official policy position of the executive board of the american society of criminology with respect to the use of uniform crime reports data . ]
columbus , oh : american society of criminology .
baumer , eric p. and janet l. lauritsen ( 2010 ) .
reporting crime to the police 1973 - 2005 : a multivariate analysis of long - term trends in the national crime survey ( ncs ) and national crime victimization survey ( ncvs ) .
_ criminology _ , 48:131 - 185 .
pepper , john v. ( 2001 ) .
how do response problems affect survey measurement of trends in drug use ? in _ informing america s policy on illegal drugs : what we do nt know keeps hurting us _ , pp .
321 - 347 .
washington , dc : national academy press .
raghunathan , trivellore e. , dawei xie , nathaniel schenker , van l. parsons , william w. davis , kevin w. dodd and eric j. feuer ( 2007 ) . combining information from two surveys to estimate county - level prevalence rates of cancer risk factors and screening
. _ journal of the american statistical association _ , 102:474 - 486 .
slocum , lee ann , terrance j. taylor , bradley t. brick , and finn - aage esbensen .
neighborhood structural characteristics , individual - level attitudes , and youths crime reporting intentions . _ criminology _ , 48:1063 - 1100 . | we consider the problem of estimating the incidence of residential burglaries that occur over a well - defined period of time within the 10 most populous cities in north carolina .
our analysis typifies some of the general issues that arise in estimating and comparing local crime rates over time and for different cities .
typically , the only information we have about crime incidence within any particular city is what that city s police department tells us , and the police can only count and describe the crimes that come to their attention . to address this ,
our study combines information from police - based residential burglary counts and the national crime victimization survey to obtain interval estimates of residential burglary incidence at the local level .
we use those estimates as a basis for commenting on the fragility of between - city and over - time comparisons that often appear in both public discourse about crime patterns . |
it is a testament to the ingenuity of nature that at all length scales there exists a multitude of complex self - assembled patterns .
the formation of patterns , from the textured dimples in a grain of pollen to the labyrinth of creases in a coral colony to the ridges in a fingerprint , is consistent and repeatable @xcite ; the mechanisms that drive them are far from understood .
mastery of such processes would revolutionize the fabrication and design of novel materials with specific properties . by harnessing elastic instabilities in elastomeric membranes
@xcite , the once lofty goal of creating self - assembled complex patterns with long range order is now a one step closer . when an elastomeric membrane with a square lattice of circular holes ( with diameter roughly half the lattice spacing ) in it is uniformly swollen or , alternatively , is compressed hydrostatically , the holes deform into a diamond plate pattern .
the order persists for upwards of @xmath0 times the original lattice spacing with only phase - slip defects , which do not change the overall symmetry of the pattern .
it is interesting to note that the same diamond plate pattern emerges from membranes with vastly different lattice spacings , ranging from @xmath1 m to 1 cm . because the same mechanism causes patterns to form over such a wide range of length scales , this technique is well suited to create devices for many fields and industries . in particular ,
when a material is in the diamond plate state , it has a photonic band - gap @xcite ; hence , merely by compressing and relaxing a membrane , we can control its band structure .
this effect occurs in the highly non - linear regime of elasticity ; thus , only finite element simulations using specific models of nonlinear elasticity capture the entire process of the holes collapsing @xcite .
however , some simulations merely predict the orientational order of the collapsed holes and can not capture the details of their final shape . with the sole assumption that each hole collapses to some elongated shape ,
our model uses only linear elasticity to successfully predict the orientational order in the diamond plate pattern and the herringbone pattern formed from an underlying triangular lattice .
not only does our model shed light on the interactions in the system , it greatly facilitates the rational design of other patterns and devices . in the next section
, we review the linear theory of elasticity and consider a hydrostatically compressed membrane containing either a single hole or two holes .
the breakdown of linear elasticity illustrates the need for non - linear analysis . in section iii ,
we adopt the spirit of the theory of cracks and show that complex behavior can be explained by recasting the problem using linear dislocation theory .
while an explicit analytic description of the shape of a collapsed hole can not be reached through this method , each hole can be modeled as an elastically interacting distribution of parallel edge dislocations , or `` dislocation dipoles , '' whose centers are fixed to the center of their corresponding hole but are allowed to rotate freely . in the simple case of a quartet of holes , the resulting configuration exactly reproduces the unit cell for the diamond plate pattern . since elastic interactions are long ranged , the collective interactions of all the holes must be included to find the true groundstate of the system the inclusion of which additionally stabilize the pattern .
we end with technical appendices which argue the validity of our approximation .
the theory of linear elasticity describes the deformation and energetics of a solid body under external force or load .
while this theory was originally developed over a century ago and has been presented time and again @xcite , this brief tutorial will illuminate a few salient features as well as familiarize the reader to our notation .
the goal is to develop the framework for a generalization of hooke s law for solid three dimensional bodies .
when a solid is deformed , the displacement of every point , @xmath2 is described by the vector @xmath3 , such that its final positions are given by @xmath4 let us first consider two points separated by @xmath5 that are very close together . after being deformed
, their separation becomes @xmath6 , where @xmath7 using the einstein summation convention . by noting @xmath8
, we may rewrite this as @xmath9 where @xmath10 is the lagrangian strain tensor .
however , for small deformations we need only consider terms linear in @xmath11 and the linearized strain tensor is @xmath12 in the following , we will rely upon orthogonal coordinates , @xmath13 , with the diagonal metric @xmath14 @xcite .
in general , the orthogonal coordinates are nonholonomic but the benefit of orthogonality outweighs this complication .
what is the linearized lagrangian strain tensor in our new coordinates ? by definition @xmath15 and @xmath16 since @xmath3 is a vector , we may write it in either coordinate system : @xmath17 from which we have @xmath18 . in order to calculate the direction cosines
, we note that if @xmath19 , then @xmath20 we recognize the radicand in the denominator as @xmath21 and we have @xmath22d\xi_\mu\ ] ] using orthogonality , @xmath23 , it is straightforward to find @xmath24/(2h_\mu h_\nu),\end{aligned}\ ] ] where we only sum over the repeated indices @xmath25 and @xmath26 , not @xmath27 or @xmath28 associated with the scale factor @xmath29 . in the following we will adopt the einstein summation convention and , only when there is ambiguity , will we specify which indices are to be implicitly summed . in the spirit of hooke s law
, there is an energy cost associated with displacing every point from its equilibrium position .
while the general form is quite complicated , to first order the energy can be constructed from the linear strain tensor . because the strain tensor is a symmetric rank two tensor , the only two possible scalar invariants that can be constructed are @xmath30 and @xmath31 . because these terms are invariants of the system
, they can not depend on coordinate system . from this
we deduce the form of the energy density @xmath32 where @xmath33 and @xmath27 are lam coefficients . since any deformation may be written in terms of uniform or hydrostatic compression and pure shear , we rewrite the strain tensor , @xmath34 the hydrostatic compression is given by the first term , as it involves only the trace of @xmath35 . and
the second term describes pure shear because its trace is zero . in terms of these quantities ,
the energy becomes @xmath36 where the bulk modulus @xmath37 , and @xmath27 is the shear modulus . when a body is deformed , the displaced internal elements experience forces which tend to restore them to their equilibrium positions .
the volume element bounded by the surfaces @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , and @xmath43 experiences a force along any surface with normal @xmath44 , @xmath45 , where @xmath46 is the stress tensor and @xmath47 in equilibrium , the internal stresses of the system must balance the external forces exerted upon it ; thus , the equilibrium condition for the system is @xmath48 in curvilinear coordinates , this entire discussion can be repeated and the stress tensor in these coordinates is merely a transformation of the stress tensor in cartesian coordinates : @xmath49 where , again , there is no sum over @xmath27 or @xmath28 .
the equilibrium conditions become @xmath50\nonumber\\ & & \quad+ \frac{1}{h_\nu}\frac{\d \s_{\mu \nu}}{\d \xi_\nu } -f^{\textrm{ext}}_\mu = 0.\end{aligned}\ ] ] where we have followed the same procedure as in the derivation of @xmath51 which requires expanding the tensor in both the @xmath52 and @xmath53 frames .
the relation between stress and strain follows the argument in cartesian coordinaties . if the system is deformed an infinitesimal amount @xmath54 , the work done by the change in internal stresses is the force times the displacement , @xmath55 when integrated by parts , the work is @xmath56 note that the surface integral vanishes because @xmath57 at infinity . thus , @xmath58 by taking the total differential of equation ( [ bulk ] ) , @xmath59 , we can rewrite the stress tensor in terms of the strain tensor , @xmath60 or conversely , the strain in terms of the stress , @xmath61 similarly , it is instructive to consider a homogeneous deformation , which has constant stain tensor everywhere in the volume . for a @xmath62-dimensional solid ,
uniform pressure is applied to the faces with normals in the @xmath63 directions . in 3-dimensions , for example
, we consider the simple compression of a rod .
this implies that @xmath64 or @xmath65 .
thus , all off diagonal components of the strain tensor are zero , and the diagonal components are @xmath66 for all @xmath67 and @xmath68 .
the relative longitudinal compression is given by @xmath69 where @xmath70 is the @xmath62-dimensional young s modulus .
the ratio of transverse extension to longitudinal compression is given by the poisson ratio @xmath71 .
conversely , the bulk and shear moduli written in terms of the young s modulus and poisson ratio are , respectively , @xmath72 and @xmath73 . the stress tensor is given in terms of the strain tensor by , @xmath74 and the converse by , @xmath75 the conventional form of the energy is given in terms of @xmath76 and @xmath77 @xmath78 now that we have derived the relations between the stress and strain tensors , the equilibrium condition may be recast in terms of the displacement vector , @xmath79 if the force only acts through the surface , then @xmath80 vanishes in the bulk and we recover the result that @xmath81 from which it follows that @xmath82 . our ultimate goal is to study the effects of uniform tension on a thin sheet of elastic material with a square lattice of circular holes cut in it , as it models an elastic sheet that is uniformly swollen .
in general , the resulting deformations are constant across the thickness of the sheet and may be considered to be purely longitudinal .
this is because forces act primarily in the plane of the film , yielding the boundary condition @xmath83 .
henceforth , we will only consider two dimensional systems with planar deformations .
let us pause for a minute to derive the equilibrium conditions of this system .
since the deformations are constant throughout the thickness of the sheet , we may assume that @xmath84 and thus @xmath85 .
we are left with the two equilibrium equations @xmath86 where @xmath87 , and @xmath29 is the thickness of the film .
when no external body forces are present the equations of equilibrium reduce to @xmath88 or , @xmath89 viewing these both as equations of the form @xmath90 , we know that @xmath91 and @xmath92 and it follows that @xmath93 ( where @xmath94 is the totally antisymmetric tensor ) .
thus @xmath95 for some scalar @xmath96 , known as the airy stress function .
we thus have @xmath97 .
moreover , since @xmath98 , we have @xmath99 .
the relation between stress and strain , ( [ stress from strain ] ) implies that @xmath100 as a result , we deduce that the stress function satisfies the biharmonic equation , @xmath101 in orthogonal coordinates , we recast the components of the stress tensor in terms of the airy stress function by taking advantage of eq .
( [ stress1 ] ) and reexpressing @xmath102 in terms of @xmath103 , @xmath104,\ ] ] where @xmath105 because @xmath106 is a right - handed orthonormal basis .
equations ( [ airy ortho ] ) are solutions to the equilibrium equations , eq .
( [ eq ortho ] ) , and the airy stress function solves the biharmonic equation in _ orthogonal _ coordinates .
the simplest system to study is an infinite elastic sheet with a circular hole , of radius @xmath107 cut in it under uniform tension @xmath108 .
this problem naturally lends itself to polar coordinates , for which the equations for the stress function become , @xmath109 a standard procedure for solving such problems is to solve first for the deformation of a continuous sheet under the proper forces .
secondly , we solve a for a second stress function respecting the symmetry broken by the force with boundary conditions @xmath110 .
the final stress function is given by the sum of the two stress functions , where the matching condition is given by the stress free boundary condition at the edge of the hole .
+ the components of the stress tensor for a continuous elastic sheet under uniform tension @xmath111 are @xmath112 and @xmath113 , which , by integrating eqs .
( [ airy cartesian ] ) yield the stress function @xmath114 from which it follows that the components of the stress tensor are @xmath115 , @xmath116 , and @xmath117 .
clearly , rotational symmetry is broken in the @xmath118-direction . in order for the second stress function to respect the broken symmetry of the system , it must have the form @xmath119 since the stress function satisfies the biharmonic equation , we can easily integrate to find @xmath120 the first boundary conditions @xmath121 dictate that @xmath122 , leaving @xmath123 using the final boundary conditions , @xmath124 the remaining constants are found to be @xmath125 , @xmath126 , and @xmath127 .
note that even though there are only two equations for three unknowns , this system is not underdetermined because the constant @xmath128 can not depend on @xmath129 . assembling this ,
the components of the stress tensor are given by @xmath130 , \nonumber\\ \sigma_{\phi\phi}(r)&= & \frac{p}{2}\left[1+\frac{r^2}{r^2}-\left(1 + \frac{3 r^4}{r^4 } \right)\cos 2 \phi \right]\nonumber\\ \sigma_{r\phi}(r ) & = & -\frac{p}{2 } \left(1 + \frac{2 r^2}{r^2 } -\frac{3 r^4}{r^4}\right ) \sin 2 \phi .\end{aligned}\ ] ] which may be rewritten , using equation ( [ strain from stress ] ) , as components of the strain tensor @xmath131 recall , the strain tensor is the relative displacement of every element from its equilibrium position . in polar coordinates we have @xmath132 ,
@xmath133 and @xmath134 , from which it follows that the displacement vector is @xmath135\nonumber\\ u_\phi&= & -\frac{p}{2y_2}\frac{(r^2+r^2)^2+\g ( r^2-r^2)^2}{r^3 } \sin 2 \phi.\end{aligned}\ ] ] however , if we repeat the above process for an infinite sheet under hydrostatic compression ( or expansion ) , to linear order the rotational symmetry of this system is not broken .
consider an annulus of inner radius @xmath136 and outer radius @xmath137 under uniform hydrostatic compression with boundary conditions given by @xmath138 and @xmath139 .
clearly , the displacements are purely radial , and we need only solve @xmath82 , subject to the boundary conditions @xmath140 and @xmath141 . the displacements are given by @xmath142 ; from which it follows @xmath143.$ ] the boundary conditions determine the values of the constants @xmath144 and @xmath145 .
the solution to an annulus under hydrostatic compression is @xmath146 with the components of the stress tensor given by @xmath147 in the case of a finite sheet or a pipe under hydrostatic compression , there is the well known von mises buckling instability at a critical pressure where the circular hole deforms into an ellipse whose major axis is chosen at random @xcite .
this critical pressure scales with the ratio of the system size to the hole radius and , thus , diverges for large systems .
the system we are studying , on the other hand , has an underlying lattice which breaks the rotational symmetry of each hole .
a superposition of the above solutions would not account for the interaction between holes .
understanding the elastic interaction between holes in an elastic sheet was , during the first half of the last century , the subject of much research @xcite .
most of which was dedicated to finding the maximum stress felt along the perimeter of each hole . while linear elasticity may provide reasonable solutions to such analysis
, we will demonstrate that linear theory breaks down upon further investigation .
and @xmath148 are circles in bipolar coordinates.,width=336 ] the simplest system accounting for the interaction between holes is an infinite elastic sheet containing two holes of radius @xmath149 , whose centers are separated by distance @xmath150 . this sheet is then subjected to uniform tension @xmath151 .
the analysis of this system will closely follow that of ling @xcite .
bipolar coordinates , defined by @xmath152 for @xmath153 are the natural choice for this problem , as lines of constant @xmath154 or @xmath148 are circles in the @xmath155-plane defined by @xmath156 and @xmath157 .
the system of two equal holes corresponds to @xmath158 , @xmath159 , and @xmath160 .
when the system is under uniform tension @xmath151 , the components of the stress tensor are @xmath161 , @xmath162 , and @xmath163 . by integrating the eqs .
( [ airy cartesian ] ) , the stress function for an infinite system under uniform tension is @xmath164 using the method outlined in the previous section , we undertake the tedious calculation , detailed in appendix [ orthogonal ] , to find the equilibrium configuration of this system .
the results for uniform compression and tension are displayed in fig .
[ bipolar pics ] . upon further analysis of the compressed system , large enough values of @xmath151 yield overlapping solutions for the displacement vectors , signaling the breakdown of the linear theory .
[ bipolar pics ]
even were there no instability in the linear theory of two elastic holes , the sheer complexity of the equations would make calculations of increasing numbers of holes a nearly impossible task , and understanding the mechanism by which the holes collapse and the shapes they form requires a nonlinear theory of elasticity .
thus , we turn to the theory of cracks for inspiration . in the linear theory of elasticity , cracks can be described by a continuous distribution of parallel dislocations @xcite .
the stresses in a body due to a crack are the same as the stresses in an isotropic body with a distribution of dislocations with the same height profile as that of the crack . as a first approximation
, we model each of the collapsed holes as a pair of oppositely charged dislocations , known as a dislocation dipole @xcite .
this formalism allows us to recover the same physics by the simple numerical minimization of algebraic equations , once described by a complex system of coupled differential equations . and @xmath165 are separated by @xmath166.,width=312 ] our model system consists of an isotropic solid where each collapsed hole is represented by a thin line of material that has been taken out of the system , or a dislocation dipole .
the burgers vector for a dislocation dipole of strength @xmath167 with dipole vector @xmath168 located at @xmath169 is @xmath170,\ ] ] or in fourier space , @xmath171\nonumber\\ & \approx&ib q\hat{\boldsymbol{z } } \times \boldsymbol{d}\ , e^{i q r \cos \theta}\cos(\theta-\theta_0).\end{aligned}\ ] ] where @xmath172 is angle of @xmath173 and @xmath174 is the direction of the dipole and we have , in the spirit of the dipole approximation , taken the lowest order term in @xmath175 .
the interaction energy of two dipoles @xmath176 and @xmath165 both of strength @xmath167 separated by @xmath166 is given by @xmath177 \cdot \left[\boldsymbol{q}\times ( \hat{\boldsymbol{z } } \times \hat{\boldsymbol{d}}_2 ) \right]}{q^4 } \left(i q \cos(\th-\th_1)\right)\left(-i q e^{-iq r \cos \th } \cos(\th-\th_2)\right),\end{aligned}\ ] ] where @xmath176 is at the origin . after carrying out the integration
( see appendix [ dipole ] ) , the pairwise interaction between two dislocation dipoles is @xmath178 note that the interaction energy is invariant under @xmath179 and @xmath180 , which reaffirms each collapsed hole is represented by a line , not a vector . the total interaction energy of an array of dislocation dipoles is merely a sum of all pairwise interactions , because we are using linear theory .
the centers of the initial holes set the position of each dislocation dipole , but they are allowed to rotate freely .
the equilibrium state minimizes the free energy over the angle each dipole makes with respect to a fixed axis .
the simplest case consists of four holes located at @xmath182 . because all four holes have the same radius , their dipole vectors should have the same magnitude .
we minimize the energy functional , composed of the sum of six pairwise terms , @xmath183,\end{aligned}\ ] ] over each of the angles , which are measured with respect to the @xmath184-axis . minimizing with respect to the four angles we find @xmath185 and @xmath186 by symmetry and @xmath187
the minimum is @xmath188 while one might have postulated that the lowest energy configuration would have @xmath189 and @xmath190 , it turns out that the energy is slightly lowered if these angles are slightly shifted .
this is due to the finite size of our system as we shall see , for larger systems the dipoles align along the crystal axes and there is a boundary effect which distorts the dipole directions at the edges .
one might wonder how we can study larger and larger systems since the interaction only falls off as @xmath192 . because the interactions is between dipoles ,
the interaction energy at large distances decreases because the angle of the dipoles rotates around the circle .
indeed , consider the interaction of a single dipole at the origin with @xmath193 other dipoles in an @xmath194 lattice .
the angular dependence in ( [ dipole energy ] ) will wash out the power law if the dipoles rotate through @xmath195 uniformly . as we will see in the next section , this is precisely what happens as shown in fig .
[ nxn ] . for these larger arrays of holes ,
the sheer number of coupled equations makes it impractical to find solutions by hand , and we turn to numerical methods to find the orientations of the ground state of each lattice .
we find that the diamond plate order of the @xmath196 plaquette persists for larger and larger samples with increasing alignment along the lattice directions .
plaquette.,width=144 ] ( a ) , @xmath197 ( b ) , @xmath198 ( c ) , and @xmath199 ( d).,width=312 ] in experimental systems @xcite , the elastic sheet was stretched in a specific direction before allowing it to swell , leading to a background stress @xmath200 @xmath201 . to calculate the coupling energy between the stretching and the dipole angle
, we will rotate the system , such that the @xmath2-axis is defined by the direction of stretching , or @xmath202 and the dipole is located at the origin , or @xmath203 it is most appropriate to use the energy functional @xmath204 . in fourier space , the dipole term is given by @xmath205 @xcite and the stretching term by @xmath206 .
the energy is @xmath207 from which the coupling energy is @xmath208 the new term causing the dislocation dipoles to align with the direction of stretching competes with original interaction energy , favoring the diamond plate pattern .
following the same minimization procedure as before , we find that for small tensions the diamond plate pattern is only slightly perturbed , and for large tensions , the dislocation dipoles align along the direction of stretching , which may be seen in fig.[stretching ] . and @xmath209 system of holes are stretched by tension @xmath210 . for small tensions ( in ( a ) and ( d ) @xmath211 ) , the diamond plate groundstate is only slightly perturbed .
whereas for large tensions ( in ( b ) , ( c ) , ( e ) , and ( f ) @xmath212 ) , the holes align along the @xmath213-axis , the direction of stretching . in ( b ) and ( e ) the angle of stretching is @xmath214 from horizontal.,width=336 ]
we have created a model system for an elastic sheet with a square lattice circular holes cut out of it .
when the sheet is swollen , or , equivalently , subjected to uniform tension , the holes snap shut .
their major axes align into a diamond plate pattern with long ranged order .
while other methods of calculating the orientational order of the holes rely upon nonlinear elasticity theory and finite element simulations , we use simple linear elasticity theory to obtain the same results .
our system also corroborates experimental results of the sheet under external forces . this method
may easily extended to holes on other lattices .
it is difficult to extend it to an infinite lattice , since the minimization would then be over an infinite number of angles .
an ewald type summation may be used for an infinite system whose unit cell is the @xmath196 plaquette .
however , this is unlikely to lead to new insight , as the diamond plate order is clearly maintained for large systems . moreover , as we show in the appendix , the dipole interactions are the dominant terms even for more general elliptical holes .
we thank g.p .
alexander , b.g .
chen , c.d .
modes , s. yang , and y. zhang for useful discussions .
this work was supported by nsf mrsec grant dmr05 - 20020 .
for clarity s sake , we take a moment to explicitly write out the useful equations in bipolar coordinates . from equations ( [ strain ortho ] ) , we simply read off the components of the strain tensor @xmath215 the stress tensor may be written as a function of the airy stress function @xmath96 , from equation ( [ airy ortho ] ) @xmath216.\end{aligned}\ ] ] or in terms of the relative displacement vector , from eq .
( [ stress from strain ] ) @xmath217 at first glance , it seems an insurmountable goal to solve the differential equations for the displacements in terms of the airy stress function .
however , following @xcite , they become much more tractable if one considers the terms @xmath218 @xmath219 or , equivalently , we arrange equations to obtain , @xmath220 from @xmath221 and @xmath222 , respectively .
there exists a function @xmath223 which satisfies equation ( [ eqn f ] ) for which @xmath224 and @xmath225 .
using these equations , we eliminate @xmath226 and @xmath227 from equation ( [ saa - sbb ] ) @xmath228 the left hand side of this equation may be written in this manner because @xmath229 .
thus , the relative displacement vectors are given by @xmath230 recall , @xmath96 is the solution to the biharmonic equation @xmath232 with respect to the symmetries in our system .
the biharmonic equation when written in terms of the function @xmath233 has the simplified form , @xmath234 similarly , the components of the stress tensor are given by , @xmath235\frac{\chi}{h}\nonumber\\ \s_{\b\b}&= & \left [ \frac{1}{h}\frac{\d^2}{\d\a^2}-\frac{\sinh \a}{a } \frac{\d}{\d\a}-\frac{\sin \b}{a } \frac{\d}{\d\b}+\frac{\cos\b}{a}\right]\frac{\chi}{h}\nonumber\\ \s_{\a\b}&=&-\frac{1}{h } \frac{\d^2}{\d\a\d\b}\left(\frac{\chi}{h}\right).\end{aligned}\ ] ] our system , while undergoing uniform hydrostatic compression , is described by the stress function in equation ( [ airy bipolar 0 ] ) , @xmath236 or , equivalently , is given by the components of the stress tensor @xmath237 we aim to find solutions to the biharmonic equation that are even in both @xmath154 and @xmath148 .
thus , the airy stress function is given by @xmath238 where @xmath239 .
the components of the stress tensor corresponding to this airy stress function are @xmath240 \cos n\b \\ \s_{\b\b}^\1&=&\frac{c}{2
a } \left(\cosh 2 \a-2 \cosh \a \cos\b+\cos 2\b\right ) + \frac{1}{a}\phi_1(\a)-\frac{1}{2 a}\phi_1''(\a)\nonumber\\ & -&\sum_{n=1}^\infty \left[\phi_{n+1}''(\a)-2 \cosh \a \phi_n''(\a)+ \phi_{n-1}''(\a)+(n+2 ) \phi_{n+1}(\a)+2 \sinh \a \phi_n'(\a)+(n-2)\phi_{n-1}(\a)\right ]
\cos n\b \quad \ \ \ \\ \s_{\a\b}^\1&= & -\frac{c}{a } \sinh \a \sin\b -\frac{1}{2 a } \sum_{n=1}^\infty \left[g_{n+1}(a)-2 \cosh \a g_n(\a)+g_{n-1}(\a ) \right ] \sin n\b,\end{aligned}\ ] ] where @xmath241 and @xmath242 .
the boundary conditions require there be no stress at infinity ( @xmath243 ) , hence @xmath244=0,\ ] ] and the normal and tangential stresses must vanish along the edges of the holes , located at @xmath158 .
thus , the constants @xmath245 , @xmath246 and @xmath247 must satisfy the following recurrence relations for @xmath248 @xmath249 subject to the conditions @xmath250 using equation ( [ gn ] ) , we find @xmath251 , where @xmath252 and @xmath253 are roots of the characteristic polynomial @xmath254 yielding @xmath255
. however , the stress must be finite everywhere , thus , @xmath256 . equation ( [ cond4 ] ) completes the relation for @xmath257 as @xmath258 . because , equation ( [ fn ] ) is a non - linear recurrence relation , we must consider @xmath259 .
two of the roots of the characteristic polynomial for this equation are degenerate , the equation is @xmath260 .
from the boundary conditions , we determine @xmath261 from the definitions of @xmath262 and @xmath263 , we find that the coefficients @xmath245 , @xmath246 and @xmath247 satsify , @xmath264 and @xmath265\right\}^{-1}.\end{aligned}\ ] ] now that we have equations for the stresses everywhere , we may now solve for the field of relative displacement vectors .
recall , our function @xmath223 is given by , @xmath266,\end{aligned}\ ] ] where @xmath267 and @xmath268 .
to calculate the displacement field , we will need to know the following relations , @xmath269 \cos n \b\right)\\ \frac{\d\chi}{\d\b}&=&h \left(\sin \b(-p h \cosh \a+c)-\sum_{n=1}^\infty\phi_n(\a)\left[\frac{h}{a}\sin
\b \cos n \b+n \sin n \b\right]\right)\\ \frac{\d g}{\d\a}&=&-\frac{2 h}{1+\s } \left(c \left(\pi \frac{h}{a } \sinh \a \cos\b + \sin \b\right)+\sum_{n=1}^\infty \left[\frac{h}{a}\psi_n(\a ) \sinh
\a -\psi_n'(\a)\right ] \sin n\b \right)\\ \frac{\d g}{\d\b}&=&-\frac{2h}{1+\s}\left(c \left(\pi \frac{h}{a}\cosh \a \sin \b-\sinh \a \right ) + \sum_{n=1}^\infty \psi_n(\a)\left[\frac{h}{a}\sin \b \sin n \b - n \cos n \b\right ] \right).\end{aligned}\ ] ] these equations , together with eqs . (
[ bipolar displacements ] ) , complete our description of the system of two holes under hydrostatic compression , which may be seen in fig .
[ two holes ] .
it should be noted that in the compressed system there is an instability for large enough values of @xmath151 wherein the displacements intersect each other , causing overlap in the system .
due to the nature of problems in the theory of linear elasticity theory , the most useful identities involve the directional cosines relating the @xmath270 to the @xmath271 coordinates .
the unit vectors in the new system are given by @xmath272 and @xmath273 .
the orthogonality condition @xmath274 implies @xmath275 the directional cosines are related because @xmath276 or @xmath277 where the sign is chosen such that both @xmath278 and @xmath279 form right - handed orthonormal triads .
thus , under the change of coordinates @xmath280 , a rank-2 tensor @xmath281 transforms as @xmath282 , which may be written as @xmath283 the derivatives of the directional cosines can be made from linear combinations of derivatives of the orthogonality condition @xmath284 and the scale functions .
for example , to find @xmath285 first note there are two ways of obtaining this derivative ; directly , @xmath286 and by taking the derivative of the product of @xmath154 scale function and orthogonality condition with respect to @xmath154 @xmath287 where we have made use of the definition @xmath288 next multiplying eqn .
[ deriv1 ] by @xmath289 and eqn .
[ deriv2 ] by @xmath290 and taking their sum , this becomes @xmath291 by following a the same procedure , the formul for the derivatives of the directional cosines are @xmath292 where @xmath293 and @xmath294 and @xmath295 .
note that the cross partial derivatives of the directional cosines are equal , @xmath296 which leads to our final identity @xmath297 we include these for completeness .
these were necessary for us to study the airy stress formalism in orthogonal coordinates and we did not find these , presumably known identities , in any reference . consider the general set of orthogonal coordinates , @xmath298 .
the new basis preserves length of the differential line element , @xmath299 defining the scale functions @xmath300 and @xmath301 however , it is often more useful to consider cartesian coordinates as functions of the new orthogonal ones , @xmath302 , which yield an equivalent statement of the scale functions @xmath303 and @xmath304 by transforming from orthogonal back to cartesian coordinates , the differential line element gives @xmath305 and @xmath306 ; from which , we obtain the first set of identities : @xmath307
we devote this appendix to the technical details of the mathematical manipulation required to compute the energetics of a lattice of dislocation dipoles . despite the complex form of the integral in the dislocation dipole interaction energy , when completed it has a surprisingly simple form .
the integral becomes tractable by transforming to polar coordinates , and then manipulating the trigonometric functions . in polar coordinates ,
the integral in eq .
( [ burgers integral ] ) becomes , our goal is to prove that we need only consider the first order dipole - dipole term when considering collapsed holes . for simplicity sake
, we will study an elongated shape that is symmetric about both its semi - major and semi - minor axes , ( see fig .
[ dipole expansion ] ) . here , we consider only shapes for which the ratio of minor and major axes @xmath311 in the theory of cracks , the height profile of a crack @xmath312 may be constructed from a continuous distribution of finite parallel edge dislocations .
a finite edge dislocation of length @xmath313 can be thought of as a two infinite edge dislocations of the same strength but opposite charge which is given by burgers vectors situated at @xmath314 and @xmath315 in other words , a dislocation dipole . given the height profile of a shape , it is trivial to construct it from such dislocation dipoles , @xmath316 thus , the interaction energy between such a shape made from dislocation dipoles and a single dislocation dipole @xmath165 located a distance @xmath317 away is given by : by repeated application of the law of cosines , the functional forms of @xmath319 @xmath320 and @xmath321 are : @xmath322 the energy density is @xmath323 where @xmath324 this may be vastly simplified by expanding the angular terms : @xmath325 the energy density becomes : @xmath326 where @xmath327 and @xmath328 and the coefficients @xmath329 are given by @xmath330 @xmath331 @xmath332 @xmath333 and @xmath334 while the above energy is for general shape of dislocations , we choose a shape to do the actual calculation . for simplicity sake
, we choose an ellipse of major axis @xmath335 and minor axis @xmath336 .
thus , @xmath337 . with the change of variables , @xmath338 our energy integral becomes : @xmath339 expanding the denominator for @xmath340 we find , @xmath341 now we need only do the integral @xmath342 for , since @xmath343 is even , this integral is zero for odd integer @xmath344 this is very simple using beta functions , which we may see by the change of variables @xmath345 : @xmath346 the interaction energy is thus : @xmath347 while this appears to be a complicated expression , let us , for the moment consider only the first order term in @xmath348 : @xmath349 this does not appear to have the same functional form as the original energy .
however , with the help of some trigonometric identities , we begin to simplify the energy : @xmath350 we now use these identities in the energy to find : @xmath351 now we substitute @xmath352 to find , @xmath353 so we have found that to first order in @xmath354 an ellipse of major axis @xmath62 and minor axis @xmath336 interacts with a dislocation dipole of strength @xmath355 a distance @xmath136 away like a dislocation dipole with effective dipole strength @xmath356 higher order terms contain more powers of @xmath354 , which can be neglected for very thin shapes or for dipoles that are very far away .
we use the convention @xmath357 to represent the @xmath358 component of a cartesian coordinate system , where the subscript can be any roman letter .
components of another orthogonal coordinate system are denoted by @xmath13 , for any greek subscript . | recent experiments have exploited elastic instabilities in membranes to create complex patterns .
however , the rational design of such structures poses many challenges , as they are products of nonlinear elastic behavior .
we pose a simple model for determining the orientational order of such patterns using only linear elasticity theory which correctly predicts the outcomes of several experiments .
each element of the pattern is modeled by a `` dislocation dipole '' located at a point on a lattice , which then interacts elastically with all other dipoles in the system .
we explicitly consider a membrane with a square lattice of circular holes under uniform compression and examine the changes in morphology as it is allowed to relax in a specified direction . |
the potential of hi ly@xmath0 to identify young galaxies at high redshift was noted over 40 years ago by @xcite .
ly@xmath0 corresponds to the @xmath9 transition and is in principle the strongest emission line originating in an hii region . in the absence of dust , the ly@xmath0/h@xmath0 line intensity ratio is expected to be in the range @xmath10 for recombination theory cases b - a @xcite , and the ly@xmath0 equivalent width , @xmath11 , is expected to be in the range @xmath12 for metallicities between @xmath13 , a normal initial mass function , a constant star formation rate of 1 m@xmath14yr@xmath7 , and ages below 10 myr @xcite . in the case of a single burst of star formation , the range of @xmath11 is @xmath15 under the same conditions .
a few times @xmath16 yr after a burst of star formation , @xmath11 drops to zero @xcite .
the above predictions do not account for the underlying stellar contribution at ly@xmath0 .
this contribution is not a concern in galaxies dominated by populations of o and early b stars . for reference ,
the mean ly@xmath0 equivalent width of the eight early and mid - o supergiants in @xcite is @xmath171 . on the other hand
, the correction can be significant for stellar populations dominated by a and late b stars .
in addition , the above predictions do not account for the effects of ly@xmath0 radiation transfer in the medium surrounding the starburst .
due to its resonant nature , a ly@xmath0 photon is absorbed and re - emitted multiple times in hi .
resonant trapping reduces the mean free path of the ly@xmath0 photons increasing significantly their probability of being destroyed by dust , shifted in frequency , or transformed to two - photon emission . on the other hand , some effects are favorable to the escape of ly@xmath0 photons from galaxies .
neutral gas outflows doppler - shift the ly@xmath0 photons from the optically thick line core to the optically thin line wings making it possible for some ly@xmath0 photons to escape @xcite .
ly@xmath0 photons may also escape through holes of sufficiently low dust and hi column densities @xcite . in spite of the various ly@xmath0
attenuation effects , today , star - forming galaxy candidates have been detected out to redshift @xmath18 using various techniques .
galaxies with redshifts of @xmath19 are detected via the drop - out technique based on the opacity at the lyman - limit of the galaxy and the intervening intergalactic medium ( igm ; @xcite ) .
galaxies with @xmath20 are detected via the drop - out technique based on the opacity of the igm at ly@xmath0 @xcite .
galaxies with @xmath21 are also detected using deep narrow band surveys that target the ly@xmath0 emission - line @xcite .
finally , galaxies at the highest redshifts are detected using gravitationally lensing clusters @xcite .
hundreds of galaxy candidates have been spectroscopically confirmed @xcite , and a few galaxies show impressively high rest - frame equivalent widths of @xmath22150 @xcite , which could alternatively be due to the presence of an agn @xcite . in principle , after removing the observational and astrophysical biases , ly@xmath0 can be used for various applications .
1 ) to probe cosmic star formation rates @xcite .
2 ) to probe the ionization fraction of the intergalactic medium during the final stages of re - ionization @xcite .
3 ) to probe the large - scale structure @xcite .
4 ) to identify potential hosts of population iii star formation @xcite . 5 ) and to constrain the nature and evolution of high - redshift galaxies @xcite .
however , the use of ly@xmath0 as a cosmological tool is limited by our understanding of the astrophysical processes that regulate the escape of ly@xmath0 photons from galaxies and affect them on their way to telescopes .
the ly@xmath0 escape problem can be approached from two complementary angles , doing statistically significant studies of the global properties of the sources and their cosmological context , or studying individual sources in detail .
the former is better done in the distant universe , whereas the latter can only be accomplished locally , where the effect of the intergalactic medium is a non - issue , the kinematical and spatial resolutions are higher , and ancillary data is abundant .
the challenges of studying ly@xmath0 locally are the necessity of a uv space - based observatory optimized for studying ly@xmath0 ; contamination with the geocoronal ly@xmath0 line , which saturates the spectra at very low redshift ; and contamination with the milky way galaxy ly@xmath0-line trough .
indeed , the interstellar hi in our galaxy produces a damped absorption along many sight - lines , which hides the potential ly@xmath0 line from star - forming galaxies with redshifts below a few @xmath1 km s@xmath7 .
there are several observational ly@xmath0 studies of galaxies at low - redshift ( [email protected] ) .
@xcite studied low resolution spectra of 21 galaxies taken with the _ international ultraviolet explorer _ ( _ iue _ ) and studied the effects of reddening and metallicity on @xmath11 , and ly@xmath0/h@xmath24 .
they found no correlation between these quantities and the uv extinction , the balmer decrement , or the oxygen abundance .
they concluded that rather than the amount of dust , the ism geometry determines the ly@xmath0 escape fraction .
unfortunately , the spectral resolution of _ iue _ is insufficient to study the effect of gas outflows on the ly@xmath0 escape .
@xcite studied high - dispersion _ hubble space telescope _
( _ hst _ ) ghrs spectra of eight nearby hii galaxies .
they found four emitters , all of which have a velocity offset between the hi and the hii gas of up to 200 km s@xmath7 , while the non - emitters show no velocity offset .
however , their study suffers from small number statistics .
@xcite analyzed the optical spectra of a sample of 31 @xmath25 ly@xmath0 emitters identified by deharveng et al .
( 2008 ) in low resolution ( @xmath26 ) _ galaxy evolution explorer ( galex ) _ data .
they tried to reproduce the stellar - absorption corrected ly@xmath0/h@xmath0 and h@xmath0/h@xmath24 ratios using different dust geometries and assumptions for the ly@xmath0 scattering in hi .
they found that the ratios are well reproduced by a clumpy dust distribution , while a uniform dust screen model results in ly@xmath0/h@xmath0 larger than the case b recombination theory value .
unfortunately , the spectral resolution of _ galex _ is insufficient for kinematical studies , and the latter authors were unable to assess the relative importance of dust geometry and outflows .
@xcite used a 3d ly@xmath0 radiation transfer code to model _
observations of the very metal - poor dwarf galaxy i zw 18 , which shows ly@xmath0 in absorption .
their analysis shows that it is possible to transform a strong ly@xmath0 emission of @xmath27ly@xmath0)@xmath2860 into a damped absorption , even with low extinction , if the hi column density is large .
finally , @xcite used _
hst _ to produce ly@xmath0 , h@xmath0 , and uv continuum maps of six galaxies covering a wide range in luminosity and metallicity , including known ly@xmath0 emitters and non - emitters .
they found that the bulk of ly@xmath0 emerges in a diffuse component resulting from scattering events .
they also found the simultaneous presence of ly@xmath0 in emission and absorption within spatial scales ranging from a few @xmath29 pc to a few @xmath1 pc .
although these studies have addressed main factors affecting the ly@xmath0 escape , we still do not understand the relative importance of starburst phase , gas kinematics , gas geometry , and dust extinction , in determining the profile and strength of the ly@xmath0 line from star - forming galaxies .
we took advantage of the high sensitivity and medium resolution of the cosmic origins spectrograph ( cos ) aboard _ hst _ in order to obtain ly@xmath0 spectroscopy of a sample of 20 galaxies located a mean redshift of 0.03 and covering a broad range of luminosity , oxygen abundance , and reddening .
we used these and ancillary data for establishing correlations between ly@xmath0 and fundamental galaxy properties . in section 2
we describe the sample , the observations , and the ancillary data ; in section 3 we present our analysis ; in section 4 we discuss our results ; and in section 5 we summarize and conclude .
the galaxies in our sample were h@xmath0-selected from the kitt peak international spectroscopic survey data release ( kissr ] , @xcite ) .
the sample is composed of 12 irregular and eight spiral galaxies distributed in the redshift range @xmath30 . at these redshifts , we avoid contamination with the geocoronal and milky way ( mw ) ly@xmath0 lines .
we cover a factor of three in redshift , which makes it possible to study the effect of distance on the observed properties of ly@xmath0 .
in addition , we cover a broad range in oxygen abundance , @xmath31 , which makes it possible to study the effect of metallicity on the ly@xmath0 escape .
metallicity is expected to correlate with the dust content , and dust destroys ly@xmath0 photons .
we also cover a wide range in h@xmath0 equivalent width , @xmath32 .
the value of @xmath33 is sensitive to the starburst evolutionary phase and to the initial mass function ( imf ) , e.g. , see @xcite . finally , we cover over an order of magnitude in _ galex _ @xcite fuv continuum luminosity , @xmath34 erg s@xmath7 @xmath7 ( uncorrected for reddening ) . for a given star formation history ( sfh ) , age , and imf ,
@xmath35 is proportional to the stellar mass in the population dominating the fuv emission of the galaxy .
therefore , @xmath35 can be used to determine the star formation rate of this population ( sfr = stellar mass / time since beginning of star formation ) .
table [ tab1 ] summarizes the properties of the sample .
the morphological types are based on sdss images .
the h@xmath0 equivalent widths are from the sdss spectra when available or from the kiss data otherwise , as indicated in column ( 10 ) of table [ tab1 ] .
llcccccccc 40 & irr & 0.026 & 112 & -18.0 & 0.03 & 8.18 & 58 & 39.29 & 1 + 108 & irr & 0.024 & 101 & -17.6 & 0.01 & 8.17 & 108 & 39.22 & 1 + 178@xmath36 & irr & 0.057 & 246 & -20.1 & 0.01 & 8.52 & 100 & 40.13 & 1 + 182 & irr & 0.022 & 96 & -17.8 & 0.01 & 8.34 & 114 & 39.33 & 1 + 218 & s & 0.021 & 90 & -19.4 & 0.01 & 8.87 & 66 & 39.65 & 1 + 242 * & irr & 0.038 & 162 & -19.4 & 0.01 & 8.38 & 435 & 40.25 & 1 + 271 * & irr & 0.023 & 97 & -17.6 & 0.02 & 8.39 & 29 & 39.13 & 2 + 298 * & s & 0.049 & 210 & -20.0 & 0.02 & 9.07 & 51 & 40.06 & 1 + 326 * & s & 0.028 & 120 & -20.0 & 0.02 & 8.74 & 78 & 39.65 & 2 + 1084 * & s & 0.032 & 137 & -20.5 & 0.05 & 8.79 & 87 & 39.7 & 1 + 1567 * & s & 0.045 & 194 & -20.0 & 0.02 & 8.75 & 29 & 40.18 & 2 + 1578 * & irr & 0.028 & 120 & -19.6 & 0.02 & 8.14 & 576 & 40.29 & 1 + 1637 & irr & 0.035 & 149 & -19.6 & 0.01 & 8.55 & 34 & 39.92 & 1 + 1785@xmath37 & irr & 0.021 & 90 & -17.6 & 0.01 & 7.99 & 199 & 39.07 & 1 + 1942 & irr & 0.039 & 169 & -19.3 & 0.03 & 8.59 & 25 & 39.83 & 1 + 2019 & irr & 0.034 & 146 & -17.9 & 0.02 & 7.96 & 465 & 39.48 & 1 + 2021 & s & 0.04 & 171 & -19.6 & 0.02 & 8.79 & 79 & 39.93 & 1 + 2023 & s & 0.036 & 155 & -19.4 & 0.02 & 9.00 & 58 & 39.77 & 1 + 2110 & irr & 0.025 & 108 & -17.3 & 0.01 & 7.86 & 511 & 39.36 & 2 + 2125 & s & 0.025 & 108 & -18.7 & 0.01 & 8.76 & 85 & 39.31 & 1 + min & & 0.021 & 90 & -20.5 & 0.01 & 7.86 & 25 & 39.07 & + max & & 0.057 & 246 & -17.3 & 0.05 & 9.07 & 576 & 40.29 & + + [ tab1 ] lcccc 40 & 2011/06/11 02:07:49 & 2011/06/11 03:26:08 & 1105 & 1105 + 108 & 2011/07/31 04:47:42 & 2011/07/31 05:09:24 & 1105 & 1105 + 178 & 2011/06/17 20:55:13 & 2011/06/17 21:16:55 & 1105 & 1105 + 182 & 2011/07/26 08:13:51 & 2011/07/26 08:35:33 & 1105 & 1105 + 218 & 2010/02/23 07:33:10 & 2010/02/23 09:09:06 & 1111 & 1111 + 242 & 2009/12/26 23:56:45 & 2009/12/27 00:18:33 & 1111 & 1111 + 271 & 2011/05/07 17:41:45 & 2011/05/07 18:03:27 & 1105 & 1105 + 298 & 2010/01/01 12:41:22 & 2010/01/01 13:03:04 & 1105 & 1105 + 326 & 2011/04/22 13:36:31 & 2011/04/22 13:58:13 & 1105 & 1105 + 1084 & 2010/02/05 05:13:19 & 2010/02/05 05:35:01 & 1105 & 1105 + 1567 & 2010/08/13 05:09:19 & 2010/08/13 05:31:46 & 1150 & 1150 + 1578 & 2010/02/18 22:07:27 & 2010/02/18 22:29:57 & 1153 & 1153 + 1637 & 2009/10/24 09:51:17 & 2009/10/24 10:13:44 & 1150 & 1150 + 1785 & 2011/08/09 10:38:26 & 2011/08/09 11:00:55 & 1142 & 1142 + 1942 & 2011/06/13 22:46:13 & 2011/06/13 23:08:28 & 1128 & 1128 + 2019 & 2011/06/19 15:57:20 & 2011/06/19 16:19:49 & 1142 & 1142 + 2021 & 2009/11/14 00:02:53 & 2009/11/14 00:25:17 & 1147 & 1147 + 2023 & 2011/06/16 14:29:53 & 2011/06/16 14:52:08 & 1128 & 1128 + 2110 & 2011/06/16 16:16:21 & 2011/06/16 16:38:50 & 1142 & 1142 + 2125 & 2011/06/16 17:52:32 & 2011/06/16 18:15:01 & 1142 & 1142
+ + [ tab2 ] we are interested in galaxies where the principal photoionization source is the strong uv radiation from ob stars .
this helps in the interpretation of the observed ly@xmath0 properties because we do not need to worry about non - stellar ionizing sources such as active galactic nuclei ( agns ) , which account for some of the high ly@xmath0 ews at @xmath38 @xcite , and low - ionization nuclear emission - line regions ( liners ) , whose nature is not well understood @xcite .
we verified the spectral class of our galaxies using the so - called bpt diagnostic diagram ( baldwin , phillips & terlevich 1981 ) .
this was not possible for kissr 271 , for which an optical spectrum was unavailable .
as figure [ fig1 ] shows , most galaxies are dominated by star formation , but kissr 1084 may be of composite type . our observations are part of cos - gto programs 11522 and 12027 ( p.i .
. descriptions of the cos and its on - orbit performance can be found in @xcite .
we used one orbit with _ hst _ to observe the brightest fuv knot detected by _
galex _ in each galaxy .
we acquired the targets using the acq / image mode and mirror a. therefore , we obtained two images of each galaxy in the observed - frame wavelength range @xmath39 .
one after the initial telescope pointing , and the second after the final telescope pointing .
we then used the g130 m grating to obtain a spectrum of each target in the observed - frame wavelength range @xmath40 .
we achieved continuous wavelength coverage by using central wavelengths , @xmath81291 and @xmath81318 at the default focal plane offset position ( fp - pos=3 ) .
table [ tab2 ] lists the epochs and exposure times corresponding to these central wavelengths .
the spectra were taken through the circular primary science aperture ( psa ) , which is 2.5 in diameter .
they contain the aberrated light from objects located up to 2 or up to @xmath41 kpc from the aperture s center , depending on the distance to the galaxy .
the data were processed with calcos v2.13.6 and combined with the custom idl co - addition routine described in @xcite .
figure [ fig2 ] shows the nuv target acquisition ( ta ) confirmation images of the galaxies with the cos aperture overlaid .
the center of the aperture is indicated by the small black cross . in 12 cases , a clear source centroid associated with young star clusters
is present .
however , in eight cases , there is no clear centroid of the emission .
this is due to the short exposure times of the ta images and in two cases , i.e. , for kissr galaxies 40 and 1567 , to ta failures . in spite of the ta failures ,
spectra were taken .
the galaxies that turned out to have net ly@xmath0 emission are marked with asterisks .
hereafter , we will call the latter galaxies the emitters . note that emitters kissr 271 , 326 , and 1567 , do not show source centroids . also note that in spite the ta failure , kissr 1567 turned out to be an emitter .
this is in agreement with the finding of @xcite that the ly@xmath0 emission is not necessarily coincident with regions of high fuv surface brightness .
we obtained information about the morphology and orientation of the galaxies from archival images . at the time of our analysis
, there were no _ hst _
images available for our targets .
the galaxies were imaged by _
galex _ in the fuv at low spatial resolution ( @xmath42 ) . unfortunately , in these images
, the galaxies look like large blobs .
however , we found useful sloan digital sky survey ( sdss , @xcite ) images of all the targets in our sample .
these are shown in figure [ fig3 ] , where we have overlaid the cos footprint at the approximate location of the telescope pointing and marked the emitters with asterisks .
figure [ fig3 ] shows that most spiral galaxies in our sample have low inclinations except kissr 2125 . while some low - inclination spirals have ly@xmath0 in emission , others do nt .
kissr 1084 , the spiral of composite spectral class is an emitter , and as expected , the highly inclined spiral is a non - emitter .
our two most metal - poor irregulars , kissr 2019 and 2110 , are ly@xmath0 absorbers , but irregulars kissr 242 and 1578 are ly@xmath0 emitters . in summary
, there are emitters and absorbers among both morphological types .
finally , note that the cos aperture only covers an area of @xmath41 kpc in radius , depending on the distance to the galaxy .
this needs to be considered when comparing with ly@xmath0 observations at higher redshifts , since at higher redshift , more of the galaxy is enclosed within the aperture and @xcite found that the bulk of the ly@xmath0 photons from nearby galaxies are in the diffuse halo .
we used ancillary optical spectroscopy of the galaxies for determining their redshift , spectral class , metallicity , reddening , sfr , and wolf - rayet ( wr ) star content .
sixteen targets have spectra from the sdss seventh data release @xcite .
the sdss spectra are of higher spectral resolution than the kissr spectra , and they were taken through a circular fiber of @xmath43 in diameter that closely matches the size of the cos aperture .
the difference between the sdss and the cos pointings is relevant to the interpretation of the ly@xmath0/h@xmath0 ratio , and to the comparison of the strength of ly@xmath0 and the value of @xmath44(h@xmath0 ) .
indeed , the cos aperture encompasses only a portion of each galaxy .
table [ tab3 ] lists the kissr identifier ( column 1 , the emitters are marked with asterisks ) , the right ascension ( ra , j2000 ) and declination ( dec , j2000 ) of the first _
hst _ pointing ( columns 2 and 3 ) , the ra and dec of the sdss spectroscopic pointing ( columns 4 and 5 ) , and the difference between the cos and the sdss pointings ( columns 6 - 8 ) . as table
[ tab3 ] shows , on average , the sdss pointings are within 0.3 `` of the cos initial pointings , with a standard deviation of 0.4 .
since the diameter of the cos aperture is 2.5 '' , we do not expect the difference in pointings to constitute a major issue .
for kissr galaxies 326 , 1567 , and 2110 , which do not have sdss spectra , we were able to recover the optical @xmath45-wide slit spectra of @xcite .
unfortunately , we do not have an optical spectrum for kissr galaxy 271 .
lccccccc 40 & 185.59833 & 29.44383 & 185.59825 & 29.44394 & 8e-05 & -1e-04 & 0.49 + 108 & 190.98042 & 29.36964 & 190.98044 & 29.36961 & -2e-05 & 3e-05 & 0.13 + 178 & 195.42306 & 29.38118 & 195.42304 & 29.38135 & 2e-05 & -2e-04 & 0.62 + 182 & 195.60688 & 28.85808 & 195.60688 & 28.85807 & -5e-06 & 1e-05 & 0.05 + 218 & 197.31725 & 29.36739 & 197.31703 & 29.36766 & 2e-04 & -3e-04 & 1.24 + 242 * & 199.01625 & 29.38161 & 199.01632 & 29.38169 & -7e-05 & -8e-05 & 0.38 + 271 * & 200.42000 & 28.88308 & & & & & + 298 * & 202.45746 & 29.57972 & 202.45748 & 29.57972 & -2e-05 & 2e-06 & 0.08 + 326 * & 204.81708 & 28.87361 & & & & & + 1084 * & 252.27192 & 29.75878 & 252.27195 & 29.75878 & -3e-05 & -2e-06 & 0.12 + 1567 * & 201.22750 & 43.79889 & & & & & + 1578 * & 202.18354 & 43.93069 & 202.18357 & 43.93070 & -3e-05 & -6e-06 & 0.10 + 1637 & 205.50002 & 42.76004 & 205.50038 & 42.76000 & -4e-04 & 4e-05 & 1.30 + 1785 & 216.62029 & 43.47083 & 216.62029 & 43.47082 & 2e-06 & 1e-05 & 0.05 + 1942 & 230.70200 & 43.77347 & 230.70197 & 43.77347 & 3e-05 & 2e-06 & 0.11 + 2019 & 236.43554 & 44.26439 & 236.43552 & 44.26439 & 2e-05 & -1e-06 & 0.08 + 2021 & 236.46992 & 44.26322 & 236.46991 & 44.26322 & 7e-06 & 2e-06 & 0.03 + 2023 & 236.67746 & 43.24367 & 236.6774 & 43.24359 & 6e-05 & 8e-05 & 0.35 + 2110 & 242.04625 & 43.63175 & & & & & + 2125 & 242.58508 & 43.00975 & 242.58507 & 43.00975 & 1e-05 & 0e+00 & 0.05 + + [ tab3 ]
the cos spectra include the intrinsic lines ly@xmath0 and a number of low - ionization ( lis ) and higher - ionization interstellar lines .
we are interested in comparing the offsets of these lines with respect to h@xmath0 .
this is for two reasons .
first , as mentioned in the introduction , velocity offsets between the hii gas traced by h@xmath0 and the hi gas traced by the lis lines have been showed to favor the escape of ly@xmath0 from galaxies .
second , in expanding - shell radiation transfer models the ly@xmath0 emission line is redshifted to a few times the shell expansion velocity ( e.g. , @xcite ) .
a similar offset is observed at high redshift when comparing the ly@xmath0 redshift to the blueshift of the lis lines , which give the expansion velocity ( e.g. @xcite ) .
thus , offset measurements are useful for establishing how the ly@xmath0 strength relates to the interstellar gas kinematics and for constraining radiation transfer models .
the accuracy of the velocity offset measurements depends on the accuracy of the wavelength zero point and on the effective spectral resolution of the data . in ultraviolet spectroscopic studies of low - redshift galaxies ,
it is common to use geocoronal emission lines to check if the wavelength zero point is shifted due to instrumental effects , and milky - way metal absorption lines to check if the wavelength zero point is shifted due to the asymmetry of the target in the aperture ( e.g. , see @xcite ) .
figures [ fig4]-[fig8 ] show the cos spectra of the galaxies .
they are contaminated with some or all of the geocoronal emission lines hi @xmath81216 , ni@xmath81200 , and/or oi @xmath81302 , 1305 ; and with the milky way ( mw ) absorption lines , siii @xmath81190 , 1193 , 1260 , 1304 , siiii @xmath81206 , oi 1302 , 1305 , and/or cii @xmath81334 , 1336 .
we measured the centroids of the contaminating lines using a custom routine developed by cos science team member k. france .
the routine takes into consideration the cos line spread function .
table [ tab4 ] lists the mean wavelength offsets of the geocoronal and mw lines .
we find that the mean velocity offsets of the geocoronal and mw lines are in the ranges @xmath46 to @xmath47 km s@xmath7 and @xmath48 to @xmath49 km s@xmath7 , respectively .
the mw lines arise from absorption in halo clouds that may have non - negligible heliocentric velocities . for this reason
, we analyzed the hi 21 cm emission from intermediate velocity clouds ( ivcs ) along the lines of sight to our targets .
the data are from the lab hi survey @xcite . in figure
[ fig9 ] we compare the centroids of the hi ivc emission lines and the mw siii @xmath81990 absorptions . the siii absorptions are corrected for the wavelength offset given by the geocoronal lines .
the hi 21 cm emission is the mean within the radio beam at each velocity .
we chose siii @xmath81990 because it is the only mw line that is not blended with other lines for any of the targets . the hi component near 0 km s@xmath7 originates in the mw disk .
we find that the maximum ivc cloud velocity is @xmath50 km s@xmath7 in the local standard of rest frame ( e.g. , kissr 40 , 1567 , and 1578 ) .
although the comparison between the 21 cm emission and the cos data is difficult in cases where the signal to noise of the cos data is low , in general , the 21 cm emission and the siii absorption overlap .
however , in some of the highest signal - to - noise cases ( e.g. , kissr 242 , 1578 , 2019 , and 2110 ) , the minimum of the absorption is blueward of the bluest 21 cm emission peak .
note that the radio beam is large in comparison with the cos aperture , which makes this wavelength calibration method imperfect .
since large offsets between the 21 cm and the cos data are unlikely to be due to shifts in the wavelength zero point , we decided to only correct the cos data for the small shift in the geocoronal lines . regarding the spectral resolution , the more extended the source , the worse the effective spectral resolution , and the worse the accuracy in the determination of line centroids .
figure [ fig2 ] shows that some targets are more extended than others .
we used the mw lines siii @xmath81190 , 1193 , 1260 , siiii @xmath81206 , and/or cii @xmath81334 for determining the effective spectral resolution of the cos spectra .
the dispersion of the g130 m grating is 9.97 m / pix and the resolution element is 6 pixels .
therefore , the nominal resolution is 0.06 or @xmath1716 km s@xmath7 .
the effective spectral resolution of each spectrum is given by the fwhm of the mw absorption lines .
for each galaxy , column 4 of table [ tab4 ] lists the mean effective spectral resolution .
the mean fwhm of the mw lines is in the range
@xmath51 km s@xmath7
. we can derive the centroids of the lis lines with an accuracy of a tenth of the effective spectral resolution , as long as the signal to noise ratio is sufficient .
llllc 40 & -32 @xmath52 21 & -78 @xmath52 27 & 179 @xmath52 91 & 287 + 108 & -21 @xmath52 5 & -1 @xmath52 5 & 109 @xmath52 37 & 351 + 178 & -27 @xmath52 5 & -62 @xmath52 29 & 197 @xmath52 24 & 332 + 182 & -21 @xmath52 13 & -30 @xmath52 12 & 132 @xmath52 16 & 210 + 218 & 10 & -26 @xmath52 51 & 173 @xmath52 45 & 391 + 242 * & 18 @xmath52 5 & -34 @xmath52 8 & 110 @xmath52 27 & 227 + 271 * & -17 @xmath52 5 & -10 @xmath52 43 & 74 @xmath52 16 & 321 + 298 * & 15 & -49 @xmath52 17 & 75 @xmath52 52 & 337 + 326
* & -33 & -90 @xmath52 60 & 107 & 274 + 1084 * & 10 @xmath52 2 & -55 @xmath52 9 & 185 @xmath52 38 & 446 + 1567 * & -17 @xmath52 2 & & & + 1578 * & -11 & -70 @xmath52 4 & 125 @xmath52 18 & 308 + 1637 & 2 & -69 @xmath52 71 & 305 @xmath52 131 & 359 + 1785 & -14 @xmath52 2 & -110 @xmath52 4 & 142 @xmath52 10 & 246 + 1942 & -7 @xmath52 21 & -95 @xmath52 61 & 179 @xmath52 39 & 345 + 2019 & -16 @xmath52 6 & -101 @xmath52 13 & 188 @xmath52 30 & 238 + 2021 & -2 @xmath52 4 & -56 @xmath52 25 & 174 @xmath52 81 & 293 + 2023 & -15 @xmath52 9 & -38 @xmath52 72 & 198 @xmath52 94 & 304 + 2110 & -8 @xmath52 15 & -77 @xmath52 9 & 138 @xmath52 21 & 186 + 2125 & -7 @xmath52 14 & -78 @xmath52 14 & 274 @xmath52 74 & 349 + min & -33 & -110 @xmath52 4 & 74 @xmath52 16 & 186 + max & 18 @xmath52 5 & -1 @xmath52 5 & 305 @xmath52 131 & 446 + + [ tab4 ] here we present the velocity offsets of the intrinsic lines mentioned in the previous section .
figure [ fig10 ] shows the ly@xmath0 profiles of the galaxies .
seven objects have net ly@xmath0 emission ( emitters ) , eleven objects have damped ly@xmath0 absorptions ( absorbers ) , and two objects , kissr 2021 and 2023 have noisy ly@xmath0 profiles blended with mw lines and that appear to be in absorption with perhaps some ly@xmath0 emission .
we measured the ly@xmath0 centroids as follows .
for the emitters , the centroid is the point of maximum ly@xmath0 flux . for the absorbers
, we used the center of the best - fit voigt profile .
the fits to the ly@xmath0 absorptions were done by eye using a custom routine written by j. tumlinson .
we overlay the voigt profiles in figure [ fig10 ] . excluding kissr 298 , the emitters have p - cygni like profiles with the emission component redshifted to a mean velocity of @xmath5349 km s@xmath7 .
kissr 298 is a face - on spiral with two ly@xmath0 peaks separated by 370 km s@xmath7 , such that one peak is redshifted and the other is blueshifted and twice as strong .
the absorbers have ly@xmath0 troughs centered at a mean velocity of @xmath54 km s@xmath7 .
the ly@xmath0 mean velocities exclude kissr 2021 and 2023 due to their noisy ly@xmath0 profiles .
the cos spectra include the hi gas tracers oi @xmath81302 , siii @xmath81190 , and cii @xmath81334 .
they also include si iii @xmath81206 , which traces gas in a higher ionization stage .
figure [ fig11 ] shows the profiles of the previous metal lines in cases where the lines are not severely blended with contaminating lines .
we omit emitter kissr 1567 , because for this object the target acquisition failed and the only observable line is ly@xmath0 .
the figure shows that cii @xmath81334 and cii @xmath81336 are resolved in the highest signal to noise spectra , e.g. , for kissr 242 and 1578 . for the emitters , the mean blueshifts of oi @xmath81302 and cii @xmath81334 are @xmath55 km s@xmath7 and @xmath56 km s@xmath7 , respectively . for the absorbers ,
the corresponding mean velocity offsets are @xmath57 km s@xmath7 and @xmath58 , which is almost consistent with a static medium .
note that for spiral kissr 2023 , whose ly@xmath0 profile is very noisy , oi @xmath81302 has a large blueshift ( @xmath59 km s@xmath7 ) , such that ly@xmath0 emission would be expected .
as previously mentioned , there appears to be some ly@xmath0 emission in this galaxy which has poor signal to noise .
interestingly , the sdss image of kissr 2023 is very similar to that of double - peaked emitter kissr 298 .
our kinematical results are shown in table [ tab5 ] , which lists the kissr i d ( column 1 ) ; the ly@xmath0 profile shape ( column 2 ) ; the centroid of the ly@xmath0 line ( column 3 ) ; the velocity offsets between h@xmath0 and intrinsic interstellar absorption lines siii @xmath81190 , siii @xmath81193 , si iii @xmath81206 , oi @xmath81302 , siii @xmath81304 , and cii @xmath81334 ( columns 4 - 9 ) ; the mean velocity offsets and standard deviations excluding si iii @xmath81206 ( columns 10 & 11 ) ; the difference between the velocity offset of si iii @xmath81206 and the value in column 10 ( column 12 ) ; and for the ly@xmath0 absorbers , the hi column density of the voigt profile ( column 13 ) . in summary , excluding the double emitter kissr 298
, we find that : i ) ly@xmath0 is in emission and redshifted when high velocity outflows corresponding to expansion velocities of @xmath60 km s@xmath7 are present , ii ) ly@xmath0 is in absorption and at a velocity close to that of h@xmath0 when null or low velocity gas flows are present , and iii ) the ly@xmath0 redshift is higher than the blueshift of the lis lines by a factor of @xmath61 .
these results are further discussed in section 4 .
in addition , in cases were the centroid of si iii @xmath81206 was measured , we find that this line tends to have a blueshift larger than that of the lis lines by about @xmath62 km s@xmath7 .
lllllllllllll 40 & ab & -22 & & -29 & & & & -29 & -29 & 0 & & 5.0e+20 + 108 & ab & 0 & & -57 & -96 & -48 & & -50 & -52 & 5 & 44 & 7.5e+20 + 178 & ab & 2 & & & & -35 & -48 & -17 & -33 & 16 & & 2.3e+21 + 182 & ab & -37 & & -46 & -47 & -34 & -33 & -47 & -40 & 8 & 7 & 2.0e+21 + 218 & ab & 10 & & & & -32 & -11 & 23 & -7 & 28 & & 9.0e+20 + 242 & p - cygni & 170 @xmath52 46 & -68 & -73 & -130 & -53 & -104 & -79 & -76 & 19 & 55 & + 271 & p - cygni & 152 @xmath52 20 & & & & -177 & & -113 & -145 & 45 & & + 298 & 2 em & -118 @xmath52 11 , & & & -185 & -162 & -192 & -100 & -151 & 47 & 33 & + & & 249 @xmath52 17 & & & & & & & & & & + 326 & p - cygni & 251 @xmath52 24 & -162 & -140 & & -83 & -175 & -88 & -130 & 42 & & + 1084 & p - cygni & 198 @xmath52 20 & -169 & -86 & -146 & -109 & -125 & -78 & -113 & 36 & 33 & + 1567 & p - cygni & 105 @xmath52 41 & & & & & & & & & & + 1578 & p - cygni & 158 @xmath52 2 & -163 & -147 & -166 & -116 & -149 & -127 & -140 & 19 & 26 & + 1637 & ab & -50 & -68 & -65 & -60 & -54 & -69 & -9 & -53 & 25 & 6 & 7.0e+20 + 1785 & ab & 50 & & & & -24 & -61 & -42 & -42 & 19 & & 1.0e+21 + 1942 & ab & -50 & -7 & -2 & & 6 & -63 & -24 & -18 & 27 & & 9.5e+20 + 2019 & ab & 50 & -54 & -64 & -37 & -25 & & -59 & -50 & 18 & -14 & 3.5e+20 + 2021 & ab & -50 & -38 & -28 & -19 & -8 & -18 & -14 & -21 & 12 & -2 & 2.5e+20 + 2023 & ab & -50 & & & & -153 & -62 & -171 & -128 & 58 & & 1.0e+20 + 2110 & ab & 50 & -13 & 6 & -76 & 36 & 15 & -2 & 8 & 19 & 85 & 3.5e+20 + 2125 & ab & 0 & -56 & -38 & & & -44 & 48 & -22 & 48 & & 1.5e+21 + + [ tab5 ] in order to characterize the strength of ly@xmath0 , we computed the ly@xmath0 luminosity @xmath64(ly@xmath0 ) , equivalent width , and escape fraction within the cos aperture . although the relevant quantity for cosmological studies is the global integrated escape fraction of the galaxy , our measurements are useful for comparing with global values derived from ly@xmath0 line images , as they give an idea of the concentration of the emission and of how coincident it is with regions of high fuv surface brightness .
such ly@xmath0 maps have been or will be obtained for a portion of our targets .
we computed two values of the escape fraction , one using h@xmath0 and the other using h@xmath24 .
both adopt the case b recombination theory predictions of @xcite for @xmath65 k and @xmath66 @xmath67 , i.e. , ly@xmath0/h@xmath0=8.1 and ly@xmath0/h@xmath24=23.1 .
@xmath64(ly@xmath0 ) .
the escape fractions were computed before and after correcting for the dust attenuation .
we also computed the cos luminosity of the continuum at 1500 , @xmath63 , for comparison with the value from _ galex _ , which is given in table [ tab1 ] . in order to obtain @xmath63 ,
we extrapolated beyond the red edge of our data which is less than 1500 . in this paper , @xmath63 is used for deriving the mass of the dominant stellar population and the star formation rate ( see below ) . for correcting the hydrogen line fluxes and
@xmath63 for the dust attenuation , we derived the total color excess , @xmath68 , from the observed h@xmath0/h@xmath24 line intensity ratio , after correcting the balmer lines for the underlying stellar absorption .
the stellar absorption correction was done by adding 2 to the equivalent widths of the balmer lines , following @xcite .
this increases the h@xmath0 flux by 2% and the h@xmath24 flux by 13% ( median values for the sample ) .
we did not correct the interstellar ly@xmath0 for the stellar contribution at ly@xmath0 , since as mentioned in the introduction , this correction is expected to be small for stellar populations dominated by o and early b stars , as is the case for the ly@xmath0 emitters in our sample .
that o and early b stars dominate for the emitters is shown in section 3.5 .
the ly@xmath0 emission line fluxes were obtained by numerical integration .
the balmer line fluxes were obtained from the sdss data when available and from the kiss data otherwise .
the sdss spectra have higher spectral resolution .
the total color excess is given by @xmath69 , where @xmath70 and @xmath71 are the galactic and intrinsic contributions , respectively .
according to the milky way reddening maps of @xcite , @xmath70 is small towards our targets ( see table [ tab1 ] ) , thus most of the extinction must be intrinsic . since @xcite and @xcite showed that by using a metallicity - dependent reddening law one could recover the theoretical ly@xmath0/h@xmath24 ratio , we tried using a mw law for the spiral galaxies and an smc law for the irregular galaxies . for this purpose
, we used the reddening laws tabulated in @xcite .
however , we found that the @xmath68 values derived from these two laws are very similar . in the end , and as in @xcite , we adopted a mw law for deriving @xmath68 . for the theoretical h@xmath0/h@xmath24 ratio we used 2.86 , which is the case b recombination theory prediction for @xmath72k and @xmath73 cm s@xmath74 of @xcite .
in order to see if the relation @xmath75 of @xcite holds for our data , we computed the color excess of the stellar continuum , @xmath76 , using the slope of the fuv continuum , @xmath24 . for obtaining @xmath76
we used the method outlined in @xcite , where the slope of the milky - way de - reddened continuum is compared to that of a dust - free model of the appropriate metallicity .
following @xcite , @xmath76 was derived using the starburst reddening law .
the comparison was done over the wavelength range from 1200 to the red edge of the data .
the red edge is redshift - dependent , as shown in figures 4 - 8 , and it is always less than 1500 .
this range is different from what is used in @xcite , i.e. , @xmath77 .
our results are shown in table [ tab6 ] , which gives the kissr i d of the galaxy ( column 1 ) ; the rest - frame value of @xmath27ly@xmath0 ) ( column 2 ) ; @xmath64(ly@xmath0 ) , @xmath63 from the cos spectra , @xmath64(h@xmath0 ) , and @xmath64(h@xmath24 ) ( columns 3 - 6 , uncorrected for dust extinction ) ; the escape fractions within the cos aperture , i.e. ly@xmath0/h@xmath0/8.1@xmath78100 and ly@xmath0/h@xmath24/23.1@xmath78100 ( columns 7 & 8) ; @xmath68 from the balmer lines and @xmath76 from the slope of the fuv continuum ( columns 9 & 10 ) ; and the reddening - corrected values of @xmath63 , h@xmath0/h@xmath24 , and the escape fractions ( columns 11 - 14 ) . in summary , we find that @xmath79 ( adopting the median of the sample ) , i.e. , that the stellar continuum is more reddened than the nebular continuum , contrary to what was found by @xcite . the different result is attributed to the larger optical and ultraviolet apertures used by @xcite , and to the different wavelength range used by the latter authors for fitting the uv continuum .
we also find that the rest - frame @xmath44s of the emitters are in the range @xmath5 , and that the reddening corrected escape fractions of the emitters are in the range @xmath5% .
the latter two results are further discussed in section 4 .
finally , we found that @xmath63 from the cos data is lower by a factor of a few compared to the value from _ galex _ ( compare column 4 of table [ tab6 ] with column 9 of table [ tab1 ] ) .
this is attributed to the larger aperture of _
lccccccccccccc 40 & @xmath80 & & 38.6 & 40.02 & 39.54 & & & 0.04 & 0.21 & 39.5 & 3.1 & & + 108 & @xmath80 & & 39.1 & 40.38 & 39.85 & & & 0.15 & 0.17 & 39.9 & 3.5 & & + 178 & @xmath80 & & 39.6 & 41.21 & 40.64 & & & 0.23 & 0.20 & 40.5 & 4 & & + 182 & @xmath80 & & 39.1 & 40.41 & 39.88 & & & 0.16 & 0.19 & 39.9 & 3.6 & & + 218 & @xmath80 & & 38.9 & 40.49 & 39.87 & & & 0.33 & 0.41 & 40.6 & 4.6 & & + 242 * & 10 & 41.20 & 40.1 & 41.61 & 41.07 & 5 & 6 & 0.18 & 0.21 & 41.0 & 3.7 & 12 & 14 + 271 * & 1 & 38.74 & 38.5 & & & & & 0.17 & 0.28 & 39.7 & & & + 298 * & 12 & 40.16 & 39.3 & 40.65 & 40.06 & 6 & 2 & 0.28 & 0.39 & 41.0 & 4.3 & 12 & 15 + 326 * & 4 & 39.30 & 38.7 & 40.22 & 39.61 & 2 & 0 & 0.16 & 0.38 & 40.3 & 4.4 & 7 & 9 + 1084 * & 1 & 38.99 & 39.0 & 41.03 & 40.37 & 0 & 1 & 0.42 & 0.44 & 41.0 & 5.2 & 1 & 1 + 1567 * & 3 & 38.71 & 38.3 & 40.25 & 39.63 & 0 & 1 & 0.15 & 0.31 & 39.6 & 4.7 & 2 & 3 + 1578 * & 7 & 41.00 & 40.0 & 41.55 & 41.04 & 3 & 4 & 0.12 & 0.10 & 40.4 & 3.4 & 6 & 7 + 1637 & @xmath80 & & 39.2 & 40.43 & 39.93 & & & 0.10 & 0.30 & 40.5 & 3.3 & & + 1785 & @xmath80 & & 38.5 & 40.38 & 39.83 & & & 0.18 & 0.19 & 39.3 & 3.7 & & + 1942 & @xmath80 & & 38.9 & 40.00 & 39.44 & & & 0.21 & 0.28 & 40.1 & 3.9 & & + 2019 & @xmath80 & & 39.1 & 40.63 & 40.17 & & & 0.01 & 0.14 & 39.7 & 2.9 & & + 2021 & @xmath80 & & 39.2 & 40.59 & 40.03 & & & 0.21 & 0.30 & 40.5 & 3.8 & & + 2023 & & & 38.6 & 40.58 & 39.91 & & & 0.45 & 0.31 & 39.9 & 5.5 & & + 2110 & @xmath80 & & 39.1 & 40.33 & 39.86 & & & 0.00 & 0.09 & 39.5 & 3 & & + 2125 & @xmath80 & & 38.9 & 40.73 & 40.03 & & & 0.49 & 0.30 & 40.2 & 5.7 & & + min & 1 & 38.71 & 38.3 & 40.00 & 39.44 & 0 & 0 & 0.00 & 0.09 & 39.3 & 2.9 & 1 & 1 + max & 12 & 41.20 & 40.1 & 41.61 & 41.07 & 6 & 6 & 0.49 & 0.44 & 41.0 & 5.72 & 12 & 15 + + [ tab6 ] we studied how the strength of ly@xmath0 relates to the starburst phase and star formation rate of the dominant stellar population by comparing the cos observations with synthetic dust - free spectra computed with the widely used package starburst99 ( s99 , @xcite ) .
the observed star formation rate is that within the cos aperture .
the value of @xmath27ly@xmath0 ) depends on the metallicity , imf , star formation history , and evolutionary phase of the stellar population ( see models by @xcite ) .
because empirical spectral stellar libraries in the uv are only available at the galactic ( @xmath81 , @xcite ) and lmc / smc ( @xmath82 and @xmath83 , @xcite ) metallicities , we have to rely on theoretical libraries at other metallicities .
we used the theoretical libraries of @xcite corresponding to @xmath84 , 0.004 , 0.008 , 0.020 , and 0.040 . for comparison , we also computed models with the empirical galactic and lmc / smc stellar libraries
. we will refer to simulations based on the theoretical and empirical libraries as the theoretical and empirical models , respectively .
the theoretical models use stellar evolution tracks corresponding to @xmath85 and model atmospheres corresponding to @xmath86 .
the empirical models use tracks corresponding to @xmath85 or @xmath87 .
our grid of models includes single stellar population ( ssp ) and continuous star formation ( csf ) scenarios .
continuous star formation is more appropriate in our case , as the observed spectra include light from regions of a few kpc in size and such large regions must contain uv - bright star clusters that span a range of ages .
the ssp models are for comparison .
we adopted a kroupa imf @xcite , as at the high - mass end , it is considered to be universally applicable @xcite .
we used the geneva stellar evolution tracks for non - rotating single - stars with high - mass loss @xcite because models that account for stellar rotation and binarity @xcite are still under development and require calibration against observations over our range of metallicities .
finally , we did not compute models for times earlier than 2 myr because this is the age of the youngest lmc main - sequence o stars , or greater than 30 myr after the beginning of star formation since the ssp models are affected by incompleteness of the stellar libraries beyond this time and the continuum slope and age - sensitive stellar - wind lines of csf models do not significantly change beyond this time .
the age of an fuv - bright stellar population can be derived spectroscopically by comparing the observed profiles of age sensitive lines originating in the winds of massive stars with model spectra .
the strong nv @xmath81238.8 , 1242.8 , siiv @xmath81393.8 , 1402.8 , and civ @xmath81548.2 , 1550.8 resonant doublets are widely used for this purpose @xcite . in summary ,
the p - cygni profiles of the nv and the civ doublets decrease in strength as the o stars evolve and expire . on the other hand ,
the siiv doublet develops a p - cygni profile in giant and supergiant o stars .
unfortunately , the civ doublet and sometimes the siiv doublet fall outside of the wavelength range that we observed with cos .
in addition , the nv doublet is sometimes contaminated with geocoronal emission . however , for targets showing clean and strong nv p - cygni profiles , the age of the dominant population is younger than @xmath1710 myr ( see figure 5 in @xcite ) .
furthermore , a large h@xmath0 equivalent also indicates the presence of a young population @xcite .
we compared the cos spectra with ssp and csf models of 5 and 10 myr .
we select this age range because between 5 and 10 myr , the strength of nv noticeably decreases and these ages are sufficient for our purpose of demonstrating how young the dominant stellar populations are .
we found the best - fit model to the nv and siiv profiles by eye .
the comparison was performed on the rectified spectra , as the models are dust - free but the data are not .
the metallicity of the model is that most appropriate for the galaxy based on its oxygen abundance$ ] , where @xmath88 corresponds to @xmath89 , i.e. , to the old solar value of @xcite . ] .
figures [ fig12 ] and [ fig13 ] show the comparisons of the observed nv and siiv profiles with the theoretical models .
similarly , figures [ fig14 ] and [ fig15 ] show the comparisons with the empirical models .
a comparison of the theoretical and empirical models in figures [ fig12 ] and [ fig14 ] shows that at both ages and at all metallicities , nv is stronger for the csf models than for the ssp models , although the difference is less at the lowest metallicity ( e.g. , for 2019 and 2110 ) .
furthermore , the theoretical csf models yield stronger nv profiles than the empirical models at similar ages and metallicities ( e.g. , 40 and 242 ) .
conversely , the theoretical siiv profiles show weaker emission components than the empirical ones . these differences do not severely affect our results .
the observed nv profiles of ly@xmath0 emitters 298 , 326 , and 1567 are contaminated with geocoronal oi lines , and the siiv profiles of 11 targets , including four of the seven emitters , are either incomplete or have low signal to noise .
for this reason , and because the siiv profiles from the theoretical models are too weak compared to observations at young ages ( e.g. , compared to the empirical models ) , in the rest of our age - dating analysis , we concentrate on the nv profiles . for targets with clean nv profiles , we find that the theoretical ssp models tend to be a better fit than the theoretical csf models , except at the lowest metallicity ( e.g. , kissr 2019 and 2110 ) , where the ssp and csf nv profiles are very similar .
this is true for the ly@xmath0 emitters as well as for the ly@xmath0 absorbers .
one exception is 1084 , which is better fitted by a csf model . in any case
, ly@xmath0 absorbers such as spiral galaxy kissr 218 and metal - poor galaxies kissr 2019 and 2110 have strong nv profiles and large h@xmath0 @xmath44s indicative of ages younger than @xmath1710 myr .
thus , they would be expected to have strong ly@xmath0 emission .
however , they are absorbers , which is consistent with their low gas - flow velocities ( see table [ tab5 ] ) .
in addition , one can not rule out attenuation of ly@xmath0 due to the presence of dust .
we also find that the emitters have ly@xmath0 equivalent widths that are consistent with their young ages , but that are small compared to model predictions for young starbursts .
thus , the emitters also require the presence of scattering in hi and attenuation by dust .
the take away point is that even at young ages ( @xmath90myr ) targets can show ly@xmath0 in absorption if the gas surrounding the starburst is almost static or if the dust column density surrounding the hot stars is significant .
table [ tab7 ] gives the stellar masses and star formation rates derived from the uv spectra by adopting the ages in the second column of the table . for a given age , metallicity , imf , and star formation history ,
the theoretical value of @xmath63 only depends of the mass of the stellar population .
we give two estimates of the mass and the star formation rate , one corresponding to the ssp and the other to the csf model .
the masses are derived from the comparison of the observed and computed luminosities of the stellar continuum at 1500 ( both dust - free ) . for the computed luminosity
, we use the mean of the theoretical and empirical predictions at the adopted age and metallicity . note that for the ssp models
, it is common practice to define a characteristic star formation rate as the ratio of the stellar mass to the age of the population .
this is what we quote in column 5 of table [ tab7 ] . for comparison
, the table also gives the star formation rate derived from the sdss h@xmath0 luminosity ( corrected for reddening using the starburst attenuation law of @xcite ) and equation ( 2 ) in @xcite .
we recall that the sdss aperture is very similar to the cos aperture .
this is not the case for the kiss data , therefore , we omit the optical sfr for targets without sdss data .
we also recall that the star formation rates are not global values .
we find that the ssp models yield higher masses and star formation rates than the csf models by a factor of 1.4 ( median ) .
this is because csf models predict larger values of @xmath63 than the ssp models at the adopted ages .
we also find that the ssp models yield systematically higher star formation rates than those derived using h@xmath0 by a factor of 5 ( median ) with a standard deviation of 7 .
the uv ssp sfrs derived from @xmath63 should be viewed with caution as they are inversely proportional to the ages , which are expected to be underestimated by a factor of a few in cases where nv is contaminated with geocoronal emission or does not show a strong p - cygni profile .
this is the case for kissr 298 and 1637 , which show the largest disagreement between the optical and the uv sfrs .
in addition , the uv ssp sfrs depend on the masses , and wofford et al .
( 2011 ) found systematically higher masses derived from @xmath63 compared to masses derived from optical photometry for individual star clusters . on the other hand , in some cases , the agreement between the optical and uv sfrs is very good .
this is the case for kissr 1578 , a target with a prominent nv doublet and a high s / n spectrum .
unfortunately , only four emitters have h@xmath0 sfrs , which makes it difficult to study trends between ly@xmath0 and the sfr .
lcccccc 40 & 10 & 7e+06 & 3e+06 & 0.7 & 0.3 & 0.1 + 108 & 5 & 7e+06 & 5e+06 & 1.4 & 0.9 & 0.3 + 178 & 10 & 7e+07 & 3e+07 & 6.8 & 2.7 & 2.6 + 182 & 10 & 2e+07 & 6e+06 & 1.6 & 0.6 & 0.3 + 218 & 5 & 4e+07 & 2e+07 & 7.2 & 5.0 & 0.7 + 242 * & 5 & 9e+07 & 6e+07 & 17.6 & 12.1 & 5.6 + 271 * & 10 & 1e+07 & 5e+06 & 1.1 & 0.5 & + 298 * & 10 & 2e+08 & 8e+07 & 19.7 & 7.9 & 0.8 + 326 * & 5 & 2e+07 & 1e+07 & 3.8 & 2.6 & + 1084 * & 5 & 9e+07 & 6e+07 & 18.5 & 12.8 & 3.1 + 1567 * & 5 & 4e+06 & 3e+06 & 0.8 & 0.5 & + 1578 * & 5 & 3e+07 & 2e+07 & 5.0 & 3.5 & 4.0 + 1637 & 10 & 6e+07 & 3e+07 & 6.5 & 2.6 & 0.3 + 1785 & 10 & 5e+06 & 2e+06 & 0.5 & 0.2 & 0.3 + 1942 & 5 & 1e+07 & 8e+06 & 2.3 & 1.6 & 0.2 + 2019 & 5 & 5e+06 & 3e+06 & 1.0 & 0.7 & 0.3 + 2021 & 5 & 3e+07 & 2e+07 & 5.2 & 3.6 & 0.6 + 2023 & 5 & 7e+06 & 5e+06 & 1.5 & 1.0 & 1.2 + 2110 & 5 & 3e+06 & 2e+06 & 0.6 & 0.4 & + 2125 & 5 & 1e+07 & 1e+07 & 2.9 & 2.0 & 1.9 + min & - & 3e+06 & 2e+06 & 0.5 & 0.2 & 0.1 + max & - & 2e+08 & 8e+07 & 19.7 & 12.8 & 5.6 + mean & - & 4e+07 & 2e+07 & 5.2 & 3.1 & 1.4 + sigma & - & 5e+07 & 2e+07 & 6.2 & 3.7 & 1.6 + + [ tab7 ] the detection of broad ( @xmath91 , @xcite ) heii @xmath84686 emission ( hereafter , heii emission ) provides another way of estimating the age of a stellar population . in the absence of an agn
, such emission originates in the dense winds of wr stars @xcite , whose progenitors are massive stars ( m@xmath92 , @xcite ) with lifetimes of @xmath93 myr @xcite .
other optical wr emission lines are ciii @xmath84650 and niii @xmath84640 , which are generally weaker . for each galaxy , we show in figure [ fig16 ] the spectral region around the heii @xmath84686 line .
the figure shows that in some cases , the [ feiii ] @xmath84658 peak is taller than the heii @xmath84686 peak . as in @xcite
, we used multiple gaussians to fit the spectral region in the vicinity of 4686 .
the gaussian centers were fixed at wavelengths where lines were expected .
we detect heii at the @xmath94 level or better in six cases , i.e. , the amplitude of heii is at least three times the standard deviation of the nearby continuum in six cases . in five of the latter cases ,
the heii line is broad , i.e. , @xmath91 .
unfortunately , for four of the seven ly@xmath0 emitters , the optical spectrum is either unavailable ( kissr 271 ) , of inadequate spectral resolution ( kissr 326 and kissr 1567 ) , or very noisy ( kissr 298 ) .
the three emitters with adequate optical spectra have heii detections with @xmath95 .
this includes kissr 1084 , which may be of composite spectral type , and whose broad heii emission is accompanied by narrow / nebular heii emission .
note that for the latter galaxy , the heii peak is taller than the [ feiii ] @xmath84658 peak , unlike in most galaxies with broad heii emission . on the other hand ,
two of the non - ly@xmath0 emitters have heii detections and @xmath96 .
the lack of ly@xmath0 emission in galaxies with broad heii emission could be due to the lack of significant gas flow in the direction towards the observer .
this is the case of kissr 178 and the highly inclined spiral kissr 2125 .
finally , the non - emitter kissr 108 only shows narrow heii emission . @xcite
interpret the presence of narrow heii emission without a broad component , as wr stars that are offset from the emitting gas .
lccc 40 & & & + 108 & 2 & [email protected] & narrow + 178 & 7 & [email protected] & + 182 & & & + 218 & & & + 242 * & 6 & [email protected] & + 271 * & & & no optical + 298 * & & & + 326 * & & & no sdss + 1084 * & 10 & [email protected] & broad + narrow + 1567 * & & & no sdss + 1578 * & 9 & [email protected] & + 1637 & & & + 1785 & & & + 1942 & & & + 2019 & & & + 2021 & & & + 2023 & & & + 2110 & & & no sdss + 2125 & 10 & [email protected] & + min & 2 & [email protected] & + max & 10 & [email protected] & + + [ tab8 ] we compared the observed values of @xmath44(heii ) with starburst99 @xcite model predictions .
note that the observed values constitute lower limits due to the dilution of the heii line in the stellar continuum of older stellar generations .
table [ tab8 ] lists fwhm(heii ) and @xmath44(heii ) .
figure [ fig17 ] shows the comparison of observed and computed values of @xmath44(heii ) corresponding to single stellar populations ( ssp ) and continuous star formation ( csf ) at the metallicities of the galaxies .
although figure [ fig17 ] shows that a given value of @xmath44(heii ) can be reached at multiple times , the range of times is restricted to @xmath97 yr for the ssp models , and @xmath98 yr for the csf models .
the presence of broad heii emission rules out an imf deficient in massive stars or a population dominated by late - b and a stars .
therefore , according to the predictions by @xcite one would expect positive values of @xmath44(ly@xmath0 ) for the galaxies with broad heii emission .
as previously mentioned , the lack of ly@xmath0 emission in kissr 178 and kissr 2125 is probably due to the dust geometry and inclination . in conclusion
, we find no one to one correspondence between the ly@xmath0 emission and the presence of wr stars .
this is in agreement with our age - dating analysis based on the nv profiles , where we concluded that some of the galaxies with young stellar populations are absorbers .
@xcite analyzed _ hst_/ghrs observations of 8 nearby hii galaxies covering metallicities from 12+log(o / h)=8.0 to solar , and reddenings from e(b - v)=0.1 to 0.55 .
they found velocity offsets between the hi and the hii gas of up to 200 km s@xmath7 in the four galaxies with ly@xmath0 emission ( p - cygni profiles ) , whereas they found broad damped ly@xmath0 absorptions at the velocity of the hii gas in galaxies with no signs of hi outflows from the lis lines .
these authors concluded that outflows are the main determining factor of the ly@xmath0 escape from galaxies .
our kinematical results are in very good agreement with these findings .
however , one still needs to assess the relative importance of scattering in hi and destruction by dust on the ly@xmath0 escape .
we also find that the ly@xmath0 redshift is higher than the blueshift of the lis lines by a factor of @xmath61 .
this is in agreement with the high redshift observations of @xcite and the expanding - shell radiation transport model of @xcite .
finally , note that the ly@xmath0 troughs are the result of a superposition of ly@xmath0 components .
therefore , the dominant component does not need to be at zero velocity .
it is interesting that for our targets and for the targets in @xcite , the ly@xmath0 absorption centroid is consistent with the velocity of the hii gas .
kissr 298 is the double - peaked ly@xmath0 emitter .
such a profile is generally interpreted as a signature of ly@xmath0 radiation transfer through a non - static medium ( see references given in section 4.5 of @xcite . for this object , we find an expansion velocity of @xmath99 km s@xmath7 from the lis lines .
this is high compared to the expansion velocities for the two double - peaked spectra in @xcite , which are almost consistent with a static medium ( they are @xmath100 km s@xmath7 ) . in kissr 298 ,
the blueshifted peak is twice as strong as the redshifted peak .
stronger blue peaks have been observed in star - forming galaxies at redshifts of @xmath101 @xcite .
they are qualitatively consistent with an inflow @xcite and could be due to in - falling gas originating in a galactic fountain , as in the model of @xcite .
the fact that in this system the dip between peaks essentially reaches the continuum is an interesting feature .
another possibility is that kissr 298 has a broad ly@xmath0 emission line blended with an absorption , and that the absorption center is shifted redward .
the shift could be caused by the relative motion of the emission region and the absorbing gas .
a system corresponding to this description is shown in panel f of figure 3 of @xcite , which corresponds to the second cloud from the top in their figure 4 .
unfortunately , the metal absorption lines of kissr 298 are noisy and it is hard to tell if there is are unexpected absorptions near the wavelength corresponding to the ly@xmath0 `` absorption '' . as can be seen in figure [ fig11 ] , the point equidistant to the two ly@xmath0 peaks of kissr 298 , which is marked with a dashed line , has a velocity coincident with the siii 1304 and cii @xmath81336 absorptions .
we checked if there is an observational bias due to distance among the ly@xmath0 emitters in our sample .
we excluded kissr 1084 and 1567 from the analysis as one is of composite type and for the other the ta failed . in order to separate the effect of distance and metallicity / reddening
, we divided the emitters in two sets : the low metallicity irregulars ( kissr 242 , 1578 , and 271 ) and the higher metallicity spirals ( kissr 298 and 326 ) .
each set spans a factor of about 2 in distance ( 1.75 for the spirals and 1.7 for the irregulars ) .
the spirals have low inclinations .
therefore the inclination angle is a secondary effect .
we find that @xmath35 , @xmath11 , and @xmath102 simultaneously increase with distance for both metallicity sets ( see table [ tab6 ] ) .
note that this result is based on small number statistics and that the general trend at much higher redshifts is that brighter objects have lower equivalent widths ( e.g. @xcite ) . for the emitters , we find that @xmath103 is low compared to values determined in other studies at similar redshifts with larger apertures .
@xcite found that @xmath104 on average for a sample of galaxies observed through the @xmath105 aperture of _
iue_. we attribute this to our smaller aperture . indeed ,
as previously mentioned , @xcite found that in nearby galaxies , the bulk of the ly@xmath0 emission is in the diffuse component of the galaxies .
we also find that the emitters in our sample have @xmath103 slightly below the completeness limit of the @xmath106 sample of @xcite , which is 15 .
however , our sample has a median absolute _
b _ magnitude of -19.4 ( see table [ tab1 ] ) , which is similar to that of the @xmath106 sample ( see their figure 23 ) .
this again could be an aperture effect . for lyman - break galaxies ( lbgs ) at @xmath107 , the median value of @xmath103 is zero ( see figure 8 of @xcite ) .
therefore , the @xmath108 of the emitters in our sample are consistent with the latter median .
if a homogeneous hi layer covers most of the hii region , the value of @xmath27ly@xmath0 ) is expected to depend on the dust extinction because the ly@xmath0 line flux is more strongly reduced ( due to multiple scattering effects ) than the adjacent continuum .
reddening due to the presence of dust is expected to increase with metallicity .
we use oxygen as a gauge of metallicity .
we re - derived the oxygen abundances of the 16 galaxies with sdss spectra using the method described in @xcite . in summary , in cases where the weak [ oiii ] @xmath84363 line was not detected , we used the mean of the oxygen abundances derived from the ep and sdss coarse calibrations , and in cases where the r23 abundance could be computed , we used the mean of the ep , sdss and r23 abundances ( all methods are defined in @xcite ) .
the oxygen abundances from the kissr and the sdss data are in agreement within the uncertainties and the final values are those given in table [ tab1 ] .
we find that the two most metal - poor and least reddened galaxies , kissr 2019 and 2110 , show ly@xmath0 in pure absorption , while the most metal - rich and most reddened galaxy , kissr 298 , has the highest value of @xmath103 .
strong ly@xmath0 emission is expected from kissr 2019 and 2110 based on the strength of their nv profiles and their high values of @xmath109 ( @xmath110 ) , which are characteristic of a young burst of star formation . however , these two targets show ly@xmath0 in absorption . this is not surprising
given that this is also the case for i zw 18 , the most metal - poor galaxy in the local universe @xcite .
although we suffer from small number statistics , the lack of one to one correspondence between the strength of ly@xmath0 and metallicity / reddening is in agreement with the work of @xcite , which is based on larger optical and ultraviolet apertures than what we use in the present study .
therefore , it is not clear that larger aperture measurements yield a different result regarding ly@xmath0 versus metallicity .
our upcoming ly@xmath0 plus h@xmath0/h@xmath24 ratio maps of three of the galaxies in our sample should help establish the role of dust and metallicity in regulating the ly@xmath0 escape .
these maps will be obtained as part of _ hst _
go-12951 and will enable the measurement of spatially resolved and integrated ly@xmath0 and reddening values .
1 . we used the _ hst _ cos to observe 20 nearby ( @xmath111 )
galaxies spanning a broad range in properties ( table [ tab1 ] ) . our data covers the wavelength range @xmath40
and samples circular regions of 0.9 - 2.4 kpc in radius within each galaxy . we studied correlations between the ly@xmath0 line properties and galaxy properties including distance , metallicity , reddening , starburst phase , and gas flow properties .
our study includes the analysis of ancillary optical spectroscopy from sdss and kiss .
based on the bpt diagram ( figure [ fig1 ] ) , the galaxies are dominated by star - formation except for kissr 1084 , which is of composite spectral type .
we found ly@xmath0 profile shapes representative of what has been observed between redshifts of @xmath112 and @xmath113 ( e.g. , @xcite ) , including strong absorptions , p - cygni profiles with redshifted emission , and double emission ( figure [ fig10 ] ) .
3 . seven out of the 20 galaxies have net ly@xmath0 emission , including spiral and irregular galaxies and composite galaxy kissr 1084 ( figures [ fig3 ] and [ fig10 ] ) .
for these seven emitters , the ly@xmath0 equivalent width is in the range @xmath5 and the range of escape fractions within the cos aperture is @xmath5 % ( table [ tab6 ] ) .
the comparison with escape fractions at higher redshifts is complicated by the fact that at higher redshifts , observations include more if not all of the galaxy in the aperture .
4 . a face - on spiral , kissr 298 , shows two peaks of ly@xmath0 emission separated by 370 km s@xmath7 , one blueshifted and one redshifted and half as strong . unfortunately , we were unable to determine if this profile is a strong emission with absorption within , as lis lines for this galaxy are too noisy .
5 . excluding kissr 298 ,
the emitters have ly@xmath0 peaks redshifted to @xmath114 km s@xmath7 with respect to h@xmath0 ( mean @xmath52 standard deviation ) .
including all emitters , the mean oi @xmath81302 and cii @xmath81334 absorption blueshifts are @xmath115 and @xmath116 km s@xmath7 .
therefore , excluding kissr 298 , the ly@xmath0 redshift is larger than the interstellar absorption blueshift by approximately a factor of @xmath61 , as found at high redshift ( e.g. shapley et al . 2003 ) , and in agreement with radiation transfer models ( e.g. , verhamme et al .
2006 ) . 6 . for the absorbers , ly@xmath0 is centered at @xmath117 km s@xmath7 and oi and cii have centroids at @xmath118 and @xmath119 km s@xmath7 . thus , most absorbers are consistent with having static or low velocity hi gas .
this supports earlier findings ( e.g. of kunth et al .
1998 ) . 7 .
the outflow velocity of si iii @xmath81206 is typically , @xmath62 km s@xmath7 greater than that of the hi gas ( table [ tab5 ] ) .
we found no one to one correspondence between the strength of ly@xmath0 and the reddening / metallicity measured within the cos aperture ( tables [ tab6 ] and [ tab7 ] ) .
in particular , we found three low - inclination spirals each showing different ly@xmath0 profiles , i.e. , pure absorption , double emission , and p - cygni profile , and the two most metal - poor and least - reddened galaxies in our sample are pure absorbers .
we find stellar populations consistent with ssps of ages in the range 5 - 10 myr among the ly@xmath0 emitters and the ly@xmath0 absorbers .
similarly , we detect signatures of the presence of wolf - rayet stars among the emitters and the absorbers .
therefore , a young starburst does not guarantee strong ly@xmath0 emission . 10 . in conclusion ,
a picture emerges where the ly@xmath0 photons escape through regions of low hi and dust column densities , where gas outflows are occurring .
11 . in _ hst _
cycle 20 , we will map three low - inclination spirals from our sample in the ly@xmath0 line and the h@xmath24/h@xmath0 ratio ( proxy for dust attenuation ) . this will make it possible to study the relative importance of gas flows and dust content in regulating the ly@xmath0 escape .
support for this work has been provided by nasa through an award to university of colorado ( boulder ) entitled `` cosmic origins spectrograph gto '' . award number nnx08ac14 g .
we thank c. danforth , k. france , and j. tumlinson for their codes , which were used for reducing ( cd ) and analyzing ( kf and jt ) the cos data .
we also thank c. thom for his help with the hi 21 cm data , and z. zheng for helpful comments regarding the ly@xmath0 profile of kissr 298 .
finally , we thank the referee for comments that have helped to greatly improve this paper . , a. , schaerer , d. , atek , h. , & tapken , c. 2007 , in astronomical society of the pacific conference series , vol .
380 , deepest astronomical surveys , ed .
j. afonso , h. c. ferguson , b. mobasher , & r. norris , 97 | the usefulness of hi ly@xmath0 photons for characterizing star formation in the distant universe is limited by our understanding of the astrophysical processes that regulate their escape from galaxies .
these processes can only be observed in detail out to a few @xmath1mpc .
past nearby ( @xmath2 ) spectroscopic studies are based on small samples and/or kinematically unresolved data .
taking advantage of the high sensitivity of _ hst _ s cos , we observed the ly@xmath0 lines of 20 h@xmath0-selected galaxies located at @xmath3 .
the galaxies cover a broad range of luminosity , oxygen abundance , and reddening . in this paper
, we characterize the observed ly@xmath0 lines and establish correlations with fundamental galaxy properties .
we find seven emitters .
these host young ( @xmath4myr ) stellar populations , have rest - frame equivalent widths in the range @xmath5 , and have ly@xmath0 escape fractions within the cos aperture in the range @xmath5% .
one emitter has a double - peaked ly@xmath0 with peaks @xmath6 km s@xmath7 apart and a stronger blue peak . excluding this object , the emitters have ly@xmath0 and oi @xmath81302 offsets from h@xmath0 in agreement with expanding shell models and lbg observations .
the absorbers have offsets that are almost consistent with a static medium .
we find no one - to - one correspondence between ly@xmath0 emission and age , metallicity , or reddening .
thus , we confirm that ly@xmath0 is enhanced by outflows and is regulated by the dust and hi column density surrounding the hot stars . |
supersymmetric grand unified theories ( susy guts ) @xcite are promising candidates of physics beyond the standard model ( sm ) .
indeed , they are strongly motivated by an experimental observation which implies that the gauge coupling constants of the sm gauge groups are to be unified at a certain high - energy scale with good accuracy @xcite . from the observation ,
the unified scale is estimated as @xmath7 gev . in the guts ,
new superheavy particles are supposed to appear around the scale , which make the couplings run together above the scale .
these particles are naturally expected to have masses of the order of @xmath8 .
they are , of course , beyond the reach of collider experiments , and there is little hope to search for them or to measure their masses directly . in refs .
@xcite , a way of constraining the masses of superheavy particles indirectly by requiring the gauge coupling unification is discussed and limits on the masses are presented . later , by applying the same method with more accurate gauge coupling constants , the authors in ref . @xcite derive a more stringent constraint on the masses of the color - triplet higgs boson , the adjoint higgs bosons , and the @xmath1 bosons in the context of the minimal susy su(5 ) gut @xcite .
they have found that while the masses of the adjoint higgs and the @xmath1 bosons are around @xmath0 gev , the mass of the color - triplet higgs boson lies in the region of @xmath9 gev , which is significantly below the gut scale .
this implies that the threshold corrections to the gauge coupling constants are still not small , even in the susy guts .
in fact , the analysis is quite sensitive to the mass spectrum in the intermediate scale , such as those of the susy particles . in ref .
@xcite , all of the particles except for gauginos are assumed to be at 1 tev .
the gaugino masses are set to be around the electroweak scale with the gut relation for the gaugino masses being assumed .
currently , on the other hand , the susy models with heavy sfermions have been widely discussed with a lot of attention @xcite .
although such models originally have been discussed as a possible candidate of susy models from a theoretical point of view , now they are also supported by the latest results of the lhc experiments ; no significant excess in the susy searches @xcite and the discovery of the 126 gev higgs boson @xcite both suggest that the susy breaking scale is somewhat higher than the electroweak scale .
while this high - scale susy scenario is difficult to be probed at the collider experiments , it still has a lot of phenomenological consequences which might be checked in other experiments .
recent studies on the subject are given in refs .
@xcite .
as mentioned above , the gut scale mass spectrum inferred from the indirect analysis presented in refs .
@xcite is highly dependent on the susy mass spectrum .
in particular , if the susy particles have masses much larger than @xmath10 tev , previous results are expected to be changed significantly . in this letter , therefore , we revisit the analysis in the high - scale susy scenario .
we carry out the calculation in the minimal susy su(5 ) gut with sfermions having masses well above the electroweak scale .
we will see that while the constraint on the adjoint higgs and @xmath1 boson masses only differs from previous ones slightly , that on the color - triplet higgs mass is found to be changed by more than an order of magnitude and actually improved in the sense that it may also be around the gut scale .
interestingly enough , this result again indicates that the high - scale susy scenario is rather supported than the traditional low - energy susy models .
this letter is organized as follows : in sec .
[ model ] , the high - scale susy model which we discuss below is presented , and its phenomenological aspects are briefly described .
the mass spectrum of the superheavy particles in the minimal susy su(5 ) gut is also displayed there . in the subsequent section ,
we discuss the method of constraining the masses of the gut - scale multiplets by means of the renormalization group equations ( rges ) as well as the threshold corrections of the gauge couplings .
then , we show some results in sec
. [ results ] .
section .
[ conclusion ] is devoted to conclusions and discussion .
let us begin by presenting a high - scale susy model discussed in this letter .
we consider the particle content of the minimal supersymmetric standard model ( mssm ) .
then , the only assumption which we adopt here is that there exists a dynamical susy - breaking sector with the khler potential having a generic structure .
then , all of the scalar bosons except the lightest higgs boson acquire masses of the order of the gravitino mass @xmath11 , while the lightest higgs boson mass is fine - tuned to be @xmath12 gev .
the higgsinos may in general have similar masses to the gravitino mass , though they might be suppressed if there are some extra chiral symmetries .
the gaugino masses are , on the other hand , generated by the quantum effects @xcite and , thus , suppressed by a loop factor compared with @xmath11 .
there are two kinds of contributions to the gaugino masses ; one is the anomaly mediation effect @xcite and the other is from the higgsino - higgs boson loop diagram .
these two effects give rise to the gaugino masses as @xmath13 with @xmath14 representing the higgsino - higgs boson loop contribution .
here , @xmath15 and @xmath16 ( @xmath17 and 3 ) are the u(1)@xmath18 , su(2)@xmath19 , and su(3)@xmath20 gaugino masses and gauge coupling constants , respectively .
we use the su(5 ) normalization for the u(1)@xmath18 coupling , _
i.e. _ , @xmath21 .
further , @xmath22 and @xmath23 denote the higgsino and the heavy higgs boson masses , respectively , which are assumed to be the same order of magnitude as sfermion masses .
@xmath24 is the ratio of the vacuum expectation values ( vevs ) of the higgs fields ; @xmath25 . since the higgsino mass is presumed to be around the gravitino mass , the higgsino - loop contribution @xmath26 is also expected to be as large as @xmath11 .
the values of the gaugino masses are , however , dependent on the relative phase between @xmath22 and @xmath11 . in the following discussion
we just assume the gaugino masses are lighter than the scalar masses by a loop factor , and regard them as free parameters .
as mentioned in introduction , this model has a lot of fascinating features from a phenomenological point of view .
they are originally discussed in refs .
@xcite , and recent development is given in refs .
@xcite . in these works , a typical scale for sfermion masses is taken to be @xmath27@xmath28 ) tev , which explains the 126 gev higgs boson mass . with such heavy particles the susy flavor and cp problems @xcite as well as the gravitino problem are considerably relaxed . in this case ,
the gaugino masses are @xmath29(1 ) tev because of an one - loop factor .
note that from eq .
it is found that with a moderate value of @xmath26 , wino becomes the lightest among gauginos , and thus the lightest susy particle in this model .
it is quite interesting since the wino dark matter with a mass of @xmath303 tev is consistent with cosmological observations @xcite .
the dark matter might be searched directly @xcite or indirectly @xcite in future dark matter experiments .
moreover , the gauge coupling unification in this high - scale susy model is found to be achieved as precisely as that in the mssm .
thus the susy gut is still promising in the case of high - scale supersymmetry . in the next section ,
we in turn require the gauge coupling unification and constrain the gut scale mass spectrum by using the requirement .
now , we briefly summarize the superheavy particles in the minimal susy su(5 ) gut , which we adopt as the working hypothesis in this paper .
the susy su(5 ) gauge theory includes twenty - four gauge superfields @xmath31 with @xmath32 . by using the su(5 )
generators @xmath33 we define a @xmath34 matrix representation of the vector superfields such that @xmath35 , with the components written as @xmath36 we collectively call @xmath37 and @xmath38 the @xmath1-bosons hereafter .
the unified gauge group su(5 ) is spontaneously broken by the vev of the adjoint higgs boson @xmath39 ( @xmath40 ) to su(3)@xmath41su(2)@xmath42u(1)@xmath18 . again , we define @xmath43 and its components by @xmath44 the mssm higgs superfields are , on the other hand , incorporated into fundamental and anti - fundamental representations as follows : @xmath45 where @xmath46 are the mssm higgs superfields . @xmath47 and @xmath48 are called the color - triplet higgs multiplets .
the superpotential of the higgs sector in the minimal su(5 ) susy gut is given by @xmath49 after the adjoint higgs field gets the vev , @xmath50 ( @xmath51 ) , the su(5 ) gauge group is broken to the sm su(3)@xmath41su(2)@xmath42u(1)@xmath18 gauge groups without breaking the supersymmetry .
also the parameter @xmath52 is fine - tuned as @xmath53 in order to realize the doublet - triplet mass splitting in @xmath54 and @xmath55 . with the condition ,
the mass of the color - triplet higgs boson is @xmath56 .
the @xmath1-boson mass is @xmath57 with @xmath58 the unified gauge coupling constant .
as regards the adjoint higgs multiplets , the components @xmath59 and @xmath60 have masses of @xmath61 and @xmath62 has @xmath63 .
the components @xmath64 and @xmath65 become the longitudinal component of the @xmath1-bosons , and thus do not show up as physical states .
in this section , we present the rges of the gauge and yukawa coupling constants , as well as the boundary conditions at each threshold .
we use the @xmath66 scheme @xcite in this work .
first , we write down the two - loop beta functions @xcite . in the mssm , the two - loop rges for the gauge coupling constants are given as @xmath67~,\ ] ] where @xmath68 and @xmath69 with @xmath70 ( @xmath71 ) the top , bottom and tau yukawa coupling constants , respectively .
since the yukawa couplings enter into the two - loop level contributions to the gauge coupling rges , it is sufficient to consider the rges for the yukawa couplings at one - loop level .
they are given as @xmath72 , \nonumber \\
\mu\frac{\partial}{\partial \mu}y_b & = \frac{1}{16\pi^2}y_b \biggl[6y_b^2+y_t^2+y_\tau^2 -\frac{7}{15}g^2_1 - 3g_2 ^ 2-\frac{16}{3 } g_3 ^ 2\biggr ] , \nonumber \\
\mu\frac{\partial}{\partial \mu}y_\tau & = \frac{1}{16\pi^2}y_\tau \biggl[3y_b^2 + 4y_\tau^2 -\frac{9}{5}g^2_1 - 3g_2 ^ 2\biggr ] .\end{aligned}\ ] ] below the susy breaking scale ( @xmath73 ) , the squarks and sleptons , the higgsinos , and the heavy higgs boson masses are decoupled so that the theory is regarded as the sm with gauginos .
the contribution of gauginos and the sm particles to the coefficients of the beta functions is given as @xmath74 @xmath75 and @xmath76 where the subscripts `` sm '' and `` gaugino '' indicate that the contributions are of the sm particles and gauginos , respectively .
the running of the yukawa couplings in this case is given as follows : @xmath77 , \nonumber \\
\mu\frac{\partial}{\partial \mu}y_b&=\frac{1}{16\pi^2}y_b \biggl [ \biggl ( \frac{3}{2}y_t^2+\frac{9}{2}y_b^2 + y_\tau^2 -\frac{1}{4}g^2_1-\frac{9}{4}g_2 ^ 2 - 8 g_3 ^ 2\biggr)_{\rm sm } ~ \biggr ] , \nonumber \\
\mu\frac{\partial}{\partial \mu}y_\tau&=\frac{1}{16\pi^2}y_\tau \biggl [ \biggl(3y_t^2 + 3y_b^2+\frac{5}{2}y_\tau^2 -\frac{9}{4}g^2_1-\frac{9}{4}g_2 ^ 2\biggr)_{\rm sm } ~ \biggr ] .\end{aligned}\ ] ] next , we consider the matching conditions at each threshold scale . at the gut scale , the gauge coupling constants in the su(3)@xmath41su(2)@xmath78u(1)@xmath18 gauge theories are equated to the unified coupling constant @xmath58 with the following threshold corrections at one - loop level @xcite : @xmath79~,\nonumber \\
\frac{1}{g_2 ^ 2(\mu)}&=\frac{1}{g_5 ^ 2(\mu)}+\frac{1}{8\pi^2}\biggl [ 2\ln \frac{\mu}{m_\sigma}-6\ln\frac{\mu}{m_x } \biggr]~,\nonumber \\
\frac{1}{g_3 ^ 2(\mu)}&=\frac{1}{g_5 ^ 2(\mu)}+\frac{1}{8\pi^2}\biggl [ \ln \frac{\mu}{m_{h_c}}+3\ln \frac{\mu}{m_\sigma}-4\ln\frac{\mu}{m_x } \biggr]~.\end{aligned}\ ] ] here , the conditions do not include constant ( scale independent ) terms since we use the @xmath66 scheme for the renormalization @xcite . from the equations it immediately follows that @xmath80 the relations allow us to evaluate the masses of the heavy particles , @xmath81 and @xmath82 , from the gauge coupling constants determined in the low - energy experiments through the rges @xcite .
while the couplings are well measured with high precision , the estimation is quite dependent on the spectrum in the intermediate region , especially on the masses of gauginos and higgsinos . for the susy breaking threshold
, we just equate the gauge couplings above and below the threshold , and change the beta functions appropriately for each region .
this approximation is valid since the particles appearing at the scale are assumed to be degenerate in mass . in the case of gauginos , on the other hand , we need to consider the threshold corrections since the mass difference among gauginos might be sizable . the condition is @xmath83 where @xmath84 are the couplings in the sm while @xmath85 are those above the gaugino threshold .
the yukawa couplings are matched as usual , _
i.e. _ , at the susy breaking scale , the yukawa couplings @xmath86 below the susy breaking scale are matched with the supersymmetric ones , @xmath87 , as follows : @xmath88 before concluding this section , we solve the rges at one - loop level and , taking the threshold corrections into account , derive relations between the superheavy masses and the low - energy gauge coupling constants .
such relations reflect the dependence of @xmath81 and @xmath82 on the mass spectrum of the susy particles . by inserting to eq .
the one - loop solutions of the rges for the gauge couplings , we have @xmath89~,\label{mhc}\\ \frac{5}{\alpha_1(m_z)}-\frac{3}{\alpha_2(m_z)}-\frac{2}{\alpha_3(m_z ) } & = \frac{1}{2\pi}\biggl [ 12\ln\biggl(\frac{m_x^2m_\sigma}{m_z^3}\biggr)+4\ln\biggl ( \frac{m_2}{m_z } \biggr)+4\ln\biggl(\frac{m_3}{m_z}\biggr ) \biggr]~.\label{mgut}\end{aligned}\ ] ] from eq .
we find that the mass of the color - triplet higgs @xmath81 gets larger as the susy breaking scale @xmath73 is taken to be higher .
this originates from the mass difference among the components of the fundamental higgs multiplets , _
i.e. _ , the triplet - higgs , higgsinos , heavy higgs bosons , and the lightest higgs boson .
therefore , the behavior of @xmath81 with respect to the susy breaking scale is universal in a sense .
further , @xmath81 depends only on the ratio of @xmath90 and @xmath91 .
@xmath92 is , on the other hand , independent of the susy braking scale @xmath73 while dependent on the scale of the gauginos , not their ratio .
this is because the right - hand side of eq .
results from the mass difference in the gauge vector multiplets and the adjoint higgs multiplet , a part of which is included as the longitudinal mode of the gauge multiplets .
it is also found that @xmath92 decreases when the gaugino masses are taken to be large values .
this is owing to the opposite sign of the contribution of gauge fields to the gauge beta functions to those of matter fields .
this feature is , therefore , again model - independent . in the subsequent section ,
we carry out a similar analysis using the two - loop rges .
as functions of the susy breaking scale @xmath73 ( pink lines ) . here , wino mass @xmath90 is fixed to be 3 tev and @xmath93
. gluino - wino mass ratio , @xmath94 , is set to be @xmath95 and @xmath96 from top to bottom , respectively .
theoretical errors coming from the strong coupling constant @xmath97 @xcite are also shown .
horizontal blue line shows a result in the case of low - energy susy ( @xmath98 tev , @xmath99 gev , and @xmath100).,height=283 ] as functions of gluino mass @xmath91 ( pink lines ) . here , @xmath93 and @xmath101 tev .
upper and lower lines correspond to @xmath102 tev and 300 gev , respectively .
error bars indicate the input error of the strong coupling constant @xmath97 @xcite .
horizontal blue line shows a result in the case of low - energy susy ( @xmath98 tev , @xmath99 gev , and @xmath100).,height=283 ] now we present some results for the rge analysis which we discuss in the previous section . as noted above ,
the running of the gauge couplings is computed at two - loop level and the threshold corrections are taken into account at one - loop level .
the masses of sfermions , heavy higgs bosons , and higgsinos are taken to be @xmath73 for brevity .
gaugino masses are assumed to be lighter than @xmath73 by an one - loop factor .
first , we consider the color - triplet higgs mass @xmath81 . in fig . [ tanb3mhc ]
, we plot the dependence of @xmath81 on the susy breaking scale @xmath73 in the pink lines . here , the wino mass @xmath90 is fixed to be 3 tev , which is favored from the thermal relic abundance , and @xmath93 .
the ratio of the gluino and wino masses , @xmath94 , is set to be @xmath95 and @xmath96 from top to bottom , respectively .
further , we show the error of the calculation coming from that of the strong coupling constant @xmath97 @xcite . the horizontal blue line shows a result in the case of low - energy susy ( @xmath98 tev , @xmath99 gev , and @xmath100 ) as a reference . in this case
, we have @xmath103 .
this figure well illustrates the feature read from the approximated expression given in eq . ;
@xmath81 increases as the susy breaking scale grows while it decreases when the ratio @xmath94 becomes large . to see the latter feature more clearly
, we show its dependence on the gluino mass @xmath91 .
again , we set @xmath93 , and the susy breaking mass is fixed to be @xmath101 tev .
the upper and lower lines correspond to @xmath102 tev and 300 gev , respectively .
these two figures show that @xmath81 is strongly dependent on @xmath73 and @xmath94 . therefore , to predict the mass with high accuracy , precise determination of the masses of gauginos as well as the susy breaking scale is inevitable . any way , in the high - scale susy scenario it is found to be possible for the mass of the color - triplet @xmath81 to be around @xmath104 gev , which is expected by the gauge coupling unification .
as functions of gluino mass @xmath91 ( pink lines ) . here ,
@xmath93 and @xmath101 tev .
upper and lower lines correspond to @xmath105 gev and 3 tev , respectively .
error bars indicate the input error of the strong coupling constant @xmath97 @xcite .
horizontal blue line shows a result in the case of low - energy susy ( @xmath98 tev , @xmath99 gev , and @xmath100).,height=283 ] next , we discuss constraints on @xmath106 derived from the relation . from now on we define @xmath107 and refer to it as the gut scale .
the equation tells us that the gut scale depends on only the gaugino masses at the leading order , so we express @xmath8 as functions of the gaugino masses . in fig .
[ m3mgut ] we plot it as functions of gluino mass . here
again we fix @xmath93 and @xmath101 tev .
the upper and lower lines correspond to @xmath105 gev and 3 tev , respectively .
again , the horizontal blue line shows a result in the case of low - energy susy with @xmath98 tev , @xmath99 gev , and @xmath100 , which gives @xmath108 gev .
the error bars indicate the input error of the strong coupling constant @xmath97 @xcite , though the effect is negligible . we see that the gut scale also has little dependence on the gaugino masses . in that sense ,
the prediction is robust compared with that for @xmath81 .
however , as discussed in the previous section , the gut scale @xmath8 gets lower when the gaugino masses become larger ( @xmath109 ) .
this feature is quite interesting when one considers the proton decay via the @xmath1-boson exchange processes .
although the change in @xmath8 is small , it might be significant since the proton decay lifetime scales as @xmath110 .
for instance , when @xmath111 gev , the proton lifetime via the @xmath1-boson exchange reduces to around @xmath112 years @xcite , which is slightly above the current experimental bound , @xmath113 yrs @xcite .
finally , we briefly comment on the @xmath114 dependence of the results .
although the one - loop computation is not related with @xmath114 , the two - loop results might be affected through the running of the yukawa couplings .
we have found , however , that the effects on the results are negligible .
in this letter , we have presented constraints on the masses of the gut scale particles from the gauge coupling unification in the case of the high - scale susy scenario . to that end , we have used the two - loop rges for the gauge couplings with one - loop threshold corrections considered . as a result ,
the mass of the color - triplet higgs multiplets @xmath81 turns out to be considerably large compared with previous results in the traditional low - energy susy scenario , while the gut scale is found to be slightly lower .
these are generic features resulting from the mass difference among the components of the same supermultiplet of the su(5 ) gauge group .
interestingly , all of the superheavy particles might be around @xmath0 gev in the high - scale susy models .
the mass spectrum of the gut scale particles predicted here stimulates us to reconsider the proton decay in the case of high - scale supersymmetry . as mentioned to above
, the relatively low gut scale enhances the proton decay rate via the @xmath1-boson exchanging process .
if the enhancement is strong enough , the proton decay in the @xmath115 channel might be searched in future experiments .
furthermore , the experiments may also reach the proton decay through the color - triplet higgs exchange . since the decay process predicts too short lifetime @xcite ,
some suppression mechanism for the process has been assumed .
nevertheless , with @xmath81 larger than those considered in the previous literature and sfermions much heavier than the electroweak scale , this process might evade the current experimental bound without any mechanism of limiting the color - triplet higgs exchanging process .
in such a case , it is possible for the @xmath116 mode , which is the main decay mode in the case of the color - triplet higgs exchange , to be searched in near future .
a detailed analysis of this decay process will be given elsewhere @xcite .
note that , in proposed models where the color - triplet higgs exchange is suppressed by some mechanism , large threshold corrections to the gauge coupling constants at the gut scale tend to appear .
one of the examples is introduction of the peccei - quinn symmetry @xcite .
it was found that when the threshold corrections at the gut scale are small , the suppression mechanism does not work @xcite . in the susy su(5 ) guts in higher dimensional space the u(1)@xmath117 symmetry
forbids the dimension - five proton decay , though the kaluza - klein particles generate large threshold corrections to the gauge coupling constants @xcite .
the high - scale supersymmetry would be another solution for the proton decay problem , in which such large threshold corrections are not required .
the work of n.n . is supported by research fellowships of the japan society for the promotion of science for young scientists .
the work of j.h . is supported by grant - in - aid for scientific research from the ministry of education , science , sports , and culture ( mext ) , japan , no .
20244037 , no . 20540252 , no .
22244021 and no . 23104011 , and also by world premier international research center initiative ( wpi initiative ) , mext , japan .
s. dimopoulos , s. raby and f. wilczek , phys .
d * 24 * , 1681 ( 1981 ) .
w. marciano and g. senjanovi ' c , phys .
d * 25 * , 3092 ( 1982 ) .
m.b . einhorn and d.r .
jones , nucl .
b * 196 * , 475 ( 1982 ) .
u. amaldi , w. de boer and h. furstenau , phys .
b * 260 * , 447 ( 1991 ) .
p. langacker and m. -x .
luo , phys .
d * 44 * , 817 ( 1991 ) .
j. hisano , h. murayama and t. yanagida , phys .
lett . * 69 * , 1014 ( 1992 ) .
j. hisano , h. murayama and t. yanagida , nucl .
b * 402 * , 46 ( 1993 ) .
h. murayama and a. pierce , phys .
d * 65 * , 055009 ( 2002 ) .
g. f. giudice , m. a. luty , h. murayama and r. rattazzi , jhep * 9812 * , 027 ( 1998 ) .
j. d. wells , hep - ph/0306127 . n. arkani - hamed and s. dimopoulos , jhep * 0506 * , 073 ( 2005 ) .
g. f. giudice and a. romanino , nucl .
b * 699 * , 65 ( 2004 ) [ erratum - ibid .
b * 706 * , 65 ( 2005 ) ] .
n. arkani - hamed , s. dimopoulos , g. f. giudice and a. romanino , nucl .
b * 709 * , 3 ( 2005 ) .
j. d. wells , phys . rev .
d * 71 * , 015013 ( 2005 ) .
l. j. hall and y. nomura , jhep * 1003 * , 076 ( 2010 ) .
l. j. hall and y. nomura , jhep * 1201 * , 082 ( 2012 ) .
g. aad _ et al . _ [ atlas collaboration ] , arxiv:1208.0949 [ hep - ex ] .
[ atlas collaboration ] , atlas - conf-2012 - 109 .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
lett . * 109 * , 171803 ( 2012 ) .
g. aad _ et al . _ [ atlas collaboration ] , phys . lett .
b * 716 * , 1 ( 2012 ) .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys . lett .
b * 716 * , 30 ( 2012 ) .
g. f. giudice and a. strumia , nucl .
b * 858 * , 63 ( 2012 ) .
m. ibe and t. t. yanagida , phys .
b * 709 * , 374 ( 2012 ) .
m. ibe , s. matsumoto and t. t. yanagida , phys .
d * 85 * , 095011 ( 2012 ) .
l. e. ibanez and i. valenzuela , arxiv:1301.5167 [ hep - ph ] .
j. hisano , k. ishiwata and n. nagata , phys .
d * 87 * , 035020 ( 2013 ) .
k. s. jeong , m. shimosuka and m. yamaguchi , jhep * 1209 * , 050 ( 2012 ) .
r. saito and s. shirai , phys .
b * 713 * , 237 ( 2012 ) .
r. sato , s. shirai and k. tobioka , jhep * 1211 * , 041 ( 2012 ) .
b. bhattacherjee , b. feldstein , m. ibe , s. matsumoto and t. t. yanagida , phys .
d * 87 * , 015028 ( 2013 ) .
l. j. hall , y. nomura and s. shirai , jhep * 1301 * , 036 ( 2013 ) .
n. arkani - hamed , a. gupta , d. e. kaplan , n. weiner and t. zorawski , arxiv:1212.6971 [ hep - ph ] .
t. moroi and m. nagai , arxiv:1303.0668 [ hep - ph ] .
d. mckeen , m. pospelov and a. ritz , arxiv:1303.1172 [ hep - ph ] .
l. randall and r. sundrum , nucl .
b * 557 * , 79 ( 1999 ) .
f. gabbiani , e. gabrielli , a. masiero and l. silvestrini , nucl .
b * 477 * , 321 ( 1996 )
. j. hisano , s. matsumoto , m. nagai , o. saito and m. senami , phys .
b * 646 * , 34 ( 2007 ) .
j. hisano , s. matsumoto and m. m. nojiri , phys .
* 92 * , 031303 ( 2004 ) .
j. hisano , s. matsumoto , m. m. nojiri and o. saito , phys .
d * 71 * , 063528 ( 2005 ) .
w. siegel , phys .
b * 84 * , 193 ( 1979 ) .
m. e. machacek and m. t. vaughn , nucl .
b * 222 * , 83 ( 1983 ) .
s. weinberg , phys .
b * 91 * , 51 ( 1980 ) .
l. j. hall , nucl .
b * 178 * , 75 ( 1981 ) .
i. antoniadis , c. kounnas and k. tamvakis , phys .
b * 119 * , 377 ( 1982 ) .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
j. hisano , d. kobayashi and n. nagata , phys .
b * 716 * , 406 ( 2012 ) .
h. nishino , k. abe , y. hayato , t. iida , m. ikeda , j. kameda , y. koshio and m. miura _ et al .
_ , arxiv:1203.4030 [ hep - ex ] .
j. hisano , d. kobayashi , t. kuwahara and n. nagata , arxiv:1304.3651 [ hep - ph ] .
r. d. peccei and h. r. quinn , phys .
lett . * 38 * , 1440 ( 1977 ) .
j. hisano , h. murayama and t. yanagida , phys .
b * 291 * , 263 ( 1992 ) .
j. hisano , t. moroi , k. tobe and t. yanagida , phys .
b * 342 * , 138 ( 1995 ) .
l. j. hall and y. nomura , phys .
d * 64 * , 055003 ( 2001 ) . | a constraint on masses of superheavy gauge and higgs multiplets at the grand unification ( gut ) scale is obtained from the gauge coupling unification in the case of high - scale supersymmetry .
we found that all of the particles may lie around a scale of @xmath0 gev so that the threshold corrections to the gauge coupling constants at the gut scale are smaller than those in the case of the low - energy supersymmetry .
in addition , the gut scale tends to be slightly lower when the gauginos are heavier and , thus , the proton decay rate via the @xmath1-boson exchange process is expected to be enhanced .
ipmu 13 - 0075 1.35 cm 1.2 cm junji hisano@xmath2 , takumi kuwahara@xmath3 , and natsumi nagata@xmath4 + 0.4 cm _
@xmath3department of physics , nagoya university , nagoya 464 - 8602 , japan _ + _ @xmath5kavli institute for the physics and mathematics of the universe ( wpi ) , todai institutes for advanced study , the university of tokyo _ , kashiwa 277 - 8568 , japan + _ @xmath6department of physics , university of tokyo , tokyo 113 - 0033 , japan _
1.5 cm |
extensive experimental studies of high - t@xmath0 materials have been able to determine a phase - diagram for hole and electron doped members of the cuprates .
undoped materials are anti - ferromagnetic ( afm ) insulators , where every second cu along [ 1,0,0 ] has opposite spin .
superconductivity on the hole doped side appears for @xmath5 , the number of holes per cu , being approximately between 0.05 and 0.25 , with a maximum of t@xmath0 for @xmath6 @xcite .
a pseudo - gap appears below a temperature t@xmath7 which is considerably larger than t@xmath0 when @xmath5 is lower ( `` underdoped '' ) than that of optimal t@xmath0 @xcite .
the width of this gap is fairly constant up to t@xmath7 , where it disappears quickly .
magnetic moments are detected in afm fluctuations as shown from studies by neutrons @xcite .
side peaks seen in neutron scattering suggests that the basic afm correlation is modulated with different periodicity as the doping is changing up to a maximum @xmath5 @xcite .
static or dynamic stripe structures appear as regions of different spin and charge along the cuo bond direction ( [ 1,0,0 ] or [ 0,1,0 ] ) .
results from angular resolved photoemission ( arpes ) , taken at low t , indicate that the cylindrical m - centred fermi surface ( fs ) `` disappears '' near the crossing with the x - m or y - m lines in underdoped cases , while the band crossing on the [ 1,1,0 ] line persists as a `` fs - arc '' @xcite .
the lattice is also influenced by doping so that oxygen phonon modes become softer as the doping is increased @xcite@xmath8@xcite .
more evidence of the importance of lattice interaction comes from studies of isotope effects , such as on t@xmath0 , where the effect is rather weak , on t@xmath7 , on the penetration depth @xcite@xmath8@xcite and recently on the peak structures seen in photoemission @xcite . on the theoretical side
the problem with local - density band calculations is that they do not find the afm insulating ground state of the undoped systems .
the calculated ground state is metallic and non - magnetic , but the fs compare well with positron annihilation experiments for the doped cuprates @xcite .
these and other features need to be understood in the complex physics of high - t@xmath0 cuprates .
the present study of properties high - t@xmath0 physics is based on electronic structure calculations for hole doped supercells of two of the simpler copper oxide materials , hgba@xmath4cuo@xmath3 ( hbco ) and la@xmath1sr@xmath2cuo@xmath3 ( lsco ) .
the fermi surfaces ( fs ) of these two cuprates contain only the m - centred barrel - like cylinder , which originates from the 2-dimensional cuo - planes .
other smaller fs - pockets or ridges are found in the more complex high - t@xmath0 materials @xcite , but it is probable that superconductivity is localized to the barrels , since they are common to all of the high-@xmath9 cuprates .
section ii describes the band calculations and the main results . in section iii
is presented different models , based on spc and the results of the band calculations , for different properties of the high-@xmath9 cuprates .
electronic structure calculations , based on the local density approximation , lda @xcite and the linear muffin - tin orbital ( lmto ) method @xcite are used to determine the bands for unit cells of different size and doping .
hole doping is introduced via the virtual crystal approximation , in which electrons and nuclear charges on la are reduced in lsco . in hbco the reduction of electronic charge is compensated by reducing the nuclear charges about equally on cu and o sites . the the effect of disorder due to local la - sr substitutions or interstitial o doping is not included .
the calculations for hbco include empty spheres ( 5 per formula unit , f.u . ) in the open part of the cells .
no empty spheres are introduced in the structures of lsco .
larger supercells were considered for lsco ( 7 sites per f.u . ) than for hbco ( 13 sites per f.u . ) .
the supercells include phonon displacements of oxygens in the cuo plane and/or spin waves with moments on cu sites .
the stacking sequence along @xmath10 is tetragonal for hbco ( [ 0,0,z ] ) , while in lsco it is body centered tetragonal ( [ 0.5,0.5,z ] )
. the present work does not investigate the coupling between different cuo planes .
the bands near @xmath11 in the calculations for the 1 f.u .
cells agree well with other calculations .
other details of the band calculations and some results on spc can be found in previous publications @xcite .
it can be anticipated from the nearly - free - electron ( nfe ) model that gaps appear at zone - boundaries @xmath12 because of small potential perturbations @xmath13 @xcite .
this has been confirmed in band calculations on hgba@xmath4cuo@xmath3 , when the potential has modulations along [ 1,1,0 ] or [ 1,0,0 ] in the cuo - plane because of atomic displacements @xcite .
atomic displacements of the planar o - atoms on both sides of a cu - atom are alternatively displaced towards ( `` compressed '' ) or away from the cu ( `` expanded '' ) . calculations with anti - ferromagnetic ( afm ) field on the cu atoms are made to model spin - waves .
it is assumed that adiabatic conditions exist , so that the electronic time scale is much shorter than that of phonons .
only relatively slow spin waves are of interest here , when the time scale of spin waves is of the same order as for phonons . in this case
one can perform `` frozen ''
spin - wave and `` frozen '' phonon calculations .
when the calculations are made for coexisting phonon and spin waves , it turns out that the magnetic moments are largest on the `` expanded '' cu - sites .
the nodes of modulated spin waves are chosen to coincide with `` compressed '' cu - sites .
figure 1 shows the cuo plane containing a phonon distortion and the spin wave , both oriented along the cuo bond direction , [ 1,0,0 ] .
the phonon mode creates a potential modulation of the coulomb potential ( a fourier component @xmath13 ) .
spin waves induce a modulation in the spin - polarized part of the potential and magnetic moments on the cu .
gaps appear at different energies depending on the wave length of the modulation .
the gain in kinetic energy is maximized when the gap is at @xmath11 , which leads to a correlation between wavelength , @xmath11 and doping .
the wavelength of the spin wave is twice that of the phonon , because the spin can be up or down at `` expanded '' cu - positions .
the bands show two gaps for modulations along [ 1,1,0 ] , one due to the phonon and one at a higher energy because of the spin wave , and there is no direct spc for electrons at @xmath11 .
however , for modulations along [ 1,0,0 ] two rows of cuo exist along the unit cell , and the phase of the spin wave along each row differ by @xmath14 .
the band results show in this case a constructive spc , where the gap from separate phonon and afm - modulations will open a gap at the same energy , i.e. at @xmath11 for the correct doping .
in addition , a phonon with large atomic displacements will increase the moments along the spin wave and the size of the gap .
this situation reminds of a ( dynamic ) peierls state , although it is not necessary that @xmath11 coincides with a peak in the density - of - states ( dos ) .
several calculations have been made where partial gaps appear at @xmath11 for different doping levels for hbco and lsco with different wave length ( periodicity ) of phonons and/or spin waves .
large cells , containing from 56 to 224 sites , are used for studies of long periodicities , and the number of k - points are reduced accordingly ( 48 to 18 points in the irreducible zone ) . sometimes this introduces structures in the dos even when the cell contain no wave .
to search for partial gaps it is necessary to compare the dos functions between cases with and without phonon and/or spin waves , and in some cases also when phonon or spin amplitudes are different . in figure 2
is shown one typical example of dos and partial gaps .
these dos functions are calculated for 16 f.u .
( 112-sites ) cells of lsco , extended 8 f.u . along [ 1,0,0 ] ( twice as long as in fig . 1 ) and 2 f.u . along [ 0,1,0 ] , and the dos for the normal 1 f.u .
cell is shown for comparison .
the periodicity of the spin and/or o - phonon wave along [ 1,0,0 ] makes a dip in the dos at @xmath11 for a doping of 4 holes per cell , i.e. for @xmath5=0.25 .
a shorter cell , as in fig 1 , opens the gap at lower energy relative to @xmath15 , the position of @xmath11 in undoped lsco , at 0.5 holes per cu .
the longer the periodicity is the smaller doping is needed for having the gap at @xmath11 .
infinite periodicity corresponds to undoped systems .
it is often seen that weaker gaps appear symmetrically above the position of @xmath15 , suggesting that partial gaps should appear also in electron doped cases .
an effort to study the dependence of the partial gap as a function of the wave length has been done for lsco using 4 , 8 and 16 cells in the [ 1,0,0 ] direction . with always 2@xmath16 along [ 0,1,0 ] and 7 sites per cell , this implies calculations with 56 ( as in fig.1 ) , 112 and 224 atoms per unit cell .
the difficulty is to have a smooth dos for the longest unit cells with only a limited number of k - points . in table
i is shown the maximal deviation of charge on cu - sites , @xmath17 , for calculations with a phonon distortion along the cells .
the distortions are 0.02@xmath16 on the oxygens between the cu sites in all cases , see fig . 1 for the cell of length 4@xmath16 . the distortions are `` step - like '' ( not sinusoidal ) for the longer cells .
the result is that the maximal deviation of cu charges increases with the length of the cell , i.e. a long phonon wavelength imply a strong charge modulation , a large potential perturbation and a large gap .
spin - polarized calculations for `` step - like '' modulated spin waves within the same structures , where magnetic fields of @xmath18 mry are applied on the cu - sites , give a distribution of magnetic moments on the cu sites .
the moments are zero on two node positions with a maximal amplitude far from the node in the interior of a magnetic stripe .
the maximal amplitudes are shown in table i , and it is seen that the cell with the intermediate length has the largest moments .
the spc seems less effective for the longest cell .
the shortest cell also has low moments because there is only one row of magnetic cu between the nodes , while in the intermediate case there are three rows of magnetic cu sites .
if the results based on these three cases are correct , one can conclude the following : the spc is optimal for the wave of intermediate length , while for less doping and longer wavelengths the phonon distortions become relatively more important .
but the interpretation can be different if perpendicular coupling is considered .
the results of the ab - initio calculations for frozen spin - phonon waves have to be reinterpreted for the real case of dynamic waves , where the partial gaps are less visible .
the idea is that spc waves exist below a certain temperature ( @xmath19 ) ( see later ) .
the spin wave is the most important of the two waves for the gap , but it needs the phonon displacements to develop and it is quenched above @xmath19 .
this scenario is not proven through ab - initio calculations of total energies , but models in the following sections are used for a qualitative account of the expected properties of such a mechanism .
it was concluded above that magnetic fluctuations are enforced on cu when oxygens are moved away from the cu , as is depicted in fig .
1 . the first column of cu atoms ( site 1 and 2 ) and the third column ( site 5 and 6 ) have large moments because there is more space around them .
zero moment nodes are on the second and fourth columns of cu , to produce a spin - wave modulation in the rows along @xmath20 .
it can be argued that spin waves are also possible along @xmath21 on the first and third columns of fig .
1 . coupling with phonons along @xmath21 is likely too .
the wavelengths along columns and rows need not to be equal .
this type of perpendicular spin - phonon coupling will generate double gaps in the dos .
a possible arrangement of spin wave and phonon configurations is shown in figure 3 , where the simple cell in figure 1 is turned vertically and repeated four times horizontally .
a short wavelength modulation along @xmath21 in fig .
3 , coexists with a twice as long spin wave along @xmath20 .
this is about the smallest possible cell for complete spin waves and phonons .
it requires 224 atomic sites for lcso ( or 416 for hbco ) , which is a bit large for well converged ab - initio calculations . in fig 4
is shown the dos for a truncated cell , half as large along @xmath20 and @xmath21 as the cell in fig .
a single phonon is oriented along @xmath21 and a spin wave along @xmath20 .
the number of k - points is insufficient for a smooth dos , but the figure shows that two gaps emerge simultaneously as function of the strength of the spin wave .
the two gaps are found at the expected positions for waves of these lengths along @xmath20 and @xmath21 .
the gap at low energy is due to the short wave along @xmath21 , while a second partial gap develops at higher energy because of a longer spin wave along @xmath20 . in figure 5
is shown the dos for two different amplitudes of the phonon distortion , but with the same strength of the spin wave .
again , it is found that both gaps appear to be stronger if the strength of one of the waves becomes stronger .
these results indicate that the spc can be mediated between perpendicular directions , to produce double gap features in the dos .
it is assumed that the gaps exist in the normal state below @xmath19 , although they should be easiest to observe at low t. the doping dependence of two gaps in bscco observed from tunneling below @xmath9 by zasadinski et .
@xcite suggests , together with the mechanism of perpendicular spc , that the lower gap is always at a fixed energy relative to @xmath15 .
however , the main upper gap moves downwards , i.e. closer to the low energy gap when the doping increases .
in other words , there are fewer states between the two gaps at large doping .
the calculated dos for one cell of hgba@xmath4cuo@xmath3 , shown in fig .
6 , is ranging from 11 to 20 states per ry and cuo unit in the relevant energy region .
a doping of 0.25 holes / cu corresponds to a spc wave of 8@xmath16 , twice as long as in fig . 1 . ( the band results can be summarized in the relation @xmath22 , where the doping @xmath5 is in units of holes / cu and
the wavelength @xmath23 is in units of @xmath16 . )
a spc wave of this length , consisting of three rows of afm cu plus one nonmagnetic cu on a node position , is often visualized , since it would give the satellite peaks in neutron scattering @xcite .
the result gives an estimation of the position of the lower gap ; about 20 mry below a reference given by @xmath15 , see fig .
6 . with dopings 0.1 and 0.15 holes per cu ,
the main gaps should be at 9 or 13 mry below the reference .
thus , the separation of the two gaps is about 7 mry ( 100 mev ) for 0.15 holes / cu and about 11 mry ( 150 mev ) for 0.1 holes / cu , respectively .
these cases correspond to wavelengths of @xmath24 and @xmath25 perpendicular to the one at 8@xmath16 in the tunneling spectra of ref .
@xcite for optimally doped and underdoped bscco , the energy separations are 100 and 140 mev , respectively .
such values fit reasonably well to the assumed perpendicular wave lengths at the correct values of doping ( 0.15 and 0.1 holes per cu ) . the overdoped sample in ref .
@xcite , showing a separation of about 60 mev , can be interpreted as a result of a doping of about 0.19 holes / cu and a perpendicular wave of length @xmath26 .
a similar analysis based on the calculated dos of la@xmath4cuo@xmath3 gives smaller energy separations since the dos is larger ( 16 - 25 states per cu ) , and @xmath11 is shifted relative to the van - hove singularity .
thus the dos peak is at 0.15 instead of at 0.25 holes per cu , as shown in fig . 6 for hgba@xmath4cuo@xmath3 . from these results
it is now possible to propose another conclusion about the spc and the length of the phonon and spin wave : since the intermediate length ( @xmath27 containing 2 phonon waves and one spin wave ) show the largest spc ( cf .
i ) , it can occur along some direction ( @xmath20 ) for all dopings .
this would fit into the observations of saturated side peaks in neutron scattering for doping larger than about 0.15 .
the additional spacing around cu within the magnetic stripe ( the middle row of three magnetic ones ) will bring those cu sites close to the critical point for a magnetic instability , and these sites may participate in modulated spin waves along @xmath21 .
the latter waves can still be assisted by phonon distortions .
if the gap from the longest wave produce the deepest gap , it will be adjusted to the position of @xmath11 .
the two gaps will merge when the doping approaches 0.25 and the two directions have similar waves .
band calculations are too tedious for very large cells containing perpendicular spin - phonon waves of realistic length .
the discussion above is based on extrapolations . from the mechanism of perpendicular spc
, it is expected that the waves come in units of @xmath28 , as in the idealized band calculations .
imperfections , twinning and dynamic effects etc . , lead to incommensurate wave lengths in real materials .
it is also clear that the calculated gaps for frozen configurations are deeper than averaged gaps from dynamic waves .
calculations of spc along [ 1,1,0 ] showed that the gaps from the two waves appear at different energies .
calculations have not been done for the nickelates , but neutron scattering has found that charge- and spin - stripes are oriented along [ 1,1,0 ] @xcite . the results for waves along this directions can therefore be used to interpret the result for the nickelates .
two gaps in the dos are expected , but the difference in energy is fixed by the factor of two of the wave lengths of parallel spin and phonon waves .
this is in contrast to the variable energy difference between the two gaps induced by perpendicular coupling .
this prediction from spc suggest studies also of nickelates through apres and stm measurements .
the correlation between incommensurability @xmath29 ( the inverse of the wave length ) and doping is found to be approximately linear for hole doping between 0.0 and 0.5 in the nickelates @xcite .
this is as the results for the spc calculations along [ 1,1,0 ] of the cuprates @xcite .
since the gaps of spin waves and phonons appear at different energies there is no constructive spc , and this might explain why the nickelates are not superconductors .
the model of spc predicts different behavior of double gap structures for waves oriented along [ 1,1,0 ] . at least if perpendicular spc is excluded .
the band calculations for hbco showed that two waves ( spin and phonons ) along [ 1,1,0 ] will open gaps at different energies @xcite .
the relation between hole doping @xmath5 and wavelength @xmath30 ( still in units of @xmath16 ) is @xmath31 .
this means that a long ( @xmath32 ) supercell with @xmath33 unit cells oriented along [ 1,1,0 ] , will have one gap ( because of the spin wave ) at an energy where the total cell contains @xmath34 holes , and a second gap ( because of the shorter phonon wave ) at a lower energy where the doping should have been @xmath35 .
the factor of two between the wave lengths of the two waves makes a difference of 2 electrons per supercell . in other words ,
the dos per supercell can harbor two electrons between the two gaps .
this leads to relatively close gaps for small doping levels and wider energy separation as the doping increases , which is in contrast to the observed trend on the cuprate @xcite .
neutron data find modulations along [ 1,1,0 ] also for the cuprate , but only for the lowest doping . for @xmath36 0.05
the modulations occur along [ 1,0,0 ] , and there is a linear relation between @xmath5 and @xmath29 for @xmath5 approximately between 0.05 and 0.15 @xcite . for @xmath5 larger than 0.15
there seems to be a saturation , with a nearly constant @xmath29 .
this behavior is compatible with perpendicular spc when no waves shorter than 4@xmath16 ( i.e. @xmath27 for the spin wave ) are possible .
calculations of total energies have not been attempted for comparisons of waves along the two directions . but a simple qualitative reason for the crossover from [ 1,0,0 ] to [ 1,1,0 ] at some low doping can be found from the rigid - band variation of the fermi surface of hbco @xcite .
the fs for the undoped case is a m - centered cylinder with slightly larger band dispersions at the fs - crossings between @xmath37 and @xmath38 than between @xmath39 and @xmath38 , see fig 2 of ref .
the radius of the cylinder increases with doping and the fs will reach the @xmath39-point , where a van hove singularity is formed for @xmath40 per cu .
this means that there are increasingly more states near the @xmath39-point for increasing doping .
a wave oriented along [ 1,0,0 ] will open a gap mostly for states near this point ( see also the nfe - model ) , which makes a large gain in kinetic energy . for less doping ,
when the fs is more like a perfect cylinder , there is a uniform distribution of states around the fs .
hence , there is no argument for a larger gain of kinetic energy by opening a gap along [ 1,0,0 ] than along [ 1,1,0 ] .
the switching between waves along the cuo bond direction or along the diagonal direction presents possibilities for tests of the spc model . a secondary gap , which is associated with @xmath19 for spc - waves in the bond direction , will change below @xmath41 when the wave turns to the [ 1,1,0 ] direction .
the fs - smearing near the @xmath39-point at low t will also change at the critical doping .
a description of the bands of nickelates from a rigid - band model based on cuprate bands puts @xmath11 on the edge of the high dos region .
full band calculations of the nickelates might be necessary to sort out if the arguments for the cuprates can be carried over to the nickelates , or if the change of fs will bring out very a different behavior of the physical properties .
the afm susceptibility are found to be stronger when a small correction to the band structure is introduced , which affects the potential and the localization .
results for the doped and undoped hbco system have been published elsewhere @xcite .
these calculations are still using lsda , but the lmto linearization energies are chosen to be near the bottom of the main bands . by doing so it is found that undoped hbco has an afm ground state and a gap at @xmath11 , and the undoped case show an enhancement of spc .
this result shows that a relatively small correction to lsda can lead to a very different ground state @xcite .
the same type of correction has a similar effect on the band results for undoped and doped lsco .
figures 7 - 8 show the dos and bands for a cell of two f.u . of paramagnetic and afm ( without magnetic field )
lsco , in which anomalous linearization energies are chosen in order to promote the afm state , as was explained above for hbco .
the highest occupied band along the @xmath42 line is susceptible to have band gaps , if potential perturbations @xmath13 are present , as in the nearly - free - electron model . when doping is considered in a modulated cell as in fig .
1 , it gives larger coupling parameters @xmath43 for spin fluctuations .
the same trend was found for hbco , where @xmath43 was increased from about 0.2 in normal lda to 0.5 in the adjusted case @xcite .
a phonon which is accompanied by a spin wave will open a partial gap at @xmath11 .
the idea is to estimate the phonon softening through the gain in kinetic energy caused by the gap .
the displacement amplitude @xmath44 of atomic vibrations is in the harmonic approximation given by @xcite @xmath45 at high t , and @xmath46 at low t. the force constant @xmath47 , where @xmath38 is the atomic mass and @xmath48 is the vibration frequency , can also be calculated from the second derivative of the total energy @xmath49 , @xmath50 . from the latter relation
it is expected that @xmath51 should be independent of the mass .
the crossover from the low - t value of @xmath52 , which is due to zero - point motion , to the high - t value is approximately given by @xmath53 .
the value of @xmath51 is typically 25 ev / @xmath54 for an o phonon with @xmath55 80 mev , which makes @xmath44 about 0.06 at room temperature .
the sum of elastic energy @xmath56 and the kinetic energy @xmath57u@xmath54 is constant in time for harmonic vibrations .
spin waves can , in analogy with phonons , be assigned a magnetic moment @xmath58 , where @xmath59 .
the band results showed that @xmath60 is larger when @xmath44 is large , and @xmath61 0.3 @xmath62/cu at room temperature for the coexisting phonon and spin waves .
a partial gap at @xmath11 removes roughly half of the dos within an energy @xmath63 100 mev around @xmath11 , which leads to a gain in kinetic energy ; @xmath64 mev / cu , i.e. about 15 - 20 percent of the elastic energy @xmath65 for a typical @xmath44 .
thus , the phonon energy is expected to decrease by the this percentage because of the coupling to the spin wave , but only in the doped case when @xmath11 is at the gap .
these results are only indicative , but they compare reasonably well with experiment @xcite . the bond - stretching mode at the zone boundary is the perfect mode for this effect .
but the bond - stretching movements are involved also for modes with shorter @xmath66-vectors , so the softening can occur for less doping . since the result depend on a partial gap at @xmath11 , it is expected that the softening should be weaker when @xmath67 .
isotope effects ( ie ) on various properties have been measured on several high-@xmath9 oxides by use of different techniques and different types for the crystal growth @xcite . as will be discussed
, many problems of interpretation will be encountered when analyzing the isotope effect . the isotope effect on @xmath9
is briefly discussed in the section about superconductivity .
the usual ie is caused by the change in vibration frequencies when the isotopes are exchanged . with spc
there is an additional effect coming from the change in vibration amplitude .
the zero - point motion makes a mass ( @xmath38 ) dependence @xmath68 , and the frequency depends as @xmath69 for a constant @xmath51 .
however , contrary to the case of simple harmonic vibrations , a small @xmath38 will decrease @xmath51 because spin waves are promoted when @xmath44 is larger .
this makes isotope effects larger than what is suggested from a constant @xmath51 .
the effect on the pseudogap will be moderated at large t , when there is no explicit mass dependence of @xmath44 , @xmath70 .
anharmonic effects are expected to mix phonon and spin contributions .
several measurements of the ie on the pseudogap has been published @xcite .
the methods determine the transition temperature through the changes in properties obtained from nqr relaxation @xcite , increase of linewidth from crystal field transitions observed in neutron scattering @xcite , xanes experiments @xcite , or muon spin rotation @xcite .
these methods provide indirect measurements of t@xmath7 , and the results depend on the rapidity of the probe . the conclusion from several measurements of the zurich group is that a heavier oxygen isotope leads to a substantial increase of t@xmath7 , more than the effect on @xmath9 @xcite .
direct observations of the dos peaks below the pseudogap has been observed recently through angle - resolved photoemission , arpes @xcite .
the mechanism of spc is compatible with
the arpes results for the ie in optimally doped bi2212 where samples of equal doping with o@xmath71 and o@xmath72 are used @xcite .
the incoherent dos peak below the pseudogap , is shifted upwards , towards @xmath11 , for the sample with the heavy isotope .
thus , a pseudogap is narrower for large o - mass , as is expected from the diminished vibration amplitude for heavy atoms and weaker spin fluctuations .
calculations for hbco show that the moment per cu - atom increases by 30 percent when the displacement amplitudes are changed by a factor 1.5 . a change of isotopes from o@xmath72 to o@xmath71 is expected to increase the phonon amplitude by about 3 percent , which roughly can be translated into an increased gap by about 2 percent .
this is considerably smaller than found experimentally , however the calculated effect will by larger if the softening of the force constant @xmath51 ( from the spinwave ) is taken into account . at larger t , of the order 100 k , gweon et .
al . finds that the isotope shift is smaller than at low t ( 25 k ) @xcite .
this is expected from the behavior of the ie on vibration amplitude .
when zero point motion is dominant at low t there is an ie on @xmath44 , while at large t , when t is approaching the debye temperature , there is not .
since the o - modes are of high energy it expected that some ie should remain at rather large t. these results are obtained when no special phonon softening ( on @xmath51 ) is present .
the immediate conclusion from the same reasoning is that t@xmath7 and the gap should decrease when heavier isotopes are used .
measurements of the cu nqr relaxation in y124 containing o@xmath71 or o@xmath72 , find only a small ie on @xmath9 and on the spin gap ( defining t@xmath7 ) @xcite , in agreement with spc .
the relative downward shifts for both temperatures are about one percent or of the order of 1 k for the heavier isotope .
however , it has been argued that the time scale of the nqr experiment is not sufficient to probe phonon vibrations @xcite .
the signatures of t@xmath7 from various fast probe experiments indicate that an increased mass lead to a large _ increase _ of t@xmath7 @xcite .
this is in contrast to what can be expected from a narrowing of the gap in the arpes results @xcite and from the results of the present spc calculations .
various hypothesis can be put forward in order to understand this discrepancy between the measured t@xmath7 and the spc mechanism .
first , a light mass will be more itinerant than a heavy one .
this can be understood from the difference in diffusion constant @xmath73 , which behaves as @xcite @xmath74 , where @xmath75 is an energy barrier for hopping to another site .
a higher hopping frequency ( @xmath48 ) together with a larger @xmath44 ( from which a barrier is easier to overcome ) for a small mass , leads to larger diffusion compared to the case with a heavier mass .
a light isotope can more easily reach interstitial positions than a heavy one , and thereby leave normal lattice positions vacant .
this can modify the effective hole doping and therefore mask the true isotope effect .
the effective doping is sensitive to many things .
for example , it is often difficult to separate the true pressure dependence on @xmath9 from indirect effects caused by the pressure such as oxygen ordering and the effective carrier concentration within the cuo - planes @xcite .
if heavier isotopes imply improved material properties , less disorder and distortion , it can make the pseudogaps sharper so that t@xmath7 appears to be larger .
in fact , experimental data often show sharper transitions with o@xmath72 than with o@xmath71 , as for the intrinsic lineshape of the crystal - field transition @xcite , and in xanes data @xcite .
an indication of a better shaped gap can be found in the arpes data for o@xmath72 of gweon et .
@xcite , which show a shoulder in the gap coming from the so - called superstructure replica in bi2212 .
the shoulder is less pronounced for o@xmath71 although the position of the incoherent peak indicates a wider gap .
other explanations of the observed increase of @xmath19 with heavy isotopes might be proposed because of the particular techniques of the different experiments . but
admittedly , the present calculations of spc provide no obvious explanation of these measurements @xcite .
the mass dependence of @xmath44 is largest at low @xmath76 , which suggests that measurements of the penetration depth @xcite @xmath23 at low @xmath76 should be good for detecting ie .
some assumptions lead to @xmath77 , where @xmath78 is the dos at @xmath11 and @xmath79 is the fermi velocity
. the 3d average of this product will be small if a partial gap appears at @xmath11 , and a light mass with large @xmath44 makes the product even smaller ( @xmath80 for a complete gap ) .
this is opposite to experiments of the zurich group @xcite .
however , the spc result can be different if only the planar directions of @xmath79 are considered , as would be the normal ansatz for a magnetic field perpendicular to the cuo planes .
the band crossing @xmath11 along @xmath37-m has larger @xmath79 than at the crossing along x - m .
phonon and spin waves along [ 1,0,0 ] will produce the gaps in the latter region and @xmath79 will increase more for the material with light isotopes .
one can estimate that an increase of @xmath44 by 3 percent will decrease @xmath78 by 1 - 2 percent .
as @xmath81 this suggests a slight increase of @xmath82 .
it is often observed , in stm for example @xcite , that the pseudogap remains rather intact as t increases .
the peaks around the gap are at fixed positions , and they are not so much getting closer as t is approaching t@xmath7 , but the peaks are smeared .
the behavior is different from that of a superconducting gap , which becomes narrower when t @xmath83 .
this suggests that the mechanism behind the pseudogap is disconnected from superconducting pairing .
it can also be noted that that t@xmath7 and @xmath9 show no scaling as function of doping , but that t@xmath7 is much larger than t@xmath0 at low doping . a sudden disappearance of the gap at a high t , can be understood if spin fluctuations , rather than phonons , are the main cause of the gap .
it can be recalled that the states below and above the gap in the nfe model are separated in real space .
the lowest state for `` spin up '' bands in an afm case is localized on sites where the spin polarized potential is attractive for spin up electrons .
the other sites have the spin down states occupied . for t=0 and @xmath11 in the gap ,
the low energy states are occupied , while states above @xmath11 ( which have opposite spin distribution ) are unoccupied .
the separation between low and high energy states leads to a real space separation of spin up and spin down electrons .
but the fermi - dirac ( fd ) distribution at high t will not separate the occupied spins below @xmath11 from the opposite spin above @xmath11 as much as at low t. the local magnetic moments ( @xmath60 ) will decrease , which will decrease the spin splitting , which will decrease @xmath60 further and so on .
this process is modeled by use of low - t parameters of the gap @xmath84 and @xmath60 , which are compatible with the band results at low t. at t=0 , when the fd occupation , @xmath85 , is a step function there is an optimal separation between the occupied majority spins at @xmath86 from the unoccupied minority spins at @xmath87 , and @xmath88 , where the dos , @xmath78 , is assumed to be constant . as function of @xmath76 , @xmath89 .
the result for @xmath84 is shown in fig .
the reason of the rapid drop of @xmath84 is the strong @xmath76-dependenence of the fd - mediated feedback of @xmath60 on @xmath84 .
this is also exemplified for two band calculations for 8 f.u .
( see fig 1 ) of lsco doped with 0.5 holes per f.u .. all parameters , except for @xmath90 in the fd - distribution , are identical .
a calculation with @xmath91t=1 mry ( t=158k ) gives an afm arrangement of local moment of 0.21 @xmath62 on cu , and a clear gap .
when @xmath91 t is increased to 4 mry , the moment decreases to less than 0.1 @xmath62 after selfconsistency , and the gap is much washed out . the example is made for a very short wave , which implies large doping and low @xmath19 .
longer unit cells corresponding to underdoped cases are required for quantitative studies of real t@xmath7 by this mechanism .
effects from thermal disorder are expected to enhance the t - dependence .
these results show the fragile nature of the magnetic state . if the gap was due to only a potential modulation from phonons , there should not be a rapid disappearance at @xmath19 , because there is no mechanism of feedback .
recent measurements of the optical conductivity on la@xmath92ba@xmath2cuo@xmath3 for @xmath93 , where ordering of `` charge - stripes '' occurs below 60k @xcite , can be interpreted as a result of a gap in the dos below this temperature .
ab - initio band calculations for large unit cells are complicated .
the size of the brillouin zone is reduced , and an unfolding of the fs to the original zone is difficult .
some insight is provided by the nfe model .
a fourier component @xmath94 of the potential ( @xmath95 ) , such as the afm spin polarized potential of undoped cuprates , will open a gap ( 2@xmath94 ) at @xmath96 @xcite .
such gaps are visible in the full band approach , see fig .
7 . an additional modulation with longer wave length ( wave vector @xmath97 ) , will modify the potential , @xmath98 , and the gap moves slightly away from @xmath99 to @xmath100 .
the gaps in the full band approach are seen at a lower energy , see fig .
2 . by introducing a perpendicular modulation along @xmath21 , it is possible obtain the bands from a 3 by 3 matrix with @xmath101 , @xmath102 , and @xmath103 in the diagonal , and @xmath94 as non - diagonal terms .
only g - vectors 0 , @xmath104 and @xmath105 are considered in the basis with different @xmath106 and @xmath107 along @xmath20 and @xmath21 .
apart from having the 2-dimensional band structure centered around @xmath37 , the lowest bands show what is expected from the ab - initio band results in terms of band dispersion , double peaks in the dos ( if @xmath108 ) , and fs broadening near the zone boundaries , see figures 10 - 12 .
the @xmath109 and @xmath110 dispersions in fig .
10 show the partial gaps at different energies when the modulation vectors are different along the two directions , while no gap appears along the diagonal direction @xmath111 $ ] .
the fs is robust on the diagonal direction , so that an arc of fs remains even with strong potential modulations , see fig .
11 . near the @xmath112 and @xmath113 lines
however , the fs will be different depending on the strength of the modulation , @xmath94 .
a dynamic modulation will smear the fs in these areas , while the fs arc remains .
this agrees with interpretations of t - dependent arpes , where only a small section of the arc is seen at the low t @xcite .
the model bands are shown for perpendicular coupling , as is motivated from the discussion of the ab - initio results .
only one gap exist if the amplitudes of @xmath106 and @xmath107 are equal .
the deepest gap in the dos ( favorable position of @xmath11 ) is for energies where the two gaps overlap as shown in fig .
12 . , while the smaller dip appears at lower energy .
twinning and dynamic effects in real materials will mix the two directions , and the gaps will be weaker due to smearing .
the importance of spc for the mechanism of superconductivity is evident , since both phonons and spin - fluctuations are mutually enhanced .
a precise theory for this mechanism is still missing , but some observations can be made .
magnetism is traditionally considered to prevent superconductivity based on electron - phonon coupling through the pair - breaking effect between electrons of opposite spins .
similarly , superconductivity mediated by spin fluctuations between electrons of equal spin is prevented by impurities or even phonons .
the effect for spc presented above , is that selected phonons create a condition for enhancement of the spin wave .
therefore , it is no longer an antagonism between the spin wave and the phonon in the case of equal spin pairing ( esp ) .
the conditions for esp is enhanced near the critical point for diverging magnetic susceptibility . but how this should be implemented into explicit calculations of @xmath9 is not yet clear . in a simple bcs - like formula , @xmath114
, @xmath115 is the coupling constant either for electron - phonon interaction ( @xmath116 ) or from spin fluctuations ( @xmath43 ) , and the prefactor @xmath48 is the energy of the phonon or the energy of the spin excitation , @xmath117 or @xmath118 , respectively . generally , if @xmath115 is large , then @xmath48 is small , and vice - versa .
in other words , a phonon softening is about equivalent to a nearly static magnetic order , and the optimal condition for a high @xmath9 is expected near such a transition . from the discussion of spc it is more probable that superconductivity is because of esp near a critical point than because of standard electron - phonon coupling .
with spc it is possible to be very near a critical point during the cycle of a phonon vibration .
the system can profit from a large @xmath43 at the extreme phonon displacement and a reasonably large @xmath118 when the displacements are not much developed .
the spc calculations showed that the conditions for having a pronounced spin wave are enhanced when the phonon is present , from which it can be assumed that @xmath118 is at least not smaller than @xmath117 .
it is possible that @xmath118 is larger than @xmath117 , but as a low limit @xmath119 80 mev for o - modes .
a value of @xmath43 of the order 0.5 was calculated from the band results for a short wave in hbco ; @xmath120 , where @xmath121 and @xmath59 as in ref .
this gives a large value ( @xmath122 100 k ) for @xmath9 from a bcs - like formula .
however , this is very indicative : the calculation of @xmath43 is for the shortest possible wave , and the lda potential needs corrections .
effects of the interplay between other phonon and spin waves , of coulomb repulsion , of anharmonicity and of strong coupling are neglected . from spc
it is quite clear that superconductivity disappears at large doping , since the limit for the wave length of a phonon is at @xmath123 , corresponding to @xmath40 holes per cu .
spin waves shorter than @xmath124 has to work without ( parallel ) phonons . in the opposite limit , at small doping
, one can see at least two reasons for the disappearance of superconductivity : first , if the mechanism of spc and the pseudogap becomes too successful at low @xmath5 , as can be deduced from the experimental trend for @xmath19 and from some of the calculated results , it will diminish @xmath78 and hence also @xmath43 and @xmath9 .
too strong effects from spc implies also that either one of the waves ( spin or phonon ) may get too soft and become static , and thereby become inefficient for superconductivity .
another reason is the change of the wave orientation from [ 1,0,0 ] to [ 1,1,0 ] which is observed for @xmath125 0.05 , which in the spc model implies the end of constructive coupling between phonons and spin waves . the isotope effect on the superconducting @xmath126 is more complex than for normal electron - phonon coupling , since @xmath43 has an isotope effect .
a pure @xmath43 tend to increase for light isotopes because of larger displacement amplitudes .
a pure phonon frequency @xmath117 behaves as @xmath127 , and above it was assumed that @xmath128 .
a lower @xmath38 should increase @xmath48 , but a softening of @xmath51 because of spc , will have a moderating effect .
these two mass dependencies work together for a positive isotope effect as in normal electron - phonon coupling , but the total effect might be small if the phonon softening is large .
in conclusion , results from ab - initio band calculations for different configurations of spin waves and phonon distortions in hbco and lsco , have been used in models aimed for an understanding of various properties . at this stage , when it is not yet clear exactly which configuration is the most important one , it is preferable to make qualitative studies of a variety of properties rather than precise studies of a single effect .
thus , it has been shown that spc leads to pseudogaps , fs broadening in the bond directions with a stable arc in between , phonon softening , and variations of the periodicity ( wave length ) with doping .
a rapid t - dependence of the gap is consistent with rapid quenching of the spin part , while this would not be explained from phonons only .
isotope shifts are expected on many properties . some experiments showing a strong increase of @xmath19 for heavy isotopes
are not explained , but the results are consistent with some other data showing smaller effects .
the nature of spc is such that different waves in perpendicular directions are probable .
this will produce double gap structures in the dos , as seen in the simple model of nearly free electrons . from the doping dependence of
observed double peaks it is speculated that the reason this phenomenon is related to perpendicular spc
. quantitative , ab - initio results are difficult to obtain for very large unit cells , because of few k - points and slow convergence of total energies .
the lda ground state is too far from the afm instability and many effects are probably underestimated in lda .
it is argued that lda need rather modest corrections in order to promote an afm , insulating state .
the gap is in this case an ordinary band gap , caused by modulations of afm within the cuo plane . the results and the mechanism of spc are sometimes very different from the conventional pictures of the physics of high - t@xmath0 superconductors .
the hope is that the mechanism of spc should give new ideas for a guidance towards the mechanism for high-@xmath9 superconductivity .
d. rubio temprano , j.mesot , s. janssen , k. conder , a. furrer , a. sokolov , v. trounov , s.m .
kazakov , j. karpinski , and k.a .
muller , eur .
j. b*19 * , 5 ( 2001 ) ; d. rubio temprano , j.mesot , s. janssen , k. conder , a. furrer , h. mutka , and k.a .
muller , phys .
lett . * 84 * , 1990 , ( 2000 ) .
.calculated maximum deviations of charges @xmath17 ( el . per cu ) , and moments @xmath60 ( @xmath62 per cu ) for lsco cells of different lengths @xmath129 along [ 1,0,0 ] .
the @xmath124 wave , is as in fig .
1 , the @xmath27 wave is like the one shown horizontally in fig . 3 , and the @xmath130 is twice as long .
the cells contain phonons or spin waves with equal distortions or magnetic field per site , so that each of the distorted o - atoms is displaced 0.02@xmath16 or each cu with imposed moment has a field of @xmath13110 mry . [
cols="^,^,^,^ " , ] | the mechanism of spin - phonon coupling ( spc ) in high - t@xmath0 copper oxides is explored from band calculations on la@xmath1sr@xmath2cuo@xmath3 and hgba@xmath4cuo@xmath3 systems .
the lmto band calculations , based on the local density approximation , are made for cells containing frozen phonon displacements and/or spin waves within the cuo plane .
the virtual crystal approximation is used for studies of hole doped systems .
the main result is that phonons are favorable for spin waves and vice - versa , and that pseudogaps appear naturally in the band structures of striped materials with strong spc .
the qualitative results are compatible with many observations showing that the properties of high - t@xmath0 superconductors depend both on lattice interactions and magnetic fluctuations .
the band results are used to model various properties , mainly of the normal state , such as isotope effects , pseudogaps , fermi surface broadening , t - dependence of the pseudogap , phonon softening and some aspects of superconductivity .
the possibility of perpendicular spc is investigated , partly by the use of a nearly free electron model . |
the next generation of dark matter detectors using liquid xenon ( lxe ) as the detection medium @xcite will probe lower interaction cross sections than ever before .
one key to success for these experiments is an improved understanding of low - energy interactions in lxe .
the interaction cross sections between weakly interacting massive particles ( wimps ) and standard model particles are generally expected to increase exponentially with decreasing interaction energy .
however , low - energy interactions in lxe , below a few kev , are not well - understood theoretically , and measurements of the fundamental properties of lxe at low energies that allow the interaction energy scale to be precisely defined are lacking . for instance
, systematic uncertainties on the scintillation - based energy scale have limited the search for axions @xcite and for annual modulation @xcite with xenon100 data .
these systematic uncertainties could be improved by using a combined energy scale ( ces ) , that is by using both the prompt scintillation ( @xmath1 ) and ionization ( @xmath2 ) signals , provided that the anti - correlation of @xmath1 and @xmath2 in lxe at low energies is well characterized .
charge yield measurements of electronic recoils ( ers ) are currently most needed .
lack of such measurements is the primary reason a ces was not adopted for studies of ers in xenon100 .
next generation detectors using a ces will benefit from these measurements .
measurements of low - energy ers from monochromatic @xmath3 rays are of particular interest for dark matter searches using lxe , as their dominant electromagnetic backgrounds in the outer layer of the detector come from such compton scatters .
direct measurements have been made of the light yield of lxe due to ers from monochromatic @xmath3 rays as a function of the recoil energy down to low energies @xcite .
these yields were derived from measurements of @xmath1 using single - phase lxe detectors operated at either zero drift field or at a single applied electric field .
the lowest mean energy previously measured for the light yield due to ers is 1.5kev @xcite .
even fewer measurements have been made of the charge yield of lxe due to ers , derived from the @xmath2 signal , since collecting the ionization electrons typically requires two - phase operation , entailing more sophisticated detectors . at the current time , no references have been found for dedicated measurements of the charge yield of ers at low energies using compton scattering of external monochromatic @xmath3-ray sources , the method employed for this work .
the charge yield of ers has been measured by several groups as a function of electric field using mono - chromatic @xmath3 rays , but only at relatively high energies , e.g. 122 , 511 , and 570kev .
see ref.@xcite for a review of er light and charge measurements .
the light and charge yield from the decay of @xmath4kr at 9.4kev and 32.1kev has been measured @xcite , though the decay products consist primarily of several internal conversion and auger electrons , and thus are not directly comparable to measurements with monochromatic @xmath3 rays . additionally , the charge and light yield from the decay of tritium ( @xmath5h ) has been measured at two drift fields @xcite . however , as tritium is a single - beta emitter with a q - value of 18.6kev , the resulting energy spectrum is continuous , and thus monte carlo ( mc ) matching is required for the recoil energy determination .
similarly , continuous spectra from the decay of @xmath6cs have been measured and an accompanying mc used to determine the charge and light yield in lxe at various drift fields @xcite . in this paper we present results of measurements of the light and charge yield of lxe due to low - energy ers as a function of applied electric field and deposited energy .
the compton coincidence technique is used to determine the amount of energy deposited in an er in the lxe .
the technique , introduced in ref.@xcite and further improved in ref.@xcite , is the same technique used for the measurement of the light yield of ers described in detail in ref.@xcite . for this work
we employ the same method , but now use a two - phase time projection chamber ( tpc ) , thus allowing both the charge and light yield to be measured simultaneously .
science data presented here were taken during two periods , referred to as runs i and ii .
the principle of the compton coincidence technique ( cct ) is to irradiate the primary detector with high - energy monochromatic @xmath3 rays , and to look for events in a second detector in prompt coincidence corresponding to the full - energy deposition of the compton - scattered @xmath3 rays .
since the initial energy is known and the final energy is measured by the second detector , the energy lost in the primary detector can be determined . by placing the second detector at various angles relative to the incoming @xmath3 rays ,
different recoil energy ranges can be probed in the primary detector .
we measure recoils from incoming 661.7kev @xmath3 rays emitted from a 123@xmath7ci @xmath6cs source , and measure the scattered @xmath3 rays with a high - purity germanium ( hpge ) detector , as shown schematically in fig.[method_schematic ] .
schematic of the experimental setup during compton coincidence measurements .
monochromatic @xmath3 rays from the decay of @xmath6cs compton scatter in the lxe detector , depositing energy which results in electronic recoils .
the energy of the @xmath3 rays scattering at an angle @xmath8 is measured by a hpge detector . ]
the primary detector used for this measurement is a two - phase lxe tpc designed to measure both the light and charge yield of nuclear and electronic recoils in lxe , and is thus dubbed nerix .
the nerix detector is the latest iteration of such detectors designed and built at columbia university for measuring the fundamental properties of lxe . as shown in fig.[detector_schematic ] , the nerix detector consists of five photomultiplier tubes ( pmts ) surrounding a sensitive volume enclosed by a series of grids supported in a polytetrafluoroethylene ( ptfe ) frame .
four 1 `` square pmts ( hamamatsu r8520 - 406-m4-sel ) , each with four channels , form a square top array of 16 channels in the gaseous xe above the sensitive volume .
one 2 '' diameter circular pmt ( hamamatsu r6041 - 406 sel ) views the lxe volume from below .
all pmts were selected for having high quantum efficiency ( qe ) , @xmath935% at the xe scintillation wavelength , 178 nm . during coincidence measurements ,
the top pmt array is used for event vertex reconstruction and the bottom pmt is used for the yield determination .
schematic drawing of the nerix detector . ]
the ptfe support pieces are stackable and compressed vertically between two stainless steel plates via polyether ether ketone ( peek ) support rods with stainless steel springs .
the springs ensure that the stack remains compressed and relatively light tight when at lxe temperature .
the nearly seamless ptfe inner surface surrounding the sensitive volume ensures optimal light collection efficiency .
the bottom and top pmts are mounted to separate stainless steel plates that can be translated vertically with respect to the tpc stack .
to produce the electric fields in the liquid and gaseous xe , high voltage ( hv ) is applied to hexagonal mesh grids with 3 mm pitch , etched from stainless steel foil .
the anode , gate , and cathode grids are 125@xmath7 m thick , and the screening grid directly above the bottom pmt is 25@xmath7 m thick , ensuring high optical transparency and good electric field uniformity .
the gate and screening grids are electrically grounded , and negative hv is applied to the pmts and the cathode , while @xmath104.5kv is applied to the anode .
a 56.5 mm diameter ring , made from 1.65 mm diameter cu wire is fixed 7 mm above the cathode and coaxial with the tpc .
two 500m@xmath11 resistors connected to the ring divide the voltage between the cathode and gate grids , improving the linearity of the electron drift field in the sensitive volume . the sensitive volume , that is the volume from which both light and charge signals from the lxe can be measured , is a cylinder 23.4 mm tall with a 43 mm diameter while at typical lxe temperature , between the gate and cathode grids .
the volume contains a lxe target mass of approximately 97 g .
a custom vacuum cryostat vessel made from 1.5 mm thick stainless steel was designed to minimize the amount of lxe and other material outside of the nerix tpc and provide the lxe level - setting mechanism .
in addition , the amount of ptfe around the sensitive volume was minimized in order to reduce the likelihood that @xmath3 rays deposit energy outside of the sensitive volume .
a total of @xmath02.2 kg of lxe are used to fill the detector vessel .
the nerix cryogenic and purification system was adapted from that used for the measurements in ref.@xcite .
an emergency recovery vessel and rupture disk were added to capture the xe gas in the case of a loss of cooling power or insulation vacuum .
circulation of the xe gas through the purification loop was achieved using a diaphragm pump ( knf 143 sn.12e ) .
the secondary detector is a p - type coaxial hpge detector ( ortec gem40 - 76 ) with a crystal of 63.7 mm diameter and 59.6 mm length .
the typical full width at half maximum ( fwhm ) energy resolution at 1.33mev is 1.81kev , with a typical peak - to - compton ratio of 71:1 . during coincidence measurements ,
the @xmath6cs source was placed 17.6 cm horizontally from the tpc center , and the hpge detector was placed such that the center of the detector endcap was 14.6 cm horizontally from the tpc center .
the positions were measured to within 1 mm using aluminum rules and an auto - leveling laser . the relative angle between the @xmath6cs and hpge detector
was fixed at either 0@xmath12 or 25@xmath12 during science data taking .
different configurations of the source and hpge detector positions were used for the two runs in order to study the impact of the lxe buffer volume .
the bottom pmt used during run i experienced a deterioration of gain , and was thus exchanged in between the runs for another pmt of the same type and comparable qe .
the data acquisition system for nerix is similar to that used in ref.@xcite .
the raw data from nerix is amplified x10 ( phillips 776 ) and one copy is digitized by three 14-bit flash adcs ( caen v1724 ) at 100ms / s with 40mhz bandwidth , and stored to disk .
the other copy is fed into the trigger logic system .
the hpge detector provides two copies of the pre - amplified hpge signal .
one copy is inverted and passed through a leading edge discriminator thus producing the hpge trigger .
the other copy is amplified with a pulse - shaping amplifier using a coarse gain of x25 and a 6@xmath7s shaping time , digitized by the same caen system , and then recorded to disk . during measurements at 25@xmath12 , the raw signals from the pmts
are not amplified , so as to mitigate the saturation of the digitizer due to large @xmath2 signals from higher energy interactions .
signals from the pmt channels pass through a leading edge discriminator which subsequently produces the @xmath1 hardware trigger .
the discriminator is updating , producing a 10ns pulse whenever an input signal is above the threshold , resulting in variable - length trigger pulses .
logical operations ensure that an @xmath2 trigger is subsequently produced only for pulses longer than 100ns .
measurements of the @xmath2 trigger efficiency were made using back - to - back 511kev @xmath3 rays from the decay of @xmath13na in prompt coincidence with a nai detector .
the average trigger efficiency for @xmath2 with 10@xmath5pe is @xmath075% , and is thus near 100% for most recoil energies measured .
fits of the observed @xmath2 spectra are corrected for this efficiency loss .
the impact on the attributed charge yield is at the percent level for 1kev recoils , and diminishes with increasing recoil energy .
the hardware trigger for coincidence events is formed by requiring that an @xmath2 trigger come within @xmath14@xmath7s after an hpge trigger , depending on the drift field being measured .
the coincidence trigger and @xmath1 trigger are recorded to disk . a stricter time coincidence requirement between the signals from the two detectors
is imposed in post - processing , as described in section[data_sel ] .
calibration measurements of the pmts single pe gain were taken biweekly using a blue led .
typical gains were @xmath15electrons / pe . regular calibrations of nerix with @xmath3 rays from @xmath6cs ( 662kev ) were taken to monitor the tpc performance .
the typical measured light yields for @xmath6cs during runs i and ii are given in table[typical_ly ] .
.typical measured light yield of nerix for the full - absorption peak of @xmath6cs ( 661.7 kev ) at various drift fields , v@xmath16 , during runs i and ii .
uncertainties represent statistical and systematic uncertainties added in quadrature . [ cols="^,^,^",options="header " , ] the posterior distributions from the mcmc
are sampled in order to determine the yields in photons / kev and electrons / kev , as well as the reconstructed mean recoil energies for each run .
the weighted average of the yields and reconstructed energies from runs i and ii is computed using the statistical and systematic uncertainties added in quadrature as the weights .
the resulting average yields are shown in fig.[ly_ab ] .
the relative deviation of the average reconstructed recoil energies from the measured recoil energies is shown in fig.[en_recon ] .
relative deviation of the recoil energy as reconstructed using eq.[anticorr ] from the recoil energy as measured by the hpge detector , shown versus the measured recoil energy .
the expectation of no deviation is shown as a dashed line . ]
due to the finite energy resolution of the hpge detector , the range of actual recoil energies in each energy bin is broader than the nominal width .
thus , while different energy bins do not share events , they are expected to overlap somewhat in terms of the actual recoil energy of events in the bins .
the selected recoil energy range for each bin is shown as solid horizontal error bars in fig.[ly_ab ] .
the estimated range of actual recoil energies , that is the selected energy range plus the @xmath17 hpge energy resolution , is shown as open horizontal brackets in fig.[ly_ab ] and as solid horizontal error bars in fig.[en_recon ] .
since the yields are determined by dividing by the energy of each bin center , a shift in the average energy of events in each bin from this center energy would result in a systematic shift the attributed mean yield .
however , it is observed that the average energy of events for each bin is within 10@xmath18 of the bin center for the lowest - energy recoil of each data type ( 1 or 2kev ) , and within 1@xmath18 of the bin center for all other bins . in all cases , the resulting shift in the attributed mean
energy is negligible compared to the bin width and energy resolution of the hpge detector .
the systematic effect of the binning definition and fit range of the measured @xmath1 and @xmath2 distributions on the yield determination was studied and found to impact the @xmath1 fits in the lowest @xmath19 bins for data taken at 0@xmath12 only .
the average relative sizes of these effects were 16% and 9% , respectively , for the lowest energy bin , and 4% and 5% , respectively , for the following bin . by the third bin ,
these effects were negligible .
the corresponding systematic uncertainties in the light yield , which are recoil - energy dependent , are added in quadrature with the statistical uncertainties of the respective bins and are shown as solid vertical bars around each point in fig.[ly_ab ] and reported in tables[345v_table]-[5500v_table ] .
the variation of the actual yield and differential scattering rate over the width of each energy bin can also result in a systematic shift of the attributed mean yield .
given the observed variation of the measured yield between consecutive bins ( taking into account the uncertainties mentioned above ) and assuming a flat differential scattering rate , we estimate this possible shift to be at most @xmath09% for 1kev recoils , @xmath20% for recoils between 2kev and 6kev , and @xmath21% for recoils @xmath22kev .
the actual size of such a systematic shift is difficult to quantify since the actual yields are unknown .
this possible shift is not included in the uncertainties reported herein . the uncertainties on the pmt gain , @xmath23 , @xmath24 , and @xmath25 affect all points of a given data type uniformly , and are herein referred to as systematic uncertainties .
the correlations between these parameters are probed by the mcmc of the global fit of the reconstructed energy , which is used to propagate the uncertainties to the final yields and reconstructed recoil energy for each run and data type .
the resulting total systematic uncertainties are shown as vertical open brackets around each point in fig.[ly_ab ] and are reported in tables[345v_table]-[5500v_table ] . in fig.[en_recon ] , the vertical error bars show the statistical and systematic errors added in quadrature .
the final results of the light and charge yield determinations are shown in fig.[ly_ab ] and summarized in tables[345v_table]-[5500v_table ] .
we observe that the light yield increases with recoil energy , attaining a maximum between @xmath26kev depending on the field , and then decreases .
a similar , but opposite , trend is observed for the charge yield .
the general dependence of the yield on the drift field is significant and is as expected for a charge recombination process , showing a decreasing light yield and increasing charge yield with increasing drift field .
we also find that the drift field dependence of the yields diminishes with decreasing recoil energy and becomes negligible , given experimental uncertainties , below @xmath27kev .
the reconstructed energies , fig.[en_recon ] , show good agreement with the anti - correlation relationship as formulated in eq.[anticorr ] for recoil energies above 3kev . below 3kev
a significant deviation from expectation is observed , increasing from an average of @xmath28 at 2kev to @xmath29 at 1kev .
see fig.[comp ] for a comparison of our results taken at 0.19kv / cm with recent measurements and predictions at comparable drift fields .
comparison plot of lxe light yield ( top ) and charge yield ( bottom ) from ers with comparable electric field strengths .
the nerix results taken at [email protected] / cm ( blue circles ) are shown together with measurements from @xcite taken at 0.18kv / cm ( magenta band and dashed line ) , measurements from @xcite taken at 0.236kv / cm ( green band and dashed line ) , and calculations from @xcite at 0.19v / cm ( blue line ) . ]
distribution of scintillation signal ( top panel ) and ionization signal ( bottom panel ) for 1kev recoils at 0.48kv / cm drift field during run ii . ]
we find that our results at 0.19kv / cm are generally in good agreement with recent measurements at comparable fields using continuous recoil energy bands .
the measured light yield agrees very well with that in @xcite , while both are systematically lower than that in @xcite by @xmath31 on average .
the measured charge yield is systematically higher than that in @xcite by @xmath32 .
our results are in similarly good agreement with the nest model predictions @xcite at low energies @xmath33kev .
the agreement for the light and charge yields at 0.48 , 1.02 , and 2.32kv / cm is very good , though for data at 0.19kv / cm the measured field dependence is less pronounced than that predicted by nest . for 2kev recoils ,
however , we observe a slightly higher light yield than predicted by nest at all measured fields , and for 1kev recoils at 0.48kv / cm , the observed charge yield is significantly higher .
fig.[en_recon ] shows that our data supports the use of a combined energy scale as defined in eq.[anticorr ] for determining er energies in lxe dark matter detectors for recoil energies above @xmath34kev .
however , for recoil energies below @xmath34kev , a significant deviation from the expectation is observed . even assuming a light yield of zero at 1kev , the measured charge yield exceeds what should be possible given the accepted , recoil - energy independent average @xmath35 value .
assuming an ionization potential of 11.7ev for lxe , the maximum number of quanta that can be produced for 1kev recoils is @xmath36 , which is greater than the measured charge yield , but in agreement given experimental uncertainties .
a threshold effect on the @xmath2 signal could plausibly cause an apparent rise in the charge yield .
however , the measured @xmath2 trigger efficiency is incorporated in the analysis .
furthermore , the observed @xmath2 spectra for 1kev recoils , see fig.[s2_dist_lowe ] and [ s2_1kev ] , show no evidence of significant asymmetry . in addition , reducing the energy bin width to 0.5kev for the data taken at 0.48kv / cm drift field results in a nearly identical charge yield at 1kev .
thus , it is unlikely that the high charge yield at 1kev is due to a rate or binning effect . a threshold effect on the @xmath1 signal ,
e.g. a loss of the @xmath1 peak identification efficiency of the raw data processing software as well as the decrease in @xmath1 detection efficiency due to the expected statistical limitations mentioned in sec[yield ] , could also potentially cause such an apparent rise in the charge yield . however , dropping the @xmath1 requirement for the @xmath2 spectra ensures that such effects can have no impact .
thus , we conclude that the observed rise in the charge yield is likely not artificial , and may instead indicate that the combined energy scale as described in eq.[anticorr ] breaks down at very low energies , namely that the average @xmath35 value may be energy dependent .
we gratefully acknowledge support from the national science foundation for the xenon100 dark matter experiment at columbia university ( grant no .
phys09 - 04220 ) .
we also thank junji naganoma for his support during data taking , and patrick de perio , qing lin , and fei gao for their useful discussions regarding the analysis .
99 e. aprile _ et al_. ( xenon collaboration ) , jcap * 2016 * , 027 ( 2016 ) .
d. c. malling _
et al_. ( lz collaboration ) , arxiv:1110.0103 .
e. aprile _ et al_. ( xenon100 collaboration ) , phys . rev . *
d 90 * , 062009 ( 2014 ) .
e. aprile _ et al_. ( xenon collaboration ) , science , vol .
349 , 851 ( 2015 ) .
e. aprile _ et al_. ( xenon collaboration ) , phys .
* 115 * , 091302 ( 2015 ) .
l. baudis _ et al_. , phys . rev . *
d 87 * , 115015 ( 2013 ) .
e. aprile _ et al_. , phys . rev .
* d 86 * , 112004 ( 2012 ) .
m. szydagis _ et al_. , jinst * 6 * , p10002 ( 2011 )
. a. manalaysay _ et al_. , rev .
sci . instrum . * 81 * , 073303 ( 2010 ) .
d. s. akerib _ et al_. ( lux collaboration ) , phys . rev .
* d 93 * , 072009 ( 2016 ) .
et al_. , phys .
d 92 * , 032005 ( 2015 ) .
j. d. valentine and b. d. rooney , nucl .
methods * a 353 * , 37 ( 1994 ) .
b. d. rooney and j. d. valentine , nuclear science , ieee transactions on * 43 * , 1271 ( 1996 ) .
et al_. , nuclear science , ieee transactions on * 55 * , 1753 ( 2008 ) .
et al_. , nuclear science , ieee transactions on * 55 * , 1073 ( 2008 ) .
g. plante _
et al_. , phys . rev .
* c 84 * , 045805 ( 2011 ) .
e. aprile _ et al_. ( xenon100 collaboration ) , j. phys .
g. * 41 * , 035201 ( 2014 ) .
e. aprile _ et al_. ( xenon100 collaboration ) , astropart .
phys . * 35 * , 573 ( 2012 ) .
d. f .- mackey _
et al_. , pub .
soc . pac . * 125 * , 306 ( 2013 ) .
m. szydagis _ et al _ , jinst * 8 * , c10003 ( 2013 ) . | the dependence of the light and charge yield of liquid xenon on the applied electric field and recoil energy is important for dark matter detectors using liquid xenon time projections chambers .
few measurements have been made of this field dependence at recoil energies less than 10kev . in this paper
we present results of such measurements using a specialized detector .
recoil energies are determined via the compton coincidence technique at four drift fields relevant for liquid xenon dark matter detectors : 0.19 , 0.48 , 1.02 , and 2.32kv / cm .
mean recoil energies down to 1kev were measured with unprecedented precision .
we find that the charge and light yield are anti - correlated above @xmath03kev , and that the field dependence becomes negligible below @xmath06kev .
however , below 3kev we find a charge yield significantly higher than expectation and a reconstructed energy deviating from linearity . |
with the turning - on of the lhc in the near future , one of the most important tasks of both general - purpose detectors atlas and cms will be the search for physics beyond the standard model , with one of the most promising candidates being supersymmetry .
in the course of this program , which will obviously be adjusted and focused with increasing luminosity and with the first indications of `` new physics '' , the determination of the properties of newly detected particles will be a major task .
indeed , only after the quantum numbers , the interaction and the mass of a new particle have been measured with high precision , this requirement can be considered fulfilled . in the present paper
we want to concentrate on the gluino , the supersymmetric partner of the gluon , and discuss the possibility to determine its quantum numbers , i.e. its spin , color , majorana nature and its mass , through the investigation of non - relativistic boundstates of two gluinos , generally denoted `` gluinonia '' .
the classification of gluinonia according to their quantum numbers and their qualitative properties , i.e. spectra and decay modes , as well as estimates of their production cross section at hadron colliders have been studied already more than two decades ago and presented in refs .
@xcite concentrating at that time on gluino masses which are by now experimentally excluded .
more recently gluinonia have been discussed in refs .
hadronic transitions between different gluinonium levels and some rare annihilation decays were evaluated in @xcite .
the production of vector and pseudoscalar states at the tevatron and the lhc was investigated in @xcite where it was argued that , given sufficiently good yet realistic jet mass resolution , pseudoscalar states could be detected for bound state masses up to 3 tev . a phenomenological study of the question , to which extent a signal for color - octet states could indeed be detected experimentally , has been performed in @xcite . in all these cases , however , the production cross section has been calculated in lowest order approximation only and a qualitative phenomenological potential for the gluino - gluino interaction has been employed .
the topic has gained renewed interest in connection with split susy @xcite , ( see , e.g. , refs .
@xcite ) .
the detection of these bound states would provide important information about gluino properties , difficult to obtain through other means .
since gluinonia only exist if gluinos do not decay too rapidly , say with decay rates less than a few gev , their observation would immediately provide at least an upper limit on @xmath1 .
the energy spectrum of the gluinonium system , e.g. the mass difference between ground state and first radial excitation ( or open gluino production ) would be sensitive to the potential and thus to the color - octet nature of the constituents .
the same is true for the production cross section which is proportional to the square of the wave function at the origin .
gluinos are expected to decay through cascades into several jets plus the lightest supersymmetric particle ( lsp ) which escapes detection .
for this reason the precise determination of its mass in direct gluino decays is limited by the missing mass or energy resolution .
in contrast , once the mass of the boundstate and the dynamics of the interaction are known with sufficient precision , the mass of the constituents is fixed with the corresponding accuracy and in any case subject to a different systematic error . the energy levels of gluinonia were estimated already in the early papers employing phenomenological color - octet potentials , obtained from phenomenological quarkonium potentials by multiplication with the ratio of the corresponding casimir operators @xmath2 and the gluino mass was interpreted as an effective mass closely related to the pole mass . during recent years
more refined studies have been performed to describe non - relativistic boundstates of color - triplet constituents .
this includes the evaluation of the static two - loop potential @xcite , first steps toward the evaluation of the three - loop potential @xcite and the calculation of mass - suppressed terms proportional @xmath3 @xcite and @xmath4 @xcite .
mass definitions better suited to control the long distance part of the potential have been introduced @xcite , and the evaluation of the spectra through analytic methods has been possible , using a perturbative approach to include higher order contributions from the potential @xcite .
last but not least , finite - width effects were incorporated for top - quark production at electron - positron @xcite and hadron colliders @xcite .
these new results can be applied also in studies of gluino bound states . in order to arrive at reliable predictions for the signals of gluinonium production and decay ,
the inclusion of a realistic qcd potential and next - to - leading order ( nlo ) corrections to boundstate production at the lhc is mandatory . considering and building on the significant progress in the perturbative treatment of non - relativistic boundstates ,
in particular quarkonium and positronium , predictions of similar quality can be made for boundstates with color - octet constituents .
this is the main content of the present paper .
we start with a brief overview and a discussion of the quantum numbers and qualitative properties of boundstates with majorana constituents and evaluate the qcd potential for a color - octet boundstate ( section [ sec : properties ] ) .
subsequently , in section [ sec : spectroscopy ] , we calculate the spectrum and the wave function at the origin for the lowest - lying levels .
these quantities are the necessary ingredients for a realistic prediction of gluinonium production and decay rates presented in section [ sec : production ] . assuming that squarks are relatively heavy , the dominant decay channel of the pseudoscalar singlet boundstate proceeds through two gluons , and it will be a difficult task to distinguish this signal from the irreducible background from the two - gluon jet continuum .
in section [ sec : photons ] we consider as an alternative the decays into top quarks and two photons , which can proceed through virtual quark ( @xmath5 ) and squark ( @xmath6 ) intermediate states .
one might , in principle , hope that the extremely clean @xmath7 signal sticks out of the irreducible background similarly to the higgs decay into two photons which has been identified as a promising signal significantly superior to the @xmath8 or two - gluon mode .
similar considerations apply to the @xmath9 mode . in order that these modes can compete with the two - gluon channel
, squarks should not be dramatically heavier than gluinos .
on the other hand , to avoid dominance of the single - gluino decay , the decay mode @xmath10 should be kinematically forbidden , corresponding to @xmath11 .
we therefore evaluate the corresponding decay rates and study if , given suitable choices for the respective masses , the two - photon mode might lead to a possible signal . in section
[ sec : background ] a rough estimate of the dominant signal versus background is presented .
section [ sec : conclusions ] contains our conclusions .
let us start with a brief recapitulation of the quantum numbers of gluino boundstates and the corresponding color , spin and orbital momentum configurations @xcite . these differ from those of quarkonia due to the restrictions arising from the majorana nature of gluinos , and due to their different color assignment .
two color - octet states can be combined into irreducible representations as follows @xmath12 where the index indicates the ( anti-)symmetry with respect to their color index
. the interaction can be either attractive , repulsive or absent ( in lowest order ) .
in lowest order the coefficient of the qcd potential is given by the expectation value of the product of the color generators @xmath13 , taken between two - particle states in the respective representation .
this product , in turn , can be expressed by the eigenvalues of the quadratic casimir operator of the constituents , @xmath14 , and the boundstate in representation @xmath15 , @xmath16 : @xmath17=\frac{1}{2}\left(c_{{r}}-2c_{a}\right ) \ , .
\label{fdotf}\end{aligned}\ ] ] the results are listed in tab .
[ table1 ] . in the following
we shall limit the discussion to the cases with negative coefficients , corresponding to attraction .
.color interaction of two @xmath18 octets .
[ cols="^,^,^,^",options="header " , ] the production rate of non - relativistic boundstates in hard collisions is proportional to the squared wave function at the origin ( for @xmath19 waves ) or its derivative ( for @xmath20 waves ) , the latter being significantly suppressed and of relative order @xmath21 . for this reason the following discussion will be limited to @xmath19 waves only .
we , furthermore , anticipate that the experimental mass resolution and , eventually , the large natural width of the boundstate will lead to a sizable smearing of the `` narrow '' resonances .
the signal to background ratio will be small and their detection difficult .
for this reason we will consider color - singlet states only , where the level spacings and production rates are enhanced approximately by the square of the color coefficient listed in tab . [ table1 ] .
let us now discuss the qcd potential .
similarly to the case of heavy - quark boundstates it can be decomposed into the following two terms @xmath22 which are ordered according to their inverse powers in the constituent mass @xmath23 and are given in coordinate space in close similarity to the quarkonium potential ( see e.g. @xcite ) . for @xmath19 waves it reads @xmath24\biggr .
\nonumber\\ & & \biggl.+\frac{\alpha_{s}(\mu^2)^2}{(4\pi)^2}\biggl[a_{2}+2\gamma_{e}\left(2 a_{1}\beta_{0}+\beta_{1}\right)+\beta_{0}^2\left(\frac{\pi^2}{3}+4\gamma_{e}^2\right)\biggr.\biggr . \nonumber\\ & & \biggl.\biggl.+\left(4a_{1}\beta_{0}+2\beta_{1}+8\gamma_{e}\beta_{0}^2\right)\ln(\mu r)+4\beta_{0}^2\ln(\mu r)^2\biggr]\biggr\ } \,,\nonumber\\ v_{\rm nc}(\vec{r}\,)&=&u_{1}(r)+u_{2}(\vec{r}\,)+u_{3}(r ) \,,\nonumber\\ u_{1}(r)&=&-\frac{\delta^2}{4m_{\tilde{g}}^3}+\frac{c_{a}\alpha_{s}(\mu^2)}{m_{\tilde{g}}^2 r}\delta \nonumber\,,\\ u_{2}(\vec{r}\,)&=&\frac{4\pi c_{a}\alpha_{s}(\mu^2)}{3m_{\tilde{g}}^2}\vec{s}^2\delta\left(\vec{r}\,\right ) \,,\nonumber\\ u_{3}(r)&=&-\frac{c_{a}^2\alpha_{s}^2(\mu^2)}{4m_{\tilde{g } } r^2 } \,,\label{vundu}\end{aligned}\ ] ] where @xmath25 c_{a}^2-\left[\frac{1798}{81}+\frac{56}{3}\zeta(3)\right]c_{a}t_{f}n_{f } \nonumber\\ & & -\left[\frac{55}{3}-16\zeta(3)\right]c_{f}t_{f}n_{f}+\left(\frac{20}{9}t_{f}n_{f}\right)^2 \,.\label{betaunda}\end{aligned}\ ] ] here @xmath26 stands for the strong coupling constant in the @xmath27 scheme , @xmath28 is the number of active quark flavors , @xmath14 , @xmath29 , @xmath30 , @xmath31 the euler constant , @xmath32 and @xmath33 is the constituent pole mass . for the coulomb part @xmath34 the transformation from color - triplet , where the result can be found in the literature , to constituents in an arbitrary representation @xmath15 , bound to a singlet state , can be understood as follows : in lo the overall normalization is changed by substituting @xmath35 .
this is the only modification also in nlo , thus leaving the constant @xmath36 and the terms proportional to @xmath37 unchanged . for the case under discussion ( @xmath38 )
this is evident from fig .
[ feynnlo ] : diagram ( a ) contributes proportional @xmath39 , diagram ( b ) proportional @xmath40 .
its @xmath39 part can be combined with ( a ) and in total the ladder plus crossed ladder can be collected in the @xmath41 part of the potential . the @xmath42 part of diagram ( b ) contributes to the @xmath43 term in @xmath44 and in the constant @xmath36 .
diagram ( d ) contributes proportional @xmath45 and is responsible for the @xmath46 terms in @xmath44 and @xmath36 .
the corresponding considerations are also applicable to the two - loop case , see e.g. fig .
[ feynnnlo ] . for diagrams with virtual gluons only ( e.g. fig .
[ feynnnlo ] ( a ) ) , the substitution @xmath47 is valid throughout . for diagrams with exactly one dressed gluon propagator exchange ( e.g. fig .
[ feynnnlo ] ( c ) , ( d ) , ( e ) ) , the same substitution is valid , as far as the coupling of the gluon to the constituent is concerned and only the external @xmath48 is replaced by @xmath49 .
diagrams with three - gluon coupling through a triangular fermion loop ( fig .
[ feynnnlo ] ( b ) ) are also proportional to @xmath50 and it is again the first factor , which is chosen to be @xmath51 in the case of quark- and @xmath52 in the case of gluino - constituents .
diagrams similar to those of fig .
[ feynnlo ] ( a)-(c ) , but with one dressed gluon ( e.g. fig . [ feynnnlo ] ( f ) ) can be handled like the nlo diagrams . in total ,
up to nnlo , the coulomb part of the potential for the binding of constituents in an arbitrary representation @xmath15 is obtained through the substitution @xmath47 in the overall factor .
( 350,50)(0,0 ) ( 0,25)[l](a ) ( 10,50)(80,50 ) ( 27,50)1.3 ( 63,50)1.3 ( 27,50)(27,0)26 ( 63,50)(63,0)26 ( 27,0)1.3 ( 63,0)1.3 ( 10,0)(80,0 ) ( 90,25)[l](b ) ( 100,50)(170,50 ) ( 117,50)1.3 ( 153,50)1.3 ( 117,50)(153,0)28 ( 153,50)(117,0)28 ( 117,0)1.3 ( 153,0)1.3 ( 100,0)(170,0 ) ( 180,25)[l](c ) ( 190,50)(260,50 ) ( 207,50)1.3 ( 243,50)1.3 ( 207,50)(225,25)24 ( 243,50)(225,25)24 ( 225,25)1.3 ( 225,25)(225,0)23 ( 225,0)1.3 ( 190,0)(260,0 ) ( 270,25)[l](d ) ( 280,50)(350,50 ) ( 315,50)1.3 ( 315,50)(315,37)21 ( 315,37)1.3 ( 315,25)(12,0,360 ) ( 315,13)1.3 ( 315,13)(315,0)21 ( 315,0)1.3 ( 280,0)(350,0 ) ( 280,140)(0,0 ) ( -10,110)[l](a ) ( 0,140)(80,140 ) ( 10,140)1.2 ( 40,140)1.2 ( 70,140)1.2 ( 10,140)(40,80)27 ( 40,140)(70,80)27 ( 70,140)(10,80)29 ( 10,80)1.2 ( 40,80)1.2 ( 70,80)1.2 ( 0,80)(80,80 ) ( 90,110)[l](b ) ( 100,140)(180,140 ) ( 115,140)1.2 ( 165,140)1.2 ( 165,140)(153,123)22 ( 115,140)(127,123)22 ( 153,123)1.2 ( 127,123)1.2 ( 140,99)1.2 ( 153,123)(127,123 ) ( 127,123)(140,99 ) ( 140,99)(153,123 ) ( 140,99)(140,80)22 ( 140,80)1.2 ( 100,80)(180,80 ) ( 190,110)[l](c ) ( 200,140)(280,140 ) ( 240,140)1.2 ( 240,140)(240,125)22 ( 240,125)1.2 ( 240,110)(15,0,360 ) ( 225,110)1.2 ( 255,110)(225,110)23 ( 255,110)1.2 ( 240,95)1.2 ( 240,95)(240,80)22 ( 240,80)1.2 ( 200,80)(280,80 ) ( -10,30)[l](d ) ( 0,60)(80,60 ) ( 40,60)1.2 ( 40,60)(40,45)22 ( 40,45)1.2 ( 40,30)(15,0,360 ) ( 32.5,42.99)1.2 ( 32.5,42.99)(32.5,17.01)23 ( 32.5,17.01)1.2 ( 40,15)1.2 ( 40,15)(40,0)22 ( 40,0)1.2 ( 0,0)(80,0 ) ( 90,30)[l](e ) ( 100,60)(180,60 ) ( 140,60)1.2 ( 140,60)(140,45)22 ( 140,45)1.2 ( 140,30)(15,0,180 ) ( 155,30)(125,30 ) ( 155,30)1.2 ( 125,30)1.2 ( 140,30)(15,180,270)22 ( 140,30)(15,270,360)22 ( 140,15)1.2 ( 140,15)(140,0)22 ( 140,0)1.2 ( 100,0)(180,0 ) ( 190,30)[l](f ) ( 200,60)(280,60 ) ( 210,60)1.2 ( 262.07,52.07)(270,60)-21 ( 210,0)(247.93,37.93)-26 ( 270,60)1.2 ( 255,45)(10,0,360 ) ( 262.07,52.07)1.2 ( 247.93,37.93)1.2 ( 210,0)1.2 ( 270,0)(210,60)-29 ( 270,0)1.2 ( 200,0)(280,0 ) the term @xmath53 of eq .
( [ vundu ] ) of order @xmath54 originates from a one - loop calculation , and the factor @xmath55 must be replaced by @xmath56 .
the terms of order @xmath57 collected in @xmath58 and @xmath59 can be translated simply by replacing the global factor @xmath48 by @xmath49 .
for our application the substitution @xmath60 is implicit .
using the potential as specified in eqs .
( [ pot ] ) and ( [ vundu ] ) , binding energies , level spacings and bound - state wave functions are easily obtained by solving schrdinger s equation numerically for @xmath61 and adding the singular terms collected in @xmath62 and @xmath53 in perturbation theory or , alternatively , by evaluating energy level and wave functions in a perturbative series in @xmath63 .
the analytic results both for binding energies @xmath64 and wave functions at the origin @xmath65 are listed in appendix [ app : energy ] . in numerical form
the results for the two lowest levels are given by @xmath66\right\ } \,,\nonumber\\ e_{2}&\hspace{-0.1cm}=&\hspace{-0.2cm}-\frac{m_{\tilde{g}}c_{a}^2\alpha_{s}^2}{16}\left\ { 1+\alpha_{s}\left(2.44l_{2}+4.42\right)\right .
\nonumber\\ & & \hspace{0cm}\hspace{1.95cm}\left.+\alpha_{s}^2\left[\left(4.47l_{2}^2 + 14.19l_{2}+20.54\right)_{c}+\left(11.95\right)_{nc}\right]\right\ } \,,\nonumber\\ \left|\psi_{1}\left(0\right)\right|^2&\hspace{-0.1cm}=&\hspace{-0.2cm}\frac{m_{\tilde{g}}^3c_{a}^3\alpha_{s}^3}{8\pi}\left\ { 1+\alpha_{s}\left(3.66l_{1}-0.43\right)\right .
\nonumber\\ & & \hspace{-0.2cm}\hspace{1.8cm}\left.+\alpha_{s}^2\left[\left(8.93l_{1}^2 - 5.11l_{1}+5.83\right)_{c}+\left(27l_{1}+57.38\right)_{nc}\right]\right\ } \,,\nonumber\\ \left|\psi_{2}\left(0\right)\right|^2&\hspace{-0.1cm}=&\hspace{-0.2cm}\frac{m_{\tilde{g}}^3c_{a}^3\alpha_{s}^3}{64\pi}\left\ { 1+\alpha_{s}\left(3.66l_{2}-0.18\right)\right .
\nonumber\\ & & \hspace{-0.2cm}\hspace{1.8cm}\left.+\alpha_{s}^2\left[\left(8.93l_{2}^2 - 3.86l_{2}+10.19\right)_{c}+\left(27l_{2}+29.53\right)_{nc}\right]\right\ } \,,\label{eundpsinum}\end{aligned}\ ] ] with @xmath67 and @xmath33 being the gluino pole mass .
the first of the @xmath68 terms give the coulombic corrections ( @xmath69 ) the second ones the non - coulombic ones ( @xmath70 ) . here and below we use @xmath71 as defined in the @xmath27 scheme and the @xmath72 dependence from the numerical solution of the three loop renormalization group equation with the starting value @xmath73 .
we use @xmath74 .
this is well justified , as long as the characteristic scale @xmath75 is smaller than the top - quark mass , an assumption well justified for all gluino masses under consideration .
top - quark effects could be included following @xcite . if not stated otherwise , in the following we use for the renormalization scale @xmath76 with @xmath77 corresponding to @xmath78 .
we have convinced ourselves that the perturbative series and the numerical results are in very good agreement as far as @xmath34 is concerned .
the subsequent analysis will therefore be based on the fully perturbative approach . in the pole mass scheme as function
of twice the constituent mass . ] in the potential subtracted mass scheme as function of twice the potential subtracted mass .
the curves have been obtained using the pole mass as input and evaluating both the potential subtracted mass ( using eq .
( [ mps ] ) ) and @xmath79 to a given order in @xmath26 .
note that the dash - dotted and dashed curves are almost on top of each other . ]
the groundstate energy @xmath80 is shown in fig .
[ energiepol ] as a function of @xmath81 for gluino masses @xmath33 between @xmath82 and @xmath83 .
the lo ( dotted ) , nlo ( dashed ) and nnlo predictions are displayed individually . for the nnlo result both
the result for @xmath34 ( dash - dotted ) and the one including the corrections collected in @xmath58 , @xmath59 and @xmath53 ( solid ) are shown .
as stated above , @xmath33 is understood as the constituent pole mass .
the poor convergence of the perturbative series can partly be traced back to the large non - coulombic correction which appear for the first time in nnlo and amount to twice the nnlo coulombic one .
qualitatively this behavior is quite similar to the one observed for quarkonia in @xcite .
it remains to be seen to which extent inclusion of orders of @xmath84 and higher will stabilize these predictions . in fig .
[ energieps ] we show as an alternative representation the predictions for the ground state energy , using as reference the potential subtracted mass @xcite , which is related to the pole mass through @xmath85\biggr . \nonumber\\ & & \biggl.\mbox{}+\left(\frac{\alpha_{s}(\mu)}{4\pi}\right)^2\biggl[a_{2}-\left(2a_{1}\beta_{0}+\beta_{1}\right)\left(\ln\left(\frac{\mu_{f}^2}{\mu^2}\right)-2\right)\biggr.\biggr .
\nonumber\\ & & \biggl.\biggl.\mbox{}+\beta_{0}^2\left(\ln^2\left(\frac{\mu_{f}^2}{\mu^2}\right)-4\ln\left(\frac{\mu_{f}^2}{\mu^2}\right)+8\right)\biggr]\biggr\ } \ , . \label{mps}\end{aligned}\ ] ] in the following we adopt for the factorization scale @xmath86 , independent of @xmath87 ( with @xmath77 as before ) .
( note , that @xmath88 now also depends on @xmath87 . )
we refrain from listing explicitly the results for the energy levels in the potential subtracted scheme since the corresponding formulae are quite bulky
. they can easily be obtained from eqs .
( [ eundpsinum ] ) and ( [ mps ] ) .
the implementation of the potential subtracted mass leads to a significantly improved convergence of the perturbative series for @xmath89 as far as the coulombic part is concerned ( see fig . [ energieps ] ) and similarly for @xmath90 .
however , the meson mass difference between the @xmath91 and the @xmath92 state , @xmath93 , is independent of the choice of the scheme and exhibits a poor convergence in both cases .
the predictions for @xmath93 are shown in fig .
[ anregungpol ] , adopting @xmath76 .
the nnlo terms are evidently important and the poor convergence of the prediction for @xmath80 is reflected in this figure . in the on - shell scheme
the @xmath91 binding energy is evidently closely related to the energy difference between the groundstate and the onset of open gluino production or , more specifically , the threshold for pair production of color - neutral @xmath94 hadrons .
however , anticipating a mass resolution of several tens of gev at least , it will be difficult to resolve the densely distributed radial excitations . for this reason
the energy difference between the groundstate and the first radial excitation is a convenient measure of the isolation of the @xmath91 state . as function
of twice the constituent mass .
] production and decay of non - relativistic @xmath19-wave bound states are proportional to the square of the wave function at the origin .
the analytic results are collected in the appendix [ app : energy ] , numerical results are listed in eq .
( [ eundpsinum ] ) and shown in fig .
[ wellenfunktion ] .
the enormous size in particular of the non - coulombic nnlo terms has also been observed for the top - antitop system and destabilizes the predictions @xcite . at present the size of these corrections
must be considered as an estimate of the theory uncertainty . in the pole mass scheme ( a ) and the potential subtracted scheme ( b).,title="fig : " ] in the pole mass scheme ( a ) and the potential subtracted scheme ( b).,title="fig : " ] . ]
the renormalization scale dependence of the lo , nlo and @xmath95 results for @xmath80 and @xmath96 are shown in figs .
[ energiemu ] and [ wellenfunktionmu ] for a constituent mass of @xmath97 tev .
the predictions are normalized to the lo prediction evaluated at the scale @xmath72 as defined above .
a significant stabilization is observed with increasing order in @xmath63 and , furthermore , the higher order contributions are reasonably small for the preferred choice of the renormalization scale , at least , as far as the coulombic part is concerned .
this is in particular true for the ground state energy parameterized in terms of the potential subtracted mass where close to @xmath98 gev the coulombic nnlo results even vanish . in the case of the wave function one
observes around @xmath99 gev corrections of the order of a few percent from nlo and nnlo@xmath100 , however , a huge contribution from the non - coulombic terms .
in the present context squarks are assumed to be significantly heavier than gluinos , such that the direct decay @xmath101 is forbidden .
an extreme example is provided by split susy @xcite with @xmath102 and we shall adopt this simplifying mass assignment throughout this section . however , this assumption could be significantly relaxed .
indeed , once @xmath103 is valid the gluinos have a very small decay rate , for @xmath104 of order @xcite @xmath105 \ , , \label{singledecay}\end{aligned}\ ] ] characterizing the rate for gluino decays into neutralinos ( similar estimates being applicable for decays into charginos ) such that the decay rate is far smaller than the annihilation decay .
the evaluation of the rate for gluinonium annihilation into gluon ( and quark ) jets proceeds similar to the one for quarkonia . in born approximation
the color factor is replaced as follows @xmath106 and another relative factor @xmath107 arises from the majorana nature of the gluinos , whence @xcite @xmath108 with @xmath109 . using the wave function
obtained in lowest order perturbation theory leads to @xmath110 which provides a qualitative estimate of the full result .
the perturbative corrections arising from virtual and real emission can be calculated similar to the ones for quarkonia @xcite .
they reduce the scale dependence of the lowest order approximation and are complementary to the higher order corrections to the wave function . in total
one finds @xmath111\biggr\ } \ , .
\nonumber\\ \label{gammanlo}\end{aligned}\ ] ] state into gluons . ] here @xmath63 is understood in the theory with @xmath112 effective flavors and all quark masses , including @xmath113 , have been neglected . the @xmath114 term present in @xmath115
is interpreted in the @xmath74 effective theory .
the predictions for @xmath116 and @xmath117 are shown in fig .
[ zerfall ] , where the nnlo wave function as shown in fig .
[ wellenfunktion ] ( excluding the non - coulombic contribution ) and @xmath118 for the renormalization scale in eq .
( [ gammanlo ] ) have been adopted . as anticipated , the decay rate of the bound state is small compared to the level spacing and large compared to @xmath119 , the decay rate of a single gluino , once @xmath120 .
hence annihilation decays will constitute the dominant signal .
however , it should be emphasized , that bound states will exist even for @xmath121 , as long as @xmath122 , which is true for the region @xmath123 .
the bound state production cross section can be calculated similarly to the one for the pseudoscalar bound state of two top quarks ( usually called @xmath124 ) @xcite . in born approximation
the reaction proceeds through gluon fusion and the radiative corrections involve additional @xmath125 , @xmath126 and @xmath127 initiated subprocesses .
the final result is quite similar to the one for quarkonia production @xcite .
we find @xmath128\biggr\ } \,,\nonumber\\ \hat{\sigma}_{gq}&=&\sigma_{0}\frac{\alpha_{s}}{\pi}\biggl\{-c_{f}\frac{z}{2}\ln z+c_{f}z+\frac{1}{2}p_{gq}(z)\left(\ln\left(\frac{4m_{\tilde{g}}^2(1-z)^2}{\mu^2}\right)-1\right)\biggr\ } \,,\nonumber\\ \hat{\sigma}_{g\overline{q}}&=&\hat{\sigma}_{gq } \,,\nonumber\\ \hat{\sigma}_{q\overline{q}}&=&\sigma_{0}\frac{\alpha_{s}}{\pi}\frac{32}{27}z(1-z ) \,,\ ] ] with @xmath129_{+}+z(1-z)-2\right ) \nonumber\\ & & + \left(\frac{11}{6}c_{a}-\frac{2}{3}n_{f}t_{f}\right)\delta(1-z ) \,,\nonumber\\ p_{gq}(z)&= & c_{f}\frac{1+(1-z)^2}{z } \,,\nonumber\\ f(z)&=&\frac{11z^5 + 11z^4 + 13z^3 + 19z^2 + 6z-12}{6z(1+z)^2}\biggr .
\nonumber\\ & & \mbox{}+\,4\left(\frac{1}{z}+z(1-z)-2\right)\ln(1-z)+4\left[\frac{\ln(1-z)}{1-z}\right]_{+ } \nonumber\\ & & \mbox{}\biggl.+\left(\frac{2(z^3 - 2z^2 - 3z-2)(z^3-z+2)z\ln z}{(1+z)^3(1-z)}-3\right)\frac{1}{1-z } \,.\label{sigmaadd}\end{aligned}\ ] ] here @xmath130 , @xmath72 and @xmath131 are the renormalization and the factorization scales , @xmath63 is again defined in the @xmath112 flavor theory , @xmath132 are the altarelli - parisi splitting functions and @xmath133_{+}$ ] denotes the usual @xmath134 distribution . for the wave function at the origin
we again adopt the nnlo result without the non - coulombic terms .
the predictions for the production cross section are shown in fig .
[ produktion ] where the parton distribution functions mstw2008lo ( mstw2008nlo ) @xcite have been used for the lo ( nlo ) calculation . leading order ( dashed ) and nlo ( solid curve ) predictions are evaluated using @xmath135 .
state at @xmath136 . ] production cross section and decomposition into contributions from various subprocesses .
the line labeled `` gg '' contains also the lo result . ] for a gluino mass of @xmath137 the @xmath72 dependence ( with @xmath138 ) of the lo and the nlo results is shown in fig .
[ produktionmu ] , together with the decomposition of the nlo result into the contributions from the different subprocesses . from figs .
[ produktion ] and [ produktionmu ] we learn that the nlo correction amounts to more than @xmath139 fairly independent of the renormalization scale .
the dominant uncertainty , not displayed in these figures , arises from the wave function at the origin , with indications for an increase by another factor of two . for an anticipated integrated luminosity of @xmath140 between @xmath141 and @xmath142 events
are expected for @xmath33 between @xmath143 and @xmath144 .
it remains to be seen , if these can be separated from the hadronic continuum background .
in the limit of extremely heavy squarks ( and conserved r - parity ) , corresponding to split susy , decay modes of single gluinos are extremely suppressed , such that gluinos ( confined in color - neutral hadrons like @xmath145 ) could even travel microscopic distances . in this case gluinonium annihilation into two gluons is the only relevant channel .
however , once squark and gluino masses are comparable , new annihilation decays become possible . at first glance
one might consider the mode @xmath146 , mediated by squark exchange .
however , for massless quarks this process vanishes as a consequence of helicity conservation . for massive top quarks one finds @xmath147 using @xmath148 . the complete result for @xmath149 and including squark mixing is given in appendix [ app : ttbar ] .
decay rate into top quarks . ]
the result is plotted in fig .
[ ratiott ] for different ratios of @xmath150 . branching ratios of up to @xmath151
are possible for gluino masses close to @xmath152 and squark masses not much bigger . for heavier squark and gluino masses
the ratio drops quickly . as a second possibility we consider the loop induced decay into two photons .
although suppressed by the electromagnetic couplings @xmath153 , the cleaner signal - to - background ratio might ( at least in principle ) compensate this disadvantage . for arbitrary squark mixing and masses the result
is given in appendix [ app : gamgam ] .
in the limit of massless quarks and degenerate squarks the decay rate simplifies considerably @xmath154 with a ratio @xmath155 around one to several @xmath156 the two - photon signal will not be detectable .
assuming the two - gluon mode to dominate the decay of gluino bound states , the resonant signal must be isolated from the two - jet continuum .
below we present an order of magnitude estimate for the signal - to - background ratio , following essentially the arguments given in ref .
@xcite . for simplicity
it is assumed that quark and gluon jets can be experimentally separated . in this case
the background is dominated by gluon - gluon scattering . furthermore , we enhance the signal - to - background ratio by excluding small scattering angles in the partonic center - of - mass system , requiring @xmath157 . to isolate a resonance in the gluon - gluon channel
, a good experimental resolution @xmath158 of the invariant dijet mass @xmath159 is essential .
let us compare the differential cross section , integrated over the interval @xmath160 for background and signal , and integrated over the same angular range .
this leads to the signal - to - background ratio @xmath161^{-1 } .\label{soverr}\end{aligned}\ ] ] . ]
using @xmath162 for the dijet resolution @xcite , one obtains the signal - to - background ratio as shown in fig .
[ back ] , varying between @xmath163 to @xmath164 . given event rates between @xmath165 and @xmath166 for @xmath33 between @xmath167 and @xmath168 the detection of gluino bound states might become feasible , in particular at a high luminosity version of the lhc .
for a large class of supersymmetric extensions of the standard model gluinos will be sufficiently stable to form color - neutral non - relativistic boundstates .
the nnlo potential between color - octet constituents forming color singlets has been derived .
just as in leading and next - to - leading order the coulombic part of the nnlo potential is related to the quarkonium potential through the substitution @xmath169 in the overall factor .
the transformation of the short distance terms involving an additional factor @xmath170 is only slightly more complicated and discussed in section [ sec : properties ] .
based on this potential , predictions for the binding energy and the wave function at the origin of the lowest - lying boundstates and the first radial excitation have been derived . for gluino masses between @xmath152 and @xmath171 the @xmath91 binding energy ( in the pole mass scheme ) varies between @xmath172 and @xmath173 , the @xmath174 energy difference between @xmath175 and @xmath176 . for the case of gluinos
being lighter than squarks the gluino decay rate is significantly smaller than the level spacing and the boundstates are thus well separated from the continuum .
also the wave function at the origin , which enters bound state production and decay rates , has been evaluated in nnlo .
for @xmath121 the decay rate of gluinonia into two gluon jets is evaluated in nlo and shown to dominate all other modes . for top squarks with masses comparable to @xmath33 the decay mode into @xmath177 may reach several percent , the decay through virtual quark loops into two photons is too small to be detectable .
we evaluate the boundstate production cross section again in nlo .
large positive corrections of around @xmath178 arise from the nnlo terms in the wave function at the origin and around @xmath179 from the nlo terms in the cross section .
the production rate depends strongly on the gluino mass , with up to @xmath142 events for gluino masses close to the lower limit around @xmath144 ( for an integrated luminosity of @xmath140 ) dropping quickly to @xmath141 events only for @xmath180 .
using standard assumptions for the dijet - mass resolution and assuming that gluon and quark jets can be separated , a signal - to - background ratio between @xmath181 and @xmath164 percent might be conceivable . for favorable gluino- and squark - mass combinations gluinonia might thus be detectable .
* acknowledgements * m.s .
thanks alexander penin for discussions about @xmath182 .
this work was supported by the bmbf through grant no . 05h09vke .
_ note added _ : after completion of this work we received a preprint by k. hagiwara and h. yokoya where the production of unstable gluinos in the threshold region is considered @xcite .
for completeness we present the perturbative analytic results for the energy levels @xmath64 of the @xmath19 waves and the corresponding values of the squared wave function at the origin @xmath65 , truncating the perturbative series at nnlo .
we present the results following the notation of @xcite .
we define @xmath183 \,,\nonumber\\ \ , , \label{eundpsis}\end{aligned}\ ] ] and parameterize the corrections as @xmath184 where @xmath185 stands for the corrections to the coulombic part of the potential and @xmath186 for the remaining one .
the lo values are given as @xmath187 the nlo and nnlo corrections are @xmath188 \nonumber\\ & & + a_{1}^2 + 2a_{2}+4s_{1}\beta_{1}+4a_{1}\beta_{0}\left(3s_{1}-1\right ) \nonumber\\ & & + \beta_{0}^2\left[s_{1}\left(12s_{1}-8-\frac{8}{n}\right)+16s_{2}-8ns_{3}+\frac{2\pi^2}{3}+8n\zeta(3)\right ] \,,\nonumber\\ e_{2}^{nc}&=&\frac{16\pi^2c_{a}^2}{n}\left(3-\frac{11}{16n}-\frac{2}{3}\vec{s}^2\right ) \,,\nonumber\\ f_{1}^{c}&=&6\beta_{0}\ln\left(\frac{n\mu}{m_{\tilde{g}}c_{a}\alpha_{s}}\right)+3a_{1}+2\beta_{0}\left(s_{1}+2ns_{2}-1-\frac{n\pi^2}{3}\right ) \,,\nonumber\\
f_{1}^{nc}&=&0 \,,\nonumber\\ f_{2}^{c}&=&24\beta_{0}^2\ln^2\left(\frac{n\mu}{m_{\tilde{g}}c_{a}\alpha_{s}}\right ) \nonumber\\ & & + \ln\left(\frac{n\mu}{m_{\tilde{g}}c_{a}\alpha_{s}}\right)\biggl\{-12\beta_{0}^2 + 6\beta_{1}+8\beta_{0}\left[3a_{1}+2\beta_{0}\left(s_{1}+2ns_{2}-1-\frac{n\pi^2}{3}\right)\right]\biggr\ } \nonumber\\ & & + 3a_{1}^2 + 3a_{2}+2a_{1}\beta_{0}\left(4s_{1}+8ns_{2}-7-\frac{4n\pi^2}{3}\right)+2\beta_{1}\left(s_{1}+2ns_{2}-1-\frac{n\pi^2}{3}\right ) \nonumber\\ & & + \beta_{0}^2\biggl[s_{1}\left(8s_{1}+16ns_{2}-20-\frac{12}{n}-\frac{8n\pi^2}{3}\right)+s_{2}\left(4n^2s_{2}+8 - 8n-\frac{4n^2\pi^2}{3}\right ) \biggr .
\nonumber\\ & & \hspace{0.3cm}\biggl.+28ns_{3}-20n^2s_{4}-24ns_{2,1}+16n^2s_{3,1}+4+\frac{(3 + 4n)\pi^2}{3}+\frac{n^2\pi^4}{9}+20n\zeta(3)\biggr ] \,,\nonumber\\ f_{2}^{nc}&=&16\pi^2c_{a}^2\biggl\{3\ln\left(\frac{n\mu}{m_{\tilde{g}}c_{a}\alpha_{s}}\right)-3s_{1}+\frac{6}{n}-\frac{15}{8n^2}+\frac{21}{4}\biggr .
\nonumber\\ & & \hspace{1.7cm}+\biggl.\vec{s}^2\left[-\frac{2}{3}\ln\left(\frac{n\mu}{m_{\tilde{g}}c_{a}\alpha_{s}}\right)+\frac{2}{3}s_{1}-\frac{4}{3n}-\frac{7}{9}\right]\biggr\ } \,.\label{eundf}\end{aligned}\ ] ] the constants @xmath189 and @xmath190 depend on the casimir operators @xmath191 and @xmath192 and are defined in eq .
( [ betaunda ] ) .
the harmonic sums @xmath193 , the nested harmonic sums @xmath194 and the zeta - function @xmath195 are given as @xmath196
( 275,50)(0,0 ) ( 0,0)(25,25)-2.53 ( 0,0)(25,25 ) ( 0,50)(25,25)2.53 ( 0,50)(25,25 ) ( 25,25)1.8 ( 25,25)(50,25)33 ( 50,25)1.8 ( 75,0)(50,25 ) ( 50,25)(75,50 ) ( 100,0)(137.5,10)-2.53 ( 100,0)(137.5,10 ) ( 100,50)(137.5,40)2.53 ( 100,50)(137.5,40 ) ( 137.5,40)1.8 ( 137.5,10)(137.5,40)3 ( 137.5,10)1.8 ( 137.5,40)(175,50 ) ( 175,0)(137.5,10 ) ( 118,25)[l]@xmath197 ( 200,0)(237.5,10)-2.53 ( 200,0)(237.5,10 ) ( 200,50)(237.5,40)2.53 ( 200,50)(237.5,40 ) ( 237.5,40)1.8 ( 237.5,40)(237.5,10)3 ( 237.5,10)1.8 ( 237.5,10)(251.5625,25 ) ( 237.5,40)(251.5625,25 ) ( 251.5625,25)(275,50 ) ( 275,0)(251.5625,25 ) ( 218,25)[l]@xmath197 in fig . [ qqbar ] the corresponding diagrams for the decay of two gluinos into top - antitop are shown .
diagram @xmath198 drops out for the case of interest because the two gluinos are projected here onto a color - singlet state .
the resulting amplitude for @xmath199 and @xmath200 reads @xmath201\left(p_{-}u_{h1}^{(\tilde{t})}-p_{+}u_{h2}^{(\tilde{t})}\right)v_{s_{2}}(k_{2},m_{t } ) \,,\label{ttbaramp}\end{aligned}\ ] ] where @xmath202 and @xmath203 are the spin indices of the top , the anti - top and of the gluinos , @xmath204 and @xmath205 are the left- and right - handed projectors as usual and the @xmath206 are the elements of the orthogonal squark - mixing matrix @xmath207 the relative minus sign between the diagrams @xmath199 and @xmath200 of fig . [ qqbar ] stems from the interchange of two fermionic operators . the amplitude in eq .
( [ ttbaramp ] ) has now , following the methods explained in refs .
@xcite , to be projected onto the spin-@xmath208 state using the clebsch - gordon coefficients @xmath209 , where the squared brackets of eq .
( [ ttbaramp ] ) become due to the antisymmetry of the spin function @xmath210\left<0,0;s,\overline{s}\right > \nonumber\\ & = & 2\sum_{s,\overline{s}}\left[u_{\overline{s}}\left(\frac{p}{2},m_{\tilde{g}}\right)\overline{v}_{s}\left(\frac{p}{2},m_{\tilde{g}}\right)\right]\left<0,0;s,\overline{s}\right > \,.\label{spinproj}\end{aligned}\ ] ] now the calculation can be done as for the case of the decay into two gluons and the rate reads @xmath211 ( 455,60)(-8,0 ) ( 0,0)(30,10)-2.53 ( 0,0)(30,10 ) ( 0,55)(30,45)2.53 ( 0,55)(30,45 ) ( 30,10)1.8 ( 30,45)1.8 ( 30,10)(30,45)3 ( 65,10)(30,10 ) ( 30,45)(65,45 ) ( 65,45)(65,10 ) ( 65,10)1.8 ( 65,45)1.8 ( 65,10)(95,0)-33 ( 65,45)(95,55)33 ( 120,0)(150,10)-2.53 ( 120,0)(150,10 ) ( 120,55)(150,45)2.53 ( 120,55)(150,45 ) ( 150,10)1.8 ( 150,45)1.8 ( 150,45)(150,10 ) ( 150,10)(185,10)3 ( 185,45)(150,45)3 ( 185,10)(185,45)3 ( 185,10)1.8 ( 185,45)1.8 ( 185,10)(215,0)-33 ( 185,45)(215,55)33 ( 240,-5)(305,10)-2.56 ( 240,-5)(305,10 ) ( 240,55)(270,45)2.53 ( 240,55)(270,45 ) ( 270,10)1.8 ( 270,45)1.8 ( 270,10)(270,45)3 ( 305,10)(270,10)3 ( 270,45)(305,45 ) ( 305,45)(305,10 ) ( 305,10)1.8 ( 305,45)1.8 ( 270,10)(335,-5)-37 ( 305,45)(335,55)33 ( 360,0)(390,10)-2.53 ( 360,0)(390,10 ) ( 360,55)(390,45)2.53 ( 360,55)(390,45 ) ( 390,10)1.8 ( 390,45)1.8 ( 390,10)(390,45 ) ( 425,27.5)(390,10)3 ( 390,45)(425,27.5)3 ( 425,27.5)1.8 ( 425,27.5)(455,0)-34 ( 425,27.5)(455,55)34 treating all the virtual quarks in fig . [ 2gluto2gam ] as massless , the onto spin and color - projected amplitude has the form @xmath212 the sum in the first equation of ( [ ampgamgam ] ) goes over the flavors ( @xmath213 ) and the indices of the squarks ( @xmath214 ) .
the @xmath215 are the momenta of the outgoing photons , the @xmath216 their polarization vectors and the @xmath217 the charges of the corresponding quarks and squarks .
the definition of the @xmath218 -functions is the same as in ref .
@xcite .
99 w. y. keung and a. khare , phys .
d * 29 * ( 1984 ) 2657 .
j. h. khn and s. ono , phys .
b * 142 * ( 1984 ) 436 .
j. t. goldman and h. haber , physica * 15d * , 181 ( 1985 ) . v. g. kartvelishvili , a. v. tkabladze and e. g. chikovani , sov .
. phys .
* 51 * , 546 ( 1990 ) [ yad .
fiz . * 51 * , 859 ( 1990 ) ] .
e. chikovani , v. kartvelishvili , r. shanidze and g. shaw , phys .
d * 53 * , 6653 ( 1996 ) [ arxiv : hep - ph/9602249 ]
. v. g. kartvelishvili , a. v. tkabladze and e. g. chikovani , z. phys .
c * 43 * ( 1989 ) 509 .
n. arkani - hamed and s. dimopoulos , jhep * 0506 * , 073 ( 2005 ) [ arxiv : hep - th/0405159 ] .
g. f. giudice and a. romanino , nucl .
b * 699 * , 65 ( 2004 ) [ erratum - ibid .
b * 706 * , 65 ( 2005 ) ] [ arxiv : hep - ph/0406088 ] .
w. kilian , t. plehn , p. richardson and e. schmidt , eur .
j. c * 39 * , 229 ( 2005 ) [ arxiv : hep - ph/0408088 ] .
k. cheung and w. y. keung , phys .
d * 71 * , 015015 ( 2005 ) [ arxiv : hep - ph/0408335 ] .
m. peter , phys .
lett . * 78 * , 602 ( 1997 ) [ arxiv : hep - ph/9610209 ] .
m. peter , nucl .
b * 501 * , 471 ( 1997 ) [ arxiv : hep - ph/9702245 ] .
y. schrder , phys .
b * 447 * , 321 ( 1999 ) [ arxiv : hep - ph/9812205 ] .
a. v. smirnov , v. a. smirnov and m. steinhauser , phys .
b * 668 * ( 2008 ) 293 [ arxiv:0809.1927 [ hep - ph ] ] .
s. n. gupta and s. f. radford , phys .
d * 24 * , 2309 ( 1981 ) .
s. n. gupta and s. f. radford , phys .
d * 25 * , 3430 ( 1982 ) .
s. titard and f. j. yndurain , phys .
d * 49 * , 6007 ( 1994 ) [ arxiv : hep - ph/9310236 ] .
b. a. kniehl , a. a. penin , m. steinhauser and v. a. smirnov , phys .
d * 65 * , 091503 ( 2002 ) [ arxiv : hep - ph/0106135 ] .
a. pineda and f. j. yndurain , phys .
d * 58 * , 094022 ( 1998 ) [ arxiv : hep - ph/9711287 ] .
a. pineda and j. soto , nucl .
* 64 * , 428 ( 1998 ) [ arxiv : hep - ph/9707481 ] .
a. v. manohar and i. w. stewart , phys .
d * 62 * , 074015 ( 2000 ) [ arxiv : hep - ph/0003032 ]
. m. beneke , phys .
b * 434 * ( 1998 ) 115 [ arxiv : hep - ph/9804241 ] .
a. h. hoang , m. c. smith , t. stelzer and s. willenbrock , phys .
d * 59 * ( 1999 ) 114014 [ arxiv : hep - ph/9804227 ] . j. h. khn , a. a. penin and a. a. pivovarov , nucl .
b * 534 * , 356 ( 1998 ) [ arxiv : hep - ph/9801356 ] .
a. a. penin and m. steinhauser , phys .
b * 538 * ( 2002 ) 335 [ arxiv : hep - ph/0204290 ] .
b. a. kniehl , a. a. penin , v. a. smirnov and m. steinhauser , nucl .
b * 635 * , 357 ( 2002 ) [ arxiv : hep - ph/0203166 ] .
v. s. fadin and v. a. khoze , jetp lett . * 46 * , 525 ( 1987 ) [ pisma zh .
teor . fiz .
* 46 * , 417 ( 1987 ) ] .
j. h. khn and p. m. zerwas , phys .
rept . * 167 * , 321 ( 1988 ) .
m. j. strassler and m. e. peskin , phys . rev .
d * 43 * , 1500 ( 1991 ) .
y. sumino , k. fujii , k. hagiwara , h. murayama and c. k. ng , phys .
d * 47 * , 56 ( 1993 ) .
m. jezabek , j. h. khn and t. teubner , z. phys .
c * 56 * , 653 ( 1992 ) .
a. h. hoang _ et al .
_ , eur .
j. direct c * 2 * , 1 ( 2000 ) [ arxiv : hep - ph/0001286 ] .
a. h. hoang and c. j. reisser , phys .
d * 71 * ( 2005 ) 074022 [ arxiv : hep - ph/0412258 ] .
v. s. fadin , v. a. khoze and t. sjstrand , z. phys .
c * 48 * , 613 ( 1990 ) .
k. hagiwara , y. sumino and h. yokoya , phys .
b * 666 * , 71 ( 2008 ) [ arxiv:0804.1014 [ hep - ph ] ] .
y. kiyo , j. h. kuhn , s. moch , m. steinhauser and p. uwer , eur .
j. c * 60 * ( 2009 ) 375 [ arxiv:0812.0919 [ hep - ph ] ] .
m. melles , phys .
d * 58 * , 114004 ( 1998 ) [ arxiv : hep - ph/9805216 ] .
m. melles , phys .
d * 62 * , 074019 ( 2000 ) [ arxiv : hep - ph/0001295 ] .
m. beneke , y. kiyo and k. schuller , nucl .
b * 714 * , 67 ( 2005 ) [ arxiv : hep - ph/0501289 ] .
a. a. penin , v. a. smirnov and m. steinhauser , nucl .
b * 716 * ( 2005 ) 303 [ arxiv : hep - ph/0501042 ] .
y. kiyo and y. sumino , phys . rev .
d * 67 * ( 2003 ) 071501 [ arxiv : hep - ph/0211299 ] .
h. e. haber and g. l. kane , phys .
* 117 * , 75 ( 1985 ) .
j. h. khn and e. mirkes , phys .
d * 48 * , 179 ( 1993 ) [ arxiv : hep - ph/9301204 ] .
a. d. martin , w. j. stirling , r. s. thorne and g. watt , arxiv:0901.0002 [ hep - ph ] . the cms collaboration , cms pas sbm 07 001 . k. hagiwara and h. yokoya , arxiv:0909.3204 [ hep - ph ] .
b. guberina , j. h. khn , r. d. peccei and r. rckl , nucl .
b * 174 * , 317 ( 1980 ) .
j. h. khn , j. kaplan and e. g. o. safiani , nucl . phys .
b * 157 * , 125 ( 1979 ) .
g. t hooft and m. j. g. veltman , nucl .
b * 153 * , 365 ( 1979 ) . | predictions for energy levels , production and decay rate of gluinonia , non - relativistic boundstates of gluinos , are presented .
the potential between color - octet constituents is derived in next - to - next - to leading order and one - loop qcd corrections are derived for the production cross section and the decay rate into gluon jets . in addition we evaluate the decay rate into top quarks and into two photons . the signal - to - background ratio is estimated for the dominant decay mode and found to be around @xmath0 . for relatively light gluinos the bound states thus might be detectable .
pacs numbers : 12.38.bx 12.60.jv 14.80.ly |
one of the most important applications of the single photon source is quantum cryptography@xcite . up to now ,
most of the experiments of quantum cryptography have been using weak coherent light as photon sources @xcite . in those experiments ,
the average photon number was kept as small as 0.1 in order to keep the probability of generating two photons in one pulse very low .
however , this meant that 90 % of the pulses were vacuum states and could not be used to send any information . therefore , single photon sources which can output single photon states with a high probability while having only small percentages of two photon states are very important for quantum cryptography .
such single photon sources are also important for quantum information processing .
there has been extensive research to produce single photon pulses using a single light emitter .
experimental demonstrations have been performed using single molecules@xcite , single color centers in diamond @xcite and single quantum dots @xcite optically pumped by short pulse lasers .
to realize electrically driven solid state single photon sources , the pioneering work using turnstile devices at very low temperatures @xcite has recently been followed by a demonstration of operation at 4.2k @xcite . in those methods , however , photons are emitted into all directions and it is difficult to collect the photons with high efficiency .
this means that effectively the percentage of single photon pulses will be small and that of vacuum states will be large . to the best of our knowledge , the highest probability @xmath0 to find a single photon at an output port where single photons are collimated was less than 10% in such devices .
recently , pelton et .
al . reported an excellent experimental result using a quantum dot embedded in a semiconductor micropost microcavity , where the efficiency of emitting a single mode traveling wave inside the cavity was 38% @xcite .
however , the overall detection efficiency was only 3% due to the technical difficulties of collecting the single photons and filtering them out from background photons .
the generation of single photon states using spontaneous parametric down - conversion ( spdc ) is an alternative approach , which was first mentioned by hong and mandel @xcite .
it was theoretically followed by yuen @xcite , who studied the parametric processes with measurement feedback for the generation of near - photon - number - eigenstates . the idea was experimentally followed by several groups@xcite
. however , these experiments focused on the observation of sub - poissonian photon statistics , and not the demonstration of a true single - photon source . therefore , they just measured the correlation function @xmath1 and determined the fano factor @xmath2 .
both properties characterize the average noise level of the photon current , but the photon number distribution can not be derived from only these parameters@xcite . in this paper , we present a single photon source based on gated spontaneous parametric down - conversion ( g - spdc ) .
in contrast to earlier work@xcite , our g - spdc source is run in a new regime in which a single - photon is emitted in each time window with high probability and a well defined repetition rate .
furthermore , the quality of our single photon source is measured by characterizing the photon number distribution of its output .
a schematic of our single photon source is shown in figure 1 .
when a continuous wave ( cw ) ultra violet ( uv ) pump beam at 351.1 nm is incident to a nonlinear crystal fulfilling phase matching condition , two pairs of single visible photons , control photons and signal photons , are created at the same time through the process of spontaneous parametric down conversion ( spdc)@xcite .
we adjusted the experimental parameters to have a sufficient number of photon pairs created in a pulse with 100 @xmath3s pulse width . in our experiment ,
an average of eight control photons is detected within the pulse width .
following the first photon detection of a control photon in the pulse , the gate timing controller opens the shutter for a very short time .
this means that only the first signal photon in the pulse can pass through the shutter and be emitted . after the emission ,
the shutter is closed to block the other photons in the pulse .
the repetition of this procedure every 100 @xmath3s results in a repetition rate of 10 khz in our experiment .
we have reported that parametric fluorescence can be emitted into two small spots and that the rate of coincident counts to singles counts can be larger than 0.8 @xcite . in this experiment
, we adopted this condition to increase the collection efficiency of photons , and also focused the pump laser beam softly with a convex lens to obtain a high yield of parametric fluorescence @xcite . during the time it takes to control the shutter according to the detection signal of the control photon , which is 150 ns in our experiment
, the signal photon should be stored with minimal losses . for this purpose
, we adopted an optical delay line using a multi mode optical fiber with a transmittance of 50 % , including the losses that occurs at the input / output fiber couplers and at the uv cut filters inserted in front of the fiber coupler .
we introduced a fast optical shutter with high transmission in this experiment .
for fast optical switching , optical shutters using electro optic modulators ( eoms ) are generally used . however
, such eom modulators are polarization dependent and not appropriate for the signal photon output from the multi - mode fiber delay line in which the polarization has been randomized .
therefore , we used the optical setup for the shutter shown in figure 1 . to begin with , the first polarizing beam splitter ( pbs ) transforms the signal photon into a superposition of the states in each of the paths according to its polarization .
then the electro optic modulators ( eom ) ( lm0202 , linos ) in both paths rotate the polarization by 90 degrees only when the shutter is controlled to be closed .
finally , the two components are combined again by the second pbs .
we used two eoms for each path in order to minimize the shutter opening time and to avoid the restriction for repetition time given by the eom drivers ( liv-8 , linos ) . in the experiment , the shutter opening time was set to 50ns with a transmittance of 83 % . with this setup we achieved an overall effective transmittance of 41 % for signal photons
. in order to find the photon number distribution of the output state of our single photon source ( in multi - mode ) , we measured the number of photons in each output pulse ( 100 @xmath3s ) by a photon number analyzer consisting of a single photon counting module ( spcm - aq - fc , perkin elmer ) and a photon counter ( sr-400 , stanford research systems ) .
the real photon number distribution of the output modified by the losses caused by the limited photon detection efficiency of the photon number analyzer is as follows@xcite .
@xmath4 where @xmath5 , @xmath6 and @xmath7 are the effective quantum efficiency of the analyzer , the probability of having @xmath8 photons in the output , and the percentage of the events where the analyzer recorded @xmath9 signals during the pulse duration , respectively .
we estimated the photon number distribution @xmath6 of the output state by applying equation ( 1 ) to the experimental values of @xmath7 for @xmath10 to 2 and the estimated effective quantum efficiency of @xmath11 , taking into account the estimated quantum efficiency of @xmath12 % for spcm and experimentally measured optical losses caused by the coupling lens and a mirror ( transmittance 90.2 % ) , the filter for the stray photons ( transmittance 49.2 % ) and the fiber coupler ( transmittance 88.2 % ) .
as a result , the photon detection efficiency of the photon number analyzer is estimated to be 27.4 % .
note that we compensated the photon detection efficiency of the photon number analyzer at the output , but we did not compensate the optical losses in the experimental setup of the g - spdc single photon source . in the estimation of @xmath7 ,
the dark count of the detector of 100 cps was also subtracted from the original data .
the number of photons which are not counted during the dead time ( 50 ns ) of the photon counter was also compensated under the assumption that the spdc has poisson distribution in multi - mode@xcite .
the photon number distribution obtained in our experiment is shown by the bars ( a ) in fig .
2 . these @xmath6 are estimated using eq.[eq : p ] from the raw photon counting data @xmath13 with the measurement of @xmath14 output pulses . the probability @xmath0 to have a single photon in a pulse is @xmath15 , which is more than three times larger than the previously reported value as far as we know . the probability @xmath16 to have two photons in a pulse
was estimated to be @xmath17 .
the average photon number of the output in a pulse is @xmath18 .
the bars ( b ) show the calculated photon number distribution of weak coherent light ( wcl ) with the same average photon number .
the sub - poisson characteristics of our single photon source is clearly observed since the @xmath0 of g - spdc is higher than that of wcl and at the same time @xmath16 of g - spdc is much smaller than that of wcl .
the probability @xmath16 of two photon states indicates the security risk for quantum cryptography .
the bars ( c ) in fig .
2 show the photon number distribution of weak coherent light having a @xmath16 of 0.011 , which is the same as that of our source .
the @xmath0 of our source is 2 times larger than that of the weak coherent light .
this result also illustrates the potential advantage of our source over weak coherent light as a single photon sources for quantum cryptography .
this is our present experimental status .
next , we will analyze this photon number distribution quantitatively .
the probability @xmath0 is restricted by the optical losses in the path of signal photons in our experimental setup .
the 41 % overall transmittance of the path suggests that the collection efficiency of the parametric fluorescence by the fiber coupler is 68 % , which is consistent with the estimation by independently performed measurement of the absolute quantum efficiency of the detection setup @xcite .
ideally , we can also eliminate the probability @xmath16 to find more than two photons in a pulse by the g - spdc method .
@xmath16 is not ideal because the optical shutter used in our experiment is not perfect in two points .
one is the non - infinitesimal shutter opening time ( 50 ns ) , and the other is the small leakage of photons through the closed optical shutter ( 0.1 % ) . the following is a discussion of the possibility for applications that requires higher repetition rates .
one of the applications could be qkd .
note that long shutter opening time ( 100 @xmath3s ) in the present setup would be one of the problems that prevent an immediate adoption of our source for conventional quantum key distribution ( qkd ) systems at 1.55@xmath3 m wavelength with photon counters gated by nano - second timing pulses . in our experiment ,
the repetition rate of 10 khz was limited by a photon pair emission rate of 10@xmath19 per second with a pump power of 200 mw . with this photon pair emission rate
, we had eight detection signals of control photons per pulse on average .
we can increase the repetition rate to several tens of khz with our present setup , however , the probability to find no control photon in a pulse becomes significant when the repetition rate is increased too much . in order to overcome this limitation ,
it may be possible to increase the photon emission rate up to @xmath20 per second by using a periodically poled lithium niobate ( ppln ) crystal @xcite and an increased pump power of 1 w. this means that the repetition rate of several tens of mhz is possible using conventional technologies .
this rate can compete with most single photon sources using single light emitters .
note also that none of the single photon sources that have been reported so far can be used to replace the weak coherent light in the conventional 1.55@xmath3 m qkd system , because of available wavelength , low collection efficiencies , limited lifetimes , low operation temperatures and so on .
our g - spdc method has two significant advantages compared to other kinds of single photon sources .
first , our method operates at room temperature and is very stable .
this compares favorably with the fact that many other proposals are suffering from a short life time or require low temperature operation .
the other advantage is that the wavelength of single photons can be continuously changed .
the spdc process occurs not only for the degenerate condition but also for other pairs of wavelengths when energy and momentum conservation laws are satisfied . in our case
, we can easily change the wavelength of photons just by changing the angle of the bbo crystal .
this tunability is a unique characteristic of the g - spdc method , since methods using single light emitters have emission wavelength that are determined by the materials or by the device structure .
because we adopted a multi - mode fiber optical delay line to minimize the optical losses , the single photons are emitted in multi modes in space . with regard to the bandwidth of the emitted photons ,
we could not measure it with our present experimental setup but we guess that it should be about 0.26 nm ( full width half maximum ) .
this is because we used a narrow band filter of 0.26 nm for control photons , and the corresponding signal photons which are emitted through the shutter should have a similar bandwidth due to energy conservation . in this case
, the associated coherence time of the emitted photons would be 6.25 ps .
therefore , the photonic wave function is supposed to be localized in time in the whole time duration ( @xmath21s ) of the pulses . note that even a source which emits single photons in multi mode may be used in quantum cryptography experiments such as the one recently reported using built - in multi - mode optical fiber telecommunication network @xcite .
generation of single photons in a single mode in space and/or in a single mode in time domain remains as a future challenge@xcite .
in conclusion , we have experimentally demonstrated the generation of single photon pulses using the g - spdc method .
the measured single photon probability was found to be @xmath0 = 0.265 at the collimated output port at a repetition rate of 10 khz after the effective quantum efficiency of 27.4 % in the detection setup was compensated .
the photon number distribution of the output pulses also clearly showed sub - poissonian characteristics .
single photon sources using our method have long lifetimes at room temperatures , and can be tuned to wide range of frequencies .
we thank k. ushizaka for assistance with the gate control circuit ; k. tsujino for assistance ; y. yamamoto and e. waks for their motivating us to perform this experiment . ;
h. f. hofmann , h. ohashi , h. fujiwara , j. hotta for discussion .
this work was supported by jst - presto project , mitsubishi electric , international communications foundation and grant - in - aid for scientific research(b ) 12555008 of jsps .
e - mail address , [email protected] . c. h. bennett and g. brassard ,
quantum cryptography : public key distribution and coin tossing , in proc .
ieee international conference on computers , systems and signal processing , 175 - 179 ( ieee press , los alamitos , california , 1984 ) .
m. pelton , c. santori , j. vuckovic , b. zhang , g. s. solomon , j. plant , and y. yamamoto , efficient source of single photons : a single quantum dot in a micropost microcavity , phys .
lett . , * 89 * , 233602 1 - 4 ( 2002 ) . for example
, a single photon source with the photon number distribution @xmath22 is much more useful for quantum cryptography than that with @xmath23 , where @xmath24 is the probability to find n photons in a pulse .
those photon number distributions , however , give you the same fano factor @xmath25 .
t. yoshikawa , n. namekata , and s. inoue , quantum key distribution using a telecommunication ( multi - mode ) fiber network , in the proc . of the 7th conference on quantum information technology , ( qit comittee of ieice , tokyo , japan , 2002 ) 161 - 164 .
s. takeuchi , twin photon beams for the single photon source , in the proceedings of the 7th international symposium on foundations of quantum mechanics in the light of new technology , isqm - tokyo 01 , ( world scientific , new jersey , 2002 ) 98 - 103 .
schematic of the experimental setup . a signal photon and a control photon
are created at the same time in a bbo crystal .
the control photons are directly detected by a single photon detector while the signal photons were guided into an optical fiber delay line .
then , the signal photons were gated by the fast optical shutter .
this shutter opens only when the first detection event of control photon is observed in each pulse ( time window of 100 @xmath3s ) and blocks other photons in the same pulse .
\(a ) photon number distribution of the output state .
@xmath0 and @xmath16 shows the probability of finding one / two photons in a output pulse ( 100 @xmath3s ) respectively .
we counted the number of detection event in each pulses and estimated a photon number distribution at the output of our source as described in the text .
( b ) photon number distribution of weak coherent light with the same average photon number of ( a ) .
( c ) photon number distribution of weak coherent light with the same @xmath16 of ( a ) . | the construction of a single photon source using gated parametric fluorescence is reported with the measurement results of the photon number distribution .
a beamlike twin - photon method is used in order to achieve high collection efficiency .
the estimated probability @xmath0 to find a single photon in a collimated output pulse is 26.5 % at a repetition rate of 10 khz when the effective quantum efficiency of 27.4 % in the detection setup is compensated . |
during the last decades , experimental achievements in astrophysics provided us with new insights about our universe .
the more precise our observations have become , the more obvious also the insufficiencies of our understanding have become .
today s research in cosmology is accompanied by the group of cosmological problems , which strongly indicate that our knowledge about the universe is incomplete . in a previous work @xcite
, a framework has been introduced to include negative gravitational sources into general relativity ( gr ) . here
, it is examined how these influence the evolution of the universe , which can provide a useful basis to address some of the cosmological problems .
the proposed introduction of an anti - gravitating sector is a relaxation of the equivalence principle .
the new anti - gravitating particles are defined through their transformation behavior under general coordinate transformations , in which they differ from the usual particles .
this leads to the introduction of a modified covariant derivate , followed by a modified equation of motion .
the so defined anti - gravitating matter has a negative gravitational stress - energy , even though its kinetic energy remains positive .
this theory therefore does not suffer from instabilities through decaying vacuum fluctuations .
the gravitational energy acts as a charge , which can take either positive or negative sign . since the interaction between both types of particles
is mediated by gravity only , it is suppressed by the large value of the planck scale , and naturally very weak at the present day .
the proposal of a gravitational charge symmetry has previously been examined in refs .
furthermore , there have been various approaches @xcite to include anti - gravitating matter into quantum theories as well as into gr .
also , the topic of negative energies has recently received attention within the context of ghost condensates @xcite .
this paper is organized as follows : in the next section , the properties of the anti - gravitating matter are briefly recalled . in section [ hub ] , we solve the geodesic equations for both types of particles in the friedmann - robertson - walker background , and derive hubble s law . in section [ evolution ]
we examine the evolution of the universe . after a brief discussion in section [ dis ] ,
we conclude in section [ concl ] . throughout this paper
we use the convention @xmath0 .
the signature of the metric is @xmath1 .
small greek indices @xmath2 are space - time indices .
the unitary representations @xmath3 of a gauge group @xmath4 define the transformation behavior of particle fields @xmath5 . for two elements of the group @xmath6 , the representation fulfils u(g ) u(g ) = u(gg ) [ multrep ] , and the field transforms as @xmath7 . from this
, it is always possible to construct a second representation , defined by ( g ) = ( u(g^-1))^t , which belongs to the charge - conjugated particle .
the anti - particle @xmath8 transforms according to the contragredient representation , @xmath9 , which is @xmath10 . in case of a local symmetry ,
these transformations lead to the introduction of gauge - covariant derivatives in the usual way .
suitable combinations of particles with anti - particles allow to construct gauge - invariant lagrangians . from the above
, one is tempted to conclude that there is no charge - conjugation for gravity .
if the gauge - group is the lorentz - group @xmath11 , then the elements @xmath12 fulfill @xmath13 , which means that in this case the second representation @xmath14 is equivalent to @xmath3
. however , this does not apply when the field transforms under a general coordinate transformations @xmath4 .
let @xmath5 be a vector field and an element of the tangent space @xmath15 . under a general coordinate transformation @xmath4
, the field and its conjugate behaves as tm & : & = g , : = g^-1 , [ stdtrafo ] where @xmath16 is the dual to @xmath15 .
the equivalence principle requires that the fields in the tangential space transform like in special relativity .
i.e. if @xmath4 is an element of the lorentz - group , the fields have to transform like lorentz - vectors
. however , the generalization to a general coordinate transformation is not unique . instead of eq .
( [ stdtrafo ] ) one could have chosen the field to transform according to :
= ( g^t)^-1 , ^ * & : & = g^t .
[ trafo ] here , the space @xmath17 is a vector - space which spans the basis for these fields , and @xmath18 is its dual . in case @xmath4
was an element of the lorentz - group , i.e. @xmath19 , both representations ( [ stdtrafo],[trafo ] ) agree . for general coordinate transformations that will not be the case .
indeed , one sees in particular that the newly introduced fields will have a modified scaling behavior . for the following ,
it is convenient to introduce a map @xmath20 which , in the vector - representation , is a vector - space isomorphism from @xmath15 to @xmath17 . for the map @xmath21 to transform adequately , @xmath22 , one finds the behavior
= ( g^t)^-1 g^-1 [ trafotau ]
. it will be useful to clarify the emerging picture of space - time properties by looking at a contravariant vector - field @xmath23 as depicted in figure [ fig1 ] .
this field is a cut in the tangent bundle , that is the set of tangent spaces @xmath15 at every point of the manifold which describes our space - time .
the field is mapped to its covariant field , @xmath24 , a cut in the co - tangent bundle , @xmath16 , by the metric tensor @xmath25 . the newly introduced field @xmath26 ( from here on named anti - gravitating ) transforms in the tangential space like a lorentz - vector in special relativity .
but it differs in its behavior under general coordinate transformations . in order not to spoil the advantages of the ricci - calculus
, it will be useful to introduce a basis for the new fields that transforms accordingly .
locally , the new field can be expanded in this basis @xmath27 .
these basis elements form again a bundle on the manifold , that is denoted with @xmath17 . to each of the elements of @xmath17
also a dual space exists , defined as the space of all linear maps on @xmath17 .
this space is denoted by @xmath18 and its basis as @xmath28 .
the map from @xmath17 to @xmath18 will be denoted @xmath29 , and defines a scalar product on @xmath17 .
the relation between the introduced quantities is summarized in figure [ fig1 ] .
note , that the underlined indices on these quantities do not refer to the coordinates of the manifold but to the local basis in the tangential spaces .
all of these fields still are functions of the space - time coordinates @xmath30 .
also , the map @xmath31 , is not a metric on the manifold , and does not measure physical distances and angles .
instead , it is a scalar product on the underlined spaces .
the transformation behavior in eq .
( [ trafotau ] ) together with the observation that in a local orthonormal basis both fields transform identical under local lorentz - transformations , gives us an explicit way to construct @xmath20 .
we choose a local orthonormal basis @xmath32 in @xmath15 , which is related to the coordinate basis by the locally linear map @xmath33 . in this basis ,
the metric is just @xmath34 and @xmath35 is just the identity .
one then finds @xmath20 in a general coordinate system by applying eq.([trafotau ] ) = ( e^t)^-1 e^-1 = ( e e^t)^-1 [ taue ] .
an example for this is given in section [ example ] .
the properties of the vector - fields are transferred directly to those of fermionic fields by using the fermionic representation and transformations . in this case , the map @xmath36 , instead of the metric , is used to relate a particle to the particle transforming under the contragredient representation , and the map relating the spinor - bundles is denoted with @xmath37 ( spinor indices are suppressed ) , see fig . [ fig2 ] . with this notation
it is @xmath38 .
it is now straightforward to introduce a covariant derivative for the new fields , in much the same way as one usually introduces the derivative for the charge - conjugated particles . we will use the notation @xmath39 for the general - covariant derivative and @xmath40 for the general covariant derivative including the gauge - derivative of the fields .
it is understood that the form of the derivative is defined by the field it acts on , even though this will not be noted explicitly . for an element @xmath41 of @xmath16 , besides the usual derivative @xmath42 , one can now apply the composite derivative @xmath43 , or , using a more abstract notation @xmath44 .
similar considerations apply for elements of products of @xmath16 and @xmath15 , or spinor bundles , with appropriate applications of the map @xmath20 , or @xmath37 respectively .
one introduces the derivative in the direction of @xmath45 on the basis in a general way by _ _ & = & ^ _ _ , _ ^ _ ^ _ _ = ( _ ^ ^ _ ^ _ ) _ .
[ stdder ] note that the connection symbols do not transform homogeneously in the first and third index and therefore these indices can not be contracted with the @xmath20 s .
alternatively , one can define the connection coefficients with respect to @xmath44 ^ _ = _ ^ ^ _ ^ _ .
[ ugamma ] the operators @xmath44 , and @xmath46 fulfill the same commutation relations as @xmath39 and @xmath47 , which means in particular that @xmath48 is symmetric in the lower two indices . to get an explicit formula for these christoffelsymbols
one commonly uses the requirement that the covariant derivative on the metric itself vanishes @xmath49 which assures that the scalar product is covariantly conserved .
similarly , we require @xmath50
. from this one finds @xcite the explicit expression ^ _ & = & ^ ( _ ^_^ _ g _ + _ ^ _ ^ _ g _ - _ ^ _ ^ _ g _ ) [ chrisagu ] , or , with eq.([ugamma ] ) ^ _ & = & ^ ( _ g _ + _ ^ ^ _ _ g _ - _ ^ ^ _ _ g _ ) [ chrisag ] .
the use of the tetrad @xmath51 , together with eq.([taue ] ) , the definition @xmath52 , and the above eq.([chrisag ] ) provides an explicit construction for the dynamics of the new fields . with these generalized covariant derivatives ,
one obtains the lagrangian of the anti - gravitational field by replacing all quantities with the corresponding anti - gravitational quantities and using the appropriate derivative for the new fields to assure homogenous transformation behavior .
one then finds additional possibilities to construct gauge- and lorentz - invariant kinetic terms in the lagrangian .
e.g. for fermionic fields @xmath5 , besides the usual @xmath53 one can now also have @xmath54 , which is not identical to the first because the covariant derivative on @xmath37 is ( in general ) non - vanishing . in case one
considers operators with higher order derivatives , more choices become possible ( see also section [ early ] ) .
it is instructive to look at the motion of a classical test particle by considering the analogue of parallel - transporting the tangent vector .
the particle s world line is denoted @xmath55 , and the anti - gravitating particle s world line is denoted @xmath56 does only indicate that the curve belongs to the anti - gravitating particle ; it is not related to the curve of the usual particle @xmath57 . ] .
in contrast to the gravitating particle , the anti - gravitating particle parallel transports not its tangent vector @xmath58 , but instead the related quantity in @xmath59 , which corresponds to the kinetic momentum , and is @xmath60 . on the particle s world line @xmath56
, it is @xmath61 which is covariantly conserved .
parallel transporting is then expressed in evaluating the derivative in direction of the curve and set it to zero . for the usual geodesic , which parallel transports the tangential vector , one has @xmath62 , whereas for the anti - gravitating particle one has ^
^ _ _ _ ^ t^= 0 , [ agpt ] which agrees with the usual equation if and only if the covariant derivative on @xmath63 vanishes and @xmath64 are equal up to a mere re - parametrization @xmath65 of the curve . in such a case
@xmath66 which could only corresponds to @xmath67 being some scalar function times the identity .
such a scalar however , is fixed by the requirement that in a local orthonormal basis @xmath20 is the identity , and therefore does not cause a reparemetrization .
we will also see later that @xmath20 is not in general proportional to the identity matrix . ] .
it is important to note that the tangent vector @xmath68 is not parallel transported along the curve given by the new eq .
( [ agpt ] ) . using the covariant derivative @xmath69 , one obtains ( _ + ^ _ ) = 0 , and by rewriting @xmath70 one finds the anti - geodesic equation @xmath71 this equation should be read as an equation for the quantity @xmath61 rather than an equation for the curve . to obtain the curve ,
one proceeds as follows * integrate eq .
( [ antigeo ] ) once to obtain @xmath61 , * translate this into the geometric tangential vector @xmath72 , * integrate a second time with appropriate initial conditions to obtain @xmath73 . it is important to note that the equations of motion eq.([antigeo ] ) are invariant under general diffeomorphism , provided that the quantities are transformed appropriately . for a particle moving in a curved spacetime ,
it is possible to choose a freely falling coordinate system , in which the usual christoffelsymbols vanish .
however , this freely falling frame for the particle will in general not also be a freely falling frame for the anti - gravitational particle .
consequently , the christoffelsymbols in eq.([chrisag ] ) will in general not also vanish in the freely falling frame of the usually gravitating particle .
both sets of symbols therefore will in general not be proportional to each other .
it is then also @xmath74 , and @xmath75 i d . in these coordinates , the derivative on @xmath20 vanishes and
the geodesic and the anti - geodesic curve agree .
since both curves are invariant under coordinate transformations , they will agree for every choice of coordinates in a globally flat spacetime . ] .
this explains how it is possible to have a relaxation of the equivalence principle .
locally , both particles experience an gravitational downwards pull like an acceleration in flat space . for the usual particle ,
inertial mass equals gravitational mass , and upward acceleration corresponds to a downward attractive gravitational pull . for the anti - gravitating particle ,
the inertial mass is the negative of the gravitational mass , and upward acceleration corresponds to a downward repulsive gravitational push .
the first is realized for particles obeying the usual transformation properties , the second for particles obeying the transformation properties eq.([trafo ] ) . a scenario with negative gravitational masses
is commonly believed to lead to inconsistencies .
when neglecting an appropriate covariant derivative , a particle with negative gravitational mass must move on a geodesic in the field of a positively gravitating mass , since the curve is independent on the particle s mass . on the other hand ,
a particle with positive gravitational mass in the field of a negatively gravitating body is repelled , as one most easily concludes using the newtonian limit .
taken together , the positive mass will attract the negative one , whereas the negative mass will repel the positive one .
this - in the limit of two point particles - leads to a self - accelerating system .
such considerations go back to @xcite , and can be resolved by noting that the curve of the negatively gravitating mass in the positively gravitating background is not a geodesic . in this case , like charges attract , unlike charges repel , and a self - acceleration can not occur .
even though both particles on their own can not distinguish between acceleration in flat space and local gravitational effects , a pair of both can .
in flat space , accelerating a pair of gravitating and anti - gravitating masses will not allow to distinguish between both as only the inertial masses are involved . in a gravitational field however , the one will move in the opposite direction from the other .
therefore , the equivalence principle is not generalized , but relaxed : it holds for both particles separately , but not for both together when they can compare their ratios of inertial to gravitational mass .
one finds the gravitational stress - energy tensors ( sets ) for the fields from variation of the action with respect to the metric s & = & ^d+4x [ full ] ,
where @xmath76 , @xmath77 is the curvature scalar , @xmath78 is the lagrangian of usual fields , and @xmath79 is the lagrangian of the new fields .
the source terms then take the form @xcite t^ & = & - g^ l , + t^ & = & - ^ _ ^ _ ( + g^ )
[ sign ] . under a perturbation of the metric , the anti - gravitational fields will undergo a transformation exactly opposite to these of the normal fields as one expects by construction .
the @xmath20-functions convert the indices and the transformation behavior from @xmath17 to the usual tangential space .
the lagrangian eq.([full ] ) consists of the gravitational contribution and two types of matter . in the following it
is assumed that the dominant type of matter in our universe is the positively gravitating one .
we will not consider the question of a transition of one dominance to the other . as will become apparent later
, this can not occur in the here examined case .
if one considers the lagrangian of the dirac - field one has to formulate the action in form of the tetrad fields .
the above used argument then directly transfers to the dirac field through the properties of the anti - gravitational field under diffeomorphisms .
for the fermions , the set then can be simplified inserting that the field fulfills the equations of motion @xmath80 .
most importantly , one sees that the set of the anti - gravitating field yields a contribution to the source of the field equations with a minus sign ( and thus justifies the name anti - gravitation ) .
this is due to the modified transformation behavior of the field components .
however , we also see that the second term , arising from the variation of the metric determinant , does not change sign .
fermionic matter will therefore display different properties than radiation .
this will turn out to be an important ingredient to the evolution of these fields in an expanding universe .
the corresponding conservation law of the derived source terms which follows from the bianchi - identities is as usual @xmath81 .
it is crucial to note that the kinetic set as defined from the noether current does not have a change in sign . here
, no variation of the metric is involved and the gravitational properties of the fields do not play a role . to clearly distinguish this canonical set from the gravitational source term , the canonical set is denoted with @xmath82 , whereas the above used @xmath83 is kept for the gravitational set .
the canonical set for a matter lagrangian @xmath84 from noether s theorem is ^ _ = _ - ^ _ l , and is covariantly conserved @xmath85 . correspondingly , one finds the conserved current for the anti - gravitational field ^
_ = ^ _ ( _ - ^ _ ) , which is also covariantly conserved @xmath86 . however , in general the quantities derived from noether s theorem are neither symmetric , nor are they traceless or gauge - invariant . for gr ,
belinfante s symmetrized tensor is the more adequate one @xcite which we will also use in the following . from the noether current , one gets a total conserved quantity for each space - time direction @xmath87 , which obey the conservation equations ^ _ + _ ^ ^ _ = 0 .
[ cons2 ] the form of the second term of eq .
( [ cons2 ] ) is readily interpreted : when the anti - gravitating particle gains kinetic energy on a world line , the gravitational particle would loose energy when traveling on the same world line .
the interaction with the gravitational field is inverted .
one thus can identify the anti - gravitating particle as a particle whose kinetic momentum vector transforms under general diffeomorphism according to eq.([trafo ] ) , whereas the standard particle s kinetic momentum transforms according to eq.([stdtrafo ] ) . in the present low energy epoch of the universes evolution ,
the interaction between gravitating and anti - gravitating matter is very weak .
since both types of matter repel , it is natural to expect that we live today in a sector of the universe with predominantly one type of matter . in this case
, one can neglect the presence of anti - gravitating matter for most purposes and the here proposed framework reduces to the standard gr .
however , the presence of anti - gravitating matter might be relevant at large distances , at high densities , and in strongly curved backgrounds .
to give an example , let us consider the familiar case of a particle moving in a schwarzschild - metric with the line - element ds^2 = - dt^2 + dr^2 + r^2 ( d^2 + ^2 d ^2 ) , with @xmath88 .
going to a local orthonormal basis yields e^i _ = ( e^i _ ) ^t = diag(,,1/r,1/(r ) ) , and so ^ _ & = & diag ( , , 1/r^2,1/(r^2 ^2 ) ) , [ tausm ] + _ & = & diag ( 1,1,1,1 ) , + g _ & = & diag ( - , , 1/r^2,1/(r^2 ^2 ) ) .
we will consider a radially moving particle with @xmath89 .
since @xmath20 is diagonal , we also have @xmath90 . assuming a small velocity of the test - particle @xmath91 , and @xmath92 , one approximates the geodesic equation as usual to t^r - ^t _ t ^r _
t t , where a dot denotes the derivative with respect to @xmath93 . with eq.([chrisag ] ) one computes the christoffelsymbols of the anti - gravitating field ( see appendix a ) . using these , one finds to first order in the newtonian limit the expression ^r - _
r g_t t = + ( r^-3 ) , integrating once and converting the index results in the approximation r & & - + ( r^-2 ) + const .
, + r & & + ( r^-3 ) .
one sees that indeed the anti - gravitating particle is repelled by the gravitational mass of the background field .
we live today in a galaxy in which one gravitational type of matter is dominant .
this local dominance , however , must not be identical to the global dominance . though we have defined the globally dominating matter to be the positive one , it is not _ a priori _ clear whether our galaxy is of the same type of matter .
it is in principle possible that we are made of anti - gravitating matter , which , in a local surrounding of the same matter type , would be indistinguishable from being positively gravitational matter in a local surrounding of positively gravitating matter . whether we are of the globally dominating matter type or not , one would expect this to reflect in cosmological observations . from analyzing the equations of motions of test particles as well as those of the cosmic fluid , we will in the following
see that we are indeed of the dominating positively gravitating type of matter .
starting point is the usual assumption of an isotropic and homogeneous universe which leads to the general line - element of the friedmann - robertson - walker metric : ds^2 = -dt^2 + a^2(t ) .
[ frw ] the christoffel symbols of this geometry for both types of particles are given in appendix b. going to a local orthonormal basis one finds e^i _ = ( e^i _ ) ^t = diag(1 , , , ) , and so ^ _ & = & diag ( 1 , , , ) , [ taufrw ] + _ & = & diag ( 1,1,1,1 ) , + g _ & = & diag ( - 1 , , , )
. we will now examine the motion of the anti - gravitating photon .
for a clear analysis , this is done parallel to the standard case .
we will denote the momentum of the gravitating particle with @xmath94 and that of the anti - gravitating with @xmath95 , in accordance with the notation of the previous section .
the curves differ in the fact that the one transports the usual tangential vector , whereas the other transports the vector which is subject to the new transformation behavior . by assuming a radial motion one can simplify the equations of motion with @xmath96 . with use of appendix b , the geodesic equations read t^t&= & - a a ( t^r ) ^2 [ stdgeo1 ] + t^r & = & - 2 t^r t^t , [ stdgeo2 ] whereas the anti - geodesic equations take the form ^t & = & ( t^r ) ^2 [ newgeo1 ] + ^r & = & 2 ^r t^t [ newgeo2 ] .
one solves eqs.([stdgeo1]),([stdgeo2 ] ) with : = t^t = , t^r = , where @xmath97 is some constant initial value . to solve the equations of motion of the anti - gravitating particle
, one makes an _ ansatz _ as powerlaw for the kinetic energy @xmath98 , which yields in eq.([newgeo1 ] ) ^r = c_i a^n+1 , where we have used that @xmath99 from eq.([taufrw ] ) . plugging this into eq.([newgeo2 ] ) one obtains @xmath100 and therefore has the solution , and so @xmath101 , which assures that @xmath102 is a constant of motion . ] = c_i a(t ) , ^r = c_i a(t)^2 .
[ antiomega ] from this one defines the red shifts for both types of photons in the usual way as z = -1 , z = -1 .
now let us examine hubble s law .
we make the expansions & = & 1 - h_0 _ t - q_0 h_0 ^ 2 _ t^2 + o(_t^3 ) + & = & 1 + h_0 _ t - h_0 ^ 2 _ t^2 ( 1 + ) + o(_t^3 ) , + with @xmath103 and the standard definitions h_0 = , q_0 = - one then finds to first order z = h_0 _
t , = - h_0 _ t .
[ hubble ] we see that the redshift goes into a blueshift and vice versa .
since we observe a redshift of the light from far galaxies , two conclusions are possible : 1 .
we are standard matter , then the universe is currently expanding with @xmath104 .
2 . or we are anti - gravitating matter , then universe is currently shrinking @xmath105 .
to draw further conclusions , it is necessary to examine the evolution of the universe .
we will formulate everything in the kinetic quantities , @xmath106 for the usual particles , and @xmath107 for the anti - gravitating particles .
these quantities are positive in the common way .
we define them in the comoving restframe and have as usual ^ & = & diag ( , , , ) + ^ & = & diag ( , , , ) [ utheta ] , and ^ & = & ^ + t^ & = & diag ( , , , ) [ ut ] , where the gravitational energy - density @xmath108 and the gravitational pressure @xmath109 fulfill an unknown equation of state .
the set of usual fermionic matter ( denoted with the upper index m ) is identical to the kinetic set ( t^m)^ = ( ^m)^ = diag(^m , 0 , 0 , 0 ) , whereas for the anti - gravitating matter one has according to eq .
( [ sign ] ) ( ^m)^ & = & diag ( ^m , 0 , 0 , 0 ) , + ( t^m)^ & = & - diag ( ^m , 0 , 0 , 0 ) . for radiation ( denoted with the upper index r ) ,
we use belinfante s symmetrized kinetic set ( ^r ) ^ = f^f_^ + g^ f^f _ , which is traceless @xmath110 , and one concludes that also the anti - gravitating matter has the ordinary equation of state for the kinetic quantities = 3 p .
the anti - gravitating matter therefore does not behave like a quintessence field .
further , it is insightful to examine the @xmath93-component of the conservation law for the kinetic sets eq.([cons2 ] ) _ ^t + ^t _ _ ^ = 0 .
[ const ] the first term yields the usual contribution _
t + 3 ( + p ) . since @xmath20 is diagonal and @xmath111 , it is furthermore sufficient to evaluate the expression _
^t = _ t ^tt + ^t _ ^ + ^ _ ^t .
[ constonlynew ] from eq.([utheta ] ) with eq.([taufrw ] ) one has ^ = _ ^
^ = diag(,p , p , p ) . inserting the christoffelsymbols from appendix b
, one finds that the second term in eq.([constonlynew ] ) changes sign , and eq.([const ] ) takes the form _
t ( + ) + 3 ( + + p - p ) = 0 .
[ consfrw ] let us first examine the case of pure matter fields .
then , the above equation simplifies to _ t ( ^m + ^m ) + 3 ( ^m + ^m ) = 0 .
[ consfrwm ] but one also has the bianchi identities 0 & = & _ ( ( ^m)^t - ( ^m)^t ) + & = & _ t ( ^m - ^m ) + 3 ( ^m - ^m ) . these together with eq.([consfrwm ] ) imply that both densities are separately conserved with the usual conservation law 0 & = & _ t ^m + 3 ^m + 0 & = & _ t ^m + 3 ^m .
this means in particular that the energy density of anti - gravitating matter @xmath112 also dilutes with the volume @xmath113 . for radiation , the situation is slightly more complicated .
examination of eq.([consfrw ] ) for purely anti - gravitating radiation , @xmath114 , yields 0 = _
t ^r + 2 ^r .
this is solved when @xmath115 dilutes with @xmath116 .
this fits very nicely with the result of the previous section that the energy of anti - gravitating radiation experiences a blue shift instead of a red shift .
one then expects the energy density to scale like @xmath117 .
one can thus draw the important conclusion that the kinetic energy density of anti - gravitating radiation dilutes _ slower _ than those of anti - gravitating matter .
this scaling behavior can not directly be used for the cosmological evolution , since the quantities that enter the friedmann - equations are the gravitational ones @xmath118 , whose properties remain to be investigated .
this however , allows us to discard the second possibility of the previous section .
apparently , the matter we consist of is not radiation dominated . in general , one will be faced with a mixture of gravitating and anti - gravitating components . to approach the problem ,
note that from eq.([sign ] ) one has the relation ( t)^ _ + ( ) ^ _
= ^ _ , or & = & - + + & = & - p - one can further use _
t + 3 ( + ) = 0 . and
rewrite this into _
t + 3 ( + p ) = _ t .
in particular , for radiation one finds from this _
t ^r = 2 ^r , and thus @xmath119 , and further @xmath120 .
therefore , one can conclude that also the gravitational energy of the anti - gravitating radiation scales like @xmath121 and behaves like a curvature component .
it consequently enters the friedmann - equations as a curvature term does and underlies similar experimental constraints . inserting these results
one obtains the modified friedmann - equations ( ) & = & - ( + 3 p + - 3 p ) [ 1 ] + ( ) ^2 & = & ( - 2 ) .
to investigate the processes in the early universe , one has to keep in mind that the lagrangian eq.([full ] ) should be interpreted as low - energy approximation of some underlying theory . at planckian temperatures , higher order terms can become important , which will provide an interaction between gravitating and anti - gravitating matter fields .
the strength of this interactions will typically grow with a power of the temperature over planck mass @xmath122 .
such higher order interaction terms might e.g. take the form ~ /d _ s^-1 /d .
however , the structure of these higher order terms , as well as the particle content of the full theory are unknown .
therefore , we will aim to discuss the consequences on a general thermodynamical level by adding an exchange term @xmath123 for the kinetic energies , which becomes relevant at planckian densities , and leaves the total kinetic energy conserved .
for such a coupled gravitating and anti - gravitating relativistic fluid , the conservation equations can then be written in the form _
t + 4 & = & q ( , ) , [ consq1 ] + _ t + 2 & = & - q(,)[consq2 ] .
the easiest choice for the exchange term is q ( , ) = , with some dimensionless constant @xmath124 of order one .
when the energy density of usual matter dominates that of the anti - gravitating matter , then the interaction term should diminish it during a contraction with @xmath125 , and so it is sign@xmath126sign@xmath127 . with this ,
( [ consq1],[consq2 ] ) can be formally recast by attributing a time - dependent equation of state to the fields with w & = & - [ w1 ] + w & = & - + [ w2 ] the friedmann equations then read ( compare to eqs.([1],[2 ] ) ) ( ) ^2 & = & ( - 2 ) + ( ) & = & - ( - 3 q ( , ) ) .
these equations can be investigated qualitatively .
suppose an initial density distribution with @xmath128 , and @xmath129 . in case
the interaction term is negligible , @xmath130 will drop faster than @xmath131 and the expansion has a bounce when @xmath132 is reached . in case the interaction term is not initially negligible , one sees from eq.([w2 ] ) that @xmath131 always dilutes , and so will the additional term in eq.([w1 ] ) .
therefore , both correction terms in eq.([w1],[w2 ] ) can eventually be neglected , and the previous case applies .
the anti - gravitating fluid acts then essentially like a positive curvature term , and a turning point is reached , at which the expansion stops . from thereon
, the universe goes into a contraction phase with @xmath125 .
@xmath130 increases again , and does so faster than @xmath131 .
both densities pass the initial configuration with a time - reversed @xmath133 .
when @xmath130 comes close to planckian densities , one sees from eq.([w2 ] ) that its presence will increase the @xmath134 of the anti - gravitating matter , and @xmath131 will eventually increase faster than @xmath130 . when @xmath135 also reaches planckian densities , it will furthermore slow the increase of @xmath130 .
that is , both densities will again approach each other .
eq.([consq1 ] ) shows that the interaction will re - distribute kinetic energy of the gravitational field into the anti - gravitating matter , thereby increase @xmath136 and decrease @xmath137 .
however , unlike the usual scenarios , the energy of the anti - gravitating field makes a negative gravitational contribution .
therefore , @xmath138 can go from a positive to a negative values when @xmath131 has increased to = .
further approach of both densities results in reaching another turning point at planckian densities , with @xmath132 . from then on ,
the universe goes again into an expanding phase with positive acceleration , until it reaches the initial configuration we started from ( which might have had either positive or negative @xmath138 , depending on the initial values ) .
taken together , the evolution is a periodic cycle .
the maximal / minimal extension that can be reached in this cycle depends on the initial values of @xmath139 and the corresponding @xmath140 .
a special case occurs for the initial values @xmath141 , for which all quantities remain constant .
figure [ fig3 ] shows some numerical results of the integration of these equations , with initial values @xmath142 of order @xmath143 , @xmath144 and @xmath145 , which confirm this qualitative investigation .
one sees that for typical initial values with @xmath146 of order @xmath147 , the oscillation period is of order a few planck times .
the larger the fraction of usual matter , the longer it will take until the anti - gravitational matter becomes important and the returning point it reached .
the fact that anti - gravitating radiation is the slowest to dilute , and that it should become important only in the late stages of the universes expansion , requires that its initial density is much smaller than that of usual matter .
this problem is similar to the coincidence problem in the context of a cosmological constant .
however , an important fact to keep in mind is that an initially homogeneous distribution of the gravitating and anti - gravitating fluid is unstable under perturbations , since the two kinds of matter want to separate .
small overdensities around the planckian density of the discussed type will not lead to large deviations from the initial configuration , and the system will attempt to wash out perturbations through the strong interaction between both fluids .
large perturbations however , can lead to a drop of temperature and reaction rate , that lasts too long to enable return to the initial configuration .
both types of matter can then separate into spatially distinct regions which predominantly consist of one type of matter .
it is therefore natural to identify the universe we currently experience as one such region with a dominantly positively gravitating matter content .
thus , the scenario investigated in the previous section applies for small perturbations in an initially planckian density . for large perturbations ,
the evolution will attempt to break out of the oscillation at a maximum value of @xmath148 . to incorporate the two kinds of matter into the description one can treat the gravitating and anti - gravitating matter as ( spatially separated ) homogenous distributions whose densities are time dependent .
the anti - gravitating matter will be repelled from a positively gravitating fluid and be attracted by the anti - gravitating one , and vice versa .
though the overall energy density is covariantly conserved , energy flux will result in additional local source terms for both fluids , arising from the flux through the boundary .
most importantly , the kinetic energy attributed to our universe is not necessarily conserved any longer . in the process of repelling the anti - gravitational matter and attracting the usual matter , the universe experiences an increase of the usual energy density .
in particular , this increase will also increase the acceleration relative to the standard cosmological model .
such a scenario can most intuitively be realized in a higher dimensional space time , as e.g. examined in @xcite . in this case
, our universe can be described as a @xmath149 dimensional submanifold with dominantly gravitating matter , beneath which dominantly anti - gravitating sub - manifolds might be present . the scale parameter @xmath148 will not only be a function of @xmath93 but also a function of the coordinates in the extra dimensions ( in the simplest case one extra dimension ) . in the limit of infinitesimally thin branes
, @xmath148 has to fulfill the common jump conditions associated with strongly localized layers of charges
. it would be interesting to further investigate the detailed properties of such a scenario .
one should also note that the content of anti - gravitating matter in our universe will never vanish completely . even though the interaction between both types of matter is suppressed by the planck scale
, it will always take place .
we have studied the influence of anti - gravitating fields on the cosmological evolution .
an examination of the geodesic equation for the anti - gravitating photon in a friedmann - robertson - walker background showed that this particle experiences a cosmological blue shift instead of a red shift .
furthermore , we have derived the modified evolution equation of the universe .
these have additional source terms arising from the anti - gravitating fields .
together with the stress - energy conservation , these equations determine the evolution of the universe .
we have shown that the anti - gravitating matter density dilutes with the volume of the universe @xmath113 and thus , like usually gravitating matter .
the anti - gravitation radiation , however , dilutes due to its blueshift with @xmath116 .
in contrast to quintessence or a cosmological constant , it was shown that the anti - gravitating matter has a usual thermodynamical equation of state .
we have examined the evolution in the early universe with a strong interaction between both kinds of matter at planckian densities .
it was shown how such a scenario leads to a periodic cycle which includes a phase of accelerated expansion .
we also briefly discussed an extension of the scenario , in which both types of matter separate on submanifolds in extra dimensions .
this work was supported by the dfg and by the department of energy under contract de - fg02 - 91er40618 .
i thank the perimeter institute for kind hospitality .
the christoffelsymbols for the anti - gravitating particle in a schwarzschild - metric can be computed using eq.([chrisag ] ) .
the coordinates are @xmath150 , and we use the notation @xmath88 . ^ t _ t r & = & - , ^ t _ r t = - , ^ r _ t t = - + ^ r _ r r & = & , ^ r _ = , ^ r _ = + ^ _ r & = & - , ^ _ r = - r , ^ _ = + ^ _ r & = & - , ^ _ = - , ^ _ r = - r ^2 + ^ _ & = & - .
all other entries are zero .
^t_rr & = & a a , ^t _ = a a r^2 , ^t _ = a a r^2 ^2 + ^r_tr & = & ^r_rt = ^_t = ^ _ = ^_t = ^_t = + ^r _ & = & -r , ^r _
= -r ^2 + ^_r & = & ^_r = ^_r = ^_r = + ^ _ & = & - , ^ _ = ^ _ = [ chris ] and ^ t _ r r & = & - , ^ t _ = - , ^ t _ = - + ^ r _ t r & = & - , ^ r _ r t = - a a , ^ r _
= + ^ r _ & = & , ^ _ t = - , ^ _ r = - + ^ _ t & = & - a r^2 a , ^ _ r = - r , ^ _
= + ^ _ r & = & - , ^ _ = - , ^ _ t = - + ^ _ t & = & - a a r^2 ^2 , ^ _ r = - r ^2,^ _ = - where @xmath151 .
all other entries are zero .
p. c. w. davies and a. c. ottewill , phys .
d * 65 * , 104014 ( 2002 ) [ arxiv : gr - qc/0203003 ] .
s. ray and s. bhadra , int .
j. mod .
d * 13 * , 555 ( 2004 ) [ arxiv : gr - qc/0212120 ] .
d. e. rosenberg , [ arxiv : astro - ph/0008166 ] .
v. m. zhuravlev , d. a. kornilov and e. p. savelova , gen .
* 36 * ( 2004 ) 1719 .
v. faraoni , phys .
d * 70 * , 081501 ( 2004 ) [ arxiv : gr - qc/0408073 ] . f. henry - couannier , int .
j. mod .
a * 20 * , 2341 ( 2005 ) [ arxiv : gr - qc/0410055 ] . c. nickner , [ arxiv : gr - qc/0603005 ] .
n. arkani - hamed , p. creminelli , s. mukohyama and m. zaldarriaga , jcap * 0404 * , 001 ( 2004 ) [ arxiv : hep - th/0312100 ] . n. arkani - hamed , h. c. cheng , m. a. luty and s. mukohyama , jhep * 0405 * , 074 ( 2004 ) [ arxiv : hep - th/0312099 ] . n. arkani - hamed , h. c. cheng , m. a. luty , s. mukohyama and t. wiseman , [ arxiv : hep - ph/0507120 ] .
r. b. mann and j. j. oh , [ arxiv : hep - th/0504172 ] .
a. krause and s. p. ng , [ arxiv : hep - th/0409241 ] . | the dynamics of a universe with an anti - gravitating contribution to the matter content is examined .
the modified friedmann equations are derived , and it is shown that anti - gravitating radiation is the slowest component to dilute when the universe expands . assuming an interaction between both kinds of matter which becomes important at planckian densities , it is found that the universe undergoes a periodic cycle of contraction and expansion .
furthermore , the possibility of energy loss in our universe through separation of both types of matter is discussed . |
understanding when and how different types of galaxies formed in the universe is an age old problem . in recent years the hierarchical merging scenario @xcite , where large galaxies are built up over time through mergers of smaller objects , has been the leading theory for the mechanism by which structure formed in the universe .
large galaxy surveys have been undertaken to provide observational evidence for how galaxies formed . from their observations , we are compiling a snapshot of galaxies througout time .
at z@xmath63 we find both massive starbursting galaxies in highly clustered regions @xcite and actively star forming galaxies elsewhere @xcite . at z@xmath62
we see galaxies that are still forming stars , but have redder colors and larger masses from continued star formation since z@xmath63 @xcite . at z@xmath61 - 2 a population of extremely red objects ( eros )
are seen by near infrared surveys such as fires @xcite , k20 @xcite , and munics @xcite , many of which appear to contain older stellar populations that have experienced passive evolution @xcite . at z@xmath61
we see a decrease in global star formation rates @xcite , an increase of stellar mass densities in early - types @xcite , and a change in galaxy morphologies to more ordered systems @xcite . at z@xmath71
we see early - type galaxies that appear to have formed at [email protected] - 2 in the field @xcite and at z@xmath62 - 3 in clusters @xcite , possibly from those that were seen as star forming galaxies at z@xmath62 and starbursts at z@xmath62 - 3 .
this observed galaxy demography , however , is not matched by the semi - analytic hierarchical galaxy formation models .
although the models do well on predicting the global star formation rate and stellar mass densities , they do not match observations on the correct proportions of galaxy types as a function of redshift @xcite .
in particular , not enough eros or submm galaxies are formed at z@xmath61 - 3 in the models .
we do not yet understand the details of how galaxies assembled their mass , although the epoch of z@xmath61 - 2 continues to stand out as an important transitional period .
detailed studies of individual galaxies around this epoch will uncover essential clues about how they assembled into the distribution of galaxies that we see today .
one such detail for study is determining ages or formation redshifts of galaxies in this epoch .
various techniques have been used in the past to determine an overall age from integrated galaxy light , including using colors to estimate age from the amounts of old ( red ) and young ( blue ) stars , using the equivalent widths ( ews ) of specific spectral lines or breaks that are known to be sensitive to stellar age @xcite , matching galaxy spectra to stellar spectra @xcite , using fits of stellar population synthesis models to spectral energy distributions ( seds ) obtained from broadband or narrowband photometry @xcite , and fitting galaxy spectra with model spectra @xcite .
we will refer to photometric seds throughout the paper as `` seds , '' not to be confused with spectra .
photometry has been widely available and is a good way to cover a larger spectral range than typically available with spectroscopy to get information about different stellar populations , but it lacks spectral details and suffers from more degeneracy between age , metallicity , and dust content .
recent surveys with spectra of hundreds or thousands of galaxies , in addition to the photometry , provide more information for detailed analyses .
furthermore , the release of population synthesis models at high spectral resolution @xcite allows one to take advantage of the detailed information contained in individual lines over the entire observed spectral range , rather than relying on only a few lines or on more degenerate broadband colors . in this work , we evaluate age and metallicity estimation techniques on globular clusters by comparing results from @xcite simple stellar population ( ssp ) model fits to the spectra , broadband photometry , and line indices of the clusters .
globular clusters provide a simpler test case than galaxies since their stars likely provide a coeval and nearly homogeneous metallicity population that has no internal dust extinction .
we test the utility of using these models to estimate ages and metallicities of high redshift galaxies by first applying them to a sample of extragalactic globular clusters from the @xcite public database that have integrated spectra , photometry , and age or metallicity estimates obtained by other methods .
galaxies will be discused in a future paper .
this paper is arranged as follows .
globular cluster spectra and photometry are described in 2 , comparison ages and metallicities from the literature are described in 3 , the stellar population synthesis models are described in 4 , our model fitting procedures are explained in 5 , age and metallicity results are presented in 6 , a discussion of issues is given in 7 , and conclusions are summarized in 8 .
the @xcite online spectral database provides our sample of extragalactic globular clusters that cover a significant range of age and metallicity , making their results applicable to evolving galaxies .
we chose those clusters that also have photometry available in the literature , which includes 79 clusters in the large magellanic cloud ( lmc ) , 4 clusters in the small magellanic cloud ( smc ) , and 18 clusters in m31 .
the spectra were obtained with the 1.52 and 2.2-meter telescopes at eso in la silla @xcite , the 2.15-m casleo telescope in argentina @xcite , the 3.6-m cfht in hawaii @xcite , and the 4.2-m william herschel telescope in la palma @xcite .
the data include near ultraviolet , optical , and near infrared spectra , covering a total wavelength range of 3200 - 10000 for some clusters , and varying in spectral resolution from 6 to 23 .
we should note that for all clusters except a few in m31 , this resolution element is larger than the @xmath5 = 10 required for lick / ids line indices .
all available spectral segments for each object are used to utilize maximum wavelength coverage , but since they originate from a number of different sources , flux calibration between regions may not be consistent .
table [ res_tbl ] lists each object with its specific wavelength coverage , spectral resolution , and source references .
we compile photometry of the clusters from various sources in the literature . @xcite
provide ubv photometry for the lmc clusters with @xmath620% of the objects taken from @xcite and the rest re - observed by the authors at ctio or casleo .
@xcite provide vjhk photometry for lmc clusters from observations on 1 - 2.5 meter telescopes at las campanas and cerro tololo observatories . for the smc clusters , @xcite provide vjhk and ubv are from @xcite who use photometry from @xcite and a few other sources .
for the m31 clusters , @xcite supply bvri from observations on the 152-cm telescope in loiano , bologna and on the 4-m kpno telescope .
@xcite provide ubjhk from observations on the 1.2-m telescope at the fred l. whipple observatory .
even for a given globular cluster , a wide range of age and metallicity estimates typically exists in the literature .
this is not completely surprising given the different methods employed to derive the parameters , some based on broadband colors and others on very narrowband spectral line strengths , but the dispersion in estimates is quite large for many clusters .
we will compare our results to age and metallicity values from a number of sources and different techniques to show where our values lie in the total spread .
literature ages and metallicities for specific clusters can be found in table [ age_tbl ] , along with our derived values .
one of the oldest techniques for measuring the age of a star cluster is using its integrated ubv colors .
when placed on color - color plots , globular clusters follow a sequence that can be related to age .
we use mc cluster u - b color ages that were determined by @xcite and @xcite .
when photometry is available for individual stars in a cluster , the location of the main sequence turnoff ( msto ) on a color - magnitude diagram ( cmd ) can be used to determine the age of the cluster .
we use a large compilation of msto ages for mc clusters from @xcite .
refinement of the msto technique is achieved by fitting theoretical isochrones to the cmds .
we take cmd fit ages from @xcite and @xcite .
we also add comparison ages determined from photometric fits to gissel96 model seds @xcite in the batc filter system , using 15 intermediate bandwidth passbands , by @xcite , and from the line index ratios h@xmath8 / fe i @xmath94045 and ca ii h / k by @xcite , which were developed to work better on younger objects than the typical lick line indices for use on older objects .
we consider the leonardi & rose line index ratio ages to be some of the most secure , given that they are determined from high quality spectra and are tailored for the younger ages of many clusters in this sample .
@xcite utilize diagnostic diagrams constructed from ew sums of metal ( ca ii k , g band , mg i ) and balmer ( h@xmath8 , h@xmath10 , h@xmath11 ) lines , which are calibrated against literature values and placed on homogeneous age and metallicity scales , to determine ages and metallicites of mc clusters .
we also include these in our comparison values of age and metallicity .
fitting theoretical ishchrones to cluster cmds is generally accepted as the most secure age determination possible when using photometry , however the results do vary between sets of ishochrones .
the absolute derived age depends on model zeropoints and uncertainties .
@xcite find that padova isochrones with agb stars included underpredict the luminosity function of giants by 0.2 - 0.3 dex when compared to observations of 47 tuc , resulting in an overprediction in the spectroscopic age of 2 - 3 gyr for old stellar populations .
young stellar populations are much less affected .
additionally , the padova isochrone uncertainty in t@xmath12 of 75 k leads to a 1 gyr age uncertainty or 0.1 dex uncertainty in the [ fe / h ] scale .
@xcite also find discrepancies between cmd and spectroscopic ages of 47 tuc due to the exclusion of @xmath13-enhancement and atomic diffusion in evolutionary models , causing a zeropoint offset . if 47 tuc is representative of other old stellar populations , for clusters in the 10 - 14 gyr age range these model isochrone issues could lead to spectroscopic based age discrepancies of 20 - 40 % when compared to cmd based values .
large homogeneous samples of globular cluster metallicities are harder to find , since metallicity can not be as accurately determined from photometry and typical methods require high quality spectra .
however , some photometric techniques have been calibrated to spectra .
of these we use [ fe / h ] values from vjk colors by @xcite , fits to cmds @xcite , and fits to gissel96 model seds by @xcite .
our more secure comparison metallicites come from spectra - based techniques .
for these we use [ fe / h ] from the @xcite line index ratios and their comparison literature values that were derived from ews of ca , fe , mg , and na lines @xcite , and from ews of the ca ii triplet at @xmath98500 @xcite .
we incorporate the comparison literature values used by @xcite , which include metal line abundances @xcite , line indices @xcite , and an infrared index at 1.6 @xmath3 m @xcite . and
finally we use a combination of line indices and spectral breaks from @xcite : d(4000 ) , cnb @xmath93883 , g - band @xmath94300 , mgh , mg _
b _ , and fe @xmath95270 .
we use the high spectral resolution ( r@xmath62000 ) ssp `` standard model '' from bruzual & charlot ( 2003 ; hereafter bc03 ) that utilizes the stelib / basel 3.1 spectral library , the padova 1994 evolutionary tracks , and the @xcite initial mass function ( imf ) with stellar mass limits of 0.1 and 100 m@xmath14 .
they provide model spectra at [ fe / h ] = @xmath22.25 , @xmath21.65 , @xmath20.64 , @xmath20.33 , + 0.093 , and + 0.56 , which we linearly interpolate at each wavelength point to a grid of spectra that spans [ fe / h ] of @xmath22.2 to + 0.5 in increments of 0.1 dex , and ages of 1 myr to 20 gyr in increments of 2 myr for 1@xmath1510 myr , 20 myr for 10@xmath15100 myr , 200 myr for 100@xmath151000 myr , and 1 gyr for 1@xmath1520 gyr .
the bc03 models include thermally - pulsing asymptotic branch stars ( tp - agb ) , making use of the multi - metallicity models of @xcite that have been calibrated on stars in the galaxy , lmc , and smc .
the importance of these stars , which have a strong influence on the integrated near infrared light from star clusters of certain ages , will become apparent later in the paper .
this phase can dredge up carbon in the stellar atmospheres , leading to carbon - rich stars , which are also very red and can dominate the near infrared light from some clusters .
although no simple prescription can be expected to match all clusters , these stars must somehow be included .
@xcite achieve this by defining the transition to carbon stars and the duration of this phase with the models of @xcite and @xcite . this semi - empirical prescription for tp - agb and carbon stars has been tested on and provides good agreement with observed colors of magellanic cloud clusters and with optical and near infrared surface brightness fluctuations of metal - poor galactic globular clusters and more metal - rich nearby elliptical galaxies @xcite .
our approach is to use the maximum amount of information possible for each object .
broadband photometry often covers a much wider spectral range than spectroscopy , providing better constraints on the possible mixes of stellar types , but suffers from age - metallicity degeneracy .
however , by combining broadband information with detailed spectral analysis , we hope to better constrain model fits and thus derived ages and metallicities .
we investigate the ability of both broadband and detailed spectral features of the models to estimate age and metallicity of the globular clusters by fitting models to the full spectrum , to the continuum - normalized spectrum , to the continuum shape , to photometry , and to spectral line indices of the clusters .
each of these fitting procedures is described in detail in the following sections . * _ full spectrum fits _ utilize information from both the line strengths and the continuum shape .
* _ continuum - normalized ( cn ) spectrum fits _ use only the information contained in the lines , losing important continuum information , but also removing any adverse effects of inaccurate flux calibration in the data or continuum shape errors in the models .
model continua problems are particularly possible in the near infrared due to the inability to properly account for the number of tp - agb or carbon stars that can contribute much of the light at these wavelengths . *
_ continuum fits _ use only the continuum shape that was removed from the spectrum . *
_ photometry fits _
cover a much broader wavelength range , but lose detailed spectral information and may also be affected by continuum shape problems in the data or models .
* _ line index fits _ move from broad to narrow and focus only on specific spectral lines that are known to be sensitive to metallicity or age .
we evaluate these specific lines in the models by calculating the indices directly from model spectra and comparing those to line indices calculated from the cluster spectra in the same way .
the spectra for each object in the @xcite database consist of near ultraviolet , optical , and near infrared segments from multiple sources at spectral resolutions of 6 - 23 .
we splice all spectra together to cover the widest possible range for each cluster .
no noise or sky spectra are included in the database , so we must estimate the noise from the spectra . to do this
we calculate the standard deviation of the spectrum in 100 bins and scale the resultant signal - to - noise ratio ( s / n ) spectrum to match the average s / n value quoted in the source publications for some objects . for those objects that do
not have quoted s / n values , the standard deviation spectrum provides the noise .
this likely overestimates the noise since absorption lines will increase the standard deviation of a bin , but the required scaling is typically less than 10% .
we prepare the data for fitting by marking regions that contain emission lines , noise spikes , and sky background residuals , as well as two regions for which the stelib stellar library used in the models has problems with telluric features ( 6850 - 6950 and 7550 - 7725 ) .
these regions are masked and deweighted for each cluster by significantly lowering its s / n spectrum over the affected wavelengths .
we smooth the grid of model spectra , which have an intrinsic resolution of 3 , to the wavelength dependent resolution of each cluster and fit its spectrum using two free parameters , age and metallicity , and a s / n - weighted normalization constant , @xmath16 , which is uniquely determined for each model .
we select the best fitting model by calculating the summed @xmath17 for each model spectrum of a given age and metallicity , relative to the cluster spectrum , and selecting the model in the grid with the minimum value . in the @xmath17 calculation
, we use the previously determined cluster noise spectrum as the @xmath18 for the flux , _ _ f__@xmath19 , at each wavelength bin , _ i _
, for each model , _
n _ , as given in equations 1 and 2 .
once the best fitting model is determined for a cluster , we inflate its noise until the @xmath20=1 to include any additional noise sources and to better estimate confidence contours and error bars .
if , after this process , the estimated errors are smaller than the distance to the next model grid point , an error bar of half this distance is assigned to the cluster for both age and metallicity .
@xmath21 @xmath22 errors on age and metallicity estimates are determined from the @xmath23 contours .
the s / n of the data controls the size of these contours , and thus the confidence of the derived parameters .
typical levels for the globular clusters are s / n@xmath6100 per resolution element .
figure [ fig1 ] shows a simulation of the effect of s / n on the confidence of the derived age and metallicity for ngc 419 in the smc , achieved by artificially adding different levels of gaussian noise to its spectrum and fitting models to each of the noisy continuum - normalized spectra .
the simulated s / n per resolution element from top left to lower right in figure [ fig1 ] is 99 , 22 , 12 , 8 , 6 , and 4 .
solid , dashed , and dotted lines represent 1 , 2 , and 3@xmath24 contours .
the color bars give the @xmath23 values for confidence levels of 68.3% ( @xmath23=2.3 ) , 90% , 95.4% , 99% , 99.73% , and 99.99% ( @xmath23=18.4 ) for two degrees of freedom .
black dots mark the locations of models in the age - metallicity grid .
the best fitting model is marked by a black cross in each panel .
the age and metallicity from the best fitting model in the highest s / n case is marked by a yellow circle in each subsequent panel for comparison .
two things become apparent from these plots .
first , as the confidence contours grow with decreasing s / n , metallicity becomes harder to constrain than age . in some low s / n conditions (
e.g. the last panel ) multiple islands of age - metallicity combinations arise as equally likely answers .
the best fitting model in each panel does not always overlap the 1@xmath24 contours of all other panels , however , this mostly happens in the [ fe / h ] dimension and those values are less secure than the ages .
this is our first indication that metallicity is poorly constrained , compared to ages .
the second point is that we find a limiting s / n below which the estimated age and metallicity become unreliable .
this happens somewhere between s / n = 22 and 12 in these plots , at which point the contours in the low s / n cases do not overlap the estimates from the highest s / n case .
we find that the s / n must be @xmath2515 per resolution element ( s / n@xmath65 @xmath4 ) for @xmath5=13 - 23 to adequately constrain the age .
metallicity is harder to constrain under all conditions .
a more detailed analysis of limiting s / n and spectral resolution can be found in
[ snr_sec ] .
errors in flux calibration of the spectra can induce errors in derived ages and metallicities .
additionally , model continua can be off in the near infrared due to stochastic contributions from tp - agb stars in @xmath6 0.1 to 1 gyr old populations and from carbon stars in populations of age 0.3 to 2.5 gyr @xcite .
these variations can result in differences of nearly 2 magnitudes in v - k ( bruzual & charlot 2003 , figure 8) .
agb stars in the thermally pulsing phase can contribute over 80% of the k - band light @xcite , the exact amount of which is hard to predict since this phase involves ejection of the outer stellar envelopes and geometry - dependent obscuration of the central stars .
the bc03 models include tp - agb and carbon stars semi - empirically , but in any given cluster the k - band light can vary depending on the numbers of stars it actually has in these phases . to avoid these issues ,
we normalize out the continua with a median - binning routine that uses bin sizes tailored to work for different classes of objects .
the young clusters of the magellanic clouds have strong balmer discontinuities and need to be sampled by small bins to correctly trace this sharp continuum break , while older objects in m31 require larger bins to smooth over absorption troughs in the red part of the spectrum .
the bins that we use are 100 wide blueward of 4000 and 300 - 700 wide redward of 4000 .
the points in each bin are sorted by flux values , the lower 1/2 of the points are ignored to reduce the weight of absorption lines , the highest few points are rejected to ignore noise spikes , and the median of the remaining flux values is assigned to that bin .
the medianed bin fluxes are linearly connected and the object spectrum is divided by the result , providing a flat continuum .
the same binning parameters used for an object are also used on each model spectrum in the grid .
removal of the continuum shape should not affect the values of line indices measured on the spectrum , providing a good check of our procedure .
we applied this test , as suggested by the referee , by choosing a random sample of 25 globular clusters spanning the entire age range and measured line indices before and after continuum normalization .
the indices used were mg__b _ _ , fe5270 , fe5335 , h@xmath11 , h@xmath26 , h@xmath27 , h@xmath28 , h@xmath29 , g4300 , @xmath0fe@xmath30 , and [ mgfe]@xmath31 .
only the positive index values that could be reliably measured were used , as in [ line_ind ] . on average
the values change by less than 1% , and all line index median differences for the 25 clusters are below 3% with no correlation to the derived cluster age .
the continuum shape that we divide out is also fit by model continua and referred to throughout the paper as `` continuum fits .
'' different types of spectral fits are illustrated in figure [ fig2 ] for an old cluster , g177 in m31 , and a young cluster , ngc 1711 in the lmc . to test ages and metallicities
derived from broadband features over a wide baseline , we fit models to the ubvjhk photometry from the literature , using all of the bands that are available for each cluster .
our model fitting procedure requires scalable fluxes , so we convert cluster magnitudes to fluxes using the ubv zeropoints from @xcite and the jhk zeropoints from @xcite .
we obtain model fluxes for each band by convolving filter transmission curves with the model spectra .
these fluxes are scaled to those of the object with an overall s / n - weighted normalization parameter during the fit .
the best fitting model is selected by @xmath17 minimization , using equations [ alpha ] and [ chi2 ] with _ i _ now representing each photometric band .
errors are not given for all literature photometry , so we initially assume a 5% photometric flux error in all bands for each cluster , which is then adjusted to make @xmath20=1 for the best - fitting model .
fitting parameters are age and metallicity .
photometric fits are also illustrated in figure [ fig2 ] .
lick line indices have been used for some time @xcite .
they are based on a very specific set of data and their application to new objects is somewhat complicated . to properly use lick / ids indices ,
some of the same stars must be observed with the instrumental setup of the new objects and the spectra of the new objects must be convolved to exactly the same wavelength dependent resolution as the original lick data , which varied from run to run during the development of the indices .
very small differences in the wavelengths can cause large errors in the indices . to investigate the robustness of these specific lines in the model spectra , we calculate line indices directly from the models , similar to the calibrations done in @xcite .
direct application of the models for this purpose would avoid the step of convolving the object spectra to the estimated lick resolutions . if the models are smoothed to the same spectral resolution as the cluster data , indices can be calculated directly from both spectra in the same manner .
however , the bandpasses of the lick indices were originally chosen for data with @xmath5 = 8 - 12 @xcite .
specific lines in lower resolution spectra may begin to lose age and metallicity information .
nevertheless , given the long history of the use of line indices , we wanted to attempt a variation of this technique on data with resolution and s / n representative of that from typical galaxy surveys .
first , we broaden the model spectra to match the resolution of each object , which may vary for wavelength regions obtained on different instruments . then , we calculate indices for the lines in the lick system using the passbands defined in @xcite and @xcite . to these
we add @xmath0fe@xmath30 , an average of fe @xmath95270 and fe @xmath95335 ; d@xmath32(4000 ) , the 4000 break strength using the narrow passbands defined in @xcite ; cnb @xmath93883 , mgh , and ca ii h+k , as defined by @xcite ; and [ mgfe]@xmath31 = @xmath33 , as defined by @xcite . the [ mgfe]@xmath31 index has been found to be a good tracer of total metallicity and to be insensitive to @xmath13/fe enhancement ( for spectral resolution near 10 ) , which may be important for some of the globular clusters in this sample .
we calculate the indices as described in @xcite , by connecting the average flux values of the pseudocontinua sidebands with a straight line and integrating the flux under that line over the index passband .
indices ( or break strengths ) are calculated for cnb , h@xmath34 , h@xmath35 , cn@xmath36 , cn@xmath37 , ca4227 , g4300 , h@xmath38 , h@xmath39 , fe4383 , ca4455 , fe4531 , c@xmath374668 , h@xmath11 , mgh , fe5015 , mg@xmath36 , mg@xmath37 , mg__b _ _ , fe5270 , fe5335 , fe5406 , fe5709 , fe5782 , na d , tio@xmath36 , tio@xmath37 , [ mgfe]@xmath31 , @xmath0fe@xmath30 , and d@xmath32(4000 ) .
indices for model spectra are calculated in the same manner .
we select the model with the smallest sum of absolute value residuals between its index values and the corresponding indices derived from a cluster s data . the age and metallicity of this closest matching model
are adopted as the estimates for the cluster ; no interpolation is performed between model grid points . to facilitate comparisons with the literature
, we repeated the model selection using different subsets of the indices listed above : all indices simultaneously ; huchra et al.s ( 1991 ) group of 6 , cnb , d@xmath32(4000 ) , g4300 , mgh , mg__b _ _ , and fe5270 ; and h@xmath11-[mgfe]@xmath31 as in @xcite .
examples of model fits are shown in figure [ fig2 ] for old and young clusters .
panels a , c,&e are the cluster g177 in m31 and panels b , d,&f are ngc 1711 in the lmc .
the full spectrum fits are shown in a&b , the continuum - normalized spectrum fits are shown in c&d , and the continuum fits are shown in e&f .
photometry fits are also shown in a&b with u , b , v , j , and k bands for g177 and u , b , v , j , h , and k bands for ngc 1711 .
the circles are the object photometry and triangles are model fluxes from the best photometric fit .
the solid lined black spectra are the clusters , the dashed cyan lines are best fitting models from the spectral fits , and the dotted green lines are spectra corresponding to the best fitting models from the photometry fits .
the model spectra corresponding to the best fitting photometry are normalized to the cluster spectra at v band in these plots . in the continuum
fits , black lines are the data and green lines are the models .
the resulting ages and metallicities from the different types of fits are given in the plot labels .
for these two clusters the ages derived from the different types of model fits nearly agree at 20 gyr for g177 and about 0.1 gyr for ngc 1711 .
the agreement between photometric and spectroscopic fits seen here is not always the case .
g177 in figure [ fig2]a illustrates a case where the photometry through k band reveals a redder sed than the optical spectrum . in this case , the best fitting photometry model has a higher metallicity than the model that best fits the spectrum to make the object redder , an effect of age - metallicity degeneracy .
a similar effect is seen in some of the young clusters , presumably due to a near infrared excess from tp - agb or carbon stars in that case .
some examples of such clusters within the affected age range of 0.1 - 2 gyr are shown in figure [ fig3 ] . to evaluate the different model fitting techniques , we first compare their results to each other . in the following sections we compare our results to the literature .
figure [ fig4 ] compares the ages derived from cn spectrum , full spectrum , photometry , and continuum fits .
clusters with average literature ages @xmath40 1 gyr are maked with yellow circles and @xmath0 1 gyr with cyan diamonds . although the general trends of the ages agree , there are differences in the results . when compared to cn spectrum fits ,
the methods that contain the continuum overestimate the ages of many clusters younger than 1 gyr .
we believe that this is due to red supergiant stars for clusters in the age range of a few to tens of myrs and tp - agb stars for clusters in the age range of 0.1 to 1 gyr
. both of these types of stars can have stocastic effects in star clusters because of their extremely high luminosities .
if more light from them appears in the integrated spectra of the cluster than is included in the models , the cluster sed will appear redder than the model , forcing an older redder model sed as the best fit .
all fitting methods that contain model continua will have this same bias .
this effect is discussed further in comparison with literature ages in [ age_results ] . for clusters with average literature ages
@xmath0 1 gyr our tightest estimated age correlation is seen between the full spectrum and continuum fits ( figure 4c ) , both of which contain the spectral shape .
the median fractional age offset , relative to the average literature age , between these fits is 0.04 for 63 clusters .
the fact that the continuum - full spectrum correlation is tighter than the cn - full spectra correlation ( fractional offset of 2.40 ) suggests that the continuum shape has a larger effect than the spectral lines in constraining the ages of these young globular clusters for our employed fitting methods .
therefore , any problems in the continuum shape of the data or models will have grave effects on the derived ages . for the clusters
older than 1 gyr our tightest estimated age correlation is between the cn spectrum and photometry fits ( figure 4b ) , with a median fractional offset of 0.18 for 12 clusters . for young clusters
the metallicities obtained from our different techniques show virtually no correlation , with a tendency for the continuum to give higher metallicities than the full spectrum . for clusters older than 1 gyr
there is a very weak correlation between metallicities obtained with continuum and full spectrum fits , possibly indicating that the derived metallicity has a stronger dependence on the continuum shape than the spectral lines for our fitting method . in figure [ fig5 ]
we compare our globular cluster ages to those obtained by other methods in the literature .
the thick cyan horizontal bars connect points that represent the same cluster , but that have ages from multiple sources in the literature .
these illustrate the large spread in previously derived ages .
the thin horizontal bars are the quoted errors on the literature ages and the thin vertical bars are our 1@xmath24 errors on ages derived from model fits .
note that the errors are much larger for the photometry fits , reflecting more degeneracy when using broadband colors than when using spectra or the continua shapes .
over the entire age range spanned by the clusters and considering all literature age estimation techniques , our cn spectrum fit ages best match those from the literature .
the average offsets and rms scatter our ages from groups of literature values based on different techniques are given in table [ rms_age_tbl ] , where column 1 gives values for all clusters in the sample that have literature ages and column 2 excludes two outliers ( for reasons explained in [ weird_ones ] ) and the clusters that only have literature values from photometric fits to bc96 models ( because these used only a few fixed metallicity values in the model grid , which could skew the derived ages , and they are based on old models ) . excluding these points , our cn spectrum fits compared to the averages of the literature values for each object have an average offset of 0.29 dex ( 0.19 gyr ) with a dispersion of 0.36 dex ( 1.00 gyr ) .
our photometric ages have an average offset of 0.59 dex ( 0.48 gyr ) and dispersion of 0.76 dex ( 1.19 gyr ) .
there are some systematic differences between the two sets of results .
the scatter of the cn spectrum fit points is more uniformly distributed around the literature values , unlike the photometric points that tend to overestimate the age , particularly for the younger clusters .
the full spectrum fits , which contain the continuum , also overestimate the ages below 1 gyr , with an average offset of 0.69 dex ( 0.54 gyr ) and a dispersion of 0.9 dex ( 3.2 gyr ) .
this is the same effect that was seen when comparing our fits with and without the continua included .
we now return to stars that can strongly affect the continua shapes .
detailed studies of such stochastic effects in star clusters can be found in @xcite , @xcite , and @xcite .
our photometric fits , in particular , show two different age ranges where our best - fit ages depart from literature ages on the high side ( figure [ fig5]c ) : 4 to 100 myr and 0.1 to 1 gyr .
the age range of 0.1 to 1 gyr would be affected by tp - agb stars .
if the model prescription for these stars does not exactly match those in the clusters , then the older ages from fitting methods that include the continuum could be explained by older , redder spectra giving the lowest @xmath17 for clusters that have this near infrared excess .
our derived ages also scatter to older values for clusters younger than 0.1 gyr when using models that include the continuum shape .
this effect is likely due to stocastic red supergiants , which have been observed with estimated masses up to 120 m@xmath14 in the lmc @xcite .
these stars will affect clusters younger than a few tens of myrs .
either the number of red supergiants is underestimated in the bc03 models , or the upper mass cutoff of 100 m@xmath14 for the chabrier imf is not quite high enough for the lmc . see
[ weird_ones ] for a discussion of two lmc clusters that are dominated by supergiants , which are also our extreme outliers in the cn spectrum fits of figure [ fig5]a . because of the large age range covered by the clusters , it is perhaps useful to look at the behavior of the fractional age errors relative to the literature values , which are plotted in figure [ fig6 ] .
overall , the cn spectrum fits provide ages with fractional offsets from the literature of 61% , while the photometry does much worse at 1048% .
if we consider younger and older clusters separately , the cn spectrum fits do better than photometry on the average age offsets for both young and old clusters , but the photometry does better on fractional errors for the older objects . for clusters with ages
@xmath0 1 gyr , the cn spectrum fits have an average offset of 63% ( 0.30 dex or 0.09 gyr ) and dispersion of 0.37 dex ( 0.35 gyr ) , while the older clusters have an offset of 35% ( 0.16 dex or 1.63 gyr ) and dispersion of 0.17 dex ( 3.78 gyr ) .
the dispersion is higher below ages of 0.1 gyr even though the cn spectra are not affected by the continua effects mentioned earlier .
this increased dispersion is likely due to the degeneracy of many spectral lines in this age range ( see figure [ fig10 ] and [ line_ind ] ) .
the photometry fits do much worse fractionally on young clusters with an average offset of 1115% ( 0.62 dex or 0.38 gyr ) and dispersion of 0.78 dex ( 1.02 gyr ) , and better on the fractional errors of older clusters with an offset of 23% ( 0.10 dex or 2.06 gyr ) and dispersion of 0.11 dex ( 2.67 gyr ) .
the lowest fractional errors for the cn spectrum fits occur for the @xcite line index ratios and for the clusters older than 1 gyr , both at 35% .
photometry fits do slightly better on the older objects with errors of 23% , but these errors rise substantially for the young objects to 1115% .
the largest fractional age offsets occur in the clusters less than 1 gyr old for the full spectrum ( 3450% ) , continuum ( 4432% ) , and photometry fits .
this is likely due to agb stars .
furthermore , the full spectrum and continuum fits have larger age offsets in the older clusters ( 425% and 480% ) than do the cn spectrum or photometry fits , which may be due to flux calibration issues , since both of these methods include the observed spectral continuum shape . to take care of both data and model continua issues
, it appears that continuum - normalized spectral fits are the most robust in producing accurate age estimates for the clusters of all ages .
the age errors do not correlate with metallicity , so they seem to be dominated by the age regime of the cluster . although photometry does a decent job on older objects , it should be noted that after excluding the bc96 photometry literature points , we only have 4 clusters in the @xmath30 1 gyr bin at ages of 1.2 , 2.2 , 12.4 , and 13.1 gyrs .
it could just be that the two younger ones have no problems with boosted near infrared emission .
in general , since there is not a good way to determine whether the 1 - 2 gyr old objects are affected by the agb star problem , using the cn spectrum fits would avoid this issue . some combination of cn spectrum and photometry fits will likely provide the most robust answer .
this will be explored further in future work .
other studies have compared results from their cluster age estimation techniques to the literature . @xcite use colors formed from ubvi photometry along with starburst99 @xcite and galev @xcite models to derive ages and compare to those in the literature from integrated colors @xcite and isochrone fitting @xcite in figure 8 of their paper .
they find an age correlation with dispersion of 0.76 gyr , which drops to 0.49 gyr when considering only the more secure literature ages derived from cmds . over a similar
age range our cn spectrum fits provide ages with an overall dispersion of 1.0 gyr about literature comparison values , with 0.12 gyr for ubv colors and 1.06 gyr for cmd fits .
our photometry fits have dispersions of 1.04 gyr for ubv colors and 0.85 gyr for cmd fits .
although @xcite use clusters in the smc , only two overlap with our sample .
our literature ages come from different sources as well , so we can not make a direct comparison to this work . to summarize our results for globular cluster age estimation , the cn spectrum model fits best match the entire range of age estimates from the literature ( figure [ fig5 ] and table [ rms_age_tbl ] ) , with average errors of 0.16 dex ( 35% ) for older clusters and 0.3 dex ( 63% ) for younger clusters .
the full spectrum , continuum , and photometry fits overestimate the ages of many clusters below 1 gyr , apparently suffering from an excess of agb stars making the cluster spectra redder than the models and forcing older aged models to provide the best fits .
although photmetric fits seem to do better for clusters older than 1 gyr ( 0.1 dex or 23% error ) , we will show in [ mets ] that cn spectrum fits are superior for simultaneously providing both age and metallicity for these older objects .
furthermore , the uncertainty of whether the clusters will have boosted near infrared emission due to agb stars makes using the photometry alone a less reliable technique than using spectra . comparing the fitting methods ( figure [ fig4 ] )
, we find a tighter correlation between ages derived from the full spectrum and continuum fits than from the cn and full spectrum fits , suggesting that the ages are more strongly driven by the continuum shape than by the spectral lines . because the derived ages are so strongly influenced by the continua and we have seen signs of problems in the models matching the cluster continua , we conclude that the best method for deriving accurate ages for globular clusters of all ages , especially when simultaneously determining metallicity , is fitting models to their continuum - normalized spectra .
there are two extreme outliers in the cn spectrum age plot of figure [ fig5]a .
the spectra of these outlying objects are shown in figure [ fig7 ] .
the upper leftmost outlier in the cn spectrum age plot is ngc 2092 in the lmc ( figure [ fig7]a , b ) , with a literature age of 4 - 12 myr ( from mstos and ubv colors ) and for which we derive an age of 10 gyr from both the cn and full spectrum fits .
its spectral coverage is only over the optical range of 3500 - 5870 .
the shape of its spectrum is fairly flat with emission lines and few absorption features .
the emission lines include [ oii]@xmath93727 , [ oiii]@xmath414959,5007 , and [ neiii]@xmath93869 nebular emission .
the emission lines are masked out in our fits , which leaves very few distinguishing features .
the cluster s ubv photometry is also essentially flat .
our best fit metallicity of this object is bottomed out at the minimum in the model grid , [ fe / h][email protected] , for both the cn and full spectrum fits , likely because nothing would fit well and the lowest metallicity model that has very few metal lines in the red produced the lowest , although high , @xmath17 .
@xcite list this object as a cluster embedded in a star forming complex with a flat continuum probably caused by red supergiants , which are stocastic in nature even for large star clusters .
this appears to be a case that can not be fit well with either spectral or photometric techniques using the bc03 ssp models .
the other outlier in the upper lefthand corner of the cn spectrum age plot in figure [ fig5]a is ngc 2096 in the lmc .
this object has a literature age of 49 myr from its u - b color .
its spectrum ( figure [ fig7]c , d ) contains a moderate sized 4000 break of d@xmath42(4000)=1.28 , which is consistent with an age of at least 1 gyr from index plots in @xcite , and shows no evidence of the characteristic spectral shape of a young object with a strong balmer discontinuity that would be expected for an age of only 49 myr . for this object
, our age is 1.2 gyr ( full spectrum and photometry fits ) to 6 gyr ( cn spectrum fit ) .
however , @xcite note this object as one where a few luminous intermediate temperature supergiants dominate the cluster s integrated spectrum and claim that its age is even younger at 6 - 12 myr .
this may be a stocastic case where the supergiant dominated spectrum is not properly matched by the models .
two out of 101 globular clusters in this sample could not be properly fit by the bc03 models .
both are very young clusters that show evidence of supergiant dominance in their integrated spectra .
the stocastic nature of these stars make it difficult for us to derive accurate ages for clusters younger than a few tens of myrs .
we compare our derived cluster metallicites to the literature values in figure [ fig8 ] .
there are fewer points on these plots than for the ages in figure [ fig5 ] because fewer of the clusters had metallicity estimates in the literature .
the magnitudes of the metallicity errors relative to literature values are a stronger function of cluster age than [ fe / h ] .
this is illustrated in figure [ fig9 ] where the [ fe / h ] errors are plotted against average literature ages .
table [ rms_met_tbl ] gives the average [ fe / h ] offsets and dispersions for different groups of clusters .
metallicity is harder to constrain than age .
this was first seen in figure [ fig1 ] when the @xmath23 contours grew primarily in the metallicity dimension as noise was added to the cluster spectrum .
nevertheless , our cn spectrum fits do well on metallicity estimates for the older clusters . in figure [ fig8 ]
we see a tight correlation of metallicity from these fits with the literature for clusters older than 1 gyr .
the large symbols mark literature estimates that are based on spectra .
there is an overall [ fe / h ] scale offset between m31 ( _ blue pluses _ ) and the mc ( _ red circles , black crosses , black inverted triangles _ ) .
if this offset of 0.45 dex is removed from the mc cluster literature metallicities , our metallicites agree to 0.12 dex with those from the @xcite line indices ( _ blue pluses _ ) , literature values on a homogeneous scale from @xcite ( _ black inverted triangles _ ) , and @xcite line index ratios ( _ red circles _ ) and their comparison literature values ( _ black crosses _ ) .
the small symbols have literature metallicity estimates that are based on colors .
there is more scatter in these values .
the weaker correlation with vjk colors ( _ green triangles _ ) can be seen , as well as its break down at low metallicities of [ fe / h ] @xmath430.7 .
the @xcite metallicities ( _ magenta squares _ ) are for older clusters in m31 and were derived from photometric fits to bc96 models using only three metallicities , [ fe / h ] = 0.0 , @xmath20.7 , @xmath21.7 . their discrete metallicity steps and any differences between the 1996 and 2003 models are the reasons for the large dispersion of these points
. for the clusters younger than 1 gyr , our metallicites from cn spectrum fits show no correlation with literature metallicity values .
we believe this is because continuum information is also necessary for determining metallicities of young clusters , particularly since younger clusters likely have higher metallicity with more line blanketing .
a hint that the continuum information might be helping to determine the metallicity of some young clusters can be seen in the plots of our full spectrum fits in figs .
[ fig8 ] and [ fig9 ] . in figure
[ fig8]c , if we ignore the points below the 1:1 correlation line for the moment , there does seem to be a metallicity trend with the literature for many of the other points .
all of the clusters that we ignored below the line show possible signs of a near infrared excess in their photometry over the model that best fits their optical spectra . some of these spectra are shown in figure [ fig3 ] .
these clusters are in the correct age range , 0.1 - 2 gyr , to be affected by tp - agb stars .
it is not clear why our derived metallicities of these objects appear too low , since it seems that redder spectra would be better fit by models with higher metallicities , but maybe it is more an indication that these results can not be trusted .
fits to the continua shape alone ( figure [ fig8]e ) result in much less correlation to the literature , however , 5 of the 6 outliers below the 1:1 correlation line in this case are common to those in the full spectrum plot . therefore , although it seems that correct continuum information might allow metallicity estimates to be made for young objects , the uncertainty in proper modeling of the continua shapes makes the full spectrum fits unreliable .
we expect our metallicity estimates that are based on broadband photometry to be less well constrained than those based on spectra .
three of our comparison literature sources make cluster metallicity estimates based on photometry ( denoted in our plots by smaller sized symbols ) : the @xcite vjk colors , the @xcite bc96 colors , and two of the @xcite homogenized literature values .
our photometrically derived metallicities ( figure [ fig8]d ) do not correlate with these literature values , except for possibly a very weak relation with the vjk colors ( _ green triangles _ ) .
as expected , our metallicities derived from broadband photometry and the continuum shape show no correlation to any values from the literature that are based on spectral lines .
@xcite use empirical relationships between the sums of ews of balmer and metal lines to the literature age and metallicity values to estimate ages and metallicities of clusters in the magellanic clouds and in the galaxy .
they find that ew(h@xmath8+h@xmath10+h@xmath11 ) and ew(cak+gband+mg ) are both sensitive to age for clusters younger than 10 gyr , while ew(cak+gband+mg ) is sensitive to [ fe / h ] only for clusters older than 10 gyr .
neither of these ew sums correlate with the ages of old clusters ( t@xmath44 10 gyr ) or with the metallicities of young ( t@xmath45 10 gyr ) clusters .
our cn spectrum fits are similar to this technique in that they use line strengths , but over the entire spectral range of the data .
our fits produce metallicities that agree with literature values only for the older clusters , supporting the @xcite result , but extending the minimum age from 10 gyr down to @xmath6 1 gyr for valid metallicity estimates . to summarize our results for metallicity estimates , we agree very well with the literature metallicities that are based on spectra for clusters older than 1 gyr .
those based on colors have much more scatter , which suggests that our cn spectrum fits are more robust in estimating [ fe / h ] than methods using colors . for clusters younger than 1 gyr , metallicity is hard to constrain .
our full spectrum fits hint that correct continua shapes would aid in metallicity estimates of younger clusters , but given the uncertainties in modeling this shape , we can not accurately derive metallicity estimates for clusters younger than @xmath6 1 gyr .
we compare line indices measured on the bc03 model spectra to those measured on the globular cluster spectra in figure [ fig10 ] for h@xmath11-[mgfe]@xmath31 and @xmath0fe@xmath30-mg_b_. panels a - d show model grids calculated at the spectral resolutions of the data , 6 , 12 , 16 , and 23 respectively .
symbols mark the indices that could be measured for the clusters in each resolution group ( indices that came out negative due to noise or emission lines were ignored ) .
the 6 and 23 resolution groups ( _ squares and crosses _ ) are older clusters in m31 , while the rest are younger clusters in the magellanic clouds .
the h@xmath11-[mgfe]@xmath31 index grids clearly show that the model indices become degenerate at @xmath46 100 myr for @xmath21.0 @xmath0 [ fe / h ] @xmath0 + 0.5 , and at @xmath47 1 gyr for [ fe / h ] @xmath0 @xmath22.0 .
most of the young mc clusters fall in the degenerate regions of the grids .
this degeneracy also occurs for the traditional lick indices ( tabulated with the bc03 model package ) in this region of parameter space , as shown by the grids in figure [ fig10]e .
this plot only includes points from clusters that have spectra with resolutions near that of lick indices ( 11 and 12 ) .
@xcite investigated how well line indices calculated directly from a library of their models containing complex stellar populations with a range of star formation histories could match galaxy spectra from the sloan digital sky survey ( sdss ) .
they concluded that those models could simultaneously fit observed strengths of h@xmath11 , h@xmath38 , h@xmath34 , [ mgfe]@xmath31 , [ mg@xmath36fe ] , [ mg@xmath37fe ] , and d@xmath48(4000 ) in high quality galaxy spectra with s / n@xmath49 30 per pixel .
their sample of moderately low redshift sdss galaxies probes a different region of age - metallicity space than does our sample of globular clusters .
local galaxies are generally old with considerably higher metallicity than these clusters , a region of parameter space where the models do well . in contrast , the m31 globular clusters are old and metal poor and the mc globular clusters are young and metal poor .
bruzual & charlot note that their ssp models using the stelib / basel 3.1 library do poorer on line strengths for [ fe / h ] @xmath0 @xmath20.7 and our index plots add model degeneracy for @xmath46 100 myr , both of which are parts of parameter space occupied by globular clusters in this sample .
figure [ fig10]f shows @xmath0fe@xmath30 vs. mg _ b _ with model grids included for the minimum and maximum spectral resolutions . some of the clusters depart from the model grid such that [ mg / fe ] may be enhanced @xcite .
since we see signs of @xmath13-enhancement and the bc03 models use scaled solar abundance ratios , the preferred metal index here is [ mgfe]@xmath31 , which was shown by @xcite to be sensitive to metallicity while insensitive to @xmath13-enhanced abundance ratios at the lick / ids spectral resolution .
however , when [ mgfe]@xmath31 is plotted against balmer indices , which are affected by @xmath13-enhancement because of metal lines in the index passbands , many of the clusters fall outside of the model grids with lower balmer indices than the models ( figure [ fig10]a - d ) . one might suspect that the lower h@xmath11 indices are caused by the lines being partially filled in by emission , however , many of these clusters are old ones from m31 ( _ squares and crosses _ ) and unlikely to have emission .
the younger mc clusters , on the other hand , could be affected by emission .
nevertheless , since our purpose in calculating line indices is for comparison of using the model indices to using the entire spectrum , and the affected mc clusters fall in the degenerate region of the model grid with unusable indices , we do not correct for emission .
further possible causes of the h@xmath11 index discrepancy could be @xmath13-enhancement or the presence of blue horizontal branch ( bhb ) stars in the clusters .
the m31 clusters do show signs of @xmath13-enhancement in figure [ fig10]f .
line indices are changed by @xmath13-enhancement , but in the wrong sense to explain the high h@xmath11 in the bc03 scaled solar abundance ratio model grids .
h@xmath11 increases with [ @xmath13/fe ] @xcite , due to extra absorption in the blue pseudo - continuum and some in the central passband .
therefore , if the clusters are @xmath13-enhanced their h@xmath11 indices should be higher than the model grid .
the second possible culprit , bhb stars in old clusters with low metallicities , also affect h@xmath11 , but in the wrong sense .
h@xmath11 should increase if a bhb exists in the cluster .
we can not explain the h@xmath11 behavior of the clusters relative to the models by either of these causes .
this same situation is seen by @xcite in their figs .
12 and 13 where they calculate lick indices directly from the pegase - hr model spectra , which use the same stellar library as the bc03 models , and compare to globular clusters of intermediate to old ages in andromeda , m31 , m33 , m81 , and m87 .
they claim that this behavior of h@xmath11 is due to the clusters being extremely metal - poor , a regime where the interpolated stellar libraries suffer from large uncertainties due to lack of sufficient numbers of stars .
this same regime of [ fe / h]@xmath500.7 was noted as a less reliable region by bruzual & charlot .
the metallicites of our m31 clusters have literature values of @xmath21.35@xmath0[fe / h]@xmath00.29 , with an average of @xmath20.73 , so many do fall in the unreliable metallicity regime of the models . that , coupled with the index degeneracy below ages of 100 myr , make indices calculated directly from the models unusable for most of the globular clusters in our sample .
because calculating line indices from the models is troublesome for the age - metallicity parameter space occupied by many of the clusters in this sample , we take a different approach to investigating the importance of these spectral features .
we fit model spectra to the clusters only over regions that include and exclude some of the typically used line indices and breaks .
these features are @xmath51(4000 ) , h@xmath8 , g4300 , h@xmath10 , h@xmath11 , mg _
b _ , fe5270 , and fe5335 .
the index definitions for the extents of the pseudocontinua passbands of lines , or for the flux ratio windows for breaks , define our wavelength ranges over which to fit the models to cluster spectra .
figure [ fig11 ] shows the results of these fits , where a&b use only the regions including these features and c&d exclude these regions .
it is clear that these features are very important for deriving cluster parameters from the models , but also that fitting the entire spectral range ( refer to figure [ fig5 ] and [ fig8 ] ) does a better overall job in estimating age and metallicity .
spectra of z@xmath61 galaxies from large surveys will most certainly have lower s / n than the spectra of the globular clusters used in the present study .
we determine the lowest s / n value that can reliably be used to estimate ages by adding gaussian noise of different levels to cluster spectra via the _ noao.artdata.mknoise _ task in iraf and fitting cn model spectra to them over the wavelength range of 3325 - 9000 .
we first use actual cluster spectra to make the test realistic and avoid any biases that might be introduced by model spectra .
the clusters in this test were chosen for their high s / n spectra and availability of both age and metallicity estimates in the literature .
further tests including effects of spectral resolution were then conducted on smoothed model spectra with added noise .
the resulting ages and metallicities as a function of s / n @xmath4 for each of four clusters , g158 , ngc 419 , ngc 2134 , and ngc 1818 , are shown by different symbols in figure [ fig12 ] .
the average spectral resolutions of these data are 23 , 13 , 13 , and 13 , respectively .
the derived ages of the clusters hover around the same values as s / n decreases until @xmath65 @xmath4 , below which the derived ages become unstable .
the horizontal line associated with each set of symbols is the average of the derived ages or metallicites for that cluster from its s / n@xmath405 spectra . for comparison ,
the spread of literature ages and metallicities for each cluster are shown as vertical lines of the same styles on the right - hand side of the plots .
our derived parameters for these four clusters fall within the literature age ranges .
if the 0.45 dex [ fe / h ] scale offset is subtracted from the mc cluster literature values , then our average [ fe / h ] lines fall within the literature values for ngc 419 and g158 , which are the clusters older than 1 gyr .
we do not match the literature [ fe / h ] of the younger clusters .
we find that the spectra need to have s / n @xmath40 5 @xmath4 to achieve stable age estimates from cn spectral fits .
examples of fits to ngc 419 with added noise and s / n = 99 , 22 , and 12 re@xmath4 ( 27 , 6 , 3 @xmath4 ) are shown in figure [ fig13 ] .
note that the last one is below the limiting s / n value .
we have also begun a study of the effects of spectral resolution on derived ages and metallicities . to accomplish this we smooth model spectra to @xmath5= 5 , 10 , 15 , 20 , 25 , and 30 , add noise in 10 realizations at each s / n level up to 60 @xmath4 for a total of 300 - 500 simulations at each @xmath5 , and fit cn model spectra to them .
the lowest resolution of @xmath5=30 was chosen to match the lowest typically achieved in galaxy surveys .
we select two input models that are consistent with age and metallicity combinations occupied by some of the globular clusters in this sample and that may also be useful for high redshift galaxies .
these combinations are : 1 gyr , [ fe / h]=0.0 ; and 10 gyr , [ fe / h]=0.0 ( similar to g158 in figure [ fig12 ] at 18 gyr , [ fe / h]@xmath520.1 , @xmath5=23 ) . the results of these fits are presented in figures [ fig14]-[fig16 ] .
plots in figure [ fig14 ] show the best fit ages ( top rows ) and metallicities ( bottom rows ) as a function of s / n @xmath4 for spectral resolutions of @xmath5 = 5 , 15 , 25 ( to conserve space @xmath5 = 10 , 20 , and 30 are not shown ) for the 1and 10 gyr , [ fe / h ] = 0 inputs .
input parameters are marked by solid lines , dashed lines represent uncertainties of 10% in age and 0.1 dex in [ fe / h ] , and dotted lines mark age uncertainties of 20% .
small filled circles show individual simulations , open circles are the average derived parameters in s / n bins of width 2 , and vertical cyan bars mark the 1@xmath24 variation within each bin .
age - metallicity degeneracy in the derived parameters can clearly be seen in figure [ fig15 ] for the two input models at the same spectral resolutions presented in figure [ fig14 ] .
the increased difficulty of distinguishing older ages is also apparent from the much larger spread in derived ages for this regime where age changes result in small spectral differences .
as spectral resolution degrades , derived ages and metallicities for the lower s / n spectra tend to be driven to the model grid limits more frequently .
we choose limiting s / n values for each spectral resolution corresponding to the point at which the standard deviation in derived ages for a s / n bin begins to increase beyond 10 or 20% of the input model age .
figure [ fig16 ] summarizes the required s / n @xmath4 as a function of @xmath5 for the 1 gyr case on the left and the 10 gyr case on the right .
solid lines denote derived ages within 10% and dashed lines within 20% of the input .
recall that the s / n is calculated for these spectra by taking the standard deviation of the flux in wavelength bins along the spectrum , which has the disadvantage that spectral features increase the apparent `` noise '' in the spectrum . because of this fact , there is a limit to the maximum s / n calculated in this manner .
these artificial limits are plotted as dotted lines in figure [ fig16 ] , and therefore the open upward facing triangles are lower limits to the required s / n at those resolutions .
the 10 gyr object requires higher s / n than the 1 gyr object at almost all resolutions , implying a strong dependence on the object s age .
however , because the age is unknown a priori , the 10 gyr model must be used to determine the minimum required s /
n . the expected trend is seen for ages within 10% of the input : higher s / n is required for lower spectral resolution . at the 20% level , however , somewhat puzzling behavior is seen for the 10 gyr old object . at @xmath53 15
the required s / n actually decreases .
this does not appear to be an aritfact of the simulations , as we see no correlations between the number of points within s / n bins and the standard deviation in derived ages .
interestingly , the globular cluster g158 at @xmath5 = 23 in figure [ fig12 ] may demonstrate similar behavior . although only one noise realization was performed at each s / n level there , its correct age is obtained at much lower s / n than the younger clusters in the plot .
perhaps this trend is real .
one could envision a scenario in which with decreasing resolution model mismatches of narrow lines in the spectrum become less important than broad features typical of old spectra , such as the 4000 break and the mg absorption trough . in order to get the correct age ,
high s / n is required , but once the allowed uncertainty is increased , the broader features can provide enough information in lower s / n spectra . clearly , this is only an initial guide for required s / n and more investigation needs to be done into the cause of this unexpected feature . at the high resolution
end , our results are consistent with those of @xcite who also fit continuum - removed spectra ( via a data compression algorithm ) and find required s / n@xmath611 - 17 @xmath4 ( 20 - 30 per 3 pixel ) at a spectral resolution of 3 .
extrapolating our 5 point down to 3 would give s / n@xmath67 - 22 @xmath4 .
in general , at the spectral resolutions of the globular clusters in figure [ fig12 ] , @xmath54 13 - 23 , our simulations indicate that a s / n @xmath55 25 @xmath4 is required for a 20% age uncertainty and @xmath55 45 @xmath4 for a 10% uncertainty .
this is higher than the s / n@xmath65 @xmath4 obtained from adding noise to actual cluster spectra .
this discrepancy is not currently understood and will be investigated further in future work .
in [ line_ind ] we looked at the results of fits to specific regions of the spectrum around typically used line indices . here
we test the broader effects of the regions blueward and redward of the 4000 break on derived ages and metallicities . this will be important for fitting galaxies at varying redshifts , since at z@xmath61 optical spectra typically reach just beyond the 4000 break in the galaxy s rest frame .
for the clusters in our sample that include near ultraviolet spectra , we analyze the range of 3200 - 5650 , which is the maximum extent common to all clusters in this group . within this range ,
we fit models to the blue ( @xmath56 4000 ) , the red ( @xmath57 4000 ) , and the region right around the break ( 3750 - 4250 , as defined in the empirical calibration of the break by @xcite ) .
the clusters in this study are listed in table [ red_blue_id ] with their ages derived from different parts of the spectrum .
correlations of ages from the different spectral regions are plotted in figure [ fig17 ]
. fits to different regions give different results . for cn spectrum
fits , the red part of the spectrum drives the result .
the red ages correlate with the ages from the entire spectrum and are slightly older than the values in the literature .
the blue region gives younger ages than both the red region and the literature .
ages from the break region correlate to the red ages with about the same amount of scatter as we saw in the red - entire spectrum correlation . since the continuum is normalized out for these fits , the break region is left with only the signatures of strong features like the ca h and k lines and h@xmath8 , similar to the d@xmath42(4000)-h@xmath8 analysis done by @xcite on sdss galaxies .
the rough agreement of ages from this region with those from the entire spectrum emphasizes the importance for the 4000 break to be in the spectrum used for estimating the age of an object .
the only correlation between [ fe / h ] is a weak one between the entire spectrum and the red spectrum ( not shown in figure [ fig17 ] ) , not surprising since most metallicity sensitive lines typically used are in the red part of the spectrum . for full spectrum fits ,
the trends change and the blue region carries more weight , demonstrating the importance of information in the continuum shape of the blue part of the spectrum .
this is especially true for young objects with strong balmer discontinuities .
ages from the blue and red regions agree , but with some scatter . however , now the entire spectrum agrees equally well with the blue and the red regions .
the 4000 break ages agree much better with the blue than with the red ages .
the red , blue , and break ages all correlate with the literature ages , but the correlation is tightest for the break region , even more so than for the full spectrum fits over the entire range .
there is no correlation between our [ fe / h ] derived from the different regions , but the red [ fe / h ] has a weak correlation with the literature . because of the importance of the continuum information blueward of the 4000 break
, we also test hybrid fits that use the full spectrum over @xmath56 4250 and the cn spectrum over @xmath57 3750 .
wavelengths below 4000 will not be affected by the tp - agb and carbon stars that dominate the near infrared , so as long as there are no continuum problems with the data this type of fit should be valid .
we expected this approach to do better on very young objects .
however , the hybrid fits match the full spectrum fits that used the entire wavelength range .
the continuum shape dominates the fits and spreads out the age correlation with the literature for objects younger than @xmath60.1 gyr .
this is the same problem seen earlier with the continua of very young clusters being redder than the models , causing older best - fitting ages . in summary ,
the continuum shape contains a lot of age information . in the blue part of the spectrum
the continuum is more important than the lines for deriving age , while in the red part of the spectrum the lines alone are sufficient to derive the age .
the most robust age estimates are obtained by using the entire cn spectrum of an object ( @xmath6 3200 - 8200 ) , but the 4000 break is an essential feature to be contained within that spectrum .
when using cn spectra to avoid problems in model and data continua , ages from only the blue region are younger than those obtained by the red region or the entire range .
age biases introduced by using different regions of the spectrum must be considered when combining objects that have different spectral coverage or different redshifts that change the rest frame wavelength coverage significantly .
additionally , for objects that can not be considered a single stellar population , the ages determined from different spectral ranges may not be straighforward to interpret .
nonetheless , near infrared spectra of high redshift galaxies will be important to provide rest frame wavelengths around and redward of the 4000 break , particularly when fitting their cn spectra .
the effects of tp - agb stars on the near infrared luminosity of objects in which they reside have long been known and discussed by groups creating stellar population synthesis models @xcite .
recent models by @xcite incorporate tp - agb stars through an empirical spectral library of c- and o - type stars by @xcite .
the effects of these stars have been reanalyzed and calibrated on mc cluster colors in the new models .
they show that over 80% of the k band light can be contributed by agb stars between the ages of 0.1 and 2 gyr for solar and lower metallicities .
this effect is down to 20% in the v band .
maraston fits photometric seds of these models to mc cluster seds .
two objects in common with our sample , ngc 1783 and ngc 419 , show discrepancies between the maraston models that best fit the photometric seds and bc03 models that best fit the spectra .
these inconsistencies are the very point being address by the new models , which show that younger ages may be obtained for objects that have a near infrared excess due to tp - agb stars if these stars are properly included
. however , analysis of the spectra may reveal a different story .
the cn spectra of the two clusters ( shown in figure [ fig18]b ) are very similar to each other .
ngc 1783 has d@xmath42(4000)=1.295 and ngc 419 has d@xmath42(4000)=1.246 .
we derive similar ages for the two clusters by fitting their cn spectra : 1.5 gyr and [ fe / h][email protected] for ngc 1783 , and 1.5 gyr and [ fe / h][email protected] for ngc 419 .
the maraston results for ngc 1783 are 0.3 gyr and [ fe / h][email protected] , and for ngc 419 are 1 gyr and [ fe / h][email protected] .
bc03 models are overplotted on the cn spectra in figure [ fig18]c , d for all the best - fitting models .
the older age of 1.5 gyr is a better fit to the cn spectra of both clusters .
our photometric fit age for ngc 1783 is 2.0 gyr and [ fe / h][email protected] , and for ngc 419 is 20 gyr and [ fe / h][email protected] .
this much older photometric age for ngc 419 could be due to the presence of tp - agb stars making the continuum redder .
the near infrared photometry of ngc 419 ( see figure [ fig3 ] ) does show an excess over best - fitting models to the spectra . it is likely that the bc03 models do not include the enough tp - agb stars in this case . on the other hand ,
the presence of agb stars was detected in ngc 1783 by @xcite , but apparently their inclusion in the bc03 models is sufficient for this case .
we conclude that the detailed information contained in spectra is necessary for breaking degeneracies that can occur when using only broadband photometry through the near infrared . in a more recent paper @xcite
apply the models with enhanced tp - agb stars to high redshift galaxies and find that some degeneracies , particularly between tp - agb stars and reddening due to dust , can be broken when mid - infrared photometry is added from spitzer .
these new models look very promising for young objects if infrared data are available , however their spectral resolution is not high enough to fit to galaxy spectra .
the ages expected for early - type galaxies at z@xmath61 fall within the range spanned by the globular clusters in this sample .
the youngest early - type galaxies at these redshifts will likely be field galaxies that were the products of mergers . if the bulk of stars in these galaxies were formed in mergers at redshifts of [email protected] - 2 , which are the formation redshifts found for field ellipticals @xcite , they would be at least 1.5 gyr old by z@xmath61 ( for @xmath58 = 0.3 , @xmath59=0.7 , @xmath60=70 km s@xmath4mpc@xmath4 ) .
our cn spectrum fits estimate ages to within 0.16 dex ( 35% ) of the literature values for globular clusters older than 1 gyr in our sample .
the metallicities of z@xmath61 galaxies will likely be higher than many of the clusters in this sample , however , a few clusters spanned most of the range expected for galaxies . at redshifts of 0.3@[email protected] , galaxies in goods -
n have metallicities from 0.3 to 2.5 times solar , or @xmath20.5 @xmath0 [ fe / h ] @xmath0 + 0.4 @xcite .
we were able to determine the [ fe / h ] of globular clusters older than 1 gyr to within 0.12 dex over the range of @xmath21.6 @xmath0 [ fe / h ] @xmath0 + 0.3 .
the only region we have not tested is the highest metallicities of [ fe / h ] @xmath30 + 0.3 .
spectra at a resolution of @xmath5 = 15 ( r @xmath6 400 at 6000 ) must have s / n @xmath6 30 - 45 @xmath4 in order to derive ages with 20 or 10% uncertainties , respectively , for old galaxies and s / n @xmath6 10 - 20 @xmath4 for young galaxies .
this is higher than typically achieved for many objects in modern galaxy surveys such as munics , deep , and tkrs .
our initial checks on tkrs spectra from deimos on keck of goods - n galaxies at [email protected] - 1.0 with magnitudes of r=23 - 23.5 find s / n @xmath6 5 @xmath4 for individual objects .
therefore , most galaxies will require coaddition of a few similar objects to reach sufficient s / n . for galaxies at z@xmath61 typical optical spectra covering @xmath64000 - 9000
will provide rest frame wavelengths of @xmath62000 - 4500 .
our tests have shown that when using cn spectra , data to longer wavelengths will provide more robust age constraints .
this means that near infrared spectra of galaxies will be important data to obtain for accurate age estimates .
many early - type galaxies appear to have @xmath13-element abundance ratios enhanced above solar @xcite .
this enhancement is a sign of short duration bursts of star formation that may well be typical for early - type galaxies formed during mergers , as we expect to find at z@xmath61 .
we see indications of @xmath13-enhancement in some of the m31 globular clusters in this sample and we still derive ages within 35% of the literature values when fitting the entire cn spectra .
this suggests that results from the entire continuum - normalized spectrum , effectively averaging over all the lines , may be less affected by @xmath13-enhancement than detailed line index methods that rely on only a few affected lines in the spectrum .
@xcite find that broadband colors in the blue are significantly affected by @xmath13-enhancement , but removal of the continuum shape must mitigate these effects to some extent .
model fitting techniques will have to be modified for galaxies .
simple stellar populations will no longer be sufficient descriptions of these complex systems .
fits will require models with complex star formation histories , multiple stellar populations , and dust .
however , this work has verified that the reliable regions of model parameter space cover most of that expected to be occupied by galaxies .
we have also proven techniques by which model fits to galaxy spectra can be expected to work .
we have investigated the utility of using the @xcite high resolution simple stellar population synthesis models for estimating the ages and metallicities of globular clusters .
we chose clusters from m31 and the magellanic clouds that cover a wide range of age and metallicity values , @xmath6 0.004@xmath0t@xmath1(gyr)@xmath020 and @xmath21.6@xmath0[fe / h]@xmath0 + 0.3 , to explore model performance in different regions of parameter space .
we tested various techniques of fitting the models to data using spectra with and without their continua and photometry .
results were compared to age and metallicity estimates in the literature from other methods to evaluate our fitting techniques .
the models are reliable in regions of parameter space where [email protected] gyr and [ fe / h ] @xmath620.7 .
these regions are limited by the incompleteness of the stellar libraries used in the models . for objects between the ages of 0.1 and 2.5 gyr , uncertainties in the amount of near infrared light coming from tp - agb and carbon stars
make it hard to match the continuum shape of individual globular clusters .
red supergiant stars affect clusters younger than 0.1 gyr .
continuum - normalized spectra can be used to get around the tp - agb star problem , as well as any flux calibration issues in the data , with the requirement that high spectral resolution models are used .
the red supergiant star problem limits accurate age estimates to objects older than @xmath60.1 gyr .
ages from our cn spectrum model fits best match the entire range of age estimates from the literature ( figure [ fig5 ] and table [ rms_age_tbl ] ) , with average differences of 0.16 dex ( 35% ) for clusters older than 1 gyr and 0.3 dex ( 63% ) for younger clusters .
the methods that depend on the continuum shape the full spectrum , continuum , and photometry fits overestimate the ages of many clusters below 1 gyr .
we believe this to be due to a near infrared excess from tp - agb stars making the cluster spectra redder than the models and forcing older aged models to provide the best fits for cluster ages of 0.1 - 1 gyr .
red supergiant stars similarly affect clusters younger than 0.1 gyr . although photometric fits seem to do well for clusters older than 1 gyr ( 0.1 dex or 23% age error relative to the literature ) , the cn spectrum fits are better for simultaneously providing both age and metallicity for these older objects
. fits to photometry alone can not reproduce the correct metallicities . comparing the fitting methods ( figure [ fig4 ] )
, we find a tighter correlation between ages derived from the full spectrum and continuum fits than from the full spectrum and cn spectrum fits for clusters younger than 1 gyr , suggesting that these derived ages are more strongly driven by the continuum shape than by the spectral lines .
clusters older than 1 gyr show a tighter correlation between ages derived from the cn spectrum and photometry fits .
because the derived ages of young clusters are so strongly influenced by the continua and we have seen signs of problems in the models matching the cluster continua , we conclude that the best method for deriving accurate ages for globular clusters , especially when simultaneously determining metallicity , is fitting models to their continuum - normalized spectra . on metallicities of clusters older than 1 gyr , we agree to 0.12 dex with the literature values that are based on spectra .
those based on colors have much more scatter , which suggests that cn spectrum fits are more robust in estimating [ fe / h ] than methods using colors .
we can not accurately derive metallicity estimates for clusters younger than @xmath6 1 gyr .
blueward of the 4000 break the continuum shape is more important than the lines for deriving age .
redward of the break the lines alone in a continuum - normalized spectrum are sufficient to derive the age .
many of the globular clusters tested here have spectra out to rest wavelengths of 1 @xmath3 m .
if fitting models to continuum - normalized spectra of z@xmath61 galaxies , this means that near infrared spectra through k band will be important to provide similar spectral coverage and constraints on age and metallicity .
the 4000 break is a very important feature to be contained in the spectrum .
it will be contained in typical z@xmath61 spectra , but higher redshifts may require near infrared spectra to correctly constrain ages and metallicites .
we should note , however , that although a strong age indicator for globular clusters and early type galaxies , the 4000 break can be significantly affected by dust in star forming galaxies .
for example , @xcite find dust imparted age errors as large as 9.5 gyr for a 0.5 gyr bc03 model ssp when ages are derived from the d@xmath32(4000)-fe4668 plane .
the spectral s / n required to derive ages by fitting the entire cn spectrum is similar to the 30 - 40 @xmath4 typically required when using line indices on old objects . for a 10 gyr
old object at @xmath5 = 5 - 25 our required s / n is 15 - 30 @xmath4 for a 20% age uncertainty and 30 - 55 @xmath4 for a 10% uncertainty .
a 1 gyr year old object requires s / n @xmath6 3 - 15 @xmath4 for a 20% age uncertainty and 10 - 25 @xmath4 for a 10% uncertainty . at @xmath5
= 30 both ages require s / n @xmath55 60 . fairly low resolution , r@xmath6240 - 1200 , spectra can be used , at the cost of increasing required s / n with decreasing resolution .
we have successfully tested most of the age - metallicity space expected to be occupied by galaxies at z@xmath61 , except for the highest metallicities of [ fe / h]@xmath30 + 0.3 .
fitting the continuum - normalized spectra of these galaxies should work well for spectra without accurate flux calibration , typical when faint galaxies are observed on multiple masks over long periods of time during surveys .
the required s / n and spectral resolution for determining ages and metallicities matches that typically achieved in recent galaxy surveys , as long as a few like objects are coadded to boost the s / n .
the techniques presented here will be applied to galaxies in future work .
bell , e.f . ,
mcintosh , d.h . ,
barden , m. , wolf , c. , caldwell , j.a.r . , rix , h.w . ,
beckwith , s.v.w .
, borch , a. , hussler , b. , jahnke , k. , jogee , s. , meisenheimer , k. , peng , c. , sanchez , s.f . , somerville , r.s . ,
wisotzki , l. , 2004b , , 600 , l11 cimatti , a. , daddi , e. , mignoli , m. , pozzetti , l. , renzini , a. , zamorani , g. , broadhurst , t. , fontana , a. , saracco , p. , poli , f. , cristiani , s. , dodorico , s. , giallongo , e. , gilmozzi , r. , menci , n. , 2002 , , 381 , l68 daddi , e. , cimatti , a. , broadhurst , t. , renzini , a. , zamorani , g. , mignoli , m. , saracco , p. , fontana , a. , pozzetti , l. , poli , f. , cristiani , s. , dodorico , s. , giallongo , e. , gilmozzi , r. , menci , n. , 2002 , , 384 , l1 hogg , d.w . , blanton , m. , strateva , i. , bahcall , n.a .
, brinkmann , j. , csabai , i. , mamoru , d. , fukugita , m. , hennessy , g. , ivezic , z. , knapp , g.r . , lamb , d.q . ,
lupton , r. , munn , j.a . ,
nichol , r. , schlegel , d.j . ,
schneider , d.p . ,
york , d.g .
, 2002 , , 124 , 646 mccarthy , p.j .
, le borgne , d. , crampton , d. , chen , h.w . ,
abraham , r.g . ,
glazebrook , k. , savaglio , s. , carlberg , r.g . ,
marzke , r.o . , roth , k. , jrgensen , i. , hook , i. , murowinski , r. , juneau , s. , 2004 , , 614 , 9 rudnick , g. , rix , h.w .
, franx , m. , labb , i. , blanton , m. , daddi , e. , frster schreiber , n.m . , moorwood , a. , rttgering , h. , trujillo , i. , van de wel , a. , van der werf , p. , van dokkum , p.g . , van starkenburg , l. , 2003 , , 599 , 847 frster schreiber , n.m .
, van dokkum , p.g . , franx , m. , labbe , i. , rudnick , g. , daddi , e. , illingworth , g.d .
, kriek , m. , moorwood , a.f.m . ,
rix , h .- w . , rottgering , h. , trujillo , i. , van der werf , p. , van starkenburg , l. , wuyts , s. , 2004 , , 616 , 40 strateva , i. , ivezi , z. , knapp , g.r . ,
narayanan , v.k . ,
strauss , m.a . ,
gunn , j.e . ,
lupton , r.h . ,
schlegel , d. , bahcall , n.a . , brinkmann , j. , brunner , r.j . , budavri , t. , csabai , i. , castander , f.j . , doi , mamoru , fukugita , m. , gyry , z. , hamabe , m. , hennessy , g. , ichikawa , t. , kunszt , p.z . , lamb , d.q . , mckay , t.a . , okamura , s. , racusin , j. , sekiguchi , m. , schneider , d.p .
, shimasaku , k. , york , d. , 2001 , , 122 , 1861 wolf , c. , bell , e.f . ,
mcintosh , d.h . ,
rix , h.w . ,
barden , m. , beckwith , s.v.w . ,
borch , a. , caldwell , j.a.r . , hussler , b. , heymans , c. , jahnke , k. , jogee , s. , meisenheimer , k. , peng , c.y . ,
snchez , s.f . ,
somerville , r.s . ,
wisotzki , l. , 2005 , , 630 , 771 contours of age and metallicity for cn spectrum fits to the smc globular cluster ngc 419 .
the s / n per resolution element from top left to lower right is 99 , 22 , 12 , 8 , 6 , and 4 . the best - fitting age and [ fe / h ]
is given in the plot titles and marked by the black crosses .
the yellow circles mark the age and [ fe / h ] from the highest s / n fit .
the solid , dashed , and dotted lines are 1 , 2 , and 3@xmath24 error contours , respectively .
color bars give the @xmath23 values for 68.3% ( @xmath23=2.3 ) , 90% , 95.4% , 99% , 99.73% , and 99.99% ( @xmath23=18.4 ) confidence levels for two degrees of freedom .
( the highest level is converted to white for plot clarity . )
black dots mark the locations of models in the age - metallicity grid .
[ fig1],title="fig : " ] contours of age and metallicity for cn spectrum fits to the smc globular cluster ngc 419 .
the s / n per resolution element from top left to lower right is 99 , 22 , 12 , 8 , 6 , and 4 . the best - fitting age and [ fe / h ] is given in the plot titles and marked by the black crosses .
the yellow circles mark the age and [ fe / h ] from the highest s / n fit .
the solid , dashed , and dotted lines are 1 , 2 , and 3@xmath24 error contours , respectively .
color bars give the @xmath23 values for 68.3% ( @xmath23=2.3 ) , 90% , 95.4% , 99% , 99.73% , and 99.99% ( @xmath23=18.4 )
confidence levels for two degrees of freedom .
( the highest level is converted to white for plot clarity . )
black dots mark the locations of models in the age - metallicity grid .
[ fig1],title="fig : " ] contours of age and metallicity for cn spectrum fits to the smc globular cluster ngc 419 .
the s / n per resolution element from top left to lower right is 99 , 22 , 12 , 8 , 6 , and 4 . the best - fitting age and [ fe / h ]
is given in the plot titles and marked by the black crosses .
the yellow circles mark the age and [ fe / h ] from the highest s / n fit .
the solid , dashed , and dotted lines are 1 , 2 , and 3@xmath24 error contours , respectively .
color bars give the @xmath23 values for 68.3% ( @xmath23=2.3 ) , 90% , 95.4% , 99% , 99.73% , and 99.99% ( @xmath23=18.4 ) confidence levels for two degrees of freedom .
( the highest level is converted to white for plot clarity . )
black dots mark the locations of models in the age - metallicity grid .
[ fig1],title="fig : " ] contours of age and metallicity for cn spectrum fits to the smc globular cluster ngc 419 .
the s / n per resolution element from top left to lower right is 99 , 22 , 12 , 8 , 6 , and 4 . the best - fitting age and [ fe / h ] is given in the plot titles and marked by the black crosses .
the yellow circles mark the age and [ fe / h ] from the highest s / n fit .
the solid , dashed , and dotted lines are 1 , 2 , and 3@xmath24 error contours , respectively .
color bars give the @xmath23 values for 68.3% ( @xmath23=2.3 ) , 90% , 95.4% , 99% , 99.73% , and 99.99% ( @xmath23=18.4 ) confidence levels for two degrees of freedom .
( the highest level is converted to white for plot clarity . )
black dots mark the locations of models in the age - metallicity grid .
[ fig1],title="fig : " ] contours of age and metallicity for cn spectrum fits to the smc globular cluster ngc 419 .
the s / n per resolution element from top left to lower right is 99 , 22 , 12 , 8 , 6 , and 4 . the best - fitting age and [ fe / h ] is given in the plot titles and marked by the black crosses .
the yellow circles mark the age and [ fe / h ] from the highest s / n fit .
the solid , dashed , and dotted lines are 1 , 2 , and 3@xmath24 error contours , respectively .
color bars give the @xmath23 values for 68.3% ( @xmath23=2.3 ) , 90% , 95.4% , 99% , 99.73% , and 99.99% ( @xmath23=18.4 ) confidence levels for two degrees of freedom .
( the highest level is converted to white for plot clarity . )
black dots mark the locations of models in the age - metallicity grid .
[ fig1],title="fig : " ] contours of age and metallicity for cn spectrum fits to the smc globular cluster ngc 419 .
the s / n per resolution element from top left to lower right is 99 , 22 , 12 , 8 , 6 , and 4 . the best - fitting age and [ fe / h ] is given in the plot titles and marked by the black crosses .
the yellow circles mark the age and [ fe / h ] from the highest s / n fit .
the solid , dashed , and dotted lines are 1 , 2 , and 3@xmath24 error contours , respectively .
color bars give the @xmath23 values for 68.3% ( @xmath23=2.3 ) , 90% , 95.4% , 99% , 99.73% , and 99.99% ( @xmath23=18.4 ) confidence levels for two degrees of freedom .
( the highest level is converted to white for plot clarity . )
black dots mark the locations of models in the age - metallicity grid .
[ fig1],title="fig : " ] derived from our different types of model fits compared to the literature .
a&b ) are from the cn spectrum fits for old and young clusters separately , c ) is from the full spectrum fits , d ) is from the photometry fits , and e ) is from the continuum fits .
literature values are from comparison values in @xcite ( which come from @xcite , @xcite , @xcite , @xcite , and @xcite ) in the mc ( _ black crosses _ ) , line indices by @xcite in m31 ( _ blue pluses _ ) , line index ratios by @xcite in the mc ( _ red circles _ ) , vjk colors by @xcite in m31 ( _ green triangles _ ) , photmetric fits to bc96 models by @xcite in m31 ( _ magenta squares _ ) , and literature [ fe / h ] on a homogeneous scale by @xcite in the mc ( _ black inverted triangles _ ) .
large symbols denote literature values based on spectra and small symbols are values based on colors .
the thick horizontal cyan lines connect points that are the same cluster with multiple literature metallicities .
thin horizontal lines are error bars on literature [ fe / h ] , thin vertical lines are 1@xmath24 error bars on our derived values .
the best agreement with literature metallicity occurs for spectra - based values of clusters older than 1 gyr from our cn spectrum fits .
[ fig8],title="fig : " ] derived from our different types of model fits compared to the literature .
a&b ) are from the cn spectrum fits for old and young clusters separately , c ) is from the full spectrum fits , d ) is from the photometry fits , and e ) is from the continuum fits .
literature values are from comparison values in @xcite ( which come from @xcite , @xcite , @xcite , @xcite , and @xcite ) in the mc ( _ black crosses _ ) , line indices by @xcite in m31 ( _ blue pluses _ ) , line index ratios by @xcite in the mc ( _ red circles _ ) , vjk colors by @xcite in m31 ( _ green triangles _ ) , photmetric fits to bc96 models by @xcite in m31 ( _ magenta squares _ ) , and literature [ fe / h ] on a homogeneous scale by @xcite in the mc ( _ black inverted triangles _ ) .
large symbols denote literature values based on spectra and small symbols are values based on colors .
the thick horizontal cyan lines connect points that are the same cluster with multiple literature metallicities .
thin horizontal lines are error bars on literature [ fe / h ] , thin vertical lines are 1@xmath24 error bars on our derived values .
the best agreement with literature metallicity occurs for spectra - based values of clusters older than 1 gyr from our cn spectrum fits .
[ fig8],title="fig : " ] derived from our different types of model fits compared to the literature .
a&b ) are from the cn spectrum fits for old and young clusters separately , c ) is from the full spectrum fits , d ) is from the photometry fits , and e ) is from the continuum fits .
literature values are from comparison values in @xcite ( which come from @xcite , @xcite , @xcite , @xcite , and @xcite ) in the mc ( _ black crosses _ ) , line indices by @xcite in m31 ( _ blue pluses _ ) , line index ratios by @xcite in the mc ( _ red circles _ ) , vjk colors by @xcite in m31 ( _ green triangles _ ) , photmetric fits to bc96 models by @xcite in m31 ( _ magenta squares _ ) , and literature [ fe / h ] on a homogeneous scale by @xcite in the mc ( _ black inverted triangles _ ) .
large symbols denote literature values based on spectra and small symbols are values based on colors .
the thick horizontal cyan lines connect points that are the same cluster with multiple literature metallicities .
thin horizontal lines are error bars on literature [ fe / h ] , thin vertical lines are 1@xmath24 error bars on our derived values .
the best agreement with literature metallicity occurs for spectra - based values of clusters older than 1 gyr from our cn spectrum fits .
[ fig8],title="fig : " ] derived from our different types of model fits compared to the literature .
a&b ) are from the cn spectrum fits for old and young clusters separately , c ) is from the full spectrum fits , d ) is from the photometry fits , and e ) is from the continuum fits .
literature values are from comparison values in @xcite ( which come from @xcite , @xcite , @xcite , @xcite , and @xcite ) in the mc ( _ black crosses _ ) , line indices by @xcite in m31 ( _ blue pluses _ ) , line index ratios by @xcite in the mc ( _ red circles _ ) , vjk colors by @xcite in m31 ( _ green triangles _ ) , photmetric fits to bc96 models by @xcite in m31 ( _ magenta squares _ ) , and literature [ fe / h ] on a homogeneous scale by @xcite in the mc ( _ black inverted triangles _ ) .
large symbols denote literature values based on spectra and small symbols are values based on colors .
the thick horizontal cyan lines connect points that are the same cluster with multiple literature metallicities .
thin horizontal lines are error bars on literature [ fe / h ] , thin vertical lines are 1@xmath24 error bars on our derived values .
the best agreement with literature metallicity occurs for spectra - based values of clusters older than 1 gyr from our cn spectrum fits .
[ fig8],title="fig : " ] derived from our different types of model fits compared to the literature .
a&b ) are from the cn spectrum fits for old and young clusters separately , c ) is from the full spectrum fits , d ) is from the photometry fits , and e ) is from the continuum fits .
literature values are from comparison values in @xcite ( which come from @xcite , @xcite , @xcite , @xcite , and @xcite ) in the mc ( _ black crosses _ ) , line indices by @xcite in m31 ( _ blue pluses _ ) , line index ratios by @xcite in the mc ( _ red circles _ ) , vjk colors by @xcite in m31 ( _ green triangles _ ) , photmetric fits to bc96 models by @xcite in m31 ( _ magenta squares _ ) , and literature [ fe / h ] on a homogeneous scale by @xcite in the mc ( _ black inverted triangles _ ) .
large symbols denote literature values based on spectra and small symbols are values based on colors .
the thick horizontal cyan lines connect points that are the same cluster with multiple literature metallicities .
thin horizontal lines are error bars on literature [ fe / h ] , thin vertical lines are 1@xmath24 error bars on our derived values .
the best agreement with literature metallicity occurs for spectra - based values of clusters older than 1 gyr from our cn spectrum fits .
[ fig8],title="fig : " ] -[mgfe]@xmath31 model indices calculated at different spectral resolutions .
grid age lines ( _ solid _ ) from bottom to top are 20 , 10 , 5 , 2 , 1 , 0.1 , 0.01 , and 0.001 gyr , and [ fe / h ] lines ( _ dashed _ ) from right to left are + 0.5 , 0.0 , -0.5 , -1.0 , and -2.0 ( these are labeled in panel a ) .
note that the model indices become degenerate at @xmath47 100 myr for -1.0 @xmath0 [ fe / h ] @xmath0 + 0.5 , and at @xmath47 1 gyr for [ fe / h ] @xmath0 -2.0 .
cluster indices are plotted as symbols , differientiated by spectral resolution : 6 ( _ squares _ ,
m31 ) , 11 ( _ triangles _ , mc ) , 12 ( _ circles _ , mc ) , 16 ( _ diamonds _ , mc ) , and 23 ( _ crosses _ , m31 ) .
e ) contains grids of the traditional lick indices provided with the bc03 model package .
they suffer from similar degeneracy at low ages and metallicities .
f ) shows an indication of [ mg / fe ] enhancement in some of the m31 clusters .
model grids for indices at the minimum and maximum spectral resolutions are included .
[ fig10],title="fig : " ] -[mgfe]@xmath31 model indices calculated at different spectral resolutions .
grid age lines ( _ solid _ ) from bottom to top are 20 , 10 , 5 , 2 , 1 , 0.1 , 0.01 , and 0.001 gyr , and [ fe / h ] lines ( _ dashed _ ) from right to left are + 0.5 , 0.0 , -0.5 , -1.0 , and -2.0 ( these are labeled in panel a ) .
note that the model indices become degenerate at @xmath47 100 myr for -1.0 @xmath0 [ fe / h ] @xmath0 + 0.5 , and at @xmath47 1 gyr for [ fe / h ] @xmath0 -2.0 .
cluster indices are plotted as symbols , differientiated by spectral resolution : 6 ( _ squares _ ,
m31 ) , 11 ( _ triangles _ , mc ) , 12 ( _ circles _ , mc ) , 16 ( _ diamonds _ , mc ) , and 23 ( _ crosses _ , m31 ) .
e ) contains grids of the traditional lick indices provided with the bc03 model package .
they suffer from similar degeneracy at low ages and metallicities .
f ) shows an indication of [ mg / fe ] enhancement in some of the m31 clusters .
model grids for indices at the minimum and maximum spectral resolutions are included .
[ fig10],title="fig : " ] -[mgfe]@xmath31 model indices calculated at different spectral resolutions .
grid age lines ( _ solid _ ) from bottom to top are 20 , 10 , 5 , 2 , 1 , 0.1 , 0.01 , and 0.001 gyr , and [ fe / h ] lines ( _ dashed _ ) from right to left are + 0.5 , 0.0 , -0.5 , -1.0 , and -2.0 ( these are labeled in panel a ) .
note that the model indices become degenerate at @xmath47 100 myr for -1.0 @xmath0 [ fe / h ] @xmath0 + 0.5 , and at @xmath47 1 gyr for [ fe / h ] @xmath0 -2.0 .
cluster indices are plotted as symbols , differientiated by spectral resolution : 6 ( _ squares _ ,
m31 ) , 11 ( _ triangles _ , mc ) , 12 ( _ circles _ , mc ) , 16 ( _ diamonds _ , mc ) , and 23 ( _ crosses _ , m31 ) .
e ) contains grids of the traditional lick indices provided with the bc03 model package .
they suffer from similar degeneracy at low ages and metallicities .
f ) shows an indication of [ mg / fe ] enhancement in some of the m31 clusters .
model grids for indices at the minimum and maximum spectral resolutions are included .
[ fig10],title="fig : " ] -[mgfe]@xmath31 model indices calculated at different spectral resolutions .
grid age lines ( _ solid _ ) from bottom to top are 20 , 10 , 5 , 2 , 1 , 0.1 , 0.01 , and 0.001 gyr , and [ fe / h ] lines ( _ dashed _ ) from right to left are + 0.5 , 0.0 , -0.5 , -1.0 , and -2.0 ( these are labeled in panel a ) .
note that the model indices become degenerate at @xmath47 100 myr for -1.0 @xmath0 [ fe / h ] @xmath0 + 0.5 , and at @xmath47 1 gyr for [ fe / h ] @xmath0 -2.0 .
cluster indices are plotted as symbols , differientiated by spectral resolution : 6 ( _ squares _ ,
m31 ) , 11 ( _ triangles _ , mc ) , 12 ( _ circles _ , mc ) , 16 ( _ diamonds _ , mc ) , and 23 ( _ crosses _ , m31 ) .
e ) contains grids of the traditional lick indices provided with the bc03 model package .
they suffer from similar degeneracy at low ages and metallicities .
f ) shows an indication of [ mg / fe ] enhancement in some of the m31 clusters .
model grids for indices at the minimum and maximum spectral resolutions are included .
[ fig10],title="fig : " ] -[mgfe]@xmath31 model indices calculated at different spectral resolutions .
grid age lines ( _ solid _ ) from bottom to top are 20 , 10 , 5 , 2 , 1 , 0.1 , 0.01 , and 0.001 gyr , and [ fe / h ] lines ( _ dashed _ ) from right to left are + 0.5 , 0.0 , -0.5 , -1.0 , and -2.0 ( these are labeled in panel a ) .
note that the model indices become degenerate at @xmath47 100 myr for -1.0 @xmath0 [ fe / h ] @xmath0 + 0.5 , and at @xmath47 1 gyr for [ fe / h ] @xmath0 -2.0 .
cluster indices are plotted as symbols , differientiated by spectral resolution : 6 ( _ squares _ ,
m31 ) , 11 ( _ triangles _ , mc ) , 12 ( _ circles _ , mc ) , 16 ( _ diamonds _ , mc ) , and 23 ( _ crosses _ , m31 ) .
e ) contains grids of the traditional lick indices provided with the bc03 model package .
they suffer from similar degeneracy at low ages and metallicities .
f ) shows an indication of [ mg / fe ] enhancement in some of the m31 clusters .
model grids for indices at the minimum and maximum spectral resolutions are included .
[ fig10],title="fig : " ] -[mgfe]@xmath31 model indices calculated at different spectral resolutions .
grid age lines ( _ solid _ ) from bottom to top are 20 , 10 , 5 , 2 , 1 , 0.1 , 0.01 , and 0.001 gyr , and [ fe / h ] lines ( _ dashed _ ) from right to left are + 0.5 , 0.0 , -0.5 , -1.0 , and -2.0 ( these are labeled in panel a ) .
note that the model indices become degenerate at @xmath47 100 myr for -1.0 @xmath0 [ fe / h ] @xmath0 + 0.5 , and at @xmath47 1 gyr for [ fe / h ] @xmath0 -2.0 .
cluster indices are plotted as symbols , differientiated by spectral resolution : 6 ( _ squares _ ,
m31 ) , 11 ( _ triangles _ , mc ) , 12 ( _ circles _ , mc ) , 16 ( _ diamonds _ , mc ) , and 23 ( _ crosses _ , m31 ) .
e ) contains grids of the traditional lick indices provided with the bc03 model package .
they suffer from similar degeneracy at low ages and metallicities .
f ) shows an indication of [ mg / fe ] enhancement in some of the m31 clusters .
model grids for indices at the minimum and maximum spectral resolutions are included .
[ fig10],title="fig : " ] .
s / n=5 is marked by vertical lines .
the clusters shown are g158 ( _ squares , solid red line _ ) , ngc 419 ( _ pluses , dashed green line _ ) , ngc 2134 ( _ triangles , dotted blue line _ ) , and ngc 1818 ( _ circles , dash - dot magenta line _ ) .
the horizontal lines mark the average age or metallicity of each cluster for the s / n@xmath405 points .
the spectral resolution of these clusters is 23 for g158 and 13 for the others .
the vertical bars on the right show the spread of literature values for each cluster , in the same line types and colors . if the 0.45 dex [ fe / h ] scale offset is subtracted from the mc cluster ( ngc 419 , 2134 , 1818 ) literature values , then our derived [ fe / h ] for ngc 419 falls within the literature range .
our [ fe / h ] values for the two younger clusters do not .
[ fig12],title="fig : " ] .
s / n=5 is marked by vertical lines .
the clusters shown are g158 ( _ squares , solid red line _ ) , ngc 419 ( _ pluses , dashed green line _ ) , ngc 2134 ( _ triangles , dotted blue line _ ) , and ngc 1818 ( _ circles , dash - dot magenta line _ ) .
the horizontal lines mark the average age or metallicity of each cluster for the s / n@xmath405 points .
the spectral resolution of these clusters is 23 for g158 and 13 for the others .
the vertical bars on the right show the spread of literature values for each cluster , in the same line types and colors . if the 0.45 dex [ fe / h ] scale offset is subtracted from the mc cluster ( ngc 419 , 2134 , 1818 ) literature values , then our derived [ fe / h ] for ngc 419 falls within the literature range .
our [ fe / h ] values for the two younger clusters do not .
[ fig12],title="fig : " ] * lmc * + h88 - 267 & & & & 3500 & 5870 & 16 & & & & 1 + hd269828 & & & & 3500 & 5870 & 16 & & & & 1 + hd32228 & & & & 3500 & 5812 & 16 & & & & 1 + ic2105 & & & & 3500 & 4704 & 12 & & & & 1 + ic2111 & & & & 3500 & 4704 & 12 & & & & 1 + ic2116 & & & & 3500 & 5878 & 16 & & & & 1 + kmhk1074 & & & & 3500 & 5878 & 16 & & & & 1 + ngc1466 & 3000 & 5652 & 15 & & & & & & & 2 + ngc1711 & & & & 3500 & 5878 & 16 & 5600 & 10000 & 14 & 1,3 + ngc1714 & & & & 3700 & 8200 & 11 & 6365 & 9809 & 12.5 & 1,4,5 + ngc1731 & & & & 3500 & 4704 & 12 & & & & 1 + ngc1735 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc1736 & & & & 3500 & 4704 & 12 & & & & 1 + ngc1743 & & & & 3500 & 4740 & 12 & & & & 1 + ngc1755 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc1761 & & & & 3500 & 5812 & 16 & & & & 1 + ngc1767 & & & & 3400 & 9700 & 12 & 5656 & 10112 & 14 & 1,3,6 + ngc1772 & & & & 3500 & 5870 & 16 & & & & 1 + ngc1774 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc1782 & & & & 3500 & 4702 & 12 & 5600 & 10000 & 14 & 1,3 + ngc1783 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & 6365 & 9809 & 12.5 & 2,4,5 + ngc1805 & & & & 3500 & 5878 & 16 & 5600 & 10000 & 14 & 1,3 + ngc1818 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc1831 & 3000 & 5652 & 15 & 3700 & 8200 & 11 & 5656 & 9800 & 14 & 2,3,5 + ngc1847 & 3000 & 5652 & 15 & 3700 & 8200 & 11 & 5656 & 9796 & 14 & 2,1,3,5 + ngc1850 & & & & 3500 & 5878 & 16 & 5600 & 10000 & 14 & 1,3 + ngc1850a & & & & 3500 & 5878 & 16 & & & & 1 + ngc1854 & & & & 3500 & 5878 & 16 & 5600 & 10000 & 14 & 1,3 + ngc1856 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & 5656 & 9800 & 14 & 2,3,5 + ngc1866 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & 5656 & 9800 & 14 & 2,3,5 + ngc1868 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & 5656 & 9800 & 14 & 2,3,5 + ngc1872 & & & & 3500 & 4704 & 12 & & & & 1 + ngc1895 & & & & 3700 & 8200 & 11 & 6365 & 9809 & 12.5 & 4,5 + ngc1903 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc1921 & & & & 3500 & 4704 & 12 & & & & 1 + ngc1935 & & & & 3500 & 4704 & 12 & & & & 1 + ngc1936 & & & & 3500 & 4702 & 12 & & & & 1 + ngc1949 & & & & 3500 & 4704 & 12 & & & & 1 + ngc1951 & & & & 3500 & 4702 & 12 & 5600 & 10000 & 14 & 1,3 + ngc1967 & & & & 3500 & 5870 & 16 & & & & 1 + ngc1970 & & & & 3500 & 4704 & 12 & & & & 1 + ngc1978 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & 6365 & 9811 & 12.5 & 2,4,5 + ngc1983 & & & & 3500 & 5874 & 16 & & & & 1 + ngc1984 & & & & 3500 & 5870 & 16 & & & & 1 + ngc1994 & & & & 3500 & 5870 & 16 & 5600 & 10000 & 14 & 1,3 + ngc2002 & & & & 3500 & 5870 & 16 & 5600 & 10000 & 14 & 1,3 + ngc2003 & & & & 3400 & 9700 & 12 & 5656 & 10112 & & 3,6 + ngc2004 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & 5656 & 9800 & 14 & 2,3,5 + ngc2006 & & & & 3500 & 5874 & 16 & 5600 & 10000 & 14 & 1,3 + ngc2009 & & & & 3500 & 5878 & 16 & & & & 1 + ngc2011 & & & & 3500 & 5870 & 16 & 5600 & 10000 & 14 & 1,3 + ngc2021 & & & & 3500 & 5878 & 16 & & & & 1 + ngc2025 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc2032 & & & & 3500 & 4704 & 12 & & & & 1 + ngc2033 & & & & 3500 & 5878 & 16 & & & & 1 + ngc2035 & & & & 3500 & 4704 & 12 & & & & 1 + ngc2041 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc2058 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc2070 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & & & & 2,5 + ngc2091 & & & & 3500 & 5878 & 16 & & & & 1 + ngc2092 & & & & 3500 & 5878 & 16 & & & & 1 + ngc2096 & & & & 3500 & 5874 & 16 & & & & 1 + ngc2098 & & & & 3500 & 5870 & 16 & 5600 & 10000 & 14 & 1,3 + ngc2100 & & & & 3700 & 8200 & 11 & 5656 & 10112 & 14 & 1,3,5 + ngc2102 & & & & 3500 & 5878 & 16 & & & & 1 + ngc2134 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc2136 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc2147 & & & & 3500 & 5878 & 16 & & & & 1 + ngc2156 & & & & 3500 & 4704 & 12 & & & & 1 + ngc2157 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & 5656 & 9800 & 14 & 2,3,5 + ngc2159 & & & & 3500 & 4704 & 12 & & & & 1 + ngc2164 & & & & 3500 & 4704 & 12 & 5600 & 10000 & 14 & 1,3 + ngc2214 & & & & 3500 & 4740 & 12 & 5600 & 10000 & 14 & 1,3 + sl153 & & & & 3500 & 4704 & 12 & & & & 1 + sl492 & & & & 3500 & 5870 & 16 & & & & 1 + sl538 & & & & 3500 & 5870 & 16 & 5600 & 10000 & 14 & 1,3 + sl586 & & & & 3500 & 5878 & 16 & & & & 1 + sl601 & & & & 3500 & 4704 & 12 & & & & 1 + sl639 & & & & 3500 & 5882 & 16 & & & & 1 + * smc * + ngc121 & & & & 3700 & 6800 & 11 & & & & 5 + ngc299 & & & & 3500 & 5877 & 16 & 5600 & 10000 & 14 & 1,3 + ngc330 & 3000 & 5656 & 15 & 3700 & 8200 & 11 & 5658 & 10116 & 14 & 2,1,3,5 + ngc419 & 3000 & 5654 & 15 & 3700 & 8200 & 11 & & & & 2,5 + * m31 * + bol104 & & & & 3200 & 6100 & 6 & & & & 7 + bol118 & & & & 3200 & 6100 & 6 & & & & 7 + bol119 & & & & 3200 & 6100 & 6 & & & & 7 + bol124 & & & & 3200 & 6100 & 6 & & & & 7 + g1 & & & & 3600 & 9900 & 23 & & & & 8 + g78 & & & & 3600 & 9900 & 23 & & & & 8 + g158 & & & & 3600 & 9900 & 23 & & & & 7,8 + g165 & & & & 3200 & 6100 & 6 & & & & 7 + g168 & & & & 3200 & 6100 & 6 & & & & 7 + g169 & & & & 3200 & 6100 & 6 & & & & 7 + g170 & & & & 3600 & 9900 & 23 & & & & 8 + g174 & & & & 3200 & 6100 & 6 & & & & 7 + g175 & & & & 3200 & 6100 & 6 & & & & 7 + g177 & & & & 3600 & 9900 & 23 & & & & 7,8 + g184 & & & & 3200 & 6100 & 6 & & & & 7 + g185 & & & & 3200 & 6100 & 6 & & & & 7 + g190 & & & & 3200 & 6100 & 6 & & & & 7 + g222 & & & & 3600 & 7934 & 8 & & & & 8 + locations & mc , m31 & mc , m31 & mc & mc & mc & mc & mc & m31 & mc , m31 & mc + references & 1,2,3,4,5,6 & 1,2,3,4,5,6 & 1 & 3,5 & 2 & 5 & 6 & 4 & 1,3,4,5,6 & 1,2,3,5,6 + no .
objects & 75 & 73 & 29 & 11 & 54 & 11 & 18 & 8 & 4 & 61 + * cn spectrum fits * + avg offset ( gyr ) & 0.84 & 0.19 & 0.12 & 0.56 & 0.03 & 0.26 & 0.59 & 4.36 & 1.63 & 0.09 + avg offset ( dex ) & 0.34 & 0.29 & 0.36 & 0.21 & 0.29 & 0.22 & 0.31 & 0.25 & 0.16 & 0.30 + rms ( gyr ) & 3.18 & 1.00 & 0.82 & 1.06 & 0.12 & 0.79 & 1.82 & 8.34 & 3.78 & 0.35 + rms ( dex ) & 0.57 & 0.36 & 0.45 & 0.31 & 0.37 & 0.27 & 0.40 & 0.46 & 0.17 & 0.37 + * full spectrum fits * + avg offset ( gyr ) & 0.93 & 0.54 & 0.71 & 2.54 & 0.42 & 2.24 & 0.70 & 3.24 & 2.76 & 0.39 + avg offset ( dex ) & 0.70 & 0.69 & 0.64 & 0.36 & 0.78 & 0.33 & 0.52 & 0.36 & 0.70 & 0.69 + rms ( gyr ) & 4.42 & 3.20 & 3.68 & 6.08 & 1.56 & 5.78 & 5.43 & 9.61 & 11.55 & 1.47 + rms ( dex ) & 0.93 & 0.90 & 0.71 & 0.51 & 0.98 & 0.44 & 0.64 & 0.42 & 0.79 & 0.91 + * photometry fits * + avg offset ( gyr ) & 0.99 & 0.48 & 0.73 & 0.67 & 0.34 & 0.37 & 0.74 & 5.24 & 2.06 & 0.38 + avg offset ( dex ) & 0.004 & 0.59 & 0.71 & 0.42 & 0.63 & 0.38 & 0.41 & 0.32 & 0.10 & 0.62 + rms ( gyr ) & 3.29 & 1.19 & 1.55 & 0.85 & 1.04 & 0.61 & 1.42 & 9.47 & 2.67 & 1.02 + rms ( dex ) & 0.52 & 0.76 & 0.90 & 0.52 & 0.80 & 0.48 & 0.49 & 0.49 & 0.11 & 0.78 + * continuum fits * + avg offset ( gyr ) & 0.87 & 1.17 & 2.05 & 4.51 & 0.46 & 4.21 & 2.90 & -2.49 & 6.06 & 0.85 + avg offset ( dex ) & 0.06 & 0.80 & 0.84 & 0.49 & 0.84 & 0.46 & 0.73 & 0.55 & 0.80 & 0.80 + rms ( gyr ) & 5.44 & 4.30 & 5.51 & 8.66 & 1.57 & 8.29 & 7.71 & 10.88 & 12.72 & 3.02 + rms ( dex ) & 0.74 & 1.03 & 1.05 & 0.68 & 1.04 & 0.58 & 0.93 & 0.65 & 0.73 & 1.04 + locations & mc , m31 & m31 & m31 & m31 & mc & mc & mc & mc , m31 & mc + references & 1,2,3,4,5,6 & 1 & 2 & 3 & 4 & 5 & 6 & 1,2,3,4,5,6 & 4,5,6 + no .
objects & 33 & 6 & 9 & 8 & 11 & 5 & 18 & 12 & 21 + * cn spectrum fits * + avg offset ( gyr ) & -0.23 & 0.11 & 0.25 & -0.14 & -0.68 & -0.69 & -0.37 & -0.10 & -0.29 + rms ( gyr ) & 0.85 & 0.12 & 0.52 & 0.71 & 1.06 & 0.81 & 0.91 & 0.41 & 1.02 + * full spectrum fits * + avg offset ( gyr ) & 0.23 & 0.31 & 0.88 & 0.40 & -0.27 & -0.05 & 0.16 & 0.48 & 0.10 + rms ( gyr ) & 0.84 & 0.50 & 1.03 & 0.59 & 0.89 & 0.53 & 0.71 & 0.78 & 0.88 + * photometry fits * + avg offset ( gyr ) & 0.28 & 0.62 & 0.77 & 0.41 & -0.18 & -0.11 & 0.15 & 0.51 & 0.14 + rms ( gyr ) & 0.87 & 0.74 & 0.84 & 0.67 & 1.10 & 0.89 & 0.72 & 0.68 & 0.96 + * continuum fits * + avg offset ( gyr ) & 0.36 & 0.81 & 1.34 & 0.85 & -0.29 & -0.35 & 0.12 & 0.85 & 0.08 + rms ( gyr ) & 1.11 & 1.03 & 1.49 & 0.92 & 1.02 & 0.92 & 0.98 & 1.09 & 1.13 + * lmc * + h88 - 267 & @xmath63 & @xmath64 & @xmath65 & 0.005 & & & & & & & & & + hd32228 & @xmath66 & @xmath67 & @xmath64 & 0.004 & & & & & & & & & + hd269828 & @xmath66 & @xmath68 & @xmath69 & 0.004 & & & & & & & & & + ic2105 & @xmath70 & @xmath71 & @xmath72 & & & & & & & & & & + ic2111 & @xmath73 & @xmath74 & @xmath75 & & & & & & & & & & + ic2116 & @xmath76 & @xmath67 & @xmath77 & & & & & & & & & & + kmhk1074 & @xmath76 & @xmath78 & @xmath79 & 0.004 & & & & & & & & & + ngc1466 & @xmath80 & @xmath81 & @xmath82 & & & & 13.100 & & & & & & -1.64 + ngc1711 & @xmath83 & @xmath84 & @xmath85 & 0.020 & & & 0.068 & & & & & & -0.68 + ngc1714 & @xmath86 & @xmath87 & @xmath88 & & & & & & & & & & + ngc1731 & @xmath63 & @xmath64 & @xmath69 & 0.004 & & & & & & & & & + ngc1735 & @xmath89 & @xmath64 & @xmath74 & 0.032 & & & & & & & & & + ngc1736 & @xmath66 & @xmath90 & @xmath91 & & & & & & & & & & + ngc1743 & @xmath92 & @xmath93 & @xmath68 & & & & & & & & & & + ngc1755 & @xmath94 & @xmath95 & @xmath96 & 0.055 & & & & & & & & & + ngc1761 & @xmath97 & @xmath64 & @xmath67 & & & & & & & & & & + ngc1767 & @xmath73 & @xmath79 & @xmath98 & 0.007 & & & & & & & & & + ngc1772 & @xmath99 & @xmath67 & @xmath100 & 0.009 & & & & & & & & & + ngc1774 & @xmath94 & @xmath101 & @xmath65 & 0.037 & & & & & & & & & + ngc1782 & @xmath89 & @xmath102 & @xmath103 & 0.036 & & & & & & & & & + ngc1783 & @xmath104 & @xmath105 & @xmath106 & & 0.200 & 1.548 & 1.300 & & & & -0.75 & -0.54 & -0.65 + ngc1805 & @xmath107 & @xmath67 & @xmath67 & 0.009 & & & 0.014 & & & & & & -0.20 + ngc1818 & @xmath89 & @xmath82 & @xmath108 & 0.014 & 0.017 & 0.020 & & & & & -0.90 & & + ngc1831 & @xmath109 & @xmath110 & @xmath111 & & 0.190 & 0.501 & 0.320 & & & & 0.01 & -0.65 & -0.62 + ngc1847 & @xmath99 & @xmath95 & @xmath79 & 0.021 & 0.016 & & & & & & & & + ngc1850 & @xmath94 & @xmath74 & @xmath74 & 0.031 & 0.040 & & 0.031 & & & & & & -0.12 + ngc1850a & @xmath66 & @xmath112 & @xmath64 & 0.004 & & & & & & & & & + ngc1854 & @xmath113 & @xmath114 & @xmath79 & 0.034 & 0.030 & & 0.034 & & & & & & -0.50 + ngc1856 & @xmath109 & @xmath74 & @xmath115 & & 0.120 & & 0.151 & & & & & & -0.17 + ngc1866 & @xmath116 & @xmath64 & @xmath117 & & 0.080 & 0.302 & 0.150 & & & & -1.20 & & -0.66 + ngc1868 & @xmath118 & @xmath95 & @xmath78 & & 0.330 & 0.933 & 0.850 & & & & -0.50 & -0.32 & -0.66 + ngc1872 & @xmath119 & @xmath64 & @xmath106 & 0.136 & & & & & & & & & + ngc1895 & @xmath89 & @xmath120 & @xmath115 & & & & & & & & & & + ngc1903 & @xmath83 & @xmath81 & @xmath95 & 0.044 & & & & & & & & & + ngc1921 & @xmath116 & @xmath64 & @xmath121 & 0.041 & & & & & & & & & + ngc1935 & @xmath66 & @xmath122 & @xmath111 & & & & & & & & & & + ngc1936 & @xmath94 & @xmath123 & @xmath124 & & & & & & & & & & + ngc1949 & @xmath73 & @xmath64 & @xmath125 & & & & & & & & & & + ngc1951 & @xmath99 & @xmath126 & @xmath103 & 0.048 & & & & & & & & & + ngc1967 & @xmath97 & @xmath103 & @xmath111 & 0.004 & 0.100 & & & & & & & & + ngc1970 & @xmath66 & @xmath64 & @xmath117 & 0.004 & & & & & & & & & + ngc1978 & @xmath127 & @xmath82 & @xmath102 & & & 2.570 & 2.200 & & & & -0.42 & -0.72 & -0.85 + ngc1983 & @xmath66 & @xmath88 & @xmath115 & 0.004 & 0.008 & & & & & & & & + ngc1984 & @xmath76 & @xmath67 & @xmath111 & 0.004 & 0.007 & & 0.004 & & & & & & -0.90 + ngc1994 & @xmath107 & @xmath88 & @xmath79 & 0.006 & 0.007 & & & & & & & & + ngc2002 & @xmath107 & @xmath87 & @xmath64 & 0.008 & & & & & & & & & + ngc2003 & @xmath73 & @xmath79 & @xmath98 & & & & & & & & & & + ngc2004 & @xmath97 & @xmath79 & @xmath79 & & 0.008 & & 0.028 & & & & & & -0.56 + ngc2006 & @xmath73 & @xmath106 & @xmath128 & 0.005 & & & & & & & & & + ngc2009 & @xmath97 & @xmath88 & @xmath95 & 0.006 & & & & & & & & & + ngc2011 & @xmath129 & @xmath130 & @xmath98 & 0.005 & 0.006 & & 0.005 & & & & & & -0.47 + ngc2021 & @xmath99 & @xmath131 & @xmath64 & 0.004 & & & & & & & & & + ngc2025 & @xmath94 & @xmath132 & @xmath133 & 0.083 & & & & & & & & & + ngc2032 & @xmath89 & @xmath64 & @xmath134 & & & & & & & & & & + ngc2033 & @xmath66 & @xmath88 & @xmath75 & 0.004 & & & & & & & & & + ngc2035 & @xmath66 & @xmath135 & @xmath136 & & & & & & & & & & + ngc2041 & @xmath94 & @xmath137 & @xmath69 & 0.058 & & & & & & & & & + ngc2058 & @xmath116 & @xmath138 & @xmath74 & 0.062 & 0.070 & & & & & & & & + ngc2070 & @xmath113 & @xmath67 & @xmath64 & & & & & & & & & & + ngc2091 & @xmath99 & @xmath79 & @xmath95 & 0.007 & & & & & & & & & + ngc2092 & @xmath139 & @xmath91 & @xmath64 & 0.012 & 0.004 & & & & & & & & + ngc2096 & @xmath140 & @xmath64 & @xmath114 & 0.049 & & & & & & & & & + ngc2098 & @xmath99 & @xmath106 & @xmath79 & 0.008 & & & & & & & & & + ngc2100 & @xmath73 & @xmath98 & @xmath88 & 0.009 & 0.010 & & 0.032 & & & & & & -0.32 + ngc2102 & @xmath76 & @xmath78 & @xmath64 & 0.004 & 0.007 & & & & & & & & + ngc2134 & @xmath116 & @xmath141 & @xmath64 & 0.124 & 0.130 & 0.631 & & & & & -1.00 & & + ngc2136 & @xmath94 & @xmath110 & @xmath64 & 0.058 & 0.040 & 0.081 & & & & & -0.40 & & + ngc2147 & @xmath99 & @xmath79 & @xmath65 & 0.021 & & & & & & & & & + ngc2156 & @xmath94 & @xmath74 & @xmath142 & 0.077 & 0.060 & & & & & & & & + ngc2157 & @xmath89 & @xmath111 & @xmath120 & & 0.030 & & & & & & & & + ngc2159 & @xmath94 & @xmath65 & @xmath133 & 0.056 & 0.060 & & & & & & & & + ngc2164 & @xmath94 & @xmath65 & @xmath131 & 0.044 & 0.050 & 0.058 & & & & & -0.60 & & + ngc2214 & @xmath89 & @xmath88 & @xmath143 & 0.038 & 0.040 & 0.052 & & & & & -1.20 & & + sl153 & @xmath116 & @xmath71 & @xmath144 & 0.054 & & & & & & & & & + sl492 & @xmath97 & @xmath88 & @xmath117 & 0.005 & & & & & & & & & + sl538 & @xmath73 & @xmath79 & @xmath87 & 0.006 & & & & & & & & & + sl586 & @xmath66 & @xmath67 & @xmath64 & & & & & & & & & & + sl601 & @xmath76 & @xmath145 & @xmath95 & 0.004 & & & & & & & & & + sl639 & @xmath76 & @xmath75 & @xmath64 & 0.030 & 0.015 & & & & & & & & + *
smc * + ngc121 & @xmath146 & @xmath95 & @xmath102 & & 13.000 & & 11.900 & & & & & & -1.19 + ngc299 & @xmath107 & @xmath85 & @xmath64 & 0.012 & & & & & & & & & + ngc330 & @xmath76 & @xmath111 & @xmath106 & 0.013 & 0.007 & & 0.025 & & & & & & -0.82 + ngc419 & @xmath104 & @xmath77 & @xmath114 & & 0.670 & 1.950 & 1.200 & & & & -0.70 & -0.90 & -0.70 + * m31 * + bo104 & @xmath127 & @xmath77 & @xmath115 & 10.000 & & & & & & & & & + bo118 & @xmath147 & @xmath111 & @xmath87 & 10.000 & & & & & & & & & + bo119 & @xmath148 & @xmath149 & @xmath67 & 10.000 & & & & & & & & & + bo124 & @xmath150 & @xmath117 & @xmath151 & 10.000 & & & & & & & & & + g1 & @xmath118 & @xmath108 & @xmath85 & & & & & & & & & & + g78 & @xmath118 & @xmath108 & @xmath78 & & & & & & & & & & + g158 & @xmath152 & @xmath78 & @xmath85 & 10.000 & & & & 19.800 & -0.26 & & & & + g165 & @xmath153 & @xmath108 & @xmath67 & 10.000 & & & & & -0.56 & -0.69 & & & + g168 & @xmath86 & @xmath108 & @xmath67 & 10.000 & & & & 6.200 & & -1.13 & & & + g169 & @xmath154 & @xmath155 & @xmath155 & 10.000 & & & & 7.500 & -1.18 & -1.35 & & & + g170 & @xmath80 & @xmath106 & @xmath98 & & & & & & & & & & + g174 & @xmath156 & @xmath88 & @xmath79 & 10.000 & & & & 7.200 & 0.29 & -0.13 & & & + g175 & @xmath80 & @xmath133 & @xmath67 & 10.000 & & & & 1.300 & & -1.16 & & & + g177 & @xmath80 & @xmath106 & @xmath79 & 10.000 & & & & 19.800 & -0.15 & -0.32 & & & + g184 & @xmath80 & @xmath111 & @xmath102 & 10.000 & & & & 19.800 & & -0.91 & & & + g185 & @xmath146 & @xmath155 & @xmath100 & 10.000 & & & & 6.500 & -1.08 & -1.19 & & & + g190 & @xmath157 & @xmath158 & @xmath100 & 10.000 & & & & & & -1.96 & & & + g222 & @xmath159 & @xmath108 & @xmath102 & & & & & & & & & & + ngc 1466 & lmc & 0.004 & 20.000 & 0.004 & 20.000 & 4.000 & 1.500 & 8.000 & 2.000 & 13.1 & 13.1 + ngc 1783 & lmc & 0.010 & 1.500 & 9.000 & 1.500 & 6.000 & 1.200 & 20.000 & 1.200 & 0.200 & 1.548 + ngc 1831 & lmc & 1.200 & 0.600 & 1.000 & 0.800 & 0.400 & 0.200 & 1.000 & 0.400 & 0.190 & 0.501 + ngc 1847 & lmc & 0.001 & 0.040 & 0.040 & 0.020 & 0.080 & 0.040 & 0.040 & 0.040 & 0.016 & 0.021 + ngc 1856 & lmc & 0.040 & 0.600 & 0.040 & 0.800 & 0.600 & 1.200 & 0.200 & 0.400 & 0.120 & 0.151 + ngc 1866 & lmc & 0.004 & 1.200 & 0.060 & 0.200 & 0.100 & 0.800 & 0.200 & 0.600 & 0.080 & 0.302 + ngc 1868 & lmc & 0.008 & 0.400 & 2.000 & 3.000 & 0.200 & 1.500 & 2.000 & 0.800 & 0.330 & 0.933 + ngc 1978 & lmc & 0.004 & 20.000 & 10.000 & 4.000 & 4.000 & 20.000 & 10.000 & 9.000 & 2.200 & 2.570 + ngc 2004 & lmc & 0.002 & 0.020 & 0.004 & 0.010 & 0.020 & 0.004 & 0.010 & 0.010 & 0.008 & 0.028 + ngc 2070 & lmc & 0.004 & 18.000 & 10.000 & 0.060 & 0.008 & 0.400 & 0.004 & 0.006 & & + ncg 2157 & lmc & 0.001 & 1.200 & 0.040 & 0.040 & 0.100 & 1.000 & 0.060 & 0.400 & 0.030 & 0.030 + ngc 330 & smc & 0.001 & 0.020 & 0.020 & 0.004 & 0.040 & 0.004 & 0.040 & 0.040 & 0.007 & 0.025 + ngc 419 & smc & 0.010 & 2.000 & 3.000 & 1.500 & 0.200 & 3.000 & 3.000 & 20.000 & 0.670 & 1.950 + | spectra of galaxies contain an enormous amount of information about the relative mixture of ages and metallicities of constituent stars .
we present a comprehensive study designed to extract the maximum information from spectra of data quality typical in large galaxy surveys .
these techniques are not intended for detailed stellar population studies that use high quality spectra .
we test techniques on a sample of globular clusters , which should consist of single stellar populations and provide good test cases , using the bruzual - charlot 2003 high resolution simple stellar population synthesis models to simultaneously estimate the ages and metallicities of 101 globular clusters in m31 and the magellanic clouds by fitting their integrated spectra and photometry .
the clusters cover a wide range of ages and metallicities , 4 myr @xmath0 t@xmath1 @xmath0 20 gyr and @xmath21.6 @xmath0 [ fe / h ] @xmath0 + 0.3 , estimated by other methods in the literature .
we compare results from model fits to both the spectra and photometry and find that fits to continuum - normalized spectra over the entire range available , typically 3500 to 1 @xmath3 m for this sample , provides the best results . for clusters
older than 1 gyr we agree with literature ages to 0.16 dex ( 35% ) and [ fe / h ] to 0.12 dex . for younger clusters we agree with literature ages to 0.3 dex ( 63% ) , but can not constrain the metallicity .
it is particularly important to use the entire continuum - normalized spectrum to avoid problems with model continua for young objects and to break age - metallicity degeneracies of broadband photometry .
our required s / n is 15 - 30 @xmath4 for 20% age uncertainties and 30 - 55 @xmath4 for 10% uncertainties over spectral resolutions of @xmath5 = 5 - 25 .
this technique should work well for the age - metallicity parameter space expected for early - type galaxies at z@xmath61 , although individual galaxy spectral s / n may require the coaddition of a few like objects .
lack of accurate flux calibration in large surveys is not an issue for the continuum - normalized spectra . |
the construction of tests for hypotheses on the coefficient vector in linear regression models with dependent errors is highly practically relevant and has received lots of attention in the statistics and econometrics literature .
the main challenge is to obtain tests with good size and power properties in situations where the nuisance parameter governing the dependence structure of the errors is high- or possibly infinite - dimensional and allows for strong correlations .
the large majority of available procedures are autocorrelation - corrected f - type tests , based on nonparametric covariance estimators trying to take into account the autocorrelation in the disturbances .
these tests can roughly be categorized into two groups , the distinction depending on the choice of a bandwidth parameter in the construction of the covariance estimator .
the first group of such tests , so - called ` hac tests ' , incorporates bandwidth parameters that lead to consistent covariance estimators , and to an asymptotic @xmath0-distribution of the corresponding test statistics under the null hypothesis , the quantiles of which are used for testing . concerning ` hac tests ' , important contributions in the econometrics literature are @xcite , @xcite , @xcite , and @xcite .
it is safe to say that the covariance estimators introduced in the latter two articles currently constitute the gold standard for obtaining ` hac tests ' .
in contrast to the estimator suggested earlier by @xcite - structurally @xmath1 times a standard kernel spectral density estimator ( @xcite , @xcite , @xcite , and @xcite section 7.9 ) evaluated at frequency @xmath2 - the covariance estimators suggested in @xcite and @xcite both incorporate an additional prewhitening step based on an auxiliary vector autoregressive ( var ) model , as well as a data - dependent bandwidth parameter . a distinguishing feature of the estimators introduced by @xcite on the one hand and @xcite on the other hand is the choice of the bandwidth parameter : @xcite used an approach introduced by @xcite , where the bandwidth parameter is chosen based on auxiliary parametric models .
in contrast to that , @xcite suggested a nonparametric approach for choosing the bandwidth parameter . even though simulation studies have shown
that the inclusion of a prewhitening step and the data - dependent choice of the bandwidth parameter can improve the finite sample properties of ` hac tests ' , the more sophisticated ` hac tests ' so obtained still suffer from size distortions and power deficiencies .
for this reason @xcite , @xcite , and @xcite suggested to choose the bandwidth parameter as a fixed proportion of the sample size .
this framework leads to an inconsistent covariance estimator and to a non - standard limiting distribution of the corresponding test statistic under the null , the quantiles of which are used to obtain so called ` fixed - b tests ' . in simulation studies it has been observed that ` fixed - b tests ' still suffer from size distortions in finite samples , but less so than ` hac tests ' . however , this is at the expense of some loss in power .
similar as in ` hac testing ' simulation results in @xcite and @xcite suggest that the finite sample properties of ` fixed - b tests ' can be improved by incorporating a prewhitening step . in the latter paper
it was furthermore shown that the asymptotic distribution under the null of the test suggested by @xcite is the same whether or not prewhitening is used . a number of recent studies ( @xcite , @xcite , @xcite , @xcite ) tried to use higher order expansions to uncover the mechanism leading to size distortions and power deficiencies of ` hac tests ' and ` fixed - b tests ' . these higher - order asymptotic results ( and also the first - order results discussed above )
are pointwise in the sense that they are obtained under the assumption of a fixed underlying data - generating - process .
hence , while they inform us about the limit of the rejection probability and the rate of convergence to this limit for a fixed underlying data - generating - process , they do not inform us about the _ size _ of the test or its limit as sample size increases , nor about the _ power function _ or its asymptotic behavior .
size and power properties of tests in regression models with dependent errors were recently studied in @xcite : in a general finite sample setup and under high - level conditions on the structure of the test and the covariance model , they derived conditions on the design matrix under which a concentration mechanism due to strong dependencies leads to extreme size distortions or power deficiencies .
furthermore , they suggested an adjustment - procedure to obtain a modified test with improved size and power properties .
specializing their general theory to a covariance model that includes at least all covariance matrices corresponding to stationary autoregressive processes of order one ( ar(1 ) ) , they investigated finite sample properties of ` hac tests ' and ` fixed - b tests ' based on _ non - prewhitened _ covariance estimators with _ data - independent _ bandwidth parameters ( covering _ inter alia _ the procedures in @xcite , sections 3 - 5 of @xcite , @xcite , @xcite , @xcite , @xcite , but _ not _ the methods considered by @xcite , @xcite or @xcite ) .
in this setup @xcite demonstrated that these tests break down in terms of their size or power behavior for generic design matrices . despite this negative result
, they also showed that the adjustment procedure can often solve these problems , if elements of the covariance model which are close to being singular can be well approximated by ar(1 ) covariance matrices .
@xcite did not consider tests based on prewhitened covariance estimators or data - dependent bandwidth parameters .
therefore the question remains , whether the more sophisticated ` hac tests ' typically used in practice ( i.e. , tests based on the estimators by @xcite or @xcite ) and the prewhitened ` fixed - b tests ' ( i.e. , tests as considered in @xcite ) also suffer from extreme size distortions and power deficiencies , or if prewhitening and the use of data - dependent bandwidth parameters can indeed resolve or at least substantially alleviate these problems . in the present paper we investigate finite sample properties of tests based on prewhitened covariance estimators or data - dependent bandwidth parameters .
in particular our analysis covers tests based on prewhitened covariance estimators using auxiliary ar(1 ) models for the construction of the bandwidth parameter as discussed in @xcite , tests based on prewhitened covariance estimators as discussed in @xcite , and prewhitened ` fixed - b ' tests as discussed in @xcite .
we show that the tests considered , albeit being structurally much more complex , exhibit a similar behavior as their _ non - prewhitened _ counterparts with _ data - independent _ bandwidth parameters : first , we establish conditions on the design matrix under which the tests considered have ( i ) size equal to one , or ( ii ) size not smaller than one half , or ( iii ) nuisance - minimal power equal to zero , respectively .
we then demonstrate that at least one of these conditions is generically satisfied , showing that the tests considered break down for generic design matrices .
motivated by this negative result , we introduce an adjustment procedure . under the assumption that elements of the covariance model which are close to being singular can be well approximated by ar(1 ) covariance matrices , we show that the adjustment procedure , if applicable , leads to tests that do not suffer from extreme size distortions or power deficiencies .
finally , it is shown that the adjustment procedure is applicable under generic conditions on the design matrix , unless the regression includes the intercept _ and _ the hypothesis to be tested restricts the corresponding coefficient . on a technical level
we employ the general theory developed in @xcite .
we remark , however , that the genericity results in particular do not follow from this general theory .
rather they are obtained by carefully exploiting the specific structure of the procedures under consideration .
the paper is organized as follows : the framework is introduced in section [ framework ] . in section [ tpwc ]
we introduce the test statistics , covariance estimators , and bandwidth parameters we analyze . in section [ neg ]
we establish our negative result and its genericity . in section [ pos ]
we discuss the adjustment - procedure and its generic applicability .
section [ concl ] concludes .
the proofs are collected in appendices [ aa]-[ad ] .
consider the linear regression model @xmath3 where @xmath4 is a ( real ) @xmath5 dimensional non - stochastic design matrix satisfying @xmath6 , @xmath7 and @xmath8 . here
, @xmath9 denotes the unknown regression parameter vector , and the disturbance vector @xmath10 is gaussian , has mean zero and its unknown covariance matrix is given by @xmath11 .
the parameter @xmath12 satisfies @xmath13 and @xmath14 is assumed to be an element of a prescribed ( non - void ) set of positive definite and symmetric @xmath15 matrices @xmath16 , which we shall refer to as the _ covariance model_. throughout we impose the assumption on @xmath16 that the parameters @xmath12 and @xmath14 can be uniquely determined from @xmath11 .
[ stationary ] the leading case we have in mind is the situation where @xmath17 are @xmath18 consecutive elements of a weakly stationary process . in such a setup a covariance model
is typically obtained from a prescribed ( non - void ) set of spectral densities @xmath19 . assuming that no element of @xmath19 vanishes identically almost everywhere , the covariance model corresponding to @xmath19 is then given by @xmath20 with @xmath21 and where @xmath22 denotes the imaginary unit .
every such @xmath23 is positive definite and symmetric .
furthermore , since @xmath23 is a correlation matrix , @xmath12 and @xmath23 can uniquely be determined from @xmath24 . as outlined in the introduction the tests we shall investigate in this paper
are particularly geared towards setups where @xmath19 is a nonparametric class of spectral densities , i.e. , where the corresponding set @xmath25 is rich .
a typical example is the class @xmath26 , which consists of all spectral densities of linear processes the coefficients of which satisfy a certain summability condition , i.e. , spectral densities of the form @xmath27 where , for a fixed @xmath28 , the summability condition @xmath29 is satisfied .
we observe that @xmath30 contains in particular all correlation matrices corresponding to spectral densities of stationary autoregressive moving average models of arbitrary large order . the linear model described in
induces a collection of distributions on @xmath31 , the sample space of @xmath32 . denoting a gaussian probability measure with mean @xmath33 and covariance matrix @xmath34 by @xmath35 and denoting the regression manifold by @xmath36 , the induced collection of distributions
is given by @xmath37 since every @xmath38 is positive definite by definition , each element @xmath35 of the set in the previous display is absolutely continuous with respect to ( w.r.t . )
lebesgue measure on @xmath39 . in this setup
we shall consider the problem of testing a linear hypothesis on the parameter vector @xmath9 , i.e. , the problem of testing the null @xmath40 against the alternative @xmath41 , where @xmath42 is a @xmath43 matrix of rank @xmath44 and @xmath45 .
define the affine space @xmath46 and let @xmath47 adopting these definitions , the above testing problem can be written as @xmath48 where it is emphasized that the testing problem is a compound one .
it is immediately clear that size and power properties of tests in this setup depend in a crucial way on the richness of the covariance model @xmath16 .
before we close this section by introducing some further terminological and notational conventions , some comments on how the above assumptions can be relaxed are in order : we remark that even though our setup assumes a non - stochastic design matrix , the results immediately carry over to a setting where the data generating processes of the design and the disturbances are independent of each other .
in such a setup our results deliver size and power properties conditional on the design .
the gaussianity assumption might seem to be restrictive .
however , as in section 5.5 of @xcite , we mention that the negative results given in section [ neg ] of the present paper immediately extend in a trivial way without imposing the gaussianity assumption on the error vector @xmath49 in , as long as the assumptions on the feasible error distributions are weak enough to ensure that the implied set of distributions for @xmath32 contains the set in equation , but possibly contains also other distributions .
furthermore , by applying an invariance argument ( explained in @xcite section 5.5 ) one can easily show that all statements about the null - behavior of the procedures under consideration derived in the present paper carry over to the more general distributional setup where @xmath49 is assumed to be elliptically distributed .
this is to be understood as @xmath49 having the same distribution as @xmath50 , where @xmath51 , @xmath38 , @xmath52 is a random vector uniformly distributed on the unit sphere @xmath53 , and @xmath54 is a random variable distributed independently of @xmath52 and which is positive with probability one .
we next collect some further terminology and notation used throughout the whole paper .
a ( non - randomized ) _ test _ is the indicator function of a set @xmath55 , i.e. , the corresponding _
rejection region_. the _ size _ of such a test ( rejection region ) is the supremum over all rejection probabilities under the null hypothesis @xmath56 , i.e. , @xmath57 throughout the paper we let @xmath58 , where @xmath4 is the design matrix appearing in and @xmath59 . the corresponding ordinary least squares ( ols ) residual vector
is denoted by @xmath60 .
the subscript @xmath4 is omitted whenever this does not cause confusion .
random vectors and random variables are always written in bold capital and bold lower case letters , respectively .
we use @xmath61 as a generic symbol for a probability measure and denote by @xmath62 the corresponding expectation operator .
lebesgue measure on @xmath39 will be denoted by @xmath63 .
the euclidean norm is denoted by @xmath64 , while @xmath65 denotes the euclidean distance of the point @xmath66 to the set @xmath67 . for a vector @xmath68 in euclidean space
we define the symbol @xmath69 to denote @xmath70 for @xmath71 , the sign being chosen in such a way that the first nonzero component of @xmath69 is positive , and we set @xmath72 . the @xmath73-th standard basis vector in @xmath39 is denoted by @xmath74 .
let @xmath75 denote the transpose of a matrix @xmath76 and let @xmath77 denote the space spanned by its columns .
for a linear subspace @xmath78 of @xmath39 we let @xmath79 denote its orthogonal complement and we let @xmath80 denote the orthogonal projection onto @xmath78 .
the set of real matrices of dimension @xmath81 is denoted by @xmath82 .
lebesgue measure on this set equipped with its borel @xmath83-algebra is denoted by @xmath84 .
we use the convention that the adjoint of a @xmath85 dimensional matrix @xmath86 , i.e. , @xmath87 , equals one . given a vector @xmath88 the symbol @xmath89 denotes the @xmath90 diagonal matrix with main diagonal @xmath91 .
we define @xmath92 i.e. , the set of @xmath93 design matrices of full rank , and whenever @xmath94 we define @xmath95 which is canonically identified ( as a set ) with the set of @xmath93 design matrices of full column rank the first column of which is the intercept @xmath96 .
in the present section we formally describe the construction of tests based on prewhitened covariance estimators .
these tests ( cf . remark [ rdef ] below and the discussion preceding it ) reject for large values of a statistic @xmath97 where @xmath98 and @xmath99 the quantity @xmath100 appearing in the definition of @xmath101 above denotes a ( var- ) prewhitened nonparametric estimator of @xmath102 that incorporates a bandwidth parameter which might depend on the data .
such an estimator is completely specified by three core ingredients : first , a _ kernel _ @xmath103 , i.e. , an even function satisfying @xmath104 , such as , e.g. , the bartlett or parzen kernel ; second , a ( non - negative ) possibly data - dependent _ bandwidth parameter _ @xmath105 ; and third , a deterministic _ prewhitening order _
@xmath106 , i.e. , an integer satisfying @xmath107 ( cf .
remark [ defp ] ) .
specific choices of @xmath105 are discussed in detail in section [ bw ] .
all possible combinations of @xmath108 , @xmath105 and @xmath106 we analyze are specified in assumption [ weightspwrb ] of section [ comb ] .
once these core ingredients have been chosen , one obtains a prewhitened estimator @xmath100 , which is computed at an observation @xmath109 following the steps ( 1 ) - ( 3 ) outlined subsequently ( cf . also @xcite ) .
we here assume that the quantities involved ( e.g. , inverse matrices ) are well defined , cf .
remark [ wdpsi ] below , and follow the convention in the literature and leave the estimator undefined at @xmath109 else . using this convention @xmath110 is obtained as follows : 1 . to prewhiten the data a var(p ) model
is fitted via ordinary least squares to the columns of @xmath111 .
one so obtains the var(p ) residual matrix @xmath112 with columns @xmath113 the @xmath114-dimensional var(p)-ols estimator is given by @xmath115 where @xmath116 and the @xmath73-th column of @xmath117 equals @xmath118 for @xmath119 . in matrix form
we clearly have @xmath120 .
2 . then , one computes the quantities @xmath121 and defines the preliminary estimate @xmath122 where in case @xmath123 one sets @xmath124 for @xmath125 and @xmath126 for @xmath127 .
3 . finally , the preliminary estimate @xmath128 is ` recolored ' using the transformation @xmath129'.\ ] ] [ wdpsi ] the construction of @xmath110 outlined above clearly assumes that ( i ) @xmath130 is well defined , which is equivalent to @xmath131 ; that ( ii ) @xmath132 is well defined , which depends on the specific choice of @xmath105 ( cf .
section [ bw ] ) ; and that ( iii ) @xmath133 is invertible .
[ defp ] by assumption , all possible var orders @xmath106 we consider must satisfy @xmath134 .
this is done to rule out degenerate cases : for if @xmath135 , then @xmath136 would follow , because of @xmath117 .
hence the covariance estimator would nowhere be well defined for such a choice , because ( i ) in remark [ wdpsi ] would then clearly be violated at every observation @xmath109 . [ varols ] in the present paper we focus on var prewhitening based on the ols estimator .
this is in line with the original suggestions by @xcite , as well as with @xcite .
alternatively , for @xmath137 , @xcite suggested to use an eigenvalue adjusted version of the ols estimator , the adjustment being applied if the matrix @xmath138 is close to being singular .
we shall focus on the unadjusted ols estimator for the following reasons : @xcite reported that the finite sample properties show little sensitivity to this eigenvalue adjustment .
furthermore , it is the unadjusted estimator that is often used in implementations of the method suggested by @xcite in software packages for statistical and econometric computing ( e.g. , its implementation in the ` r ` package ` sandwich ` by @xcite , or its implementation in ` eviews ` , e.g. , @xcite , p. 784 . ) .
we remark , however , that one can obtain a negative result similar to theorem [ thmlrvpw ] , and a positive result concerning an adjustment procedure similar to theorem [ tu_3 ] , also for tests based on prewhitened estimators with eigenvalue adjustment .
furthermore , we conjecture that it is possible to prove ( similar to proposition [ generic ] ) the genericity of such a negative result , and to show that one can ( similar to proposition [ genericadj ] ) generically resolve this problem by using the adjustment procedure .
we leave the question of which estimator to choose for prewhitening to future research . in a typical asymptotic analysis of tests based on
prewhitened covariance estimators the event @xmath139 is asymptotically negligible ( since @xmath101 converges to a positive definite , or almost everywhere positive definite matrix ) .
hence there is no need to be specific about the definition of the test statistic for @xmath140 , and one can work directly with the statistic @xmath141 which is left undefined for @xmath140 . in a finite sample setup , however , one has to think about the definition of the test statistic also for @xmath140 . our decision to assign the value @xmath2 to the test statistic for @xmath140 is of course completely arbitrary . that this assignment does not affect our results
at all is discussed in detail in the following remark .
[ rdef ] given that the estimator @xmath101 is based on a triple @xmath108 , @xmath105 , @xmath106 that satisfies assumption [ weightspwrb ] introduced below ( which is assumed in all of our main results , and which is satisfied for covariance estimators using auxiliary ar(1 ) models for the construction of the bandwidth parameter as considered in @xcite , for covariance estimators as considered in @xcite , and for covariance estimators as considered in @xcite ) , it follows from lemma [ n*pw ] that @xmath139 is either a @xmath142-null set , or that it coincides with @xmath143 . in the first case , which is generic under weak dimensionality constraints
as shown in lemma [ nulln*pw ] , the definition of the test statistic on @xmath139 does hence not influence the rejection probabilities , because our model is dominated by @xmath142 ( @xmath16 contains only positive definite matrices ) .
therefore , size and power properties are not affected by the definition of the test statistic for @xmath140 .
in the second case , i.e. , if @xmath139 coincides with @xmath143 , the statistic in is nowhere well defined , and hence , regardless of which value is assigned to it for observations @xmath140 , the resulting test statistic is constant , and thus the test breaks down trivially . in the following we describe bandwidth parameters @xmath105
that are typically used in step 2 in the construction of the prewhitened estimator @xmath100 as discussed above : the parametric approach ( based on auxiliary ar(1 ) models ) suggested by @xcite and @xcite , the nonparametric approach introduced by @xcite , and a data - independent approach which was already investigated in @xcite in simulation studies and which has recently been theoretically investigated by @xcite .
since the bandwidth parameter @xmath105 is computed in step 2 in the construction of @xmath110 , we assume that @xmath108 , @xmath106 and @xmath109 are given and that step 1 has already been successfully completed , i.e. , all operations in step 1 are well defined at @xmath109 , in particular @xmath144 is available for the construction of @xmath105 . if not , we leave the bandwidth parameter ( and hence the covariance estimator ) undefined at @xmath109 .
we also implicitly assume that the quantities and operations appearing in the procedures outlined subsequently are well defined and leave the bandwidth parameter ( and hence the covariance estimator ) undefined else .
a detailed structural analysis of the subset of the sample space where a prewhitened estimator @xmath101 is well defined is then later given in lemma [ npwrb ] in section [ struct ] . finally , we emphasize that the bandwidth parameters discussed subsequently all require the choice of additional tuning parameters .
these tuning parameters are typically chosen independently of @xmath109 and @xmath4 , an assumption we shall maintain throughout the whole paper ( but see remark [ rtuning ] for some generalizations ) .
let @xmath145 be such that @xmath146 and @xmath147 for @xmath148 , i.e. , @xmath149 is a _
weights vector_. based on this weights vector the bandwidth parameter is now obtained as follows : first , univariate ar(1 ) models are fitted via ols to @xmath150 for @xmath148 , giving @xmath151 where we note that @xmath152 holds as a consequence of @xmath6 and @xmath153 .
then , one calculates @xmath154 finally , bandwidth parameters are obtained via @xmath155 where to obtain a bandwidth parameter , one has to fix the constants @xmath156 , @xmath157 and @xmath73 and where @xmath158 .
typically the choice of these constants and the choice of @xmath73 depends on certain characteristics of @xmath108 ( for specific choices see @xcite , section 6 , in particular p. 834 ) . for example , if @xmath108 is the bartlett kernel one uses @xmath159 , @xmath160 and @xmath161 , or if @xmath108 is the quadratic - spectral kernel one would use @xmath162 , @xmath163 and @xmath164 .
since we do not need such a specific dependence to derive our theoretical results , we do not impose any further assumptions on these constants beyond being positive ( and independent of @xmath109 and @xmath4 ) .
we shall denote by @xmath165 the set of all bandwidth parameters that can be obtained as special cases of the method in the present section , by appropriately choosing - functionally independently of @xmath109 and @xmath4 - a weights vector @xmath149 , constants @xmath156 , @xmath157 and a @xmath166 .
since @xmath18 , @xmath167 and @xmath168 are fixed quantities , the tuning parameters @xmath149 , @xmath169 for @xmath170 and @xmath73 might also depend on them , although we do not signify this in our notation .
a similar remark applies to the constants appearing in section [ mnw94 ] and in section [ mr ] .
although we do not provide any details , we furthermore remark that one can extend our analysis to bandwidth parameters as above , but based on estimators other than @xmath171 , e.g. , all estimators satisfying assumption 4 of @xcite such as the yule - walker estimator or variants of the ols estimator .
let @xmath145 be as in section [ ma92 ] and let @xmath172 for @xmath173 be real numbers such that @xmath174 .
for example , @xcite suggested to use rectangular weights , i.e. , @xmath175 where @xmath176 denotes the floor function .
define for every @xmath177 @xmath178 a bandwidth parameter is then obtained via @xmath179 ^ 2 n \right)^{\bar{c}_3},\ ] ] where @xmath180 is a positive integer , where @xmath181 and @xmath182 are positive real numbers and where @xmath183 .
these numbers are constants independent of @xmath109 and @xmath4 and have to be chosen by the user .
the choice typically depends on the kernel ( for the specific choices we refer the reader to @xcite , section 3 ) . as in the previous section
, we do not impose any assumptions beyond positivity ( and independence of @xmath109 and @xmath4 ) on the constants .
furthermore , we shall denote by @xmath184 the set of all bandwidth parameters that can be obtained as special cases of the method in the present section , by appropriately choosing - functionally independently of @xmath109 and @xmath4 - a weights vector , numbers @xmath172 for @xmath173 , @xmath180 a positive integer , @xmath185 and @xmath186 .
\(i ) the method described here is the ` real - bandwidth ' approach suggested in @xcite , as opposed to the ` integer - bandwidth ' approach . in the latter approach
one would use @xmath187 instead of @xmath188 .
both approaches are asymptotically equivalent ( @xcite , theorem 2 ) for most kernels ( including the bartlett kernel which is suggested in @xcite ) .
therefore , they are equally plausible in terms of their theoretical foundation . for the sake of simplicity and comparability with the bandwidth parameter
as suggested by @xcite , which is not an integer in general , we have chosen to focus on the ` real - bandwidth ' approach .
+ ( ii ) @xcite , p. 637
, in principle also allow for @xmath189 ( @xmath190 in their notation ) in the definition of their estimator .
we do not allow for such a choice .
however , note that @xmath189 implies @xmath191 .
this is a data - independent bandwidth parameter .
these parameters are separately treated in section [ mr ] .
+ @xcite and @xcite studied properties of prewhitened ` fixed - b tests ' . here one sets @xmath192 where @xmath193 $ ] is functionally independent of @xmath109 and @xmath4 .
for example , in @xcite the choice @xmath194 is studied .
these approaches all lead to bandwidth parameters @xmath195 , that are functionally independent of both @xmath4 and @xmath109 .
we denote the set of such bandwidth parameters by @xmath196 .
different combinations of kernels @xmath108 , bandwidth parameters @xmath105 and var orders @xmath106 obviously lead to different estimators .
we indicate the dependence of the estimator on these quantities by writing @xmath197 . in the present paper we shall consider estimators @xmath197 based on a triple @xmath108 , @xmath105 , @xmath106 which satisfies the following assumption : [ weightspwrb ] the triple @xmath108 , @xmath105 , @xmath106 satisfies : 1 .
@xmath103 is an even function and @xmath104 .
furthermore , @xmath108 is continuous , satisfies @xmath198 , and for every real number @xmath199 and every positive integer @xmath200 the @xmath201 symmetric toeplitz matrix with @xmath202-th coordinate @xmath203 is positive definite .
2 . @xmath204 .
@xmath106 is an integer satisfying @xmath107 .
[ rweights ] first , we remark that the positive definiteness assumption in part 1 of assumption [ weightspwrb ] is natural in our context , because it guarantees that @xmath197 is nonnegative definite whenever it is well defined .
furthermore , it allows us to derive simple conditions for positive definiteness of @xmath197 ( cf .
lemma [ n*pw ] ) .
it is well known that many kernels used in practice satisfy the positive definiteness assumption , e.g. , the bartlett , parzen , and quadratic - spectral kernel .
secondly , we note that in principle assumption [ weightspwrb ] does not prohibit a combination of @xmath205 with a second order kernel , or the combination of @xmath206 with a first order kernel .
it also allows for a combination of elements of @xmath184 with a prewhitening order @xmath207 and for the combination of elements of @xmath196 with a kernel other than the bartlett kernel .
this goes well beyond the original suggestions in @xcite , @xcite and @xcite , but we include these additional possibilities for convenience .
we also remark that since we assume throughout that @xmath208 , the set of var orders satisfying the third part of assumption [ weightspwrb ] always includes the order @xmath137 .
[ rtuning ] the tuning parameters used in the construction of @xmath204 , e.g. , the weights vector @xmath149 used in the construction of @xmath209 , are by definition functionally independent of @xmath109 _ and _ @xmath4 . requiring that the tuning parameters are independent of @xmath4
is not a restriction in all results of the present paper in which the design matrix @xmath4 is _ fixed _
( i.e. , theorem [ thmlrvpw ] , proposition [ excpw ] , and theorem [ tu_3 ] ) . to see this ,
suppose that a design matrix @xmath4 as in is given , that @xmath108 and @xmath106 satisfy the first and third part of assumption [ weightspwrb ] , respectively , and that @xmath105 is constructed as in one of the sections [ ma92 ] , [ mnw94 ] , [ mr ] , but with a vector of tuning parameters @xmath210 , say , that is not constant on @xmath211 .
the triple @xmath212 hence does not satisfy assumption [ weightspwrb ] .
let @xmath213 be the bandwidth parameter that is obtained from @xmath105 by replacing the vector of tuning parameters @xmath210 by @xmath214 .
clearly , @xmath215 satisfies assumption [ weightspwrb ] , and the test statistics as in equation based on @xmath216 and @xmath217 , respectively , coincide for this specific @xmath4 .
the study of finite sample properties of a test based on the statistic in equation with @xmath218 requires a detailed understanding of definiteness properties of the covariance estimator @xmath197 , and of the structure of the set @xmath219 . denoting the subset of the sample space @xmath143 where @xmath197 is not well defined by @xmath220
, we can write @xmath221 as a first step we study @xmath220 in the subsequent lemma , where it is shown that @xmath220 is algebraic . the lemma also characterizes the dependence of @xmath220 on the design matrix , which will later be useful for obtaining our genericity results .
[ npwrb ] assume that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] . then , @xmath222 where @xmath223 is a multivariate polynomial ( explicitly constructed in the proof ) . as a consequence @xmath220 is an algebraic set . furthermore , @xmath224 does not depend on the hypothesis @xmath225 .
the subsequent lemma discusses definiteness and regularity properties of @xmath197 and shows that @xmath219 is an algebraic subset of @xmath143 .
again the dependence of this algebraic set on the design is clarified .
given a prewhitening order @xmath106 satisfying part 3 of assumption [ weightspwrb ] , we define for every @xmath226 such that @xmath130 is well defined and such that @xmath227 is invertible the matrix @xmath228 [ n*pw ] assume that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] . then the following holds : 1 . @xmath229 is nonnegative definite if and only if @xmath230 .
@xmath229 is singular if and only if @xmath230 and @xmath231 .
3 . @xmath232 if and only if @xmath230 and @xmath233 .
@xmath229 is positive definite if @xmath230 and @xmath234 .
we have @xmath235 where @xmath236 is a multivariate polynomial ( explicitly constructed in the proof ) . as a consequence @xmath219 is an algebraic set .
furthermore , @xmath237 is independent of @xmath238 .
it is a well known fact that an algebraic subset of @xmath143 is either a closed @xmath142-null set , or coincides with @xmath143 ( for a proof see , e.g. , @xcite ) .
the latter case occurs if and only if a ( multivariate ) polynomial defining the algebraic set vanishes everywhere .
together with part 5 of lemma [ n*pw ] this implies that @xmath219 is either a closed @xmath142-null set , or coincides with @xmath143 , depending on whether @xmath239 or @xmath240 holds , respectively . in the latter case ,
every test based on the test statistic defined in equation with @xmath218 trivially breaks down , because in this case the test statistic vanishes identically on @xmath143 .
obviously , studying size and power properties of tests based on this test statistic in a sample of size @xmath18 is only interesting , if we can guarantee that @xmath239 holds for a sufficiently large set of design matrices . that this is indeed the case is the content of the subsequent lemma .
more precisely it is shown that @xmath239 is generically satisfied whenever @xmath18 exceeds a certain threshold .
it is also shown that the threshold we give can not be substantially improved .
the notion of genericity employed is further discussed in remark [ rgenericityomega ] following the lemma .
[ nulln*pw ] assume that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] . then the following holds : 1 .
if @xmath241 and @xmath242 , then @xmath243 2 . if @xmath244 , then @xmath245 if @xmath246 we have in particular @xmath247 .
3 . if @xmath94 and @xmath248 , where @xmath249 , then @xmath250 [ rgenericityomega ] ( 1 ) part 1 demonstrates that if @xmath18 is too small in the sense that @xmath241 , then for every @xmath251 the test statistic in equation with @xmath218 vanishes identically if @xmath242 holds , because the estimator @xmath197 is either not well defined or singular at every observation @xmath109 .
this shows that one can in general not expect that @xmath219 is generically a @xmath142-null set in case @xmath241 .
\(2 ) under the assumption that @xmath244 holds , part 2 establishes genericity of @xmath252 in that it shows that the statement holds for @xmath253- almost every @xmath251 .
this notion of genericity is obviously related to situations , where the data - generating process underlying the design matrix @xmath4 is assumed to be absolutely continuous w.r.t .
@xmath253 . in this situation , a bandwidth parameter @xmath209
would typically be based on the weights vector @xmath254 . as a specific result of independent interest
it is also shown that if @xmath246 then @xmath247 holds , which means that in the location model the set @xmath219 is a @xmath142- null set .
\(3 ) under the assumption that @xmath94 and @xmath248 holds , part 3 establishes genericity of @xmath255 by showing that the statement holds for @xmath256 almost every @xmath257 .
this is a genericity statement concerning design matrices the first column of which is the intercept .
in contrast to ( 2 ) this notion of genericity is related to situations , where the first column of the design matrix is fixed and the data - generating process underlying the remaining columns is absolutely continuous w.r.t .
in such a setup the construction of a bandwidth parameter @xmath209 would typically be based on the weights vector @xmath258 .
in the first part of this section we obtain our main negative result concerning finite sample properties of tests based on prewhitened nonparametric covariance estimators . for this result to hold , we have to impose a richness assumption on the covariance model @xmath16 .
let @xmath259 denote the set of all correlation matrices corresponding to stationary autoregressive processes of order one , i.e. , @xmath260 , where @xmath261 for @xmath262 .
the assumption is as follows .
[ aar(1 ) ] @xmath263 .
[ singboundary ] assumption [ aar(1 ) ] implies in particular that the singular boundary of @xmath264 , i.e. , the set of singular matrices in @xmath265 , contains at least the two elements @xmath266 and @xmath267 , where @xmath268 and @xmath269 .
we note that these two singular matrices can be approximated by sequences @xmath270 with @xmath271 and @xmath272 , respectively , where @xmath273 . since the procedures we study in the present paper are geared towards situations such as @xmath274 for some @xmath28 ( cf .
remark [ stationary ] ) , covariance models which clearly satisfy the above assumption , assumption [ aar(1 ) ] is mild in our context ( cf . also the discussion in section 3.2.2 of @xcite ) . under this assumption and
given a hypothesis @xmath225 , the subsequent theorem provides four sufficient conditions on the design matrix under which a test based on a test statistic as in equation with @xmath218 , together with an arbitrary ( but data - independent ) critical value @xmath275 , breaks down in terms of its finite sample size and/or power properties .
more precisely , conditions ( 1 ) and ( 4 ) imply that the test has size equal to one , condition ( 3 ) implies that the test has size not smaller than @xmath276 , and condition ( 2 ) implies that the nuisance - minimal rejection probability equals zero at every point @xmath277 .
[ thmlrvpw ] suppose that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] and that @xmath16 satisfies assumption [ aar(1 ) ] .
let @xmath278 be the test statistic defined in with @xmath218 .
let @xmath279 be the rejection region , where @xmath280 is a real number satisfying @xmath281 .
then the following holds : 1 .
suppose @xmath282 and @xmath283 holds for some ( and hence all ) @xmath284 , or @xmath285 and @xmath286 holds for some ( and hence all ) @xmath287
. then @xmath288 holds for every @xmath289 and every @xmath290 .
in particular , the size of the test is equal to one .
2 . suppose @xmath282 and @xmath291 holds for some ( and hence all ) @xmath284 , or @xmath285 and @xmath292 holds for some ( and hence all ) @xmath287
. then @xmath293 holds for every @xmath289 and every @xmath290 , and hence @xmath294 holds for every @xmath295 . in particular , the test is biased .
furthermore , the nuisance - infimal rejection probability at every point @xmath296 is zero , i.e. , @xmath297 in particular , the infimal power of the test is equal to zero .
suppose @xmath282 , @xmath298 and @xmath299 exists for some ( and hence all ) @xmath284 , or @xmath285 , @xmath300 and @xmath301 exists for some ( and hence all ) @xmath287 .
then @xmath302 holds for every @xmath289 and every @xmath290 .
in particular , the size of the test is at least @xmath276 .
4 . suppose that @xmath252 .
suppose further that @xmath303 and @xmath304 holds , or @xmath305 and @xmath306 holds .
then @xmath307 holds for every @xmath289 and every @xmath290 .
in particular , the size of the test is equal to one .
[ remarkthmlrvpw ] ( i ) lemma [ a567pw ] in appendix [ aneg ] shows that the rejection probabilities @xmath308 depend on @xmath309 only through @xmath310 , where @xmath311 is uniquely determined by @xmath312 .
+ ( ii ) obviously , the conclusions of the preceding theorem also apply to any rejection region @xmath313 which differs from @xmath314 only by a @xmath142-null set .
+ ( iii ) in part 1 of the theorem the condition @xmath282 ( @xmath285 ) is superfluous , because it is already implicit in @xmath315 ( @xmath316 ) , which is readily seen from the definition of @xmath278 in equation . a similar comment applies to part 3 of the theorem , where the condition @xmath282 ( @xmath285 ) is already implicit in @xmath317 ( @xmath318 ) .
the conditions are included for the sake of comparability with part 2 of the theorem .
+ ( iv ) in case @xmath319 , the assumption concerning the existence of the gradient can be dropped in part 3 of the theorem .
this follows from lemma [ gradientt ] in appendix [ aneg ] , where it is shown that if @xmath319 , then the existence of @xmath320 and @xmath321 is already implied by @xmath282 and @xmath285 , respectively .
+ ( v ) throughout the theorem , assumption [ aar(1 ) ] can be replaced by the weaker assumption that there exist two sequences @xmath322 and @xmath323 of ar(1 ) correlation matrices in @xmath16 , such that @xmath324 and @xmath325 . in parts 1 and 2 of the theorem
it is even enough to assume that there exist sequences @xmath326 for @xmath170 with @xmath327 and @xmath328 .
therefore , in these parts it is only important that - and not how - these singular matrices can be approximated from within @xmath16 .
we shall now provide some intuition for theorem [ thmlrvpw ] ( cf . also the discussion preceding theorem 5.7 in @xcite ) .
the repeated appearance of the vectors @xmath329 and @xmath330 in the theorem stems from the fact that both @xmath331 and @xmath332 are elements of the singular boundary of @xmath333 ( cf .
remark [ singboundary ] ) .
furthermore , for every @xmath334 and every @xmath13 we have that @xmath335 weakly as @xmath336 with @xmath38 , and similarly that @xmath337 weakly as @xmath338 with @xmath38 .
these limiting measures are absolutely continuous w.r.t . @xmath339 and @xmath340 , respectively . as a consequence
we see that the mass of @xmath341 concentrates on ` neighborhoods ' of certain one - dimensional affine spaces as @xmath14 approximates @xmath331 or @xmath332 from within @xmath16 . from
that it is intuitively clear that size and power properties crucially depend on the behavior of the tests on ` neighborhoods ' of these spaces .
the first and second part of the theorem provide sufficient conditions under which these spaces are almost surely ( w.r.t . @xmath339 and @xmath340 ) contained in the interior or exterior of the rejection region , respectively .
the former case then leads to size distortions , the latter to power deficiencies .
the situation in the third part of the theorem is quite different and more complex . in this case
the one - dimensional affine space supporting the respective limiting measure is neither almost surely contained in the interior , nor almost surely contained in the exterior of the rejection region .
rather it is almost surely contained in the boundary of the rejection region .
therefore , in contrast to parts 1 and 2 , it is not only important that the measures concentrate on the respective one - dimensional space , but also _ how _ they concentrate ( cf . remark [ remarkthmlrvpw ] ( v ) ) .
the concentration turns out to be such that eventually the measures put roughly equal weight onto the rejection region and onto its complement , resulting in rejection probabilities as large as @xmath276 under the null .
we point out that the proof idea used to establish part 3 is inspired by the proof of theorem 2.20 in @xcite .
the last part of the theorem considers the case where one of the vectors @xmath329 or @xmath330 is an element of @xmath342 that is also ` involved ' in the hypothesis .
it is then shown that the size of the test is one if the global condition @xmath252 is satisfied .
we recall that if this condition fails to hold , then the test @xmath278 based on @xmath197 breaks down in a trivial way , because @xmath278 is then zero everywhere .
therefore we see that under assumption [ aar(1 ) ] one simply can not test a hypothesis involving @xmath343 or @xmath344 by means of a test @xmath278 based on @xmath197 with @xmath108 , @xmath105 , @xmath106 satisfying assumption [ weightspwrb ] ( this in particular covers the location model where @xmath345 , cf .
also lemma [ nulln*pw ] , part 2 ) .
[ boundaway ] suppose that it is _ known a priori _ that for some ( fixed ) @xmath346 $ ] the covariance model @xmath16 does not contain ar(1 ) correlation matrices @xmath347 with @xmath348 ; i.e. , instead of assumption [ aar(1 ) ] the covariance model @xmath16 satisfies @xmath349 inspection of the proof of theorem [ thmlrvpw ] then shows that a version of theorem [ thmlrvpw ] holds , in which all references to @xmath330 are deleted in parts 1 - 4 .
for example , part 4 of this version of theorem [ thmlrvpw ] reads as follows : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` suppose that @xmath252 .
suppose further that @xmath303 and @xmath304 holds .
then @xmath307 holds for every @xmath289 and every @xmath290 .
in particular , the size of the test is equal to one . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this statement covers ( in particular ) the important special case of testing a restriction on the mean in a location model .
we make the following observations concerning this version of theorem [ thmlrvpw ] : * since @xmath332 is not necessarily an element of the singular boundary of the covariance model considered here , the result just described does not contain `` size equal to one''- or `` nuisance - minimal - power equal to zero''-statements that arise from covariance matrices approaching @xmath332 .
note , however , that the original theorem [ thmlrvpw ] implies by a continuity argument that if @xmath350 is small ( compared to sample size ) , then considerable size distortions or power deficiencies will nevertheless be present for covariance matrices in @xmath16 that are close to @xmath332 .
* consider the case where @xmath343 , i.e. , the regression contains an intercept , and where the hypothesis does not involve the intercept , i.e. , @xmath351 : then we see that parts 1 - 4 of the version of theorem [ thmlrvpw ] just obtained do not apply .
in fact , in this case we can establish a positive result concerning a test based on @xmath278 with @xmath218 , and based on a non - standard critical value that depends on @xmath350 .
this positive result , together with its restrictions , is discussed in remark [ rposbound ] . given a hypothesis @xmath225 the four sufficient conditions provided in the preceding theorem are conditions on the design matrix @xmath4 .
they depend on observable quantities only .
how these conditions can be checked is discussed in the subsequent paragraph : the first three parts of the theorem operate under the local assumption that the multivariate polynomial @xmath352 does not vanish at the point @xmath329 or @xmath330 , respectively .
the multivariate polynomial @xmath352 is explicitly constructed in the proof of lemma [ n*pw ] .
therefore , the condition that it does not vanish at specific data points can readily be checked .
some additional conditions needed in parts 1 - 3 of the theorem are formulated in terms of @xmath353 and @xmath354 , which are in fact independent of the specific @xmath334 chosen and therefore easy to calculate .
part 3 of the theorem requires the existence of @xmath320 or @xmath321 ( which is immaterial if @xmath319 as discussed in the preceding remark ) . again
the existence of the gradients is independent of the specific choice of @xmath334 .
sufficient conditions for the existence of the gradient , under the assumption that @xmath108 is continuously differentiable on the complement of a finite number of points , are provided in lemma [ gradientt ] in appendix [ aneg ] .
these conditions amount to checking whether or not @xmath355 or @xmath356 , respectively , is an element of a certain set determined by @xmath108 consisting of finitely many points .
in contrast to parts 1 - 3 , the fourth part of the theorem operates under the global assumption that the multivariate polynomial @xmath352 is not the zero polynomial . since the polynomial @xmath352 is explicitly constructed in the proof of lemma [ n*pw ] , the global assumption @xmath252 can either be checked analytically , or by using standard algorithms for polynomial identity testing .
in addition to this global assumption , the fourth part needs additional assumptions on the structure of @xmath342 and the hypothesis @xmath225 which can of course be easily checked by the user .
the preceding theorem has given sufficient conditions on the design matrix , under which the test considered breaks down in terms of its size and/or power behavior .
however , for a given hypothesis @xmath225 there exist elements of @xmath357 to which the theorem is not applicable . as a consequence , the question remains to ` how many ' elements of @xmath211 the theorem can be applied once @xmath225 has been fixed .
this question is studied subsequently .
it is shown that generically in the space of all design matrices at least one of the four conditions of theorem [ thmlrvpw ] applies .
the first part of the proposition establishes this genericity result in the class of all design matrices of full column rank , i.e. , @xmath211 .
the remaining parts establish the genericity result in case @xmath94 and the first column of @xmath4 is the intercept , i.e. , @xmath358 with @xmath257 .
before we state the proposition , we introduce two assumptions on the kernel @xmath108 .
the first assumption is satisfied by all kernels typically used in practice .
[ cd ] the kernel @xmath108 is continuously differentiable on the complement of @xmath359 , a set consisting of finitely many elements .
the second assumption , which is used in some statements of the second part of the genericity result , imposes compactness of the support of the kernel .
this is satisfied by many kernels used in practice , e.g. , the bartlett kernel or the parzen kernel , but is not satisfied by the quadratic - spectral kernel .
[ cdl ] the support of @xmath108 is compact .
the genericity result is now as follows , where several quantities are equipped with the additional subindex @xmath4 to stress their dependence on the design matrix .
[ generic ] fix a hypothesis @xmath225 such that @xmath360 .
let @xmath108 , @xmath105 , @xmath106 satisfy assumption [ weightspwrb ] .
for @xmath251 let @xmath361 be the test statistic defined in with @xmath362 and let @xmath363 be arbitrary ( the sets defined below do not depend on the choice of @xmath364 ) .
fix a critical value @xmath280 such that @xmath275 .
then , the following holds . 1 .
suppose that @xmath244 , define @xmath365 and similarly define @xmath366 and @xmath367 .
then , @xmath368 and @xmath366 are @xmath369-null sets . if @xmath319 or if @xmath108 satisfies assumption [ cd ] , then @xmath370 and @xmath367 are @xmath371-null sets .
if assumption [ aar(1 ) ] holds , then the set of all design matrices @xmath372 for which the first three parts of theorem [ thmlrvpw ] do not apply is a subset of @xmath373 and hence is a @xmath369-null set if @xmath319 or if @xmath108 satisfies assumption [ cd ] ; it thus is a ` negligible ' subset of @xmath374 in view of the fact that @xmath374 differs from @xmath375 only by a @xmath369-null set .
2 . let @xmath94 and assume further that @xmath376 , where @xmath377 .
define @xmath378 then , @xmath379 is a @xmath380-null set .
furthermore , @xmath381 is a @xmath380-null set under each of the following conditions : 1 .
2 . @xmath382 and @xmath108 satisfies assumptions [ cd ] and [ cdl ] .
@xmath383 , @xmath106 is odd , @xmath384 for some @xmath385 and @xmath108 satisfies assumptions [ cd ] and [ cdl ] .
4 . @xmath108 satisfies assumption [ cd ] and @xmath386 on @xmath387 .
+ suppose that the first column of @xmath42 consists of zeros and that assumption [ aar(1 ) ] holds .
then , the set of all matrices @xmath388 such that the first three parts of theorem [ thmlrvpw ] do not apply to the design matrix @xmath389 is a subset of @xmath390 and hence is a @xmath391-null set if one of the conditions in ( a)-(d ) holds ; it thus is a ` negligible ' subset of @xmath392 in view of the fact that @xmath392 differs from @xmath393 only by a @xmath394-null set .
3 . suppose @xmath395 , that the first column of @xmath42 is nonzero and that assumption [ aar(1 ) ] holds .
then theorem [ thmlrvpw ] ( part 4 ) applies to the design matrix @xmath396 for every @xmath397 satisfying @xmath398 .
[ rgeneric ] ( i ) if @xmath241 and @xmath242 holds , the first part of lemma [ nulln*pw ] shows that the test trivially breaks down , since for every element @xmath4 of @xmath211 the test statistic @xmath361 is then constant on @xmath143 .
therefore , the assumption on @xmath18 in the first two parts of the proposition can in general not be substantially improved .
\(ii ) in the second part of the proposition , the analogously defined sets @xmath399 and @xmath400 clearly satisfy @xmath401 and @xmath402 .
\(iii ) in the third part of the proposition , if @xmath358 does not satisfy @xmath398 , then the test breaks down in a trivial way , since @xmath361 is then constant .
the first part of the preceding genericity result shows that if @xmath319 , or if the kernel satisfies assumption [ cd ] , then theorem [ thmlrvpw ] can be applied to generic elements of @xmath211 , i.e. , to all elements besides a @xmath253-null set . since all kernels used in practice , in particular the kernels emphasized in @xcite and @xcite , i.e. , the quadratic - spectral kernel and the bartlett kernel , respectively , satisfy assumption [ cd ] , this additional restriction on @xmath108 is practically immaterial
the second part of the proposition considers the situation where the first column of the design matrix is the intercept , which in addition is assumed not to be involved in the hypothesis in the sense that the first column of @xmath42 is zero .
in this situation it is shown that theorem [ thmlrvpw ] can generically be applied to design matrices of the form @xmath403 with @xmath257 , under certain sets of conditions on the triple @xmath108 , @xmath105 , @xmath106 .
we first discuss conditions ( a)-(c ) : 1 . in case @xmath319 no additional condition is needed for establishing generic applicability of theorem [ thmlrvpw ] .
2 . if @xmath382 , generic applicability of the negative result follows if the kernel satisfies assumptions [ cd ] and [ cdl ] , which applies to many kernels used in practice , but not to the quadratic - spectral kernel which is emphasized in @xcite .
3 . in case @xmath383
, the result shows that the procedure breaks down generically if @xmath106 is odd , @xmath384 for some @xmath385 and @xmath108 satisfies assumptions [ cd ] and [ cdl ] .
this seems to be restrictive .
however , the recommended procedure in @xcite is obtained by choosing @xmath108 the bartlett kernel , @xmath137 and @xmath404 , because in part 2 the first column of @xmath4 is the intercept .
therefore , we see that the recommended procedure in @xcite satisfies this condition . summarizing , we see that the conditions ( a)-(c ) in the proposition cover the recommended choices of @xmath108 , @xmath105 and @xmath106 in @xcite and @xcite . for all procedures that are not covered by conditions ( a)-(c ) , e.g. , the procedure in @xcite based on the quadratic - spectral kernel , one can typically obtain the genericity result by applying condition ( d ) , which ( under assumption [ cd ] ) is always satisfied apart from at most one exceptional critical value @xmath405 .
this is seen as follows : clearly , condition ( d ) depends on the critical value @xmath280 .
we see that if assumption [ cd ] is satisfied , then ( d ) can be violated for at most a single @xmath406 .
if this @xmath405 happens to coincide with @xmath280 , the condition is not satisfied and we can not draw the desired conclusion for this specific value of @xmath280 .
moreover , we immediately see that the condition must then be satisfied for any other choice @xmath407 , say .
therefore , generic applicability of the negative result follows for any value @xmath408 in that case .
this shows that even if one chooses a triple @xmath108 , @xmath105 , @xmath106 that does not allow for an application of ( a)-(c ) , one can not expect to obtain a procedure that has good finite sample size and power properties , because for all but at most one exceptional critical value the corresponding test is guaranteed to break down generically .
the third part of the proposition considers the case where the first column of the design matrix is the intercept , and where the coefficient corresponding to the intercept is restricted by the hypothesis . in this case
it follows that one can either apply part 4 of theorem [ thmlrvpw ] , or the test statistic is constant and hence the test breaks down in a trivial way ( cf .
remark [ rgeneric ] ) .
in the previous section we have established a ( generically applicable ) negative result concerning tests as in based on a prewhitened covariance estimator @xmath197 . in the present section
we first present a positive result concerning these tests under a _ non - generic _ condition on the design matrix .
then we introduce an adjustment procedure and establish a condition on the design matrix under which the adjustment procedure leads to improved tests .
finally we prove that this condition holds generically in the set of all design matrices .
both , the positive result concerning tests as in based on a prewhitened covariance estimator @xmath197 , and the results concerning the adjustment procedure are established under the following assumption on the covariance model @xmath16 .
[ approxaar(1 ) ] the set @xmath264 is norm - bounded and satisfies @xmath409 .
furthermore , for every sequence @xmath410 that converges to @xmath411 satisfying @xmath412 there exists a corresponding sequence @xmath413 such that @xmath414 as @xmath415 .
[ rapproxaar(1 ) ] ( i ) we first note that assumption [ approxaar(1 ) ] is stronger than assumption [ aar(1 ) ] .
therefore , under the former assumption the negative result established in section [ neg ] concerning tests as in based on a prewhitened covariance estimator @xmath197 _ does _ apply a fortiori . as a consequence ,
if @xmath16 satisfies assumption [ approxaar(1 ) ] , then positive results concerning size and power properties of tests of the form can only be established under non - generic assumptions on the design matrix .
however , as we shall show , positive results can generically be established for an _ adjusted _ version of such tests .
\(ii ) boundedness of @xmath16 is typically satisfied in our setup , as it is always satisfied if @xmath16 consists only of correlation matrices .
\(iii ) the last part of the assumption states that elements of @xmath16 that are ` close ' to being singular can be well approximated by ar(1 ) correlation matrices .
this , together with @xmath259 being a subset of @xmath16 , readily implies that the singular boundary of @xmath16 must coincide with @xmath416 .
therefore , we see that the assumption rules out the existence of rank deficient elements of @xmath417 with rank strictly greater than one . as an example , this rules out the case where @xmath16 is the correlation model corresponding to all stationary autoregressive processes of order less than or equal to two ( cf .
lemma g.2 in @xcite ) .
if this is not ruled out , however , further obstructions to good size and power properties can arise along suitable sequences approximating these boundary points ( cf . section 3.2.3 in @xcite ) .
the possibility of establishing positive results in settings like that is beyond the scope of the present paper and will be discussed elsewhere .
\(iv ) we note that assumption [ approxaar(1 ) ] is clearly satisfied for every covariance model of the form @xmath418 , where @xmath419 is a closed set consisting of positive definite correlation matrices . as an example , let @xmath420 be fixed and let @xmath421 denote the set of all correlation matrices corresponding to stationary moving average processes of an order not exceeding @xmath422 , i.e. , @xmath423 where @xmath424 denotes the @xmath15-dimensional correlation matrix corresponding to the spectral density @xmath425 ( cf .
equation ) . then @xmath426 satisfies assumption [ approxaar(1 ) ] , because every element of the closure of @xmath421 is a positive definite correlation matrix ( the latter statement follows from equation , compactness of the unit sphere in @xmath427 , and the dominated convergence theorem ) . under assumption [ approxaar(1 ) ] we shall subsequently establish a positive result concerning tests based on a test statistic @xmath278 as in with @xmath218 . in light of part ( i ) of the preceding remark
we already know that such a positive result can only be established under non - generic conditions on the design matrix .
in particular , the subsequent positive result considers the non - generic case where - besides @xmath428 , a condition that is generically satisfied under a mild constraint on @xmath18 ( cf .
lemma [ nulln*pw ] ) - the column span of the design matrix includes the vectors @xmath329 and @xmath330 and where @xmath429 holds .
[ excpw ] suppose that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] , and that @xmath16 satisfies assumption [ approxaar(1 ) ] .
let @xmath278 be the test statistic defined in equation with @xmath218 .
let @xmath430 be the rejection region , where @xmath280 is a real number satisfying @xmath431 .
suppose further that @xmath432 , @xmath433 and @xmath428 .
then , the following holds : 1 .
the size of the rejection region @xmath314 is strictly less than @xmath434 , i.e. , @xmath435 furthermore , @xmath436 2 .
the infimal power is bounded away from zero , i.e. , @xmath437 3 .
for every @xmath438 @xmath439 holds for @xmath440 and for any sequence @xmath441 satisfying @xmath442 with @xmath443 a singular matrix .
furthermore , for every sequence @xmath444 @xmath445 holds for @xmath440 whenever @xmath446 , @xmath447 , and @xmath448 is a closed subset of @xmath16 .
[ the very last statement even holds if one of the conditions @xmath449 and @xmath429 is violated .
] 4 . for every @xmath450 , @xmath451 ,
there exists a @xmath452 , @xmath453 , such that @xmath454 under the maintained assumptions on the hypothesis and the design proposition [ excpw ] shows that given any level of significance @xmath455 , a critical value can be chosen in such a way that the test obtained holds its size , while its nuisance - minimal power at every point @xmath456 in the alternative is bounded away from zero .
as theorem [ thmlrvpw ] in combination with proposition [ generic ] shows , this is impossible for generic elements of the space of all design matrices .
additionally , part 3 of the proposition shows that the power approaches one in certain parts of the parameter space corresponding to the alternative hypothesis .
these parts are characterized by @xmath457 being bounded away from zero and @xmath458 with @xmath443 being singular , or @xmath459 and @xmath458 with @xmath443 positive definite , and where in both cases @xmath460 is the parameter vector corresponding to @xmath456 ( note that @xmath461 is bounded from above and below by multiples of @xmath462 , where the constants involved are positive and depend only on @xmath4 , @xmath42 and @xmath238 ) .
[ rposbound ] suppose that instead of assumption [ approxaar(1 ) ] it is known that the covariance model satisfies the following variant of assumption [ approxaar(1 ) ] that rules out ar(1 ) correlation matrices @xmath347 with @xmath463 arbitrarily close to @xmath464 : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the covariance model @xmath264 is norm - bounded and there exists an @xmath346 $ ] such that @xmath465 ( cf .
remark [ boundaway ] ) .
furthermore , for every sequence @xmath410 that converges to @xmath411 satisfying @xmath412 there exists a corresponding sequence @xmath466 such that @xmath414 as @xmath415 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath278 and @xmath314 be defined as in proposition [ excpw ] above , and suppose further that @xmath343 , i.e. , the regression contains an intercept , that @xmath351 , i.e. , the hypothesis does not involve the intercept , and that @xmath252
. then one can show ( using essentially the same argument as in the proof of proposition [ excpw ] ) that the conclusions 1 - 4 of proposition [ excpw ] hold in this setup .
in particular , for any given @xmath467 there exists a critical value @xmath468 such that the test with rejection region @xmath469 has size not greater than @xmath470 .
this establishes a _ positive result _ concerning a test based on the test statistic @xmath278 with @xmath218 , and based on the non - standard critical value @xmath471 , which in practice can be obtained as explained in the discussion following theorem [ tu_3 ] .
however , the test obtained critically depends on @xmath350 , which in practice will typically be difficult to choose .
if one uses a test with critical region @xmath472 where @xmath473 , then the size of this test might exceed @xmath470 . in light of this drawback
, it is important to stress that autocorrelation - robust - testing is possible _ without _ imposing such an artificial condition that rules out ar(1 ) correlation matrices @xmath347 with @xmath463 arbitrarily close to @xmath464 : theorems [ tu_3 ] and [ genericadj ] in the present section show that ( at the small cost of including the artificial regressor @xmath330 ) a generic positive result can be obtained under the more natural assumption [ approxaar(1 ) ] . theorem [ tu_3 ] furthermore shows that including the artificial regressor @xmath330 leads to good power properties of the resulting test for covariance matrices close to @xmath474 .
proposition [ excpw ] assumes , among others , that @xmath4 satisfies @xmath475 , an assumption which is satisfied only by non - generic elements of the set of all design matrices . in the following
we shall now consider the ( generic ) situation where @xmath475 is _
violated_. proposition [ excpw ] is then clearly not applicable .
however , as we shall see , a positive result similar to proposition [ excpw ] can be established for an _ adjusted _ test statistic . to explain how the adjustment procedure works , suppose that we have a triple @xmath108 , @xmath105 , @xmath106 satisfying assumption [ weightspwrb ] which we want to use for covariance estimation .
suppose further that @xmath476 holds , which is typically satisfied .
the following theorem now shows that under certain conditions on the design matrix - which are shown to be generically satisfied in proposition [ genericadj ] below , and which in particular require @xmath475 to be violated - one can work with an _ adjusted _ test statistic that has improved size and power properties , and which is constructed as follows : instead of basing the construction of the test statistic on the true design matrix @xmath4 and on @xmath42 , we first construct an artificial design matrix @xmath477 of full column rank satisfying @xmath478 by adding the vectors @xmath329 and/or @xmath330 to @xmath4 .
furthermore , we construct a corresponding matrix @xmath479 , where zero columns are added to @xmath42 such that @xmath479 and @xmath477 have the same number of columns . then we construct a test statistic @xmath480 as in equation , but with @xmath4 and @xmath42 replaced by @xmath477 and @xmath479 , respectively , and where the covariance estimator is based on @xmath108 , @xmath105 and @xmath106 as above besides some minor updates in the construction of @xmath105 described below .
the subsequent theorem shows that if @xmath475 is violated , if every @xmath481 satisfies @xmath482 , if @xmath483 , and if an assumption on @xmath477 and @xmath479 analogous to the assumption @xmath428 in proposition [ excpw ] is satisfied , then for every critical value @xmath275 the test with critical region @xmath484 has the same properties ( 1)-(4 ) as @xmath314 in proposition [ excpw ] .
[ tu_3 ] suppose that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] , that @xmath106 additionally satisfies @xmath476 , and that @xmath16 satisfies assumption [ approxaar(1 ) ] .
suppose that one of the following ( mutually exclusive ) scenarios applies : 1 .
@xmath485 with @xmath486 , @xmath487 and @xmath488 .
let @xmath489 and define @xmath490 .
2 . @xmath491 , @xmath492 with @xmath493 and @xmath488 .
let @xmath494 and define @xmath490 .
3 . @xmath491 , @xmath495 with @xmath496 and @xmath497 .
let @xmath498 and define @xmath499 .
4 . @xmath491 , @xmath495 with @xmath500 and @xmath488 .
let @xmath494 and define @xmath490 .
then in all cases @xmath477 is a matrix of full column rank .
define @xmath501 0 & \text{else},\end{cases}\ ] ] where @xmath502 is the estimator one would obtain following steps 1 - 3 in section [ tpwc ] if @xmath477 was the underlying design matrix , @xmath503 was the hypothesis to be tested and where @xmath504 is defined as follows : in case @xmath319 we set @xmath505 ; in case @xmath382 we compute @xmath504 as outlined in section [ ma92 ] ( using as input @xmath506 as opposed to @xmath507 , and replacing @xmath167 by @xmath508 ) , with the same constants @xmath509 and @xmath510 and @xmath73 as used in the construction of @xmath105 , but with @xmath149 replaced by @xmath511 ; in case @xmath383 we compute @xmath504 as outlined in section [ mnw94 ] ( using as input @xmath506 as opposed to @xmath507 ) , with the same constants @xmath512 for @xmath513 and the same weights @xmath514 as used in the construction of @xmath105 , but with @xmath149 replaced by @xmath511 .
let @xmath515 be the rejection region where @xmath280 is a real number satisfying @xmath281 .
if @xmath516 ( where @xmath517 is the function obtained from lemma [ n*pw ] applied to @xmath108 , @xmath504 and @xmath106 and acting as if @xmath477 was the underlying design matrix and @xmath503 was the hypothesis to be tested ) , or equivalently if @xmath518 , then in each of the four scenarios above the conclusions ( 1)-(4 ) of proposition [ excpw ] hold with @xmath314 replaced by @xmath519 . the procedure outlined in the preceding theorem
is based on an artificial design matrix @xmath477 which is obtained from @xmath4 by adding either one or both elements of the set @xmath520 to the columns of @xmath4 .
if the so - obtained matrix @xmath477 satisfies @xmath521 , then the results from proposition [ excpw ] carry over to the rejection regions derived from @xmath480 .
in particular the adjusted test statistic @xmath480 leads to rejection regions the size of which is bounded away from one and such that the nuisance minimal power is bounded away from zero .
besides these improvements , the adjustment procedure is extremely convenient from a computational perspective , as the adjusted test statistic @xmath480 does not require any additional implementations .
it is based on the same algorithm as the calculation of @xmath278 , only with a different design matrix .
the theorem shows that for every level of significance @xmath455 , there exists a critical value @xmath522 such that the rejection region @xmath523 has size smaller than @xmath470 .
the critical value can be determined as follows : first of all , due to certain invariance properties of @xmath480 ( cf .
the proof of theorem [ tu_3 ] ) , the probabilities @xmath524 do not depend on @xmath525 and @xmath12 .
hence , for any fixed @xmath275 , the maximal rejection probability under the null can be approximated numerically by simulating the rejection probabilities from a finite subset of @xmath16 , and then doing a grid search . in a second step @xmath522
can be approximated by a line search exploiting monotonicity of @xmath524 in the critical value .
the adjustment procedure described in theorem [ tu_3 ] is applicable and yields an improved test under the assumption that @xmath475 is violated ( and hence the positive result in proposition [ excpw ] concerning the _ unadjusted _ test does not apply ) , that every @xmath481 satisfies @xmath482 , that @xmath526 and that @xmath521 .
given a hypothesis @xmath527 these are conditions on the design matrix @xmath4 .
our final result now shows ( under mild constraints on @xmath18 ) that these conditions are generically satisfied in the set of all design matrices @xmath211 ; and also in @xmath528 , the set of all design matrices the first column of which is the intercept , under the additional condition that the first column of @xmath42 is zero . under assumption [ approxaar(1 ) ] we hence see that although rejection regions based on @xmath278 generically break down as a consequence of proposition [ generic ] , this problem can generically be resolved by using rejection regions based on the adjusted test statistic @xmath480 , unless the regression includes an intercept and the first column of @xmath42 is nonzero .
[ genericadj ] fix a hypothesis @xmath225 with @xmath360 , suppose that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] , that @xmath106 additionally satisfies @xmath476 and that @xmath16 satisfies assumption [ approxaar(1 ) ] . then the following holds , where @xmath377 . 1 . if @xmath529 , then for @xmath253-almost every design matrix @xmath530 scenario 3 in theorem [ tu_3 ] applies , and the conclusions ( 1)-(4 ) of proposition [ excpw ] hold for any critical value @xmath275 with @xmath314 replaced by @xmath531 , where @xmath480 is constructed as outlined in theorem [ tu_3 ] .
2 . suppose that the first column of @xmath42 is zero , that @xmath94 and assume that @xmath532 holds .
then for @xmath256-almost every @xmath257 scenario 1 of theorem [ tu_3 ] applies to @xmath358 , and the conclusions ( 1)-(4 ) of proposition [ excpw ] hold for any critical value @xmath275 with @xmath314 replaced by @xmath531 , where @xmath480 is constructed as outlined in theorem [ tu_3 ] .
we have shown that tests for based on prewhitened covariance estimators and possibly data - dependent bandwidth parameters break down in finite samples in terms of their size or power properties .
this breakdown arises already for comparably simple covariance models such as @xmath533 .
we have also shown how a simple adjustment procedure can generically solve this problem in many cases . the test statistic obtained by applying
the adjustment procedure is of the same structural form as the test statistic based on estimators suggested by @xcite and @xcite and the test statistic in @xcite , but it is based on an artificial design matrix . therefore , the adjustment procedure does not only lead to improved size and power properties , but is also convenient from a computational point of view . for the adjustment procedure to work , assumption [ approxaar(1 ) ] has to be satisfied , which requires that elements of the covariance model @xmath16 that are close to being singular are well approximated by ar(1 ) correlation matrices . if and how the adjustment procedure can be extended to settings where this approximation condition is not satisfied is currently under investigation .
appendices * additional notation : * for the sake of clarity we shall repeatedly stress the dependence of @xmath534 , @xmath535 , @xmath536 , @xmath537 , @xmath538 , @xmath539 , @xmath197 and @xmath540 on the design matrix @xmath4 by writing @xmath541 , @xmath542 , @xmath543 , @xmath544 , @xmath545 , @xmath546 , @xmath547 and @xmath548 in the following proofs . at various places
we shall use the following notation : given a matrix @xmath549 and indices @xmath550 and @xmath551 we denote by @xmath552_{ij } = m_{ij}$ ] the @xmath202-th coordinate of @xmath105 , by @xmath552_{\cdot j } = m_{\cdot j}$ ] the @xmath553 column of @xmath105 and by @xmath552_{i\cdot } = m_{i\cdot}$ ] the @xmath554 row of @xmath105 . in case
@xmath555 we write @xmath552_{i } = m_i$ ] instead of @xmath552_{i1}$ ] .
since @xmath4 is a matrix of full column rank by assumption , we clearly have @xmath556 . from the definition of @xmath547
we see that @xmath557 , i.e. , @xmath558 is not well defined , if and only if one of the following conditions is satisfied ( cf .
remark [ wdpsi ] ) : using @xmath563 we see that the coordinates of @xmath564 and of @xmath565 are values of certain multivariate polynomials defined on @xmath566 evaluated at the point @xmath567 .
since ( i ) is equivalent to @xmath568 this shows that ( i ) is equivalent to @xmath569 , say , where @xmath570 is a multivariate polynomial which is clearly independent of @xmath225 . using this equivalence , condition ( ii )
is seen to be equivalent to @xmath571 and @xmath561 .
because of @xmath571 we have @xmath572 where @xmath573 . using this together with similar arguments as above we see that pre - multiplying @xmath574 by @xmath575 results in a matrix , the entries of which are values of certain multivariate polynomials , defined on @xmath566 , evaluated at the point @xmath567
it follows that the second equation in ( ii ) can be replaced by @xmath576^k \det\left(i_k-\sum_{l = 1}^p \hat{a}^{(p)}_{l , x}(y)\right ) = 0,\ ] ] where @xmath577 is a multivariate polynomial which is independent of @xmath225 either . summarizing our observations concerning ( i ) and ( ii )
we see that @xmath578 & \hspace{1.5 cm } \cup { \left\ { y \in { \mathbb{r}}^n : g_1(y , x)g_2(y , x ) \neq 0 \text { and } m(y ) \text { not w.d . } \right\}}.\end{aligned}\ ] ] the set in the second line of the previous display depends on the specific bandwidth @xmath105 .
hence , we have to distinguish three cases : suppose first that @xmath579 , i.e. , @xmath105 is a constant which is functionally independent of @xmath109 , @xmath4 and thus everywhere well defined on @xmath143 .
define @xmath580 , so that @xmath581 is a multivariate polynomial . noting that @xmath582 then proves the statement in case @xmath319 , because @xmath583 is independent of @xmath225 .
next we consider the case @xmath584 , where we write @xmath585 instead of @xmath586 , because the argument and the resulting function @xmath587 do not depend on @xmath73 and @xmath588 .
we partition @xmath589 where @xmath590 and @xmath591 are disjoint and defined as @xmath592 the equality in is readily seen from the definition of @xmath585 .
we want to obtain more suitable characterizations of @xmath590 and @xmath591 and proceed in two steps : ( i ) first , we claim that @xmath593 if and only if @xmath594_{ij } [ \hat{z}_x(y)]_{i(j-1)}\right]^2 - \left [ \sum_{j = 1}^{n - p-1 } [ \hat{z}_x(y)]_{i j}^2\right]^2 \right ) = 0.\ ] ] to see this assume that @xmath595 holds : suppose that @xmath596 is not well defined .
the latter occurs if and only if @xmath597_{{i^*}j}^2=0 $ ] , i.e. , all summands are zero , which immediately implies @xmath598_{{i^*}j } [ \hat{z}_x(y)]_{{i^*}(j-1 ) } = 0 $ ] . therefore , the factor corresponding to index @xmath599 vanishes and thus the product defining the second equation in vanishes .
that @xmath600 implies that the product vanishes is obvious . to prove the other direction assume that @xmath595 and that the product vanishes
this implies that at least one factor with index @xmath599 , say , equals zero , which implies that either @xmath596 is not well defined or @xmath600 holds .
this proves the claim .
secondly , we recall that if @xmath595 , then @xmath601 . using an argument as above it is then easy to see that @xmath602 pre - multiplied by @xmath603 gives a matrix , the entries of which are values of certain multivariate polynomials defined on @xmath566 evaluated at the point @xmath567 .
thus , if we multiply the second equation in by the @xmath604-th power of the expression in the previous display we see that equation can equivalently be written as @xmath605 where @xmath606 is a multivariate polynomial that is independent of @xmath225 . summarizing
, we have shown that @xmath607 ( ii ) first we observe that @xmath608 if and only if @xmath609_{ij } - \hat{\rho}_i(y ) [ \hat{z}_x(y)]_{i(j-1)}\right)^2 = 0,\ ] ] where we recall that by assumption @xmath149 is functionally independent of @xmath109 and @xmath4 . because @xmath610 implies that @xmath611 is well defined for @xmath148 , which is equivalent to @xmath597_{{i}j}^2\neq0 $ ] for @xmath148 , the second equation in the previous display can be replaced by @xmath612_{{i}j}^2\right ] [ \hat{z}_x(y)]_{ij } - \left[\sum_{j = 2}^{n - p } [ \hat{z}_x(y)]_{{i}j } [ \hat{z}_x(y)]_{{i}(j-1 ) } \right ] [ \hat{z}_x(y)]_{i(j-1)}\right)^2 = 0.\ ] ] we now multiply the function defining this equation by the @xmath613-th power of the expression in equation and denote the resulting function by @xmath614 .
the statement in equation is then seen to be equivalent to @xmath615 where @xmath616 is a multivariate polynomial .
we also see that @xmath617 is independent of @xmath225 .
we conclude that @xmath618 now let @xmath619 . by what has been shown above @xmath620
is a multivariate polynomial .
furthermore , @xmath621 does not depend on @xmath225 .
we observe that @xmath622 this proves the lemma in case @xmath382 . finally , we consider @xmath623 , where we write @xmath624 instead of @xmath625 , because the argument and the resulting polynomial are independent of @xmath626 .
we use a similar argument as in the previous case .
we observe that if @xmath595 , then the function @xmath624 is not well defined if and only if @xmath627 where @xmath628_{\cdot j } \left([\hat{z}_x(y)]_{\cdot ( j-|i|)}\right)'\omega \text { for } |i| = 0 , \hdots n - p-1.\ ] ] since @xmath149 and @xmath514 are both functionally independent of @xmath4 and @xmath109 , we can pre - multiply equation by the square of the expression in equation to see that the statement @xmath595 and @xmath629 not being well defined is equivalent to @xmath630 where @xmath631 is a multivariate polynomial . the function @xmath632 is independent of @xmath225 . using these properties , defining @xmath633 , a function which does not depend on @xmath225 , and noting that @xmath634 then proves the claim in case @xmath383 . to establish parts 1 - 4 of the lemma we apply a similar argument as in the proof of lemma 3.1 in @xcite .
we observe that if @xmath635 , or equivalently @xmath230 , we can write @xmath558 as @xmath636 where @xmath637 is the symmetric toeplitz matrix with ones on the main diagonal , and where for @xmath638 its @xmath202-th coordinate is given by @xmath639 whenever @xmath640 , and by @xmath2 else .
recall that @xmath641 .
if @xmath123 we have @xmath642 . if @xmath643 the matrix @xmath644 is positive definite by assumption [ weightspwrb ]
therefore , in both cases the matrix @xmath644 is positive definite .
this immediately establishes parts 1 - 4 , where @xmath360 is used in proving part 4 ( we emphasize that @xmath558 can be nonnegative definite , singular , zero or positive definite only if it is well defined , i.e. , only if @xmath230 holds ) .
it remains to prove part 5 .
we recall that @xmath645 = 0 \right\}}.\ ] ] from part 2 of the present lemma we know that we can rewrite the second set to the right as @xmath646 = 0 \right\}}.\ ] ] for every @xmath647 we have with @xmath648 that ( using the same notation as in the proof of lemma [ npwrb ] ) @xmath649 can be written as @xmath650^{-1 } d(p ) \right)^{-1 } \\ & \hspace{1cm}\times \left(\hat{v}_{p , x}(y ) - \bar{v}_{p , x}(y ) \bar{v}'_{1 , x}(y)\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right]^{-1 } \hat{v}_{1 , x}(y)\right ) \\[7pt ] = & r(x'x)^{-1 } \det(x'x)^{-1 } \left(\det(\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] ) i_k - \bar{v}_{p , x}(y ) \bar{v}'_{1 , x}(y)\operatorname{adj}\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] d(p ) \right)^{-1 } \\ & \hspace{1 cm } \times \left(\det\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] \bar{v}_{p , x}(y ) - \bar{v}_{p , x}(y ) \bar{v}'_{1 , x}(y)\operatorname{adj}\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] \bar{v}_{1 , x}(y)\right ) \\[7pt ] = & \det(x'x)^{-2 } \det
\left(\det(\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] ) i_k - \bar{v}_{p , x}(y ) \bar{v}'_{1 , x}(y)\operatorname{adj}\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] d(p ) \right)^{-1 } \\ & \hspace{1cm}\times r\operatorname{adj}(x'x ) \operatorname{adj}\left(\det(\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] ) i_k - \bar{v}_{p , x}(y ) \bar{v}'_{1 , x}(y)\operatorname{adj}\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] d(p ) \right ) \\ & \hspace{1cm}\times \left(\det\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] \bar{v}_{p , x}(y ) - \bar{v}_{p , x}(y ) \bar{v}'_{1 , x}(y)\operatorname{adj}\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] \bar{v}_{1 , x}(y)\right ) \end{aligned}\ ] ] we therefore see that the coordinates of the matrix @xmath651 , say , which is obtained by pre - multiplying @xmath649 by the factor @xmath652 ) i_k - \bar{v}_{p , x}(y ) \bar{v}'_{1 , x}(y)\operatorname{adj}\left[\bar{v}_{1 , x}(y)\bar{v}'_{1 , x}(y)\right ] d(p ) \right)\ ] ] ( for later reference we note that @xmath653 is a multivariate polynomial ) , are values of certain multivariate polynomials defined on @xmath566 evaluated at @xmath567 .
furthermore , we can replace @xmath654 in equation by @xmath651 without changing the set .
this follows because @xmath655 implies @xmath656 if we combine this equivalent expression for with and lemma [ npwrb ] we obtain @xmath657 = 0 \right\}}.\ ] ] we next define @xmath658 $ ] . by lemma [ npwrb ]
we see that @xmath659 is a multivariate polynomial that does not depend on @xmath238 .
[ auxconstr ] let @xmath8 , @xmath6 and let @xmath225 be a hypothesis .
suppose that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] .
assume that the tuple @xmath660 satisfies for some @xmath661 : 1 .
@xmath662 has exactly @xmath663 nonzero columns with indices @xmath664 .
2 . @xmath665 for @xmath666 , and @xmath667 .
3 . if @xmath668 , then @xmath669 . otherwise ,
@xmath670_{\cdot j_i } : i = 1 , \hdots , t \right\ } } ) = \operatorname{span}({\left\ { [ \hat{v}_x(y)]_{\cdot j_i } : i = 2 , \hdots , t+1 \right\ } } ) = { \mathbb{r}}^k.\ ] ] 1 .
2 . under each of the following three conditions it follows that @xmath672 ( or equivalently @xmath673 ) : 1 . @xmath319 ; 2 .
@xmath382 , and every row vector of the matrix obtained from @xmath674 by deleting its last column is nonzero [ this is in particular satisfied if @xmath675 ; 3 .
@xmath383 , and either each coordinate of @xmath676 is non - negative , or each coordinate of @xmath676 is non - positive .
3 . for every @xmath677 such that @xmath678
, the tuple @xmath679 is an element of @xmath680 that satisfies ( a1 ) , ( a2 ) and ( a3 ) .
if @xmath94 and either @xmath681_{{1j_{i } } } > 0 $ ] for @xmath682 or @xmath681_{{1j_{i } } } < 0 $ ] for @xmath682 holds , then there exists a regular matrix @xmath683 such that the first columns of @xmath4 and @xmath684 , respectively , coincide and such that @xmath685 ( or equivalently @xmath686 ) .
denote the column vectors of @xmath662 by @xmath687 for @xmath688 . if @xmath689 , then by ( a3 ) the set @xmath690 and @xmath691 , respectively , spans @xmath692 . using ( a3 ) , we now show that this is automatically satisfied in case @xmath668 . to prove this claim , we first recall that @xmath693_j x_{j \cdot}'$ ] .
we see that @xmath694 implies @xmath695_j x_{j \cdot } ' = \sum_{j = 1}^n v_j = \sum_{i = 1}^{k+1 } v_{j_i},\ ] ] where the third equality follows from ( a1 ) ( @xmath668 ) .
this shows that @xmath696 by ( a3 ) @xmath669 , which together with ( a1 ) implies that @xmath697 .
therefore , it follows from the two equations in the previous display that @xmath698 and @xmath699 , respectively , spans @xmath692 .
hence the claim follows .
we next show that @xmath700 is well defined . for this
we have to verify that @xmath701 ( cf . remark [ wdpsi ] ) . the @xmath73-th column ( @xmath119 ) of @xmath702
is given by @xmath703 which is to be interpreted as @xmath704 if @xmath137 , as @xmath705 if @xmath706 etc .
for @xmath707 we define the @xmath708 dimensional auxiliary matrix @xmath709 , where @xmath710 denotes the @xmath711-th element of the canonical basis of @xmath712 ( and @xmath713 denotes the kronecker product ) .
the following claims are immediate consequences of the structure of @xmath662 implied by ( a1 ) - ( a2 ) : 1 .
@xmath714 is the first column of @xmath702 ; 2 . if @xmath715 , then @xmath716 for @xmath717 and @xmath707 are columns of @xmath702 ; 3 . if @xmath718 , then @xmath719 for @xmath720 are columns of @xmath702 . to see parts ( i ) and
( ii ) , we observe that ( a1 ) and ( a2 ) imply that for @xmath666 , there are at least @xmath106 zero columns between the columns @xmath721 and @xmath722 of @xmath662 .
equation together with @xmath723 then immediately implies parts ( i ) and ( ii ) .
now we consider part ( iii ) and hence assume that @xmath724 .
we start with the case @xmath725 .
every column of @xmath662 with index greater than @xmath726 is zero by assumption ( a1 ) .
by assumption ( a2 ) we have @xmath727 . together
, this implies that the column @xmath728 is followed by at least @xmath729 zero columns .
since @xmath730 by assumption ( a2 ) , the column @xmath728 is preceded by at least @xmath106 zero columns .
the assumption @xmath727 is equivalent to @xmath731 .
hence , denoting the @xmath732-dimensional zero matrix by @xmath733 , we can use equation with @xmath734 to see that @xmath735 is a column of @xmath702 , where in deriving the equality we made use of the already established fact that the column @xmath728 of @xmath662 is preceded by at least @xmath207 zero columns ( which implies that @xmath736 is the zero vector ) , and followed by at least @xmath729 zero columns ( which is used in case @xmath737 ) .
this proves the statement concerning @xmath738 . in case @xmath706
we are done .
if @xmath737 , then the statements concerning @xmath719 for @xmath739 follow from equation together with the equation in the previous display and the fact that @xmath728 is preceded by at least @xmath106 zero columns .
we now use ( i)-(iii ) together with @xmath740 to show that @xmath701 : the matrix @xmath702 is @xmath741 dimensional .
therefore , we must show that it has full row rank .
assume existence of a row vector @xmath742 , where @xmath743 for @xmath744 , such that @xmath745 holds . part ( i ) shows that @xmath746 . if @xmath715 , then part ( ii ) applied with @xmath747 shows that @xmath748 for @xmath717 . summarizing the cases @xmath749 and
@xmath715 we obtain @xmath750 for @xmath666 .
because @xmath690 spans @xmath692 , it follows that @xmath751 . if @xmath724 , part ( iii ) implies that @xmath752 for @xmath720 . if @xmath715 part ( ii ) implies @xmath753 for @xmath720 and @xmath717 .
summarizing again the cases @xmath749 and @xmath715 we obtain that for @xmath720 we have @xmath754 because @xmath691 spans @xmath692 , it follows from the previous display that @xmath755 for @xmath756 . since we already know that @xmath751 , we obtain @xmath757 and thus @xmath701 . therefore @xmath758 is well defined . to see that @xmath759 we observe that every nonzero column of @xmath662 besides the first one is preceded by at least @xmath106 zero columns . the matrix @xmath760 is obtained from @xmath662 by deleting the first @xmath761 columns .
this together with equation immediately implies @xmath762 and thus @xmath759 . to show that @xmath673 under the conditions ( ckv ) , ( cam ) and ( cnw ) , respectively , we first note that @xmath763
this follows , because @xmath759 implies @xmath764 , which together with @xmath765 shows that the vectors @xmath721 for @xmath682 ( which span @xmath692 ) are column vectors of @xmath674 . as a consequence of part 4 of lemma [ n*pw ] positive definiteness of @xmath558 and hence @xmath673 follows if we can show that @xmath558 is well defined . since @xmath759 implies invertibility of @xmath766 , it remains to show that @xmath105 is well defined at @xmath109 ( cf .
remark [ wdpsi ] ) .
this is trivially satisfied under condition ( ckv ) because @xmath319 is everywhere well defined .
suppose that condition ( cam ) holds , i.e. , @xmath382 and every row vector of the matrix obtained from @xmath674 by deleting the last column is nonzero . that the latter condition is satisfied
if @xmath767 follows because in that case the last column of @xmath674 is the zero vector and @xmath763 . under the assumption that every row vector of the matrix obtained from @xmath674 by deleting the last column is nonzero
it is obvious that the denominators in the definition of @xmath611 for @xmath148 , i.e. , @xmath768_{ij}^2 ~~ \text { for } ~~ i = 1 , \hdots , k,\ ] ] do not vanish .
therefore , @xmath611 for @xmath148 are well defined . using assumptions ( a1 ) and ( a2 ) together with @xmath761 and @xmath764
, it follows that there is always at least one zero column between two nonzero columns of @xmath674 .
therefore , it is clear that the numerators appearing in the definition of @xmath611 for @xmath148 , i.e. , @xmath769_{ij}[\hat{z}_x(y)]_{i(j-1 ) } ~~ \text { for } ~~
i = 1 , \hdots , k,\ ] ] must vanish .
it follows that @xmath770 for @xmath148 .
we finally show that @xmath771 for @xmath148 .
we note that @xmath770 for @xmath148 implies @xmath772 ^ 2_{ij } \text { for } i = 1 , \hdots , k,\ ] ] because of @xmath764 it follows from assumptions ( a1 ) and ( a2 ) that the first column of @xmath674 must be zero .
furthermore , we already know that @xmath763 .
this implies that the matrix @xmath773 , say , which is obtained from @xmath674 by deleting the first column , must be of full row rank @xmath167 .
consequently all rows of @xmath773 must be non - zero .
the previous display shows that @xmath774 for @xmath148 is , up to a positive factor , the squared euclidean norm of the @xmath775-th row of @xmath773 .
therefore @xmath771 for @xmath148 must hold .
therefore , we have shown that @xmath132 is well defined ( we even see that @xmath123 holds ) .
now we consider the case where condition ( cnw ) holds , i.e. , @xmath383 and every coordinate of @xmath676 is non - negative ( non - positive ) .
we have to show that @xmath776 the non - negativity ( non - positivity ) condition immediately implies @xmath777 for @xmath778 . furthermore , since @xmath763 and @xmath146 , by the definition of the weights vector
, we have @xmath779 .
this implies @xmath780 . by assumption @xmath174 and @xmath172 for @xmath781 .
therefore , the quantity in the previous display does not vanish and thus @xmath132 is well defined .
this proves the second part of the lemma .
we next prove part 3 .
let @xmath782 be a regular @xmath783 dimensional matrix .
first , we obviously have @xmath784 , because @xmath251 and @xmath782 is regular . secondly , since @xmath785
, using regularity of @xmath782 , we have that @xmath786 .
this immediately entails @xmath787 therefore , the tuple @xmath788 satisfies ( a1 ) , ( a2 ) and ( a3 ) , because @xmath567 does so and @xmath782 is regular .
it remains to prove part 4 .
we do this by constructing a @xmath782 as in part 3 such that the tuple @xmath788 satisfies condition ( ckv ) , ( cam ) or ( cnw ) , respectively , if @xmath319 , @xmath382 or @xmath383 , respectively . if @xmath319 we can obviously choose @xmath789 .
suppose that @xmath790 .
let @xmath677 be such that @xmath678 .
we specify this matrix later on .
from part 2 of the present lemma we see that the tuple @xmath679 satisfies ( a1 ) , ( a2 ) and ( a3 ) and therefore we can conclude from part 1 of the present lemma that @xmath791 , which implies @xmath792 .
together with the equation in the previous display , we see that @xmath793 we now want to choose @xmath677 ( regular ) such that 1 .
every row vector of the matrix obtained from @xmath794 by deleting the last column is nonzero , and 2 .
either every coordinate of @xmath795 is non - negative , or every coordinate of @xmath795 is non - positive , holds , and that additionally the first columns of @xmath4 and @xmath684 , respectively , coincide . by assumption
we either have @xmath681_{{1j_{i } } } > 0 $ ] for @xmath682 , or we have @xmath681_{{1j_{i } } } < 0 $ ] for @xmath682 .
consider the former ( latter ) case : let @xmath796 which is a regular matrix for every @xmath797 .
post - multiplying @xmath4 by @xmath798 has the same effect as adding @xmath799-times the first column of @xmath4 to all other columns , without changing the first column .
we observe that since @xmath800 is obtained from @xmath662 by deleting the first @xmath106 columns , the nonzero columns of @xmath760 are precisely the vectors @xmath681_{{\cdot j_{i}}}$ ] for @xmath682 because @xmath567 satisfies assumptions ( a1 ) and ( a2 ) .
therefore , it is obvious from equation , together with the assumed @xmath681_{{1j_{i } } } > 0 $ ] for @xmath682 ( @xmath681_{{1j_{i } } } < 0 $ ] for @xmath682 ) , that by choosing @xmath801 large enough , we can enforce that all nonzero columns of @xmath802 are coordinate - wise positive ( negative ) . consider ( i ) : using equation we see that @xmath803_{\cdot j_2}$ ] is a column of the matrix obtained by deleting the last column of @xmath802 .
this follows from @xmath804 ( a consequence of the assumptions @xmath805 and @xmath723 ) , which shows that @xmath803_{\cdot j_2}$ ] is a column of @xmath802 , together with @xmath806 ( a consequence of @xmath94 ) , which shows that it is not the last column of @xmath802 .
since all coordinates of @xmath803_{\cdot j_2}$ ] are positive ( negative ) by construction of @xmath807 , it follows that @xmath807 satisfies ( i ) above . to show ( ii ) we recall that @xmath149 is nonzero and coordinate - wise nonnegative .
since all nonzero columns of @xmath802 are coordinate - wise positive ( negative ) , it immediately follows that every coordinate of @xmath808 is non - negative ( non - positive ) . by construction
the first columns of @xmath4 and @xmath809 coincide .
this proves the claim .
we start with the first part .
let @xmath530 and @xmath226 be arbitrary but fixed .
we show that @xmath810 , which is equivalent to @xmath811 by part 5 of lemma [ n*pw ] .
if @xmath812 we are done
. suppose @xmath635 which is equivalent to @xmath230 by lemma [ npwrb ] .
we claim that @xmath813 must hold .
assuming this claim and using @xmath814 , @xmath251 which implies @xmath815 , and @xmath635 which implies @xmath816 , it then follows from the definition of @xmath817 in equation that @xmath818 . as a consequence , part 2 of lemma [ n*pw ] then shows that @xmath558 is singular , which implies @xmath810 . to prove @xmath813 we note that @xmath819 = \hat{v}_{p , x}(y)\pi_{\operatorname{span}(\hat{v}'_{1 , x}(y))^{\bot}}.\ ] ] we see from the previous display that @xmath763 , i.e. , @xmath674 having full row rank , is equivalent to @xmath820 and @xmath821 . using @xmath701 , a consequence of @xmath635 ( cf .
remark [ wdpsi ] ) , this implies @xmath822 but this is impossible , because the matrix @xmath823 is @xmath824 dimensional , which together with @xmath825 implies @xmath826 .
next , we prove part 2 of the lemma . under the present assumptions it is shown in part 1 of proposition [ generic ] that for @xmath253 almost every @xmath251 we have @xmath827 and therefore in particular @xmath252 .
this proves the first statement . to show the remaining statement we construct a @xmath109 such that @xmath828 .
note first that @xmath829 obviously implies @xmath830 .
let @xmath226 satisfy @xmath831 , @xmath832 and @xmath833 else .
[ note that this is feasible , i.e. , @xmath834 holds , because of @xmath830 and @xmath761 .
] we intend to apply part 2 of lemma [ auxconstr ] with @xmath835 to the tuple @xmath836 .
we first have to show that the tuple @xmath836 , which is clearly an element of @xmath680 , satisfies assumptions ( a1 ) , ( a2 ) and ( a3 ) . for this
we observe that @xmath837 , which implies @xmath838 and therefore @xmath839 hence ( a1 ) is satisfied , because @xmath840 and @xmath109 has only two nonzero coordinates .
the corresponding indices are @xmath723 and @xmath841 .
the first part of assumption ( a2 ) is therefore obviously satisfied .
the second part , i.e. , @xmath842 , follows immediately from @xmath830 .
assumption ( a3 ) follows from @xmath843 together with the previous display and @xmath668 .
therefore , @xmath844 follows as an application of part 1 of lemma [ auxconstr ] .
obviously ( ckv ) holds if @xmath319 . since @xmath845 it is also obvious that ( cnw ) holds if @xmath383 .
suppose now that @xmath382 holds .
in this case @xmath846 and therefore @xmath847 holds .
the latter implies @xmath848 .
consequently the statement in brackets in ( cam ) shows that the condition is satisfied .
[ a567pw ] assume that the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] .
assume further that @xmath252 .
then , @xmath852 and @xmath197 satisfy assumptions 5 , 6 , and 7 in @xcite with @xmath853 .
in fact , @xmath854 is nonnegative definite for every @xmath855 , and is positive definite @xmath856-almost everywhere .
the test statistic @xmath278 defined in equation , with @xmath218 , is invariant under the group @xmath851 and the rejection probabilities @xmath857 depend on @xmath858 only through @xmath859 ( in fact , only through @xmath860 ) , where @xmath861 corresponds to @xmath862 via @xmath863 .
the assumption @xmath252 together with part 5 of lemma [ n*pw ] implies that the algebraic set @xmath219 is a closed @xmath142-null set .
therefore , by lemma [ npwrb ] , it follows that the algebraic set @xmath864 is a closed @xmath142-null set as well .
we claim that @xmath197 is continuous ( and obviously well defined by definition ) on @xmath865 , because it can be written as a composition of continuous functions on this set : we recall from equation the representation @xmath866 we first observe that @xmath867 ( which was defined in equation ) is continuous on @xmath865 , because @xmath868 and hence @xmath869 and @xmath870 are continuous on @xmath865 . by considering each of the cases @xmath319 , @xmath383 and @xmath382 separately , it is easy to see that @xmath871 is continuous on @xmath865 .
the main diagonal entries of @xmath872 are by definition constant on @xmath865 .
therefore , it remains to show that all off - diagonal entries are continuous on @xmath865 .
each of them is of the form @xmath873 for some fixed @xmath781 if @xmath640 , and @xmath2 if @xmath123 .
since @xmath108 is a continuous function by assumption [ weightspwrb ] , and @xmath871 is continuous on @xmath865 and satisfies @xmath641 , it remains to check that @xmath874 as @xmath875 , which is a part of assumption [ weightspwrb ] .
this proves the claim .
since @xmath852 is well defined and continuous everywhere on @xmath143 , it follows that both @xmath852 and @xmath197 are well - defined and continuous on @xmath865 .
clearly , @xmath197 is symmetric on @xmath865 .
this proves part ( i ) of assumption 5 in @xcite . to prove the second part let @xmath876 , @xmath877 and @xmath878 .
we have to show that @xmath879 .
note that @xmath880 , which implies @xmath881 and @xmath882 .
the latter immediately leads ( considering each of the cases @xmath319 , @xmath383 and @xmath382 separately ) to @xmath883 , which in turn implies @xmath884 .
it then follows from the previous display and the definition of @xmath885 that @xmath886 .
therefore , we clearly have @xmath887 , which proves part ( ii ) of assumption 5 in @xcite , and where we have also established the equivariance property of @xmath197 required in part ( iii ) of assumption 5 in @xcite . that @xmath852 satisfies
the equivariance property in part ( iii ) of assumption 5 in @xcite is obvious .
it remains to show that @xmath197 is @xmath142-almost everywhere nonsingular on @xmath865 .
this is equivalent to @xmath888 being a @xmath142-null set .
this is obvious , since we have already observed that @xmath219 is a @xmath142-null set under the maintained assumptions .
this proves the claim concerning assumption 5 . that @xmath229 is nonnegative definite for every @xmath655 ( which is equivalent to @xmath230 by lemma [ npwrb ] )
has been shown in part 1 of lemma [ n*pw ] .
it follows that @xmath197 is positive definite on the complement of @xmath219 .
hence @xmath197 is @xmath142- almost everywhere positive definite .
this immediately shows that assumptions 6 and 7 in @xcite are satisfied .
the remaining two claims in the lemma now follow immediately from what has been established together with lemma 5.15 part 3 and proposition 5.4 in @xcite . in each part of the theorem we have @xmath252 .
in part 4 this is an explicit assumption . in the other parts this
is implied by the assumption that @xmath889 does not vanish at a specific point . as a consequence lemma [ a567pw ]
is applicable in all parts of the theorem .
we shall now apply the first two parts of corollary 5.17 in @xcite to prove the first two parts of the present theorem .
lemma [ a567pw ] shows that @xmath852 and @xmath197 satisfy assumption 5 in @xcite with @xmath890 .
furthermore , note that the set @xmath891 figuring corollary 5.17 of @xcite coincides with @xmath219 . by assumption [ aar(1 ) ] the spaces @xmath892 and @xmath893 are concentration spaces of @xmath16 ( cf .
lemma g.1 in @xcite ) .
hence , parts 1 and 2 of the present theorem now follow by applying the first two parts of corollary 5.17 in @xcite and remark 5.18(i ) in @xcite to @xmath894 as well as to @xmath895 , and by noting that part 5 of lemma [ n*pw ] shows that the statement @xmath896 translates into @xmath897 , with a similar translation if @xmath898 is replaced by @xmath899 . that the test is biased in part 2 of the theorem follows immediately from part 5 of lemma 5.15 in @xcite ( note that assumptions 5 and 6 in @xcite are satisfied by lemma [ a567pw ] ) showing that @xmath314 contains a non - empty open set . to prove part 4 we apply theorem 5.19 in @xcite .
lemma [ a567pw ] shows that @xmath852 and @xmath197 also satisfy assumption 7 in @xcite .
we consider the case where @xmath343 and @xmath900 .
the other case can be handled analogously . from remark 5.20 in @xcite
we see that all conditions on the covariance model in theorem 5.19 in @xcite are satisfied with @xmath901 , @xmath902 and @xmath903 .
clearly , @xmath904 and @xmath905 holds @xmath906-a.e .
this shows that equation ( 33 ) in @xcite holds in the present setup . to conclude
, it remains to observe that @xmath907 in this equation equals one .
this follows from the discussion preceding theorem 5.19 in @xcite , because @xmath197 is almost everywhere positive definite by lemma [ a567pw ] .
now we consider part 3 of the theorem .
we prove the case where @xmath908 , @xmath298 and @xmath299 exists for some @xmath284 .
the other case works analogously .
the statement in the theorem saying that if @xmath299 exists and @xmath298 holds for _ some _ @xmath284 , then @xmath299 exists and @xmath298 holds for _ all _ @xmath284 follows at once from invariance of @xmath278 w.r.t .
@xmath909 , which holds as a consequence of lemma [ a567pw ] . in a first step
we now show that the linear functional on @xmath143 corresponding to the row vector @xmath910 does not vanish everywhere on @xmath911 : arguing by contradiction , assume that @xmath912 for every @xmath913 , which is equivalent to @xmath914 and therefore @xmath915 holds , i.e. , @xmath916 for some @xmath917 . since @xmath278 is @xmath909 invariant , it holds for every @xmath918 that @xmath919
hence on the set @xmath920 the mapping @xmath921 is constant , thus showing that the directional derivative of @xmath278 at the point @xmath922 in direction @xmath329 is zero .
the latter is equivalent to @xmath923 , and hence @xmath924 holds which implies @xmath925 .
to arrive at a contradiction it remains to show that there is a vector @xmath91 such that the directional derivative of @xmath278 at @xmath926 in direction @xmath91 does not vanish . to this end ,
recall that assumption 5 in @xcite is satisfied , hence the discussion following that assumption in @xcite shows that @xmath219 is invariant w.r.t .
therefore , @xmath928 implies @xmath929 .
since @xmath219 is closed by lemma [ n*pw ] , there exists an open ball @xmath930 of radius @xmath931 centered at @xmath932 such that @xmath933 .
additionally , we note that @xmath934 , because @xmath935 always holds ( see the discussion in @xcite after assumption 5 ) . therefore , @xmath936 is well defined and for @xmath937 we have @xmath938 .
assume that @xmath937 .
the ols estimator @xmath852 clearly satisfies @xmath939 where the second equality follows from @xmath940 . since @xmath197 satisfies the equivariance condition in assumption 5 of @xcite we can furthermore write @xmath941 by definition of @xmath278 ( cf . equation and recall that @xmath942 ) and the assumed equality @xmath298 , the relations derived above allow us to show that @xmath943 holds for every @xmath937 .
this implies that the directional derivative of @xmath278 in direction @xmath91 at the point @xmath944 equals @xmath945 , which is nonzero as a consequence of @xmath946 . in a second step
we shall now derive an expansion of @xmath278 at points of the form @xmath947 for @xmath109 satisfying @xmath948 : for every @xmath949 we have @xmath950 where @xmath951 as @xmath952 and @xmath953 . recall that @xmath278 is invariant under the group @xmath909 .
in particular for every @xmath109 such that @xmath954 we have @xmath955 where the first equality holds because of @xmath956 and the second follows from invariance of @xmath278 w.r.t .
this means that whenever @xmath948 holds , we can combine the equation in the previous display and equation with @xmath957 to see that @xmath958 holds and that @xmath959 for any sequence @xmath960 satisfying @xmath961 , @xmath962 and @xmath963 .
now , we choose a sequence @xmath413 such that @xmath271 and apply assumption [ aar(1 ) ] to obtain @xmath270 for every @xmath964 .
we intend to show that @xmath965 along a subsequence .
the last statement in lemma [ a567pw ] then implies @xmath966 for every pair @xmath967 and @xmath13 along this subsequence . in part 3 of lemma g.1 in @xcite
it is shown that @xmath968 , @xmath969 , where @xmath970 is a sequence of numbers such that @xmath971 , @xmath972 and @xmath86 is regular on @xmath911 .
furthermore , it is shown that @xmath973 .
we can use these relations to derive three useful facts : ( i ) observe that the matrix @xmath974 is an @xmath15-dimensional nonnegative square root of the symmetric matrix @xmath975 .
therefore , we can find an orthogonal matrix @xmath976 such that @xmath977 holds .
the sequence @xmath978 converges to @xmath979 as a consequence of @xmath980 together with the continuity of taking the nonnegative definite symmetric matrix square root of a symmetric and nonnegative definite matrix .
since @xmath976 is orthogonal we can choose a subsequence @xmath981 along which @xmath976 converges to @xmath982 , say . without loss of generality we henceforth assume @xmath983 . using the relation in the previous display
we find that @xmath974 converges to @xmath984 and we recall from above that @xmath985 is regular on @xmath911 .
( ii ) we note that @xmath986 implies @xmath987 ( iii ) we show that @xmath988 must hold : note that @xmath973 can be rewritten as @xmath989 that the term in brackets converges to @xmath990 by ( i ) and that the other term converges to @xmath991 by ( ii ) . therefore , @xmath992 or equivalently @xmath993 . but
this implies @xmath994 .
now , we are ready to show that @xmath965 .
let @xmath995 be a random n - vector defined on some underlying probability space such that the probability measure induced by @xmath995 on @xmath996 equals @xmath997 .
consequently , the random vector @xmath998 induces the distribution @xmath999 on @xmath996 . for notational convenience
we write @xmath1000 .
consequently , we have @xmath1001 \geq 0\right),\end{aligned}\ ] ] where we used @xmath1002 and @xmath1003 in deriving the second equality .
note that @xmath1004 and @xmath1005 on an event of probability one . using the expansion developed in equation
we see that with probability one @xmath1006 $ ] can be written as @xmath1007 \\ = & ~~ \frac{n}{e_{+}'\mathbf{g}_m } \operatorname{grad}t ( e_{+ } + \mu_0^ * ) s_m^{-1/2 } \pi_{\operatorname{span}(e_{+})^{\bot}}\mathbf{g}_m + { \vert { \frac{n}{e_{+}'\mathbf{g}_m } s_m^{-1/2 } \pi_{\operatorname{span}(e_{+})^{\bot}}\mathbf{g}_m } \vert } \\ & ~~~~~~~ \times ~~~~~ { \vert { \left(\frac{n}{e_{+}'\mathbf{g}_m } \pi_{\operatorname{span}(e_{+})^{\bot}}\mathbf{g}_m\right ) } \vert}^{-1 } q\left(\frac{n}{e_{+}'\mathbf{g}_m } \pi_{\operatorname{span}(e_{+})^{\bot}}\mathbf{g}_m\right).\end{aligned}\ ] ] to derive the almost sure limit as @xmath415 of the expression in the previous display we first observe that @xmath1008 converges point - wise to @xmath1009 because of ( ii ) . from that
it follows that @xmath1010 converges point - wise to @xmath1011 and that @xmath1012 converges point - wise to zero .
an application of the continuous mapping theorem hence shows that @xmath1013 almost surely as @xmath415 , which immediately implies @xmath1014 almost surely as @xmath415 as a consequence of equation together with @xmath1015 .
we also observe that ( i ) above implies @xmath1016 point - wise and thus , using the continuous mapping theorem again , we see that @xmath1017 almost surely as @xmath415 ( where the limiting random vector is well defined almost - surely ) .
this finally shows that @xmath1018 \rightarrow \frac{\sqrt{n}}{e_{+ } ' \mathbf{g } } \operatorname{grad}t ( e_{+ } + \mu_0^ * ) d^ * \mathbf{g},\ ] ] almost surely .
we already know from equation that @xmath1019 , where @xmath982 is an orthogonal matrix .
furthermore @xmath979 maps @xmath143 onto @xmath911 , and @xmath1020 does not vanish everywhere on @xmath911 .
hence , we see that the probability that the limiting random variable in the previous display takes on the value @xmath2 vanishes because @xmath1021 is a gaussian random variable with mean zero and positive variance .
hence , equation together with portmanteau theorem shows that @xmath1022 the covariance between the gaussian mean - zero random variables @xmath1021 and @xmath1023 is given by @xmath1024 where the equality follows from ( iii ) .
therefore , @xmath1023 and @xmath1021 are independent . since the probability to the right in equation equals the probability that the random variables @xmath1023 and @xmath1021 have the same sign it is now obvious that the limit equals @xmath276 . 1 . if @xmath319 and @xmath1025 , then @xmath1026 exists .
2 . suppose that @xmath790 , that @xmath108 is continuously differentiable on the non - void complement of a closed set @xmath359 , and that @xmath1025 .
assume further that one of the following conditions is satisfied : 1 .
@xmath640 and @xmath1027 for @xmath781 .
2 . @xmath123 and @xmath108 has compact support .
+ then @xmath1026 exists .
3 . if @xmath790 , then for every @xmath1028 we have @xmath1029 where @xmath1030 is a multivariate polynomial ( explicitly constructed in the proof ) that does not depend on the hypothesis @xmath225 .
we first verify parts 1 and 2 .
let us start by deriving a convenient expression for @xmath1031 under the assumption @xmath1025 .
by lemma [ n*pw ] the assumption @xmath1025 is equivalent to @xmath1032 .
an application of lemma [ a567pw ] shows that @xmath197 satisfies assumption 5 in @xcite .
an application of part ( ii ) of this assumption shows that @xmath1033 .
we can therefore use equation together with @xmath1034 and @xmath1035 ( both following from assumption 5 in @xcite ) to see that @xmath1032 implies @xmath1036 where in deriving the second equality we made use of the representation of @xmath229 developed in equation .
the function @xmath1037 is linear and hence totally differentiable on @xmath143 .
furthermore , in the proof of lemma [ n*pw ] it is shown that the coordinates of the matrix @xmath867 are multivariate rational functions ( without singularities ) on @xmath1038 .
in particular the coordinates of @xmath867 are continuously partially differentiable on @xmath1038 . to show that @xmath1026 exists at a given point @xmath1039
it is therefore sufficient to show that each off - diagonal element ( recall that the diagonal is constant ) of @xmath1040 is continuously partially differentiable on an open neighborhood of @xmath109 in @xmath1038 .
recall that the @xmath775-th off - diagonal element ( @xmath1041 ) of @xmath1040 evaluated at some @xmath1039 is given by @xmath1042 if @xmath319 the sufficient condition above is obviously satisfied , because in this case @xmath1043 is constant and therefore @xmath1044 is constant on @xmath1038 .
this proves part 1 of the lemma .
consider now part 2 . by considering separately the cases @xmath382 and @xmath383
, we observe that @xmath871 is continuously partially differentiable in an open neighborhood of any element @xmath109 of @xmath1038 satisfying @xmath640 .
we start with condition ( a ) .
let @xmath109 satisfy @xmath1032 and @xmath640 .
fix an @xmath1041 . by assumption @xmath108
is continuously differentiable on an open neighborhood of @xmath1045 .
hence there exists an open neighborhood @xmath982 of @xmath109 in @xmath1038 on which @xmath871 is strictly greater than zero and such that @xmath1046 is continuously partially differentiable on @xmath982 .
it hence follows that @xmath1044 is continuously partially differentiable on @xmath982 because it coincides with @xmath1046 on this set . to establish existence of the gradient under condition ( b )
let @xmath109 satisfy @xmath1032 and @xmath123 .
let @xmath775 be as before and recall that the support of @xmath108 is compact by assumption .
since @xmath105 is continuous on @xmath1038 , there exists an open neighborhood of @xmath109 in @xmath1038 such that for every point @xmath1047 in this neighborhood we either have @xmath1048 or that @xmath1049 is not contained in the support of @xmath108 .
it follows that the function @xmath1050 is constant equal to @xmath2 , and thus is in particular continuously partially differentiable , on this neighborhood . to prove the third part of the lemma consider first the case @xmath1051 , where we dropped the index @xmath588 because the argument and the resulting polynomial do not depend on it .
suppose @xmath109 satisfies @xmath1052
. then @xmath1053 is well defined and by definition @xmath1054 if and only if @xmath1055 holds , where @xmath509 and @xmath510 are positive constants .
this can equivalently be written as @xmath1056 which , after multiplying both sides of the equation by @xmath1057 ( which is nonzero ) , is seen to be equivalent to @xmath1058 \\ & \hspace{2 cm } - ~~ \delta^ * { \sum \limits_{i = 1}^{k}}\omega_i \left[\hat{\sigma}^4_i(y ) ( 1+\hat{\rho}_i(y))^2 ( 1-\hat{\rho}_i(y))^2 \prod_{j \neq i}^k ( 1-\hat{\rho}_j(y))^6(1+\hat{\rho}_j(y))^2 \right ] = 0 \end{aligned}\ ] ] by multiplying both sides of this equation by a suitably large power of the products of the denominators of @xmath611 ( which are nonzero ) , we can write the preceding equation equivalently as @xmath1059 where each @xmath1060 for @xmath148 is a multivariate polynomial . in a final step
we multiply both sides of the equation by a suitably large power of the non - vanishing factor in equation to obtain an equivalent equation of the form @xmath1061 where each @xmath1062 is a multivariate polynomial . therefore , the condition @xmath1063 can be equivalently stated as @xmath1064 finally , we note that the multivariate polynomial @xmath1065 does not depend on the hypothesis @xmath225 .
this proves the last part of the lemma in case @xmath1066 .
the proof of the case @xmath1067 is almost identical and therefore we omit it .
we finally note that similar arguments can be used to prove the statement in case @xmath383 , but we omit details .
we first prove that the sets @xmath1068 , @xmath1069 , @xmath1070 and @xmath1071 do not depend on the specific choice of @xmath1072 .
this follows from an invariance argument .
consider for example the set @xmath1068 , which is by definition a subset of @xmath1073 .
every element @xmath4 of this superset satisfies @xmath1074 .
hence , for every such @xmath4 the corresponding test statistic @xmath361 is invariant w.r.t .
@xmath1075 by lemma [ a567pw ] .
it now immediately follows that @xmath1076 does not depend on the specific choice of @xmath1077 .
the same argument shows that the statement in part 2 condition ( d ) is independent of the specific choice of @xmath1078 .
we shall now prove the three main parts of the proposition and start with the first : \1 ) we begin with the statement concerning @xmath368 . under the maintained assumptions we know from part 5 of lemma [ n*pw ]
that @xmath1079 is a multivariate polynomial .
this immediately implies that @xmath1080 is a multivariate polynomial , showing that @xmath1081 is an algebraic superset of @xmath368 .
it hence suffices to show that the set in the previous display is a @xmath253- null set , or equivalently to show that @xmath1082 . to this end
we shall use lemma [ auxconstr ] ( with @xmath668 ) to construct a matrix @xmath251 such that @xmath1083 .
let @xmath1084 be an auxiliary matrix the column vectors of which span @xmath1085 , where @xmath1086 is the vector obtained from @xmath329 by selecting the coordinates with indices @xmath1087 for @xmath1088 ( we shall need a similar construction for @xmath330 later on ) .
note that this selection is feasible because @xmath1089 holds as a consequence of the assumption @xmath1090 - \mathbf{1}_{\mathbb{m}_{am}}(m ) \geq 0 $ ] together with @xmath761 .
we also note that @xmath1091 does not contain a row consisting of zeros only .
if the construction of @xmath105 involves a weights vector @xmath149 ( which is assumed to be functionally independent of the design ) we choose the columns of @xmath1091 in such a way that @xmath1092 which is possible because this vector is an element of @xmath1085 and @xmath146 holds .
we now let @xmath1093 be the matrix the non - zero rows of which are precisely @xmath1094 for @xmath1088 , i.e. , @xmath1095 where @xmath1096 denotes the @xmath732-dimensional zero matrix ( here @xmath1097 vanishes if @xmath1098 ) .
obviously @xmath1099 holds , which implies @xmath251 .
furthermore , @xmath1100 holds , which follows immediately from @xmath1101 , because the non - zero columns of @xmath4 have column indices @xmath1102 for @xmath1088 by construction .
therefore , we see that @xmath1103 showing that @xmath1104 now we apply lemma [ auxconstr ] with @xmath668 to @xmath1105 . that the tuple @xmath1106 satisfies ( a1 ) of that lemma is obvious from the preceding display and equation .
we also see that ( a2 ) is satisfied because @xmath1107 for @xmath1108 , and @xmath244 implies @xmath1109 . that ( a3 ) is satisfied follows from the preceding display together with @xmath1099 .
to infer @xmath1110 from part 2 of lemma [ auxconstr ] , we consider three cases : first , if @xmath319 ( ckv ) is obviously satisfied and we are done .
secondly , assume that @xmath382 . in this case
we have by assumption @xmath1111 which implies @xmath1112 .
this shows that ( cam ) is satisfied .
thirdly suppose that @xmath383 .
since @xmath1113 follows from part 1 of lemma [ auxconstr ] , we see that @xmath1114 and hence that the nonzero columns of @xmath1115 are precisely @xmath1116 for @xmath1117 . by construction
we have @xmath1118 and @xmath1119 for @xmath1120 .
this shows that exactly one coordinate of @xmath1121 is non - zero which implies that ( cnw ) holds . to show that @xmath1122 is a @xmath1123null set , we can use a similar construction : we replace @xmath329 by @xmath330 throughout .
if @xmath106 is even we then have @xmath1124 .
if @xmath106 is odd we then have @xmath1125 . furthermore ,
if the construction of @xmath105 involves a weights vector we choose @xmath1091 such that @xmath1126 if @xmath106 is even , and @xmath1092 if @xmath106 is odd .
the remaining arguments are identical .
now consider @xmath1127 .
using part 1 of lemma [ gradientt ] we see that in case @xmath319 the set @xmath1068 is empty , because @xmath1128 exists whenever @xmath1129 .
next consider the cases where @xmath790 and where @xmath108 satisfies assumption [ cd ] . from part 2a of lemma [ gradientt ]
we know that for @xmath1129 the non - existence of @xmath1128 implies either @xmath1130 or @xmath1131 for some @xmath781 .
the latter two cases can clearly be summarized as @xmath1132 , where @xmath1133 is a set consisting of finitely many elements .
therefore , @xmath1134 we use part 3 of lemma [ gradientt ] to rewrite the latter set as @xmath1135 which is clearly a subset of @xmath1136 part 3 of lemma [ gradientt ] shows that @xmath1137 is a multivariate polynomial . we consider two cases : first assume that @xmath1138 .
consequently , the set in the previous display is a @xmath253-null set .
it hence follows that @xmath1068 is a @xmath253-null set and we are done .
next , assume that @xmath1139 .
it follows that there must exist a single index @xmath775 such that @xmath1140 holds [ this is easily shown by contradiction ] .
part 3 of lemma [ gradientt ] hence shows that @xmath251 and @xmath827 , i.e. , @xmath1129 , implies @xmath1141 .
clearly , @xmath1142 if @xmath1129 , then ( cf . the argument in the beginning of the proof of lemma [ gradientt ] applied to @xmath1143 ) @xmath1144 furthermore , since @xmath1129 implies @xmath1141 , the matrix @xmath1145 is constant @xmath1146 , say , on @xmath1073 . hence , for @xmath1129 , the statement @xmath1147 is equivalent to @xmath1148'r'\operatorname{adj}(\bar{b}_{p , x}(e_+)\bar{\mathcal{w}}_{n - p}\bar{b}'_{p , x}((e_+ ) ) ) r \left[\det(x'x ) \hat{\beta}_x(e_+ ) \right ] \\[7pt ] & \hspace{3 cm } = ~~ ~\frac{n}{n - p}\det(x'x)^2 \det(\bar{b}_{p , x}(e_+)\bar{\mathcal{w}}_{n - p}\bar{b}'_{p , x}((e_+ ) ) ) c\end{aligned}\ ] ] where @xmath1149 and @xmath1150 have been defined in the proof of lemma [ n*pw ] , where it is shown that @xmath1151 and @xmath251 ( which is weaker than @xmath1129 ) implies @xmath1152 .
furthermore , it is shown in the proof of lemma [ n*pw ] that @xmath1153_{ij } : { \mathbb{r}}^{n \times k } \rightarrow { \mathbb{r}}$ ] ( for @xmath1154 and @xmath1155 ) is a multivariate polynomial , and that @xmath1156 is a multivariate polynomial as well .
it is easily seen that the coordinates of @xmath1157 as a function of @xmath4 are multivariate polynomials . putting this together
we have shown that @xmath1158 where @xmath1159 is a multivariate polynomial . therefore , if we can show that @xmath1160 we obtain that @xmath1068 is a @xmath253-null set . in the proof of the part concerning @xmath1076 above
we have already constructed an @xmath1161 that satisfies @xmath1100 which implies @xmath1162 .
together with equation this shows that @xmath1163 holds for this specific @xmath4 . but
this immediately shows that @xmath1164 and we are done .
clearly , we can use an almost identical argument to prove the statement concerning @xmath1165 . under assumption [ aar(1 ) ] the set of matrices @xmath251 for which the first three cases of theorem [ thmlrvpw ] do not apply is obviously a subset of @xmath1166 .
hence the first part of the proposition follows .
\2 ) we start with the statement concerning @xmath1167 . under the maintained assumptions we know from part 5 of lemma [ n*pw ]
that @xmath1168 is a multivariate polynomial .
this immediately implies that @xmath1169 is a multivariate polynomial , which shows that @xmath1170 is an algebraic superset of @xmath1167 .
it hence suffices to show that the set in the previous display is a @xmath256- null set , or equivalently to show that @xmath1171 .
again , we shall use lemma [ auxconstr ] with @xmath668 to construct a matrix @xmath257 such that @xmath1172 .
the situation here is more complicated than in the first part , because the first column of the design matrix we seek has to be the intercept . for our construction
we need some additional ingredients : by definition @xmath1173 if @xmath106 is odd , and @xmath1174 if @xmath106 is even .
if @xmath106 is odd set @xmath1175 , where @xmath1176 is the vector obtained from @xmath330 by selecting the coordinates @xmath1177 for @xmath148 and @xmath1178 , where @xmath1087 for @xmath1088 was defined in part 1 above .
this selection is feasible , because by assumption we have @xmath1179 , which , since @xmath106 is odd , gives @xmath1180 , implying that @xmath1181 , because of @xmath761 . if @xmath106 is even set @xmath1182 , the vector obtained from @xmath330 by selecting the coordinates @xmath1183 for @xmath1088 .
next , define @xmath1184 we claim that @xmath91 and @xmath1185 satisfy @xmath1186 , @xmath1187 is linearly independent of @xmath1188 and @xmath1185 is orthogonal to @xmath1189 .
the latter property is clearly always satisfied , regardless of whether @xmath106 is even or odd .
we thus only have to verify the first two conditions .
we start with the case @xmath106 odd . here
we have @xmath1190 and therefore @xmath1191 .
furthermore @xmath1192 can not equal @xmath1193 for some @xmath917 , because the last and the last but one coordinate of @xmath91 are unequal . for @xmath106 even @xmath1194 , where @xmath1195 equals either @xmath2 ( if @xmath167 is even ) or @xmath434 ( if @xmath167 is odd ) , therefore @xmath1196 holds and thus @xmath1191 .
furthermore @xmath1197 can not equal @xmath1193 for some @xmath917 , because the second and third coordinate of @xmath91 are unequal .
this proves the claim . using these properties
, we see that @xmath1198 and @xmath1199 where we have used that @xmath1185 is orthogonal to both @xmath1189 and @xmath68 to derive the third equality .
we shall now define an auxiliary matrix .
let @xmath1200 denote a @xmath1201-dimensional matrix such that @xmath1202 , @xmath1203 and such that the remaining @xmath1204 columns @xmath1205 for @xmath1206 are linearly independent and orthogonal to @xmath1207 .
since @xmath1189 and @xmath68 are linearly independent , we have @xmath1208 . for later use
we observe that @xmath1209 where the first equality follows immediately from @xmath1205 for @xmath1206 being linearly independent and orthogonal to @xmath1207 .
this immediately shows @xmath1210 define the two @xmath167-vectors @xmath1211 and @xmath1212 .
let @xmath1093 be such that @xmath1213 for @xmath1088 , and if the index @xmath1214 , then let @xmath1215 if @xmath1216_j = 1 $ ] , and let @xmath1217 if @xmath1216_j = -1 $ ] . by construction
the matrix @xmath4 is of the form @xmath358 .
we claim that @xmath1218 . to see this denote the set of indices @xmath1219 such that @xmath1216_j = -1 $ ] by @xmath1220 , and the set of indices @xmath1219 such that @xmath1216_j = 1 $ ] by @xmath1221 .
the sum of squares @xmath1222 can be written as @xmath1223_{j^*_i } - x_{j^*_i \cdot } \beta)^2 + \sum_{j \in \mathcal{i}_- } ( -1 - r_- \beta)^2 + \sum_{j \in \mathcal{i}_+ } ( 1 - r_+ \beta)^2 \\ & = \sum_{i = 1}^{k+1 } ( v_i - l_{i \cdot } \beta)^2 + \sum_{j \in \mathcal{i}_- } ( -1 - r_- ' \beta)^2 + \sum_{j \in \mathcal{i}_+ } ( 1 - r_+ ' \beta)^2 \\ & = { \vert { v - l\beta } \vert}^2 + \sum_{j \in \mathcal{i}_- } ( -1 - r_- ' \beta)^2 + \sum_{j \in \mathcal{i}_+ } ( 1 - r_+ ' \beta)^2 \end{aligned}\ ] ] if we now plug in @xmath1224 and note that @xmath1225 and @xmath1226 we see that @xmath1227 this immediately proves the claim @xmath1218 .
hence , the residual vector satisfies @xmath1228_j = \begin{cases } u_i & \text { if } j = j^*_i \text { for some } i = 1 , \hdots , k+1 \\ 0 & \text{else}. \end{cases}\ ] ] this immediately entails that @xmath1229 equals @xmath1230}),\ ] ] where the indices of the nonzero columns of this matrix are precisely @xmath1231 for @xmath1088 , because the first column of @xmath1200 is @xmath1189 and @xmath1232 for @xmath1088 , the latter following since @xmath1198 and @xmath1233 for @xmath1088 by definition .
in deriving the dimension of @xmath1234}$ ] we used @xmath1235 now we apply lemma [ auxconstr ] ( with @xmath668 ) .
clearly , @xmath1236 . from equation , and the discussion following it , we see that the tuple @xmath1237 satisfies assumption ( a1 ) . assumption ( a2 ) is satisfied , because @xmath1238 for @xmath148 , and because we see from the previous display that @xmath1239 $ ] , which together with the assumption @xmath1240 implies @xmath1241 . assumption ( a3 ) is satisfied because @xmath1242 . if @xmath319 we are done .
consider the case @xmath382 .
we show that condition ( cam ) is satisfied .
but this is obvious , because the assumption @xmath1243 immediately implies @xmath1244 .
suppose @xmath383 .
we apply part 4 of lemma [ auxconstr ] . for this
we claim that either @xmath1245_{1j_i^ * } > 0 $ ] for @xmath1117 or @xmath1245_{1j_i^ * } < 0 $ ] for @xmath1117 . assuming that this claim is true
, the lemma shows that there exists a regular matrix @xmath677 such that @xmath784 , the first column of @xmath1246 is @xmath329 and @xmath1247 , and we are done . to prove the claim recall that by construction @xmath1198 holds , which shows that either @xmath1248 for @xmath1117 or @xmath1249 for @xmath1117 .
furthermore , the first column of @xmath1200 is the vector @xmath1250 .
equation now shows that @xmath1245_{1j_i^ * } = u_i$ ] for @xmath1117 .
this proves the claim .
the part of the statement concerning @xmath1251 is established by exploiting an argument similar to the one given in part 1 of the proof .
firstly , if @xmath319 , then we know from part 1 of lemma [ gradientt ] that @xmath1252 implies existence of @xmath1253 .
therefore , @xmath1251 is empty in case ( a ) .
it remains to prove the remaining three cases , in all of which assumption [ cd ] holds .
clearly we can also assume that @xmath790 .
we start with the following observation : combining assumption [ cd ] with part 2a of lemma [ gradientt ] as in part 1 of the proof , we see that there exists an integer @xmath1254 and real numbers @xmath1255 , such that @xmath1256 it follows with the same argument as in part 1 that either @xmath1251 is a @xmath256-null set , or there exists an index @xmath775 such that @xmath1257 . in the former case
we are done . in the latter case
one can show , with a similar argument as in part 1 of the proof , that @xmath1258 where @xmath1259 is a multivariate polynomial .
either we have that @xmath1260 and @xmath1071 is a @xmath1261-null set , or @xmath1262 and the superset in the previous display coincides with @xmath1263 . consider condition ( d ) .
if the function @xmath1264 is not constant @xmath280 on @xmath1263 , then @xmath1260 , and hence the superset in equation is a @xmath1261-null set .
this shows that @xmath1265 is a null set under condition ( d ) . for condition ( b )
we consider again the inclusion in equation .
either the superset is a null set and we are done , or there must exist a real number @xmath1266 such that @xmath1257 . assume the latter .
we exploit a property of the matrix @xmath358 with @xmath1267 constructed above . in the proof of part 2 ( cam ) of lemma [ auxconstr ]
it is shown that for this specific @xmath4 we have @xmath1268 .
therefore , @xmath1269 , or equivalently @xmath1268 for every design matrix @xmath358 with @xmath1267 .
but since the kernel satisfies assumption [ cdl ] , the existence of @xmath1270 for every @xmath1267 then follows from part 2b of lemma [ gradientt ] .
hence , @xmath1271 .
consider condition ( c ) .
we use a similar argument as under condition ( b ) : we establish the existence of a sequence of matrices @xmath1272 with @xmath1273 eventually in @xmath1263 , such that @xmath1274 as @xmath415 . from an argument as in the proof under condition ( b ) this then implies that either @xmath1071 is a null set , or that @xmath1275 on @xmath1263 .
but since @xmath108 satisfies assumption [ cdl ] , it then follows from part 2b of lemma [ gradientt ] that in the latter case @xmath1071 is empty .
this then proves the claim .
it remains to construct a sequence @xmath1273 as claimed . by assumption @xmath384 for some @xmath385 .
assume without loss of generality that @xmath1276 ( otherwise we have to interchange the columns of the @xmath1273 sequence to be constructed accordingly ) .
let @xmath358 be as constructed above .
let @xmath1277 be a sequence diverging to @xmath1278 . recall that by construction @xmath1279 for @xmath1088 .
since @xmath106 is odd , a simple calculation shows that @xmath1280 equals @xmath1281 let @xmath1282 and note that @xmath1283 define the @xmath783 dimensional regular matrix @xmath1284 clearly , post - multiplying a matrix with @xmath167 columns by @xmath1285 has the same effect as adding @xmath588 times the first column to the second column , then multiplying the column so obtained by @xmath1286 and leaving all other columns unchanged . since @xmath1202 and @xmath1287 , the expression for @xmath1288 above shows that the second column of @xmath1289 , has precisely two nonzero elements with indices @xmath434 and @xmath1290 , respectively .
since @xmath1291 and @xmath1292 satisfies ( a1)-(a3 ) as established above , part 3 of lemma [ auxconstr ] shows that @xmath1293 and that the tuple @xmath1294 satisfies ( a1 ) , ( a2 ) and ( a3 ) . hence @xmath1295 holds . as a consequence @xmath1296
is well defined for every @xmath964 . using @xmath1297 ( cf .
the proof of lemma [ auxconstr ] part 4 ) and @xmath1298 , which was shown above , we see that @xmath1299 must hold , which together with @xmath146 immediately implies @xmath1300 . since @xmath1301 is obtained from @xmath1302 by deleting its first @xmath106 columns , we observe , using the remark concerning the second column of @xmath1289 above together with equation , that the second row of @xmath1303 has exactly one non - zero coordinate , namely @xmath1304 .
consider @xmath1305_{\cdot j}[\hat{z } _
{ x_m}(y)]_{\cdot ( j-|i|)}'\omega}{\sum_{j=1}^{n - p } \omega'[\hat{z } _ { x_m}(y)]_{\cdot j}[\hat{z } _ { x_m}(y)]_{\cdot j}'\omega } \\
= & \frac{\sum_{j=|i| + 1}^{n - p } \bar{\omega}_m'[q ' \hat{z } _ { x}(y)]_{\cdot j}[q ' \hat{z }
_ { x}(y)]_{\cdot ( j-|i|)}'\bar{\omega}_m}{\sum_{j=1}^{n - p } \bar{\omega}_m'[q ' \hat{z } _ { x}(y)]_{\cdot j}[q ' \hat{z } _ { x}(y)]_{\cdot j}'\bar{\omega}_m } , \end{aligned}\ ] ] for @xmath781 ,
where @xmath1306 .
clearly @xmath1307 .
since the second row of @xmath1308 contains by construction precisely one nonzero entry , it follows that the limit of must be @xmath2 for @xmath1309 .
furthermore we have @xmath1310 .
it immediately follows from @xmath174 and the definition of @xmath105 that @xmath356 is well defined for @xmath964 large and that it converges to @xmath2 as @xmath415 .
the remaining part of the proposition is obvious .
\3 ) let @xmath849 and assume that @xmath1311 satisfies @xmath398 .
obviously , @xmath1312 .
note that @xmath1313 .
the first column of @xmath42 is non - zero .
therefore @xmath1314 .
thus we can ( since assumption [ aar(1 ) ] holds ) apply part 4 of theorem [ thmlrvpw ] .
we verify the assumptions of theorem 5.21 in @xcite with @xmath1315 and @xmath1316 . because @xmath252 by assumption , and since the triple @xmath108 , @xmath105
, @xmath106 satisfies assumption [ weightspwrb ] , we can use lemma [ a567pw ] to conclude that @xmath852 and @xmath197 satisfy assumptions 5 , 6 and 7 in @xcite , that @xmath197 is almost everywhere positive definite and that @xmath278 is invariant w.r.t .
@xmath909 . assumption [ approxaar(1 ) ] , remark [ rapproxaar(1 ) ] ( part ( iii ) ) together with remark 5.14 ( ii ) in @xcite now shows that @xmath1317 and that all assumptions on @xmath16 appearing in theorem 5.21 in @xcite are satisfied .
because @xmath449 is assumed we have @xmath1318 .
the assumption @xmath429 even implies @xmath1319 ( for some arbitrary @xmath967 ) .
invariance of @xmath278 w.r.t .
@xmath909 then shows that equation ( 34 ) in theorem 5.21 of @xcite is satisfied .
the assumptions on @xmath1320 appearing in parts 2 and 3 of that theorem are satisfied , because @xmath197 is positive definite almost everywhere .
the theorem now follows from theorem 5.21 in @xcite , using a standard subsequence argument , positive definiteness of every element of @xmath16 and compactness of @xmath448 , to obtain the second statement in part 3 from the corresponding part of theorem 5.21 in @xcite .
the claim in parenthesis in part 3 follows from the corresponding claim in parenthesis in theorem 5.21 of @xcite , together with the observation that the conditions on @xmath329 and @xmath330 have only been used to verify the condition in equation ( 34 ) of theorem 5.21 of @xcite ( cf .
the proof of theorem 3.7 in @xcite ) .
we apply theorem 5.21 of @xcite with the estimators @xmath1321 and @xmath1322 . obviously , the test statistic defined in equation ( 28 ) of @xcite based on these estimators coincides with the test statistic @xmath480 as defined in the statement of the present theorem . since the assumptions concerning @xmath16 in the present theorem are the same as in proposition [ excpw ] , we see from the proof of this proposition that it suffices to verify that @xmath1323 and @xmath1324 satisfy assumption 5 in @xcite , that @xmath1323 is almost everywhere positive definite ( implying that assumptions 6 and 7 in @xcite are satisfied ) , and that the invariance condition in equation ( 34 ) of @xcite is satisfied by @xmath480 . by definition , @xmath1323 is the estimator one would obtain following steps 1 - 3 of the construction in section [ tpwc ] based on @xmath108 , @xmath504 and @xmath106 , if @xmath477 was the underlying design matrix ( observe that @xmath477 is of full column rank ) and @xmath503 was the hypothesis to be tested . by assumption , the triple @xmath108 , @xmath105 , @xmath106 satisfies assumption [ weightspwrb ] w.r.t . the ( dimensions @xmath167 and @xmath18 of the ) design matrix @xmath4 and additionally @xmath1325 holds . from @xmath1326 , and the definition of @xmath504
it follows that the triple @xmath108 , @xmath504 , @xmath106 satisfies assumption [ weightspwrb ] w.r.t .
( the dimensions @xmath508 and @xmath18 of ) @xmath477 . furthermore , it is assumed that @xmath521 .
therefore , we can apply lemma [ a567pw ] , acting as if @xmath477 was the underlying design matrix , to conclude that @xmath1327 and @xmath1323 satisfy assumption 5 in @xcite with @xmath1328 and @xmath167 replaced by @xmath508 , @xmath4 replaced by @xmath477 and @xmath342 replaced by @xmath1329 ) .
furthermore , lemma [ a567pw ] shows that @xmath1323 is almost everywhere positive definite .
we now apply part 1 of proposition 5.23 in @xcite to obtain that @xmath1323 and @xmath1324 satisfy ( the original ) assumption 5 in @xcite , and that the invariance condition is satisfied . to this end , it suffices to verify that in each of the four cases we have @xmath1330 ( for some arbitrary @xmath967 ) .
this is obvious in the first three cases . for case 4 we can use exactly the same argument as in the proof of part 4 in theorem 3.8 in @xcite .
we begin with the proof of the first statement .
we note that for @xmath253- almost every @xmath251 case 3 of theorem [ tu_3 ] applies : since by assumption @xmath1331 and by definition @xmath1332 , we have @xmath1333 .
therefore , the set of matrices @xmath4 in @xmath1334 such that @xmath1335 holds is a @xmath253- null set
. hence , @xmath1336 and @xmath1337 holds for @xmath253- almost every @xmath251 .
it remains to verify that @xmath521 for almost every @xmath251 , where @xmath1338 , @xmath1339 and @xmath504 is constructed as outlined in theorem [ tu_3 ] .
for that it suffices to find a matrix @xmath1093 and a vector @xmath1340 such that @xmath1341 . to see this note that the triple @xmath108 , @xmath504 , @xmath106 satisfies assumption [ weightspwrb ] w.r.t .
( the dimensions of ) @xmath477 ( cf . the proof of theorem [ tu_3 ] )
. therefore , lemma [ n*pw ] shows that @xmath1342 is a multivariate polynomial .
if we can find a matrix @xmath4 and a vector @xmath109 as above , this implies that the multivariate polynomial @xmath1342 is not the zero polynomial , and therefore the zero set of this multivariate polynomial is a @xmath1343- null set .
it follows that for @xmath253- almost every @xmath4 we must have @xmath521 [ assuming the opposite , there exists a set @xmath1344 of positive @xmath253- measure such that @xmath1345 for every @xmath1346 , which implies that @xmath1347 is a subset of the zero set of @xmath1348 .
but clearly @xmath1349 has positive lebesgue measure , a contradiction . ] . in the following
we shall construct such a pair @xmath567 as above : let @xmath1350 and define @xmath1351 as @xmath1352 , @xmath1353 and @xmath1354
. by construction @xmath1355 is orthogonal to @xmath1356 and @xmath1357 . noting that @xmath1358_i = [ w_2]_i$ ] for @xmath1359 a dimensionality argument implies existence of @xmath167 normalized vectors @xmath1360 , that are functionally independent of @xmath470 , linearly independent and orthogonal to @xmath1361 , @xmath1362 , @xmath1356 , @xmath1357 and @xmath1355 ( for every @xmath1350 ) .
recall that @xmath1361 and @xmath1362 are the first two elements of the canonical basis of @xmath1363 .
hence , the first two coordinates of @xmath1364 for @xmath1365 are zero .
these orthogonality properties readily imply @xmath1366 and @xmath1367 for @xmath1368 . inserting zero coordinates and rows , respectively
, we shall now suitably embed @xmath1369 and @xmath1370 into @xmath143 and @xmath1334 .
define @xmath1371 as @xmath1372 and @xmath4 as @xmath1373 where @xmath1374 , a number that does not exceed @xmath18 by assumption .
we emphasize that by construction @xmath4 does _ not _ depend on @xmath470 .
furthermore , if we delete from @xmath329 and @xmath330 those coordinates that correspond to the zero coordinates that have been inserted to obtain @xmath1371 from @xmath1355 , we obtain the vectors @xmath1356 and @xmath1357 .
therefore , it follows from equation that @xmath1371 is orthogonal to @xmath1375 , that @xmath1376 and that for every @xmath1350 we have @xmath1377 as an immediate consequence we obtain @xmath1378 where we recall that all coordinates of @xmath1355 are nonzero and therefore @xmath1379 has precisely @xmath1380 nonzero columns .
we now intend to apply lemma [ auxconstr ] with @xmath1381 , acting as if @xmath1382 was the underlying design , @xmath503 was the hypothesis to be tested and with the triple @xmath108 , @xmath504 , @xmath106 which obviously satisfies assumption [ weightspwrb ] with respect to @xmath477 ( a matrix with @xmath1383 columns ) , since by assumption we have @xmath476 ( note that due to interpreting @xmath477 as the underlying design , @xmath1383 corresponds to the ` k ' in lemma [ auxconstr ] ) .
we note first that removing the first or last row of @xmath1384 does not reduce its rank , because @xmath1355 ( a vector all coordinates of which are nonzero ) is orthogonal to every column of this matrix ( cf . the argument in the beginning of the proof of lemma [ auxconstr ] ) . using @xmath1385
we hence see that assumptions ( a1)-(a3 ) in lemma [ auxconstr ] are satisfied by construction .
now consider the case @xmath319 . by definition @xmath504
is an element of @xmath196 ( acting as if @xmath477 was the underlying design matrix ) .
therefore , condition ( ckv ) is satisfied for @xmath1350 arbitrary , and @xmath1386 follows . consider the case where @xmath382 . since @xmath1387 and because of @xmath1388 it follows that @xmath1389 .
therefore , condition ( cam ) in lemma [ auxconstr ] is satisfied and therefore @xmath1386 for @xmath1350 arbitrary . it remains to consider the case where @xmath383 .
it suffices to find a @xmath1390 such that @xmath1391 is well defined ( see the proof of part 2 of lemma [ auxconstr ] ) .
the latter statement is equivalent to the denominator in the fraction appearing in the definition of @xmath1391 being nonzero , i.e. , @xmath1392 where @xmath1393_{\cdot j } [ \hat{z}_{\bar{x}}(y(\delta^*))]_{\cdot ( j-|i| ) } ' \bar{\omega } ~~~ \text { for } |i| = 0 , \hdots , n - p-1.\ ] ] by definition @xmath1394 and we recall that @xmath1395 which implies via equation that @xmath1396 equals @xmath1397 the only coordinate of this vector that depends on @xmath470 is @xmath1398 , the latter equation following from @xmath1399 and @xmath1400 . since @xmath1401 , the denominator appearing in the definition of @xmath1402 interpreted as a function of @xmath470 is now seen to be a polynomial of degree @xmath1403 in @xmath470 .
hence , there must exist a @xmath1390 such that the denominator does not vanish .
it follows that @xmath1404 .
concerning the second statement we observe that @xmath1405 implies @xmath1406 , and therefore we have @xmath1407 for @xmath256-almost every @xmath257 . by assumption
the first column of @xmath42 is zero .
therefore , for @xmath256-almost every @xmath1408 we have @xmath1409 , @xmath1410 and @xmath1411 , i.e. , for @xmath256-almost every @xmath1412 scenario ( 1 ) in theorem [ tu_3 ] applies .
as above , it suffices to construct a pair @xmath1340 and @xmath1413 ( recall that @xmath94 ) , such that @xmath1414 , where @xmath1415 and @xmath1416 . here , the matrix @xmath1417 is @xmath1418 dimensional . by assumption @xmath1419
obviously satisfies @xmath1420 . to construct the matrix @xmath1417 we can thus use the same argument as was used to construct @xmath4 in the proof of the first statement ( @xmath1421 replacing @xmath167 )
the matrix @xmath477 so obtained has ( after a permutation of its columns ) the same structure as has the matrix @xmath477 constructed in the proof of the first statement .
we can therefore use almost the same arguments to conclude that @xmath1422 for some @xmath1390 and @xmath1371 as constructed in the first part of the proof . | we analytically investigate size and power properties of a popular family of procedures for testing linear restrictions on the coefficient vector in a linear regression model with temporally dependent errors .
the tests considered are autocorrelation - corrected f - type tests based on prewhitened nonparametric covariance estimators that possibly incorporate a data - dependent bandwidth parameter , e.g. , estimators as considered in @xcite , @xcite , or @xcite . for design matrices that are generic in a measure theoretic sense
we prove that these tests either suffer from extreme size distortions or from strong power deficiencies . despite this negative result
we demonstrate that a simple adjustment procedure based on artificial regressors can often resolve this problem . * ams mathematics subject classification 2010 * : 62f03 , 62j05 , 62f35 , 62m10 , 62m15 ; * keywords * : autocorrelation robustness , hac test , fixed - b test , prewhitening , size distortion , power deficiency , artificial regressors . |
seeking a suitable thank to the organizers of this unique conference and volume , we chose to present a nascent work in which rigor and speculation are hopefully well balanced .
accordingly , it has two parts : + i ) quantum oblivion ( qo ) refers to a very simple yet hitherto unnoticed type of quantum interaction , where momentum appears not to be conserved .
the problem s resolution lies within a brief critical interval , during which more than one interaction takes place .
rapid self - cancellation , however , leaves only one interaction completed . while the paradox s resolution is novel , the interaction itself turns out to underlie several well - known quantum peculiarities .
this gives a new , realistic twist to one of qm s most uncanny hallmark , namely , the _ counterfactual_. although rigorously derived from standard qm , oblivion offers some new experimental predictions , as well as new insights into other quantum oddities .
+ ii ) quantum hesitation ( qh ) proceeds further to theorizing , as to _ how _ qo takes place .
it reformulates oblivion within time - symmetric interpretations of qm , mainly aharonov s two - state - vector formalism ( tsvf ) .
we assume that , beneath several momentary interactions , out of which only one is completed , there were several possible _ histories _ , which have left only one finalized . from the tsvf
we then take one of its most exotic features , namely , weak values unbounded by the eigenvalue spectrum .
this allows too - large / too - small / negative weak values appearing under special pre- and post - selections .
such `` unphysical '' values are assumed to evolve along both time directions , over the same spacetime trajectory , eventually making some interactions `` unhappen '' while prompting a single one to `` complete its happening , '' until all conservation laws are satisfied over the entire spacetime region .
+ naturally , the qo phenomenon and the qh hypothesis should be considered separately .
their combination , however , strives to make quantum mechanics more comprehensible and realistic , at the price of admitting that spacetime has some aspects still unaccounted for in current physical theory .
it is the quantum effects themselves , regardless of any interpretation , which are best illustrated by the cynical yiddish aphorism quoted in this section s title . in classical physics ,
grandmother s wheels , which she never had , trivially play no role in her dynamics .
a quantum grandmother , in contrast , manages somehow to employ them , even to the point of outrunning vehicles .
this is the quantum counterfactual @xcite , illustrated by the following examples : + i ) in the elitzur - vaidman experiment @xcite , a bomb is prepared and positioned such that , if struck by an appropriately prepared photon , it will explode .
surprisingly , even when no explosion happens , its mere _ potentiality _ suffices to disturb the photon s interference : a non - explosive bomb leaves it undisturbed .
+ ii ) hence , in any interaction of a particle with more than one possible absorber , each absorber s capability of absorbing the particle takes part in determining its final position ( `` collapse '' ) .
consider , e.g. , a position measurement of a photon whose wave - function spreads from the source towards a circle of remote detectors : every detector s _ non - clicking _ incrementally contribute to the photon s eventual position , just like the single detector that does click .
+ iii ) such counterfactuals underlie also quantum non - locality .
bell s theorem @xcite implies the following counterfactual : had bob chosen to measure his particle s spin along an axis other than the one he actually chose , then alice s spin measurement would give the opposite value .
+ this peculiar status of counterfactuals in qm stems from the uncertainty principle .
consider the simple setting in fig .
[ fig1 ] . a beam - splitter ( bs )
is positioned between four equidistant mirrors , left / right / up / down .
we know only that , within the device , a single photon crosses the bs back and forth .
this situation , with four possible positions and momenta of the photon , encapsulates the uncertainty @xmath0 , well - known from the standard mach - zehnder interferometer ( mzi ) . ] to see that , loosen mirror @xmath1 to enable it to measure the photon s possible interaction with it . in @xmath2 of the cases ,
the detector will remain silent .
this means , with certainty , that the photon hits only @xmath3 and then reflected back to bs , where it splits towards _ both _ @xmath4 and
@xmath5 , then reflected back , reunited by bs and returns _ only _ to @xmath3 .
we can therefore leave the detector in @xmath1 and be sure that , despite the photon s endless oscillations , it will _ never _ hit this mirror .
what has happened here ?
apparently nothing .
a possible event did not occur . yet the fact that it _
could _ occur endows it a physical say .
for if we turn also mirror @xmath4 into a detector , then again in @xmath2 cases it will not click , but now , sooner or later , @xmath1 is bound to click !
the second non - event gave rise to a real event , which until that moment , was impossible .
these four possible positions and momenta are typical quantum counterfactuals . as such , they present an acute duality : + i ) as long as they are not verified , they take an essential causal part in the system s dynamics .
+ ii ) once , however , one of them is verified , all others vanish without a trace . + how , then , can non - events play part in the quantum process ?
the challenge s acuity is reflected in the various radical moves it has elicited from the reveal interpretations of qm , of which two famous schools have desperately resort to the extreme opposites : + _ abandonment of ontology : copenhagen_. since counterfactuals are facts of our knowledge , just like actual facts , let us define qm as dealing only with knowledge , information , etc .
, rather than with objective reality .
+ _ excess ontology : many worlds_. the counterfactual does occur , but in a different world , split from ours at the instant of measurement .
+ in what follows we offer an explanation to quantum counterfactuals which is purely physical , derived from quantum theory alone : at the sub - quantum level , grandmother nature utilizes her proverbial wheels by taking advantage of an otherwise - unhappy consequence of her age : _ amnesia_.
we begin with a very simple yet surprising quantum interaction where one particle emerges from it visibly affected , while the other seems to `` remember '' nothing about it .
let an electron and a positron , with spin states @xmath6 and momenta @xmath7 , @xmath8 , enter two stern - gerlach magnets ( sgms ) ( drawn for simplicity as beam - splitters ) positioned at @xmath9 and @xmath10 respectively ( see fig .
[ fig2 ] ) .
the sgms split the particles paths according to their spins in the @xmath11-direction : @xmath12 and @xmath13 let technical care be taken such that , if the particles turn out to reside in the intersecting paths , they would meet , at @xmath14 or @xmath15 , ending up in annihilation .
two nearby detectors @xmath16 , @xmath17 are set to measure the photons emitted upon pair annihilation , thereupon they would change their states to @xmath18 or @xmath19 .
let us follow the time evolution of these particles .
initially , at @xmath20 , the total wave - function is the separable state : latexmath:[\[\label{initial } depending on the two particles positions at times @xmath14 or @xmath15 , they may ( not ) annihilate and consequently ( not ) release photons , which would in turn ( not ) trigger one of the detectors . at @xmath22 , then , the superposition is still unchanged , as in [ initial ] .
but at @xmath23 , either photons are emitted , indicating that the system ended up in @xmath24 _ or not _ , and then @xmath25|ready_1\rangle|ready_2\rangle,\ ] ] which is entangled : one component of it is a definite state , while the other is a superposition in itself . ] similarly at @xmath26 : if a photon pair is emitted , we know that the particles ended up in paths @xmath27 and @xmath28 , i.e. @xmath29 , otherwise we find the non - entangled state @xmath30 which is peculiar . the positron is observably affected : if we time - reverse its splitting , it may fail to return to its source .
its momentum has thus changed . not so with the electron : it remains superposed , hence its time - reversibility remains intact ( fig .
[ fig3 ] ) .
+ summarizing , one party of the interaction `` remembers '' it through momentum change , while the other remains `` oblivious , '' apparently violating momentum conservation .
this is quantum oblivion ( qo ) . ]
it is obviously the intermediate time - interval @xmath31 that conceals the momentum conservation in qo .
the details , however , are no less interesting . the two particles , during this interval , become entangled in their positions and momenta .
suppose , e.g. that we reunite the two halves of each particle s wave - function through the original bs , to see whether they return to their source .
either one of the particles may fail to do that , on which case the other must remain intact . similarly for their positions .
this is entanglement , identical to that of the electron - positron pair in hardy s experiment @xcite , @xmath32 .
it is during this interval that the positions and momenta of the two particles become , in contrast with ordinary quantum measurement , first entangled and then non - entangled ( for greater detail of this effect and its implications see @xcite ) .
it remains to be shown that something similar occurs also to the two macroscopic detectors that finalize a experiment .
this is done next . rather than a curious effect of a specific interaction , oblivion
is present in every routine quantum measurement .
its elucidation can therefore shed new light on the nature of measurement , further enabling some novel varieties thereof .
+ ordinary quantum measurement requires a basic preparation often considered trivial .
consider e.g. , a particle undergoing simple detection ( as routinely employed during spin measurements @xcite ) .
the detector s pointer , positioned at a specific location , reveals the particle s arrival by receiving momentum from it .
this , by definition , requires the pointer to have considerable momentum certainty ( usually @xmath33 ) . in return , however , the pointer must have _
position uncertainty_. let this tradeoff be illustrated with a slight modification of our first experiment . in the original version ( fig .
[ fig2 ] ) , the two possible interactions were annihilations , which were mutually exclusive .
for the present purpose , however , let us replace annihilation by mere ( elastic ) collision ( fig .
[ fig4 ] ) . in other words , two superposed atoms @xmath34 and @xmath35
interact like the electron and positron in fig .
[ fig2 ] , but instead of annihilating , they just collide .
this can now happen on _ both _ possible occasions at @xmath14 and @xmath15 , namely at the two locations where @xmath34 can reside . with annihilations thus dropped
, the outcome is even more interesting .
suppose that the detector on path @xmath36 remains silent ( fig .
[ fig4]b ) : we are now certain that the two atoms have collided , but remain oblivious about this collision s location . what we thus measure is an _ ordinary momentum exchange _ : both atoms momenta have been reversed along the horizontal axis . here ,
oblivion is small , affecting only the two atoms positions at the time of the collision . yet , because @xmath35 has vanished from the distant location on @xmath36 , the final outcome is a coarse position measurement of @xmath35 . s position .
[ fig4 ] ] the situation is much more surprising when we _
fail _ to detect @xmath34 and @xmath35 in the paths to which they would be diverted in case of collision , namely @xmath37 , @xmath38 , @xmath39 , @xmath40 .
we are now certain that @xmath35 s superposition is reduced to path @xmath36 while @xmath34 has returned to its initial superposition over @xmath27 and @xmath41 ( fig . [
in other words , @xmath35 undergoes momentum oblivion , which is again an ordinary position measurement of @xmath34 , but this time the measurement is interaction - free , offering a special opportunity to observe how a counterfactual takes an integral part in the quantum evolution . to summarize : we have studied an asymmetric interaction between two atoms , where two halves of @xmath34 s wave - function interact with one half of @xmath35 .
two momentum exchanges between the atoms can occur : either + i ) @xmath35 turns out to have collided with @xmath34 .
this amounts to @xmath35 undergoing position measurement .
the price exacted by the uncertainty principle is a minor position oblivion of @xmath34 and @xmath35 .
+ or + ii ) @xmath35 turns out to have not collided with @xmath34 .
this again amounts to @xmath35 undergoing position measurement- collapsing it to the remote @xmath36 path .
hence its momentum ( interference ) is visibly disturbed .
this time , however , if @xmath34 serves as a detector s pointer , its oblivion is amplified to a macroscopic scale , making the measurement interaction - free .
+ as both ( i ) and ( ii ) occur under unusually high space- and time - resolution , they enable a novel study of the critical interval . during this interval ,
entanglement between the two atoms has ensued , as they have assumed new possible locations @xmath42,\ ] ] which have remained undistinguished until the macroscopic detectors indicated that no annihilation has occurred .
this has broken the symmetry between the two wave - functions , finalizing the interaction and sealing the oblivion .
the generalization is therefore natural : during every quantum measurement , the detector s pointer interacts with the particle in the same asymmetric manner as atoms @xmath43 and @xmath44 above : part of the particle s wave - function with the whole wave - function of the pointer . to make the analogy complete , recall that in reality the pointer s superposition is continuous rather than discrete .
as the pointer thus resides over a wide array of locations , momentum measurement becomes much more precise .
not only is quantum oblivion essential for every quantum measurement , it is also present beneath several well - known variants thereof . in @xcite
we have demonstrated qo s underlying ifm @xcite , hardy s paradox @xcite , weak measurements @xcite , partial measurement @xcite , the aharonov - bohm @xcite and the quantum zeno @xcite effects .
this passage from discrete to continuous superposition also opens the door for several interesting interventions , studied in @xcite .
let us summarize the oblivion effect with the commonest and most basic example .
every time a detector s pointer is set to measure the momentum of a particle that hits it , the following happens .
+ i ) the detector s pointer , prepared with ( almost ) precise zero momentum , is consequently superposed in space .
+ ii ) therefore , not only one particle - pointer interaction takes place ; rather , several interactions occur , one after another , in all the pointer s superposed locations , as the particle proceeds along them during the critical interval .
+ iii ) in all these interactions , the photon and the pointer momentarily `` measure '' each other s state , + iv ) and each of these mutual `` measurements '' mixes position and momentum , hence being very inaccurate .
+ v ) yet none of them is amplified to a full measurement .
a complex superposition of correlated states thus builds up during the critical interval .
+ vi ) only then does the pointer s state undergo amplification into a full measurement , as follows : all its possible positions fall prey to oblivion , while all possible momenta add up to a precise value .
+ vii ) all the above equally holds for ifm , where the momentum measured is zero .
+ our counterfactuals , then , become demystified . during the ci ,
grandmother s wheels do appear and vanish time and again beneath the quantum noise . in the present case , these wheels are the pointer s possible locations .
it is only because we choose to measure the pointer s momentum that we give up the options to extract these positions , that have now turned into counterfactuals .
otherwise , as shown in the previous section , had we chosen to measure them instead of the momentum , other quantum surprises would emerge .
like two tunnels dug underneath towards one another until merging into one , the above line of investigation turns out to complement another research program , much older and renowned for its surprising predictions , and also verified time and again in laboratories . in what follows
we introduce the two - state vector formalism and its offshoot , weak measurement , before proceeding to describe how the two tunnels meet .
tsvf originates from the work of aharonov , bergman and lebowitz @xcite .
it asserts that every quantum system is determined by two wave - functions : one ( also known as the pre - selected wave - function ) evolves forward in time while the other ( post - selected ) evolves backward .
the forward- and backward - evolving wave - functions , @xmath45 and @xmath46 respectively , define the so called two - state vector @xmath47 .
the two wave - functions are equally important for describing the present of the quantum system via the weak value of any operator @xmath48 defined by : @xmath49 here a logical catch ensues : _ `` the state between two measurements '' can not be revealed by measurement .
_ weak measurement @xcite was conceived in order to bypass this obstacle as well as to test other tsvf s predictions .
this led to numerous intriguing works , both theoretical and experimental .
apart from its time - symmetry , tsvf reveals even subtler symmetries with respect to values usually considered inherently positive .
consider the surprising values for mass and momentum in the following two experiments + i ) the three boxes paradox @xcite is a by - now familiar surprise yielded by tsvf , of which the underlying logic can serve as an introduction to the idea of qh .
a particle is prepared with equal probability to reside in one out of three boxes : @xmath50 later it is post - selected in the state @xmath51 what is the particle s state between these two strong measurements ? by definition , _ projective measurement is unsuitable for answering this question _ , as it reveals the state upon the intermediate measurement .
it is _ weak _ measurement , again , that comes to help .
tsvf predicts the following weak values of the projection operators , @xmath52 for @xmath53 : @xmath54 therefore , the total number of particles is @xmath43 , as it should be , but it is a sum of two ordinary particles plus one odd .
the last equation denotes negative weak value for the very exitance in the third box . in order to fully grasp the paradoxical nature of this term ,
let us consider within the context of its standard versions : + if `` probability @xmath43 '' means `` the particle certainly resides within this box '' ; + and `` probability @xmath33 '' means `` the particle has never resided within this box '' ; + then `` probability @xmath55 '' means `` the particle certainly _
un_resides within this box . ''
+ as absurd as third expression may sound , this is its simplest non - mathematical meaning , the alternative being dismissing it as meaningless .
the choice in @xcite to trust the mathematics , has led to assigning a negative sign to every interaction involving the third box , as long as it is weak enough @xcite .
obviously , this can not be related to the particle s charge , as it played no role from the beginning .
the remaining choice is mass .
the simplest way to prove this prediction is through the particle s momentum : a collision with another particle must give the latter a `` pull '' rather than a `` push , '' even though their initial velocities were opposite . in passing ,
it is worth comparing this step of the tsvf to dirac s choice to trust the mathematics upon encountering the negative value for the electron s charge , following the dual solution for his famous equation . that choice has later led to the discovery of the positron
this may be the case with aharonov s present choice as well .
can this prediction be put to test ?
a preliminary version has been carried out by steinberg s group @xcite + ii ) the importance of negative weak values becomes highly visible through hardy s experiment @xcite .
two mzis overlap in one corner ( see fig . [ fig5 ] ) .
they are tuned such that electron entering the first mzi will always arrive at detector @xmath56 , while a positron entering the second mzi will always arrive at detector @xmath57 . therefore ,
when the electron and positron simultaneously traverse the setup , they might annihilate at the intersection or make their partner reach the `` forbidden '' detector @xmath58 for the electron / positrion respectively . in case no annihilation was detected we can exclude the case they booth took the overlapping path @xmath59 and their state becomes @xmath60 _ i.e. _ , at least one of the particles took the non - overlapping ( @xmath61 ) state .
the interferometers were tuned such that @xmath56 clicks for the electron constructive interference state @xmath62 , @xmath63 clicks for the position constructive interference state @xmath64 state , and similarly for @xmath57 and @xmath65 .
therefore , observing clicks at @xmath63 and @xmath65 is equivalent to post - selection of the state @xmath66 this post - selection is rather peculiar since a click at @xmath63 naively tells us that the positron took its overlapping path , while click at @xmath65 naively tells us that the electron took its overlapping path .
this scenario , however , is impossible , because we know annihilation did not occur .
the paradox can be resolved within the tsvf .
when we calculate the weak values of the various projection operators we find out that @xmath67 and @xmath68 while @xmath69 this leads us to conclude that although the number of pairs is @xmath43 , we have two `` positive '' pairs and one `` negative '' pair- a pair of particles with opposite properties . the pair in the `` no - no '' path creates a negative `` weak potential '' @xcite , that is , when weakly interacting with any other particle in the intermediate time , its effect will have a negative sign .
moreover , we can see now the cancelation of positive and negative weak values by considering projections on the non - overlapping paths : @xmath70 @xmath71 ] \iii ) we can now derive from these experiments precisely the conclusion needed for our thesis : _ the negative values that they reveal exist also in ordinary measurements , only perfectly hidden_. + consider then , a photon split by a bs , into @xmath72 and @xmath73 beams , then split the @xmath73 beam further into two @xmath72 beams ( this setup was recently suggested by vaidman to answer the question `` where have the photons been ? ''
perform weak measurements on all three thirds of the wave - function . now delay all three parts in separate boxes .
this delay enables you to choose between two options : either + a ) reunite the three beams for interference , and post - select destructive interference with negative sign of the third .
you will get the results of in the above three boxes experiment , namely , two photon and one negative - mass photon .
+ or + b ) reunite only the two split @xmath72s back into the original @xmath73 beam , then perform strong position measurements on each beam .
you will get a photon on one side and nothing on the other .
no post - selection can reveal nothing unusual in the earlier weak measurements .
what was so far a trivial `` nothing '' thus turns out to be something much more profound , namely , a perfect mutual cancellation of positive and negative masses .
similarly for the real photon detected on the other half : it turns out to be the sum of two odd weak values .
so far , our discussion went strictly within standard quantum formalism .
oblivion straightforwardly follows under a finer time - resolution of the quantum interaction . encouraged by tsvf
, we now take a step further asking : how does oblivion evolve in space and time ?
this is admittedly a dissembling question , pretending not to know that quantum measurement itself still poses unresolved issues ; that `` collapse of the wave - function '' is hotly debated , etc .
we thus enter the daunting minefield of interpretations of qm , where each choice carries its penalty . equally , however
, these difficulties may serve as positive incentives : to the extent that qo and tsvf illuminate the nature of quantum measurement , it is worth rephrasing older questions in the terms of these frameworks .
our proposal focuses on aspect ( ii ) of the quantum counterfactuals dual nature ( section 2 ) , namely , their perfect self - annihilation . recall that any possibility to observe a counterfactual after the measurement , such as retrieving information about a particle s momentum after its position was measured , entails straightforward violation of the uncertainty principle .
the challenge , therefore , is to explain not only how the counterfactual exerts such an essential role , but also how , having done that , it manages to disappear without a trace .
the first clue is already in our possession : qo has shown that counterfactuals can be observed , although indirectly beneath great quantum noise , during the critical interval ( sec .
3.2 ) .
we now return to aharonov s two state - vector formalism described in sec .
4.1 above , where causality is highly time - symmetric at the quantum level .
elsewhere @xcite we have explained our endorsement of this approach .
first , it is appealingly simple .
take for example the epr experiment : one experimenter s choice of measurement obviously affects the outcome of another measurement chosen at that moment , far away , by the other experimenter .
this appears disturbing , however , only as long as one assumes the familiar causation , going from the past emission of the particles to the later measurements .
if , in contrast , one allows causal relations to somehow go back and forth between the past event and its successors , much of the mystery dissolves .
_ what appears to be nonlocal in space , becomes perfectly local in spacetime_. similar explanations , equally elegant , await many other quantum paradoxes , from schrdinger s cat to the quantum liar paradox @xcite . to be sure , the above `` somehow '' , by which causality runs along both time directions , still needs much clarification .
the model also takes the price of admitting that there is something to spacetime that still lies beyond physics comprehension .
these issues will be discussed in consecutive works . yet
we find the advantages of time - symmetric quantum causality compelling enough to pursue them , even before they mature into a fully - fledged physical theory .
our second clue is therefore : _ quantum peculiarities can be better explained by allowing causality to be time - symmetric at the quantum level_. while time - symmetric causality has been conceived by several quantum physicists ( most notably cramer s transactional interpretation @xcite ) , surprising experimental derivation of it are solely due to the tsvf .
these are the odd weak values described in section 4.2 above . although they could be derived from standard quantum theory
, the fact is that they never crossed anybody s mind unless guided by tsvf .
paradoxically , it is these exotic features that go beyond interpretation , having won several laboratory verifications worldwide . here
then is our third clue : _ unusual quantum values that appear to be mere noise due to the pointer s uncertainty , turn out to be physically real , revealed by weak measurements
. they may be unusually large , unusually small , and negative , even in the case of the mass , which in classical physics is known to be only positive_. we have what we need . the established oblivion effect ( section 3 ) , plus
the hypothesis of time - symmetric quantum causality ( section 4.1 ) , plus the latter s more specific discovery of odd values ( section 4.2 ) , merge into the following coherent picture : oblivion is an evolution not only of states , but of histories .
`` collapse , '' whereby quantum counterfactuals first play an essential role and then succumb to oblivion , occurs by ( i ) retarded and advanced actions complementing one another along the same spacetime trajectory , and ( ii ) negative values taking place alongside with positive ones . to prove this assertion
, we have to show that its most exotic ingredient , namely the odd values , rather than being fringe phenomena , are in fact omnipresent .
as paradoxical as this may sound , the mathematical proof is fairly straightforward .
as in the case of oblivion , we now show that weak values , rather than an exotic curiosity , are part and parcel of every quantum value .
the proof relies on two ranges of continuum .
suppose that we gradually move ( i ) from weak measurements to strong ones , and/or ( ii ) from special pre- and post - selections to non - special ones .
would odd weak values vanish ?
this indeed appears to be the case , but in reality they become stronger , only counterbalanced by opposite weak values so as to give the familiar quantum values . moreover , as one of us has shown in @xcite there exists a continuous mapping between weak and projective measurement , allowing to decompose the latter into a long sequence of weak measurements .
performing a sequence of weak measurements on a single particle , is amount to a biased random walk with incremental .
finally , the measured state is randomly driven into a definite vector out of the measurement basis .
ergo , upon gradually moving from weak to strong measurements , odd values do not diminish , but rather increase .
its only through their addition with other weak values that they add up or cancel out , either completing or canceling each other to give the familiar quantum values , including @xmath33 .
the above idea seem to be independently emerging in the very current literature on weak values . the main argument of this work nothing short of astonishing , but perfectly converges with ours : weak values take part in the formulation of the most salient features of the quantum realm . +
i ) contextuality : when you measure a particle s spin along a certain direction and get the value `` up , '' you can be sure that , had the sgm been oriented to the opposite direction , the results would have been `` down . ''
similarly for other directions : the outcome should deviate from your `` up '' by the angle difference between the directions .
this is common logic .
quantum contextuality , however , poses an odd restriction on these counterfactual outcomes : they have no objective existence , because they depend on the very orientation you chose first .
in other words , there can not be some objective `` up / down '' along a certain spin direction which has existed prior to your choice , such that the angle you choose would be related to it .
on the contrary , it is the direction that you chose for whatever reason that serves as the basis for all other possible choices . although contextuality and non - locality of quantum phenomena are dominating strong values , the work of pusey @xcite , as well as our previous works @xcite have recently shown respectively that interfering weak values can well account for the contextuality and non - locality of qm .
+ ii ) interference : classical waves interfere in strict accordance with everyday intuition .
single - particle interference , in contrast , is paradoxical in that the location of a single particle after passing the interference device is strictly determined by counterfactual paths . here too , mori and tsutumi @xcite have recently argued that weak values take part in the formation of this quantum phenomenon .
their `` weak trajectories , '' which the single particle is supposed not to have traversed , are shown on one hand to be detectable by weak measurement and on the other hand to add up to give the wave - function s familiar undulatory motion .
this article recounts a search for a better understanding of the unique causal efficacy of quantum events that failed to occur .
we have revisited two opposite characteristics unique to quantum physical values . in terms of information theory ,
classical values constitute , by definition , signals .
quantum values , in contrast , are sometimes equivalent to lack of signals , namely , silence , and sometimes to a surplus of signals , namely , noise . perhaps , then , instead of ignoring the former and trying to filter out the latter , it is time to take them as complementary to signals , even as their constituents ?
we have therefore studied two such quantum phenomena .
+ _ interaction - free measurement _ indicates that a detector s silence is as causally potent as its click . seeking to understand this potency , we pointed out quantum oblivion as the basic mechanism underlying it . upon further study , qo turned out to underlie every standard quantum measurement .
+ _ weak values
_ appear amid enormous quantum noise inflicting the quantum weak measurement .
tsvf , however , has extracted from them a great deal of novel information about the quantum state .
they reveal , e.g. negative weak values emerging in many quantum states .
recently , it has been argued that these weak values take part also in more common quantum phenomena , such as contextuality and interference .
+ our synthesis of these two lines of research has led to the model of quantum hesitation : _ the weak values , taking part in the time - symmetric evolution of the quantum state , are responsible for the very phenomenon of quantum measurement which we take as real_. following the tsvf , we suggest that the measurement s outcome , randomly coming out of several possible ones , is the sum of the two state vectors going back and forth in time .
this outcome may therefore emerge from weak values , even odd ones , brought by these two state vectors .
similarly for the even more intriguing disappearance of the other possible outcomes : it may be negative weak values , contributed by one state vector , which precisely cancel the positive weak value of the other state vector .
hence , perhaps , the apparently innocuous `` @xmath33 , '' which nevertheless endows quantum grandmother such intriguing mobility .
+ can this account boast greater rigor than just one more interpretation of qm ?
in conclusion we want to point out what we consider to be the greatest difficulty hindering such an advance .
the above account invokes two distinct causal chains contacting between past and future events , _ traversing the same spacetime trajectory_. such a far - reaching idea may invoke more new questions than answers for the old ones .
the most acute problem concerns the nature of time : is there more to physical time then the geometrical characteristics assigned to it by special and general relativity ? or does it allow some yet unknown `` becoming '' @xcite that involves a privileged `` now '' moving from past to future ? + the two major time - symmetric interpretations of quantum mechanics , namely tsvf and the transactional interpretation , openly admit that these issues are sorely obscure .
recently both schools produced novel accounts that deal with time s nature in radical ways @xcite .
as our own contribution to these issues has been only to stress their acuity @xcite , we can only hope for this kind of theorizing to be even more radical in the future .
vaidman l 2009 counterfactuals in quantum mechanics , in compendium of quantum physics : concepts , experiments , history and philosophy greenberger d , hentschel k and weinert f ( eds . ) ( berlin : springer - verlag ) p 132 ( _ preprint _
arxiv:0709.0340 ) aharonov y and cohen e 2015 weak values and quantum nonlocality , to be published in `` quantum nonlocality and reality '' , mary bell and gao shan ( eds . ) , cup , ( _ preprint _ http://www.ijqf.org/wps/wp-content/uploads/2015/01/aharonovcohen.pdf ) | among the ( in)famous differences between classical and quantum mechanics , quantum counterfactuals seem to be the most intriguing . at the same time , they seem to underlie many quantum oddities . in this article
, we propose a simple explanation for counterfactuals , on two levels .
quantum oblivion ( qo ) is a fundamental type of quantum interaction that we prove to be the origin of quantum counterfactuals .
it also turns out to underlie several well - known quantum effects .
this phenomenon is discussed in the first part of the article , yielding some novel predictions . in the second part
, a hypothesis is offered regarding the unique spacetime evolution underlying qo , termed quantum hesitation ( qh ) .
the hypothesis invokes advanced actions and interfering weak values , as derived first by the two - state - vector formalism ( tsvf ) . here too , weak values are argued to underlie the familiar `` strong '' quantum values . with these ,
an event that appears to have never occurred can exert causal effects and then succumb to qo by another time - evolution involving negative weak values that eliminate them .
we conclude with briefly discussing the implications of these ideas on the nature of time . |
most experts do not believe that effective optical depths of @xmath11 measured in the @xmath12 hi ly@xmath1 forest imply that hi reionization was occurring ( e.g. @xcite ) .
observations of the @xmath13 heii ly@xmath1 forest also find @xmath11 in select regions , and again the view of experts is that these observations do not necessarily indicate that heii reionization is happening ( e.g. @xcite ) .
however , there are several key differences that make heii ly@xmath1 absorption at @xmath14 a more sensitive probe of diffuse heii than hi ly@xmath1 absorption at @xmath10 is of diffuse hi : 1 ) the abundance of helium is @xmath15 times smaller than that of hydrogen , 2 ) a photon redshifts through the heii resonance @xmath16 times faster than the hi resonance , 3 ) the intergalactic medium ( igm ) is less dense and has more voids at @xmath14 than @xmath12 , and 4 ) the hi ly@xmath1 forest absorption is available for each heii sightline and reveals the sightline s density structure .
previous theoretical studies of the heii ly@xmath1 forest modeled this absorption as a set of discrete absorbing clouds and/or focused on the @xmath17 statistic @xcite .
this letter employs better - suited techniques to study heii ly@xmath1 absorption near saturation .
the discrete - cloud approximation works best in overdense regions .
the low - density igm is better approximated as a less clumpy continuum of gas that has not decoupled from the hubble flow , and it is the absorption in the least dense regions that are the most constraining with regard to heii reionization . furthermore , the local measure for the heii fraction studied here is more powerful than global measures like @xmath18 , for which the effect of heii reionization is highly model - dependent .
our measure uses hi absorption to target the heii absorption in the most evacuated voids .
the advantages of using the hi absorption were realized in many observational studies of the heii forest @xcite , and we argue here that this information can be used to obtain tighter constraints on the amount of intergalactic heii than previous analyses have found .
this study is timely because the hubble space telescope ( hst ) reservicing mission installed the cosmic origins spectrograph ( cos ) in may 2009 .
cos will enable absorption measurements at @xmath19 for heii ly@xmath1 and at @xmath20 for heii ly@xmath3 , where @xmath21 is the redshift of the quasar .
it is able to achieve higher signal - to - noise ratios and higher spectral resolutions than previous instruments on @xmath22 existing heii ly@xmath1 sightlines , in addition to providing @xmath23 new sightlines .
many candidate sightlines have been identified as promising for cos spectroscopy @xcite . in this letter ,
we assume a flat @xmath24cdm cosmology with @xmath25 , @xmath26 , @xmath27 , @xmath28 , and @xmath29 , consistent with recent measurements @xcite .
in contrast to hi ly@xmath1 gunn - peterson absorption at @xmath30 , which saturates for neutral fractions of @xmath31 , the gunn - peterson optical depths for a photon to redshift through the heii ly@xmath1 , ly@xmath3 and ly@xmath32 resonances are @xmath33 where @xmath34 is the gas density in units of the cosmic mean and @xmath35 is the local fraction of helium in heii .
lower redshift heii ly@xmath1 absorption obscures the heii ly@xmath3 absorption ( and both ly@xmath1 and ly@xmath3 obscure the ly@xmath32 ) .
if the foreground ly@xmath1 absorption is from @xmath36 ( where @xmath37 ) , this obscuration should not significantly weaken bounds that ignore it .
the focus of this letter is on absorption in underdense regions and what this absorption implies about the ionization state of denser , neighboring regions .
approximately 13% of the volume has @xmath38 at @xmath39 , 5% has @xmath40 @xcite , and a larger fraction in redshift space .
equation ( [ eqn : gp_heii ] ) , combined with the knowledge that the least dense regions in the @xmath41 igm have @xmath42 , suggest that the heii ly@xmath1 forest is sensitive to heii fractions in such underdensities of the order @xmath43 ( and higher series lines to @xmath44 )
. will be an overestimate ( by we estimate from cosmological simulations @xmath45 for @xmath46 ) .
accounting for peculiar velocities does not change our constraints on @xmath35 because peculiar velocities have the same effect on both @xmath47 and @xmath48 . ] in order to achieve such constraints , the locations of underdense regions must be known .
the hi ly@xmath1 forest reveals this information .
the hi ly@xmath1 gunn - peterson optical depth is @xmath49 where we have assumed photoionization equilibrium with @xmath50 the hi photoionization rate in units of @xmath51 s@xmath52 , and a power - law temperature - density relation given by @xmath53 where @xmath54 @xcite . the term @xmath55 in equation [ eqn : tauhi ] arises from the temperature dependence of the recombination rate .
we use the values @xmath56 , @xmath57k , and @xmath58 , consistent with observations @xcite .
@xmath50 is expected to be nearly spatially invariant owing to the long mean free path ( mfp ) of hydrogen - ionizing photons and the large number of sources within a mfp .
the smaller the value of @xmath34 that can be reliably located in the hi ly@xmath1 forest , the better one can constrain the heii fraction from the heii lyman forest . however , continuum fitting tends to artificially remove flux such that the flux in the lowest density pixels is set to zero , thereby preventing one from distinguishing between small values of @xmath34 .
@xcite estimated that continuum fitting on average removes @xmath59% of the transmission at @xmath60 .
therefore , a value of @xmath61 , which yields @xmath62 or @xmath63 absorption , is approximately the minimum density contrast over @xmath64 that can be discriminated at @xmath41 .
while a more rigorous derivation of the minimum @xmath34 would be worthwhile , we employ @xmath61 for this study .
figure [ fig : spectra ] shows the hi ly@xmath1 spectrum ( dotted curve , @xmath65 ) and heii ly@xmath1 spectrum ( thick solid curve , @xmath66 ) of the @xmath67 quasar q0302 - 003 @xcite , one of the most - studied heii ly@xmath1 sightlines .
note that for @xmath68 a region spanning @xmath69 comoving mpc ( cmpc )
@xmath70 ( except perhaps near @xmath71 ) such that there is no significant heii ly@xmath1 transmission .
the short - dashed horizontal line in figure [ fig : spectra ] represents the transmission for @xmath72 .
many of the elements with @xmath70 in heii ly@xmath1 absorption admit @xmath73 hi ly@xmath1 transmission ( see the cyan highlighted regions in figure [ fig : spectra ] ) .
these regions should have @xmath46 from the above discussion .
equation ( [ eqn : gp_heii ] ) and the limit @xmath70 implies that the underdense elements with @xmath40 have @xmath74 .
keck hires hi ly@xmath1 forest spectrum of q0302 - 003 . since underdense regions tend to have several hundred km s@xmath52 widths , their character does not change with higher resolution data . ] throughout this letter , we adopt this bound on @xmath35 for the most underdense , saturated elements in the heii ly@xmath1 spectrum of q0302 - 003 , although a more detailed analysis could improve upon it .
another well - studied heii forest sightline , the @xmath75 quasar he2347 - 4342 ( not shown ) , has a gunn - peterson trough at @xmath76 with no detected heii ly@xmath1 transmission and provides a similar bound .
sometime before @xmath77 , the second electron of intergalactic helium was reionized by a source of ultraviolet photons ( most probably quasars ) , and afterward the helium was kept doubly ionized by the metagalactic radiation background . if a gas parcel is exposed to a radiation background with heii photoionization rate @xmath78 , the heii fraction as a function of elapsed time @xmath79 is @xmath80 where @xmath81 is the heii fraction at @xmath82 , @xmath83 , @xmath84 @xmath85 is the electron density , and @xmath86 is the recombination coefficient
. appears implicitly in @xmath87 and @xmath1 is the case a recombination coefficient . ]
the @xmath41 recombination time @xmath88^{-1}$ ] is roughly half the hubble time @xmath89 at mean density .
figure [ fig : heiifrac ] plots @xmath90 as a function of @xmath78 for @xmath91 and @xmath92 .
we have shown that a measurement with percent - level precision of the heii ly@xmath1 forest can be used to set a percent - level constraint on the heii fraction in underdense regions that are opaque to heii ly@xmath1 photons . in this section
, we argue that such a measurement can place an even stronger limit on the local heii fraction in a @xmath6cmpc region surrounding such underdensities .
our argument requires two propositions to hold : 1 . the ionization state of gas elements in a volume that is exposed to the same @xmath93 scales as @xmath94 , the scaling expected for photoionization equilibrium .
factors that appear in the denominator of equation [ eqn : equil ] for notational convenience ( see footnote 5 ) .
our calculations include this additional scaling , which increases in importance with density . ] 2 . a @xmath6cmpc region surrounding an underdensity in the forest
is exposed to approximately the same @xmath93 as the underdensity .
from these two propositions , one can place an interesting lower bound on @xmath35 in @xmath6cmpc regions . to see this ,
let us take our constraint of @xmath95 at @xmath96 ( section [ sec1 ] ) .
using the two propositions and equation ( 4 ) to infer the ionization state of neighboring gas elements ( which provides the scaling @xmath94 ) , we find that nearby elements with @xmath97 should have @xmath98 .
see the intersection of the solid vertical line in figure [ fig : heiifrac ] with the thick solid curve .
in addition , the intersection of this vertical line with the two dashed curves in figure [ fig : heiifrac ] represents the lower limit on the volume- and mass - weighted heii fractions in the region surrounding the opaque voids with @xmath95 at @xmath91 , or @xmath99 and @xmath100 ( excluding @xmath101 in these averages ) .
this calculation of the volume- and mass - weighted heii fractions uses the gas density probability distribution function ( pdf ) measured from simulations in @xcite , with more recent simulations with an updated cosmology in the range @xmath102 @xcite .
] and it uses the fact that a @xmath103cmpc region is fairly representative of the igm at @xmath39 . averaged over a sphere of radius @xmath104 ( @xmath105 ) cmpc is @xmath106 ( @xmath107 ) at @xmath60 . ]
it is clear that these constraints can be improved with a more careful analysis and better data .
we expect that the constraint @xmath108 in a @xmath23 cmpc region is possible by utilizing smaller @xmath34 ( probably in a statistical manner ) or by using higher lyman - series resonances .
for example , the measurement of @xmath109 at @xmath91 in ly@xmath3 absorption would constrain @xmath108 ( dashed vertical line in figure 2 ) . +
we have shown that if two propositions hold , the heii lyman forest can place strong constraints on @xmath35 in a region .
next , we argue that the two propositions are generally satisfied .
we define heii reionization as being over in a region when @xmath110 and @xmath111 is much greater than @xmath112 , such that the region never significantly recombines .
the scaling @xmath94 will apply a few equilibrium times after heii reionization ends in such a region even if photoionization equilibrium does not to a precision of @xmath113 ( equation [ eqn : xheii ] ) , where @xmath114 represents a time average from the time heii reionization ended , @xmath79 .
furthermore , once this scaling is achieved _
after _ heii reionization , this scaling is maintained even if photoionization equilibrium is not . during or a few @xmath115 after heii reionization this scaling
may not be achieved , but it is still difficult to significantly evade the constraint on @xmath35 that assumes this scaling . to see this ,
let us consider a region with uniform @xmath93 that has an initial heii fraction of @xmath116 and is being reionized ( @xmath35 is decreasing ) . in this case , assuming photoionization equilibrium and using underdense regions to infer the largest allowed photoionization rate @xmath117 could result in an overestimate of the heii fraction if the incident @xmath118 .
this effect is illustrated by the long- and short - dashed curves in figure [ fig : xd ] , which are not equilibrium curves but which represent @xmath59 and @xmath105 times larger @xmath78 than the corresponding solid curve ( which assumes equilibrium ) .
however , it is unlikely that a region will have @xmath119 because the timescale to reach the new equilibrium state is @xmath120 , which is typically much less than @xmath89 .
for example , the constraint @xmath121 at @xmath97 and @xmath39 would imply @xmath122 .
figure [ fig : allowed ] illustrates this argument for a toy example where @xmath78 is taken as constant after heii reionization begins .
plotted are contours of fixed heii fraction in the @xmath78 versus @xmath123 plane , where @xmath123 is the time since the beginning of heii reionization in the region .
the top thick solid curve represents our constraint @xmath124 at @xmath125 . only the colored area left of this curve
is allowed .
the dashed curves represent @xmath126 , and @xmath127 .
this illustrates that in order for @xmath128 to be significantly less than the limit we set assuming photoionization equilibrium ( or @xmath129 ; section 4 ) , the heii needs to have been recently reionized .
for example , to evade this limit such that @xmath130 , the heii in the region must have been reionized over the last @xmath9 myr ( @xmath131 ) .
this proposition holds because the mfp of a heii - ionizing photon , which sets the correlation length for @xmath78 , should be @xmath132cmpc .
the mfp for a heii lyman - limit photon to be absorbed in diffuse gas at @xmath39 is @xmath133cmpc ( assuming a homogeneous igm ) , which is @xmath6cmpc for the value of @xmath128 that saturates the previous bounds we have quoted , @xmath134 .
the spectrally - averaged mfp , weighting by contribution to @xmath78 , is @xmath59 and @xmath135 times larger than the lyman - limit mfp for photon spectral indexes of @xmath136 ( expected in the case of no absorption ) and @xmath137 , respectively .
in addition , estimates for the mfp of heii - ionizing photons to be absorbed in dense systems la @xcite are similar at the @xmath78 that yield @xmath138 ( appendix a in @xcite plus followup work ) . in agreement with this proposition ,
the heii reionization simulations of @xcite find @xmath78-fluctuations on @xmath139cmpc scales ( even in heii regions ; see their figure 5 ) .
the previous paragraph argued that @xmath78 fluctuates as @xmath6cmpc when our bound on @xmath140 is saturated .
however , @xmath140 could be larger than our lower bound immediately around the underdensity such that the fluctuation scale is @xmath141cmpc .
one could then design a situation in which the average @xmath142 in a @xmath143cmpc region around an underdensity is less than our bound .
however , the situation would have to be very contrived in order to significantly evade our bound , especially since current spectra show saturated voids throughout their gunn - peterson troughs . in apparent contradiction to this proposition
, @xcite detected large fluctuations in @xmath78 on @xmath144cmpc scales in the spectra of he2347 - 4342 , with the least dense pixels yielding smaller @xmath78 .
however , @xcite replicated the @xcite analysis on mock data with the same noise properties , and they showed that the observed fluctuations could be explained by poisson fluctuations in the distribution of quasars and by a @xmath145cmpc mfp .
it appears the reason that a @xmath146cmpc correlation length is inferred in @xcite owes to noise in the heii forest spectrum , which often results in a large overestimate for @xmath78 in pixels with @xmath147 .
in fact , @xcite demonstrated the presence of this bias and , when it was corrected for , estimated a fluctuation scale for @xmath78 of @xmath148cmpc from he2347 - 4342 .
related to proposition 2 is the misconception that @xmath78 should correlate strongly with the density of an absorber .
the rays from a quasar will traverse many ( essentially uncorrelated ) absorption systems prior to traveling @xmath149cmpc to a typical absorber , making it implausible that the typical overdense element will be exposed to a larger @xmath78 from these sources than an underdense one .
this argument ignores that overdense absorbers are more likely to exist closer to a quasar where there is more radiation .
however , the correlation between the locations of @xmath150 quasars ( @xmath151 per @xmath152 cmpc@xmath153 ) and the density of a typical absorber is negligible @xcite .
even if @xmath150 quasars were not the primary source for @xmath78 [ contrary to what the quasar luminosity function and the amplitude of @xmath78-fluctuations in the heii forest suggest @xcite ] , it would be difficult to make @xmath78 correlate strongly with the density of absorbers : as the number density of the sources increases , the fluctuations in @xmath78 should decrease .
[ conclusions ] this letter showed that current heii ly@xmath1 forest data imply that @xmath154 in regions surrounding opaque voids , while @xmath155 should be possible with future heii lyman forest data .
these limits derive from assuming photoionization equilibrium and that the maximum allowed @xmath78 in an underdense gas element is also the maximum in a surrounding @xmath6cmpc region the expected fluctuation scale for @xmath78 .
they may be weakened somewhat , but only if the region in question is being reionized . however , since heii reionization is the epoch when a non - negligible heii fraction exists in the low - density igm , by definition it would be occurring if @xmath155 .
therefore , the heii forest may be able to address whether heii reionization is occurring at @xmath13 .
our constraint of @xmath156 is a significant improvement over previous efforts , which set the bound @xmath157 in the regions of highest opacity @xcite .
our constraint on @xmath128 in select regions is @xmath9 times stronger than the tightest constraint on the volume - averaged hi fraction from the @xmath10 hi ly@xmath1@xmath32 forests of @xmath158 @xcite .
it can be improved by better observations , by including redshift - space effects , by measuring @xmath35 in pixels with @xmath159 , or by utilizing absorption information from higher lyman - series resonances .
+ we would especially like to thank gabor worseck for providing the stis and fors2 spectra and michael rauch for the hires . we also thank jamie bolton , claude - andr faucher - gigure , adam lidz , and gabor worseck for useful discussions . | many experts believe that the @xmath0 heii ly@xmath1 forest will suffer from the same saturation issues as the @xmath2 hi ly@xmath1 forest and , therefore , will not be a sensitive probe of heii reionization .
however , there are several factors that make heii ly@xmath1 absorption more sensitive than hi ly@xmath1 .
we show that observations of heii ly@xmath1 and ly@xmath3 gunn - peterson troughs can provide a relatively model - independent constraint on the volume - averaged heii fraction of @xmath4 .
this bound derives from first using the most underdense regions in the heii forest to constrain the local heii fraction and , then , assuming photoionization equilibrium with the maximum allowed photoionization rate to calculate the ionization state of nearby gas .
it is possible to evade this constraint by a factor of @xmath5 , but only if the heii were reionized recently .
we argue that heii ly@xmath1 gunn - peterson troughs observed in the spectra of q0302 - 003 and he2347 - 4342 signify the presence of @xmath6 comoving mpc patches in which @xmath7 .
this is a factor of @xmath8 improvement over previous constraints from these spectra and @xmath9 times stronger than the tightest constraint on the hi volume - filling fraction from the @xmath10 hi lyman forest . |
follow - up radial velocity ( rv ) observations of sun - like stars with planets have revealed a number of extrasolar multi - planet systems .
many of them are involved in low - order mean motion resonances ( mmrs ) . in particular , at least four extrasolar systems are involved in a strong first order 2:1 mmr : gliese 876 @xcite , hd 82943 @xcite , hd 128311 @xcite , and hd 73526 @xcite .
a considerable effort has been devoted to study the origin @xcite and dynamical stability of these intriguing systems ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
yet dynamical studies of the resonant configurations often rely on the 2-keplerian coplanar fits by the discovery teams .
it has been demonstrated that the 2-keplerian models can be of a very limited use for systems involved in strong mutual interactions ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
even if a model incorporates mutual interactions , due to typically short time - span and a limited number of the observations , the orbital inclinations are barely constrained and usually only coplanar , edge - on configurations are considered . however , the recent results of @xcite and @xcite suggest that a significant fraction of planetary systems involving giant planets may be substantially non - coplanar .
dynamical mechanisms which lead to fast amplification of the relative inclination are especially effective in the first order resonance configurations @xcite .
also dynamical relaxation and collisional scattering of the protoplanets may favor large relative inclinations in such systems , even if they initially emerge in a flat protoplanetary disk .
the interpretation of the rv data for multi - planet system may be difficult .
the determination of the number of planets and their orbital periods can be problematic for some systems .
in particular , for those involved in a 1:1 mmr .
a periodogram of the rv signal for such a system @xcite is basically indistinguishable from that of a single planet in an eccentric orbit or , as we show in this paper , from a periodogram of a 2:1 mmr orbital configuration . @xcite and
@xcite have demonstrated that a coplanar 1:1 mmr of jovian planets may be stable in a wide range of their orbital parameters .
the results of hydrodynamic simulations by @xcite indicate that trojan planets in tadpole or horseshoe orbits might readily form and migrate within a proto - planetary disk .
presumably , a 1:1 configuration may also emerge as a result of dynamical relaxation or migration frequently used to explain 2:1 mmr configurations . in the solar system there
exist a number of moons involved in this type of resonance famous janus - epimetheus system ( co - orbital moons of saturn , exchanging orbits ) , helene - polydeuces ( trojans of dione , a moon of saturn ) , telesto calypso ( trojan moons of tethys , yet another moon of saturn ) .
dynamically , these configurations mimic planetary systems in the 1:1 mmr . in this paper
, we perform an independent analysis of the rv data for and to verify if the observed rv variations can be explained not only by a configuration in a 2:1 mmr but also 1:1 mmr .
the inevitable problem with modeling such systems is that due to a limited number of data and relatively large measurement errors , the best - fit orbital elements often and easily lead to catastrophically unstable configurations . in order to solve this problem
one needs a method of fitting which incorporates a stability criterion .
without such a constraint , one can find a stable best - fit orbits basically by chance .
when we deal with a 1:1 mmr configuration , an appropriate stability control is essential as the planets share the same ( or a very similar ) orbit and a multi - parameter dynamical model is highly nonlinear . in this paper
we use the term `` trojan planets '' not only for tadpole , close to coplanar and circular configurations but for all configurations characterized by a 1:1 mmr so having not only similar semi - major axes but also ( possibly ) large relative inclinations and variable eccentricities .
due to strong mutual interactions , the planetary systems with giant planets have to avoid the unstable zones of the mmrs , a proximity of the collision zone and the zone of global instability where the overlapping of mmrs occurs .
otherwise , the chaotic diffusion quickly leads the planets to collisions between each other or with the parent star .
the overall picture of the phase space of a planetary system is predicted by the fundamental kolmogorov - arnold - moser theorem @xcite : the phase space is not continuous with respect to the stability criterion .
hence , commonly used ( in particular , gradient - like ) algorithms of exploring the phase space are poorly designed for this task because they are `` blind '' to a sophisticated fractal - like structure of the phase space .
the kam stability is described in terms of stable ( regular , quasi - periodic ) and unstable ( chaotic ) motions . at first
, the use of such a formal criterion in the fitting process may be problematic .
almost any planetary system , including our own , can be very close to a chaotic state .
nevertheless , we expect that even if chaos appears , it should not impair the astronomical stability @xcite meaning that a system is bounded over a very long time and any collisions or ejections of planets do not occur .
however , for configurations involving jupiter - like companions in close orbits with large eccentricities , the formal stability seems to be well related to the astronomical stability of the system , i.e. , chaotic motions mean a fast destabilization of a planetary configuration over short - time scale related to the most significant , low - order mmrs .
it has been already demonstrated by dynamical analysis of systems residing in the regions of the phase space where the low - order mmrs are possible ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we should note that there is not known any general relation between the lyapunov time ( a characteristic time - scale of the formal instability ) and the event time ( the time after which a physical change of a planetary system happens ; see , e.g. * ? ? ?
* ; * ? ? ?
to put these ideas into action and to search for stable best - fit solutions in a self - consistent and optimal fashion , we treat the dynamical behavior ( in terms of chaotic and regular or mildly - chaotic states ) as an additional observable at the same level of importance as the rv measurements .
this powerful approach has been already described in detail and successfully applied in @xcite .
the kernel of our approach is the the genetic algorithm scheme ( ga ) implemented by @xcite in his publicly available code pikaia .
the gas are ideal for our purpose because of their global non - gradient nature and their proven ability to efficiently explore a multidimensional noncontinuous parameter space .
the rv data are modeled by a synthetic signal of the full @xmath2-body dynamics @xcite .
the @xmath3 function is modified by a stability penalty term employing an efficient fast indicator megno @xcite .
the ga fits are finally refined by yet another very accurate non - gradient minimization scheme by nelder and mead @xcite widely known as the simplex method .
this greatly reduces the cpu usage .
we give the algorithm an acronym gamp ( genetic algorithm with megno penalty ) .
the simplex method finds local minima of @xmath3 .
the code may be also trapped in resonance islands surrounded by strongly chaotic motions even if @xmath3 inside such islands is larger than in the neighboring ( but unstable ) areas . by collecting solutions to which the gamp converged in many independent runs , we gather an ensemble of the local best - fit solutions .
it helps us to illustrate the multidimensional properties of @xmath3 and to obtain realistic estimates of the parameter s errors by choosing the solutions within prescribed limits of the overall best - fit @xmath3 found in the entire search . at the end ,
some of the selected best fits can be refined with longer integration times and much lower simplex tolerance than is used during the search phase .
finally , the stability of the best - fit solutions is examined in planes of selected orbital osculating elements using the spectral number method ( @xmath4 ) by @xcite .
it is an efficient fast indicator completely independent on megno .
this enables us to verify and illustrate the best - fit solutions in a robust way and to examine dynamical properties of such configurations in wide ranges of neighboring initial conditions .
note that the @xmath4 is related here to the short - term dynamics .
thus the spectral signal analyzed is a time series @xmath5\}$ ] where @xmath6 and @xmath7 are respectively an osculating _ canonical _ semi - major axis and longitude of a planet .
such an analysis makes it possible to resolve the proper mean motion @xmath8 as one of the fundamental frequencies of the system . note
also that a stability criterion in the gamp code may be basically arbitrary .
we use the formal kam criterion as the most general and well defined one . [ [ section ] ] the hd 128311 is an active k0 star @xcite . in the discovery paper , @xcite found an indication of a jovian planet and a linear trend in the rv data .
they concluded that due to photospheric activity ( @xmath9 ) the stellar jitter is large @xmath10 m / s and the signal variability may be explained exclusively by the jitter . using an updated set of 76 rv measurements
, @xcite found that the observations can be modeled by a system of two jupiter - like planets involved in a 2:1 mmr .
the current estimate of the stellar jitter by these authors is @xmath11 m / s but still uncertain with a @xmath12 error .
we rescale the measurement errors by adding this estimate in quadrature to the formal rv errors .
the discovery team reports that the best - fit 2-keplerian model yielding @xmath13 and an rms @xmath14 m / s is catastrophically unstable .
using our hybrid ga / simplex code @xcite driven by the keplerian model of the rv , we found a different , apparently better 2-planet solution which has @xmath15 and an rms @xmath16 m / s .
the model parameters @xmath17 , i.e. , the semi - amplitude , orbital period , eccentricity , the argument of periastron and the time of periastron passage , for this fit are as follows : ( 51.948 m / s , 459.870 d , 0.362 , 59.401@xmath18 , 2474.867 d ) and ( 77.214
m / s , 917.371 d , 0.248 , 5.541@xmath18 , 2310.806
d ) for the inner and outer planet , respectively ; @xmath19 jd 2,450,000 and the velocity offset @xmath20 m / s .
it is argued that the fit parameters of a multi - planet system should be interpreted in terms of osculating keplerian elements and minimal masses related to the jacobi coordinates @xcite . adopting the date of the first observation as the osculating epoch
, we recalculated the inferred _ astrocentric _ osculating elements ( @xmath21 ) as follows ( 1.639 , 1.101 au , 0.362 , 59.37@xmath18 , 272.79@xmath18 ) and ( 3.194 , 1.746 au , 0.249 , 5.39@xmath18 , 199.62@xmath18 ) for the inner and outer planet , respectively .
still , the derived configuration is also unstable and disrupts during about 200,000 yr .
nevertheless , we found that its megno signature is characteristic for a system residing on the border of a stable region rather than a collisional configuration .
thus one may suspect that in its proximity , a rigorously stable solutions can be easily found . in order to deal with the problem of an unstable 2-kepler fit ,
@xcite applied a method of fitting which incorporates the mutual interaction between planets @xcite and , additionally , explicitly involves stability criterion .
as such the authors use the maximal eccentricity attained by the companions during an integration time .
they report many stable solutions corresponding to the 2:1 mmr . according to the authors
, their best fit yields an rms @xmath22 m / s and @xmath23 .
we could not reproduce that value of @xmath3 .
the quoted @xmath3 could be misprinted or a jitter estimate larger than @xmath11 m / s was used in the calculations . for a comparison with that result and as a background for a further analysis
, we performed the gamp search for the best - fit solution to the rv data from @xcite assuming that a 2:1 resonance is indeed present in the coplanar and edge - on system .
we also explored a more general model in which the orbits are mutually inclined but we did not find substantially better fits . in fig .
[ fig : fig1 ] , the elements of the best - fit solutions from the gamp runs are shown in a few representative planes of the osculating elements at the date of the first observation , jd , 2,450,983.827 . in the gamp code ,
the megno was evaluated over @xmath24 orbital periods of the outer planet that is both efficient enough and makes it possible to withdraw strongly chaotic , unstable solutions .
the best - fit initial conditions are marked with symbols of different sizes larger circles indicate a smaller @xmath3 ( a better fit ) . only _
stable _ solutions within the @xmath25 confidence interval of the best - fit solution ( given in table 1 ) are shown . in overall , the statistics of initial conditions shown in fig . [
fig : fig1 ] is in accord with the results of @xcite , see their fig .
the permitted _ initial _ eccentricities of the fits span skewed and narrow valley in the @xmath26)-plane .
let us note that we represent the @xmath2-body initial conditions in terms of astrocentric , osculating kepler elements at the epoch of the first observation .
the orbital elements for the outer planet have larger errors than for the inner one .
both semimajor - axes and phases are already very well determined .
the parameters of the stable best - fit solution are given in table 1 ( fit i ) .
note that this strictly regular solution has @xmath27 and rms @xmath28 m / s .
a very similar value of @xmath3 in the @xmath2-body and keplerian fits means that the mutual interactions between planetary companions are not evident in the doppler signal spanning 10 yr .
nevertheless , we stress that the stability constraints are essential for obtaining stable configuration of the @xmath2-body model . in figure
[ fig : fig2 ] , we show the stability analysis the best fit solution corresponding to a 2:1 mmr ( fit i given in table 1 ) . the spectral number , @xmath29 , as well as @xmath30 , the maximal eccentricities of both planets , and @xmath31 , the maximal @xmath32 ( where @xmath33 are the longitudes of pericenters ) , attained during the integration over @xmath34 orbital periods of the outer body are shown in figure [ fig : fig2 ] .
it turns out that the best - fit solution lies on the border of an island related to the corotation of apsides ( @xmath35 as well as the critical arguments of the 2:1 mmr are librating about @xmath36 with a large amplitude ) .
the border of the stable resonance zone which is present in the @xmath4-map can be also seen in all other maps , in particular in the @xmath37-maps .
this is a strong argument that the formal stability criterion is in the one - to - one relationship with the behavior of the system .
clearly , the search zone for the best - fit solution should be limited to the resonance island which constantly changes its shape when we change the orbital parameters .
thanks to the instability penalty in our approach , we have confidence that the obtained solution is indeed optimal ( i.e. , it minimizes @xmath3 and is dynamically stable ) . in our next test , we carried out a search for a stable trojan configuration .
our model was extended to 14 osculating orbital elements including the inclinations and one nodal longitude as free parameters .
note that due to very similar orbital semi - major axes , the planets are numbered by giving the symbol `` b '' to that planet which has a smaller initial eccentricity . as one can see in fig .
[ fig : fig3 ] , a well defined minimum of @xmath3 is present in the @xmath38 and @xmath39 planes .
the best - fit inclinations are not very well constrained nevertheless their concentration is quite evident , in spite of a moderate time span of the observations . the osculating elements of the best - fit solution are given in table 1 ( fit ii ) .
its @xmath40 and rms @xmath41 m / s are very close to those of the 2:1 mmr configuration . the synthetic rv signals of both solutions are shown in fig . [
fig : fig4 ] .
they can be barely distinguished one from another .
we also computed the lomb - scargle periodograms of both synthetic signals and we plotted them together with the periodogram of the data set in fig .
[ fig : fig5 ] .
it shows that periodograms of the 2:1 and 1:1 configurations almost perfectly match each other .
it would be very difficult to distinguish between the configurations by looking only at the periodograms .
the best - fit trojan configuration resides in a wide stable zone in the plane of the eccentricities which extends up to 1 for both of these elements ( see fig . [
fig : fig6 ] ) .
the resonance area in the @xmath42-plane covers about 0.2 au .
this width is even larger than for the 2:1 mmr configuration ( see fig . [
fig : fig2 ] ) . obviously , the stable 1:1 mmr is possible due to the corotation of the apsides seen in the @xmath31-maps ( @xmath35 librates about @xmath43 ) .
our choice of the stability criterion enables us to obtain very sharp borders of the resonance area .
if this criterion was violated , the system would quickly disrupt ( because both @xmath44 go to 1 ) . in fig .
[ fig : fig7 ] we show the stability maps for a solution which has an rms of about @xmath45 m / s ( slightly more than the best one ) and corresponds to much smaller masses ( both initial inclinations are about @xmath46 ) .
essentially , all the dynamical features of the system do not change but the width of the resonance zone shrinks substantially .
this is also an argument that an appropriate stability criterion has to be an integral part of the fitting tool .
due to an extremely nonlinear nature of the system , even a small change of its initial elements may lead to a significant change in the shape of the resonance zone .
simultaneously , @xmath47 becomes very `` flat '' in the regions of large @xmath44 so @xmath47 would not be a convenient stability indicator in a gamp - like code .
another argument is that the variable rate of the chaotic diffusion which leads to the changes of the eccentricity can be sometimes to small to detect a collision or a qualitative change of the configuration over relatively short integrations which , for efficiency reasons , have to be limited to the time - scale of the mmrs .
one might think that ( again , mainly for efficiency reasons ) @xmath31 would be a better choice than both @xmath47 and megno as a stability criterion .
however , that test may also fail as in the resonance zone the critical angle @xmath35 may librate about different centers ( usually , about of @xmath36 or @xmath48 ) but it can circulate in some marginally unstable regions as well ( see appendix for details ) .
[ [ section-1 ] ] the planetary system @xcite has drawn attention of many researchers ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the 2-keplerian solutions produced by the discovery team correspond to catastrophically unstable configurations @xcite .
the discovery team did not publish the observations of hd 82943 in source form .
the method of dealing with the problem of unavailable rv data relies on digitizing the published figures depicting the measurements .
this somewhat unusual approach has already become an accepted procedure ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
first , we digitized 142 data points from the figures of @xcite .
they slightly differ from the real observations .
in particular , it is difficult to recover the exact moments of the observations @xcite but such digitized measurements still properly describe the overall shape of the observed rv curve and its characteristic features .
we also graphically derived the measurement errors and rescaled them by adding in quadrature the stellar jitter which we estimated as @xmath49 m / s on the basis of @xcite .
having such `` measurements '' , we recovered the best - fit 2-kepler solutions published by the discovery team @xcite .
the only problem seems to be a little larger value of the rms of @xmath50 m / s ( compared to 6.8 m / s quoted in the original work ) . using the digitized rv measurements , @xcite showed that stable 2:1 mmr configurations are possible .
their orbital parameters of coplanar , edge - on systems are similar to those we found with gamp ( fit iii in table 1 ) .
we note that these authors looked for the best - fit solutions by minimizing the rms rather than @xmath3 , and they did not increased the internal errors by jitter .
also the discovery team did not account for the jitter in their solutions .
we extended the search for the best - fit solution assuming that a 1:1 mmr can be present in the system .
as in the previous case , we did not find any stable and strictly coplanar , edge - on configuration of this type .
however , using the generalized model in which masses , inclinations and one nodal longitude are free parameters , we found many stable solutions .
their quality is not as good as for the 2:1 mmr the best 1:1 mmr fit has @xmath51 and the rms is @xmath52 m / s which is about of 1 m / s worse than for our best 2:1 mmr solution .
still , the 1:1 mmr solution may be plausible ( note that we use digitized `` observations '' ) .
also , since the mass of the parent star can not be determined precisely @xcite , and when new measurements are available , the best - fit parameters and their @xmath3 may change . we demonstrate it in the next section . the best - fit 1:1 mmr configuration ( fit iv in table 1 ) is characterized by initially large mutual orbital inclination because , although the inclinations for both planets are almost the same , the nodal longitude is about @xmath48 ( and the apsidal lines are anti - aligned ) .
the orbital evolution leads to quite large variations of the orbital inclinations ( a few tens of degrees ) .
the stability maps shown in figure [ fig : fig8 ] reveal that , apparently , such a system would be locked in an extremely large zone of stable motions which extends up to @xmath53 .
this means that the eccentricities could reach extremely large values but the system would be still stable .
the width of the resonance with respect to @xmath54 is also relatively large , about @xmath55 au .
a zone of strictly periodic motions can be seen in the map for @xmath31 , close to its diagonal .
we extended the analysis of after gaining access to the same data set used by @xcite ( and also lee 2005 , private communication ) .
these authors also used the `` digitized '' measurements from @xcite but added new observations obtained with the keck / hires .
these new very accurate data fill in some gaps in the coralie measurements as well as significantly increase the time span of the observations . for consistency with @xcite
we adopted the jitter estimate of @xmath56 m / s .
the results of fitting 2-planet keplerian models and the analysis of their stability by @xcite strongly confirm the possibility of a stable 2:1 mmr in the system .
still , new questions may be asked .
the best fit of the 2:1 mmr configuration yields an rms @xmath57 m / s which is unexpectedly larger by 1 m / s from that quoted for the coralie data by the discovery team @xcite as well as by @xcite , and in this work . for the updated data set , we recovered all the best fit keplerian solutions quoted by @xcite using our ga / simplex code .
some of them appear to be formally chaotic or strongly unstable .
thus we searched for a stable @xmath2-body solution using gamp assuming that the velocity offsets for the coralie and keck / hires data are independent .
the found best - fit , _ rigorously stable _ solution corresponding to the 2:1 resonance of an edge - on system is given in table 2 ( fit v ) .
its quality is not very different from the best solutions found by @xcite but the initial eccentricities are significantly different .
our fit v is most similar to the 2-planet kepler fit ii of @xcite .
we also found other solutions which are similar to their fit iii and iv with respect to small initial @xmath58 . according to the results of @xcite and @xcite , @xmath58 and @xmath59 are the less constrained parameters of the kepler fits to the rv data of hd 82943 .
thus we computed dynamical maps for the relevant solutions ( see fig .
[ fig : fig9 ] and its caption ) as well as for fit ii of @xcite .
these maps reveal two narrow zones of stability in which the best - fit solutions reside .
all acceptable ( stable ) 2:1 mmr fits likely belong to these two distinct islands .
we label them with a and b in fig .
[ fig : fig9 ] .
note that the positions and shape of the resonance areas are significantly altered when the fit parameters are adjusted . inside the resonance zone
a , the amplitudes of the critical angles may vary in wide ranges .
@xcite found that their fit ii has very small amplitudes of the critical angles of the 2:1 mmr , @xmath60 .
our best @xmath2-body fit v yields much larger amplitudes , @xmath61 .
the large amplitudes of the critical angles are also reported by @xcite .
the resonance island a is characterized by the corotation of apsidal lines .
we found that in this zone @xmath31 may be very close to the libration center @xmath36 .
for instance , for the dynamical map of fit v , at ( @xmath62 ) the variations of @xmath63 , indicating a strictly periodic solution . in the second island labeled by b in fig .
[ fig : fig9 ] , @xmath31 also librates about @xmath36 but with large amplitudes .
we may speak about dynamical similarity of the best - fit solutions found so far , having in mind their position in the two resonance zones .
the results of the dynamical analysis done by @xcite and in this paper favor the 2:1 mmr fits located in the island a ( about @xmath64 ) .
this zone is extended with respect to the not well constrained @xmath58 and very small amplitudes of the 2:1 mmr critical angles are possible .
we also found many stable mutually inclined configurations using the orbital inclinations and one nodal argument as free parameters . still , all these fits have the rms @xmath57 m / s . by releasing the stability requirements in the gamp code ,
one finds the best 2-planet fit yielding the rms @xmath65 m / s ( fit vii in table 2 ) .
however , this configuration disrupts in a few hundred years .
curiously , such a solution involves a brown dwarf and a jovian planet on inclined orbits ( mutual inclination of @xmath66 ) .
in fact , our main goal was to perform a possibly extensive search for 1:1 configurations .
the statistics of stable solutions gathered in this search is illustrated in fig .
[ fig : fig10 ] , in a similar manner as for system .
qualitatively , the solutions do not differ from the ones we found using the coralie data only .
the semi - major axes and the eccentricities are very well constrained .
also the initial inclinations and masses are bounded to two well determined local minima of @xmath3 .
the best fit solution is given in table 2 ( fit vi ) .
its rms @xmath67 m / s is even closer to that of the 2:1 mmr best fit than in the case of the coralie data alone .
its megno signature which indicates perfectly stable , quasi - periodic configuration , and the evolution of orbital elements during 3 myr are shown in fig .
[ fig : fig11 ] .
the relevant dynamical maps of the best fit in the ( @xmath68)- and ( @xmath69)-plane are shown in fig .
[ fig : fig12 ] .
we also compared periodograms ( fig .
[ fig : fig13 ] ) of the synthetic curves for the 2:1 mmr , 1:1 mmr and the measurements ( shown in fig .
[ fig : fig14 ] ) .
also in this case the periodograms for the 2:1 and 1:1 mmr perfectly match each other .
the solutions described so far do not explain the curious rms excess that is present in the extended data set .
it seems unlikely the problem is caused by some inconsistency of the measurements from the two spectrographs .
another possible explanation is that there is a new , unknown object in the system .
that possibility is suggested by @xcite .
looking at their fig . 4 which shows the residual signal to the 2-planet solutions , we can see a quasi - sinusoidal modulation with the period about of 1000 d. yet the jitter estimates of hd 82943 are uncertain by @xmath12 @xcite . thus by adopting values as high as 67 m / s ,
one would obtain @xmath70 and the larger rms would be not necessarily unreasonable . still , the hypothesis about the third planet is a very attractive explanation of the rms excess .
below we try to find out whether such a configuration would be consistent with a stable dynamics .
first , we searched for 3-planet solutions using the hybrid kepler code and `` blindly '' assuming the same bounds of the orbital periods of [ 10,1200 ] d and eccentricities of [ 0,0.8 ] for every planet .
the use of multi - planet keplerian model enables us to quickly localize regions of orbital parameters in which potentially stable , @xmath2-body fits can be found .
the code was restarted thousands of times . in this search
the algorithm converged to a few distinct local minima yielding similar rms and @xmath3 .
remarkably , two of the keplerian best fits , yielding @xmath71 and an rms @xmath72 m / s correspond to _ coplanar _ configurations involving two ( of three ) jovian planets in 1:1 mmr .
the primary parameters ( @xmath73 ) of these fits are given in table 3 ( fit x and fit xi , respectively ) .
their planets c and d would have similar periods but the eccentricities of the two outer planets in 1:1 mmr are significantly different .
unfortunately , these fits are highly unstable .
we did not succeed in `` stabilizing '' them by gamp ; nevertheless , the search was not very extensive and we suspect that stable solutions involving mutually inclined orbits may exist .
the fit ix ( the mathematically best fit found in this paper ) yields @xmath74 and an rms @xmath75 m / s that indicates a almost `` perfect '' solution .
it could be interpreted as a configuration of the outermost planet accompanying the confirmed giants involved in the 2:1 mmr .
unfortunately , this solution is very unstable due to a large eccentricity of the outermost planet .
we tried to refine it with gamp . in the relevant range of semi - major axes we found stable solutions , and
the one we selected is given as fit viii in table 2 .
let us note that the stability criterion forces @xmath76 of this solution to a small value @xmath77 which also increases the rms to about 7.35 m / s .
the dynamical maps shown in fig .
[ fig : fig15 ] reveal narrow islands of stability in which the solution is found . the megno signature ( the left - upper panel of fig .
[ fig : fig16 ] ) uncovers a weakly chaotic nature of this solution .
nevertheless , there is no sign of a physical instability over at least 250 myr ( fig.[fig : fig16 ] is for the initial 5 myr integration period ) .
a peculiarity of this fit is that @xmath76 remains small in spite of a close proximity to the two larger companions in eccentric orbits . for a comparison with the previously found 1:1 mmr and 2:1 mmr solutions , the synthetic curve of fit viii
is shown in the bottom panel of fig .
[ fig : fig14 ] .
the described results might indicate that our knowledge of system is still limited in spite of much effort devoted to study the rv data of the parent star .
the currently available measurements permit many qualitatively different orbital solutions well fitting the measurements .
still , the use of stability criterion in the fit process seems to be essential to resolve the degeneracy between very good but strongly unstable keplerian and newtonian fits which , as we have shown above , can easily appear .
we have demonstrated the rv measurements for and harboring 2-planet systems may be successfully modeled with two qualitatively different orbital configuration .
one is already recognized configuration corresponding to a 2:1 mmr .
we show that these observations are equally well modeled with trojan pairs of planets ( a 1:1 mmr ) .
both these types of orbital configurations produce very similar periodograms of the rv signal .
a common feature of the trojan solutions for both systems is the possibility for large eccentricities of the orbits , reaching @xmath78 .
still , the best - fit trojan configurations reside in extended zones of rigorously stable quasi - periodic motions .
the ease of maintaining stability and the large zones of regular motions may strengthen the hypothesis about the 1:1 mmr configurations .
it is difficult to explain finding two systems in a 1:1 mmr in the sample of only @xmath10 multi - planet systems .
a most promising mechanism that might produce such a configuration is the dynamical relaxation and planetary scattering @xcite . in particular ,
an argument supporting such a hypothesis for the hd 82943 system is the evidence of a planet engulfment by the parent star @xcite .
that event indicates planetary scaterring in the past
. however , its effect on the currently observed configuration of the system would be hard to predict .
to the best of our knowledge there are no works that could explicitly explain an inclined 1:1 mmr configuration as a result of a migration . on the other hand ,
the origin of the 2:1 mmr is a well recognized problem as we know of at least four systems presumably involved in such a resonance .
our work demonstrates that the 1:1 mmr configurations can be used to describe the observations of and .
this hopefully will encourage others to study the origin of such systems .
the best - fit trojan configurations were found using of our approach of modeling the rv data which incorporates a stability indicator .
for this purpose we use a formal kam criterion which is closely related to a physical behavior of a planetary system . this criterion generalizes the @xmath47 and @xmath79 maps .
still , these maps help us to determine the character of motions e.g. a type of the corotation of the apsides .
obviously , all the three indicators are strictly related .
presumably , the stability maps would change if the model of motion included the relativistic and tidal interactions with the star . although these factors are orders of magnitude smaller than the leading gravitational interactions , their influence might change the overall stability picture of the systems .
our work on this subject is ongoing .
we appreciate discussions with prof .
sylvio ferraz - mello and his help with obtaining the rv data of the hd 82943 system .
we thank dr .
man hoi lee for a detailed review , critical remarks that improved the manuscript and for providing the full set of rv measurements of hd 82943 .
this work is supported by the polish ministry of education and science , grant no .
1p03d 021 29 .
m.k . is also supported by nasa through grant nng04gm62 g . here
we discuss the problem of a proper choice of the stability indicator in a gamp - like fitting code .
we analyze two 1:1 initial conditions for the system which are marked in the stability maps ( fig .
[ fig : fig8 ] , the upper row ; one is our best - fit 1:1 mmr configuration ) .
let us recall that the calculations of @xmath4 were conducted over @xmath80 orbital periods of the planets ( about of @xmath81 yr ) .
apparently , both initial conditions are localized in an extended resonance zone .
figure [ fig : fig17 ] ( upper row ) illustrates the temporal megno , @xmath82 , as a function of time but computed over a much longer time - span , @xmath83 orbital periods . for the best - fit solution
, the behavior of @xmath82 ( oscillations about 2 ) corresponds to a strictly quasi - periodic system while a slow divergence of this indicator can be observed for the modified initial condition ( marked with a diamond in the @xmath4 map , fig .
[ fig : fig8 ] ) .
it indicates that in this case the system is in fact weakly chaotic ( the lyapunov exponent is relatively small , @xmath84 ) .
an inspection of the @xmath31-map for this initial condition reveals that @xmath35 in this case circulates ( fig .
[ fig : fig8 ] and fig .
[ fig : fig17 ] , right column ) but on the average the apsides are anti - aligned and this helps to maintain the stability .
it means that @xmath31 would not be a good choice as a stability indicator . nevertheless , it is a valuable tool for resolving the complex structure of the resonance .
.the best - fit 2-planet initial conditions for the @xcite and @xcite planetary systems found with gamp ( megno was calculated over @xmath85@xmath86 periods of the more distant companion ) .
jitter estimates are @xmath87 m / s for and @xmath88 m / s for .
astrocentric osculating elements are given for the date of the first observation from @xcite and @xcite , respectively .
the masses of the parent stars are of @xmath89 for and of @xmath90 for . [ cols="<,^,^,^,^,^,^,^,^ " , ] | two nearby stars , hd 128311 and hd 82943 , are believed to host pairs of jupiter - like planets involved in a strong first order 2:1 mean motion resonance ( mmr ) . in this work
we reanalyze available radial velocity ( rv ) measurements and demonstrate that it is also possible to explain the observed rv variations of the parents stars as being induced by a pair of trojan planets ( i.e. , in a 1:1 mmr ) .
we show that these trojan configurations reside in extended zones of stability in which such systems may easily survive in spite of large masses of the planets , large eccentricities and nonzero mutual inclinations of their orbits .
we also show that hd 82943 could harbor a previously unknown third planet of @xmath00.5 jupiter masses in @xmath1 au orbit . |
the ground states of certain two - dimensional lattice hamiltonians of a type first introduced by kitaev@xcite can be used as quantum error - correcting codes known as surface codes .
quantum information can be stored and protected using these codes when they are defined on lattices with holes ( defects).@xcite fault - tolerant gates can then be carried out either transversally or by deforming the code in order to braid these defects while staying entirely within the code subspace.@xcite one downside to using the kitaev surface codes , for which defects behave as abelian anyons , is that to realize a universal set of fault - tolerant gates at least one gate using a resource costly `` magic state '' distillation process@xcite is required . the same is true for fault - tolerant quantum computation using the so - called color codes.@xcite nevertheless , quantum computation using these surface codes has a number of appealing features , notably the need for only nearest - neighbor gates between qubits in a two - dimensional array and high error thresholds , e.g. @xmath2 for the kitaev surface code.@xcite recently knig , kuperberg , and reichardt@xcite ( kkr ) outlined a method for fault - tolerant quantum computation using _ non - abelian _ surface codes .
these codes , which are defined mathematically in terms of the turaev - viro topological invariants for 3-manifolds,@xcite can be viewed physically as ground states of levin - wen models,@xcite two - dimensional lattice models which generalize the kitaev model.@xcite these models can be used to realize so - called `` doubled '' versions of any consistent anyon theory , including theories of non - abelian anyons for which braiding is universal for quantum computation .
the simplest such universal anyons are the fibonacci anyons . here
we refer to the corresponding levin - wen model as the fibonacci levin - wen model and , following kkr,@xcite refer to the ground states of this model as the fibonacci code . as shown in ref . , when using the fibonacci code , fibonacci anyons can be associated with holes in the lattice subject to certain boundary conditions and proper initialization .
these fibonacci anyons can then be used to encode logical qubits and universal quantum computation can be carried out purely by braiding them,@xcite without the need for magic state distillation .
the levin - wen models are defined by a set of commuting vertex and plaquette projection operators which act on qubits ( more generally , qudits ) associated with the edges of a two - dimensional trivalent lattice .
when using the ground states of these models as quantum codes it will be necessary to continually measure these vertex and plaquette operators in order to check for errors , which would then have to be corrected without disturbing the quantum information stored in the topological degrees of freedom of the code .
for the kitaev surface code , quantum circuits which can be used to measure these operators are known and are fairly straightforward.@xcite for either an @xmath0-sided plaquette , or a vertex where @xmath0 edges meet , these measurement circuits each require a single initialized syndrome qubit which is measured after carrying out @xmath0 controlled - not ( cnot ) gates .
the simplicity of the quantum circuits used to measure the vertex and plaquette operators for the kitaev surface code reflects the abelian nature of this code .
it is natural to ask how complex the quantum circuits need to be to measure the vertex and plaquette operators for the non - abelian fibonacci code . in this paper
we present explicit quantum circuits for performing such measurements .
these circuits are built in part out of smaller circuits which carry out unitary transformations which have been described both in kkr@xcite and , in the context of entanglement renormalization , in ref . .
our goal here is to explicitly construct these circuits in terms of standard elements ( toffoli gates , cnot gates and single - qubit rotations ) in an attempt to quantify their complexity .
the purpose of this work is not to argue that non - abelian surface codes are viable competitors to the kitaev surface code .
indeed , we share the view of many in the field that quantum computation using the kitaev surface code , given its clear advantages over other fault - tolerant quantum computation schemes , may well provide the best practical route to building a functioning quantum computer.@xcite here our goal is the more modest one of making a first pass at determining the complexity of syndrome extraction for the significantly less well understood fibonacci code , which we believe is of intrinsic interest in its own right .
an additional goal of this work is to begin developing a dictionary " for translating the mathematical structures which appear in general anyon theories into interesting quantum circuits , some of which we find require only a few qubits and might feasibly be carried out experimentally in the near future .
the levin - wen models@xcite are defined on two - dimensional trivalent lattices such as the hexagonal lattice shown in fig .
[ lattice](a ) .
the degrees of freedom of the models are associated with lattice edges which can take on a finite number of labels .
these labels can , in general , be oriented , meaning for each label @xmath3 there is a dual label @xmath4 . if @xmath5 then the edge is unoriented . for the fibonacci levin - wen model
there are only two labels @xmath6 and @xmath7 and the edges are unoriented ( @xmath8 , @xmath9 ) .
thus , for this model , as for the kitaev surface code , we simply associate a qubit with each edge of the lattice .
the two states of each qubit @xmath10 and @xmath11 then correspond to the two labels 0 and 1 , respectively . for a given trivalent lattice
the levin - wen hamiltonian has the form @xmath12 here @xmath13 and @xmath14 are projection operators associated with the vertices ( labeled @xmath15 ) and plaquettes ( labeled @xmath16 ) of the lattice .
the vertex operator @xmath13 acts on the three qubits associated with the edges connected to vertex @xmath15 and is diagonal in the standard @xmath17 , @xmath18 basis .
( here we focus on the fibonacci levin - wen model and so only consider the case when a single qubit is assigned to each edge . ) if these qubits are in the states @xmath19 , @xmath20 and @xmath21 the result of applying @xmath13 is determined by the tensor @xmath22 ( see fig .
[ lattice](b ) ) which , for the fibonacci levin - wen model , is given by , @xmath23 the plaquette operator @xmath14 is significantly more complex than @xmath13 . for example , for a hexagonal plaquette , @xmath14 acts on the six qubits on the edges of plaquette @xmath16 in a way determined by the state of the six qubits on the edges connected to the plaquette .
@xmath14 is therefore a twelve - qubit interaction ( in general a @xmath24-qubit interaction for an @xmath0-sided plaquette ) . for the fibonacci levin - wen model
the precise form of the plaquette projection operator is , @xmath25 where @xmath26 for @xmath27 and 1 are plaquette operators associated with the label @xmath28 and @xmath29 is the golden ratio .
the action of @xmath26 on an @xmath0-sided plaquette is shown in fig .
[ lattice](b ) where , @xmath30 here the six - indexed tensor @xmath31 , along with @xmath22 , forms the basic data of a so - called tensor category the mathematical framework for a general anyon theory , in this case the theory of fibonacci anyons .
the @xmath32 and @xmath33 tensors satisfy certain self - consistency conditions which , among other things , guarantee that the operators @xmath26 and @xmath13 all commute with each other.@xcite note that since the fibonacci levin - wen model is unoriented , in ( [ eq : bps ] ) we have assumed @xmath34 for all labels
. the precise form of the @xmath32 tensor for this model is given in sec .
[ fmove_sec ] . when using the ground states of the levin - wen model as quantum error - correcting codes the commuting vertex and plaquette projection operators @xmath13 and @xmath14 should be viewed as stabilizers .
the code space is then defined by the requirement that @xmath35 on each vertex and @xmath36 on each plaquette .
for the fibonacci code the constraint @xmath35 projects the hilbert space onto the space spanned by states in which edges in the state @xmath11 form branching loop configurations ( see fig .
[ lattice](a ) ) , while the plaquette constraint @xmath36 leads to particular quantum superpositions of these states .
as described in kkr,@xcite when these code states are defined on lattices with holes that have certain boundary conditions on their edges , these holes ( or defects ) can realize a `` doubled '' version of the anyon theory characterized by the @xmath32 and @xmath33 tensors .
for the fibonacci code this means that these defects can encode two types of fibonacci anyons with opposite chiralities .
as further shown in kkr,@xcite with proper initialization these defects can be forced to encode fibonacci anyons of a particular chirality .
these chiral anyons can then be used to encode qubits and braided in order to carry out universal quantum computation . in this paper
we focus on the problem of how to measure the stabilizers @xmath13 and @xmath14 for the fibonacci code . in the passive approach to the levin - wen model envisioned in kkr,@xcite rather than engineering the levin - wen hamiltonian to realize the fibonacci code
it will be necessary to continually measure these operators in order to detect errors which can then be corrected .
the measurement of @xmath13 for the fibonacci code is straightforward and not significantly more difficult to carry out than the analogous measurement for the kitaev surface code . a quantum circuit which carries out a quantum non - demolition measurement of @xmath13
is shown in fig .
the circuit acts on the three qubits associated with a given vertex as well as a fourth syndrome qubit initialized in the state @xmath10 . after carrying out the circuit
the syndrome qubit is measured . if it is found to be in the state @xmath10
then @xmath35 for this vertex and the vertex constraint is satisfied , if not then @xmath37 and the vertex constraint is violated .
the most difficult part of the @xmath13 circuit to carry out is likely to be the four - qubit toffoli gate which performs a not gate on the syndrome qubit if and only if the state of each of the three vertex qubits is @xmath11 .
( here and throughout it should be understood that an @xmath0-qubit toffoli gate is a gate with @xmath38 control qubits and one target qubit . )
this four - qubit toffoli gate is the first of several @xmath0-qubit toffoli gates required in our constructions , all of which are directly related to the non - abelian nature of the fibonacci code . here
this gate is needed to allow for the loop branching associated with the fact that @xmath39 . for the fibonacci code .
] in what follows we will be interested in quantifying the complexity of the quantum circuits we construct .
of course the notion of quantum circuit complexity is somewhat ill - defined and depends , among other things , on what we take as our primitive gate set .
this in turn will depend on the particular hardware of the quantum computer being considered .
accurate three - qubit toffoli - class gates have recently been been carried out experimentally using superconducting qubits@xcite and trapped ions.@xcite motivated by this , we take one primitive gate set to consist of three - qubit toffoli gates , cnot gates and single - qubit rotations .
an @xmath0-qubit toffoli gate can then be carried out using @xmath40 three - qubit toffoli gates if @xmath41 additional qubits are available.@xcite these additional qubits need not be initialized and their states are left unchanged once the full @xmath0-qubit toffoli gate is carried out .
thus nearby code qubits which are not being acted on directly by the operator under measurement can be used . with this construction
we can count the total number of three - qubit toffoli gates ( or , simply , toffoli gates ) , cnot gates and single - qubit rotations required to carry out a given circuit . for the case of the four - qubit toffoli gate appearing in our @xmath13 circuit
this count gives @xmath42 toffoli gates .
the total gate count for our @xmath13 circuit is then 4 toffoli gates and 3 cnot gates .
this can be contrasted with the analogous circuit for the kitaev surface code which , when acting on a trivalent vertex , would require only 3 cnot gates ( it is , in fact , identical to the circuit shown in fig .
[ qv ] with the four - qubit toffoli gate removed).@xcite for a second gate count we assume that the @xmath0-qubit toffoli gates which appear in our circuits are themselves primitive gates . by this count ,
our @xmath13 circuit consists of 1 four - qubit toffoli gate and 3 cnot gates .
we note that there are proposals for carrying out single - step @xmath0-qubit toffoli - class gates using trapped ions,@xcite superconducting qubits,@xcite and neutral atoms interacting with cavity photons;@xcite in addition , it has been observed that these gates are efficiently achieved if one of the qubits has @xmath0 available quantum levels.@xcite of course @xmath0-qubit toffoli gates can also be simulated using the usual primitive gate set consisting of cnot gates and single - qubit rotations.@xcite however , as we have seen with our @xmath13 measurement circuit , and as will become more clear in what follows , the ability to directly carry out accurate @xmath0-qubit toffoli gates ( with @xmath43 and 5 ) will give a strong advantage when carrying out quantum computation using the fibonacci code .
-move , a five - qubit unitary operation defined in terms of the tensor @xmath44 .
( b ) action of an @xmath32-move on the abstract trivalent lattice of the fibonacci code which illustrates the decoupling of this lattice from the physical qubits . in this example
the qubits ( open circles ) are arranged in a kagome lattice and lie on the edges of an initial trivalent ( hexagonal ) lattice .
after the @xmath32-move the edges of the new trivalent lattice must be distorted if they are forced to coincide with the physical qubit lattice . ] despite requiring a four - qubit toffoli gate , the @xmath13 measurement circuit shown in fig .
[ qv ] is relatively simple , reflecting the simplicity of the vertex operator . in what follows
we turn to the more difficult problem of measuring the plaquette operator @xmath14 . for this case a brute force approach to constructing a circuit which measures the appropriate operator acting on the edges of a plaquette for each possible state of the edges connected to that plaquette is problematic .
fortunately , there is a useful resource which simplifies the problem greatly the @xmath32-move .
when using the fibonacci code , the physical qubits of a quantum computer may be fixed in space and may even form a rigid lattice . however , this physical lattice need not be the same as that formed by the edges of the abstract trivalent lattice used to define the code .
indeed , as emphasized in kkr,@xcite this abstract trivalent lattice should be thought of as fluid and constantly changing throughout the computation .
these changes are accomplished by carrying out @xmath32-moves , processes which locally redraw the trivalent lattice while reassigning the physical qubits to new lattice edges and carrying out an appropriate unitary operation .
specifically , when carrying out an @xmath32-move five edges of the lattice are redrawn as shown in fig .
[ f_lattice](a ) while a unitary transformation determined by the six indexed tensor @xmath44 ( the same @xmath32 tensor which appears in ( [ eq : bps ] ) ) is applied to the five qubits associated with these edges .
this five - qubit unitary is a controlled operation on the qubit labeled @xmath45 in fig .
[ f_lattice](a ) contingent on the states of the other four qubits ( labeled @xmath46 ) .
the usefulness of the @xmath32-move here derives from the fact that if one starts in a ground state of a given levin - wen model on a particular trivalent lattice then , after performing an @xmath32-move , the resulting state will be a ground state of the new levin - wen model defined on the new trivalent lattice.@xcite this is true even though this lattice has decoupled from the physical qubits , as illustrated in fig .
[ f_lattice](b ) .
it was shown in kkr@xcite that the ability to decouple the abstract trivalent lattice from the physical qubits with @xmath32-moves is an important resource for carrying out quantum computation using the fibonacci code .
for example , by carrying out sequences of @xmath32-moves one can deform the code to perform dehn twists on the trivalent lattice which can then be used to braid defects encoding fibonacci anyons.@xcite since the braiding of fibonacci anyons is universal for quantum computation , this means that one can perform a universal set of gates while staying inside the fibonacci code subspace without the need for magic state distillation .
the @xmath32-move for the fibonacci code is represented graphically in fig .
[ f_move ] .
this figure , together with fig .
[ f_lattice](a ) , can serve as a definition of the @xmath32 tensor for fibonacci anyons .
the effect of carrying out an @xmath32-move is only shown for those states which satisfy the vertex constraint ( i.e. for which @xmath13 = 1 for all vertices ) .
when defining the levin - wen models , the @xmath32 tensor is assumed to vanish when acting on those states which violate the vertex constraint.@xcite here we will assume before applying any @xmath32-move that it has been verified that @xmath35 on each relevant vertex of the initial trivalent lattice .
the structure of the @xmath32-move then guarantees that the vertex constraint will continue to be satisfied on the new trivalent lattice .
-move for the fibonacci code ( the @xmath47 matrix @xmath32 is given in eq .
[ fmatrix ] ) .
the labels @xmath48 refer to the same labels in fig .
[ f_lattice](a ) .
( b ) five - qubit controlled-@xmath32 gate expressed in terms of a five - qubit toffoli gate . here
@xmath49 are single - qubit rotations about the @xmath50 axis with @xmath51 for which @xmath52 . ] a quantum circuit which acts on five qubits at a time and which carries out the @xmath32-move defined in fig .
[ f_move ] for those states satisfying the vertex constraint is shown in fig .
[ f ] . in this figure the labels
@xmath48 refer to the same labels shown in fig .
[ f_lattice](a ) .
although it is not immediately apparent from its structure , one can readily check that this circuit has the symmetries of the @xmath32 tensor@xcite ( e.g. , @xmath53 ) .
note also that the circuit squares to 1 ( since @xmath54 , see below ) , so the same circuit can be used for the inverse transformation .
as described above , this @xmath32 circuit carries out a particular operation on the qubit labeled @xmath45 depending on the state of the other four qubits labeled @xmath46 which are themselves left unchanged at the end of the circuit .
the @xmath32 circuit can therefore be viewed as a generalized toffoli - class gate . because the four control qubits are not equivalent , it is important to label these qubits in our @xmath32 circuit as we have done in the green box in fig .
[ f_move ] .
this notation will be useful when we embed the @xmath32 circuit into larger circuits acting on more than five qubits .
( b ) to the state @xmath11 . ] at the heart of the @xmath32 circuit is the five - qubit controlled-@xmath32 gate where @xmath32 is the @xmath47 unitary matrix acting on qubit @xmath45 when @xmath55 , @xmath56 the remaining toffoli gate and cnot gates take care of all other cases for which the outcome is essentially fixed by the vertex constraint .
as stated above , this circuit is designed to carry out an @xmath32-move only on those states which satisfy the @xmath57 constraint on all vertices . in what follows
we will always assume it has been verified that the vertex constraint is satisfied before applying the @xmath32 circuit .
figure [ f](b ) shows how to carry out the five - qubit controlled-@xmath32 gate using a five - qubit toffoli gate and two single - qubit rotations .
this simple construction is possible because @xmath54 and @xmath58 . as for the four - qubit toffoli gate appearing in the measurement circuit for @xmath13 ,
the appearance of this five - qubit toffoli gate can be traced back to the fact that loops are allowed to branch in the fibonacci code and is a direct consequence of the non - abelian nature of this code . using the construction of ref .
described above this five - qubit toffoli gate can be carried out using 8 conventional toffoli gates .
the total gate count for our @xmath32 circuit is then 9 toffoli gates , 4 cnot gates and 2 single - qubit rotations .
alternatively , if we treat @xmath0-qubit toffoli gates as primitives , the gate count is 1 five - qubit toffoli gate , 1 toffoli gate , 4 cnot gates and 2 single - qubit rotations . given the importance of carrying out @xmath32-moves when using the fibonacci code,@xcite the ability to accurately carry out this five - qubit toffoli gate can be viewed as an important experimental threshold for realizing this type of quantum computation .
the @xmath32-move satisfies an important self - consistency condition known as the pentagon equation .
the pentagon equation can be represented as a sequence of @xmath32-moves on a seven - edged trivalent lattice as shown in fig .
[ pentagon](a ) . in a quantum computer ,
the lattice edges would be associated with qubits , labeled 1 through 7 in fig .
[ pentagon](a ) .
as one follows this sequence of @xmath32-moves , the trivalent lattice is repeatedly redrawn while the qubits , which can be considered fixed in physical space , are reassigned to the new lattice edges after each @xmath32-move . by the time
one has gone all the way around the pentagon the trivalent lattice has returned to its original form .
however , the qubits associated with two of the edges ( labeled @xmath59 and @xmath60 in the figure ) are swapped . the process of carrying out this sequence of five @xmath32-moves and the resulting qubit swap can be translated into the quantum circuit identity shown in fig . [ pentagon](b ) .
we refer to the left - hand side of this identity as the pentagon circuit .
the solid green rectangles in the pentagon circuit represent the five - qubit @xmath32 circuit shown in fig .
[ f ] and the corresponding @xmath48 labels are the same as the labels shown in fig .
again we assume that before carrying out the pentagon circuit it has been verified that @xmath35 on each of the two vertices of the initial trivalent lattice .
it is only for this case that the circuit identity shown in fig .
[ pentagon](b ) holds ( for clarity these vertices are labeled by their associated qubits inside the red box under the equals sign in this figure ) . in the pentagon circuit two of the qubits ( qubits @xmath59 and @xmath60 )
are acted on while the remaining qubits play the role of control qubits .
simpler quantum circuits can be constructed by fixing these five effective control qubits to be in a particular state .
for example , if we fix all the qubits except for @xmath59 and @xmath60 to be in the state @xmath11 then the pentagon circuit reduces to the simple two - qubit circuit shown on the left - hand side of the circuit identity in fig .
[ pentagon_simple ] .
this simplified pentagon circuit consists of five controlled-@xmath32 gates with alternating control qubits , and the net effect of this sequence of gates is a swap gate . note that when qubits 5 and 6 are both in the state @xmath10 and all other qubits are in the state @xmath11 the vertex constraint is violated in the full seven - qubit pentagon circuit .
however , in this case the simplified pentagon circuit merely carries out the identity operation , which is consistent with swapping the two qubits . therefore the expression shown in fig .
[ pentagon_simple ] is an exact circuit identity , regardless of the vertex constraint .
we note the resemblance of this circuit identity to the familiar three cnot construction of the swap gate.@xcite in our case , the circuit identity shown in fig .
[ pentagon_simple ] represents the nontrivial part of the pentagon equation which uniquely fixes the form of the matrix @xmath32 ( up to an arbitrary and irrelevant phase choice for the off - diagonal matrix elements ) .
we envision that this circuit identity may be useful for calibrating the @xmath32 operation .
for example , one can imagine tuning @xmath32 until it can be verified by quantum process tomography that five controlled-@xmath32 gates with alternating control qubits indeed produce a swap gate .
we now turn to constructing a quantum circuit to measure the plaquette operator @xmath14 .
to do this we use a method inspired by the entanglement renormalization scheme of ref . .
the essential idea is that through a sequence of @xmath32-moves any @xmath0-sided plaquette can be reduced to a 1-sided plaquette with a single external line , i.e. a `` tadpole . ''
one such sequence of @xmath32-moves which reduces a hexagonal plaquette to a tadpole is shown in fig .
[ plaquette_reduction ] .
note that the final @xmath32-move in this sequence acts on four qubits rather than five . a quantum circuit which carries out this reduced @xmath32-move , obtained by identifying the qubits labeled @xmath61 and @xmath62 in the circuit shown in fig .
[ f ] , is shown in fig .
[ fprime ] .
( gate counts for this reduced @xmath32 circuit : 5 toffoli gates , 4 cnot gates , and 2 single - qubit rotations , or 1 four - qubit toffoli gate , 1 toffoli gate , 4 cnot gates , and 2 single - qubit rotations . )
it was shown in ref . that the plaquette operator @xmath14 commutes with @xmath32-moves , i.e. after each @xmath32-move shown in fig . [ plaquette_reduction ]
the value of @xmath14 is unchanged even as the plaquette size is reduced .
this is equivalent to the statement that if we start with a plaquette in a ground state of the levin - wen model ( meaning @xmath35 on each vertex and @xmath36 for the plaquette ) then , after each @xmath32-move , the qubits will continue to be in the ground state of the levin - wen model for the new lattice .
thus , after each @xmath32-move , it will still be true that @xmath35 on each vertex and @xmath36 on the reduced plaquette .
this means that after performing the `` disentangling '' reduction of the @xmath0-sided plaquette to a tadpole one need only measure @xmath14 for the tadpole to measure @xmath14 for the original plaquette .
since the tadpole only consists of two qubits this measurement is straightforward .
-move obtained by identifying the qubits labeled @xmath61 and @xmath62 in fig . [ f ] . ] transformation acting on a two - qubit tadpole .
the tensor @xmath63 is defined in the text .
( b ) @xmath64 circuit which carries out an @xmath64 transformation .
the @xmath47 matrix @xmath64 is given in eq .
[ smatrix ] . here
@xmath65 are single - qubit rotations about the @xmath50 axis with @xmath66 for which @xmath67 .
( c ) quantum circuit which uses the @xmath64 circuit to measure @xmath14 for a two - qubit tadpole . ] for a tadpole there is a unique eigenstate of @xmath14 with eigenvalue 1,@xcite @xmath68 which satisfy the vertex constraint will then have @xmath69 .
this space is spanned by the states latexmath:[\[\begin{aligned } to measure @xmath14 for this simple two - qubit system we first rotate the head qubit of the tadpole so that it is in the state @xmath10 if @xmath36 and in the state @xmath11 if @xmath69 .
this can be done by carrying out a single - qubit rotation @xmath64 on the head qubit if and only if the state of the tail qubit is @xmath10
where@xcite @xmath71 this transformation corresponds to the diagram shown in fig .
[ s](a ) and is defined in terms of the tensor @xmath72 which is equal to the matrix @xmath64 when @xmath73 and for which @xmath74 ( the case @xmath75 with @xmath76 or @xmath77 violates the vertex constraint ) . a quantum circuit which carries out this transformation ( and its inverse since the circuit squares to 1 )
is shown in fig .
this circuit can be carried out with 1 cnot gate and 2 single - qubit rotations . like the @xmath32 circuit ,
this simple construction is possible because @xmath78 and @xmath79 .
if the two tadpole qubits are initially in the state @xmath80 the result of carrying out this circuit is the state @xmath81 .
if the two tadpole qubits are initially in the two - dimensional hilbert space spanned by the states @xmath82 then after carrying out this circuit they will be in the space spanned by the states @xmath83 . in either case the state of the second qubit , i.e. the rotated head of the tadpole , will be equal to @xmath84 . after carrying out the @xmath64 circuit on the tadpole , a cnot gate can be done with the head qubit as the control qubit and a syndrome qubit , initialized to the state @xmath10 , as the target qubit .
the syndrome qubit can then be measured and if the result is @xmath6 then @xmath36 for the tadpole ( and hence for the original plaquette ) , and if the result is @xmath7 then @xmath69 .
after measuring @xmath14 for the tadpole , the final step is to reconstruct the full plaquette .
this can be done by undoing the @xmath64 circuit on the tadpole and then undoing the @xmath32-moves .
putting everything together the resulting measurement circuit for the case of a hexagonal plaquette is the palindromic circuit shown in fig .
[ bp ] . in this circuit
the notation is the same as in the pentagon circuit , with each box corresponding to either the full or reduced @xmath32 circuit , or the @xmath64 circuit , and the letters labeling the various `` inputs '' as defined in figs .
[ f],[fprime ] , and [ s ] . from the structure of the circuit
it is clear how this construction generalizes to the case of an arbitrary @xmath0-sided plaquette .
we again emphasize that the circuit shown in fig .
[ bp ] only measures @xmath14 correctly if the vertex constraint @xmath35 is satisfied on each vertex of the initial plaquette at the start of the circuit .
if the vertex constraint is violated on any of these vertices then by definition @xmath69 for the plaquette;@xcite but the circuit will , in some cases , give the wrong result of @xmath36 . to see this , consider the action of this circuit on the full @xmath85-dimensional hilbert space of the @xmath24 qubits associated with an @xmath0-sided plaquette , including those states which violate the vertex constraint .
from the structure of the circuit , which performs a unitary transformation on @xmath24 qubits and then measures the state of a single qubit to determine @xmath14 , it is clear that the dimensionalities of the hilbert spaces for which @xmath36 or @xmath69 would both be @xmath86 , i.e. half that of the full hilbert space .
however , once the vertex constraint is taken into account the hilbert space is greatly reduced .
the dimensionalities of the projected hilbert spaces for which @xmath35 on each of the @xmath0 vertices and @xmath87 or @xmath69 for the plaquette are @xmath88 = f_{2n-1}$ ] and @xmath89 = f_{2n+1}$ ] , respectively , where @xmath90 is the @xmath0th fibonacci number ( @xmath91 , @xmath92 , @xmath93 , @xmath94 , etc . ) . for the case of a hexagonal plaquette
this means the full @xmath95 dimensional hilbert space of twelve qubits is projected down to a space of dimensionality @xmath96 with an 89 dimensional space of states satisfying the plaquette constraint with @xmath36 . the reader will be reassured to know we have numerically checked that the circuit shown in fig .
[ bp ] performs the correct measurement of @xmath14 on this projected space .
it should be noted that the requirement that @xmath35 on each vertex before measuring @xmath14 may cause problems when extracting error syndromes .
for example , if a faulty measurement of @xmath13 gives 1 for a particular vertex on a plaquette , but the actual value of @xmath13 is 0 for that vertex , then , as described above , the @xmath14 measurement circuit for the plaquette will , in some cases , give @xmath36 even though the correct value ( as it is for any plaquette in which a vertex constraint is violated ) is @xmath69 . in this paper
we have not addressed the important question of whether it is possible to extract error syndromes for the fibonacci code fault tolerantly , nor the question of precisely how these errors would be corrected .
our goal has been to construct circuits which , in the absence of errors , can be used to measure @xmath13 and @xmath14 in order to begin to get a measure of their complexity .
we can now give our final gate counts for measuring @xmath14 . if we choose to reduce all @xmath0-qubit toffoli gates to conventional three - qubit toffoli gates ( using @xmath97 toffoli gates , following ref .
as described in sec .
[ sec : qv ] ) then we find that our procedure for an @xmath0-sided plaquette ( with @xmath98 ) requires @xmath99 toffoli gates , @xmath100 cnot gates and @xmath101 single - qubit rotations . alternatively ,
if we consider @xmath0-qubit toffoli gates as primitives , then our procedure requires @xmath102 five - qubit toffoli gates , 2 four - qubit toffoli gates , @xmath103 toffoli gates , @xmath100 cnot gates and @xmath101 single - qubit rotations.@xcite not surprisingly , this is significantly more demanding than the analogous requirement for the kitaev surface code , for which only @xmath0 cnot gates are needed to measure the plaquette operator for an @xmath0-sided plaquette .
finally we note that there are , of course , many different ways to reduce a given plaquette to a tadpole using @xmath32-moves , all of which can be used to measure @xmath14 and some of which will be more `` parallelizable '' than others .
one of the motivations of the present work is to find simple quantum circuits which might feasibly be carried out in the near term and which begin to test some of the key properties of the fibonacci code .
we have already seen one example of such a circuit , the sequence of five controlled-@xmath32 gates which results in a two - qubit swap gate discussed in sec .
[ sec : pentagon ] .
this circuit is a simplified version of the full seven - qubit pentagon circuit shown in fig .
[ pentagon](b ) and can potentially be used to calibrate the @xmath32 operation . in this section
we give a similar example a four - qubit circuit which first initializes a tadpole into a state with either @xmath36 or @xmath6 , and then pulls this tadpole through a line using @xmath32-moves to produce a new tadpole which can be measured to verify that the value of @xmath14 has not changed . as for the pentagon circuit
, this four - qubit circuit can be simplified to a two - qubit circuit which , in this case , can be used to calibrate the @xmath64 operation .
the sequence of operations we consider is illustrated in fig .
[ sf_circuit](a ) .
the system consists of a four - edged trivalent lattice and so uses four qubits , labeled 1 through 4 in the figure .
initially two qubits ( 1 and 2 ) are assigned to edges which form a line and the other two qubits ( 3 and 4 ) form a tadpole attached to this line . as always , in what follows we assume that it has been verified that @xmath35 on each of the two vertices of this lattice at the start of the process .
the first step is to initialize the tadpole in a state with either @xmath36 or @xmath6 .
then , using two @xmath32-moves , as shown in fig .
[ sf_circuit](a ) , the tadpole can be pulled through the line .
the @xmath32-moves preserve @xmath14 , and so the intermediate state of this process is a 2-sided plaquette which has been initialized either into the code space if @xmath36 or outside of the code space if @xmath69 . after the tadpole has been pulled through the line , the two qubits forming the initial tadpole have swapped places the head of the tadpole is now the tail and vice versa . if @xmath14 is now measured for the new tadpole the result should yield the same value of @xmath14 that the tadpole was initialized to at the start of the process .
the left - hand side of the circuit identity shown in figure [ sf_circuit](b ) is a four - qubit circuit which carries out the procedure described above .
if qubit 4 is initially in the state @xmath104 then the first @xmath64 circuit initializes the tadpole in a state with @xmath105 .
a reduced @xmath32 circuit then carries out the first @xmath32-move and produces a 2-sided plaquette with @xmath105 .
next , a second reduced @xmath32 circuit carries out the second @xmath32-move producing a new tadpole with @xmath106 but with the head and tail of the tadpole interchanged .
finally , after carrying out an @xmath64 circuit on this tadpole the state of qubit 3 will be @xmath104 .
note that if qubit 4 is initially in the state @xmath11 so that the tadpole is initialized to a state with @xmath107 then qubit 3 can initially be in either the state @xmath10 or @xmath11 while still satisfying the vertex constraint .
after the first @xmath64 circuit on the left - hand side of fig .
[ sf_circuit](b ) is carried out the tadpole will then be placed in a quantum superposition of @xmath108 and @xmath109 ( see eqns .
( [ bp2 ] ) and ( [ bp3 ] ) ) . at the end of the circuit , after being pulled through the line formed by qubits 1 and 2 , the tadpole will still be in the two - dimensional @xmath69 hilbert space , but the particular superposition will in general have changed .
direct calculation shows that if qubits 1 and 2 are both in the state @xmath11 then the operation acting on the two - dimensional @xmath110 space when pulling the tadpole through the line is given by @xmath111 otherwise , if either qubit 1 or qubit 2 ( or both ) are in the state @xmath10 the state of the final tadpole will be the same as that of the initial tadpole with the head and tail qubits swapped .
these cases are all accounted for by the swap gate and four - qubit controlled-@xmath112 operation on the right - hand side of the circuit identity in fig .
[ sf_circuit](b ) . as for the pentagon circuit ,
this identity only holds when @xmath35 on the two vertices of the initial lattice ( again these vertices are labeled inside the red box in the figure ) . in the circuit identity shown in fig .
[ sf_circuit](b ) .
( b ) equivalent circuit identity obtained by moving the two not gates from the left side of the identity shown in ( a ) to the right side . ]
this four - qubit circuit , which essentially represents initializing a 2-sided plaquette into a state with a given value of @xmath14 and then producing a state which can be measured to determine @xmath14 after carrying out a different @xmath32-move than the one used to initialize it , is much simpler than the full circuit for measuring @xmath14 for a hexagonal plaquette .
however , it still involves the four - qubit toffoli gate which appear in the reduced @xmath32 circuit .
as for the pentagon circuit , a simpler two - qubit circuit identity can be found by fixing the states of the qubits which act effectively as control qubits ( qubits 1 and 2 in fig .
[ sf_circuit ] ) . if we fix these qubits to both be in the state @xmath11 we obtain the simplified two - qubit circuit identity shown in fig .
[ sf_simple](a ) .
this circuit identity can be simplified further by multiplying both sides on the left and right by not gates which act on the top and bottom qubits , respectively , to obtain the equivalent circuit identity shown in fig .
[ sf_simple](b ) .
note that if the initial state for the circuit shown in fig .
[ sf_simple](a ) is @xmath113 , where the first qubit is the top qubit ( qubit 3 in fig .
[ sf_circuit ] ) , then the vertex constraint is not satisfied for the full four - qubit circuit .
however , the simplified circuit identity is readily seen to be satisfied in this case . for all other cases
the vertex constraint is satisfied , and so it follows that the expression shown in fig . [ sf_simple](a ) and the equivalent expression in fig .
[ sf_simple](b ) are exact circuit identities , independent of whether or not the vertex constraint is satisfied .
the key action of the two - qubit circuit on the left - hand side of fig . [ sf_simple](b ) occurs when the tadpole is initialized in a state with @xmath36 for which the tail qubit must start in the state @xmath10 . for this case , after pulling the tadpole through the line the new tadpole must again be in the state with @xmath36 .
thus , after accounting for the removal of the two not gates , this circuit must take the state @xmath114 to the state @xmath115 .
like the five controlled-@xmath32 swap circuit in fig .
[ pentagon_simple ] , which can be used to calibrate the @xmath32 matrix , the circuit identity shown in fig .
[ sf_simple](b ) can be used to calibrate the @xmath64 matrix . once @xmath32 has been fixed by the pentagon circuit , the requirement that the circuit on the left - hand side of fig .
[ sf_simple](b ) takes the state @xmath113 to the state @xmath115 fixes the form of the matrix @xmath64 ( up to an overall phase which is irrelevant for our purposes ) .
note that in performing this calibration it is not necessary to carry out a full quantum process tomography .
it is sufficient to verify that the circuit identity holds for the initial state @xmath113 . for this case
, only the swap gate on the right - hand side is relevant since the controlled-@xmath116 gate enters only when the initial state of the second qubit is @xmath11 .
in this paper we have constructed explicit quantum circuits for measuring the vertex and plaquette operators , @xmath13 and @xmath14 , in the fibonacci levin - wen model .
these operators can be viewed as stabilizers for the fibonacci code,@xcite a surface code for which defects can behave as fibonacci anyons the simplest non - abelian anyons for which braiding alone is universal for quantum computation . while the @xmath13 measurement is not significantly more difficult than the analogous measurement for the kitaev surface codes ( for which the defects behave as abelian anyons ) , the @xmath14 measurement scheme we present here _ is _ significantly more difficult than its abelian counterpart . while the present scheme is almost certainly not the most efficient one for performing this measurement , given the complexity of the operator @xmath14 it is likely that even the most efficient schemes will require a large number of primitive gate operations .
this cost in circuit complexity will then need to be weighed against the gain of not requiring magic state distillation .
the situation is somewhat analogous to comparing the relative merits of performing topological quantum computation with ising anyons ( which requires magic state distillation ) to fibonacci anyons.@xcite it is clear that further work will be needed before such a direct comparison of the resources needed to carry out fault - tolerant quantum computation using the fibonacci code with that using the kitaev surface code will be possible .
while recent progress strongly suggests that the kitaev surface code is the most promising from the practical point of view of trying to build an actual fault - tolerant quantum computer , we believe it is too early to rule out the possibility that the fibonacci code may have some practical implications . even if it does not , we believe the fibonacci code is of intrinsic interest , in part because computing with it can be viewed as essentially simulating a non - abelian state of matter on a quantum computer .
our measurement circuit for @xmath14 is built out of circuits which realize @xmath32-moves and the action of the @xmath64 matrix on a trivalent lattice .
in addition to our measurement circuits we have also given simpler circuits built out of these @xmath32 and @xmath64 circuits .
the first is a seven - qubit circuit which can be used to verify that the @xmath32 circuit satisfies the pentagon equation , as well as a simpler two - qubit circuit which contains the nontrivial content of this equation and fixes the form of the @xmath32 matrix .
the second is a four - qubit circuit which uses the @xmath64 circuit to initialize a tadpole ( 1-sided plaquette ) in a state with either @xmath36 or @xmath6 , carries out a sequence of @xmath32-moves to pull the tadpole through a line , and then produces a state which can be measured to determine @xmath14 for the new tadpole . for this circuit
we have also given a simpler two - qubit circuit which , once @xmath32 has been fixed by the pentagon circuit , fixes the form of the @xmath64 matrix .
these simple two - qubit circuits ( figs .
[ pentagon_simple ] and [ sf_simple ] ) may be useful for calibrating the @xmath32 and @xmath64 operations .
a recurring theme in this work has been the need for @xmath0-qubit toffoli gates ( with @xmath1 and 5 ) when computing with the fibonacci code .
these @xmath0-qubit toffoli gates arise as a natural consequence of the non - abelian nature of this code .
we believe the possibility of using non - abelian surface codes such as the fibonacci code for fault - tolerant quantum computation provides further motivation for developing experimental techniques to directly carry out accurate @xmath0-qubit toffoli - class gates .
the connection between the turaev - viro invariants and the levin - wen models is discussed in sec .
7 of ref . .
see also z. kadar , a. marzuoli , m. rasetti , int . j. quantum inform .
* 7 * 195 ( 2009 ) ; and f. j. burnell and s. h. simon , ann . of phys . *
325 * , * 11 * , 2550 ( 2010 ) .
@xmath117 is the identity operator when acting on states which satisfy the vertex constraint .
thus only @xmath118 contributes nontrivially to @xmath14 for the fibonacci code with the particular linear combination ( [ eq : bp ] ) forming a projection operator ( see ref . ) .
m. mariantoni , h. wang , t. yamamoto , m. neeley , r. c. bialczak , y. chen , m. lenander , e. lucero , a. d. oconnell , d. sank , m. weides , j. wenner , y. yin , j. zhao , a. n. korotkov , a. n. cleland , and j. m. martinis , science * 334 * , 61 ( 2011 ) .
@xmath0-qubit toffoli gates can be carried out efficiently ( for @xmath119 ) using @xmath120 cnot gates and @xmath121 single - qubit rotations ( see ref . ) . thus our @xmath13 measurement circuit ( fig .
[ qv ] ) can be carried out using 17 cnot gates and 16 single - qubit rotations .
if all @xmath0-qubit toffoli gates are simulated using cnot gates and single qubit rotations following ref . , ( see ref . )
we find the total gate count for measuring @xmath14 for an @xmath0-sided plaquette with @xmath122 is @xmath123 cnot gates and @xmath124 single - qubit rotations .
this gate count should be viewed as an upper bound because we have not considered possible gate merging . | we construct quantum circuits for measuring the commuting set of vertex and plaquette operators that appear in the levin - wen model for doubled fibonacci anyons .
such measurements can be viewed as syndrome measurements for the quantum error - correcting code defined by the ground states of this model ( the fibonacci code ) .
we quantify the complexity of these circuits with gate counts using different universal gate sets and find these measurements become significantly easier to perform if @xmath0-qubit toffoli gates with @xmath1 and 5 can be carried out directly .
in addition to measurement circuits , we construct simplified quantum circuits requiring only a few qubits that can be used to verify that certain self - consistency conditions , including the pentagon equation , are satisfied by the fibonacci code . |
the response of a fermi gas to an external , time - dependent perturbation is a well - known hard problem .
it has repeatedly been studied in various contexts , mostly on systems exhibiting the orthogonality catastrophe , fermi - edge singularities , or the kondo effect@xcite .
the technical difficulties are connected to the necessity of summing an infinite number of logarithmically divergent diagrams , and has led to the development of a multitude of methods to tackle this task , including diagram summations , singular integral equation approaches , bosonization , and summation schemes based on slater determinants@xcite .
many of these approaches are technically involved or require assumptions such as , for instance , the separability of the external potential .
in addition , the recent controversies concerning nonequilibrium fermi systems@xcite have shown that extensions of these methods are problematic and hard to achieve .
a unified approach , depending on a minimum of assumptions is , therefore , desirable .
very recently , a major step in this direction has been made in the form of the _ _ riemann - hilbert approach__@xcite , and applied to the calculation of fermi - edge singularities at @xmath0 in fermi gases under nonequilibrium constraints@xcite . in this approach
, the calculation of general response functions in the perturbed fermi gas is transformed into the problem of solving a ( generally non - abelian ) riemann - hilbert ( rh ) boundary value problem : the response functions are rewritten in terms of complex matrix functions @xmath1 that satisfy boundary conditions , for @xmath2 , expressed by a matrix function @xmath3 . in most cases of interest , @xmath3 is directly related to the scattering matrix of the time - dependent potential . in this paper , we extend the rh technique to finite temperatures . we first give a precise definition of the finite temperature rh problem .
we then show that the finite temperature problem is equivalent to a different zero temperature rh problem through the bijective mapping of the time variable @xmath4 , with @xmath5 . in simple ( i.e. abelian ) cases ,
this allows us to construct the explicit solution , but generally the resulting rh problem remains highly nontrivial .
we show , however , on the concrete example of the nonequilibrium problem of refs . how the equivalence between finite and zero temperature solutions can be explored to construct approximate solutions to the corresponding rh problem in the high and low temperature limits .
the paper is organized as follows : in sec .
[ sec : zerotemp ] , we summarize the essential steps in the zero temperature rh approach . in sec .
[ sec : finitetemp ] , the finite temperature rh problem is defined , and the equivalence to the zero temperature case is shown . finally , in sec .
[ sec : discussion ] , we discuss the finite temperature extensions for some simpler examples and for the nonequilibrium fermi - edge singularity problem discussed in refs . .
we start with the introduction of the required concepts and notations for the zero temperature rh approach , following ref . .
the approach has been developed for the calculation of response functions in a nonadiabatically perturbed system of fermions .
we consider a gas of noninteracting fermions that is exposed to a time - dependent potential @xmath6 . the hamiltonian is assumed to take the form @xmath7 where @xmath8 labels the single - particle energies of the electrons , and @xmath9 is a vector of the electron operators @xmath10 , where @xmath11 is a channel index labeling any further classification of the single particle states .
the potential @xmath12 is a @xmath13 matrix in channel space , which we assume to be nonzero only for some finite time interval @xmath14 , during which the external perturbation is switched on .
the quantities of interest are response functions of the type @xmath15 at zero temperature , @xmath16 denotes the average over the ground state @xmath17 ; at finite temperatures , as considered in sec .
[ sec : finitetemp ] , it represents the thermal average over the noninteracting system states at @xmath18 .
the operator @xmath19 is related to the time - evolution operators @xmath20 and @xmath21 of @xmath22 and @xmath23 , respectively , or equivalently to the scattering matrix @xmath24 for the potential @xmath6 . for the present discussion , the precise form of @xmath19 is unimportant .
a summary of necessary conditions on @xmath19 is given in appendix [ sec : time_repr_r ] , while a detailed discussion can be found in refs . and ; some examples are given at the end of this paper .
since the hamiltonian is quadratic in the electron operators , the action of the many - body operator @xmath19 is fully specified by its action on single - particle states ( see the appendices ) .
let @xmath25 be the matrix formed of the matrix elements of @xmath19 between the single - particle states @xmath26 , @xmath27 , with @xmath28 the true vacuum .
for the following treatment , it is essential to assume that @xmath29 is diagonal in the time representation , @xmath30 ( with @xmath31 the variable conjugated to the energy representation @xmath8 ) .
for the response functions relevant for fermi - edge singularity problems , it can be shown@xcite ( see appendix [ sec : time_repr_r ] ) that this assumption is valid as long as the variation of the scattering matrix is slow compared to the delay time of the scattering process itself . for the noninteracting system
, @xmath17 is a slater determinant in the single particle states . as shown in appendix [
sec : deriv_eq3 ] , we can write @xmath32 where @xmath29 represents the matrix elements of @xmath19 taken between the single - particle states , and @xmath33 is the fermi function , which has the matrix elements @xmath34 in this expression , we have set the chemical potential @xmath35 to zero . this is possible without loss of generality even in the nonequilibrium case of several fermi seas @xmath36 with different chemical potentials @xmath37 , such as for biased tunneling barriers . for the unperturbed system with uncoupled fermi seas , we can shift the single - particle energies by @xmath38 and set all chemical potentials to zero .
in other words , the fermion operators are replaced by @xmath39 .
the hamiltonian keeps the same form except that the potential @xmath40 , and thus @xmath3 , acquires an additional dependence on time@xcite , @xmath41 .
the logarithm of the determinant in eq . can be written in the form @xmath42 + { \mathrm{tr}}[\ln(1-f+fr ) - f \ln r ] \\ & \equiv \ln\chi_r^{(1 ) } + \ln\chi_r^{(2 ) } , \end{aligned}\ ] ] with @xmath43 being the trace over channel indices @xmath36 and energy .
@xmath44 $ ] consists of the diagonal terms of @xmath45 summed over the occupied states of the unperturbed system . depending on the system under investigation
, it expresses , for instance , a shift of the ground state or threshold energy ( fumi s theorem@xcite if @xmath29 is the scattering matrix ) together with an imaginary correction for nonequilibrium conditions@xcite , or the average transfer of charge across a barrier@xcite .
the second term , @xmath46 , contains the nontrivial effects close to the fermi surface which are , for instance , associated with fermi - edge singularities . in the following ,
we will focus on @xmath46 only .
it is convenient to switch from the energy representation to the time representation of the trace , in which @xmath29 is diagonal .
the fermi function , however , becomes nondiagonal in time and reads @xmath47 ( here and henceforth we set @xmath48 . ) the logarithm of an infinite matrix is represented by introducing a dependence on a parameter @xmath49 as @xmath50 and by writing @xmath51
\frac{\mathrm{d}r}{\mathrm{d}\lambda } \right],\ ] ] where @xmath52 denotes the trace over the channel index @xmath36 only .
the difficult task is to find the inverse of @xmath53 . for this ,
a complex @xmath13 matrix function @xmath1 is introduced , which solves an auxiliary _ rh problem _ ,
defined by : a. [ prop:1 ] @xmath1 is a nonsingular @xmath54 matrix function , which is analytic and nonzero in the complex plane , except on the interval @xmath55 , where it satisfies the boundary condition @xmath56 for @xmath57 , and except at the end points @xmath58 , where @xmath1 may vanish or be weakly singular as @xmath59 for @xmath60
. b. [ prop:2 ] @xmath1 has the asymptotic condition @xmath61 from these two properties , two important identities follow : @xmath62 which equally hold if @xmath63 is replaced by @xmath64 . in the time representation ,
. reads @xmath65 which is proved by closing the contour in the upper half - plane .
a similar contour in the lower half - plane proves eq . .
in terms of the functions @xmath66 , the inverse of @xmath53 is expressed by @xmath67,\ ] ] which can be verified by multiplication by @xmath53 from the right or left , and by making use of the identities and .
finally , the following result @xmath68 b\bigr\ } & = \int { \mathrm{d}t}\lim_{t'\to t } { \mathrm{tr}}\frac{i}{2\pi } \left[\frac{a(t)-a(t')}{t - t'+i0}\right ] b(t ' ) \nonumber \\ & = \frac{i}{2\pi } \int { \mathrm{d}t}\ , { \mathrm{tr}}\left\{\frac{\mathrm{d}a(t)}{{\mathrm{d}t } } b(t ) \right\ } \label{eq : relation}\end{aligned}\ ] ] for any differentiable matrix functions @xmath69 and @xmath70 , allows to write eq . in the compact form
@xmath71 the response function @xmath72 , therefore , is fully expressed in terms of the solution @xmath1 of the rh problem .
we show in this section , that we can define a finite temperature version of the rh problem such that the important formula ( and , therefore , eq . ) remains valid .
subsequently , we show that this new rh problem is equivalent to a zero temperature rh problem .
note that eq
. remains valid if we generalize the identities and to @xmath73 with the same @xmath74 , a ( finite ) matrix function , appearing in both identities . at finite temperatures @xmath75 ,
the fermi function becomes @xmath76 with @xmath5 . with this function , eq
. reads explicitly @xmath77 the fermi function is antiperiodic in the imaginary time direction , @xmath78 .
if we impose that eq . and are invariant under such transformations , @xmath1 and @xmath79 must be ( anti)periodic in the imaginary time direction as @xmath80 and @xmath81 .
this allows us to restrict the analysis to the strip @xmath82 .
the _ finite temperature rh problem _ can then be formulated as follows : a. [ prop : i ] @xmath1 is a nonsingular @xmath54 matrix function , which is analytic in the strip @xmath83 , except on the interval @xmath55 , where it satisfies eq . , and except at the end points @xmath84 , where it may vanish or be weakly singular as @xmath85 with @xmath86
. b. [ prop : ii ] @xmath1 tends to definite finite values as @xmath87 in @xmath88 ( the values on the far right or far left , however , are generally different ) . c. [ prop : iii ] @xmath1 can be analytically continued through @xmath89 , for real @xmath31 , such that @xmath90 . for convenience ,
we normalize @xmath1 ( by multiplication by a constant matrix from the right ) such that @xmath91 .
condition is absent in the zero temperature case , but is essential for the validity of eqs . and .
indeed , let us first focus on eq . .
we integrate @xmath92 over the contour @xmath93 shown in fig .
[ fig : contour1 ] . in the figure
, @xmath94 is a large number , which we eventually send to infinity .
since @xmath1 remains bounded as @xmath95 , the integrals over the vertical lines @xmath96 and @xmath97 vanish in this limit , and we obtain eq .
with @xmath98 given by the integral of @xmath92 over the line through @xmath99 , parallel to the real axis .
a similar contour in the lower part of @xmath88 leads to eq . , with @xmath98 given by the integral over the line through @xmath100 , parallel to the real axis . both integrals for @xmath98 are identical due to the periodicity and analyticity of @xmath1 , and eqs . and are fulfilled .
the solution of the finite temperature rh problem , together with eqs . and , immediately leads to the solution for the response function @xmath101 by means of eq . .
( note that eq
. remains true when replacing @xmath102 by @xmath103^{-1}$ ] . )
the finite temperature rh problem , defined by conditions - , is equivalent to the rh problem at zero temperature , defined by conditions and , upon a redefinition of the boundary condition .
the equivalence is shown by means of the mapping @xmath104 which is often used to switch between zero and finite temperatures , and which describes a bijection between the strip @xmath88 and the slit plane @xmath105 , w \not\in [ 1,\infty ) \}$ ] .
let us set @xmath106 for @xmath107 .
* the boundary condition for @xmath1 and @xmath108 becomes a boundary condition of the same type for @xmath109 and @xmath110 , @xmath111 , and the matrix @xmath112 . *
the analyticity of @xmath1 in @xmath113 $ ] ensures the analyticity of @xmath109 in @xmath114 $ ] .
the periodicity @xmath90 for real @xmath31 is nothing but the condition of analyticity of @xmath109 through the real axis at @xmath115 .
* @xmath116 maps onto infinity in the @xmath117plane . with the normalization @xmath118
, we have @xmath119 as @xmath120 . on the other hand
, the boundedness of @xmath1 as @xmath87 in @xmath88 implies that @xmath109 is finite at @xmath121 .
this proves that the zero and finite temperature versions of the rh problem are equivalent , upon the change of boundary condition @xmath122 . with the mapping , the identities and are mapped onto and , too .
the infinite @xmath31-integrals over the real axis , map onto @xmath123 in the @xmath124-plane ( including the cut @xmath125 $ ] ) , while the integrals over the contours @xmath126 and @xmath127 provide the remaining parts of the real axis in the @xmath124-plane , @xmath128 and @xmath129 , respectively , at which @xmath109 is analytic .
the integrands can be identified through the formula @xmath130 the dependence on @xmath131 in the integrand cancels with @xmath132 .
after factorization of @xmath133 , we see that eqs . and are identical to and .
a solution to the finite temperature rh problem allows the calculation of the response function .
we illustrate this with a few simpler examples of how the equivalence to the zero temperature rh problem can be used to construct explicit finite temperature solutions .
we then discuss approximate finite temperature solutions for the nonequilibrium rh problem of ref . .
the rh problem appeared early in the study of fermi - edge singularities .
it is part of the solution of the dyson equation for the green s function , formulated by nozires and de dominicis as a singular integral equation@xcite , for which the standard method of solution consists in the transformation into a rh problem@xcite . in nozires and
de dominicis case , the boundary condition @xmath29 is a constant , for @xmath108 , given by the single - particle single - channel scattering matrix @xmath134 , where @xmath135 is the scattering phase shift ( @xmath136 ) .
the solution is given by @xmath137 the finite temperature extension to this expression was achieved by yuval and anderson@xcite in their application of the nozires - de dominicis approach to the kondo problem .
it resulted from writing @xmath138 , the free green s function in the kernel of the integral equation , as the sum of @xmath139 over integer @xmath140 in the kernel of the integral equation .
the corresponding rh problem becomes the product of zero - temperature rh problems with branch cuts shifted by @xmath141 .
hence , the finite temperature solution found by yuval and anderson reads @xmath142 with the replacement @xmath4 in eq . and the use of eq .
, this solution can immediately be verified , as constant matrices @xmath29 and @xmath143 coincide .
note that with the mapping we obtain an additional constant factor @xmath144 , reflecting the fact that a solution of the rh problem remains a solution upon multiplication by a constant matrix from the right . with our normalization
this factor is suppressed .
such constants are unimportant since they drop out in the physical solution .
eqs . and
remain valid if @xmath29 is a time - independent diagonalizable matrix . in this case
, @xmath135 becomes a matrix with eigenvalues representing the phase shifts of the multi - channel scattering matrix @xmath29 . for scalar , but time - dependent @xmath3 ,
the solution of the rh problem can be constructed in a standard way@xcite , and is given by @xmath145 at finite temperatures , the mapping allows here the explicit solution , too , which has the same form as the previous equation , with @xmath29 replaced by @xmath143 , @xmath146 notice that all these expressions become singular at @xmath147 or @xmath148 .
these divergences are spurious and due to the approximation of an infinite bandwidth ( given by the form of the fermi function ) , and have a natural cutoff in the true inverse bandwidth@xcite , @xmath149 . in these simple examples
, the response function can explicitly be calculated .
let us , for instance , consider eq . .
since @xmath150 , we obtain @xmath151 \\
= - \frac{{\mathrm{tr}}\{\delta^2\}}{\pi^2 } \ln\left[\frac{\sinh(\pi t_f/\beta)}{\sinh(\pi/\beta\xi)}\right],\end{gathered}\ ] ] again with the cutoff at the inverse bandwidth @xmath149 . for short times
@xmath152 , we obtain a power - law similar to the well known zero temperature result @xmath153 , which , however , crosses over into an exponential decay as @xmath154 , @xmath155 . as a more involved example , let us consider the non - abelian rh problem , in which @xmath3 is a time - dependent matrix that does not allow a time - independent diagonalization .
such situations naturally arise in nonequilibrium systems , and have recently been studied with the rh approach@xcite : consider , for instance , the coupling between a defect state in a tunneling barrier and the connecting electrodes that are kept at a voltage bias @xmath156 .
the defect is assumed to be either in its ground state @xmath157 or in in an excited state @xmath158 .
we further assume that in the ground state the defect is fully screened and does not interact with the conduction electrons , whereas in the excited state it acts on the conduction electrons through the potential @xmath6 .
the diagonal matrix elements of this potential describe the scattering of the electrons in the electrodes , while the nondiagonal matrix elements represent a modification of the tunneling rate .
the matrix @xmath159 is the scattering matrix for @xmath6 .
the response function @xmath160 is related to the transition rate of the defect into the excited state , @xmath161 , by @xmath162 which is the golden rule expression rewritten as a time integral .
@xmath163 is the bare energy difference between ground state and excited state of the defect , and @xmath164 is the bare amplitude in the hamiltonian for the transition @xmath165 .
a closed solution to the resulting time - dependent rh problem is not known . in the short time
limit @xmath166 and in the long time limit @xmath167 , however , it is possible to construct approximate solutions@xcite .
we first give an example for such an approximation at zero temperature , and then discuss how it extends to finite temperatures . in the nonequilibrium fermi - edge singularity problem@xcite , @xmath3 is a unitary time - dependent scattering matrix coupling two fermi liquids , @xmath168 and @xmath169 , at the left and at the right of a tunneling barrier .
the nonequilibrium constraint is introduced by a finite voltage bias @xmath156 , keeping the left and right chemical potentials at the values @xmath170 and @xmath171 , respectively ( the electronic charge @xmath158 is set to unity ) . as noted in sec .
[ sec : zerotemp ] , we can shift @xmath172 to 0 in the unperturbed system by passing to the interaction representation @xmath173 , where @xmath174 are the number operators for the electrons in the electrodes 1 and 2 .
this is equivalent to a gauge transformation , transforming the electron operators as @xmath175 .
the voltage difference now appears only as an additional time dependent phase factor in the matrix elements of @xmath3 that couple electrodes 1 and 2 : @xmath176 where @xmath177 and @xmath178 are the backscattering matrix elements , and @xmath179 and @xmath180 describe transmission through the barrier . for @xmath181 ,
the scattering matrix @xmath29 is time - independent during @xmath55 , and the explicit solution of the rh problem is given by eq .
, with @xmath135 the matrix of phase shifts , @xmath182 .
if @xmath183 , an explicit solution to the rh problem is not known .
the naive solution , obtained by multiplication of eq
. by @xmath156-dependent phase factors @xmath184 would solve eq . but violate condition , expressed by eq .
, because of the divergence of @xmath184 for large complex - valued arguments @xmath31 .
such a divergence can not be compensated by multiplication by @xmath31-dependent analytic matrix functions from the right .
the solution would become invalid .
an approximate solution , however , can be constructed as follows@xcite : for @xmath185 , we decompose @xmath3 into the product @xmath186 with @xmath187 , @xmath188 , @xmath189 , @xmath190 , and @xmath191 . now define the function @xmath192\right),\ ] ] where @xmath193 .
this function satisfies the boundary condition @xmath194 at @xmath108 . to approximate the unknown solution @xmath1 , we consider the following function @xmath195 : @xmath196 where the functions @xmath197 are defined in the regions @xmath198 indicated in fig .
[ fig : zones ] .
@xmath199 has branch cuts between the boundaries of all 4 regions @xmath200 .
most importantly , between regions @xmath201 and @xmath202 , it satisfies @xmath203 furthermore , @xmath204 as @xmath87 .
hence @xmath195 satisfies the conditions and of the rh problem , at the price of the additional vertical branch cuts through @xmath205 and @xmath206 .
these branch cuts , however , become exponentially small , @xmath207 , with increasing distance to the real time axis .
for instance , consider the vertical cut at @xmath208 in the upper half - plane .
we have @xmath209 similar exponential suppressions hold for the remaining vertical cuts .
therefore , @xmath210 is a good approximation to the solution of the rh problem if we consider times @xmath31 such that @xmath211 .
notice that in the boundary condition only @xmath188 and @xmath212 are involved , the matrix elements of backscattering into the left electrode at @xmath170 and into the right electrode at @xmath171 .
this reduction of the full scattering matrix can be explained in a simple picture based on the uncertainty relation : for long times @xmath167 , the energy uncertainty @xmath213 of the scattered particles is smaller than the voltage , @xmath214 .
the fermi surface of the other electrode , therefore , is not seen by the scattered particle , and only the backscattering part of the scattering matrix is of importance . for the left electrode ,
this is @xmath177 , while for the right electrode , the pauli principle forbids virtual trajectories into the left electrode .
after removal of these , the scattering amplitude becomes @xmath215 .
since @xmath177 and @xmath178 alone are not unitary the real phase shifts @xmath135 are replaced by complex numbers .
the implications , mainly , a broadening of the fermi - edge singularity , are shortly discussed below . for more details
we refer to refs . . in the limit @xmath216 or @xmath217 , on the other hand , nonequilibrium constraints play no major role , and the equilibrium expression is expected to become appropriate .
this indeed is confirmed by the same argument of the uncertainty relation : for small times , @xmath166 , the energy uncertainty @xmath213 of the scattered particles becomes large , @xmath218 , and both fermi surfaces at @xmath219 and @xmath220 are coupled .
the full scattering matrix is involved .
let us focus now on low temperatures , @xmath221 .
an approximate solution to @xmath1 to the rh problem can be constructed in the same way as for the zero temperature case : we decompose @xmath3 according to eq . , and consider the rh problem for @xmath222 , described by eq . .
since this problem is diagonal with a time - independent boundary condition , it can be exactly solved by the mapping on the zero - temperature problem .
its solution is given by eq .
with @xmath223 replaced by @xmath224 , @xmath225\right).\ ] ] the function @xmath195 is given by the same conditions as before , but where the regions @xmath198 now are defined in the strip @xmath88 , with additional branch cuts at @xmath226 ( see fig .
[ fig : zones ] ) the validity of the approximation on the vertical branch cuts remains unchanged from the zero temperature case , and corresponds to @xmath227 . on the additional horizontal cut ,
@xmath228 and @xmath229 differ in their nondiagonal parts , which are of the order of @xmath230 .
this leads to corrections for the identities and of the form @xmath231 where @xmath232 is the integral over all coinciding parts of @xmath233 along @xmath226 .
@xmath234 are the contributions due to the mismatch of @xmath235 and @xmath236
. both , @xmath237 and @xmath238 , are finite quantities of the order of unity .
since @xmath239 , their contribution is negligible compared with the error arising from the vertical cuts . on the other hand , for @xmath240 , the voltage ( i.e. the nonequilibrium situation ) as well as the temperature become unimportant
the solution to the rh problem , therefore , is that of a zero temperature equilibrium system , expressed by eq . .
at high temperatures , @xmath241 , we expect that @xmath156 plays no longer an important role .
indeed , since @xmath242 , the naive solution of the rh problem , @xmath243 now becomes accurate . because of the restriction of @xmath31 to the strip @xmath88 ,
@xmath1 can not diverge exponentially , so that @xmath244 remains bounded as @xmath87 in @xmath88 . yet , eq .
violates the condition of periodicity , @xmath245 due to the exponentials @xmath246 .
again , this leads to different integrals @xmath98 from the integration along @xmath247 and @xmath248 , respectively .
since @xmath242 , however , we can expand the exponential functions , and we are led to similar identities as before @xmath249 with @xmath250 of the order of unity .
the solution @xmath63 , and hence the inverse of the matrix , which is solved by the rh problem , as well as the logarithm of the response function are accurate within corrections of the order of @xmath251 .
the results above allow us to extend the nonequilibrium fermi - edge singularity problem to finite temperatures .
we consider the case where @xmath3 has the form given in eq . with @xmath252 .
the high temperature limit leads to an expression , which is unchanged from the equilibrium result . at lower temperatures @xmath221 , on the other hand
, we can use the approximation @xmath253 for for @xmath254 . with these functions and the decomposition of @xmath29 in eq . , the matrix products in eq .
can be evaluated to @xmath255 where @xmath256\ ] ] and @xmath257 is a time - independent function of @xmath49 .
the term proportional to @xmath257 leads to a contribution to @xmath258 that is linear in @xmath259 , and hence consists in a voltage dependent ( generally complex ) shift of the ground state energy , similar to that of @xmath260 .
the time integration over @xmath261 has to be cut off at @xmath262 and @xmath263 , where the solution @xmath1 crosses over into the equilibrium solution .
the remaining integral can be estimated by integrating the equilibrium solution from @xmath264 to @xmath265 .
this leads to @xmath266 \\ - \frac{{\mathrm{tr}}\{\delta^2\}}{\pi^2 } \ln\left[\frac{\sinh(\pi/\beta v)}{\sinh(\pi/\beta \xi ) } \right],\end{gathered}\ ] ] where we recall that @xmath267 . since @xmath268 ,
this can be rewritten as @xmath269 \\ - \frac{{\mathrm{tr}}\{\delta^2\}}{\pi^2 } \ln(\xi / v).\end{gathered}\ ] ] for @xmath270 , this function crosses over into the equilibrium solution . since both @xmath135 and @xmath271 are time - independent ( for @xmath108 ) we see that , in this limit @xmath272 , the extension from zero to finite temperatures , originally achieved by the mapping @xmath273 , can here be obtained by a rule of thumb of replacing @xmath206 by @xmath274 in the logarithms . for @xmath275
this means that the zero temperature power laws of the form @xmath276 are replaced by @xmath277^\alpha$ ] .
therefore , finite temperatures replace the power - law tails by an exponential decay with a decay time proportional to the inverse temperature @xmath278 . in other words , while the system at zero temperature keeps a long - term memory to the initial perturbation , finite temperature fluctuations suppress it within a characteristic time set by @xmath279 .
the transition rate @xmath161 is determined by the fourier integral .
an explicit calculation of the integral is possible and leads to a beta function , yet the following estimates are physically more instructive : at zero temperature both , the singularity at @xmath147 as well as the power - law tails for @xmath280 contribute equally to the integral . in the equilibrium situation , we obtain the well - known fermi - edge singularity , a power - law dependence on energy , @xmath281 . for @xmath183 , @xmath163 and the exponent , acquire an imaginary correction .
the nonequilibrium effects dominate for small energies @xmath282 , and the imaginary corrections to @xmath163 lead to a broadening of the fermi - edge singularity .
the high energy tails , @xmath283 , coincide with the equilibrium solution . at finite but small temperatures , @xmath221 ,
the exponential decay of the response functions for @xmath284 causes an additional , lorentzian , broadening of the singularity for @xmath285 . at high temperatures , @xmath241 ,
the nonequilibrium effects are completely washed out . for small energies , @xmath286 ,
the exponential decay then dominates the integral , and the singularity is broadened to a lorentzian with a width proportional to @xmath287 . on the other hand , for high energies @xmath288 , the exponential decay is slow compared to the quickly oscillating phase factor .
the behavior of the integral can be estimated through a saddle point approximation , leading to @xmath289 , corresponding to an activated transition such as for a noninteracting system . in summary , nonequilibrium constraints and temperature
have the similar effect of a broadening of the fermi - edge singularity with two different shapes : a lorentzian broadening from the temperature with a width given by @xmath287 , and a lorentzian to the power of an expression depending on @xmath271 from the nonequilibrium constraints with a width determined , to first order , by @xmath156 .
away from the resonance , but still for @xmath286 , the transition rate has a power - law dependence on energy , which eventually turns into an activated behavior for @xmath290 .
we have shown that the rh approach developed in refs . and has an exact finite temperature extension .
this finite temperature case can be mapped onto a different zero temperature problem , for which much is mathematically known@xcite , and which allows , in some simpler cases , an exact solution .
based on this equivalence we have constructed approximate solutions for the nonequilibrium rh problem at finite temperatures .
these results provide a necessary tool for a general application as well as for further developments of the rh approach and the study of fermi - edge singularities or related phenomena in nonequilibrium or nonadiabatic situations .
i thank d. e. feldman , j. b. marston , j. merino , and t .- k .
ng for helpful discussions and comments .
this work is supported in part by the nsf under grant number dmr-0213818 .
the presented method requires that the time - representation of @xmath29 , the matrix elements of the many - body operator @xmath19 between single - particle states , is diagonal , @xmath30 . for clarity and completeness , we here shortly state the conditions necessary for this , summarizing the arguments of ref .
( see also refs . and , in particular , ) .
for a more extended discussion we refer to the cited references . in most cases of interest , @xmath19 is connected to the evolution operators @xmath20 and @xmath21 for the hamiltonians @xmath291 and @xmath292 , respectively .
for instance , in the case of the fermi - edge singularities , @xmath19 represents the overlap between perturbed and unperturbed states , @xmath293 . on the other hand , for the shot noise spectrum of a tunneling barrier
, @xmath19 can be the moment generating function for the distribution of charge transfered out of electrode @xmath168 during the time @xmath18 to @xmath206 . if @xmath294 measures the charge of this electrode , @xmath295 .
since the hamiltonian is quadratic , the action of @xmath20 on the many - body states can fully be described by its action on the single - particle states ( see also appendix [ sec : deriv_eq3 ] ) , @xmath296 where @xmath297 is a @xmath54 matrix in channel space , and @xmath28 the true vacuum .
the matrix @xmath298 can be related to the single - particle scattering matrix @xmath299 for a particle of energy @xmath300 calculated for the instantaneous value of the potential @xmath40 . in general , this relation is complicated , but becomes simple when the following condition of adiabaticity is met : @xmath301 this equation encodes nothing than the condition that the scattering matrix has to vary slowly during the characteristic scattering time of a particle ( the wigner delay time ) . in this case , the connection to the matrix @xmath298 is simple , @xmath302 where @xmath303 with @xmath304 and @xmath305 the density of states in channel @xmath36 . the time representation @xmath3 is obtained by a similar fourier transform of @xmath306 .
@xmath307 as noted above , the definition of @xmath29 involves @xmath298 or simple combinations of @xmath298 , so that the same condition of adiabaticity applies .
the transformations in the rh method do not involve the dependence of these matrices on @xmath300 . at low temperatures and in equilibrium ,
the relevant physics is restricted to energies close to the chemical potential @xmath35 , so that we can fix @xmath308 . for nonequilibrium situations ,
the treatment remains valid as long as the dependence of @xmath29 on @xmath300 is weak .
in this appendix , we demonstrate the equality @xmath309 = \det\bigl ( f r + ( 1-f ) \bigr),\ ] ] stated in eq .
that connects a many - body description on the left - hand side to a single - body representation on the right - hand side . @xmath19
is the many - body operator of interest , @xmath310 the initial many - body density matrix at @xmath18 , @xmath311 , and @xmath43 the trace over all many - body states . on the other hand , @xmath312 is the @xmath313 matrix between the @xmath314 possible single - particle states in the system , @xmath315 ( for simplicity of notation , we here label single - particle states by a single index @xmath316 only in the main text this is split into energy @xmath8 and channel indices @xmath36 ) .
@xmath28 is the true vacuum , not the ground state , @xmath33 the corresponding fermi function , and @xmath317 the determinant over the @xmath313 matrices .
the hamiltonian @xmath318 is quadratic in the single - particle operators for all times @xmath31 . at time
@xmath18 and at zero temperature , the density matrix projects on the ground state @xmath17 , which is a slater determinant in the single - particle states @xmath319 .
if the many - body operator @xmath19 conserves particle numbers , it is , therefore , fully described by its action on the single - particle constituents of the ground state , represented by the matrix @xmath29 .
explicitly , for a system of @xmath320 particles , the ground state is @xmath321 , where the sum runs over all permutations @xmath322 , @xmath323 is the sign of the permutation , and where we assume that @xmath324 labels the @xmath325 single - particle states of lowest energy .
we have @xmath326 , where @xmath327 is the restriction of @xmath19 to the single - particle state @xmath316 .
further @xmath328 , from which follows that @xmath329 \\&\ \times \hat{r } \bigl[|\psi_{\pi(1)}\rangle \otimes \dots \otimes |\psi_{\pi(n)}\rangle \bigr ] \\ & = \frac{1}{n ! } \sum_{\pi \pi ' } ( -1)^{\pi+\pi ' } \bigl [ \langle\psi_{\pi'(1)}| \otimes \dots \otimes \langle\psi_{\pi'(n)}| \bigr ] \\&\ \times \bigl [ \sum_{j_1,\dots , j_n=1}^{n_s } r_{j_1\pi(1 ) } |\psi_{j_1}\rangle \otimes \dots \otimes r_{j_n \pi(n)}|\psi_{j_n}\rangle \bigr ] \\ & = \frac{1}{n ! } \sum_{\pi \pi ' } ( -1)^{\pi+\pi ' } r_{\pi'(1)\pi(1 ) } \dots r_{\pi'(n)\pi(n ) } \\ & = \sum_{\pi } ( -1)^{\pi } r_{1\pi(1 ) } \dots r_{n\pi(n)}. \end{aligned}\ ] ] this is the determinant of @xmath29 , taken over the occupied states only , and thus corresponds to the right - hand side of eq . .
the generalization to nonzero temperatures is immediate : every excited state of @xmath325 particles yields a similar determinant of @xmath29 for a different selection of @xmath325 particles @xmath330 .
the density matrix @xmath310 weights these states with the boltzmann factors @xmath331 .
since these are exponentials of single - particle operators , they just provide the weights @xmath332 for the occupied single - particle states that sum up to the fermi function when summing over all excited states .
therefore , eq . remains valid at nonzero temperatures . 9 for reviews
see : k. ohtaka and y. tanabe , rev .
phys . * 62 * , 929 ( 1990 ) ; g. d. mahan , _ many particle physics _ , 3rd ed .
( kluwer / plenum , ny , 2000 ) . t
ng , phys .
b * 54 * , 5814 ( 1996 ) ; m. combescot and b. roulet , phys . rev .
b * 61 * , 7609 ( 2000 ) ; b. braunecker , phys . rev .
b * 68 * , 153104 ( 2003 ) . b. a. muzykantskii and y. adamov , phys . rev .
b * 68 * , 155304 ( 2003 ) .
n. dambrumenil and b. muzykantskii , phys .
b * 71 * , 045326 ( 2005 ) .
b. muzykantskii , n. dambrumenil , and b. braunecker , phys .
lett . * 91 * , 266602 ( 2003 ) .
y. adamov and b. muzykantskii , phys .
b * 64 * , 245318 ( 2001 ) .
f. g. fumi , philos . mag . * 46 * , 1007 ( 1955 ) .
p. nozires and c. t. de dominicis , phys . rev . * 178 * , 1097 ( 1969 )
. n. i. muskhelishvili , _ singular integral equations _ ( p. noordhoff ltd . , groningen , nl , 1953 ) ; n. p.
vekua , _ systems of singular integral equations _ ( p. noordhoff ltd . , groningen , nl , 1967 ) ; r. estrada and r. p. kanwal , _ singular integral equations _
( birkhuser , basel , 2000 ) . g. yuval and p. w. anderson , phys . rev .
b * 1 * , 1522 ( 1970 ) . of definition of the functions @xmath334 ( eq . ) .
branch cuts are indicated by the wiggled lines .
( a ) zero temperature case .
( b ) modification for finite temperatures with additional branch cuts at @xmath335 .
[ fig : zones ] ] | we provide an exact finite temperature extension to the recently developed riemann - hilbert approach for the calculation of response functions in nonadiabatically perturbed ( multi - channel ) fermi gases . we give a precise definition of the finite temperature riemann - hilbert problem and show that it is equivalent to a zero temperature problem . using this equivalence
, we discuss the solution of the nonequilibrium fermi - edge singularity problem at finite temperatures . |
the background caused by scattering of neutrons off the irradiated sample is a serious issue in neutron capture experiments . through subsequent neutron interactions with the materials surrounding the sample ,
secondary reaction products are created such as @xmath0 rays and/or charged particles which may be detected alongside the capture @xmath0 rays emitted from the sample , contributing to the total background .
this particular contribution , referred to as the _ neutron background _
, is most notable for the samples characterized by a large neutron scattering - to - capture cross section ratio . in general , neutron background is characteristic of environments which are strongly affected by the neutron scattering .
it is intensified by the presence of any neutron - sensitive material in the immediate vicinity of the detectors , and especially by the detector proximity to the walls of the experimental hall .
the neutron background is determined by two distinct components , one being the sample itself , serving as the primary neutron scatterer , and other being the sample - independent _ neutron sensitivity _ related to the entire experimental setup .
the neutron sensitivity may be generally defined as the detector response to reaction products created by the interaction of scattered neutrons with the surrounding materials . a nontrivial effect of the neutron sensitivity on the neutron background and , consequently , the entire capture measurement has been demonstrated by koehler et al .
@xcite in capture measurement on @xmath1sr , where the reduction of the neutron sensitivity of the experimental setup has led to significant improvements in the acquired capture data . at the neutron time - of - flight facility n_tof at cern ,
neutron sensitivity considerations have been followed since the start of its operation .
this was reflected through the development of specially optimized c@xmath2d@xmath2 ( deuterated benzene ) liquid scintillation detectors , exhibiting a very low intrinsic neutron sensitivity @xcite . however , the neutron background at the n_tof facility is heavily affected by the surrounding massive walls , serving as the prime candidates for the enhanced neutron scattering . furthermore , the much higher neutron energies available from the n_tof spallation source introduce an additional contribution to the neutron background , when compared to the neutron sources based on electron linacs , where the neutron energies are usually limited to @xmath310 mev .
details on the n_tof facility can be found in refs .
@xcite .
recently , geant4 @xcite simulations were developed for determining the neutron background in the measurements with c@xmath2d@xmath2 detectors at n_tof @xcite .
the results of these simulations were first applied in the analysis of the experimental capture data for @xmath4ni @xcite and the analysis of the integral cross section measurement of the @xmath5c(@xmath6)@xmath5b reaction @xcite .
an earlier @xmath4ni capture measurement by guber et al .
@xcite has already revealed that previous experimental results and adopted evaluations of the @xmath4ni capture cross section have been heavily affected by the neutron background , that was in the past inadequately suppressed or accounted for . at n_tof the neutron background
was accurately determined by means of dedicated simulations benchmarked against the available measurements @xcite , and was subtracted from the @xmath4ni data @xcite .
the aim of this paper is to demonstrate that deriving the neutron background from the neutron sensitivity ( sections [ sensitivity ] and [ resonance ] ) or even the dedicated measurements ( section [ natcarbon ] ) is not a trivial issue and requires a suitable procedure .
we address the issue by developing an improved method for determining the neutron background , which is based on an advanced treatment of the simulated neutron sensitivity ( section [ novel ] ) .
the improvements regard both the event tracking in the simulations and the subsequent data analysis .
in particular , we propose to study the neutron sensitivity by keeping track of the total time delays between the neutron scattering off the sample and the detection of counts caused by the neutron - induced reactions .
the limitations of the method are addressed in section [ uranium ] .
section [ summary ] summarizes the results and conclusions of this work . a detailed mathematical formalism underlying the proposed method is reported throughout the appendices , , and .
when comparing the neutron background to the neutron sensitivity , a clear distinction has to be made concerning the neutron energies .
the _ primary neutron energy _ is the true energy of the neutron ( from the incident neutron beam ) that has caused the reaction or the chain of reactions leading to the neutron background .
the _ reconstructed energy _ is the energy determined from the total time delay between the neutron production and the detection of secondary particles generated by the neutron - induced reactions . in case of the _ prompt counts
_ , caused by the reaction products immediately produced in the sample ( e.g. @xmath0 rays from neutron capture ) , the reconstructed energy is equal to the neutron kinetic energy , due to the total time delay being equal to the neutron time - of - flight . in case of neutron scattering inside the experimental hall or some other delay mechanism , such as the decay of radioactive products created by neutron - induced reactions , the total time delay may be large and may significantly affect the reconstructed neutron energy . for these _ delayed counts _ contributing to the neutron background , the reconstructed energy will be lower than the primary neutron energy , often by orders of magnitude . while the reconstructed energy is experimentally accessible , the primary neutron energy is not , and can only be determined by simulations .
the neutron background estimation methods laid out in sections [ sensitivity ] , [ resonance ] and [ natcarbon ] neglect the difference between the primary neutron energy @xmath7 ( before the scattering ) , the scattering energy @xmath8 ( sampled in the simulations ) and the reconstructed energy @xmath9 .
hence , throughout these sections the notation @xmath8 will be used as the universal one for the neutron energy .
we follow this approach for consistency with refs .
@xcite , freely combining the considerations strictly valid either for @xmath7 , @xmath8 or @xmath9 . starting from section [ novel ] , these distinctions will be explicitly taken into account . in that
, it should be noted that the neutron sensitivity has conventionally been expressed in terms of the scattering neutron energy @xcite . on the other hand ,
the neutron background as appearing in the experiments is a function of the reconstructed energy , suggesting at once an incompatibility between the two . in order to calculate the neutron background from the neutron sensitivity
, one needs to determine the neutron detection efficiency @xmath10 , i.e. the efficiency for detecting a neutron through the detection of particles produced in secondary neutron reactions .
this is commonly achieved by running the dedicated simulations , wherein the neutrons are isotropically and isolethargically generated from a point source at the sample position .
we note that the pulse height weighting technique @xcite has to be applied in calculating the efficiency , in order to compensate for the lack of correlations between @xmath0 rays in the simulated cascades following neutron captures .
this issue has already been addressed in ref .
furthermore , the central role of applying the pulse height weighting technique to the simulated capture data was unambiguously confirmed in ref .
@xcite by comparing the simulated and the experimental capture data for @xmath11au .
a detailed description of the pulse height weighting technique applied at n_tof may be found in ref .
@xcite .
we adopt the definition of the neutron sensitivity from , which uses the ratio @xmath12 , taking into account the maximum detection efficiency @xmath13 as an additional constant factor . in order to be able to use the weighted neutron detection efficiency @xmath14 ,
we further generalize the definition of the neutron sensitivity @xmath15 , by introducing the average weighting factor @xmath16 : @xmath17 the weighted neutron detection efficiency @xmath14 is : @xmath18 with @xmath19 as the appropriate weighting factors from the pulse height weighting technique , dependent on the energy @xmath20 deposited in detectors .
@xmath21 is the number of detected counts caused by neutrons of scattering neutron energy @xmath8 , while @xmath22 is the total number of neutrons simulated at this energy .
the average weighting factor @xmath16 is obtained by taking into account all neutron energies sampled : @xmath23 it may be noted that without weighting ( @xmath24 for all counts ) the generalized neutron sensitivity reverts to the original @xmath12 ratio . the weighted efficiency @xmath14 has been calculated for two c@xmath2d@xmath2 detectors used at n_tof .
one is the modified version of a commercial bicron detector and the other one was custom built at forschungszentrum karlsruhe and denoted as fzk detector @xcite . for the sake of simplicity and clarity , in this paper we will only show the results for the bicron detector , with the condition @xmath25 kev , as usually imposed on the experimental data .
furthermore , the reader s attention may be drawn to noticeable fluctuations apparent in multiple figures presented throughout this paper . with the exception of clearly recognizable resonances in the displayed spectra , the fluctuations are purely statistical in nature a simple consequence of a finite runtime dedicated to the computationally intensive simulations .
they are also naturally enchanted by the application of the pulse height weighting technique and by a fine binning that was selected for displaying the data , in order to preserve the clear appearance of some of the very narrow resonances . in accordance with the laid out considerations , fig .
[ fig1 ] shows the weighted neutron detection efficiency of the bicron detector .
the generalized neutron sensitivity , offset by a constant factor @xmath26 , is also shown , since it will be used later on . to estimate the neutron background in terms of the weighted counts per neutron bunch , the yield @xmath27 of elastically scattered neutrons ( i.e. the scattering probability ) and the neutron flux @xmath28 ( normalized to the number of neutrons per neutron bunch ) have to be taken into account .
for purely illustrative purposes , we will assume a simple relation for the yield : @xmath29 where @xmath30 is the areal density of the sample in number of atoms per unit surface .
@xmath31 and @xmath32 denote the elastic scattering cross section and the total cross section , respectively
. this expression does not take into account the multiple scattering effects , the angular distribution of scattered neutrons nor the contribution from inelastic reactions , which may all be accounted for in dedicated simulations , as in ref .
the weighted neutron background @xmath33 may be expressed as : @xmath34 this estimate is compared in fig .
[ fig2 ] against the true neutron background for @xmath4ni , obtained from dedicated simulations @xcite . outside the resonance region the scale of the true background is , indeed , very well reproduced by the estimated one . however , under the capture resonances the background calculated from the neutron detection efficiency is clearly overestimated .
this is precisely because the estimated background is expressed in terms of the primary neutron energy instead of the reconstructed energy , thus missing the time delays following the neutron scattering off the sample . since the background overestimation seems to be more pronounced for strong resonances , subtracting the neutron background estimated from the neutron detection efficiency may significantly affect the measured capture resonances
this is particularly troublesome because the strongest capture resonances constitute the dominant contribution to the maxwellian averaged cross sections ( macs ) , which represent the basic input for astrophysical models of stellar nucleosynthesis .
ni and the one estimated from the weighted neutron detection efficiency of the bicron detector . ]
as is shown in fig .
[ fig2 ] , the neutron background estimated from the neutron sensitivity of the experimental setup exhibits strong resonances due to the elastic scattering cross section of a given sample .
since each capture resonance is accompanied by the corresponding resonant component in the elastic scattering cross section , an erroneous estimate of the neutron background may heavily affect the capture data . here
we discuss the correct method of estimating the neutron background , focusing on the illustrative example of the strong 15.3 kev resonance in the @xmath4ni(@xmath35 ) reaction .
the first estimate of the neutron background under the capture resonances relies on simple neutron sensitivity considerations . to demonstrate the limitations of this method
, we benchmark it against the true neutron background under the 15.3 kev resonance ( obtained by dedicated simulations ; shown in fig . [ fig2 ] ) .
we assume that the capture yield @xmath36 under the resonance is determined by its radiative width @xmath37 : @xmath38 .
the reaction yield is translated into the number of the detected capture counts @xmath39 through the average detection efficiency @xmath40 as : @xmath41 . by the same reasoning ,
the yield of elastically scattered neutrons @xmath42 is determined by the neutron width @xmath43 as : @xmath44 .
the neutron background counts @xmath45 are similarly affected by the normalized weighted neutron detection efficiency @xmath46 , so that : @xmath47 . in order to establish the link with the adopted definition of the generalized neutron sensitivity @xmath48 , we replace the average detection efficiency @xmath40 by the maximum one @xmath13 .
the relative contribution of the neutron background to the total number of counts measured at the given resonance may then be estimated as : @xmath49 for the @xmath4ni resonance at 15.3 kev the values from endf / b - vii.1 @xcite are : @xmath37 = 1.104 ev and @xmath43 = 1354.062 ev . with the neutron sensitivity for the bicron detector of @xmath50 , determined by averaging the data from fig .
[ fig1 ] within the resonance range from 14 kev to 17 kev , the ratio from eq .
( [ eq1 ] ) amounts to approximately 68% . on the contrary ,
dedicated simulations of the neutron background indicate that when expressed in terms of the primary ( note : _ not _ reconstructed ! ) neutron energy , the neutron background between 14 kev and 17 kev amounts only to 24% of the total detected counts .
the difference relative to the result of eq .
( [ eq1 ] ) is due to the following reasons : ( 1 ) in eq .
( [ eq1 ] ) the maximum detection efficiency was used instead of the average one ( which , in fact , lowers the estimated value ) ; ( 2 ) in small part because the neutron sensitivity was calculated assuming an isotropic distribution of scattered neutrons ( isotropic in laboratory frame ) , instead of a more realistic one ; ( 3 ) most importantly , the elastic scattering cross section has a different shape than the neutron capture cross section , showing pronounced interference patterns and more extended resonance tails , strongly affecting not only the partial contribution to the yield of scattered neutrons within the limited energy range between 14 kev to 17 kev , but also the proportionality between the overall scattering yield and the neutron width , relative to the capture counterparts ( @xmath51 ) .
these simple considerations already indicate that a simplified approach , such as the one of eq .
( [ eq1 ] ) , may lead to large errors in the estimates of the neutron background .
ni within the 14 kev 100 kev range , compared to the spectrum of primary energies for the neutrons that have caused it . ]
a very different value for the relative contribution of the neutron background is obtained if we consider the reconstructed neutron energy .
the true neutron background , integrated between 14 kev and 17 kev , makes only 6.2% of the total resonance area .
moreover , this reduced value consists of both the prompt and the delayed component , while the neutron sensitivity considerations apply only to the prompt component .
( the delayed component was defined by considering events with more than 1% relative difference between the primary neutron energy @xmath7 and the reconstructed energy @xmath9 , i.e. @xmath52 . increasing this relative difference to 10%
did not produce any notable changes between the two components . ) if treated separately , the delayed component , which can not be estimated on the basis of simple neutron sensitivity considerations , contributes 2.6% to the 15.3 kev resonance , while the prompt component amounts to only 3.6% , in contrast with the initial value of 24% from the neutron background expressed as a function of the primary neutron energy .
ni caused by neutrons with primary energies between 14 kev and 100 kev . ] in order to understand the difference between the value of 24% from the primary neutron energies and the value of 6.2% from the reconstructed energies , we focus on the origin of the neutron background within the 14 kev 100 kev range , where two strong capture resonances ( 15.3 kev and 63.3 kev ) are located .
figure [ fig5 ] compares the total neutron background in that region with the energy spectrum of primary neutrons that have generated it .
neutrons all the way up to 10 mev contribute notably to the background below 100 kev , while the contribution from neutrons of higher energy is negligible .
on the other hand , fig .
[ fig6 ] shows the background produced by neutrons of primary energies between 14 kev and 100 kev . the comparison of figs .
[ fig5 ] and [ fig6 ] shows that a large fraction of the background in the 14 kev 100 kev range is produced by higher - energy neutrons .
simultaneously , the neutrons of primary energy from the considered range produce a background mostly contained at lower reconstructed energies , down to 0.1 ev .
only a small fraction of the background remains in the same energy range ( in particular , only 15% , as the ratio between previously quoted values of 3.6% and 24% ) . from both figures , together with fig .
[ fig2 ] , we reach the following conclusion : contrary to the past assumption , not even the prompt component of the neutron background under the capture resonances may be safely estimated from neutron sensitivity considerations alone . instead , complete simulations of the neutron propagation throughout the experimental hall
must be performed , taking the full temporal evolution of the neutron - induced reactions into account .
another common method used for estimating the neutron background without relying on simulations , except for possible higher order corrections consists in scaling the neutron background measured with a neutron scatterer , such as a sample of natural carbon .
this method is based on the assumption that the measured background is proportional to the yield of elastically scattered neutrons for a given sample : @xmath53 , with the constant of proportionality assumed to be equal for both the scatterer and the sample under investigation . under such assumptions
the weighted background related to a given sample ( we use the example background @xmath54 related to the @xmath4ni sample ) may be estimated from the weighted background @xmath55 , measured with the carbon sample , as : @xmath56 where @xmath57 and @xmath58 are the yields of elastically scattered neutrons for @xmath4ni and @xmath59c , respectively .
ni and the one estimated from the weighted neutron background of @xmath59c sample . ]
figure [ fig8 ] compares the actual neutron background for @xmath4ni with the one estimated by eq .
( [ carbon ] ) , using the simulated neutron background for @xmath59c . under strong resonances
the neutron background is again overestimated , for the same reasons that were covered in section [ resonance ] .
however , a startling agreement may be observed outside the resonant region of @xmath4ni , which may be surprising at first , since the principle of scaling the backgrounds for different samples by the portion of elastically scattered neutrons ( at a given primary neutron energy ) can only be applied to the prompt components .
though , in principle , an excellent agreement outside the resonant region may not be _ a priori _ expected , it may be understood from the fact that the sources of the neutron background other than the sample itself ( i.e. the experimental components and the walls of the experimental area ) are equal for both samples . if figs . [ fig2 ] and [ fig8 ] have shown anything , it is that in estimating the neutron background directly from the elastic scattering cross section , it is by far more appropriate to use the smoothed modification of the cross section , in which the ( strong ) scattering resonances have been flattened and replaced by the smooth sections connecting to the cross section at surrounding energies . unfortunately , fig .
[ fig8 ] may betray an overly optimistic picture , since for estimating the neutron background of @xmath4ni , the simulated neutron background of @xmath59c was used , instead of the experimental one .
the reason is that the measured carbon data from experimental area 1 of the n_tof facility reveal the pure neutron background only at the reconstructed energies above 1 kev @xcite . below this energy
the measurements are strongly affected by the detection of @xmath60 rays from the decay of radioactive @xmath5b residuals produced by the inelastic @xmath5c(@xmath6)@xmath5b reaction @xcite , opening above the reaction threshold of 13.6 mev .
this is an intrinsic feature of @xmath5c itself and can not be projected to the other samples .
for this reason , in experimental area 1 the neutron background for @xmath4ni ( or any other sample ) can not be estimated in the energy range below 1 kev directly from the measured carbon data . at the same time ,
any reliable extrapolation toward lower energies is extremely hard , especially since the shape of the background below 10 ev decidedly departs from the one set above 1 kev .
the considerations up to this point reveal the limitations of the currently available methods for determining the neutron background .
herein , we propose an improved approach to estimating the neutron background , based on simulations of the complete experimental setup , barring the sample itself .
these simulations need to be performed only once for a given experimental setup , with the output results reflecting the detector response to the scattered neutrons
being adjusted to the scattering properties of a particular sample by means of a proper normalization technique .
the simulations are basically identical to those used for the neutron sensitivity in section [ sensitivity ] and are described in ref .
@xcite . in short ,
the overall experimental setup is irradiated by neutrons isolethargically and isotropically emitted from a point source at the sample position .
note that the neutrons are emitted as if they have been already scattered off the sample .
for this reason , we will treat them as the _ scattered neutrons _
, as opposed to the _ primary neutrons _ from the neutron beam .
the distinction and its importance are elaborated in .
the sampled neutron energies span the full energy range of the n_tof neutron beam : from thermal up to 10 gev .
each count detected by any of the two c@xmath2d@xmath2 detectors is characterized by the following set of parameters extracted from simulations : @xmath61 . here
@xmath8 is the initial neutron energy , @xmath20 is the energy deposited in the detector , @xmath62 is the time delay between the neutron emission from the sample position and the detection of secondary particles , while @xmath63 and @xmath64 are conventionally defined polar and azimuthal angles of the initial neutron emission , relative to the primary beam direction .
the most important feature of the improved approach is that the neutron background is calculated from the time delays @xmath62 , and expressed as a function of the reconstructed energy @xmath9 , instead of the initially sampled energy @xmath8 .
in addition , a more involved data analysis will be included in the procedure .
figure [ fig11 ] shows examples of the detector response to neutrons isotropically scattered in the laboratory frame , for several selected scattering energies @xmath8 , as functions of the time delay @xmath62 .
the detector response function @xmath65 and its relation to the neutron background are explained in detail in .
[ fig11 ] shows the angle - integrated response @xmath66 .
the counts at very large time delays are caused by the detection of particles ( mostly @xmath60 rays ) emitted in the decay of long - lived radionuclides produced by neutron activation . only counts with a time delay of up to @xmath3100
ms contribute significantly to the neutron background affecting the n_tof measurements ( since @xmath3100 ms is the width of the data acquisition window at n_tof facility ; the exact vale depends on the adopted sampling rate of the data acquisition system ) .
the later counts may only contribute ( with sharply decreasing probability ) through the wrap - around process in subsequent neutron pulses .
the backbone of the new method is the proper normalization of the simulated data , which consists in weighting each detected count by the appropriate weighting factors @xmath67 , dependent on more parameters than just the reconstructed energy @xmath9 , which appears as the main argument of the neutron background .
since the cosine of the scattering angle plays a dominant role , in order to abbreviate relevant expressions we use the following notation : @xmath68 together with an abbreviation for the set of all relevant parameters : @xmath69 we report here the central expression for the weighting factors , while the complete mathematical formalism underlying the method is treated in detail in a series of appendices ( , , ) .
in particular , an overview of considerations leading to the correct expression for the weighting factors @xmath67 is presented in , with the detailed derivation given in . adopting the notation from eqs .
( [ short_cos ] ) and ( [ star ] ) , the expression for the weighting factors may be written as : @xmath70 here @xmath71 is the weighting function from the pulse height weighting technique .
@xmath72 is the yield of elastically scattered neutrons , given by eq .
( [ yield ] ) , but dependent on the _ primary neutron energy
_ @xmath7 from before the scattering off the sample .
the primary neutron energy @xmath7 is given as a function of the _ neutron scattering energy _ @xmath8 and the scattering angle @xmath73 by eq .
( [ primary ] ) in . the neutron flux @xmath74 is given in units of lethargy .
the angular correction factor @xmath75 given by eq .
( [ eq3 ] ) in translates the isotropically simulated angular distribution @xmath76 of the scattered neutrons into the more realistic laboratory distribution @xmath77 : @xmath78 which is the relativistically transformed angular distribution of neutrons isotropically scattered in the neutron - nucleus center of mass frame . finally , @xmath79 is the total number of simulated neutrons , with @xmath80 and @xmath81 as the minimum and maximum sampling energy , respectively .
we remark that the derivative @xmath82 may be easily calculated from eq .
( [ primary ] ) .
ni ( top panel ) and @xmath59c ( bottom panel ) , compared with the background obtained by the newly proposed estimation method . the estimated background obtained by properly normalizing the simulated data , on the basis of eq .
( [ weighting ] ) , is compared against the one obtained by a simplified normalization based on eq .
( [ approx]).,title="fig : " ] ni ( top panel ) and @xmath59c ( bottom panel ) , compared with the background obtained by the newly proposed estimation method . the estimated background obtained by properly normalizing the simulated data , on the basis of eq .
( [ weighting ] ) , is compared against the one obtained by a simplified normalization based on eq .
( [ approx]).,title="fig : " ] since the weighting factors from eq .
( [ weighting ] ) are dependent on more parameters than appear as the arguments of the final distribution , the weighting procedure can not be directly applied to an overall distribution of unweighted counts .
rather , it must be applied to a set of discrete data , on a count - to - count basis . presents the simple formalism establishing the link between the continuous distributions and the associated set of discrete data .
figure [ fig10 ] shows the neutron background for @xmath4ni and @xmath59c , estimated using the new method , i.e. applying the weighting factors from eq .
( [ weighting ] ) . an excellent agreement with the neutron background determined from the dedicated simulations for the two samples
is observed in both cases . while eq .
( [ weighting ] ) represents the fully relativistic approach , reasonable results can also be obtained with a simplified approach in which the angular corrections and the difference in neutron energy before and after the scattering are ignored : @xmath83 in the reconstructed - energy range , the ratio @xmath84 between the average proper weighting factor @xmath85 from eq .
( [ weighting ] ) and the average simplified factor @xmath86 from eq .
( [ approx ] ) is very close to 1 for both @xmath4ni and @xmath59c .
this may be easily understood from the fact that the major contribution to these particular backgrounds comes from essentially nonrelativistic neutrons of energies below 10 mev , with only a minor contribution from higher energies @xcite .
however , while the differences between the proper and the simplified weights may average out , they may become significant on the level of single counts ( being more pronounced for lighter nuclei , since the boost from a neutron - nucleus center of mass frame into the laboratory frame depends on the mass of the target ) .
this is illustrated in fig .
[ fig12 ] for the counts from the reconstructed - energy range @xmath87 , for both @xmath4ni and @xmath59c .
evidently , the corrections are mostly limited to @xmath8810% in case of @xmath4ni and to @xmath8820% in case of @xmath59c , but both distributions show the long tails far beyond the central parts . from eq .
( [ weighting ] ) and the simplified weighting factors @xmath89 from eq .
( [ approx ] ) , for @xmath4ni and @xmath59c . ]
the advantage of the method proposed herein is the universality of the simulated data that need to be obtained only once for a particular experimental setup . estimating the neutron background for a given sample requires only the proper selection of the elastic scattering cross section and the introduction of an appropriate nuclear mass into eqs .
( [ primary ] ) and ( [ eq4 ] ) from [ considerations ] , used for evaluating the central eq .
( [ weighting ] ) .
in this section we present an additional complication that may affect the neutron background . the method proposed in section [ novel ] considers only the elastic neutron scattering off the sample . in case of an additional background component , related to the sample properties other than the elastic scattering
, there is no alternative to running the dedicated simulations specifically adapted to the particular sample .
this is discussed in the example of the neutron background in a measurement of the @xmath90u(@xmath35 ) cross section at n_tof @xcite .
figure [ fig9 ] shows the simulated neutron background for a @xmath90u sample , clearly separating a portion of the background caused exclusively by the neutron scattering off the sample .
the presence of a strong additional component is immediately evident . according to the simulations ,
this component is caused by high energy neutrons ( above 600 kev ) inducing fission reactions on @xmath90u .
a variety of radioactive fission fragments is produced in the process , most so short - lived that their decay falls within @xmath3100 ms data acquisition window of the n_tof facility .
the products of these decays ( mostly @xmath60 rays ) are then detected alongside the capture @xmath0 rays , contributing to the neutron background .
the detection of these secondary products is similar to the detection of @xmath60 rays from the @xmath5c(@xmath6)@xmath5b reaction on the carbon sample .
this background component , although related to the activation of the sample , can not be measured separately in beam - off runs , since the half - lives of the produced radioactive isotopes span just a few neutron bunches .
the sudden increase in the total neutron background above several hundreds of kev is due to the prompt @xmath0 rays released by neutron inelastic scattering and neutron - induced fission .
u , compared against the background caused exclusively by neutrons elastically scattered off the sample .
two background estimates are also shown : one obtained from the neutron detection efficiency of the experimental setup , the other one by the improved estimation technique ( see the main text for details ) . ]
figure [ fig9 ] also shows the neutron background estimated by the older technique represented by eq .
( [ estimate ] ) comparing it against the background obtained by the new technique from section [ novel ] .
the superiority of the new technique in reconstructing the portion of the background caused by the neutron scattering is clearly evident .
note that in the energy range above several hundred kev the angular distribution of elastically scattered neutrons from @xmath90u departs from isotropy @xcite . while the dedicated simulations properly account for the angular distribution of the scattered neutrons , the new technique at its basic level explicitly assumes isotropic scattering in the neutron - nucleus center of mass frame , leading to a more pronounced discrepancy between the true and the estimated background
however , the correction for the realistic angular distribution may be introduced directly through the angular correction factor @xmath75 . assuming the angular distribution @xmath91 of neutrons scattered in the center of mass frame to be known and normalized to unity ( with @xmath92 as the neutron scattering angle in the neutron - nucleus center of mass frame ) ,
the angular correction factor simply becomes : @xmath93 thus accommodating even the most general form of the angular distribution .
of course , @xmath94 needs to be treated here as a function of @xmath73 , which is achieved directly by applying eq .
( [ eq4 ] ) from [ considerations ] . in conclusion ,
the neutron background for @xmath90u is dominated by an additional component caused by the radioactive decay of short - lived fission products ( together with the fission neutrons further enhancing the neutron background ) .
however , it should be noted that the gravity of this issue is strongly related to the length of the particular neutron flight - path ( @xmath3185 m for experimental area 1 at n_tof ) , which determines the time - energy correlation @xmath95 .
thus , the only way to correctly estimate the background is by means of dedicated simulations , taking into account the full framework of the known neutron reactions induced both within and without the sample , and closely following their complete temporal evolution . nevertheless , the newly proposed method from section [ novel ] may still be used for a fast and reliable estimation of the neutron background originating exclusively from elastic neutron scattering off the sample .
following ref .
@xcite , we have performed a close investigation of the relationship between the neutron background in neutron capture measurements and the neutron sensitivity related to the experimental setup .
the neutron background for a @xmath4ni sample was used in order to illustrate the difficulties in estimating the neutron background from neutron sensitivity considerations , i.e. from the neutron detection efficiency related to the experimental setup . as opposed to the neutron sensitivity being a function of the primary neutron energy ,
the neutron background is a function of the reconstructed neutron energy , which is affected by the temporal evolution of the neutron - induced reactions . as a consequence
, the neutron background may be overestimated under the capture resonances when the estimation is attempted on a basis of the neutron sensitivity , or even on a basis of the surrogate measurements with the neutron scatterer of natural carbon .
the reason is that correlations between the primary neutron energy and the reconstructed energies are neglected in considerations based only on the neutron sensitivity . outside the resonance region
a good agreement is found between the true neutron background and the background estimated from the neutron detection efficiency related to the experimental setup , and from the neutron background for natural carbon ( we remind that in the experimental area 1 of the n_tof facility the pure neutron background for the carbon sample is experimentally available only above 1 kev ) .
therefore , these methods may still be used , provided that a smoothed elastic scattering cross section is used , instead of the resonant one .
an improved neutron background estimation technique was presented , relying on the calculation of the neutron sensitivity as a function of the reconstructed neutron energy .
supplemented by an advanced data analysis procedure , taking into account the fully relativistic kinematics of the neutron scattering , the proposed procedure yields excellent agreement between the true and the estimated neutron backgrounds for @xmath4ni and @xmath59c .
the above considerations apply only to the background caused by elastically scattered neutrons . in the presence of reactions leading to the production of short - lived radioactive nuclides within the sample itself
, an additional background component may be present .
this has been illustrated in the case of the @xmath90u sample , for which neutron - induced fission at primary neutron energies above 600 kev translates into a strong background component at reconstructed neutron energies in the thermal and epithermal region , through the detection of radioactive residuals produced by fission .
the same case can also be made for the carbon sample measured in the experimental area 1 of the n_tof facility , if one considers the @xmath5c(@xmath6)@xmath5b reaction as interfering with the neutron background measurements . in all these cases particularly , in measurements of capture cross sections of actinides the only way of properly estimating the total neutron - induced background is by means of detailed monte carlo simulations .
+ * acknowledgements * + this work was supported by the croatian science foundation under project no .
the geant4 simulations have been run at the laboratory for advanced computing , faculty of science , university of zagreb .
let us suppose we have the function @xmath96 of the detector response to _ primary _ neutrons of energy @xmath7 , scattered by an angle @xmath73 ( we adopt the notation from eq .
( [ short_cos ] ) for the cosine of the scattering angle ) , @xmath96 being a distribution of time delays @xmath62 between the neutron scattering off the sample ( at @xmath97 ) and the detection of counts caused by reactions of the scattered neutrons . in parallel ,
let us consider the function @xmath65 of the detector response to neutrons _ scattered _ with the energy @xmath8 .
evidently , the following must hold : @xmath98 since the effect must be the same whether we regard the primary neutron of energy @xmath7 as the ultimate source of the detected counts , or the same neutron after the scattering by the angle @xmath73 , with the associated scattering energy @xmath99 .
let us now consider the contribution @xmath100 to the detected counts caused by the primary neutrons of energy @xmath7 which were scattered by the angle @xmath73 , where the counts themselves are detected at time @xmath101 after the primary neutron production ( at @xmath102 ) .
the total time delay @xmath101 is given by the neutron time - of - flight @xmath103 along the flight - path between the neutron source ( in particular , the n_tof spallation target ) and the irradiated sample , and the time delay @xmath62 between the neutron scattering off the sample and detecting the count : @xmath104 the detector response @xmath105 to the neutrons characterized by @xmath7 and @xmath73 determines the number of the detected counts as : @xmath106 with @xmath74 as the neutron flux , @xmath72 as the yield of the sample - scattered neutrons , given by eq .
( [ yield ] ) , and @xmath107 as the angular distribution of the neutrons scattered in the laboratory frame , such as the one from eq .
( [ ang ] ) .
( throughout the rest of the paper the neutron flux @xmath74 is given in units of lethargy , requiring the transition @xmath108 . )
the detector response function is extracted from the dedicated simulations . however , when the neutrons are simulated not as coming from the primary beam , but as already having been scattered off the sample which is a backbone of the improved method from section [ novel ] then the simulations yield the detector response function @xmath65 , instead of @xmath96 . fortunately , the relation from eq . ( [ relation ] ) overcomes this difficulty , allowing to rewrite eq .
( [ counts ] ) and to express the neutron background @xmath109 ( untreated by the pulse height weighting technique ) as the function of the total time delay @xmath101 : @xmath110 evidently , attempting to follow this procedure would be a formidable computational task , requiring the identification of the detector response @xmath65 as a function of no less than three variables , with a sufficient statistical accuracy
. however , the data analysis procedure laid down in section [ novel ] relying on the proper weighting of the individual counts
circumvents this problem by immediately building the integral from eq .
( [ background ] ) , instead of first requiring the extraction of the detailed multidimensional detector response function .
the ultimate confirmation of this claim comes in form of eq .
( [ ultimate ] ) from .
in the simulations the neutrons are treated as if already having been scattered off the sample .
though they are simulated isotropically in the laboratory frame , it is more justifiable to assume that scattering is isotropic in the center of mass frame of the primary neutron and the target nucleus .
furthermore , if the primary neutron beam was assumed to be isolethargic , the energy distribution of the scattered neutrons would not be such .
therefore , the detected counts caused by an over - idealized stream of scattered neutrons need to be weighted in a manner which will make them appear as if they were caused by an isolethargic beam of primary neutrons , which have been scattered isotropically in the neutron - nucleus center of mass frame . after this correction
, the isolethargic distribution may be reliably corrected by the actual energy dependence of the neutron flux , applying the correction to the energy distribution of the primary neutron beam , rather than to the distribution of scattered neutrons . to apply this correction properly , we will have to treat the primary neutron flux as a function of the _ primary energy _
@xmath7 that the neutron must have had in the primary beam in order to be scattered by the scattering angle @xmath63 and with the _ scattering energy _ @xmath8 .
we remind that @xmath63 and @xmath8 have been independently sampled in the simulations : the scattering angle isotropically , the scattering energy isolethargically . adopting the notation from eq .
( [ short_cos ] ) and employing the relativistic scattering kinematics , the expression for the primary energy @xmath7 may be obtained : @xmath111 ^ 2}\times\\ & \bigg\{(m+m)\big[e_n-(m - m)c^2\big]+\\ & \big(e_n+2mc^2\big){\,\xi\,}\big[\sqrt{m^2-m^2\big(1-{\,\xi\,}^2\big)}-m{\,\xi\,}\big]\bigg\ } \end{split}\end{aligned}\ ] ] with @xmath112 as the mass of the neutron , @xmath113 as the mass of the scattering nucleus and @xmath114 as the speed of light in vacuum .
the isotropically simulated angular distribution of the scattered neutrons can be translated into the more realistic laboratory distribution ( which is the relativistically transformed distribution of the neutrons isotropically scattered in the neutron - nucleus center of mass frame ) by means of the angular correction factor @xmath75 , as used in eq .
( [ ang ] ) : @xmath115 here @xmath92 is the neutron scattering angle in the center of mass frame , relative to the initial beam direction . the term @xmath94 is found by employing the relativistic scattering kinematics : @xmath116 the terms required for the evaluation of eq .
( [ eq4 ] ) are given by : @xmath117 @xmath118\ ] ] @xmath119 here @xmath120 is a conventional relativistic term for the center of mass speed in the laboratory frame ( , with @xmath121 as the actual speed ) .
additionally , @xmath122 and @xmath123 are the momentum and total energy of the scattered neutron in the laboratory frame .
though by plugging eqs .
( [ eq4])([eq7 ] ) into eq .
( [ eq3 ] ) , a correction factor @xmath75 may be analytically calculated , the exact expression is long and tedious .
therefore , the derivative from eq .
( [ eq3 ] ) is best calculated numerically .
furthermore , the reader may note that the scattered neutron energy @xmath123 from eq .
( [ eq7 ] ) should correspond to @xmath124 , where the kinetic energy @xmath8 of the scattered neutron has been directly sampled in the simulations .
however , the correction factor @xmath75 must be calculated for a fixed primary energy @xmath7 , instead of @xmath8 .
hence , during the calculation of the correction factor @xmath75 , the momentum @xmath122 from eq .
( [ eq6 ] ) and energy @xmath123 from eq .
( [ eq7 ] ) must be treated as functions of @xmath7 and @xmath73 , and have to be varied accordingly . for brevity of expressions
, we adopt the notation @xmath125 from eq .
( [ star ] ) for the set of all relevant parameters .
the first step in determining the weighting factors @xmath67 , dependent on any combination of these parameters , is the normalization of the data by the total number @xmath126 of neutrons simulated with scattering energy @xmath8 and scattering angle @xmath127 . to obtain the contribution to the detected counts per single neutron bunch
, the statistical effect of a single scattered neutron isolated by the previous normalization must be amplified by the number @xmath128 of primary neutrons ( those from the primary beam ) that can be scattered by the angle @xmath127 , with the energy @xmath8 .
furthermore , it is necessary to account for the probability that the primary neutron of energy @xmath7 will indeed be scattered .
this probability may be expressed as the yield @xmath72 of elastically scattered neutrons from eq .
( [ yield ] ) .
finally , since we are applying the pulse height weighting technique , the most evident weighting factor is given by the weighting function @xmath71 , dependent on the energy @xmath20 deposited in the detector .
combined , the previous considerations give rise to the total weighting factor @xmath67 that has to be applied to each detected count : @xmath129 since the scattered neutrons have been simulated isolethargically and isotropically ( in the laboratory frame ) , the term @xmath126 is simply determined as : @xmath130 where @xmath79 is the total number of simulated neutrons , with @xmath80 and @xmath81 as the minimum and maximum sampling energies , respectively .
the angular distribution of neutrons scattered from the primary beam under the assumption of isotropic scattering in the neutron - nucleus center of mass frame is already known from eq .
( [ ang ] ) . with the neutron neutron flux
@xmath74 conveniently given in units of lethargy , the term @xmath128 is also easily expressed : @xmath131 combined , eqs .
( [ term1 ] ) and ( [ term2 ] ) lead to : @xmath132 since the primary neutron energy @xmath7 may be calculated for any combination of the scattering energy and angle ( up to the possible kinematic limitations , dependent on the projectile and target masses ) , @xmath8 and @xmath73 are independent , making the derivative from eq .
( [ ratio ] ) a partial one : @xmath133 finally , plugging eqs .
( [ ratio ] ) and ( [ der ] ) back into eq .
( [ w_start ] ) yields the full expression for the overall weighting factor : @xmath134 we remind that the derivative @xmath82 may be easily calculated from eq .
( [ primary ] ) .
we present a convenient formalism establishing the link between discrete data and continuous distributions that are built from these data .
this formalism is especially useful when each datum must be applied its own weighting factors , dependent on more parameters or other parameters than those that appear as the arguments of the final distribution .
we demonstrate the formalism by immediately applying it to the estimated neutron background from section [ novel ] . for this purpose ,
let us consider the contribution @xmath135 to the detected counts weighted by the weighting factors @xmath67 , dependent on any combination of the physical parameters .
as before , we adopt the notation from eqs .
( [ short_cos ] ) and ( [ star ] ) . within the formalism of continuous distributions ,
the contribution @xmath135 to the detected counts from an element of the parameter space defined by @xmath136 , @xmath137 , @xmath138 and @xmath139 , may be written as : with @xmath142 as the elementary contribution to the unweighted counts . on the other hand , when building the distribution from discrete data , the contribution @xmath135 appears as a union of all appropriately weighted counts @xmath143 from an associated element of the parameter space : with @xmath145 as the total number of detected counts from the parameter space volume @xmath146 . in eqs .
( [ rep1 ] ) and ( [ rep2 ] ) @xmath9 holds a prominent place because we are expressing the final distribution ( the neutron background from section [ novel ] ) as a function of the reconstructed energy . combining eqs .
( [ rep1 ] ) and ( [ rep2 ] ) leads to a direct correspondence between the set of discrete data and their representation by the continuous distribution : 00 p. e. koehler , r. r. winters , k. h. guber , et al .
c 62 ( 2000 ) 055803 .
r. plag , m. heil , f. kppeler , et .
al . , nucl .
instr . and meth . a 496 ( 2003 ) 425 .
c. guerrero , a. tsinganis , e. berthoumieux , et al . , eur .
j. a 49 ( 2013 ) 27 . c. wei , e. chiaveri , s. girod , et al .
instr . and meth . a 799 ( 2015 ) 90 .
s. barros , i. bergstrm , v. vlachoudis and c. wei , j. instrum .
10 ( 2015 ) p09003 .
s. agostinelli , j. allison , k. amako , et al .
instr . and meth . a 506 ( 2003 ) 250 .
p. ugec , n. colonna , d. bosnar , et al .
instr . and meth . a 760 ( 2014 ) 57 .
p. ugec , m. barbagallo , n. colonna , et al .
c 89 ( 2014 ) 014605 .
p. ugec , n. colonna , d. bosnar , et al .
c 90 ( 2014 ) 021601(r ) .
k. h. guber , h. derrien , l. c. leal , et al .
, phys . rev .
c 82 ( 2010 ) 057601 .
f. gunsing , e. berthoumieux , g. aerts , et al .
c 85 ( 2012 ) 064601 .
r. l. macklin , j. h. gibbons , phys .
159 ( 1967 ) 1007 .
u. abbondanno , g. aerts , h. alvarez , et al .
instr . and meth . a 521 ( 2004 )
m. b. chadwick , m. herman , p. obloinsk , et al .
data sheets 112 ( 2011 ) 2887 . the n_tof collaboration , _ neutron capture cross section measurements of @xmath90u , @xmath151am and @xmath152am at n_tof _ , proposal to the isolde and neutron time - of - flight committee , cern - intc-2009 - 025 / intc - p-269 20/04/2009 .
f. mingrone , c. massimi , s. altstadt , et al .
data sheets 119 ( 2014 ) 18 . | the relation between the neutron background in neutron capture measurements and the neutron sensitivity related to the experimental setup is examined .
it is pointed out that a proper estimate of the neutron background may only be obtained by means of dedicated simulations taking into account the full framework of the neutron - induced reactions and their complete temporal evolution .
no other presently available method seems to provide reliable results , in particular under the capture resonances .
an improved neutron background estimation technique is proposed , the main improvement regarding the treatment of the neutron sensitivity , taking into account the temporal evolution of the neutron - induced reactions .
the technique is complemented by an advanced data analysis procedure based on relativistic kinematics of neutron scattering .
the analysis procedure allows for the calculation of the neutron background in capture measurements , without requiring the time - consuming simulations to be adapted to each particular sample .
a suggestion is made on how to improve the neutron background estimates if neutron background simulations are not available .
neutron sensitivity , neutron background , geant4 simulations , neutron time - of - flight , n_tof , neutron capture , neutron scattering |
triple gauge boson couplings constitute a sensitive probe of nonstandard interactions @xcite . in particular , triple
neutral gauge boson couplings ( tngbc ) @xmath6 , with @xmath7 , vanish when the three gauge bosons are on mass shell and bose symmetry and lorentz invariance are simultaneously satisfied @xcite .
eventhough these vertices do not reflect directly the non - abelian nature of the electroweak gauge group , since the yang - mills sector does not induce them at the tree or one - loop level , they constitute a window to new physics @xcite .
these vertices must be induced by loop effects in any renormalizable theory since they do not possess a renormalizable structure . at the one - loop level
, they can be generated only by fermionic triangles @xcite . in the standard model ( sm ) and the minimal supersymmetric standard model ( mssm )
they are rather small , of order @xmath8 @xcite , but new heavy quark generations may enhance them by about one order of magnitude @xcite .
experimental measurements of possible tngbc @xmath0 and @xmath1 have reached an accuracy at the few percent level at the tevatron @xcite , even better than the accuracy obtained for nonstandard deviations of the @xmath9 couplings @xcite .
persuasive arguments within the effective lagrangian approach indicate that tngbc are not expected to be larger than @xmath10 @xcite .
indirect limits on these vertices have been obtained from the muon @xmath11 value , the known bound on the electron electric dipole moment @xcite and the measurement of the rare decay mode @xmath12 obtained at lep @xcite .
the process @xmath13 has been studied also at the energies expected in future linear colliders in order to search for its sensitivity to the tngbc through the effects of @xmath14 polarization and initial state radiation @xcite , as well as possible cp - violating effects in the decay mode @xmath15 @xcite .
the process @xmath16 has been studied previously in connection to the determination of the number of light neutrino species @xcite and its possible electric and magnetic moments @xcite , as well as in the search of extra neutral gauge bosons @xcite . in the present paper
we study the sensitivity to the tngbc in the process @xmath17 at the next generation of linear colliders , namely the international linear collider ( ilc ) @xcite and the compact linear collider ( clic ) @xcite .
we expect to get better limits on the anomalous @xmath6 couplings through the @xmath18 channel because its larger branching ratio induces a cross section for @xmath16 about a factor 3 larger than the combined rates for @xmath19 @xcite .
the process @xmath20 is also competitive with respect to @xmath21 production at the lhc because the inclusive nlo qcd corrections to the @xmath21 cross section are quite large at high photon transverse momenta in the sm and reduce the sensitivity to the tngbc @xmath6 @xcite .
we will obtain in the present paper that the next generation of @xmath22 colliders may reach in fact sensitivities that are close to the sm predictions for the tngbc . in this respect ,
our study can be considered complementary to previous studies on the tngbc @xcite .
this paper is organized as follows : in sect .
ii we present the calculation of the cross section for the process with anomalous couplings @xmath0 and @xmath1 and , finally , we summarize our results and conclusions in sec .
the most general anomalous @xmath23 vertex function is given by @xcite @xmath24,\end{aligned}\ ] ] where @xmath25 is the @xmath14 boson mass and the respective @xmath1 general vertex function can be obtained from eq .
( 1 ) by the following replacements : @xmath26 the form factors @xmath27 are dimesionless functions of the @xmath28 momenta and in order to avoid violation of partial wave unitarity , generalized dipole form factors will be considered @xcite @xmath29 we will assume that the new physics scale @xmath30 is above the collision energy @xmath31 .
the feynman diagrams that contribute to the process @xmath32 are depicted in fig .
1 . we will consider only the effect due to the anomalous @xmath33 vertices for various ilc / clic energies and luminosities .
the sm contribution to this process would arise from initial state radiation coming from the @xmath22 beams and photons emitted from a virtual @xmath34 boson exchanged in the @xmath35-channel . in order to suppress the sm contribution we impose the cuts @xmath36 @xmath37 and @xmath38 , where @xmath39 is the polar angle of the photon with respect to the beam direction @xcite .
the expression for the respective cross section with anomalous couplings @xmath0 is given by @xmath40[(h^z_{10})^2f_1(s , e_\gamma , \cos\theta_\gamma)\nonumber\\ & + & ( h^z_{30})^2f_2(s , e_\gamma , \cos\theta_\gamma ) ] e_\gamma de_\gamma d\cos\theta_\gamma,\end{aligned}\ ] ] where @xmath41 and @xmath42 are the energy and scattering angle of the photon .
the kinematics is contained in the functions @xmath43 } { ( s - m^2_z)^2+m^2_z\gamma^2_z},\\ f_2(s , e_\gamma , \cos\theta_\gamma)&\equiv & \frac { [ -se^2_\gamma+2\sqrt{s}e^3_\gamma + \frac{3}{2 } se^2_\gamma\sin^2\theta_\gamma - \sqrt{s}e^3_\gamma\sin^2\theta_\gamma ] } { ( s - m^2_z)^2+m^2_z\gamma^2_z},\end{aligned}\ ] ] while the coefficient @xmath44 is given by @xmath45\equiv \frac{[1 - 4x_w+8x^2_w ] } { x^2_w(1-x_w)^2},\ ] ] where @xmath46 . in the case of the anomalous coupling @xmath1 , in order to get the respective expression for the cross section , we have to use the substitution given in eqs .
( 2 - 4 ) and take the appropriated photon propagator into account . we do not reproduce the analytical expressions here because they are rather similar to the terms given in eqs .
( 5)-(8 ) .
we expect that the contribution arising from the @xmath6 anomalous vertices is enhanced in the process @xmath2 at high energies with respect to the sm contribution @xcite , as it is shown in fig .
2 . in this section
we derive those values of @xmath47 , @xmath48 , which would give rise to a deviation from the sm prediction at the level of two standard deviations @xmath49 c.l . ) . as it is shown in eq .
( 5 ) , the couplings @xmath50 do not contribute to the total cross section @xcite and the calculation of the sensitivity bounds is facilited by the fact that the cp - conserving and cp - violating couplings @xmath51 do not interfere .
even more , cross sections and sensitivities are nearly identical for equal values of these couplings @xcite . for simplicity , we shall take then one of the vertex contributions at the time . in figs . 3 and 4 we present the dependence of the sensitivity limits of the @xmath52 vertices with respect to the collider luminosity for three different values of the c.m .
a similar dependence is obtained for the sensitivity limits of the @xmath53 vertices with respect to the collider luminosity . in figs .
5 and 6 we summarize the respective limit contours for these vertices at the @xmath54 c.l . in the @xmath55 plane for a luminosity of 500 @xmath3 and the c.m .
energies 0.5 - 1.5 @xmath4 .
finally , in tables 1 and 2 we include the @xmath54 c.l .
limits obtained for both vertices for two different values of the new physics scale energy @xmath56 and 5 @xmath4 and the collider luminosity of 500 @xmath3 . in conclusion
, we have found that the process @xmath60 at the high energies and luminosities expected at the ilc / clic colliders can be used to probe for nonstandard @xmath6 vertices . in particular
, we can appreciate that the @xmath54 c.l .
sensitivity limits expected for the @xmath51 vertices at 1 - 1.5 @xmath4 c.m .
energies already can provide tests of these vertices of order @xmath5 , one order of magnitude lower than the sm and mssm predictions @xcite .
these sensitivity limits are of the same order of magnitude that those expected in the process @xmath61 , also at future @xmath22 colliders , through the effects induced by the polarization of the @xmath14 gauge boson and initial state radiation @xcite .
j. m. hernndez _
et al . , phys .
rev . _ * d60 * , 013004 ( 1999 ) ; f. larios _ et al .
rev . _ * d63 * , 113014 ( 2001 ) ; m. a. prez , j. j. toscano , and g. tavares - velasco , _ int . j. mod
. phys . _ * a19 * , 159 ( 2004 ) .
e. ma and j. okada , _ phys .
_ * d8 * , 4219 ( 1978 ) ; m. carena , _ et al .
_ , _ phys .
_ * d68 * , 113007 ( 2003 ) ; a. gutirrez - rodrguez , _ et al .
_ , arxiv : 0806.2163 [ hep - ph ] , to be published in _ mod .
lett . _ * a*. | we study the sensitivity for testing the anomalous triple gauge couplings @xmath0 and @xmath1 via the process @xmath2 at high energy linear colliders . for integrated luminosities of 500 @xmath3 and center of mass energies between 0.5 and 1.5 @xmath4
, we find that this process can provide tests of the triple neutral gauge boson couplings of order @xmath5 , one order of magnitude lower than the standard model prediction . |
quasiparticles of the @xmath0-fqh state live on the quotient lattice @xmath16 , where the @xmath9-matrix is given in .
when @xmath2 , the theory has the same anyon content as the @xmath1 kitaev toric code and the quotient lattice decomposes into @xmath179 as all quasiparticles can be uniquely be written as @xmath64 where @xmath18 and @xmath19 .
when @xmath22 is odd , 2 is invertible in @xmath180 by @xmath70 and the transformation @xmath181 gives an isomorphism of the quotient lattice @xmath182 with inverse @xmath183 as @xmath184 mod @xmath9 .
thus @xmath29 where each quasiparticle has a unique decomposition @xmath23 where @xmath11 and @xmath25 are the charge and bilayer / spin vectors .
the fqh quasiparticle @xmath79-symbols are chosen to be multilinear homomorphisms @xmath185 .
( i ) @xmath186 for @xmath1 toric code when @xmath2 .
( ii ) @xmath187 for @xmath0-fqh states with @xmath22 odd . under this convention
, @xmath75-symbols can be chosen to be trivial one @xmath34 and the quasiparticle strings satisfy the crossing , bootstrap and bending rules . @xmath132{rxy1.pdf}}}=\vcenter{\hbox{\includegraphics[width=0.35in]{rxy2.pdf}}}r^{{\bf
a}{\bf b}}=\vcenter{\hbox{\includegraphics[width=0.35in]{rxy3.pdf}}}r^{{\bf a}{\bf b}}r^{{\bf b}{\bf a}},\quad\vcenter{\hbox{\includegraphics[width=0.45in]{f3.pdf}}}=\vcenter{\hbox{\includegraphics[width=0.45in]{f4.pdf}}},\quad\vcenter{\hbox{\includegraphics[width=0.35in]{t1.pdf}}}=\vcenter{\hbox{\includegraphics[width=0.35in]{t2.pdf}}}=\vcenter{\hbox{\includegraphics[width=0.35in]{t3.pdf}}}\end{aligned}\ ] ] the double loop operator @xmath188 carries an anyon @xmath189 twice around a twist defect as shown in fig .
we have the relations ( i ) @xmath190 , ( ii ) @xmath191 and ( iii ) @xmath192 , where @xmath193 is the eigenvalue for the bare defect ( @xmath194 ) .
@xmath195 $ ] satisfies ( i iii ) , where the constant @xmath196 is fixed by requiring that @xmath197 , so that @xmath198 is well - defined on the anyon lattice .
the result is eq .. basis transformation between different fusion trees with defect channels can be generated by a consistent set of fundamental @xmath75-moves . in table
[ tab : fsymbols ] , we list all the @xmath75-symbols for @xmath1 toric code and @xmath0-fqh states .
these @xmath75-symbols are calculated by gluing the splitting states defined in figure [ fig : splittingstates ] by joining quasiparticle strings .
here we illustrate the calculation of the @xmath75 symbol that relates @xmath199 in @xmath0-fqh states for @xmath22 odd . -matrix
@xmath200_{\bf a}^{\bf b}$ ] .
the sum over wilson loops @xmath201 ( blue ) is restricted to an arbitrary set of @xmath61 quasiparticles @xmath202 with distinct spin components @xmath203 as @xmath204 ( see figure [ fig : doubleloop ] ) .
the operator @xmath205 flips the branch cut configuration from joining @xmath206 to connecting @xmath207.,width=624 ] the splitting state for @xmath208 is represented by the first diagram and that for @xmath209 is at the last by setting @xmath210 .
the @xmath75-symbol is given by the product of all the crossing phases and the normalization factor @xmath211 .
the basis transformation between any two fusion trees can be achieved by a series of @xmath75-symbols , and the result is independent from the choice of moves .
this is ensured by the pentagon identity @xmath212 and the maclane s coherence theorem @xcite .
@xmath75-symbols listed in table [ tab : fsymbols ] can be shown to satisfy the @xmath213 identities .
the @xmath79-symbol for exchanging a pair of defects is evaluated by rotating the splitting space for twist defect in figure [ fig : splittingstates](c ) .
we illustrate this for the case when @xmath22 is odd .
is a product of crossing phases and a double wilson operator . | we describe the braiding statistics of topological twist defects in abelian bosonic bilayer @xmath0 fractional quantum hall ( fqh ) states , which reduce to the @xmath1 toric code when @xmath2 .
twist defects carry non - abelian fractional majorana - like characteristics .
we propose local statistical measurements that distinguish the fractional charge , or species , of a defect - quasiparticle composite .
degenerate ground states and basis transformations of a multi - defect system are characterized by a consistent set of fusion properties .
non - abelian unitary exchange operations are determined using half braids between defects , and projectively represent the sphere braid group in a closed system .
defect spin statistics are modified by equating exchange with @xmath3 rotation .
the braiding @xmath4 matrix is identified with a dehn twist ( instead of a @xmath5 rotation ) on a torus decorated with a non - trivial twofold branch cut , and represents the congruent subgroup @xmath6 of modular transformations .
exotic anyonic excitations @xcite , such as ising and fibonacci anyon , in non - abelian topological phases @xcite have promising implications in topological quantum computing @xcite due to their non - abelian braiding properties and ability to store quantum information non - locally .
more recently topological defects that carry fractional majorana - like characteristics are theoretically predicted to be present in abelian topological systems such as fractional topological insulator - ferrormagnet - superconductor heterostructures @xcite , and dislocations in abelian bilayer fractional quantum hall states @xcite .
they can be conceptually studied as twist defects in symmetry enchanced exact solvable spin / rotor models like the kitaev toric code @xcite , the @xmath1 wen plaquette model @xcite , the bombin - martin color code @xcite and its @xmath1 generalization @xcite .
twist defects are semiclassical point - like objects that violate an ungauged anyonic symmetry of the system by twisting the anyon label of a circling quasiparticle by the symmetry operation , such as the electric - magnetic duality of the toric code or bilayer symmetry in the @xmath0-fqh state . unlike quantum deconfined anyons in a topological phase , twist defects are not excitations of a quantum hamiltonian and there is no superposition between quantum states in different defect configurations .
the semiclassical distinction of defects from quantum anyons manifest in braiding behavior , and we demonstrate this by twofold twist defects in abelian @xmath0-fqh states @xcite with a @xmath7 bilayer or spin symmetry .
this includes the @xmath1 toric code as a special case for @xmath2 .
an abelian quantum hall state with bilayer symmetry is described by the effective topological field theory @xmath8 characterized by the symmetric integral coefficient @xmath9-matrix @xmath10 and the charge vector @xmath11 .
the theory is invariant under the twofold bilayer symmetry represented by the pauli matrix @xmath12 , and consists of bosonic fundamental charged particles when the diagonal entry @xmath13 is even .
it has a ground state degeneracy of @xmath14 on a torus .
the quasiparticles ( up to fundamental bosons ) are labeled by two dimensional integral vector @xmath15 on a periodic lattice @xmath16 , and fusion rules are set by lattice addition @xmath17\times[{\bf b}]=[{\bf a}+{\bf b}]$ ] modulo @xmath9 .
bilayer defects in this letter are demonstrated in one of the following two classes of quantum hall states to avoid algebraic complications , however we expect the results and arguments apply to the general case .
( i ) @xmath2 .
the theory has the identical anyon structure as the @xmath1 kitaev toric code , where quasiparticles are generated by @xmath18 and @xmath19 with braiding characterized by the exchange phases @xmath20 and @xmath21 .
( ii ) @xmath22 is odd .
quasiparticles @xmath23 are composition of the charge sector @xmath24 and neutral spin " sector @xmath25 with exchange phases @xmath26 , @xmath27 , and @xmath28 . in either of the two cases ,
the quasiparticle lattice splits into @xmath29 so that there are unique decomposition of anyons by @xmath30 or @xmath31 .
the general exchange phase @xmath32 can be chosen to be multilinear in its two entries so that the hexagon identities of the ribbon category are satisfied with the trivial @xmath33-symbols @xmath34 , which gives trivial bootstraps and bendings for quasiparticle string - nets .
when @xmath22 is odd , this gauge is facilitated by the _ invertibility _ of 2 modulo @xmath35 and we denote the abbreviation @xmath36 for the integer @xmath37 while @xmath38 means the regular fraction @xmath39 . for instance @xmath40 is the exchange phase in case ( ii ) .
the twisting action of a defect on an encircling quasiparticle is given by the twofold symmetry operation @xmath41 that interchanges @xmath42 for @xmath1 toric code or layer / spin conjugation @xmath43 for @xmath0-states .
the bilayer symmetry leaves the @xmath24 vector and hence the charge of a quasiparticle invariant .
since charge is fractionalized in units of @xmath44 , there are @xmath45 species of twofold defects , denoted by @xmath46 for @xmath47 , distinguished by the fractional charge they carry .
species mutation is driven by absorbing or emitting quasiparticles .
fusion or splitting outcome should be insensitive to the detail of how many times a quasiparticle circles around the defect during the absorption or emission process .
defect species are therefore topological quantities that would change by the same amount upon fusion with the quasiparticle @xmath15 or its symmetry partner @xmath48 .
species label lives in the cyclic quotient group @xmath49 , which is generated by the charge vector @xmath24 and coincides with the quasiparticle charge sector . by moving a quasiparticle @xmath15 twice around a twofold defect ( cross ) .
( b ) a linking phase @xmath50 from joining two double wilson operators @xmath51 around a pair of defects .
the two unlinked @xmath15 and @xmath48 loops can then be absorbed in the condensate if the defects fuse to the vacuum.,width=316 ] it should be noticed that the distinction of defect species is not a consequence of @xmath52 charge symmetry but rather the twofold symmetry of the anyonic system @xcite .
defect species can be measured by the braiding phase accumulated by moving a quasiparticle @xmath15 twice around a twofold defect @xmath46 .
the quasiparticle trajectory leaves behind a wilson string operator @xmath53 depicted in figure [ fig : doubleloop ] .
as the size of @xmath53 could be arbitrary small , it does not intersect with any non - trivial wilson operators in the system and corresponds the good quantum number @xmath54}\label{doubleloopphase}\end{aligned}\ ] ] around a defect with species label @xmath55 . the eigenvalue is uniquely determined by self - intersection @xmath56 , invariance under twofold symmetry @xmath57 , and a linking phase shown in figure [ fig : doubleloop ] that results in the phase @xmath58 around a pair of self - conjugate twofold defects @xmath59 .
the fusion rules @xcite between twofold defects and quasiparticles conserve charge and are given by @xmath60 there are @xmath61 charge conserving abelian quasiparticle fusion channels in and therefore the quantum dimension of defects is given by @xmath62.@xcite .
branch cuts ( wavy lines ) switch labels @xmath63 of passing quasiparticles and end at twofold defects ( crosses ) .
( i ) for @xmath1 toric code where @xmath2 , quasiparticle @xmath64 is attached with strings @xmath65 and @xmath66 , twofold defect @xmath46 is attached with string @xmath67 .
( ii ) for @xmath0-fqh states with @xmath22 odd , quasiparticle @xmath15 is attached with strings @xmath68 , and twofold defect @xmath46 is attached with string @xmath69 , @xmath70.,width=316 ] quantum states of a system of defects ( on a closed sphere ) are labeled by good quantum numbers represented by admissible fusion channels @xmath71 on a fusion tree with defects @xmath72 as open branches .
@xmath73 , one for each trivalent vertex on the fusion tree representing the splitting of object @xmath74 .
splitting states are local states generated by open wilson strings around the splitting objects , and a gauge is fixed by the particular choice of string configurations shown in figure [ fig : splittingstates ] .
the overall state can then be constructed by piecing the strings together between local splitting states on the fusion tree by matching boundary conditions .
basis transformation between different fusion trees are generated by a consistent set of fundamental @xmath75-moves @xmath76{f1.pdf}}}\right\rangle=\sum_\eta\left[f^{\psi_3\psi_2\psi_1}_{\psi_4}\right]_\xi^\eta\left|\vcenter{\hbox{\includegraphics[width=0.5in]{f2.pdf}}}\right\rangle\end{aligned}\ ] ] where the state at the left ( and right ) hand side is the ( entangled sum over boundary conditions of ) tensor product of splitting states @xmath77 ( resp .
@xmath78 ) . these tensor product states can be deformed into each other by crossing , bending and rearranging bootstraps of quasiparticle strings using the @xmath75 and @xmath79-symbols of the underlying abelian topological phase .
the reults of defect @xmath75-symbols for @xmath1 toric code and @xmath0-fqh state are summarized in table [ tab : fsymbols ] .
ll + @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 & @xmath85 + @xmath86 , @xmath87 & @xmath88 + @xmath89_{\bf a}^{\bf b}$ ] & @xmath90 ll + @xmath80 & @xmath85 + @xmath84 , @xmath91 , & + @xmath82 , @xmath92 & + @xmath86 , @xmath93 & @xmath94 + & @xmath95 + & @xmath96 non - abelian defect braiding operations are generated by fundamental exchange permutations of adjacent defects known as @xmath97-moves .
@xmath76{bmatrix1.pdf}}}\right\rangle=\sum_{\eta}\left[b^{\psi_2\psi_1}_{\phi_2\phi_1}\right]_{\xi}^{\eta}\left|\vcenter{\hbox{\includegraphics[width=0.5in]{bmatrix2.pdf}}}\right\rangle\label{bmatrixdef1}\end{aligned}\ ] ] they can be evaluated by a series of fundamental @xmath75 and @xmath79-moves @xmath98_{\xi}^{\eta}&=\sum_{\bf a}\left\{\left[f^{\phi_2\sigma_2\sigma_1}_{\phi_1}\right]_{\eta}^{\bf a}\right\}^\ast r^{\sigma_1\sigma_2}_{\bf a}\left[f^{\phi_2\sigma_1\sigma_2}_{\phi_1}\right]_{\bf a}^{\chi}\end{aligned}\ ] ] where the @xmath79-symbol relates the state before and after exchanging a pair of defects @xmath99 with fixed fusion outcome @xmath15 .
it can be derived by twisting the quasiparticle strings of the splitting state @xmath100 in figure [ fig : splittingstates ] by @xmath101 and deform back into the untwisted configuration through crossing strings and introducing a double wilson operator @xmath102 .
@xmath103{r222.pdf}}}\left|\vcenter{\hbox{\includegraphics[width=0.7in]{r221.pdf}}}\right.\right\rangle\nonumber\\&=\left\{\begin{array}{*{20}c}e^{\frac{2\pi i}{n}a_2\left(\lambda_2+\frac{n}{2}-\frac{1}{2}a_2\right)}\hfill\\\theta_{\bf
s}^{2^{-1}\alpha^2}\theta_{\bf t}^{(2^{-1})^3(\lambda_1+\lambda_2)(3\lambda_2-\lambda_1)}\label{rsymbol}\end{array}\right.\end{aligned}\ ] ] where the first line applies to @xmath1 toric code and the second to @xmath0-fqh states with @xmath22 odd .
ll + @xmath104 & @xmath105}$ ] + @xmath106_{\bf a}^{\bf b}$ ] & @xmath107}$ ] + @xmath108 & @xmath109}\delta_{\bf a}^{\sigma_x{\bf b}}$ ] ll + @xmath104 & @xmath110 + @xmath106_{\bf a}^{\bf b}$ ] & @xmath111 + @xmath108 & @xmath112 the fundamental @xmath97-moves in table [ tab : bsymbols ] are consistent exchange operations that satisfy the yang - baxter equation for adjacent permutations @xmath113 , where @xmath114 is the exchange braid between the @xmath115 and @xmath116 defect .
an exchange is gauge
independent whenever it is performed between defects of identical species , and a full braid is always gauge independent .
they generate non - abelian unitary operations that _ projectively _ represent the braid group @xmath117 of @xmath118 points on a sphere .
classical braid groups are restricted by a compactification relation if the system lives on a closed sphere , or equivalently the overall fusion channel of the defect system is the vacuum . for instance one expect the largest braid that moves an object once around all others to be trivial as it should be contractible on the other side of the sphere . @xmath119{braidaroundonce.pdf}}}=1\label{braidoncearound}\end{aligned}\ ] ] the violation of this is shown by the non - trivial @xmath120 matrices in table [ tab : bsymbols ] in a closed system of four defects with canceling species labels @xmath121 , where the @xmath61-fold ground state degeneracy is labeled by the restricted fusion channel @xmath122 . the large braid operation @xmath120 can be interpreted as a flip operator that interchanges the bilayer as the braid , and consequently the twofold branch cut left behind the defect trajectory , is contracted on the other side of the closed sphere .
the @xmath120-matrix commutes with all braiding operations and generates a center that extends the sphere braid group @xmath117 .
it squares to a scalar phase @xmath123 of products of defect exchange spins @xmath124 discussed below .
@xmath120 can thus be viewed as a braiding signature of interchanging bilayer in a defect system .
the exchange matrix defines the spin of a twofold defect by averaging over admissible quasiparticle fusion channel @xmath125 where the sign in the @xmath126 phase is given by @xmath127 .
a defect can be represented as an open end of a twofold branch cut which switches the bilayer , and therefore a @xmath128 rotation leaves behind an uncancelable branch cut surrounding the defect .
the exchange spin however identifies with @xmath129 rotation up to the constant @xmath126 phase .
the defect @xmath46 is equipped with a quasiparticle string @xmath15 that gives the fractional charge @xmath130 to the defect . upon rotation , the string wraps around the defect twice , and can be untwisted by a crossing @xmath131 and absorbing a double wilson operator @xmath53 into the condensate .
@xmath132{z2spin1.pdf}}}&=\theta_{\bf a}\vcenter{\hbox{\includegraphics[width=0.32in]{z2spin2.pdf}}}=\theta_{\bf a}\theta^0_{\bf a}\vcenter{\hbox{\includegraphics[width=0.17in]{z2spin3.pdf}}}=e^{\frac{2\pi i}{m+n}\lambda\left(\frac{\lambda+m+n}{2}\right)}\vcenter{\hbox{\includegraphics[width=0.17in]{z2spin3.pdf}}}\label{720rotation}\end{aligned}\ ] ] we notice that the ribbon ( or pair of pants " ) identity for braiding a pair of defects is violated @xmath133 , which is different from @xmath134 when the defects carry fractional charge and have non - trivial spin .
it differentiates semiclassical defects from quantum deconfined non - abelian anyons , and forms an obstruction to a fully braided category unless the bilayer structure is melted .
defect braiding statistics is encoded by the following average over admissible fusion channel within the defect sector that mimics a modular @xmath4-matrix .
@xmath135{topologicalsmatrix.pdf}}}=\frac{1}{\mathcal{d}_{\sigma}}\sum_{\bf a}r^{\overline{\sigma_{\lambda_2}}\sigma_{\lambda_1}}_{\bf a}r^{\sigma_{\lambda_1}\overline{\sigma_{\lambda_2}}}_{\bf a}\label{defectsmatrix}\end{aligned}\ ] ] where @xmath136 is the total quantum dimension of the defect sector , and @xmath137 is the anti - particle of @xmath46 . the gaussian sum has closed form solution @xmath138 where @xmath70 , and the scalar @xmath139 is 1 if @xmath140 mod 4 , @xmath141 if @xmath142 mod 4 , or @xmath143/\sqrt{2}$ ] if @xmath22 even . along the horizontal direction that leaves the branch cut ( wavy line ) invariant .
( b ) @xmath144-action on a vertical wilson loop @xmath145 .
( c ) dehn twist @xmath146 along the vertical ( longitudinal ) direction and its action on a double wilson loop @xmath53.,width=288 ] defect braiding @xmath4 and exchange @xmath144 matrices can alternatively be evaluated as geometric _ congruent _ transformation of ground states on a torus with a twofold branch cut , which occupies a particular cycle , say the @xmath147-direction , on the torus ( see figure [ fig : dehntwist ] ) .
the subgroup of modular transformations that leaves the branch cut invariant [ i.e. fixes the @xmath147-cycle with @xmath7 coefficient in @xmath148 is known as a congruent subgroup @xmath149 where @xmath150 are dehn twists on the torus along the @xmath151 direction .
the wilson operator algebra is generated by quasiparticle strings @xmath145 of moving @xmath15 along the vertical cycle and @xmath53 twice along the horizontal cycle ( see figure [ fig : dehntwist ] ) .
@xmath145 corresponds the quasiparticle string attaching to a twofold defect and gives rise to its species @xmath130 , and @xmath53 is equivalent to the double wilson operator in figure [ fig : doubleloop ] that measures the species label .
as these operator can pass across the branch cut and change its label , they obey the twofold symmetry @xmath152 , @xmath153 and satisfy the fusion relations @xmath154 , @xmath155 .
they commute up to the braiding phase of intersection @xmath156 and transform under the congruent generators @xmath4 and @xmath144 according to @xmath157 and @xmath158 ( see figure [ fig : dehntwist ] ) while keeping @xmath159 and @xmath160 .
the @xmath45-fold degenerate ground states are simultaneous eigenstates of @xmath53 and @xmath161 that corresponds defect species , @xmath162 , @xmath163 , where the eigenvalue @xmath164 is given by .
@xmath165 where @xmath15 is any quasiparticle with charge @xmath130 , and @xmath166 is the eigenvalue @xmath167 . and
therefore up to a constant phase , the congruent @xmath144-matrix that represents double dehn twist @xmath168 is diagonal with spin entries , @xmath169 .
the commuting @xmath145 and @xmath170 share the simultaneous eigenstate @xmath171 , which generates all ground states by @xmath172 , where @xmath130 and the sum is restricted to any collection of @xmath45 quasiparticles @xmath173 with distinct fractional charges .
the congruent @xmath4-matrix that represents dehn twist @xmath174 has entries @xmath175 where @xmath176 is the eigenvalue @xmath177 . after evaluating the gaussian sum ,
equals the braiding @xmath4-matrix up to a constant phase .
the braiding @xmath4 and exchange spin @xmath144 matrices in and generate a unitary representation for the congruent subgroup @xmath6 where @xmath178_{\lambda_1\lambda_2}=sgn(m+n)\delta_{\lambda_1,-\lambda_2}\end{aligned}\ ] ] serves as a charge conjugation operator to defect species .
_ acknowledgements _ this work was supported by the simons foundation ( jt ) and national science foundation through grant dmr 09 - 03291 ( ar ) and dmr-1064319 ( xc ) .
44 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , * * , ( ) . ,
_ _ ( , ) . , * * , ( ) . , , ,
* * , ( ) .
, * * , ( ) . ,
* * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , _ _ ( ) , http://www.theory.caltech.edu/people/preskill/ph229/#lecture . , , , , ( ) . , , , , , * * , ( ) . ,
_ _ ( , ) . , * * , ( ) . , , , ,
* * , ( ) . , , , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , ( ) . ,
* * , ( ) . , * * , ( ) . ,
* * , ( ) . ,
* * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , ( ) . ,
* * , ( ) . ,
* * , ( ) . , * * , ( ) . ,
_ _ ( , ) . , _ _ ( , ) , ed . , , , , * * , ( ) . , , , , , ,
* * , ( ) . , , , , , * * , ( ) . , _ _ ( , ) , ed . ,
_ _ ( , ) . , , , * * , ( ) . |
in recent times cellular automata based simulations of traffic flow have gained considerable importance . by extending the range of rules from nearest neighbours to a range of 5 grid sites and introducing 6 discrete velocities
@xmath0 nagel and schreckenberg @xcite have found a striking resemblance of simulation and realistic traffic behaviour . for @xmath1 schadschneider and schreckenberg
have found an analytic solution @xcite . for higher @xmath2 ,
these analytic approaches lead to good approximations for the average behavior @xcite .
further analytic results can be found in @xcite .
nagel @xcite pointed out the strong connections between particle hopping models and fluid - dynamical approaches for traffic flow .
much less is known about modelling of multi - lane traffic .
this statement is not only true for particle hopping models for traffic flow , but for traffic flow theory in general . queueing models are not truly multilane , but emulate multiple lanes by switching the order of vehicles on one lane whenever a passing would have occurred in reality @xcite .
fluid - dynamical models incorporate multi - lane traffic only by parametrization @xcite , although sometimes based on kinetic theory @xcite .
traditional car - following theory ( see @xcite ) by and large never dealt with multi - lane traffic .
modern microscopic traffic simulation models ( e.g. @xcite ) obviously handle multi - lane traffic by necessity .
cremer and coworkers @xcite even treat multi - lane traffic in the context of cellular automata models . yet
, all these papers approach the problem by using heuristic rules of human behavior , without checking which of these rules exactly cause which kind of behavior . in validations then ( e.g. @xcite ) , it often enough turns out that certain features of the model are not realistic ; and because of the heuristic approach it is difficult to decide which rules have to be changed or added in order to correct the problem . for that reason
, a more systematic approach seems justified .
our approach here is to search for _ minimal _ sets of rules which reproduce certain macroscopic facts .
the advantage is that relations between rules and macroscopic behavior can be more easily identified ; and as a welcome side - effect one also obtains higher computational speed .
we again choose particle hopping models as starting point for this investigation because their highly discrete nature reduces the number of free parameters even further .
it is clear that a similar analysis could be applied to continuous microscopic models , hopefully benefiting from the results obtained in this and following papers .
nagatani examined a two lane system with completely deterministic rules and @xmath1 @xcite , where cars either move forward or change lanes .
a very unrealistic feature of this model are states where blocks of several cars oscillate between lanes without moving forward at all .
he corrected this by introducing randomness into the lane changing @xcite .
latour has developed the two lane model which served as the basis for the one discussed here @xcite .
rickert used a more elaborate rule set for two lane traffic which reproduced the phenomenon of increased flow with an imposed speed limit @xcite .
for the convenience of the reader we would like to outline the single lane model .
the system consists of a one dimensional grid of @xmath3 sites with periodic boundary conditions .
a site can either be empty , or occupied by a vehicle of velocity zero to @xmath2 .
the velocity is equivalent to the number of sites that a vehicle advances in one update provided that there are no obstacles ahead .
vehicles move only in one direction .
the index @xmath4 denotes the number of a vehicle , @xmath5 its position , @xmath6 its current velocity , @xmath7 its maximum speed .
we now allow for different esired velocities @xmath7 to include an inhomogeneous fleet ] , @xmath8 the number of the preceding vehicle , @xmath9 the width of the gap to the predecessor . at the beginning of each time step the rules are applied to all vehicles simultaneously ( parallel update , in contrast to sequential updates which yield slightly different results ) .
then the vehicles are advanced according to their new velocities . * * if * @xmath10 * then * @xmath11 ( * s1 * ) * * if * @xmath12 * then * @xmath13 ( * s2 * ) * * if * @xmath14 * and * @xmath15 * then * @xmath16 ( * s3 * ) * s1 * represents a linear acceleration until the vehicle has reached its maximum velocity @xmath17 .
* s2 * ensures that vehicles having predecessors in their way slow down in order not to run into them . in * s3
* a random generator is used to decelerate a vehicle with a certain probability modelling erratic driver behaviour .
the free flow average velocity is @xmath18 ( for @xmath19 ) .
the single lane model is not capable of modelling realistic traffic mainly for one reason : a realistic fleet is usually composed of vehicles types having different desired velocities . introducing such different vehicle types in the single lane model only results in _ platooning _ with slow vehicles being followed by faster ones and the average velocity reduced to the free flow velocity of the slowest vehicle @xcite .
we introduce a two lane model consisting of two parallel single lane models with periodic boundary conditions and four additional rules defining the exchange of vehicles between the lanes .
the update step is split into two sub - steps : 1 .
check the exchange of vehicles between the two lanes according to the new rule set .
vehicles are only moved _
sideways_. they do not _ advance_. note that in reality this sub - step regarded by itself seems unfeasible since vehicles are usually incapable of purely transversal motion .
only together with the second sub - step our update rules make physically sense .
+ this first sub - step is implemented as strict parallel update with each vehicle making its decision based upon the configuration at the beginning of the time step .
2 . perform independent single lane updates on both lanes according to the single lane update rules . in this second sub - step
the resulting configuration of the first sub - step is used .
the most important parameters of the two lane model are as follows : symmetry : : : the rule set defining the lane changing of vehicles can be both symmetric and asymmetric .
the symmetric model is interesting for theoretical considerations whereas the the asymmetric model is more realistic .
stochasticity : : : the single lane model proved that a strictly deterministic model is not realistic : the model did not show the desired spontaneous formation of jams . in the case of the two lane model the lack of stochasticity in combination with the parallel update results in strange behaviour of slow platoons occupying either lane : since none of the vehicles has reached its maximum velocity and all evaluate the other lane to be better there is collective change sidewise which is usually reversed over and over again until the platoon dissolves or the platoon is passed by other vehicles .
+ we introduce stochasticity into the two lane rule set to reduce the effective number of lane changes and thus dissolve those platoons .
the simulation also revealed that is effect is also important in the asymmetric free flow case ( see [ pingpong ] ) .
direction of causality : : : in the single lane model a vehicle only looks ahead (= downstream = in the direction of vehicle flow ) so that causality can only travel upstream (= in the opposite direction of vehicle flow ) . a reasonable lane changing
rule must include a check of sites _ upstream _ in order not to disturb the traffic of the destination lane .
this would result in causality travelling downstream .
a somewhat generic starting point for modeling passing rules is the following : ( t1 ) you look ahead if somebody is in your way .
( t2 ) you look on the other lane if it is any better there .
( t3 ) you look back on the other lane if you would get in somebody else s way .
technically , we keep using @xmath20 for the number of empty sites ahead in the same lane , and we add the definitions of @xmath21 for the forward gap on the other lane , and @xmath22 for the backward gap on the other lane .
note that if there is a vehicle on a neighbouring site both return -1 .
the generic multi - lane model then reads as follows .
a vehicle @xmath4 changes to the other lane if all of the following conditions are fulfilled : * @xmath23 ( * t1 * ) , * @xmath24 ( * t2 * ) , * @xmath25 ( * t3 * ) , and * @xmath26 ( * t4 * ) .
@xmath27 , @xmath28 , and @xmath29 are the parameters which decide how far you look ahead on your lane , ahead on the other lane , or back on the other lane , respectively . according to the before mentioned characteristics we associate the parameters of our rule set : [ cols="^,^,^",options="header " , ]
as an example , we start with @xmath30 , @xmath31 , @xmath32 , and @xmath33 .
that means that both @xmath27 and @xmath28 are essentially proportional to the velocity , whereas looking back is not : it depends mostly on the expected velocity of other cars , not on one s own . in the symmetric version of this model ,
cars remain on their lane as long as they do nt `` see '' anybody else .
if they see somebody ahead on their own lane ( i.e. @xmath34 ) , then they check on the other lane if they can switch lanes and do so if possible . afterwards ,
if they are satisfied , they remain on this lane until they become dissatisfied again . in the asymmetric version , cars always try to return to the right lane , independent of their situation on the left lane .
space - time - plots both of the symmetric and the asymmetric version are shown in figs .
[ spacetime_symm_lb5 ] and [ spacetime_asym_lb5 ] . for these plots
, we simulated a system with a length of 12,000 sites of which we plot 400 sites in 400 consecutive time - steps .
the density is 0.09 which is slightly above the density of maximum flow ( see below ) .
vehicles go from left to right ( spatial axis ) and from top to bottom ( time axis ) .
traffic jams appear as solid areas with steep positive inclination whereas free flow areas are light and have a more shallow negative inclination .
each plot is split into two parts : the left part containing the left lane and the right part containing the right lane , respectively .
note that plot [ spacetime_asym_lb5 ] ( left lane ) gives a good impression of the great number of lane changes through the high frequency of short vehicle life lines appearing and disappearing : these are vehicles that temporarily leave the right lane to avoid an obstacle .
they go back to their old lane as soon as the obstacle has been passed .
it will be confirmed quantitatively that indeed the rate of lane changes is much higher for the asymmetric model than for the symmetric model . before going on
, we would like to describe our standard simulations set - up for the following observations .
note that quantitative simulation results were obtained with a much larger system than the qualitative space - time plots .
we simulated a system of length @xmath35 with closed boundary conditions , i.e. traffic was running in a loop .
we started with random initial conditions , i.e. @xmath36 cars were randomly distributed on both lanes around the complete loop with initial velocity @xmath37 .
since the system is closed , the average density per lane is now fixed at @xmath38 where the `` 2 '' stands for the number of lanes .
the simulation was then started , 1000 time steps were executed to let the transients die out , and then the data extraction was started .
the flow which is found in the fundamental diagrams is both space and time averaged , i.e. @xmath39 values for lane change frequency and ping pong lane change frequency ( see below ) are obtained by the same averaging procedure except that statistics are gathered every time step , since _ by definition _
ping pong lane changes occur in subsequent time steps .
we usually used @xmath40 , and the same procedure was repeated for each density found in the plots . with a resolution of @xmath41 an average plot took about 22 hours of computation time on a sparc 10 workstation . by comparing these models with each other and with earlier results ,
we make the following observations ( fig . [ fig_flow_lb5_c1_0 ] unless otherwise noted ) : \(i ) both for the symmetric and the asymmetric version , maximum flow is higher than twice the maximum flow of the single lane model ( fig .
[ fig_single_multi ] ) . which means that , in spite of the additional disturbances which the lane changing behavior introduces into the traffic flow , the general effects are beneficial , probably by diminishing large deviations from `` good '' flow patterns .
\(ii ) both for the symmetric and the asymmetric version , the combined 2-lane flow reaches a maximum at @xmath42 , which is at or near a sharp bend of the flow curve .
\(iii ) for the asymmetric model , flow on the left lane keeps increasing slightly for @xmath43 , but this is over - compensated by the decreasing flow on the right lane .
\(ii ) and ( iii ) together lead one to the speculation that maximum flow in the asymmetric case here actually is connected to a `` critical '' flow on the right lane and a `` sub - critical '' flow on the left lane .
any addition of density beyond here leads to occasional break - downs on the right lane and thus to a much lower flow there .
obviously , such interpretations would have to be clarified by further investigations , and the word `` critical '' would have to be used with more care , such as is pointed out in @xcite for the single lane case .
\(iv ) for both lanes combined , the curves for symmetric and asymmetric traffic actually look fairly similar .
if the above interpretation is right , this means that the overall density of maximum flow is a fairly robust quantity , but one can stabilize one lane at a much higher density if this density is taken from the other lane .
\(v ) at very low densities in the asymmetric case , flow on the left lane , @xmath44 , only slowly builds up .
this is to be expected , since at least two cars have to be close to each other to force one of them on the left lane , leading to a mean field solution of @xmath45 for @xmath46 .
\(vi ) for @xmath47 , flows on both lanes in the asymmetric models are fairly similar and similar to the lane flows in the symmetric models .
to get some further insight into the lane changing dynamics , fig .
[ fig_change_lb5 ] shows the frequency of lane changing both for the asymmetric and the symmetric model .
\(i ) note that in the asymmetric case there is a sharp bend in the curve , which is not found for the symmetric case .
this bend is also near @xmath48 , giving further indication that the dynamics above and below @xmath48 are different .
\(ii ) for the symmetric case , lane changing occurs with less than half the frequency compared to the asymmetric case .
\(iii ) in the symmetric case , the lane changing frequency per site for small densities increases approximately quadratically up to rather high densities , whereas the same quantity for the asymmetric model grows approximately linearly already for fairly low densities .
suggests that for the symmetric case a mean field description of interaction , @xmath49 , would be valid up to comparably high densities .
obvious that this does not work .
since the vehicles have a strong tendency to be on the right lane , already a density of 0.04 per lane would be a density of 0.08 if everybody were on the right lane .
yet , @xmath50 is known to be already a density of high interaction in single lane traffic .
since this high interaction tends to spread vehicles out @xcite , each additional vehicle simply adds its own share of lane changes , making the relation roughly linear .
\(iv ) the maximum number of lane changes occurs at densities much higher than @xmath48 .
the lane changing probability _ per vehicle _ , however , reaches a maximum below the critical density ( fig .
[ fig_lanechangespercar ] ) .
an artifact of the so far described algorithm is easily recognizable when one starts with all cars on the same lane , say the right one . assuming fairly high density , then all cars see somebody in front of them , but nobody on the left lane . in consequence , everybody decides to change to the left lane , so that _ all _ cars end up on the left lane . here , they now all decide to change to the right lane again , etc .
, such that these coordinated lane changes go on for a long time ( `` cooperative ping - pong effect '' ) .
this effect has already been observed by nagatani for the much simpler two - lane model @xcite .
one way around this is to randomize the lane changing decision @xcite .
the decision rules remain the same as above , but even if rules t1 to t3 lead to a yes , it is only accepted with probability @xmath51 . with this fourth lane changing rule ,
patterns like the above are quickly destroyed . in order to quantify the effects of a different @xmath51 ,
simulations with @xmath52 were run .
the observations can be summarized as follows : \(i ) the flow - density curves are only marginally changed ( fig . [ fig_flow_lb5 ] ) .
\(ii ) the frequency of lane changes is decreased in general , but , except for @xmath53 in the asymmetric case , by much less than the factor of two which one would naively expect ( fig .
[ fig_change_lb5 ] ) .
that means that usually there is a _ dynamic _
reason for the lane change , that is , if it is not done in one time step due @xmath54 , then it is re - tried in the following time step , etc .
\(iii ) to better quantify in how far a @xmath55 actually changes the pattern of vehicles changing lanes back and forth in consecutive time steps , we also determined the frequency of `` ping pong lane changes '' , where a car makes two lane changes in two consecutive iterations .
obviously , there are left - right - left ( lrl ) and right - left - right ( rlr ) ping pong lane changes .
[ fig_pingpong_lb5 ] shows that reducing the probability to change lanes , @xmath51 , from 1 to 1/2 has indeed a beneficial effect : the number of ping pong lane changes decreases by about a factor of five .
yet , for the symmetric case , the frequency of ping pong lane changes is more than an order of magnitude lower in both cases anyway .
this indicates that in simulations starting from random initial conditions , the cooperative effect as described further above as cooperative ping - pong effect does not really play a role for the statistical frequency , because this effect should be the same for the symmetric and the asymmetric model . instead ,
the cause of the ping pong lane changes in the asymmetric model is as follows : assume just two cars on the road , with a gap of 5 between them .
with respect to velocity , both cars are in the free driving regime , and their velocities will fluctuate between 4 and 5 . now assume that the following car has velocity 5 from the last movement .
that means that it looks 6 sites ahead , sees the other car , and changes to the left lane .
then , assume that in the velocity update , the leading car obtains velocity 5 and the following car obtains velocity 4 .
then , after the movement step , there is now a gap of 6 between both cars , and in the lane changing step , the follower changes back to the right lane . and this can happen over and over again in the asymmetric model , but will not happen in the symmetric model : once the following car in the above situation has changed to the left lane , it will remain there until it runs into another car on the left lane . to investigate this second kind of ping pong lane changes we ran simulations recording whether a ping pong lane change was made at low velocities @xmath56 or high velocities @xmath57 .
[ fig_slowfastpingpong ] shows a very distinct peak for _ fast _ ping pong changes at low densities whereas _
slow _ ping pong changes have a lower peak at higher densities similar to that of the symmetric case .
this gives a strong indication that most lane changes are actually caused by the `` tailgating effect '' as described above , which is an artifact of the rules .
it is , though , to be expected that this behavior does not have a strong influence on the overall dynamics : it mostly happens in the free driving regime ; as soon as , for example , another car is nearby in the left lane , it is suppressed by the looking back and forward on the other lane .
we would like to mention two other parameter combinations .
they are presented because they generate artifacts which contradict the common sense one would apply to the phenomena of traffic flow .
\(i ) in the first case the lookahead is reduced to @xmath58 instead of @xmath30 . while this change is negligible for vehicles at higher velocities it becomes crucial to a vehicles stopped in a jam : assuming the current velocity to be zero the vehicle looks _ zero _ sites ahead and decides to remain in the current lane due to the non fulfilled rule @xmath59 .
this state will persist until the predecessor moves even if the other lane is _ completely _ free ! fig .
[ fig_model1 ] shows the impact the reduced look ahead on overall flow : for density @xmath60 there is no perceptible flow in the right lane which corresponds to a traffic jam that occupies more or less the whole right lane .
\(ii ) in the second case the look - back is reduced to @xmath61 .
vehicles no longer check whether their lane changing could have a disadvantageous effect on the other lane which corresponds to a very egoistic driver behaviour .
[ fig_flow_lbx_c1_0 ] shows flow density relationships for look - back @xmath62 and @xmath61 .
it is obvious that the decrease in look - back also decreases the maximum flow at critical densities .
moreover @xmath63 seems to split the curves of the symmetric and asymmetric cases which used to be almost identical for @xmath64 : the lack of look - back is much more disadvantageous for asymmetric than for symmetric rules . in figs .
[ spacetime_symm_lb0 ] and [ spacetime_asym_lb0 ] we used @xmath63 for the symmetric and asymmetric rule sets with one plot per lane .
it is clearly visible ( compare to figs .
[ spacetime_symm_lb5 ] and [ spacetime_asym_lb5 ] ) how @xmath63 completely disrupts the laminar flow regime .
vehicles change lanes without looking back ; and due to the formulation of the model this does not cause accidents , but causes the obstructed vehicles to make sudden stops . since these stops are caused more or less randomly , the regime becomes much more randomly disturbed than before , somewhat reminiscent of the asymmetric stochastic exclusion process ( see @xcite ) . as seen before the effect is even more drastic for the asymmetric rule set since the number of lane changes is higher than in the symmetric case . in fig .
[ spacetime_asym_lb0 ] with @xmath63 dynamics are dominated by small traffic jams caused by lane changes , while in fig .
[ spacetime_symm_lb5 ] there are still some fairly laminar areas .
compared to reality ( e.g. @xcite ) , the lane change frequency in the asymmetric models presented here is by about a factor of 10 too high . using @xmath65 would correct this number , but is dynamically not a good fix : it would mean that a driver follows a slower car in the average for 10 seconds before he decides to change lanes .
besides , it was shown that about 90% of the lane changes in the asymmetric models here are produced by an artificial `` tailgating dance '' , where a follower changes lanes back and forth when following another car .
it remains an open question in how far artifacts like this can be corrected by the current modeling approach or if it will be necessary to , e.g. , introduce memory : if one remembers to just have changed lane from right to left , one will probably stay on that lane for some time before changing back .
another defect of the models presented in this paper is that the maximum flow regime is most probably represented incorrectly . both measurements ( e.g. @xcite or fig .
3.6 in @xcite ) or everyday observation show that real traffic shows a `` density inversion '' long before maximum flow , that is , more cars drive on the left lanes than on the right lanes .
this effect is more pronounced for countries with higher speed limits .
let us denote by @xmath66 the density of maximum flow of the _ single lane _ case .
it follows for the real world two lane case that at a certain point the left lane will have a density higher than this density @xmath66 whereas the right lane has a density lower than @xmath66 . when further increasing the overall density , then the flow on the left lane will decrease whereas it still increases on the right lane .
it is unclear if the net flow here increases or decreases ; but it should become clear that instabilities here are caused by the left lane first .
this is in contrast to the models of this paper , where the right lane reaches the critical density first .
work dealing with this problem is currently in progress .
also the effect of different maximum velocities will be addressed in later papers .
we have presented straightforward extensions of the cellular automata approach to traffic flow so that it includes two - lane traffic .
the basic scheme introduced here is fairly general , essentially consisting of two rules : look ahead in your own lane for obstructions , and look in the other lane if there is enough space .
the flow - density relations of several realizations of this scheme have been investigated in detail ; possible artifacts for certain parameter choices have been pointed out . in general , there seem to be two important lessons to be drawn from our investigations : * checking for enough space on the other lane ( `` look - back '' ) is important if one wants to maintain the dynamics consisting of laminar traffic plus start stop waves which is so typical for traffic . * especially in countries with high speed limits , observations show a density inversion near maximum flow , that is , the density is higher on the left lane than on the right lane .
this effect is not reproduced by our models ( work in progress ) . yet
, in general , it seems that the approach to multi lane traffic using simple discrete models is a useful one for understanding fundamental relations between microscopic rules and macroscopic measurements .
we thank a. bachem and c. barrett for supporting mr s and kn s work as part of the traffic simulation efforts in cologne ( `` verkehrsverbund nrw '' ) and los alamos ( transims ) .
we also thank them , plus ch .
gawron , t. pfenning , s. rasmussen , and p. wagner for help and discussions .
computing time on the parsytec gcel-3 of the zentrum fr paralleles rechnen kln , on the paragon xp / s 5 and xp / s 10 of the zentralinstitut fr angewandte mathematik ( zam ) of the forschungszentrum jlich and on the sgi-1 of the regionales rechenzentrum kln is gratefully acknowledged .
we further thank all persons in charge of maintaining the above mentioned machines .
m. mcdonald , m.a .
brackstone , simulation of lane usage characteristics on 3 lane motorways , paper no . 94at051 , in : proceedings of the 27th international symposium on automotive technology and automation ( isata ) ( automotive automation ltd , croydon , england , 1994 ) , p. 365 . | we examine a simple two lane cellular automaton based upon the single lane ca introduced by nagel and schreckenberg .
we point out important parameters defining the shape of the fundamental diagram .
moreover we investigate the importance of stochastic elements with respect to real life traffic . 0.3 |
the nature of qcd matter at high temperatures and densities has been of interest due to the experiments at the relativistic heavy ion collider(rhic ) in bnl , new york and the upcoming large hadron collider(lhc ) in cern , geneva .
theoretically , it is expected that the spontaneously broken chiral symmetries at low temperatures are restored at high temperatures . while lattice methods have been very effective in predicting the transition temperature as well as the nature of the the strongly interacting matter near the transition temperature , fermions with exact chiral symmetry on lattice are important for such a study of the chiral phase transition .
kaplan@xcite proposed to define fermions with exact chiral symmetry on a five - dimensional ( 5-d ) lattice with a mass term @xmath2 in the form a step function(domain wall ) and with an infinite extent along the fifth dimension .
the massless 4-d fermions are obtained as localized on the wall , and are hence known as the domain wall fermions . on a finite lattice needed for numerical simulations , however ,
fermions of both chiralities exist with an exponentially small overlap between the respective chiral states@xcite .
currently , the most popularly used fermions in qcd simulations at finite temperatures / densities are the staggered fermions which have only a remnant chiral symmetry on the lattice .
moreover , they explicitly break spin and flavor symmetries .
the full chiral symmetry for these fermions is recovered only in the continuum limit , i.e. , in the limit of vanishing lattice spacing . in spite of the ( exponentially small in @xmath0 , the number of sites in the fifth dimension ) chiral violation on the lattice , the domain wall fermions are more promising than the staggered fermions due to their exact flavor and spin symmetry on the lattice . on the other hand
these are more expensive to simulate as the computational cost increases linearly with @xmath0 .
one has to optimize @xmath0 and @xmath2 for full qcd simulations . in order to gain insights on ways to minimize the lattice cut - off effects we study various thermodynamic quantities of free domain wall fermions as a function of @xmath2 and @xmath0 with an aim to optimize the irrelevant lattice parameters for faster convergence to their continuum values .
we find that by adjusting the domain wall height @xmath2 in the range @xmath7 rather than the frequently used choice of @xmath8 , a faster convergence to the continuum results for both finite and infinite values of @xmath0 is achieved .
however , the cut - off effects are seen to be quite large on small lattices with temporal extent of 4 - 6 where most of the current qcd simulations are being done .
we therefore examine modifications of the domain wall , as well as the overlap kernel to minimize such corrections for small lattice sizes .
the plan of the paper is as follows : in section ii , we compute the energy density of free domain wall quarks on the lattice analytically and verify that it yields the correct continuum limit . in section iii , the same quantity is computed numerically and the various lattice parameters for which the convergence to the continuum is fastest are estimated . in section iv , we repeat the calculations of energy density in the presence of chemical potential and susceptibility and confirm that this optimum m - range does not shift significantly . in section
v , we propose a method of reducing the lattice cut - off corrections to thermodynamic quantities on small lattice sizes , computed using both the chiral fermions , namely the domain wall at infinite @xmath0 and the overlap fermions .
this helps in faster convergence to the continuum results even for @xmath8 .
the domain wall fermions@xcite in the continuum are defined on a 5d space - time with the mass term in the fifth dimension in form of a domain wall @xmath9 .
this helps in localizing a fermion of definite chirality on the domain wall .
the domain wall operator in the continuum is given as , @xmath10 the massless fermion modes in 4-d are obtained when the following conditions are simultaneously satisfied .
@xmath11 it was shown that only one normalizable solution exist , bounded to the wall at @xmath12 where the @xmath13 changes abruptly .
the corresponding analog of the domain wall term on the lattice is of the form @xmath14 on the lattice we do not get a single massless mode by discretizing eq . ( [ eqn : dwc ] ) .
this is because the lattice regulator is anomaly free , so massless fermions of both handedness exist on the lattice .
a wilson term is needed to spatially separate the left and the right handed fermions in the 5th dimension by localizing them on the domain wall and the anti - domain wall respectively which are separated from each other by the lattice extent in the fifth dimension @xmath0 . to obtain thermodynamical quantities of free fermions with exact chiral symmetry on the lattice in 4-d , we need to divide out contribution of the heavy fermion modes which exist in the fifth dimension .
this is done by subtracting a pseudo - fermion action@xcite from the standard 5-d action .
following shamir@xcite , the domain wall fermion action on a @xmath15 anisotropic lattice with lattice spacings of @xmath16 , @xmath17 and @xmath18 in the three spatial , the temporal and the fifth dimension respectively can be written as , @xmath19\psi(x',s'),\end{aligned}\ ] ] with the boundary conditions @xmath20 where @xmath21 are the chiral projectors and @xmath22 is the bare quark mass in lattice units .
@xmath23 is the wilson - dirac operator defined on a 4-d lattice .
the volume of the system is @xmath24 and @xmath25 is its temperature .
the chemical potential @xmath26 is usually introduced as a lagrange multiplier corresponding to the conserved number density in the expression for the lagrangian . for the domain wall fermions
we do not have a good prescription for obtaining the conserved number density .
following bloch and wetting @xcite , we incorporate the chemical potential in @xmath23 but in a general form using the functions @xmath27 and @xmath28 @xcite defined below .
these multiply the @xmath29 factors in the wilson - dirac operator leading to , @xmath30 in this paper we consider the non - interacting fermions , i.e. , @xmath31 . introducing @xmath32 and @xmath33 by @xmath34 the free wilson - dirac operator in eq .
( [ eqn : wilop ] ) can be diagonalized in the momentum space in terms of the functions , @xmath35 such that @xmath36 to study thermodynamics of fermions one has to necessarily take anti - periodic boundary conditions along the temporal direction .
assuming periodic boundary conditions along the spatial directions we obtain @xmath37 it is to be noted that @xmath2 , the height of the domain wall on the lattice , is bound to @xmath38 to simulate one flavor quark on the lattice .
to suppress the heavy mode contributions and recover a single chiral fermion , pseudo - fermion fields are introduced which have the same action but with @xmath39 i.e with anti - periodic boundary condition in the fifth dimension@xcite .
the fifth dimensional degrees of freedom can be integrated out to yield an effective domain wall operator@xcite @xmath40 where the transfer matrix @xmath41 is @xmath42 since @xmath41 can be shown to be hermitian for @xmath43 , and therefore has real eigenvalues , @xmath44 has only positive eigenvalues for even @xmath0 . introducing @xcite a notation @xmath45 , the function @xmath46 in the domain wall operator
can be expressed in the form of a tanh function as in eq .
( [ eqn : tanh ] ) .
@xmath47 the above derivation of the effective domain wall operator assumes that @xmath48 does not have any zero eigenvalues . for
if it does , then the contribution of the heavy modes is zero .
if @xmath49 be an eigenvalue of t , then this assumption requires that @xmath50 this is clearly true for @xmath43 for even the interacting fermions where @xmath41 is hermitian and thus any @xmath49 is real . however , once chemical potential is introduced in the wilson - dirac operator , as above , @xmath23 and @xmath41 are not hermitian any longer for the free fermions themselves , leaving open the possibility that this condition will not be met .
it is easy to see that three distinct limits are of interest in which we should compute the various thermodynamic quantities for massless domain wall operator .
these are as follows : 1 .
@xmath51 , where one obtains exact chiral fermions for @xmath52 , 2 .
@xmath53 , such that @xmath54 , leading to an approximate form for the overlap fermions@xcite , and 3 .
@xmath55 , corresponding to the form of domain wall operator directly relevant for practical simulations on the lattice . in this limit ,
the tanh function in eq .
( [ eqn : tanh ] ) becomes sign function and the resultant effective domain wall operator is given as @xmath56 for @xmath57 , this form of the domain wall operator satisfies the ginsparg - wilson relation @xcite , as shown in the appendix [ app1 ] .
indeed , it is just like the overlap operator , but with a different argument of the sign function .
the finite size corrections to various thermodynamic quantities computed with this lattice operator are expected to be different from the overlap case . for this type of ginsparg
wilson fermion too the introduction of chemical potential necessarily leads to chiral symmetry breaking @xcite on the lattice because the action in presence of @xmath58 is not invariant under the chiral transformations@xcite on lattice . like in the case of the overlap fermions , chiral symmetry
is exactly realized for these wall fermions only in the absence of chemical potential .
the energy density @xmath59 of the domain wall fermions in the chiral limit is evaluated from the temperature partial derivative of the partition function , @xmath60 .
this is equivalent to taking a partial derivative with respect to @xmath17 on a lattice of fixed size @xmath61 .
the energy density , @xmath62 can be evaluated analytically in terms of the quantities @xmath63 and @xmath64 defined below in eq .
( [ eqn : def1 ] ) , where the dash denotes the @xmath17-derivative of the respective quantities .
defining @xmath65 one has @xmath66 setting @xmath67 after evaluating the @xmath17-derivatives , the summation over the discrete matsubara frequencies can be evaluated analytically by the standard contour integral technique or numerically by explicitly summing over them and the momenta @xmath68 . for the former , we need to determine the singularities of the summand @xmath69 in eq .
( [ eqn : sgne ] ) .
we outline below briefly the results one obtains for the zero and finite temperature cases .
[ [ t-0-mu - ne-0 ] ] @xmath70 , @xmath71 : in order to obtain a general condition for eliminating the spurious @xmath72-divergences , we first calculate the energy density at zero temperature in the limit @xmath73 at finite a. the frequency sum @xmath74 in eq . ( [ eqn : sgne ] ) gets replaced by the integral @xmath75 in this limit .
subtracting the vacuum contribution corresponding to @xmath43 , i.e. @xmath76 , the energy density at zero temperature is given by @xmath77~.\ ] ] for brevity , we suppress from now on the explicit @xmath68-dependence of the function @xmath69 although we retain the overall sign to remind of it . .
] choosing the contour shown in figure [ zmuc ] , the expression above can be evaluated in the complex @xmath78-plane as @xmath79~.~\end{aligned}\ ] ] the second and fourth terms cancel since @xmath69 satisfies @xmath80 .
hence , we obtain @xmath81~,~\end{aligned}\ ] ] where -i@xmath82 is the residue of the function @xmath83 at the pole
@xmath84 sinh@xmath85 .
it is clear from eq .
( [ eqn : emuz ] ) that the condition @xmath86 cancels the integrals , yielding the canonical fermi surface form of the energy density . for @xmath87
, there will in general be violations of the fermi surface on the lattice .
moreover , in the continuum limit @xmath88 , one will in general have the @xmath5-divergences for @xmath87 in the energy density .
the condition to obtain the correct continuum values of @xmath89 turns out to be @xmath90 . that this effective domain wall fermion satisfy the same condition as the overlap @xcite suggest that such condition may be generically true for ginsparg - wilson fermions
also that one obtains identical condition in the staggered case@xcite suggests that the behavior near the continuum limit dictates this condition .
note also that the form used by bloch and wettig @xcite , namely , exp(@xmath91 ) for @xmath27 , @xmath28 , also satisfies the condition @xmath92 .
[ [ t - ne-0 ] ] @xmath93 : in order to choose the appropriate contour in the @xmath93 case , note that the function @xmath94 has poles at @xmath95 .
these turn out to contain the physical poles in the continuum limit . however , there are additional(unphysical ) poles at @xmath96 , @xmath97 for @xmath98 and at @xmath99 where @xmath100 .
the definitions of the quantities @xmath101,@xmath102,@xmath103 are @xmath104 note the obvious similarity with the case of the overlap fermions@xcite , where the same set of poles contribute as well .
unlike in the overlap case , however , there are no branch cuts present in this case .
the contour chosen for evaluating the frequency sum shown in figure [ econt ] , is thus different from that chosen for overlap fermions .
the residue of the pole enclosed by the contour for @xmath69 comes out to be , @xmath105 with the first term yielding the continuum value of the energy density in the limit of vanishing lattice spacing @xmath16 .
the energy density expression comes after performing the contour integral comes out to be , @xmath106\\ & \times&\frac{1}{e^ { n_t\sinh^{-1}\sqrt{f}}+1 } + \epsilon_3+\epsilon_4~,~\end{aligned}\ ] ] which again turns out to be similar to the overlap case . due to a different functional form of @xmath69 and a different choice of contour , the corresponding lattice correction terms @xmath107 which are the line integrals of @xmath69 along lines 3,4 in the figure [ econt ] , are different , leading to different finite size corrections . in the continuum limit
the unphysical poles are pushed to infinity and the values of @xmath107 vanish , leaving only the contribution of the physical poles to the energy density : in the square bracket , only first term gives the usual continuum expression with the other term vanishing as @xmath108 .
the same treatment goes through in presence of @xmath58 only the contour has to be shifted along the imaginary @xmath78 plane by an amount dependent on @xmath58 with the position of the poles in the complex @xmath78-plane remaining unchanged . . ] in the case when the lattice spacing in the fifth direction @xmath110 and the number of sites @xmath111 such that @xmath112 is finite , the effective domain wall operator reduces to @xmath113 starting from the above expression we recover the overlap operator when @xmath114 . with this effective domain wall operator ,
the energy density can be evaluated@xcite as , @xmath115 ( \left(-h_4 h_5 \alpha + h^2 \gamma \right ) \cosh[\frac{sl_5}{2}]}{sn^3n_t}\\\nonumber & & \frac{+\left(h_4 h_5 \alpha + ( h_5 ^ 2+s^2)\gamma+2h_5 s^2 t\right)\cosh ( \frac{3 s l_5 } { 2})- 2 s\sinh[\frac{s l_5}{2 } ] } { ( h^2+\left(s^2 + h_5 ^ 2\right ) \cosh[2 sl_5]}\\\nonumber & & \frac { ( h_5 ^ 2 t+h_5\gamma + ( h^2 t+h_4 \alpha + 2 h_5(h_5 t+\gamma ) ) \cosh [ sl_5 ] ) ) } { -2 h_5 s \sinh[2 s l_5])}~,\\\end{aligned}\ ] ] where @xmath116 and @xmath117 are the same as defined previously and @xmath118 is now defined as , @xmath119 it was checked that the overlap energy density is obtained back when @xmath114 .
we use the expression above for our numerical work presented in section iii . while performing monte carlo simulations with domain wall fermions one needs to work on lattices with finite number of sites in the fifth dimension . for finite @xmath0 ,
the chiral symmetry is broken and it is important to ascertain the dependence of the correction terms with @xmath0 . evaluating the matrix @xmath120 in eq .
( [ eqn : tanh ] ) various thermodynamic quantities of free domain wall fermions on the lattice can be evaluated .
the energy density in the massless limit then is @xmath121 where the quantities @xmath122 and @xmath123 are functions of h s defined in eq .
( [ eqn : hdef ] ) defined as , @xmath124 the partial derivatives of the above variables are represented as the same variables with a dash , and are functions of h s , @xmath116 and @xmath117 .
@xmath125\\\nonumber & + & n_5\left(\frac{t+q}{1-h_5}\right)^{n_5 - 1}\left [ \frac{t^{'}+q^{'}}{1-h_5}+\frac{\gamma(t+q)}{(1-h_5)^2}\right]~,\\\nonumber x^{'}&=&n_5\left(\frac{t - q}{h_5 - 1}\right)^{n_5 - 1}\left [ \frac{t^{'}-q^{'}}{h_5 - 1}-\frac{\gamma(t - q)}{(h_5 - 1)^2}\right]\\\nonumber & -&n_5\left(\frac{t+q}{1-h_5}\right)^{n_5 - 1}\left [ \frac{t^{'}+q^{'}}{1-h_5}+\frac{\gamma(t+q)}{(1-h_5)^2}\right]~.~\end{aligned}\ ] ] again , we shall use these expressions for obtaining the numerical results presented below where we also show the results for quark number susceptibility .
the same set of formulae remain valid for calculation of the susceptibility except for the fact that @xmath126 and @xmath127 then are the derivatives with respect to @xmath58 and are defined as , @xmath128
the goal of our numerical study is to find the optimum range of @xmath2 for which the finite lattice spacing corrections are minimum and compare it with that for the dirac - neuberger case @xcite .
we do this in the chiral limit and set @xmath130 . the lattice energy density given by eq .
( [ eqn : sgne ] ) was computed numerically by summing over the momenta along the spatial and temporal directions .
the zero temperature part of the energy density was determined in the limit @xmath73 on a lattice with a very large spatial extent @xmath131 by numerically evaluating the @xmath132 integral . holding the physical volume constant in units of @xmath41 by keeping @xmath133 fixed , we define the continuum limit by @xmath134 .
the thermodynamic limit is then achieved in the limit of large @xmath135 .
we first determine the acceptable range of @xmath135 by looking for @xmath135-independence .
the @xmath59 obtained by subtracting the zero temperature part from the lattice energy density expression was normalized by its continuum value @xmath136 .
figure [ dwzeta ] displays the ratio @xmath137 as a function of @xmath61 for different values of @xmath135 at a fixed @xmath138 .
one notices that for @xmath139 the energy density plots lie on top of each other , suggesting the thermodynamic limit to have reached by @xmath140 .
dependence of the energy density of domain wall fermions for @xmath138 , @xmath141 and in the limit @xmath142 . ] in order to highlight the deviations in the continuum limit , the same ratio is exhibited for different m values for @xmath143 in figure [ dwsgn ] as a function of @xmath144 and listed in table [ dwsgne ] for a range of @xmath61 likely to be used in simulations .
we choose to define the optimum range of m as the values of m for which the thermodynamic quantities are within 3% of the continuum values for the smallest possible @xmath61 .
one sees from both the figure [ dwsgn ] and the table [ dwsgne ] that the order @xmath144 corrections are minimum for m between 1.45 - 1.50 and @xmath145 .
.@xmath137 values for different m for @xmath143 [ cols="^,^,^,^,^,^",options="header " , ] [ dwen ]
it should be noted that in this case @xmath41 is no longer hermitian but as long as the condition given in the eq .
( [ eqn : cond ] ) is satisfied the effective operator in eq .
( [ eqn : sgne ] ) is well defined .
we shall restrict the range of @xmath58 to ensure that it is so .
we choose @xmath27 and @xmath28 to be @xmath147 respectively in our numerical computations as suggested in @xcite .
our aim again is to find the optimum @xmath2 for which the continuum results are obtained with least computational effort , and compare it with our the range obtained from the energy density above .
we consider two observables here .
one is the change in the energy density due to nonzero @xmath148 : @xmath149 . in the continuum limit
this is @xmath150 another observable we studied was the quark number susceptibility at @xmath43 .
it is defined for any @xmath58 by , @xmath151 and in the continuum is given by , @xmath152 we will focus on @xmath153 here due to its importance in the applications to the heavy ion collisions .
we estimated numerically @xmath154 for @xmath155 fixed at 0.5 .
the upper and lower panels of the figure [ dwsgndele ] display our results for this observable in the units of @xmath156 for @xmath157 and 18 respectively for the @xmath2 values indicated .
the horizontal line in each case shows the expected result in the continuum limit from eq.([eqn : delemu ] ) . from the figure [ dwsgndele ]
it is evident that there are no @xmath5 divergences on the lattice , as expected .
the deviations from the continuum limit are due to the @xmath2 dependent finite size effects .
these correction terms are again seen to be small for the same optimum range of @xmath158 for both the cases , as obtained in the zero chemical potential case in section iii . for different @xmath2 for @xmath159 and infinite @xmath0(upper panel ) and for @xmath160 ( lower panel).,title="fig : " ] for different @xmath2 for @xmath159 and infinite @xmath0(upper panel ) and for @xmath160 ( lower panel).,title="fig : " ] the @xmath0 dependence of the quark number susceptibility at @xmath43 is plotted in figure [ dwchi ] .
it too exhibits a convergence to the infinite @xmath0 results for @xmath161 , indicating that @xmath162 can again be used safely to approximate the infinite @xmath0 .
figure [ dwchi ] shows the @xmath2-dependence of the quark number susceptibility at @xmath43 .
both the @xmath163 ( upper panel ) and 18 ( lower panel ) show small deviations from the stefan - boltzmann value of @xmath164 for @xmath165 range and for @xmath1 . for @xmath0 values as indicated for @xmath166 and @xmath167 .
] for @xmath2 values as indicated for @xmath167 and @xmath163 ( upper panel ) and 18 ( lower panel).,title="fig : " ] for @xmath2 values as indicated for @xmath167 and @xmath163 ( upper panel ) and 18 ( lower panel).,title="fig : " ] recent computations of this susceptibility@xcite for the interacting domain wall fermions were performed with @xmath168 . of course , one expects some shift in @xmath2 due to the interactions , which should however be small for large enough temperature where one expects those computations to approach the free quark gas results . in all our plots we find that for the optimum @xmath2 range , the deviations from the ideal gas results at smaller @xmath44 - 8 are quite significant but with a relative mild @xmath2-dependence for @xmath169 .
thus a slightly larger value of @xmath2 than the optimum range we found may not change the finite size effects drastically for small @xmath61 .
what one does need to be careful about though is the extrapolation to the continuum limit .
for the optimal range of @xmath2 and @xmath170 , the smallness of corrections compared to other errors in the computations may make it a less important issue .
in the previous sections we observed that the fermions with exact chiral symmetry on the lattice have large @xmath144 corrections for small @xmath61 . while we found that the continuum limit for various thermodynamic quantities can be approached faster by choosing the irrelevant parameter @xmath2 in the range 1.45 - 1.55 , the correction terms for @xmath44 - 6 are about twice that of the stefan - boltzmann result for such a choice of @xmath2 too . here we describe our attempts to improve the convergence to the continuum results for small @xmath61 and even for @xmath171 .
having the option of the choice of @xmath8 may be useful since it has been noted previously@xcite that the residual mass for such a choice of @xmath2 is zero for a range of @xmath0 at the tree level . the domain wall operator given in eq .
( [ eqn : tanh ] ) is a matrix - function of the wilson - dirac operator as in eq .
( [ eqn : wilop ] ) .
it is clear that its improvement may lead to a better domain wall operator , or indeed even a better overlap operator , one is looking for .
inspired by the attempts to improve the staggered fermions in the so - called naik - action @xcite , we add three - link terms to the @xmath23 as below . @xmath172
comparing with the eq .
( [ eqn : wilop ] ) , it is clear that the modification amounts to replacing @xmath173 by @xmath174 .
the wilson mass term , added to remove the doublers , is kept unchanged .
note that the modified @xmath23-operator is still @xmath175-hermitian for arbitrary real values of the coefficients @xmath176 and @xmath177 .
the new domain wall operator can therefore be derived in the same way as eq .
( [ eqn : tanh ] ) was obtained .
we fix the coefficients by demanding the dispersion relation for free fermions on the lattice to be the same as in the the continuum up to @xmath178 .
we find that all the terms at @xmath179 are eliminated for the coefficients @xmath180 .
we employ them below for the calculation of the thermodynamic quantities .
the ratio of quark number susceptibilities , @xmath181 , computed using the above modified domain wall operator @xcite , is plotted as a function of @xmath144 as in figure [ dwimp ] along with that for the unimproved domain wall operator of eq .
( [ eqn : tanh ] ) .
we used @xmath182 , @xmath143 , @xmath183 and @xmath141 for this computation .
one clearly notices that the large correction terms ( @xmath184% ) at @xmath46 - 8 for the usual domain wall operator go down to about 7 - 8 % .
indeed , the size of corrections go down further as @xmath61 increases .
similarly , the the energy density of such improved fermions also exhibited smaller , about 15 - 5% , deviations from the continuum for @xmath46 - 10 , as compared to about 30 % in the lower panel of figure [ dwfin ] .
as a function of @xmath144 . ] from section ii , we know that the overlap operator can be derived as a special limiting case of the domain wall operator . it would be thus interesting to check how the improvement in the wilson dirac operator in eq .
( [ eqn : impwilop ] ) fairs in the overlap case . for that purpose
we compute the quark number susceptibility for non - interacting fermions on a @xmath185 lattice numerically with the corresponding improved overlap operator .
the @xmath181 does have lower @xmath144 corrections for @xmath186 than for the conventional overlap operator with @xmath182 as shown in figure [ impoc ] .
we also observe a faster approach to the continuum result with such improved overlap operator than with the neuberger overlap operator even with optimum @xmath187 reported in @xcite .
another advantage is that the thermodynamic quantities calculated from this improved operator are free from oscillations at odd - even values of @xmath61 exhibited @xcite by the usual overlap operator .
the improvement in the energy density is marginal up to @xmath188 but substantial for @xmath189 onwards . as a function of @xmath144 for the improved and the canonical overlap operators at @xmath8 . ]
since the chiral violations vanish exponentially with the number of sites @xmath0 in the fifth dimension , the domain wall fermions offer a more practical alternative to the overlap fermions and yet have exact flavor and spin symmetry .
we have computed the energy density and susceptibility at zero chemical potential of such fermions numerically for both finite and infinite @xmath0 .
the chiral symmetry is exact in the latter case and a choice @xmath2 between 1.45 - 1.50 allows faster convergence to the continuum results .
we have also verified analytically that the energy density has the correct continuum value in the chiral limit .
varying the number of lattice sites in the fifth dimension , we have shown that @xmath160 is sufficient to restore chiral symmetry .
we found that introducing chemical potential @xmath58 in domain wall operator leads to chiral symmetry breaking even for infinite @xmath0 .
but if we do allow that , there exist a large class of functions @xmath190 and @xmath191 , with @xmath192 , for which there are no @xmath58-dependent divergent terms in the physical observables . from the numerical evaluation of the energy density in presence of @xmath58 ,
we conclude that the optimum range of @xmath2 remains the same .
the lattice cut - off effects are however very large for small @xmath61 = 4 - 8 . by systematically removing the dominant correction terms to the continuum value of the chiral fermion operators we have achieved a faster convergence to the continuum as well as small @xmath144 corrections for small lattice sizes even for @xmath8 .
this set of optimum parameters is anticipated to produce similar results in full qcd simulations with chiral fermions though an explicit check needs to be done .
s.s would like to acknowledge the council of scientific and industrial research(csir ) for financial support .
we show here that the effective domain wall operator eq .
( [ eqn : effdw ] ) satisfies the ginsparg wilson relation : @xmath193 hence @xmath194 10 d. b. kaplan , _ phys .
b288 , 342 ( 1992 ) .
m. vranas , _ phys .
d57 , 1415 ( 1998 ) .
y. shamir , _ nucl .
b406 , 90 ( 1993 ) .
j. bloch and t. wettig , _ phys . rev . _
d76 , 114511 ( 2007 ) .
r. v. gavai , _ phys .
d32 , 519 ( 1985 ) .
h. neuberger , _ phys .
d57 , 5417 ( 1998 ) .
r. g. edwards and u. m. heller , _ phys .
d63 , 094505 ( 2001 ) .
r. narayanan and h. neuberger , _ phys .
_ 71 , 3251 ( 1993 ) ; h. neuberger , _ phys .
lett . _ 417b , 141 ( 1998 ) .
p. h. ginsparg and k. g. wilson , _ phys .
d25 , 2649 ( 1982 ) . m. luscher , _ phys .
b428 , 342 ( 1998 ) .
d. banerjee , r. v. gavai and s. sharma , _ phys .
d78 , 014506 ( 2008 ) .
r. v. gavai and s. sharma , _ j. phys .
_ g 35 , 104097 ( 2008 ) .
g. t. flemming , _ nucl .
_ 94 , 393 ( 2001 ) .
p. hegde _ et al .
_ , _ eur .
j. _ , c55 , 423 ( 2008 ) .
p. hasenfratz and f. karsch , _ phys .
lett . _ 125b , 308 ( 1983 ) .
p. hegde , f. karsch and c. schmidt , e - print arxiv:0810.0290[hep - lat ] .
s. capitani , _ phys .
d75 , 054505 ( 2007 ) .
s. naik , _ nucl .
b316 , 238 ( 1989 ) . following @xcite , we use @xmath195 and @xmath196 for introducing @xmath58 for the 3-link terms .
r. v. gavai , _ nucl .
_ 119 , 529 ( 2003 ) . | studying various thermodynamic quantities for the free domain wall fermions for both finite and infinite fifth dimensional extent @xmath0 , we find that the lattice corrections are minimum for @xmath1 for both energy density and susceptibility , for its irrelevant parameter @xmath2 in the range 1.45 - 1.50 .
the correction terms are , however , quite large for small lattice sizes of @xmath3 .
we propose modifications of the domain wall operator , as well as the overlap operator , to reduce the finite cut - off effects to within 10% of the continuum results of the thermodynamic quantities for the currently used @xmath46 - 8 lattices .
incorporating chemical potential , we show that @xmath5 divergences are absent for a large class of such domain wall fermion actions although the chiral symmetry is broken for @xmath6 . |
since the seminal discovery of unconventional superconductivity at @xmath6 1 k , sr@xmath1ruo@xmath2 ( sro214 ) has been studied extensively @xcite .
the crystal structure of sro214 is of k@xmath1nif@xmath2-type at low temperature as shown in fig .
[ structure ] and its normal phase is a paramagnetic metal .
its intriguing electronic behavior and superconductivity are still a subject of active study @xcite .
srruo@xmath3 ( sro113 ) is a ferromagnetic metal with a transition temperature of t@xmath7 k. the observed magnetic moment is @xmath8= 1.11.7 @xmath9f.u . @xcite and the stable structure at low temperature is an orthorhombic perovskite with a gdfeo@xmath3-type distortion .
the electronic structure of sro113 is located near to the half - metallicity as reported by density functional theory ( dft ) + @xmath10 calculations @xcite .
the thin film sro113 is of particular interest as a widely - used bottom electrode @xcite and is intensively studied for the possibility of a new field - effect device @xcite .
considering their fundamental and technological importance , it is essential to understand correctly the electronic structures of sro214 and sro113 . for sro214
, the early electronic structure calculations @xcite showed that the states near the fermi level ( @xmath11 ) consist of antibonding combinations of ru-@xmath12 and o-2@xmath13 , and the fermi surface is composed of the @xmath14-centered electron - like sheets ( called as @xmath15 and @xmath16 ) and the x - point - centered hole - like sheet ( called as @xmath17 ) @xcite .
the lda ( local density approximation ) fermi surface topology was in good agreement with that of de haas - van alphen ( dhva ) experiments @xcite , but does not seem to be with the angle - resolved photoemission spectroscopy ( arpes ) @xcite in the sense that one electron - like ( @xmath15 ) and two hole - like ( @xmath17 and @xmath16 ) fermi surface were observed by arpes @xcite .
it however turns out to be a surface effect @xcite later .
therefore the overall features of the electronic structure can be regarded as being described well by the conventional electronic structure calculation techniques such as lda and gga ( generalized gradient approximation ) .
however the details of the electronic structure are still not clearly understood .
a series of recent studies that identify sro214 as a ` hund s metal ' @xcite can be one example showing that the previous understanding based on the conventional electronic structure calculations was not enough .
also , there was a debate regarding the detailed feature of electronic levels at @xmath18 ev revealed by x - ray photoemission spectroscopy ( xps ) @xcite . according to xps and resonant xps ,
the indication of lower hubbard band ( lhb ) of ru-4@xmath5 states is observed while the lda calculation predicts that those features should be attributed mainly to the oxygen states @xcite .
the experimental effective mass ( @xmath19 25 ) is also significantly larger than the lda value @xcite .
while dmft can be a reasonable alternative in this situation @xcite , its parameter dependency has caused the discrepancy even among the dmft results .
for example , the calculations by pchelkina _ et al .
_ @xcite and by liebsch _
@xcite seem to give different answers to the nature of the states at @xmath20 ev most likely due to their different choices of @xmath10 and @xmath21 values . beside the ambiguity in the double - counting term in dft+dmft , this parameter dependency is a well known problem for the first - principles description of the correlated electron systems .
it is also noted that another well - established technique , dft@xmath22 , is not useful for paramagnetic sro214 because of its hartree - fock nature which always prefers the magnetic solution @xcite .
for sro113 the situation is similar with sro214 ; while the overall features can be described by the conventional methods , the details are not clearly understood .
note that dft@xmath22 can be used for this case as the ground state of sro113 is ( ferro ) magnetic . for the bulk phase , lda , lda+@xmath10 , and self - interaction correction ( sic ) give the correct ferromagnetic solution although the calculated moment shows some deviations @xcite .
for the thin film sro113 , however , the experimentally observed metal - to - insulator transition ( mit ) and the ferromagnetic - to - nonmagnetic transition as a function of layer thickness are not consistently reproduced by these techniques @xcite .
for example , the single - layer sro113 is predicted to be either ferromagnetic or nonmagnetic depending on the choice of the exchange - correlation functional and the @xmath10 values @xcite . as in the case of sro214
, dmft has made important contributions for sro113 @xcite , but it is still not completely satisfactory because of the limitations such as the parameter dependency .
also , the details of the spectroscopic data of sro113 needs further clarifications @xcite . in this paper , we re - investigate the electronic structure of these two classical @xmath23 transition - metal oxide systems by using quasiparticle self - consistent @xmath0 ( qs@xmath0 ) @xcite method .
to the best of our knowledge , this well - established calculation technique has never been applied to the ruthenates while @xmath0 self - energy can be expected to give a good description for the metallic and weakly correlated systems . despite of its limitation to take the hubbard - like on - site coulomb interactions into account , qs@xmath0 has a distinctive advantage as a fully self - consistent ` parameter - free ' technique . due to the recent experimental progress in making ` complex oxide ' structures such as thin film and heterointerface ,
the parameter - free description of correlated systems has become more imperative . in this
context , qs@xmath0 study of these classical @xmath23 systems can be of significant importance , and its ability and limitation should be carefully investigated .
our calculations show that the effect of electronic correlations beyond lda and gga is reasonably well captured by qs@xmath0 procedure , and the noticeable bandwidth reduction is observed . for the paramagnetic sro214 , the effect of @xmath0 self - energy is clearly distinctive from the other many - body calculation ( _ e.g. , _
dmft ) in that the spectral weight transfer and the lhb feature do not appear in the qs@xmath0 result . for the ferromagnetic sro113 ,
qs@xmath0 band structure is quite similar with the results of dft+@xmath10 and the half - metallic band structure is reproduced .
in qs@xmath0 , the self - energy @xmath24 is calculated within the @xmath0 approximation , and @xmath25 ( the noninteracting hamiltonian describing quasiparticles or band structures ) and @xmath26 ( dynamically - screened coulomb interactions between quasiparticles within random phase approximation ( rpa ) ) is updated self consistently @xcite .
the static nonlocal one - particle exchange - correlation potential @xmath27 is generated from @xmath24 as @xmath28\delta(\omega - h^0 ) + c.c .
\\ & = & \sum_{ij } |\psi_i\rangle\langle\psi_i| \frac{{\rm re}[\sigma(\varepsilon_i)+\sigma(\varepsilon_j)]}{2 } |\psi_j\rangle\langle\psi_j| \end{aligned}\ ] ] where @xmath29 and @xmath30 refer to the eigenvalues and eigenfunctions of @xmath25 , respectively , and @xmath31 $ ] is the hermitian part of the self - energy @xcite . with this @xmath32 , one can solve a new static one - body hamiltonian @xmath25 , and continue to apply @xmath0 approximation until self - consistency is achieved .
the distinctive feature of qs@xmath0 compared with dft+@xmath10 and dmft is that it does not require any _ ad hoc _ parameters .
previous studies , ranging from atoms @xcite , semiconductors @xcite to the various 3@xmath5 transition metal oxides @xcite and 4@xmath33 electron systems @xcite have demonstrated its capacity to investigate many different types of correlated materials .
cage ) , and o atoms , respectively.,scaledwidth=45.0% ] our implementation of qs@xmath0 in ` ecalj ' code @xcite is based on the ` augmented plane wave ( apw ) + muffin - tin orbital ( mto ) ' method , designated by ` pmt ' @xcite .
the accuracy of this scheme was proven to be satisfactory @xcite .
a key feature of this scheme for qs@xmath0 is that the expansion of @xmath32 can be made with mtos , not apws , which enables us to make the real space representation of @xmath32 at any @xmath34 point .
in contrast to the previous approach based on fp - lmto ( full potential linearized muffin - tin orbital ) @xcite , our scheme is free from the fine tuning of mto parameters .
we used the in - plane and out - of - plane lattice constant of 3.87 and 12.73 , respectively , for the body - centered tetragonal sro214 .
for sro113 , the cubic perovskite structure is considered as shown in fig .
[ structure](b ) .
the pseudocubic lattice constant of the orthorhombic structure is 3.93 @xcite .
we used 6@xmath356@xmath356 @xcite and 8@xmath358@xmath358 * k * points for the self - energy calculation in the first brillouin zone of sro214 and sro113 , respectively .
for cubic sro113 we also performed dft+@xmath10 calculations using openmx code @xcite to make a comparison with previous studies .
both lda @xcite and gga @xcite functionals were used in combination with dudarev form of dft+@xmath10 @xcite .
16@xmath3516@xmath3516 * k * points were adopted for dft@xmath22 calculations .
( red solid lines ) .
the spectral function @xmath36 calculated by qs@xmath0 self - energy is plotted in ( e ) ( green solid line ) along with the non - interpolated total dos ( red solid line ) .
the fermi level is set to zero.,scaledwidth=48.0% ] pdos for sro214 by ( a ) lda and ( b ) qs@xmath0 .
the red , cyan , blue , green and gray lines refer to the ru @xmath37 , @xmath38 , @xmath39 , @xmath40 and @xmath41 states , respectively .
the yellow lines represent the o and sr states ( not shown in dos ) .
the thickness of the bands corresponds to the amount of the corresponding orbital character .
the red and blue arrows represent the bandwidth of each band .
the fermi level is set to zero .
the insets highlight the band dispersion at around @xmath42 ev along @xmath14 to m ( @xmath43,0,0 ) by ( a ) lda and ( b ) qs@xmath0 . a sudden change of its slope at ( 0.55@xmath43,0,0 ) is observed only in qs@xmath0 ( red arrow ) .
this feature is reflected in pdos ( see fig .
[ t2g](b)).,scaledwidth=45.0% ]
our lda result is in good agreement with the previous calculations @xcite , see fig .
[ pdos ] ( black dotted lines ) and fig . [ fat_band ] .
the van hove singularity ( vhs ) is located at @xmath44 mev above the @xmath11 and the density of states ( dos ) at @xmath11 is @xmath45 states/(ev f.u . ) .
the ru-@xmath12 state is dominant near @xmath11 and is hybridized with o-2@xmath13 . from fig .
[ pdos](a ) , ( c ) , and ( d ) , one can see that the total dos in the range of @xmath46 @xmath47
ev is largely attributed to the apical o-2@xmath13 .
the calculated band dispersion is presented in fig .
[ fat_band](a ) .
the bandwidth of the 2d - like @xmath37 state is about two times larger than that of @xmath48 states .
the nonbonding o-2@xmath13 state is located in @xmath46 @xmath47 ev as can also be seen in fig . [
pdos](d ) .
the qs@xmath0 dos is presented in fig .
[ pdos ] ( red solid lines ) @xcite . while the overall shape of the dos is not much different from the lda results , some differences are observed .
the two vhs located at @xmath49 and @xmath50 mev in lda @xcite become closer in qs@xmath0 shifting to @xmath51 and @xmath52 mev , respectively ( inset in fig .
[ pdos](b ) ) .
both the bonding ( at @xmath53 ev ) and the antibonding ( around the @xmath11 ) part of ru-4@xmath5 states becomes flatter in their dispersion compared to the lda result .
the @xmath12 bandwidth is slightly reduced ( see fig .
[ pdos ] and [ fat_band ] ) .
an interesting feature is found in the @xmath37 band dispersion along @xmath14 to m. the different dispersion at @xmath54 ev ( see fig . [ fat_band ] insets ) is related to the more pronounced peak in the qs@xmath0 pdos ( fig .
[ t2g](b ) ) .
the o-@xmath55 states are shifted to the lower energy regions in qs@xmath0 ( fig .
[ pdos](c)-(d ) ) .
the sr-4@xmath5 state is further pushed up to @xmath56 ev ( not shown ) .
( red solid lines ) .
the @xmath57 states in the previous lda and dmft study were identical @xcite and are not shown here .
( d ) the integrated @xmath5 dos defined as @xmath58 , where @xmath59 is the dos at energy @xmath60 . each value was normalized to have an equal number of states at @xmath61 .
@xmath11 is set to zero.,scaledwidth=48.0% ] the further details of ru-@xmath12 state is of particular interest as the lhb feature at around @xmath623 ev has been a subject of debate @xcite .
[ t2 g ] compares the calculated pdos by lda , lda+dmft , and qs@xmath0 .
note that the states in the range of @xmath624
@xmath621 ev is noticeably larger in lda+dmft than lda and qs@xmath0 especially for @xmath48 .
it is related to the spectral weight transfer from the near-@xmath11 regions to the lhb .
this is one of the main features of dmft calculations and was the main point of the previous debate between lda , dmft @xcite and the xps experiment @xcite , while the qs@xmath0 result in this region is similar with the lda rather than the dmft .
namely , qs@xmath0 procedure does not capture the dynamic ` mott - hubbard physics ' well as in dmft while it still takes some correlation effect into account as reflected in the bandwidth reduction and the mass enhancement .
this can also be seen in fig .
[ spectra](a ) .
one can notice the shoulder - like lhb feature in xps spectrum at @xmath63 3 ev being consistent with lda+dmft , while both lda and qs@xmath0 give the less states in this energy region .
the theoretical spectra in fig .
[ spectra](a ) were broadened as in ref .
not surprisingly , this lhb feature of dmft is more pronounced in @xmath48 states than in @xmath37 because of the narrower bandwidth .
this point is highlighted in the integrated dos ( fig .
[ t2g](d ) ) .
while the total number of ru states ( the integrated value up to @xmath11 ) is basically the same in all three calculations , the dmft value gets increased in a wider energy range . on the other hand , lda and qs@xmath0 values
are much more rapidly increased in the narrower energy range of @xmath64 ev . since the qs@xmath0 bandwidth is narrower than lda , this increase is more pronounced in qs@xmath0 .
our result demonstrates the characteristic feature of qs@xmath0 to take into account of electron correlations distinctive from dmft especially regarding the incoherent states at @xmath4 ev .
the ru-@xmath57 bands are also affected by qs@xmath0 self - energy ( see fig . [ fat_band ] ) .
the slight up - shift of the @xmath57 anti - bonding bands is consistent with the previous study of srvo@xmath3 @xcite . as a result
, the @xmath40 band does not touch the @xmath11 even at @xmath14 and @xmath65 point ( see fig .
[ fat_band ] ) .
the o-2@xmath13 nonbonding state is located at @xmath66
@xmath67 ev in qs@xmath0 ( fig .
[ fat_band ] ) , which is in better agreement with the recent arpes result @xcite than lda . plane .
the color represents the calculated effective mass defined as @xmath68.,scaledwidth=27.0% ] the overall shape of the calculated fermi surface by qs@xmath0 ( see fig .
[ fs_mass ] ) is consistent with the lda @xcite , dhva @xcite and arpes @xcite .
the inclusion of spin - orbit coupling as a perturbative correction within qs@xmath0 induces relatively small modifications at the band crossing points ( not shown ) @xcite .
the calculated effective mass at each * k * direction is shown in fig .
[ fs_mass ] with color plot where @xmath69 is estimated simply by taking the derivative of band dispersions . along the @xmath14 to x ( @xmath43,@xmath43,0 ) line ,
the mass enhancement by qs@xmath0 is about 15% ( @xmath70 ) for @xmath37 and 22% ( @xmath71 ) for @xmath48 , respectively . for the @xmath14 to m ( @xmath43,0,0 )
line , @xmath72 and @xmath73 .
it is instructive to compare the qs@xmath0 result with the dmft calculations and experiments while the dmft effective mass strongly depends on the @xmath10 and @xmath21 values .
the value reported by mravlje _
et al . _ with @xmath74 and @xmath75 0.1 ev is @xmath76 for both @xmath37 and @xmath48 @xcite .
this value is significantly smaller than the other dmft calculations , _
e.g. _ , by pchelkina _
et al . _ , @xmath69 @xmath77 and 2.3 for @xmath37 and @xmath48 , respectively ( with @xmath78 and @xmath79 ev obtained from constrained lda ) @xcite , and by the same group but with @xmath80 and @xmath81 ev , @xmath69 is @xmath82 and 2.0 for @xmath37 and @xmath48 , respectively @xcite .
these dmft values were calculated from @xmath83_{\omega = 0}$ ] . our estimation from qs@xmath0 self - energy yields @xmath84 for @xmath37 and 1.71 for @xmath48 @xcite which is not much different from the dmft values of 1.7 calculated by mravlje _
et al . _ with @xmath74 and @xmath75 0.1 ev @xcite .
note that @xmath80 and @xmath85 ev were suggested to be reasonable @xcite in comparison to dhva measurements @xcite , and that the experimental values are larger than the qs@xmath0 .
the early arpes reports @xmath86 @xcite and the more recent measurement by iwasawa _
is @xmath87 and @xmath88 @xcite .
there are two reports from the dhva experiment ; @xmath89 and @xmath90 by mackenzie _ et al . _
@xcite , and @xmath91 and @xmath92 by bergemann _
_ @xcite .
the lda result for cubic sro113 is in good agreement with the previous studies @xcite ( see fig . [
113_pdos ] and fig . [ 113_fat_band ] ) .
the calculated magnetic moment of @xmath93/f.u .
also reasonably well agrees with the literature values of @xmath94/f.u .
( calculated by vasp ) and @xmath95/f.u .
( calculated by siesta ) @xcite .
the sr-4@xmath5 state is located above @xmath56 ev ( not shown ) and the nonbonding state of o-2@xmath13 is in @xmath46
@xmath47 ev .
the antibonding ru-@xmath12 character dominates the near-@xmath11 region and the bonding complex is located at @xmath96 @xmath46 ev . a clear splitting between the up and down spin dos
is noticed and is responsible for the ferromagnetism in this material .
the effect of orthorhombic distortion ( not taken into account in our calculation ) has been investigated previously .
for example , rondinelli _ et al . _
@xcite showed that the result from orthorhombic structure is similar to the cubic case other than the slightly reduced exchange splitting at @xmath14 point and bandwidth reduction by @xmath97 ev for @xmath12 , 1.5 ev for @xmath57 , and 0.6 ev for o 2@xmath13 .
( red solid lines ) .
the fermi level is set to zero .
the positive and negative dos represent the up- and down - spin parts , respectively.,scaledwidth=48.0% ] -orbital pdos for cubic sro113 by ( a ) lda majority spin band , ( b ) lda minority spin band , ( c ) qs@xmath0 majority spin band , and ( d ) qs@xmath0 minority spin band .
the red , cyan , blue , and gray lines refer the ru @xmath37 , @xmath38 , @xmath39 , and @xmath57 states , respectively .
the yellow lines represent the o and sr states ( not shown in dos ) .
the thickness of the bands corresponds to the amount of the corresponding orbital character .
the fermi level is set to zero . , scaledwidth=48.0% ] and @xmath57 pdos of cubic sro113 by lda+@xmath10 ( @xmath98 ev ) ( blue dashed lines ) , gga+@xmath10 ( @xmath99 ev ) ( green dotted lines ) and qs@xmath0 ( red solid lines ) .
the fermi level is set to zero .
the positive and negative dos represent the up- and down - spin parts , respectively.,scaledwidth=48.0% ] several distinctive features are found in the qs@xmath0 results .
first , the notable bandwidth reduction is observed ( see fig .
[ 113_pdos ] and fig . [ 113_fat_band ] ) .
the majority spin bandwidth is reduced by @xmath100 ev ( from @xmath101 ev ( lda ) to @xmath102 ev ( qs@xmath0 ) ) and the minority spin bandwidth is by @xmath103 ev ( from @xmath101 ev ( lda ) to @xmath104 ev ( qs@xmath0 ) ) .
the exchange splitting is enhanced to be @xmath105 ev which is significantly larger than the lda value of @xmath106 ev .
we found the naive comparison of the effective mass for lda and qs@xmath0 based on the * k*-derivative can be misleading in sro113 due to the different fermi wave vectors caused by the enhanced exchange splitting in qs@xmath0 .
the calculated @xmath107 for the up and down spin is 1.33 and 1.68 , respectively .
the effective mass of @xmath108 estimated from the bandwidth ratio is reasonably well compared with some of dmft results while the dmft values show significant deviations ranging from 1.1 to 4.5 @xcite .
the specific heat measurements report the mass enhancement of @xmath109 @xcite and 4.5 @xcite .
the discrepancy reflects not only the limitation of rpa - qs@xmath0 correlations but also the effect of orthorhombic distortion @xcite .
together with the bandwidth reduction and the enhanced exchange splitting , the qs@xmath0 electronic structure becomes half - metallic with a gap in the majority spin state .
note that this qs@xmath0 band structure is quite similar with dft@xmath22 result @xcite as shown in fig .
[ 113_u ] .
the calculated magnetic moment of qs@xmath0 is @xmath110f.u . which is significantly larger than the lda value and comparable with the dft@xmath22 results @xcite .
while we only considered the cubic structure , the effect of orthorhombic distortion is basically to enhance the on - site correlations as reported by the previous study @xcite .
the experimental verification of the half - metallicity may not be easy because of the large magnetic fields required to overcome the magnetic anisotropy @xcite .
this possibility in sro113 has been actively discussed based on dft@xmath22 calculations @xcite and hybrid functional @xcite .
therefore our qs@xmath0 result adds a new promising aspect toward this direction . the detailed comparison with dft@xmath22 result is given in fig . [ 113_u ] .
the optimized values of @xmath111 are favorably compared with a recent constrained rpa result of @xmath112 ev although it is calculated from the orthorhombic structure @xcite .
for comparison one can recall that sic - lda result is reproduced by @xmath113 ev in the case of orthorhombic structure of sro113 @xcite . as for the dmft calculations ,
several different choices of @xmath10 and @xmath21 have been made as in the case of sro214 . in terms of @xmath111=@xmath114
, it ranges from 1.75 to 2.4 ev @xcite .
several photoemission spectroscopy ( pes ) experiments report the lhb - like feature at @xmath632 ev @xcite .
the pes spectrum of orthorhombic phase @xcite are plotted along with the calculated dos by lda , lda+dmft @xcite , and qs@xmath0 in fig . [ spectra](b ) .
the lhb - like peak near @xmath62 1.5 ev is observed in the pes spectra , while not in lda and qs@xmath0 .
it has been argued that this lhb - like feature is not related to the magnetic fluctuations or the orthorhombic structural distortion @xcite .
lda+dmft calculation of orthorhombic sro113 with @xmath115 and @xmath116 ev predicts the lhb at @xmath4
@xmath117 ev @xcite ( fig .
[ spectra](b ) ) . as in sro214
, lhb is the characteristic feature of dmft distinctive from qs@xmath0 ( also from dft@xmath22 ) while some dmft results do not seem to support this feature @xcite . the absence of this state in qs@xmath0 and dft@xmath22 ( see @xmath12 states of fig .
[ 113_u ] ) indicates that it is related to the dynamic aspects of ru-@xmath23 correlation hybridized with o-@xmath55 as in the case of zhang - rice band in cuprates .
also , regarding the half - metallic band structure , dmft calculations do not seem to give a consistent prediction @xcite .
these issues that is likely related to the intrinsic parameter dependency require further investigations .
using qs@xmath0 calculations , we re - investigated the electronic structure of sro214 and sro113 . without any _ ad hoc _ parameter or the ambiguity related to the double - counting and downfolding issues , some of the important features of electron correlations were reasonably well captured such as the bandwidth renormalization and the exchange splitting . in the case of sro113 ,
qs@xmath0 result is in good agreement with dft+@xmath10 in a reasonable range of @xmath10 and @xmath21 parameter .
while the qs@xmath0 shows the limitation in describing the detailed features such as the ru-@xmath23 spectral weight transfer to lhb , it can be improved in combination with other techniques as reported in recent studies @xcite .
our result sheds new light on the possibility and the limitation of the first - principles electronic structure calculations of the moderately correlated transition - metal oxides systems .
s.r . , s.w.j . , and m.j.h .
were supported by basic science research program through the national research foundation of korea ( nrf ) ( 2014r1a1a2057202 ; ministry of education ) and by samsung advanced institute of technology ( sait ) .
the computing resource is supported by national institute of supercomputing and networking / korea institute of science and technology information with supercomputing resources including technical support ( ksc-2013-c2 - 024 ) and by computing system for research in kyushu university .
t.k . was supported by the advanced low carbon technology research and development program ( alca ) , the `` high - efficiency energy conversion by spinodal nano - decomposition '' program of the japan science and technology agency ( jst ) , and the jsps core - to - core program advanced research networks ( computational nano - materials design on green energy ) .
y. maeno , h. hashimoto , k. yoshida , s. nishizaki , t. fujita , j. g. bednorz , and f. lichtenberg , nature ( london ) * 372 * , 532 ( 1994 ) ; g. m. luke , y. fudamoto , k. m. kojima , m. i. larkin , j. merrin , b. nachumi , y. j. uemura , y. maeno , z. q. mao , y. mori , h. nakamura , and m. sigrist , nature ( london ) * 394 * , 558 ( 1998 ) ; k. ishida , h. mukuda , y. kitaoka , k. asayama , z. q. mao , y. mori , and y. maeno , nature ( london ) * 396 * , 658 ( 1998 ) ; t. m. rice and m. sigrist , j. phys . : condens .
matter * 7 * , l643 ( 1995 ) ; g. baskaran , physica b * 224 * , 490 ( 1996 ) .
h. n. lee , h. m. christen , m. f. chisholm , c. m. rouleau , and d. h. lowndes , appl .
. lett . * 84 * , 4107 ( 2004 ) . c. h. ahn , t. tybell , l. antognazza , k. char , r. h. hammond , m. r. beasley , .
fischer , and j .-
triscone , science * 276 * , 1100 ( 1997 ) . c. h. ahn , r. h. hammond , t. h. geballe , m. r. beasley , j .-
triscone , m. decroux , .
fischer , l. antognazza , and k. char , appl .
. lett . * 70 * , 206 ( 1997 ) .
t. oguchi , phys .
b * 51 * , 1385 ( 1995 ) .
a. p. mackenzie , s. ikeda , y. maeno , t. fujita , s. r. julian , g. g. lonzarich , j. phys .
. jpn . * 67 * , 385 ( 1998 ) .
a. p. mackenzie , s. r. julian , a. j. diver , g. j. mcmullan , m. p. ray , g. g. lonzarich , y. maeno , s. nishizaki , and t. fujita , phys .
lett . * 76 * , 3786 ( 1996 ) .
h. iwasawa , y. aiura , t. saitoh , i. hase , s. i. ikeda , y. yoshida , h. bando , m. higashiguchi , y. miura , x. y. cui , k. shimada , h. namatame , and m. taniguchi , phys .
b * 72 * , 104515 ( 2005 ) . c. bergemann , s. r. julian , a. p. mackenzie , s. nishizaki , and y. maeno , phys . rev .
* 84 * , 2662 ( 2000 ) .
d. h. lu , m. schmidt , t. r. cummins , s. schuppler , f. lichtenberg , and j. g. bednorz , phys .
lett * 76 * , 4845 ( 1996 ) .
t. yokoya , a. chainani , t. takahashi , h. ding , j. c. campuzano , h. katayama - yoshida , m. kasai , and y. tokura , phys .
b * 54 * , 13311 ( 1996 ) .
a. damascelli , d. h. lu , k. m. shen , n. p. armitage , f. ronning , d. l. feng , c. kim , z .- x .
shen , t. kimura , y. tokura , z. q. mao , and y. maeno , phys , rev .
lett * 85 * , 5194 ( 2000 ) .
k. m. shen , a. damascelli , d. h. lu , n. p. armitage , f. ronning , d. l. feng , c. kim , z .- x .
shen , d. j. singh , i. i. mazin , s. nakatsuji , z. q. mao , y. maeno , t. kimura , and y. tokura , phys .
b * 64 * , 180502(r ) ( 2001 ) . v. i. anisimov , f. aryasetiawan , and a. i. lichtenstein , j. phys . : condens .
matter * 9 * , 767 ( 1997 ) ; g. kotliar , s. y. savrasov , k. haule , v. s. oudovenko , o. parcollet , and c. a. marianetti , rev .
phys . * 78 * , 865 ( 2006 ) .
kim , h .- j.noh , k. h. kim , and s .- j .
oh , phys . rev .
lett * 93 * , 126404 ( 2004 ) ; j. park , s .-
oh , j .- h . park , d. m. kim , and c .- b .
eom , phys .
b * 69 * , 085108 ( 2004 ) ; w. siemons , g. koster , a. vailionis , h. yamamoto , d. h. a. blank , and m. r. beasley , phys . rev .
b * 76 * , 075126 ( 2007 ) ; e. b. guedes , m. abbate , k. ishigami , a. fujimori , k. yoshimatsu , h. kumigashira , m. oshima , f. c. vicentin , p. t. fonseca , and r. j. o. mossanek , phys .
b * 86 * , 235127 ( 2012 ) .
t. kotani , j. phys .
* 83 * , 094711 ( 2014 ) ; t. kotani and m. van schilfgaarde , phys .
b * 81 * , 125117 ( 2010 ) ; t. kotani and h. kino , j. phys .
. jpn . * 82 * , 124714 ( 2013 ) ; t. kotani , h. kino , and h. akai , j. phys .
jpn . * 84 * , 034702 ( 2015 ) .
it is found that the inclusion of m - point ( @xmath43,0,0 ) is important due to the existence of van hove singularity . as a result
, the result of 7@xmath357@xmath357 * k * points shows the unusually large deviation from the 6@xmath356@xmath356 result . due to the large computation cost , we could not further increase the number of * k * points . | using quasiparticle self - consistent @xmath0 calculations , we re - examined the electronic structure of sr@xmath1ruo@xmath2 and srruo@xmath3 .
our calculations show that the correlation effects beyond the conventional lda ( local density approximation ) and gga ( generalized gradient approximation ) are reasonably well captured by qs@xmath0 self - energy without any _ ad hoc _ parameter or any ambiguity related to the double - counting and the downfolding issues . while the spectral weight transfer to the lower and upper hubbard band is not observed , the noticeable bandwidth reduction and effective mass enhancement are obtained .
important features in the electronic structures that have been debated over the last decades such as the photoemission spectra at around @xmath4 ev in sr@xmath1ruo@xmath2 and the half - metallicity for srruo@xmath3 are discussed in the light of our qs@xmath0 results and in comparison with the previous studies .
the promising aspects of qs@xmath0 are highlighted as the first - principles calculation method to describe the moderately correlated 4@xmath5 transition metal oxides along with its limitations . |
a number of recent observations @xcite suggest that the @xmath0 , the ratio of the matter density(baryonic+dark ) to the critical density , is significantly less than unity suggesting that either the universe is open or that there is some other sources of this missing energy which makes @xmath1 .
the recent findings of boomerang experiments @xcite strongly suggests the second possibility of a flat universe . at the same time , the measurements of the luminosity - redshift relations observed for the 50 newly discovered type ia supernova with redshift @xmath2 @xcite indicate that at present the universe is expanding in an accelerated fashion suggesting a net negative pressure for the universe .
initial suggestions were to identify this missing energy density to a cosmological constant @xmath3@xcite . for a flat matter dominated universe with @xmath3 in einstein gravity observations strongly suggest @xmath4 . however , this possibility that @xmath3 could be the dominant energy density has the drawback that the energy scale involved is lower than normal energy scale predicted by the most particle physics models by a factor of @xmath5 .
an alternative source of energy density that may be admissible for this acceleration could be a dynamical @xmath3@xcite in the form of a scalar field with some self interacting potential @xcite .
if the energy density of this kind of source varies slowly with time , it mimics an effective cosmological constant .
the idea of this candidate , called _ quintessence _
@xcite , is borrowed from the inflationary paradigm of the early universe .
the difference , however , is that this new field evolves at a much lower energy scale
. the energy density of this field , though dominant at present epoch , must remain sub - dominant at very early stages and should have evolved in such a way that it becomes comparable to the matter density @xmath6 now . when quintessence is modeled using a minimally coupled scalar field , in general , parameters need to be fine - tuned so as to ensure that @xmath6 and @xmath7 are of the same order today .
this fine tuning problem has been termed as the cosmic coincidence problem . a new form of quintessence field called `` _
tracker field_''@xcite has been proposed to solve the cosmic coincidence problem .
it has an equation of motion with an attractor like solution in the sense that for a wide range of initial conditions the equation of motion converges to the same solution .
there are a number of quintessence models which have been suggested and most of these involve scalar fields with minimal coupling with potentials dominating over the kinetic energy of the field .
a purely exponential potential is one of the widely studied cases @xcite .
inspite of the several advantages the energy density is not enough to make up for the missing part .
inverse power law is another form of the potential ( @xcite-@xcite ) that has been considered extensively for quintessence models , in particular , for solving the cosmic coincidence problem .
though many of the problems are resolved successfully with this potential , the predicted value for the equation of state for the quintessence field , @xmath8 , is not in good agreement with the observed results . in search of suitable models that would eliminate the problems , new types of potentials , like @xmath9^p$]@xcite and @xmath10@xcite have been considered , which have asymptotic forms like the inverse power law or exponential ones .
different physical considerations have lead to the study of other types of the potentials also@xcite .
recently saini _ et al _ @xcite have reconstructed the potential in context of general relativity and minimally coupled quintessence field from the expression of the luminosity distance @xmath11 as function of redshift obtained from the observational data . however
, none of these potentials are entirely free of problems .
hence , there is still a need to identify appropriate potentials to explain current observations @xcite .
also it has been recently shown by pietro and demaret@xcite that for constant scalar field equation of state , which is a good approximation for a tracker field solutions , the field equations and the conservation equations strongly constrain the scalar field potential .
most of the widely used potential for quintessence , such as inverse power law one , exponential or the cosine form , are incompatible with these constraints .
the cdm is in general considered to be a perfect fluid . however , in some scenarios , certain physical processes can make the cdm fluid effectively a dissipative one .
in such a situation the fluid has an effective pressure that is negetive .
recently it has been proposed that the cdm must self interact in order to explain the detailed structure of the galactic halos @xcite .
this self interaction will create a viscous pressure whose magnitude will depend on the mean free path of the cdm particles . in a recent work chimento et.al
have shown that a mixture of minimally coupled self interacting scalar field and a perfect fluid is unable to drive the accelerated expansion as well as solve the cosmic coincidence problem at the same time @xcite . however
, a mixture of a dissipative cdm with bulk viscosity and a minimally coupled self interacting scalar field can successfully achieve both features simultaneously .
also , as demonstrated in a recent paper by zimdahl @xcite one can also have a negative @xmath12 if there exists an interaction which does not conserve particle numbers .
this may be due to the particle production out of gravitational field . in this case , the cdm is not a conventional dissipative fluid , but a perfect fluid with varying particle number .
substantial particle production is an event that occurs in the early universe .
but zimdahl have shown that even extremely small particle production rate can also cause the sufficiently negative @xmath12 to violate the strong energy condition . in this paper
we have used a minimally coupled scalar field with a self interacting potential together with a matter fluid having a dissipative pressure over and above its positive equilibrium pressure .
we have not assumed any particular model for this negative pressure .
instead , we have investigated what kind effects it has in the expansion of the universe .
unlike other works in scalar field cosmology with a dissipative pressure @xcite we have neither assumed the behaviour of the scale factor nor have we assumed any specific form of the potential .
rather we have expressed all the variables in terms of what we call the ` _ generating function _ ' .
for this we have followed the method described by chimento et.al .
@xcite with some additional assumptions .
we have proceeded with a particular choice of the generating function for which the potential is constructed using a combination of different power - law functions of @xmath13 . from the behaviour of decelerating parameter
it has been shown that one can indeed generate both inflationary era in the early time and also an late time accelerating phase with a decelerating period in between .
we have also investigated the stability and attractor structure of the general solutions of the field equations with this kind of potential and have found that for certain choices of the constants the solutions indeed exhibit attractor behaviour in the late times .
let us consider a homogeneous , isotropic , spatially flat frw universe with a line element ds^2 = - dt^2 + a(t)^2 _
i=1 ^ 3 dx_i^2 , where @xmath14 is the scale factor .
the energy density consists of a massive scalar field with a self - interacting potential @xmath15 s _
= _ , _ , - g _ ( _ , ^ , + v ( ) ) , together with a dissipative fluid having bulk viscosity as the only dissipative term : t _ = u _ u _ + h _
( p + ) , where @xmath16 is the energy density , @xmath17 is the equilibrium pressure , @xmath12 is the bulk viscous pressure , and @xmath18 is the projection tensor . the independent field equations describing the system are [ kgeqn ] + 3 h + = 0 , + [ friedeqn ] 3 h^2 = ^2 + v + , + [ conseq ] + 3 h ( + p + ) = 0 after some straightforward calculations one can construct the equation [ friedkgeqn ] 2 h = - ^2 - ( + p + ) . for @xmath19 ,
the system consists of three independent equations , viz . , ( [ conseq ] ) , ( [ friedeqn ] ) and ( [ friedkgeqn ] ) and we have six unknowns , @xmath20 and @xmath12 .
hence , three constraint equations are required in order to have a closed system of equations .
we first assume that @xmath17,@xmath16 and @xmath12 are related to @xmath21 by , [ assumeqn1 ] + p + = m ^2 , where @xmath22 is an arbitrary constant .
we emphasize that this choice is considered purely because it makes the system of equations simple to solve .
we use this to see whether such a choice leads to any physically acceptable solutions .
the nature of the potential determines the evolution of @xmath13 .
however , instead of choosing a particular form for @xmath15 , we can alternatively describe the evolution of the scalar field by expressing @xmath21 as a function of @xmath13 as , @xmath23 where @xmath24 is called the generating function . in this approach , we start with a specific trajectory , @xmath25 , in phase space for @xmath13 and determine the potential that evolves the scalar field in that manner . however , choosing a potential @xmath15 or choosing a generating function , @xmath24 is not completely equivalent .
choosing a form for @xmath26 is more restrictive than choosing a potential due to the following reason .
a specific @xmath24 refers to a specific class of initial conditions .
hence the set of solutions represented by a specific form for @xmath24 is a subset of the full set of solutions for the corresponding potential .
having constructed the potential function , we study the evolution of the system in phase space for a more general initial condition ,
_ i.e. _ ones which are not restricted by any specific form for @xmath24 .
using equation ( [ assumeqn1 ] ) , equation ( [ friedkgeqn ] ) and equation ( [ chosphi ] ) one can write , 2 = - ( m+1 ) f ( ) . in this paper , we first choose @xmath24 and then calculate the @xmath15 that is consistent with our choice of @xmath24 as well as with the einstein s equations .
the solutions for which the phase space trajectory is given by @xmath27 is a subset of the of the general class of solutions for this potential .
( later in subsection [ stability ] we will draw the phase space trajectory for a set of general solutions ( _ i.e. _ with different initial conditions ) including the one for which @xmath27 ) and check the conditions for this to be a stable solution . ) for any given @xmath24 which we term as the `` generating function '' , one can , in principle , solve the system as follows : h ( ) & = & -f ( ) d + h_0 , [ heqn ] + ( ) & = & - 3 m h ( ) f ( ) d+_0 , [ rhoeqn ] + a ( ) & = & a_0 exp [ aeqn ] + v ( ) & = & 3 h^2 ( ) - f()^2 - ( ) , [ poteqn ] + p + & = & m f()^2 - ( ) . [ ppieqn ] where @xmath28 and @xmath29 are integration constant.with the assumption of an equation of state @xmath30 , one can also calculate @xmath12 . hence , our main aim is to choose @xmath24 properly to have some physically acceptable behaviour for different variables .
with this choice of the generating function , the differential equation governing the time evolution of @xmath13 is @xmath32 in this case , the exact solutions turn out to be , ( t ) & = & _ 0 exp(t),[phires ] + h(t ) & = & - _ 0 ^ 2 exp(2 t ) + h_0[hres ] + a(t ) & = & a_0 exp , [ ares ] + q(t ) & = & - 1 , [ qres ] + ( t ) & = & + & - & + _ 0 , [ rhores ] + v ( ) & = & ( 3 h_0 ^ 2-_0 ) + - ( + 3 h_0).[vres ] for an expanding universe we must have @xmath33 . to ensure this , either @xmath34 or @xmath35 .
we choose @xmath35 .
it may be noted from equation ( 19 ) that the proper volume becomes zero at @xmath36 and hence this is taken to be the initial time for our model .
the potential @xmath15 given in equation ( 22 ) is the standard renormalizable tree level potential arising in the perturabative regime of quantum field theory . with @xmath35 , the shape of the potential depends on the factor ( @xmath37 ) . when @xmath38 , it is the most simplified version of the potential for the hybrid inflation @xcite .
similarly when @xmath39 , it is the inverted quadratic potential and has been discussed by many authors for inflationary models ( see @xcite and references therein ) . as has been pointed out before , the solution represented by equation ( [ phires ] ) is only one of the classes solutions ( corresponding to the potential given in equation ( [ vres]))for which the the initial values of @xmath40 and @xmath13 are related by @xmath41 . 0.1 cm expressing the equation of state as , @xmath42 , where @xmath43 , the bulk viscous pressure is given by , & = & m_0 ^ 2 exp(2t)(+ ) + & - & ^2_0 ^ 4 exp(4 t ) -_0 .
[ visc ] the behaviour of the deceleration parameter is shown in figure [ figure1 ] .
the evolution of @xmath44 has the basic feature that we require for an acceptable form for the deceleration parameter .
@xmath44 is negative to begin with resulting in an inflationary phase .
it increases and subsequently @xmath44 becomes positive . at a later time , it drops below @xmath45 and finally saturates at a constant value below zero .
this last feature of @xmath44 at late times when it becomes a negative constant produces the present day accelerated expansion .
in fact it asymptotically becomes a de sitter universe .
the behaviour of the equation of state for the scalar field @xmath46 in figure ( [ figure 2 ] ) shows that at late times @xmath47 which essentially supports the existence of a cosmological constant in late time .
an acceptable model should also satisfy constraints arising from cosmological nucleosynthesis and from structure formation . in order to keep these intact
, the matter energy density should dominate over the energy density of the scalar field in the early universe .
however , at late times the scalar field energy density should be more than that due to matter so that one can explain the _ missing energy _ of the universe . in figure ( [ figure4 ] )
we have plotted the energy density for the matter @xmath48 and for the scalar field @xmath49 .
we see the above constraints are satisfied .
the matter energy density is dominant in the early time and so nucleosynthesis and structure formation are unaffected .
but at late time @xmath7 decreases more slowly than @xmath16 and hence ultimately it becomes greater than @xmath16 this explains the missing energy associated with accelerated expansion of the universe .
it can also be seen that although in early period the energy densities are of different order of magnitude but in late time they are of the same order .
this gives a dynamical solution to the cosmic coincidence problem . also at late times
the ratio of these two energy densities becomes constant showing the tracking behaviour .
we have also plotted the two density parameter @xmath50 and @xmath51 in figure 4 .
the variation of the viscous pressure @xmath12 is shown in [ figure6 ] .
the solutions we discussed were those for which @xmath31 .
in other words , for these set of solutions the initial conditions are constrained by the relation @xmath52 , where @xmath53 and @xmath54 are the initial values of @xmath13 and @xmath21 , respectively .
these solutions are significant only if these are stable , _
i.e. _ , only if more general initial conditions asymptotically evolve towards these phase space trajectories .
this solution corresponds to the case when the expression for the hubble parameter ( @xmath55 ) , scale factor ( @xmath14 ) , deceleration parameter ( @xmath44 ) and the density ( @xmath56 ) given by equations ( [ hres ] ) , ( [ ares ] ) , ( [ qres ] ) and ( [ rhores ] ) . in this section
we proceed to investigate whether or not the solutions described are stable and whether there are any constraints on the parameters for this .
the general equation of motion for the potential given in equation ( [ vres ] ) is = -3h -^2 ^ 3+^2 + 3h_0together with the equation for the evolution of @xmath55 , we can write the following set of three coupled differential equations , & = & x + x&=&-3hx-^2 ^ 3+^2 + 3h_0 + h&=&-x^2 + by equating the right hand sides of the equations to @xmath45 we obtain the following three critical points in the ( @xmath57 ) plane .
( 0,0 ) , ( , 0 ) , ( -,0 ) .
the position of the critical points in the phase diagram changes with the value of @xmath58 .
further , at @xmath59 all the three critical points merge .
there is a transition in the nature of stability of the trajectories in @xmath60 space at this value of @xmath58 .
we discuss below the nature of the trajectories for different values of @xmath58 .
first of all , we should note that @xmath35 .
in figures ( [ figure7 ] ) and ( [ figure8 ] ) , we have plotted the phase space trajectories for the scalar field with a variety of initial conditions that are not constrained by the particular choice of the generating function .
figure [ figure7 ] corresponds to the case of @xmath61 . as we have chosen @xmath62 to be @xmath63 , this is a case in which the potential has a local maximum .
on the other hand , figure [ figure8 ] corresponds to the case of @xmath64 in which @xmath65 is a minimum for the potential . in the first case
the phase space trajectories converge to either of the two non - zero values of @xmath13 and is not an attractor like solution as the asymptotic nature of the solution depends on the initial conditions .
but in the second case , however , the trajectories converge to @xmath65 , @xmath66 independent of the initial conditions and , the de - sitter solution is an attractor for this case .
the most important conclusion is that models with a self - interacting scalar field and a matter fluid having a negative pressure in addition to its positive equilibrium pressure can produce a scenario for the cosmological evolution in which one can have an inflationary phase to begin with , an accelerated phase at late times ( like the present era ) and a decelerating phase in - between . recently
lopez and matos @xcite have shown that this kind of complete history for the scale factor can be described by a hyperbolic potential .
but the physical origin of such potential is still not well known .
but here we have shown that such kind of behaviour for the scale factor can be generated with a potential given in ( 22),which has been widely used by many authors for inflationary models . the behaviour of @xmath16 and @xmath7 in our model shows that although in early universe , @xmath16 is greater than @xmath7 which is necessary for different physical phenomena like nucleosynthesis and structure formation etc , in the late times , @xmath7 starts dominating .
this feature explains the missing energy density and also the ration of two energy densities becomes a constant in late time showing the `` _ tracking nature _ '' . one should note that both the assumptions ( 8) and ( 16 ) play crucial role in our model . given a barotropic equation of state between @xmath16 and @xmath17 one can not assume ( 8) and ( 16 ) at the same time if the dissipative pressure @xmath12 is zero in our model as t hat will lead to an over - determined problem ( the number of unknowns will be less than the number of independent equations ) .
even if one assumes these two condition one can check that will lead to an negative equilibrium pressure which is not desirable .
h ence the existence of dissipative pressure also plays an important role in our model .
we have also studied the general equation of motion for the scalar field ( equation ( 24 ) ) for the potential ( 22 ) and have shown that for the choices of constant for which the potential is minimum at @xmath65 , the phase space diagram exhibit a attractor behaviour towards the asymptotic de - sitter solution .
we want to mention that previously tracker and attractor solutions have been studied for scalar fields having inverse power law , exponential , cosine potential .
but in all of these cases the equation of state @xmath67 is a constant in radiation era as well as in a matter dominated era .
it was later shown by pietro and demaret @xcite that these kind of potential with a constant @xmath67 is not consistent with the field equations . in our case ,
the equation of state for the scalar field @xmath67 is not a constant but it varies with the cosmic evolution and approaches towards -1 asymptotically showing the existence of a cosmological constant in late times .
99 electronic address : [email protected] electronic address : [email protected] electronic address : [email protected] j. p. ostriker and p. j. steinhardt , _ nature _ , * 377 * , 600,(1995 ) .
de bernardis et al , _ nature _ , * 404 * , 955 , ( 2000 ) .
s. perlmutter et.al .
, _ astro - ph_/971221 ; a. g. riess et.al , _ astro - ph_/9805201 .
n.a.bahcall , j.p.ostriker , s.perlmutter and p.j.steinhardt , _ science _ , * 284 * , 1481 ( 1988 ) v.sahni and a.starobinsky , _ int
d _ to appear ( 2000 ) _
astro - ph_/9904398 .
r.r.caldwell , r.dave and p.j.steinhardt , _ phys .
lett _ , * 80 * , 1582 ( 1998 ) p.j.e.peebles and b.ratra , _
astrophys.j.lett._ , * 325 * , l17 , ( 1988 ) ; p.g.ferreira and m.joyce , _
phys.rev.lett._ , * 79 * , 4740 ( 1987 ) ; e.j.copeland , a.r.liddle and d.wands , _
phys.rev.d_ , * 57 * , 4686 ( 1988 ) p.j .
steinhardt , l.wang and i.zlatev , _ phys.rev.lett._ , * 59 * , 123504 ( 1999 ) i.zlatev , l.wang and p.j.steinhardt , _ phys.rev.lett._ , * 82 * , 896 ( 1999 ) .
v.sahni and l.wang , _ astro - ph_/9910097 l.a.u.lopez and t.matos , _ astro - ph_/0003364 j.p.uzan , _
phys.rev.d_ , * 59 * , 123510 ( 1999 ) t.d .
saini , s. raychauchaudhury , v. sahni and a.a .
starobinsky , _
phys.rev.lett_ , * 85 * , 1162 ( 2000 ) . p.g.ferreira and m.joyce , _
phys.rev.d_ , * 58 * , 023503 ( 1998 ) ; b. ratra and p.j.e .
peebles , _ phys .
* 37 * , 3406 ( 1988 ) ; t.barreiro , e.j.copeland and n.j.nunes _ astro - ph_/9910214 . elisa de pietro and jacques demaret gr - qc/9908071 .
d.n.spergel and p.j.steinhardt phys.rev.lett , * 82 * , 896 ( 1999 ) ; j.p.ostriker _ astro - ph_/9912548 ; s.hannestad _ astro - ph_/9912558 ; b.moore et.al astrophys.j * 535 * , l21 ( 2000 ) .
l.p.chimento , alejandro s.jakubi and diego pavon phys.rev.d * 52 * , 063509 ( 2000 ) .
w.zimdahl , d.j.schwarz , a.b.balakin and d.pavon astro - ph/0009353 . n.banerjee and s.sen phys.rev.d*57 * , 4614 ( 1998 ) ; l.p.chimento , v.mendez and n.zuccala , int.j.mod.phys.d . * 5 * , 313 ( 1996 ) .
l. chimento , v. mendez and n. zuccala , class .
quantum grav .
* 16 * 3749 ( 1999 ) .
d.lyth and a.riotto phys.rep .
* 314 * ( 1999 ) .
l.arturo urena - lopez and tonatiuh matos , astro - ph/0003364 | we have investigated cosmological models with a self - interacting scalar field and a dissipative matter fluid as the sources of matter .
different variables are expressed in terms of a _ generating function_. exact solutions are obtained for one particular choice of the _ generating function _ the potential corresponding to this generating function is a standard tree - level potential arising in the perturbative regime in quantum field theory . with suitable choice of parameters , the scale factor in our
model exhibits both inflationary behaviour in the early universe as well as an accelerating phase at late times with a decelerating period in between .
it also satisfies the constraints for primeval nucleosynthesis and structure formation and seems to solve the cosmic coincidence problem .
the solution exhibits a attractor nature towards a asymptotic de - sitter universe . |
much of the power of information theory , classical and quantum , comes from the separation of information from its physical carriers .
this level of abstraction favors a black box approach describing physical processes by their input - output relations in discrete time steps . for quantum systems
the most general black box is described by a trace preserving and completely positive map a quantum channel .
this might describe the application of a gate in a quantum processor , a quantum storage device , a communication channel , or any open systems dynamics where one merely has partial access to the relevant degrees of freedom . in any case
it can be considered a snapshot of a physical evolution after a certain time . in the present letter
we investigate what such a snapshot reveals about the intermediate continuous time evolution , in particular regarding the markovian and hence memoryless or non - markovian character of the process .
this will link the black box approach to the dynamical theory of open quantum systems .
remarkably , fixing a single point in time enables us to gain non - trivial information about the path along which the system has or has not evolved , even without making additional assumptions about the physics of the environment and its coupling to the system . on the one hand
this analysis thus provides a model - independent means of investigating non - markovian features . on the other hand it tells us which type of evolution is required for the continuous realization of theoretically given quantum channels . on the experimental side recent progress in the field of quantum information science showed more and more precise determination of input - output relations via _ quantum process
tomography_. this has by now been achieved in various systems including nmr @xcite , ion traps @xcite , linear optics implementations @xcite and solid state qubits @xcite .
some of them , in fact , rely on the a priori assumption that the process in fact is markovian . in the study of open quantum systems as such , non - markovian processes also move to the center of interest @xcite .
in general , a precise understanding of decoherence and dissipation processes is vital for further improvements and for the design of adapted error correcting codes and fault - tolerant schemes .
most notably in this context non - markovian effects are known to require a careful analysis @xcite .
before we start , some remarks concerning the central notions ` markovian ' and ` time - dependent markovian ' are in order
. we will call a quantum channel _ markovian _ if it is an element of any one - parameter continuous completely positive semigroup , i.e. , a solution of a master equation with generator in lindblad form .
if the generator depends on time , we use the term _ time - dependent markovian _ instead . in both cases
the continuous evolution is memoryless in that at any point in time the future evolution only depends on the present state and not on the history of the system .
our findings are : * the sets of ( time - dependent ) markovian channels are strictly included within the set of all quantum channels and exhibit a non - convex geometry . * for arbitrary finite dimensions
there is an efficient algorithm for deciding whether or not a quantum channel is markovian . * a computable measure is introduced which quantifies the markovian part of a channel . * for qubits a simple criterion for time - dependent markovianity is given together with a detailed analysis of the geometry of the sets of quantum channels .
* examples of non - markovian processes are discussed . *
an application in the theory of renormalization group ( rg ) transformations on quantum spin chains is outlined .
throughout we will consider quantum channels on finite dimensional systems , i.e. , linear maps @xmath1 on @xmath2 ( density ) matrices , @xmath3 referred to as _ dynamical maps _ , reflecting the snapshot in time @xcite .
when occasionally changing from schrdinger to heisenberg picture we will denote the respective map by @xmath4
. it will be convenient to consider @xmath5 as a hilbert space @xmath6 equipped with the scalar product @xmath7 $ ] . on this space
the map @xmath8 is represented by a matrix @xmath9={\langlef_\alpha|t|f_\beta\rangle}_{{\mathfrak{h}}},\ ] ] where @xmath10 is any orthonormal basis in @xmath6 .
unless otherwise stated , we will use matrix units @xmath11 as basis elements . note that a concatenation of two maps @xmath12 simply corresponds to a product of the respective matrices @xmath13 and that a density matrix @xmath14 in this language becomes a vector with entries @xmath15 .
a useful operation is the involution @xmath16 @xcite .
it connects the matrix representation @xmath17 of the map to its choi matrix @xmath18 where @xmath19 is a maximally entangled state @xmath20 , @xmath21 .
complete positivity is then equivalent to @xmath22 . _ quantum channels
_ are completely positive and trace preserving maps .
the workhorse in the dynamical theory of open quantum systems are semigroups @xmath23 depending continuously on one parameter @xmath24 ( time ) and giving rise to completely positive evolution for all time intervals .
two equivalent standard forms for the respective generators have been derived in refs .
@xcite : @xmath25 + \sum_{\alpha,\beta } g_{\alpha,\beta } \bigl(f_\alpha\rho f_\beta^\dagger - \frac12\{f_\beta^\dagger f_\alpha,\rho\}_+\bigr ) , \label{normal2}\ ] ] so @xmath26 , where @xmath27 , @xmath28 , @xmath29 is completely positive and @xmath30 .
a channel will be called _ time ( in)dependent markovian _ if it is the solution of any master equation @xmath31 with time ( in)dependent liouvillian in _
lindblad form _ ( [ normal2 ] ) , i.e. , @xmath32
given a quantum channel @xmath8 when is it markovian , i.e , of the form @xmath33 ? a priori , this might be a trivial question : as the channel fixes only one point within a continuous evolution , there might always be a ` markovian path ' through that point .
as the attentive reader might already guess , this turns out to be not the case .
one attempt to decide whether @xmath8 is markovian could be to start from a markovian ansatz and then calculate @xmath34 , e.g. , by numerical minimization
. the major drawback of such an approach is the non - convex geometry of the set of markovian channels @xcite which inevitably leads to the occurrence of local minima .
the following approach circumvents this problem and guarantees to find the correct answer efficiently by first taking the @xmath35 of @xmath8 and then deciding whether this is a valid lindblad generator of the form ( [ normal2 ] ) .
the latter can easily be decided : a map @xmath36 can be written in lindblad form iff ( a ) it is hermitian @xcite , ( b ) @xmath37 corresponding to the trace preserving property and ( c ) @xmath38 is _ conditionally completely positive _ ( ccp ) @xcite , i.e. , @xmath39 where @xmath40 is the projector onto the orthogonal complement of the maximally entangled state ( see appendix ) . before applying ( [ eq : ccp ] ) to @xmath41 we need to discuss some spectral properties of quantum channels . for simplicity we will restrict ourselves to the generic case where @xmath17 has non - defective and non - degenerate jordan normal form .
hermiticity of a channel implies that its eigenvalues are either real or come in complex conjugate pairs .
jordan normal form_achievable via a similarity transform in the orthonormal basis in which the channel is expressed is then @xmath42 where @xmath43 labels the real and @xmath44 the complex eigenvalues respectively .
the @xmath45 s are orthogonal ( but typically not self - adjoint ) spectral projectors and @xmath46 is the flip - operator ( @xmath47 ) .
projectors corresponding to complex conjugate eigenvalues are related via @xmath48 due to hermiticity of the channel which can in turn be expressed as @xmath49
. now @xmath8 is markovian iff there is a branch of the logarithm @xmath50 , defined via the logarithm of the eigenvalues in eq .
( [ eq : jordan ] ) , which fulfills the above mentioned conditions ( a ) ( c ) .
note that ( b ) is always satisfied if we start from a trace preserving map .
moreover , hermiticity holds iff there is no negative real eigenvalue and branches for each complex pair of eigenvalues are chosen consistently so that the eigenvalues remain complex conjugates of each other .
if @xmath17 has negative eigenvalues , the dynamics will not be markovian .
the set of hermitian logarithms is then characterized by a set of integers @xmath51 , @xmath52 where @xmath53 denotes the principal branch .
the infinity of discrete branches looks a bit awkward at first glance , but the problem can now be cast into a familiar form . defining the matrices @xmath54 and applying eq .
( [ eq : ccp ] ) to @xmath55 yields that @xmath8 is markovian iff @xmath56 for any set of integers @xmath57 .
these matrices @xmath58 will be hermitian if @xmath17 has only positive real eigenvalues .
note that the real solutions @xmath59 of eq.([eq : semi0 ] ) form a convex set @xmath60 ( @xmath61 being the number of complex pairs of eigenvalues ) . the present problem is then to decide whether @xmath62 contains an integer point .
fortunately , this has an efficient solution in terms of a semi - definite integer program .
that is , there is an algorithm @xcite which either finds a solution or guarantees that none exists within a run time of order @xmath63 where @xmath64 is the number of digits to which the input is specified @xcite . in practice
it turns out that checking the vicinity of the principle branch ( @xmath65 ) is typically sufficient . for qubit channels the problem simplifies further since it becomes ( at most ) one - dimensional as @xmath66 .
hence , one has merely to maximize the smallest eigenvalue ( a concave function ) of eq .
( [ eq : semi0 ] ) with respect to real @xmath67 and then check positivity for the two neighboring integers . note that again
, this criterion does not necessarily reveal whether the dynamics was truly markovian , but whether it is consistent with it .
when applying the above criterion to random quantum channels one finds that only a small ( but remarkably nonzero ) fraction of them is markovian , see fig .
[ figure : set ] . in the following
we will show how one can quantify the deviation from markovianity for the remaining channels .
desirable properties of a measure of markovianity @xmath68 are : * some form of normalization , e.g. , @xmath69 $ ] with @xmath70 iff @xmath8 is markovian , * computability , * continuity , * basis independence , i.e. , @xmath71 for all unitary channels @xmath72 and * an operational or physical interpretation .
one possibility would again be to start from a distance measure @xmath34 which , however , looses much of its appeal by the apparent difficulties in computing it .
we therefore propose a different approach based on the criterion in eq .
( [ eq : semi0 ] ) . to this end
let us regard @xmath73 as liouvillian of a master equation .
if the channel is not markovian this will not give rise to completely positive , i.e. , physical , evolution for all times .
however , adding an additional dissipative term might yield a physical markovian evolution .
if we choose isotropic noise of the form @xmath74 with corresponding generator @xmath75 then @xmath76 becomes a valid lindblad generator for some @xmath67 iff @xmath77 exceeds @xmath78 hence @xmath79 is the minimum amount of isotropic noise required to make the channel markovian .
note that @xmath79 can again be calculated by semidefinite integer programming and that it is basis independent in the sense of ( iv ) . in order to meet the normalization condition and to add an intuitive geometric interpretation to the physical one we use @xmath80\;\in[0,1]\ ] ] as a _ measure for markovianity_. if @xmath81 is not hermitian we assign @xmath82 .
this turns out to be precisely the factor by which the additional dissipation shrinks the output space of the channel in order to make it markovian . in order to see this note that the volume of the output space ( one might think in terms of the bloch sphere for @xmath83 ) is quantified by the determinant of the channel @xcite .
moreover , @xmath84 , since @xmath85 $ ] is independent of @xmath67 . in this sense @xmath86 quantifies the markovian part of the channel @xcite .
[ fig : plot ] ( a ) shows the markovianity of a convex combination @xmath87 of a unitary channel @xmath88 corresponding to @xmath89-rabi oscillation ( with hamiltonian @xmath90 ) and a dephasing process @xmath91 with @xmath92 this confirms the _ non - convex geometry _ in fig . [ figure : set ] and shows that non - markovian effects can arise from an environment which is in a mixture of states each of which leads to a markovian evolution .
interestingly , there also exist non - markovian processes that could have arisen from a markovian process when judged from a snapshot in time : the _ spin - star network _ in ref .
@xcite has the property that for all times @xmath93 , and hence the channel is consistent with markovianity .
this is perfectly physical , as in each time step there could have been a different memoryless evolution .
[ fig : plot ] ( b ) in turn depicts the deviation from markovianity for the _ damped jaynes - cummings model _ , where the non - markovian character of the dynamics is clearly displayed @xcite .
further examples where this competing effect of time scales can be observed are non - markovian models arising from spins coupled to _ structured baths _ with an energy gap as studied , e.g. , in ref .
for deciding whether a channel is a solution of a time - dependent master equation we resort to generic qubit channels @xcite and content ourselves with presenting results whose technical proofs are published elsewhere @xcite .
a necessary and sufficient criterion is most easily expressed in the basis of pauli matrices @xmath94 .
let @xmath95 and @xmath96 be the ordered square roots of the eigenvalues of @xmath97 .
then @xmath8 is time - dependent markovian iff @xmath98 and @xmath99 @xcite .
this set contains all extremal qubit channels ( fig .
[ figure : set ] ) , but only 17% of all channels @xcite . a paradigmatic example outside
this set is @xmath100{\mathbbm{1}}+\rho^t\big)/3,\ ] ] which is the best physical approximation to matrix transposition ( or time - reversal or in optics
phase conjugation when rotated by @xmath101 ) .
this channels fails the criterion since @xmath102 @xcite .
in fact , it belongs to the peculiar set of _ indivisible channels _ that can not be decomposed into a concatenation of two channels unless one of them is unitary .
let us finally outline an application of the above results within an entirely different field : rg transformations for translational invariant states on quantum spin chains .
the transformation introduced in ref .
@xcite amounts to a coarse graining consisting of two steps : ( 1 . )
build equivalence classes of states which only differ by a change of local basis , ( 2 . )
merge neighboring sites and then iterate .
both steps can be carried out explicitly if the state s matrix product representation @xcite uses matrices of finite dimension @xmath103 .
each equivalence class then corresponds to a quantum channel @xmath8 on @xmath104 and merging neighboring sites is reflected by replacing @xmath8 with @xmath105 . in this sense
the renormalization group is identified with a semigroup of quantum channels .
if @xmath8 is markovian , we can replace the discrete block spin transformation by an equivalent continuous rg flow @xmath106\ ] ] with @xmath107 parameterizing the scale of observation , see eq .
( [ eq : semi0 ] ) .
this allows then not only to coarse - grain ( @xmath108 ) but also to zoom in ( @xmath109 ) .
indivisible channels are also special in this context : since there is no @xmath110 such that @xmath111 they correspond to starting points , i.e. , ultraviolet limits , of rg flows . in fact , the best physical approximation to time reversal in this way distinguishes the well known ground state of the spin-1 aklt hamiltonian @xcite .
in this letter , we have introduced a framework to assess whether a given dynamical map describing a physical process could have arisen from markovian dynamics . to test this property we have provided necessary and sufficient conditions .
we also introduced a natural measure of markovianity , quantifying the markovian content of a process .
as such , we have provided the means to judge the forgetfulness of a physical process from a snapshot in time .
mw thanks g. giedke and i.de vega for valuable discussions .
this work has been supported by the elitenetzwerk bayern , the eu ( qap , compas , scala ) , the epsrc , microsoft research , and the euryi .
the following is a modification of the argument given in ref .
@xcite and allows to decide whether or not a map @xmath38 can be written in lindblad form eq .
( [ normal2 ] ) .
it is obvious from ( [ normal2 ] ) that @xmath38 has to be hermitian , i.e. , @xmath112 for all @xmath113 , and that @xmath37 .
necessity of eq .
( [ eq : ccp ] ) is seen by exploiting eq .
( [ normal2 ] ) : @xmath114 d.\ ] ] on the r.h.s .
the first term is positive and the other terms vanish when projected onto @xmath115 . conversely any hermitian matrix fulfilling inequality ( [ eq : ccp ] ) is of the form @xmath116 and @xmath117 .
to arrive at eq .
( [ normal2 ] ) we interpret @xmath45 as choi - matrix of a completely positive map @xmath29 and set @xmath118 .
nielsen , e. knill , and r. laflamme , nature * 396 * , 52 ( 1998 ) ; l.m.k . vandersypen and i.l
chuang , rev .
phys . * 76 * , 1037 ( 2004 ) ; n. boulant et al .
a * 67 * , 042322 ( 2003 ) ; j. emerson et al . ,
science * 317 * , 1893 ( 2007 ) .
note that the closure of the set of markovian channels is also non - convex .
simple examples like convex combinations of the identity channel and a flip of two levels can depending on the regime be proven to be markovian and non - markovian . for small dimensions
@xmath122 , this leads to a practical and efficient algorithm for testing for markovianity of a quantum channel . note that the computational hardness in the system s dimension @xmath122 is not due to any specific algorithm provided here , but in fact due to the nature of the problem as such . for practical purposes , yet , one is typically faced with channels of a given dimension @xmath122 , and not with a family of channels of increasing @xmath122 . note
that this measure is merely continuous up to a set of measure zero : the logarithm becomes discontinuous at zero so that @xmath86 can become discontinuous at singular channels . in the degenerate case
a second set of discontinuities appears ( again of measure zero ) : namely if a markovian channel with pairwise negative eigenvalues is perturbed such that the eigenvalues remain real but the degeneracy is removed . in the damped jaynes cummings model ( fig .
[ fig : plot ] ( b ) ) , the hamiltonian is @xmath123 with @xmath124 being the annihilation operator of the field mode , undergoing markovian damping with a lindblad generator @xmath125 and @xmath126 ( see , e.g. , ref .
the non - markovian dynamics is the reduced dynamics of the spin .
d. perez - garcia et al . , quant .
inf . comp .
* 7 * , 401 ( 2007 ) ; m. fannes , b. nachtergaele , r.f .
werner , commun .
phys . * 144 * , 443 ( 1992 ) ; a. klmper , a. schadschneider , and j. zittartz , j. phys .
a * 24 * , l955 ( 1991 ) . | we investigate what a snapshot of a quantum evolution
a quantum channel reflecting open system dynamics reveals about the underlying continuous time evolution .
remarkably , from such a snapshot , and without imposing additional assumptions , it can be decided whether or not a channel is consistent with a time ( in)dependent markovian evolution , for which we provide computable necessary and sufficient criteria . based on these , a computable measure of ` markovianity '
is introduced .
we discuss how the consistency with markovian dynamics can be checked in quantum process tomography .
the results also clarify the geometry of the set of quantum channels with respect to being solutions of time ( in)dependent master equations .
[ theorem]lemma [ theorem]corollary [ theorem]proposition [ theorem]definition [ theorem]example @xmath0 @xmath0 |
volume - phase holographic ( vph ) gratings potentially have many advantages over classical surface - relief gratings ( barden , arns , & colburn 1998 ; see also barden et al . 2000 ) , and are planned to be used in a number of forthcoming instruments ( e.g. , aa@xmath2 ; bridges et al .
while applications to optical spectrographs only are currently being considered , vph gratings will also be useful to near - infrared spectrographs if the performance at low temperatures is satisfactory .
in particular , its diffraction efficiency and angular dispersion should be confirmed .
contraction of dichromated gelatin with decreasing temperature could cause variations in the line density and profile of diffraction efficiency ( the thickness of the gelatin layer is one of the parameters defining diffraction efficiency ) . since cooling and
heating cycles might cause some deterioration of a vph grating and reduce its life time , we also need to see whether these characteristics vary with the successive cycles . in this paper , results from measurements of a sample grating at 200 @xmath0 and at room temperature are presented .
a picture of the grating investigated is shown in figure [ vph ] .
this grating was manufactured by ralcon development lab , and its diameter is about 25 cm .
the line density is 385 lines / mm and thus the peak of diffraction efficiency is around 1.3 @xmath3 m at the bragg condition when the incident angle of an input beam to the normal of the grating surface is 15@xmath4 .
the measurements are performed at wavelengths from 0.9 @xmath3 m to 1.6 @xmath3 m . the target temperature , size and line density of the grating , and wavelengths investigated
are nearly the same as those adopted for the fibre - multi object spectrograph ( fmos ; e.g. , kimura et al . 2003 ) , which is one of the next generation instruments for the 8.2 m subaru telescope with commissioning expected in 2004 : this instrument will be observing at wavelengths from 0.9 @xmath3 m to 1.8 @xmath3 m , and a vph grating will be used as an anti - dispersing element in the near - infrared spectrograph which is operated at @xmath5 200 @xmath0 to reduce thermal noise .
before starting measurements using the cryogenic test facility ( see next section ) , we investigated the diffraction efficiency of the vph grating at a room temperature .
the measurements were performed manually on the optical bench .
the procedures and results obtained are summarised below . in figure [ config ] , the overall configuration of the optical components used for the measurements is indicated ( the detailed information for the main components are listed in table [ comps ] ) .
light exiting from the monochromator is collimated and used as an input beam to illuminate the central portion of the vph grating .
the spectral band - width of this input beam is set by adjusting the width of the output slit of the monochromator . the slit width and the corresponding spectral band - width
were set to 0.5 mm and @xmath5 0.01 @xmath3 m , respectively , throughout the measurements ; the beam diameter was set to @xmath5 2 cm by using an iris at the exit of the lamp house .
the input beam is diffracted by the grating and the camera , composed of lenses and a near - infrared detector ( 320 @xmath6 256 ingaas array ) , is scanned so as to capture the diffracted beam .
the output slit of the monochromator is thus re - imaged on the detector .
since the detector has some sensitivity at visible wavelengths , a visible blocking filter which is transparent at wavelengths longer than 0.75 @xmath3 m is inserted after the monochromator to reduce contamination of visible light from a higher order .
the basic measurement procedures are as follows .
first , the brightness of the lamp and the wavelength of light exiting from the monochromator are fixed ( the brightness of the lamp is kept constant by a stabilised power supply during the measurement cycle at a given wavelength ) , and the total intensity included in the image of the slit is measured without the vph grating . then , the vph grating is inserted at an angle to the optical axis , and the intensities of the zero and first order ( @xmath7 ) diffracted light are measured .
the diffraction angle is also recorded .
next , the grating is set at a different incident angle and the intensities of the diffracted light and diffraction angles are measured .
after these measurements are repeated for all the incident angles of interest , a different wavelength is chosen and the same sequence is repeated . the brightness of the lamp can be changed when moving from one wavelength to another : a higher brightness was used at shorter wavelengths because the system throughput is lower . in figure
[ eff_c ] , the diffraction efficiencies measured are plotted against wavelengths for the cases where incident angles are 15@xmath4 ( upper panel ) and 20@xmath4 ( lower panel ) .
open and solid dots show the efficiency profiles for the zero and first order ( @xmath7 ) diffracted light , respectively .
since random errors are dominated by fluctuations of the bias level of the detector on a short time scale ( @xmath8 sec ) , the error bars plotted are calculated from a typical value of the fluctuations .
it is found from this figure that particularly for the incident angle of 15 @xmath4 , the peak of the diffraction efficiency reaches @xmath5 80% and the efficiency exceeds 50% over the wavelength range from 0.9 @xmath3 m to 1.6 @xmath3 m .
the profiles can be well reproduced by theoretical calculations based on the coupled wave analysis ( kogelnik 1969 ) , which are shown by the solid lines in the figure . in these calculations ,
a thickness of the dichromated gelatin layer of 12 @xmath3 m and a refractive index modulation amplitude of 0.05 are assumed .
energy losses by surface reflections at boundaries between glass and air ( @xmath5 10% ) are also included .
although the energy losses at boundaries between glass and gelatin are likely to be much smaller because their refractive indices are very similar , they might explain that the measured diffraction efficiency tends to be slightly lower than the theoretical calculation .
( the energy lost by internal absorption of the dichromated gelatin layer is estimated to be @xmath9 1% below 1.8 @xmath3 m ; e.g. , barden et al .
in the following , we describe the measurements at 200 @xmath0 as well as those at a room temperature , both of which were performed using the cryogenic test facility as shown below . throughout these measurements
, we used a different vph grating from that used in the pre - test at room temperature , although both have the same specifications . in figures
[ config1 ] , schematic views of the fore - optics and the optics inside the cryogenic chamber are indicated .
pictures of these facilities are shown in figure [ pic ] .
the fore - optics and light path before the window of the cryogenic chamber and the camera are the same as those used in the warm pre - test .
the slit width and the spectral band - width were set to 0.1 mm and @xmath5 2.0 @xmath10 m , respectively , throughout the measurements , and the beam diameter was set to @xmath5 2 cm by using an iris before the window .
the light path in the cryogenic chamber is described as follows : the input beam illuminates the central portion of the vph grating and is diffracted by the grating .
the diffracted light is captured by scanning the pick - off arm and is delivered to the camera on the top of the chamber by 3 pick - off mirrors .
the output slit of the monochromator is thus re - imaged onto the detector .
this procedure enables measurements to be made at a variety of incident and diffraction angles without having to mount the detector inside the cryostat .
the measurement procedures are the same as those in the pre - test , except that all the measurements were performed with the vph grating in place .
we initially perform the measurements at a room temperature ( @xmath5 280 @xmath0 ) . then
, we repeat the measurements at 200 @xmath0 before returning to 280 @xmath0 to repeat the cycle . when we cool the vph grating , we monitor the temperature of the grating with a sensor on the surface , close to the edge of the grating but unilluminated by the input beam .
when the temperature reaches @xmath11 , we switch off the compressor and cold heads before starting the measurements .
although we do not have any thermostatic systems to maintain a given temperature , it takes several hours for the temperature of the grating to start increasing and go above 200 @xmath0 after the compressor and cold heads are switched off .
thus the temperature of the grating stays approximately at 200 @xmath12 5 @xmath0 for the duration of the measurement cycle .
the following results were obtained when the incident angle was set to @xmath13 , which gives the peak of diffraction efficiency around 1.3 @xmath3 m when satisfying the bragg condition .
we note that the same trends are obtained from measurements when different incident angles are adopted . in the upper panel of figure [ eff ] ,
differences of diffraction efficiencies in a sequence of measurements at 200 @xmath0 from those obtained in the first warm test ( 280 @xmath0 ) are plotted against wavelengths .
the error bars are calculated from the typical fluctuation of the bias level of the detector .
the errors are larger at shorter wavelengths because the system throughput is lower so that the bias level fluctuation is larger compared to the intensity of the slit images ( the brightness of the lamp is kept constant with a stabilised power supply throughout the measurements at all the wavelengths ) .
it is found that the differences are close to zero at all the wavelengths , suggesting that there is no significant variation in profile of diffraction efficiency such as a global decrease in the efficiency or a lateral shift of the peak . in the lower panels ,
the differences in diffraction efficiency are averaged over the wavelength range , and the averaged difference from the first warm test is plotted against cycle number .
open triangles and solid dots represent the measurements at 200 @xmath0 and those at 280 @xmath0 , respectively .
the error bars indicate the standard deviation of a distribution of the differences around the average value .
these results suggest that the diffraction efficiency of a vph grating is nearly independent of temperature , at least between 200 @xmath0 and 280 @xmath0 , and that no significant deterioration is caused by a small number of heating and cooling cycles . in the upper panel of figure [ dispersion ] , difference of diffraction angle from the prediction for the line density of 385 lines / mm ( the nominal line density of the vph grating )
is plotted against wavelength ; solid line ( zero level ) corresponds to the relationship between diffraction angle and wavelength for the line density of 385 lines / mm . dotted and dashed lines indicate the relationships predicted for a line density of 375 and 395 lines / mm , respectively .
the data points show the actual measurements at 280 @xmath0 . as the arrow shows , if the gelatin layer shrinks with decreasing temperature , the line density would increase and the data points would shift upwards on this plot . in the lower panel ,
the difference of diffraction angle from that measured in the first warm test is plotted against the number of cycles .
the symbols have the same meanings as those in figure [ eff ] .
again , each data point represents the difference averaged over the wavelength range , and the error bars indicate the standard deviation of the distribution of the differences around the average value .
the lower panel suggests that at 200 @xmath0 , the diffraction angle is slightly larger than that at 280 @xmath0 .
this is equivalent to a slight increase of the line density of the grating , and the simplest explanation for this is a shrinkage of the grating with decreasing temperature . by using this increment of diffraction angle ( @xmath5 0.1@xmath4 ) ,
the amount of shrinkage is estimated to be @xmath14% of the diameter of the grating , which is consistent with the amount of shrinkage of the glass substrate expected when the temperature is decreased by 80 @xmath0 ( the amount of shrinkage of gelatin would be larger by an order of magnitude ) .
one needs to keep in mind , however , that only a small portion of the vph grating was illuminated throughout the measurements ( the beam diameter was @xmath5 2 cm while the diameter of the vph grating is @xmath5 25 cm ) and thus a variation of the line density might be difficult to be detected . investigating larger portions over the vph
grating would be an important future work .
other cryogenic tests of vph gratings are also in progress by the golem group at brera astronomical observatory ( bianco et al .
their preliminary results suggest that diffraction efficiency is significantly reduced ( @xmath5 20% around the peak ) at @xmath5 200 @xmath0 compared with room temperature , which is inconsistent with our results .
one consideration here is that there is a significant difference in speed of the cooling and heating processes between the tests of the golem group and ours . in the golem case , they require only 1 hour to cool a vph grating down to their target temperature from a room temperature ( zerbi 2001 ) . with the cryogenic chamber used for our experiments , it takes about 15 hours to cool down to 200 @xmath0 from 280 @xmath0 .
this may imply that rapid cooling and/or heating can cause some deterioration of a vph grating .
further experiments are required in this area .
in this paper , results from the cryogenic tests of a vph grating at 200 @xmath0 are presented .
the aims of these tests were to see whether diffraction efficiency and angular dispersion of a vph grating are significantly different at a low temperature from those at a room temperature , and to see how many cooling and heating cycles the grating can withstand .
we have completed 5 cycles between room temperature and 200 @xmath0 , and find that diffraction efficiency and angular dispersion are nearly independent of temperature .
this result indicates that vph gratings can be used in spectrographs cooled down to 200 @xmath0 such as fmos without any significant deterioration of the performance . in future
, we will be trying more cycles between 200 @xmath0 and 280 @xmath0 to mimic more realistic situations of astronomical use .
measurements at a much lower temperature ( e.g. , @xmath5 80 @xmath0 ) will also be necessary to see whether vph gratings are applicable to spectrographs for use in the @xmath0-band .
we will report on these issues in a forthcoming paper .
we thank colleagues in durham for their assistance with this work , particularly paul clark , john bate , and the members of the mechanical workshop .
we are also grateful to the anonymous referee for the comments to improve our paper .
this work was funded by pparc rolling grant ( ppa / g / o/2000/00485 ) .
kimura , m. , maihara , t. , ohta , k. , iwamuro , f. , eto , s. , iino , m. , mochida , d. , shima , t. , karoji , h. , noumaru , j. , akiyama , m. , brzeski , j. , gillingham , p. r. , moore , a. m. , smith , g. , dalton , g. b. , tosh , i. a. j. , murray , g. j. , robertson , d. j. , & tamura , n. 2003 , proc .
4841 , 974 with that at 280 @xmath0 . in the upper panel ,
differences of diffraction efficiencies in a sequence of measurements at 200 @xmath0 from those obtained in the first warm test ( 280 @xmath0 ) are plotted against wavelengths .
the error bars are calculated from the typical fluctuation of the bias level of the detector . in the lower panels ,
the differences in diffraction efficiency as shown above are averaged over the wavelength range , and the averaged difference from the first warm test is plotted against cycle number .
open triangles and solid dots represent the data at 200 @xmath0 and those at 280 @xmath0 , respectively .
the error bars indicate the standard deviation of a distribution of the differences around the average value.,width=302 ] | we present results from cryogenic tests of a volume - phase holographic ( vph ) grating at 200 @xmath0 measured at near - infrared wavelengths .
the aims of these tests were to see whether the diffraction efficiency and angular dispersion of a vph grating are significantly different at a low temperature from those at a room temperature , and to see how many cooling and heating cycles the grating can withstand .
we have completed 5 cycles between room temperature and 200 @xmath0 , and find that the performance is nearly independent of temperature , at least over the temperature range which we are investigating . in future , we will not only try more cycles between these temperatures but also perform measurements at a much lower temperature ( e.g. , @xmath1 ) . |
entanglement is the hallmark as well as most counterintuitive feature of quantum mechanics . among various ways to quantify entanglement , bipartite block entanglement entropy emerged as a concept of central importance in quantum information science@xcite , and has been receiving growing attention in other branches of physics recently .
for example , it was suggested to be a possible source of black hole entropy@xcite . in condensed matter or many - body physics ,
the entanglement entropy has been increasingly used as a very useful and in some cases indispensable way to characterize phases and phase transitions , especially in strongly correlated phases@xcite . in this context
the most important result is perhaps the so - called area law@xcite , which states that in the thermodynamic limit , the entropy should be proportional to the area of the boundary that divides the system into two blocks .
there are a few very important examples@xcite in which the area law is violated , most of which involve quantum criticality@xcite ; the specific manner with which the area law is violated is tied to certain universal properties of the phase or critical point . in some other cases , important information about the phase can be revealed by studying the leading _ correction _ to the area law@xcite ; for example this is the case for topologically ordered phases@xcite .
comparatively there have been relatively few studies of the behavior of entanglement entropy in states with traditional long - range order@xcite .
this is perhaps because of the expectation that ordered states can be well described by mean - field theory , and in mean - field theory the states reduce to simple product states that have no entanglement . in particular in the limit of perfect long - range order
the mean - field theory becomes exact " , and the entanglement entropy should vanish . in this paper
we will show that this is _ not _ the case , and interesting entanglement exists in states with perfect long - range order .
we will study two exactly solvable spin-1/2 models : ( i ) an unfrustrated antiferromagnet with infinite range ( or constant ) antiferromagnetic ( afm ) interaction between spins in opposite sublattices , and ferromagnetic ( fm ) interaction between spins in the same sublattice ; ( ii ) an ordinary spin-1/2 ferromagnet with arbitrary fm interaction among the spins@xcite .
while the ground states have perfect long - range order for both models , we show that they both have non - zero entanglement entropy that grow logarithmically with the size of the subsystem .
the paper is organized as follows . in sec .
ii we introduce the model ( i ) and present its exact solution . in sec .
iii we calculate the reduced density matrix of a subsystem to obtain the exact expression of the entanglement entropy for model ( i ) . based on those results , in sec .
iv we study the scaling behavior of the entropy as the system size tends to infinity under two different partition limits and compare it with numerical calculations .
section v is devoted to model ( ii ) . finally , in sec . vi we summarize and discuss the results of this paper .
some mathematical definitions , notations and details are given in the appendices .
we consider a lattice model composed of two sublattices interpenetrating each other as in fig . [ fig.lattice ] , with interaction of infinite range , i.e. , every spin interacts with all the other spins in the system , with interaction strength independent of the distance between the spins . within each sublattice ,
the interaction is ferromagnetic , and between the sublattices the interaction is antiferromagnetic ; as a result there is no frustration . the hamiltonian is written as , @xmath0 with @xmath1 .
the ground state of eq .
( [ ham ] ) can be solved in the following manner@xcite .
let @xmath2 then the hamiltonian can be written as , @xmath3 since @xmath4=[\s_a , h]=[\s_b , h]=0 $ ] , eigenstates of @xmath5 can be chosen to be simultaneous eigenstates of @xmath6 , @xmath7 and @xmath8 , with angular momentum eigenvalues @xmath9 , @xmath10 and @xmath11 respectively , such that the energy eigenvalue is , @xmath12 to minimize the energy , we must first maximize @xmath10 and @xmath11 , and then minimize @xmath9 , i.e. , @xmath13 . here
@xmath14 and @xmath15 represent the number of corresponding sublattice sites .
physically this means that spins in the same sublattice are all parallel to each other , while spins in opposite sublattices are antiparallel .
thus the ground state is , @xmath16 + in this paper we only consider the simplest case with @xmath17 , thus the total system size is @xmath18 .
then the ground state is reduced to an antiferromagnetic ground state which has zero total spin , |s_as_b;sm=|;00 .
[ singletneel ] this ground state has perfect neel order , as manifested by the spin - spin correlation function , @xmath19
we divide the system spatially into two subsystems which are labeled 1 and 2 respectively , as shown in fig .
[ fig.division ] , and study the ground state entanglement entropy @xmath20 between these two subsystems .
@xmath20 is defined to be the von neumann entropy of either of the subsystems , @xmath21 or @xmath22 , which can be calculated from the reduced density matrix , @xmath23 where @xmath24 is the reduced density matrix of the subsystem 1 , obtained from the density matrix @xmath25 of the whole system by tracing out degrees of freedom of the other subsystem , @xmath26 , and vice versa . to solve for the explicit form of the reduced density matrix , we proceed as follows .
first , we further decompose the system into four parts , @xmath27 , with
_ ia i 1 _ i ,
_ a_2 = _ i a i 2 _ i ; + _ b_1 = _ ib i 1 _ i ,
_ b_2 = _ ib i 2 _ i . therefore , these operators satisfy the following relations , @xmath28 here we note that , as discussed in the previous section , the spin state within each sublattice is ferromagnetic .
this means that not only must the total spin quantum numbers of @xmath29 and @xmath30 take their maximum values , but the total spin quantum numbers of @xmath31 and @xmath32 must also take their maximum values .
more importantly , these values are thus fixed , which enables us to treat the operators @xmath33 and @xmath32 as four single spins , and in what follows we shall denote these operators by their corresponding spin quantum numbers .
the problem is then that we are given a four spin state in which the spins @xmath34 and @xmath35 are combined into a state with total spin @xmath10 and the spins @xmath36 and @xmath37 are combined in a state with total spin @xmath11 , and then these two states are combined into a total singlet ( resulting in the ground state of our long - range afm model ) , and we must express this state in a basis in which the spins @xmath34 and @xmath36 are combined into a state with total spin @xmath38 and @xmath35 and @xmath37 are combined into a state with total spin @xmath39 .
this change of basis involves the familiar ls - jj coupling scheme .
we proceed by first rewriting the density matrix describing the ground state in terms of the basis states @xmath40 by appropriate insertions of the unit operator @xmath41 [ eqn : rdm ] & = |s_as_b;sms_as_b;sm| = | ; 00 ; 00| + & = _
|s_1 m_1 s_2 m_2 s_1 m_1 s_2 m_2|s_as_b;sm + & s_as_b;sm| _ |s_1 m_1 s_2 m_2 s_1 m_1 s_2 m_2| + & = _ _ s_1 m_1 s_2 m_2 _ s_1 m_1 s_2 m_2^*|s_1 m_1 s_2 m_2 s_1 m_1 s_2 m_2| , + where @xmath42 .
we then insert another unit operator @xmath43 to express this matrix element as the product of a clebsch - gordan coefficient and an ls - jj coupling coefficient@xcite , @xmath44 here @xmath45 is the ls - jj coupling coefficient or x - coefficient defined as , l_1 & s_1 & j_1 + l_2 & s_2 & j_2 + l & s & j + = l_1 l_2 s_1 s_2 l s j m | l_1 s_1 l_2 s_2 j_1 j_2 j m. since @xmath46 , we have @xmath47 .
so @xmath48 , and for simplicity , we can now suppress the @xmath39 and @xmath49 indices and represent @xmath50 by @xmath51 without causing any ambiguity .
therefore , @xmath52 in the following calculation , for consistency , we adopt the convention of wigner 6-j and 9-j symbols , the definitions of which , and the relation to the racah coefficients and ls - jj coupling coefficients ( or x - coefficients ) , are given in appendix a. following wigner s convention , eq .
( [ eqn : lsjj ] ) can be written as , @xmath53 in our case @xmath54 , so the 9-j symbol can be expressed more simply in terms of a 6-j symbol@xcite ( see also appendix b ) , @xmath55 in our particular case , within each sublattice , the state is ferromagnetic , @xmath56 .
let @xmath57 be the size ( number of lattice sites ) of subsystem 1 . without losing any generality
, we can let @xmath58 .
applying eq .
( [ eqn : szero ] ) we then have , @xmath59 using a symmetry property of the racah coefficients@xcite @xmath60 , we have , @xmath61 then we employ the following relation@xcite : @xmath62 ^{\frac{1}{2}}.\\ \end{split}\ ] ] this gives @xmath63^\frac{1}{2}. \end{split}\ ] ] now with the expression above , we can directly trace out the degrees of freedom in subsystem 2 from eq .
( [ eqn : rdm ] ) , with the result _ 1 = tr_(2 ) ( ) = _ s_1 m_1 |_s_1 m_1|^2|s_1 m_1 s_1 m_1| . the bipartite entanglement entropy between subsystems 1 and 2
is then given by @xmath64 here we note that , although @xmath51 is written with an explicit @xmath65 dependence , the actual expression is independent of @xmath65 . as a result , we can eliminate the summation over @xmath65 from ( [ eqn : entropy1 ] ) by multiplying by a factor of @xmath66 .
the final expression for the entanglement entropy is then @xmath67 in the following section we will first derive the asymptotic behavior of the entanglement entropy @xmath20 in certain limits using the exact results above , and then present the results of our numerical calculation of @xmath20 with finite system sizes .
in this section we consider two limiting cases : ( i)@xmath68 , this case gives the saturated entropy at fixed @xmath69 since intuitively @xmath20 should increase with the subsystem size .
\(ii ) @xmath70 , in this limit we are considering system s entanglement with its ( much larger ) environment , and generically we should be able to find that the entropy should be independent of the total system size as @xmath71 , which is indeed what we find . to obtain the asymptotic behavior of the entanglement entropy , we first consider the asymptotic behavior of @xmath51 , then in the large @xmath69 limit . then the summation over @xmath38 which runs from 0 to @xmath72 can be replaced by an integral .
as we will see , the distribution of @xmath51 is proportional to a gaussian function with respect to @xmath38 , thus the bounds of integration can be extended to from 0 to @xmath73 . our concern is the distribution of @xmath51 with respect to @xmath38 , therefore , we extract the dependence on @xmath38 and then use the normalization condition @xmath74 to obtain the normalization constant . from eq .
( [ eqn : lambda ] ) we can write , _ s_1m_1 ^ 2 & ~ + & + & ~e^- - .
+ this approximation is generally good as long as both @xmath75 and @xmath76 are large . in our work ,
@xmath77 , and @xmath75 is assumed to be @xmath78 ( equal partition ) or @xmath79 , so this criteria is satisfied quite well in both cases . then by the normalization condition &
_ s_1a(2s_1 + 1)e^-s^2_1(+ ) + & _ 0^a(2s_1 + 1)e^-s_1 ^ 2(+)ds_1 = 1 , + we obtain the ( inverse ) precoefficient @xmath80 . therefore , _
s_1m_1 ^ 2 e^- - .
the entanglement entropy is therefore given by @xmath81 here for simplicity we set @xmath69 even ( note that the total system size is @xmath18 ) , thus @xmath82 , and @xmath83 , then @xmath84 and the entanglement entropy becomes , @xmath85 compared with our numerical result , as shown in fig
. [ fig.ep ] , we see that they agree very well , not only for the prefactor of the @xmath86 dependence , but the intersection coefficient as well . in this case , @xmath70 , we can expand the entropy as follows , e & + & = n_1 + ( 1 - + ) + & n_1 - + o(()^2 ) .
from this expression we see that when the assumed condition is satisfied the entropy indeed depends only on the subsystem size to leading order .
to check to what extent our approximation is still valid , we plot the expression as a function of total system size @xmath69 with different subsystem size @xmath75 , as shown in fig . [ fig.uep.n ] .
this figure shows that when @xmath69 is large enough the entropy becomes independent of @xmath69 , and the first order term , @xmath87 dominates except for an apparently constant shift from the exact curve .
we have also plotted the asymptotic behavior as a function of @xmath75 while keeping @xmath69 fixed , shown in fig.[fig.uep.l ] .
in this section we consider a ferromagnetic ( fm ) spin-1/2 model on an arbitrary lattice with @xmath69 sites , h=-_ijj_ij_i_j , with @xmath88 .
the ground state is the fully magnetized state @xmath89 with @xmath90 and @xmath91 , and is clearly long - range ordered : @xmath92 .
however there is a crucial difference between the fm ground state and the afm ground state studied earlier : the fm ground state has a finite degeneracy , and thus the system exhibits a non - zero entropy even at zero temperature , @xmath93 , resulting from the density matrix of the entire system , & = _ m =-
s^s | sm sm | + & = _
s^s | sm sm | + & = ( ccccc 1 & 0 & & & + 0 & 1 & 0 & & + & 0 & 1 & 0 & + & & & & + & & & 0 & 1 ) + in this case the entanglement entropy between two subsystems ( 1 and 2 ) is defined in the following manner .
we first obtain reduced density matrices for subsystems 1 and 2 by tracing out degrees of freedom in 2 and 1 from @xmath94 : & _ 1=tr_(2)= _ m = -s^s
tr_(2 ) ( | sm sm | ) + & = _ m = -s^s _ 1 m , + & _
2= tr_(1)= _ m = -s^s tr_(1 ) ( | sm sm | ) + & = _ m = -s^s _ 2 m , and calculate from them the entropy of the subsystems , @xmath95 and @xmath22 .
the entanglement entropy is defined as@xcite e = ( e_1 + e_2 - e_0 ) / 2 . for the present case @xmath21 and @xmath22
can be easily obtained from the following observations .
( i ) because the total spin is fully magnetized , so are those in the subsystems : @xmath96 and @xmath97 .
thus this is a two - spin entanglement problem .
( ii ) because the total density matrix @xmath94 is proportional to the identity matrix in the ground state subspace , it is invariant under an arbitrary rotation in this subspace .
( iii ) as a result the reduced density matrix @xmath24 is also invariant under rotation in the subspace of subsystem 1 with @xmath96 , and is proportional to the identity matrix in this subspace .
thus _ 1 & = _ m_1=-^ | m_1 m_1 | + & = ( ccccc 1 & 0 & & & + 0 & 1 & 0 & & + & 0 & 1 & 0 & + & & & & + & & & 0 & 1 ) , + and @xmath98 ( in agreement with ref .
@xcite ) . similarly @xmath99 . thus e = [ ( n_1 + 1)+(n_2 + 1)-(n+1)]/2 .
we find in both the equal partition ( @xmath100 , here again we set @xmath69 even for simplicity ) and unequal partition ( @xmath101 ) limits , the entropy grows logarithmically with subsystem size @xmath75 , _
n_1e ( n_1 ) .
in this paper we have studied antiferromagnetic ( afm ) and ferromagnetic ( fm ) spin models with perfect long - range magnetic order in their ground states , and calculated the ground state block entanglement entropy when the system is divided into two subsystems ( or blocks ) . in both cases
we find the entropy grows logarithmically with block size . in the following
we discuss the significance of our results .
first of all , there _ is _ entanglement , despite the perfect long - range order in the ground state .
this is somewhat surprising as one might think that in such hamiltonians one can obtain certain properties of the system exactly using mean - field theory , in which the ground state is approximated by a product state with no entanglement .
our results indicate that mean - field approximation is _ not _ appropriate for entanglement calculation , even if it is exact " for other purposes .
this point is particularly striking for the afm model , in which the ground state is unique .
the source of the discrepancy is there still is quantum fluctuations even in such a model with super long - range interaction , which render the ground state a singlet , even though quantum fluctuation does not reduce the size of the order parameter .
the entanglement is due to the quantum fluctuation of the _ direction _ of the order parameter , which is a collective mode with zero wave - vector ( or a zero mode ) ; this is missed by any mean - field approximation .
secondly , the entropy does _ not _ obey the area law .
the reasons for that are different for the two cases we studied . in the afm model ,
the interaction does not depend on distance in the hamiltonian , thus there is no notion of distance or area in this model . in the fm model , on the other hand , the ground state is _ independent _ of the hamiltonian , as long as all interactions are fm .
the fully magnetized ground states are invariant under permutation of spins , thus there is no notion of distance or area in the _ ground states _ , even though the terms in hamiltonian can have distance dependence .
thirdly , the absence of an area law is special to the cases we studied , again each in their own ways . for the fm model ,
it is specific to zero temperature .
as soon as a finite temperature is turned on , one expects the entanglement entropy ( or mutual information ) to grow with the area separating the two subsystems or blocks , as long as the interaction is not long - ranged@xcite .
for the afm model at zero temperature , we do expect an area - law contribution to the entropy for short - range or even certain power - law long - ranged interaction due to quantum fluctuations .
this is most easily seen within spin - wave approximation , which is a version of mean - field theory . within the spin wave approximation spins
are mapped onto bosons , and the hamiltonian is mapped onto coupled harmonic oscillators .
detailed recent studies have established the area law of entanglement for such systems@xcite . in this regard
the super long - range afm model we study here is very special , in that all spin - wave degrees of freedom at finite wave - vector disappear , and the only degrees of freedom contributing to the ground state are zero wave - vector modes represented by @xmath102 and @xmath103@xcite ; as a result there is no area - law contribution from quantum fluctuations of spin - waves .
we conclude by speculating that in the more general cases that do have an area - law contribution to the entanglement entropy , as long as long - range spin order is present , the logarithmic contribution due to fluctuations of the order parameter zero - modes we find here will remain and show up as a sub - leading ( yet singular ) correction to the area law
. if that is the case , then conventional long - range order contributes to the entanglement entropy in a way similar to the much subtler topological order@xcite or quantum criticality@xcite .
6-j symbol & racah coefficient : @xmath104^{-\frac{1}{2}}\\ & \times \bra j_1,j_{23};j|j_{12},j_3;j\ket.\\ \end{split}\ ] ] where @xmath105 and @xmath106 refer to the coupling of @xmath107 and @xmath108 or @xmath108 and @xmath109 respectively .
in this appendix we give a slightly simplified derivation of the entanglement entropy for the singlet ground state of the infinite - range afm model defined in sec .
ii . while this derivation is not as general as that given in the main text ( which can , in principle , be applied to states with nonzero total spin )
, it has the advantage of clarifying the reason for the appearance of the wigner 6j - symbol in the final result for the entanglement entropy . following the notation of sec .
ii and iii , we consider the case of @xmath18 spin-1/2 particles divided equally into two sublattices @xmath111 and @xmath112 .
the spins on each sublattice are taken to be fully polarized , so that @xmath113 and @xmath114 . in what follows
we will use a notation in which , for example , the state for which the spins @xmath10 and @xmath11 form a singlet is represented as @xmath115 . in this notation ,
pairs of spins contained within a set of parenthesis form a state whose total spin is equal to the subscript labeling the parenthesis .
thus , @xmath116 is a singlet state equivalent to the state defined in eq .
( [ singletneel ] ) in sec .
( needless to say , for this state to exist it is necessary to have @xmath117 . )
now we consider what happens if these spins are divided into two different subsystems labeled @xmath118 and @xmath119 .
following sec .
ii , if @xmath120 is the total spin quantum number of the @xmath111 sublattice spins in subsystem @xmath121 , and @xmath122 is the total spin quantum number of the @xmath112 sublattice spins in subsystem @xmath121 , then the state of the spins on the @xmath111 sublattice can be written @xmath123 and the state of the spins on the @xmath112 sublattice can be written @xmath124 .
the total singlet state for the entire system , @xmath125 , can then be expressed as @xmath126 note that we have not included the @xmath127 quantum numbers associated with the @xmath128-components of @xmath10 and @xmath11 in writing the above expression for @xmath125 .
this is not necessary because the requirement that the total spin of the two sublattices combine to form a singlet uniquely determines the state .
it should , of course , always be understood that when we write eq .
( [ singlet ] ) what is really meant ( in obvious notation ) is , @xmath129 equation ( [ trivial ] ) effectively ( up to irrelevant phase factors ) gives the schmidt decomposition ( see , e.g. , @xcite ) of the state @xmath125 into the two subsystems consisting of the @xmath111 and @xmath112 sublattices .
given such a schmidt decomposition it is straightforward to determine the entanglement between these two subsystems .
however , here we are interested not in the trivial entanglement between subsystems @xmath111 and @xmath112 , but the entanglement between subsystems 1 and 2 . to find this we need the schmidt decomposition of @xmath125 into subsystems 1 and 2 . note
that because @xmath125 is a total spin singlet it follows that @xmath130 where , for convenience , we have rearranged the order of the spins in the second equality .
the key point here is that the three spins @xmath34 , @xmath35 and @xmath37 must be in a state which is a total spin eigenstate with total spin @xmath36 . if this were not the case it would be impossible to form a total spin singlet with the remaining spin @xmath36 .
next we use the wigner @xmath131-symbol to express the three spin state @xmath132 as a superposition of states of the form @xmath133 . the result is @xmath134 where the coefficients are given by @xmath135 ( see appendix a ) . rearranging the spins again , and using the fact that the total spin of all four spins is 0 ,
we can express the resulting four spin basis states as follows , @xmath136 where , obviously , @xmath137 .
we therefore conclude that @xmath138 finally , after writing the @xmath127 sum explicitly ( as in eq .
( [ trivial ] ) ) , we have the desired schmidt decomposition ( again , up to irrelevant phases ) of the state @xmath125 into subsystems 1 and 2 , @xmath139 where @xmath140 when the values of @xmath113 , @xmath141 , @xmath142 are substituted into this expression it can be seen to be equivalent to that given in eq .
( [ schmidt ] ) in sec .
the derivation of the entanglement entropy given in the main text follows . c. holzhey , f. larsen , and f. wilczek , nucl .
b * 424 * , 443 ( 1994 ) ; g. vidal , j. i. latorre , e. rico , and a. kitaev , phys .
lett . * 90 * , 227902 ( 2003 ) ; p. calabrese and j. cardy , j. stat .
( 2004 ) p06002 ; g. refael and j.e .
moore , phys .
lett . * 93 * , 260602 ( 2004 ) ; r. santachiara , j. stat .
mech . : theory exp .
, l06002 ( 2006 ) ; adrian feiguin , simon trebst , andreas w. w. ludwig , matthias troyer , alexei kitaev , zhenghan wang , michael h. freedman , phys .
lett . * 98 * , 160409 ( 2007 ) ; n. e. bonesteel and k. yang , phys .
* 99 * , 140405 ( 2007 ) .
the entanglement property of this model has bee previosuly studied in ref .. here we introduce a different definition of the entanglement entropy that properly takes into account the ground state degeneracy , and can be calcualted much more straigtforwardly using symmetry consideration .
see sec . v. this definition is probably not unique .
it is one half of the mutual information " introduced in ref . , and reduces to eq .
( [ entanglemententropy ] ) when @xmath94 is that of a pure state .
the same definition was used by castelnovo and chamon [ claudio castelnovo , and claudio chamon , phys .
b * 76 * , 174416 ( 2007 ) ] . the name mutual information " may be first coined by adami and cerf [ c. adami and n.j .
cerf , phys .
a * 56 * , 3470 ( 1997 ) ] and vedral , plenio , rippin and knight [ v. vedral , m.b .
plenio , m.a .
rippin , and p.l .
knight , phys .
lett . * 78 * , 2275 ( 1997 ) ] , although stratonovich [ r. l. stratonovich , izv .
, radiofiz .
* 8 * , 116 ( 1965 ) ; probl .
* 2 * , 35 ( 1966 ) ] considered this quantity already in the mid-1960s . | in this paper we calculate the block entanglement entropies of spin models whose ground states have perfect antiferromagnetic or ferromagnetic long - range order . in the latter case the definition of entanglement entropy is extended to properly take into account the ground state degeneracy .
we find in both cases the entropy grows logarithmically with the block size .
implication of our results on states with general long - range order will be discussed . |
over the years we watched ourselves working back and forth between writing equations for clocks and signals on a blackboard and working with lasers , lenses , and electronics on a work bench . in the course of this experience we noticed the role in physics of memories , both the memories of the investigators and the memories of the digital computers they employ , and our eyes opened to unsuspected vistas .
we speak of _ memory _ as belonging to a _ party _ , which can be a person , a computing machine , _ etc_. as we mean it , a memory is a device in which symbols are recorded and manipulated . by _ memory _ we mean no static photograph , but a dynamic device in which the symbols recorded can undergo changes from moment to moment . by _ symbols _ we mean what is recorded in a memory of a party , distinct from whatever propagates externally from one party to another party , which we call a _ signal_. the elemental symbol carries a binary distinction : the bit .
what we call a party or a symbol or a signal depends on the level of description , which can be finer or coarser . by a change in level of description ,
what is termed `` a memory '' belonging to a single party can become several memories belonging to distinct parties , with communications among them , and _
vice versa_. thus the distinction between _ symbol _ and _ signal _ is relative to the memory of a party , and both the memory and the party are relative to a level of description . as noted in sec .
[ subsec:5.1 ] changes in levels of descriptions will be seen to correspond to morphisms of graphs .
regardless of how one imagines mathematical entities , their expression in symbols is physical , _
e.g. _ as ink on paper or voltages in a computer memory .
symbols in formulas and symbols of evidence from experiments live in memories .
thinking of symbols as physical attributes of memory , with associated dynamics and rhythms , offers a physical analog of gdel coding : one can inquire into the timing and location of symbols , both symbols of theory expressing classical or quantum states and also symbols expressing evidence extracted from experiments .
a familiar blackboard " picture of memory is the turing - machine tape , divided into squares ; on each square a symbol `` 0 '' or a symbol `` 1 '' can be written or erased . now lift up this abstraction to recall that the physical mechanism of computer memory is a single device
what engineers call a clocked set - reset flip - flop @xcite that recognizes a binary symbol carried by a signal , and as part and parcel of the act of recognition also acts as a memory device by recording the symbol .
the flip - flop works as a damped inverted pendulum , a hinge if you will , with its exposure to signals from outside cycled by a driven adjustable pendulum , in effect a clock .
noticing that a physical implementation of a turing machine depends on the flip - flop allows one to see symbols as physical objects .
then one can inquire into the motion of symbols , and into the relation of that motion to concepts of spatial and temporal order .
the paired - pendulum mechanism acts as a physical unit of computation and also , through its participation in the machinery of radar , as a physical unit of geometry .
we make a distinction between _ recognizing a symbol in a signal _ and _ measuring the signal_. in recognition , the hinge in the memory of a party falls one way or the other to express the symbol ; further , the hinge - position - as - symbol can be copied to flip - flops in the memories of other parties .
in contrast , measurement is idiosyncratic , characterized by error bars , and no two instances of a measurement can be expected to agree exactly .
the results of a measurement , though idiosyncratic , can be expressed in symbols ( digitized ) , but only after waiting for hinges to fall one way or the other , and with a ( usually small ) risk of confusion . out of the buzzing world of experience , fingers on knobs , tweaking adjustments to bring optics into alignment _
etc_. , comes , one way or another , _ evidence _ from an experiment . by _ evidence
_ we mean expressions in mathematical language taken as reflecting experience on the work bench .
theory , quantum or otherwise , offers explanations , such as explanations in terms of quantum state vectors and operators or explanations in terms of a general - relativistic 4-manifold .
an explanation asserts ( rightly wrongly ) properties of evidence .
experience evades direct comparison with theory , but in memories symbols for evidence reflecting experience can be compared against assertions about evidence implied by explanations .
( see fig .
[ fig:1 ] . ) recognizing the role of memory as the holder of evidence written in symbols splits the question of the relation between theory and experience into two questions : 1 .
how well does evidence in a memory reflect the experience of an investigator ? 2 .
how well does an assertion about evidence implied by an explanation fit actual evidence ? mathematical structures ( _ e.g. _ axioms ) for explanations have been much studied .
we raise the parallel question : what mathematical structures are to be found or invented to express evidence ? in this report we concentrate on structures of evidence recordable in the memories of communicating parties , to do with the timing ( not the content ) of their communications .
one party communicates a symbol from its memory to the memory of another party via a signal . in propagating from one party to another ,
a signal deforms unpredictably , so the recognition of a symbol carried by a signal must be insensitive to a range of deformations .
the damped inverted pendulum of the paired - pendulum recognition mechanism offers this insensitivity , provided that the _ rhythm of communication meshes the arrival of the part of a signal that carries a symbol with the phase of symbol recognition_. the receiver must look at the signal when the symbol is present , within some leeway but not too much earlier or too much later . by its dependence on the meshing of the part of a signal that carries a symbol with a receiving party s phase of recognition , the paired - pendulum mechanism of the flip - flop shapes evidence recordable from a communications network by imposing discrete phases of the adjustable pendulum for signal reception , leading to a single form of evidence , regardless of whether explanations for evidence
are stated in quantum terms or in terms of general relativity .
the symbol recognized in a signal can not be a function of the signal alone .
to communicate , two parties must share some axioms in common , and also _ share a rhythm _ that meshes their clocks with the signal propagating from one to the other .
the rhythm , once acquired , must be maintained , and its maintenance depends on reaching beyond the logic of symbol recognition : the rhythm of symbol exchange is maintained not by recognitions but by _ measurements of signal arrivals relative to pendulum phases_. these measurements are subject to idiosyncrasies of each party , on which the other party must rely : an _ intimacy _ necessary to the communication of a symbol from one party to another .
radar as the instrument by which spacetime is conceived will be shown to have an analog in the timing of symbols communicated among memories of a synchronized network .
( indeed a working radar depends on the communication of symbols , such as those identifying targets . )
evidence of the timing of signals transmitted and received in a network of communicating parties , recorded in their memories , has a mathematical form independent of metric assumptions involved in explanations based on the special or general theory of relativity .
we show this form in a record format which we relate functorially to colored , directed graphs .
the graphs expressing records of communication networks , such the global positioning system ( gps ) , assume neither a general - relativistic geometry , nor quantum states . because of this freedom from additional assumptions needed for one or the other form of explanation
, we will show how the graphical expression of evidence offers a platform on which to negotiate the joint participation of quantum theory and general relativity in explanations of evidence from networks of communicating parties .
we think of a memory as belonging to a communicating _ party _ , a person or a machine . as a first cut , model a _ party _ by a turing machine moved by a _ driven adjustable pendulum_a clock with a faster - slower lever . following turing @xcite , we think of the history of a party s memory as segmented into moments interspersed by moves , but , unlike turing s history of a memory as a sequence of snap shots at successive moments , we need to inquire into what happens during a move in which the symbols in memory can change . thus we view turing s `` move '' not as something structureless but as a phase of positive duration , during which there can be measurements of clock readings . picture the clock that moves a turing machine as moving its hand cyclically around a circle marked in subdivisions of the unit interval , so that a reading of the hand position is the clock reading modulo integers .
take the phase ` move ' to be an interval of the circle that includes the position `` 12 oclock '' at the top of dial and the phase ` moment ' as a disjoint interval that includes the `` 6 oclock '' position at the bottom of the dial . at the level of description appropriate to an engineer who probes the operation of a computer memory ,
the memory itself becomes a network of communicating `` sub - parties , '' each with a piece of the computer memory .
symbols are conveyed by signals from one piece of memory to another .
because of uncontrolled deformations as the signal propagates and because , on the workbench , no two things ever get built quite alike , the signal that carries a symbol to a receiving sub - party is subject to unpredictable deformations ; and beyond these practicalities , lower limits to signal variability are implied by quantum theory . for this reason , the mechanism for recognizing a symbol carried by a signal must be made insensitive to small variations in the signal ; _
i.e. _ , the signal has to be allowed a certain _ leeway _ in both its shape and its timing . in terms of differential equations , recognizing a single symbol
regardless of a certain variation in the signal requires an attractor leading to each symbol , with the implication that between attractors there are unstable equilibria .
the insensitivity to variations in the signal requires damping , in conflict with any quantum explanation that invokes only the unitary evolution of a schrdinger equation . physically , the simplest memory element for recording a choice between two symbols consists of paired pendulums , one inverted and damped , with two stable positions and an unstable equilibrium between them , the other the adjustable pendulum of a clock , swinging through phases as part of a rhythm of communication , opening and shutting a gate to allow a signal to flip over or not to flip over the inverted pendulum that holds a bit .
a `` bit '' is thus a snap shot of a livelier creature a recognition - and - memory device that not only can display a `` 0 '' or a `` 1 '' but , when the rhythm of its operation is disturbed , can teeter in an unstable equilibrium like a flipped coin landing on edge , where it can hang , lingering , with no sharp limit on how long it can take to show a clear head or tail .
we are to think of a bit not as a 0 or 1 on a turing tape but as the position of the inverted pendulum at a moment . in computer hardware , the inverted pendulum gated by a clock
is called a clocked set - reset ( s - r ) flip - flop @xcite . without adequate maintenance of the rhythm of communication
the part of a signal in which a bit is to be recognized can arrive at a receiving party in a race with the closing of the gate , resulting in `` runt signal '' squeaking through the gate , big enough to push the inverted pendulum ( think of a hinge ) part way but not all the way over , leaving the hinge hung up in an unstable `` in between '' state @xcite .
we say the signal straddles a timing boundary .
known to engineers concerned with the synchronization of digital communications , such hang - up causes logical confusion .
computation requires acts of copying symbols : a symbol in flip - flop a at one moment is copied into two flip - flops , say b and c , at a later moment , so that whatever bit value was in a at the earlier moment appears in both b and c at the later moment both hold 0 or both hold 1 ; one speaks of `` fan - out . ''
if flip - flop a hangs up in an unstable equilibrium , then flip - flops b and c not only may hang up , but can `` fall differently '' so that the symbol in b , instead of matching that in c , conflicts with it . in revealing conflicts in response to an unstable condition of a flip - flop a , the fan - out from a to flip - flops b and c also offers a means of detecting unstable conditions , which has been used to show a roughly exponential decline with waiting time of the probability of disagreement between b and c , resulting in the measurement of a _ half - life _
@xmath0 of the instability @xcite .
( for silicon integrated circuits we found @xmath0 to be close to 1 ns .
modern gallium arsenide circuits operate much faster , and efforts to shorten their half - life are underway , but so far their ratio of half - life to cycle period is not much less than that for silicon @xcite . ) in quantum explanations , one describes the inverted pendulum by a wave function , putting planck s constant into the relation between the short time constant required for rhythmic operation of the flip - flop and the long time that must be waited for it to settle down when subject to the straddling of timing boundaries and the ensuing runt pulses @xcite.=-1 in its use to decide a race among more than two signals , the teetering hinge of a flip - flop has a noteworthy consequence .
consider the case of a three - way race among signals @xmath1 , @xmath2 , and @xmath3 arriving at a clock .
a world line in a general - relativistic explanation of this clock corresponds on the workbench not to one device but to several interconnected devices .
each of the three signals fans out to allow three separate pairwise comparisons of `` which came before which '' . in a close race , teetering in all three pairwise comparisons can result in finding : @xmath4 , and @xmath5 , violating the transitivity of an ordering relation , and suggesting a limit on the validity of even local temporal ordering .
making sense out of temporal order requires distinguishing the question of which cycle a symbol recognition occurred from the question of when within a cycle did a signal arrive.=-1 remarks : 1 . to reduce the risk of disagreement between b and c
, it suffices to wait after the setting of a to the reading of b and c. the literature on digital circuits discusses the related use of `` arbiters''of which there are two types , one that might take forever , the other that might generate confusion .
2 . weeks after a given day , gps publishes corrections to coordinates for events that it issued on that day , derived from subsequent cross comparisons among its clock readings recorded at the transmission and reception of radio signals .
although the process of comparing and correcting may yet be greatly speeded , not only does the delay in communicating comparisons limit how quickly one can determine what the clock readings `` should have been , '' but an additional delay is imposed by the balancing instrument used to convert analog measurements to digital signals suitable for communication .
for theoretical purposes , we assume the conceptually simplest ( but not the most used ) scheme for digital communications , called _ synchronous communication _ @xcite , which offers the fastest response . in synchronous communication
a receiver recognizes symbols one by one ( without use of sample - and - hold techniques @xcite ) .
synchronous communication from a party a to a party b , moved by clocks a and b , respectively , requires that a symbol be transmitted from clock a while a s clock hand is in the 12 oclock `` move '' phase and must arrive a b while b s clock hand is also in a 12 oclock `` move '' phase .
clocks , including the atomic clocks used to generate international atomic time ( tai ) , drift unpredictably in rate , leading eventually to unbounded phase drift between two nearby clocks , with the result that clocks function only in a network of comparisons that guide adjustments of clock rates over some ( possibly small ) range .
in addition , communications involve other perturbing circumstances , including doppler shifts among parties in motion .
unless the clock of a receiving party can be maintained so the phases of reception are aligned with the arrivals of symbol - carrying signals , the recognition of a symbol carried by a signal fails .
suppose that the conditions of phasing allowing synchronous communication between two parties have been brought into being a story in itself @xcite .
to maintain these conditions over a succession of symbols requires more or less continual adjustment of the motion of the clocks : their accelerations , their rates of ticking , or both . in all cases
the adjustment is guided by departures from nominal behavior of the arriving symbols relative to an imagined center of the phase of reception , much as steering an automobile toward the center of a lane depends on noticing and responding to its departure from the center.=-1 how then to determine the departures in the clock reading of a receiving party at a signal arrival ?
let the reading of the clock of the a receiving party relative to the 12 oclock center of the receptive move , modulo integers , be symbolized by @xmath6 .
in order to guide adjustment necessary to the maintenance of synchronous communication , the offset symbolized by @xmath6 has to be made to act on a `` lever '' ( as in the lever on the back of a wind - up clock by which its rate of ticking is adjusted ) .
( if @xmath7 the receiver clock needs to be showed down relative to the arriving symbols , and speeded if @xmath8 . ) in hardware , the symbol `` @xmath6 '' never appears .
for example , one way to guide adjustment is by `` bang - bang '' control that responds to whether the part of a signal that carries a signal arrives before or after a nominal clock reading within the receptive phase . for this a logical and gate
is used not as part of a device for recognizing symbols but as a measuring device in a feedback loop that controls the rate of ticking of the receiving party .
the and gate is opened at the beginning of the cycle to the arrival of the signal but turned off at the nominal reading . if the signal arrives well before the turn - off it passes through the and gate to put a pulse of charge on a capacitor . a running average of the charge on the capacitor controls the faster - slower lever of the party s clock .
close races between the arriving signal and the turn - off produce runt pulses without causing any logical confusion , for the runt pulses never need to be recognized as symbols ; instead the pulses , runt or not , pile up like gravel that is shoveled without the stones being counted .
the point is that the fine - grained determination of clock reading within a receptive phase at the arrival of a symbol can not be _ recognized _ as a symbol but requires something distinctly different , which we call _ measurement _ , as follows .
symbol recognition depends not only on leeway but also on avoiding `` straddling of boundaries . ''
to recognize a symbol , such as the arrival of a pawn on a square of a chess board , the act of looking must be coordinated with the arrival so that a party looks while the pawn is in the square and not sliding over a boundary that it straddles as it moves . it is these conditions of `` no straddle '' and `` leeway '' that allow two parties to agree exactly in their recognitions of symbols .
in contrast to the recognition of symbols , we speak of _ measurement _ as in the determination of a mass in a balance , for which no two parties can expect to agree exactly ; instead one speaks of error bars .
the idiosyncratic variations among parties resulting in error bars are inescapable precisely because of the straddling of boundaries and the lack of leeway .
although distinct , ` measuring ' and ` recognizing ' depend on one another .
for example , in measuring using a balance instrument , i have to recognize weight a , weight b , and that `` they balance '' or `` the balance tips toward a , '' and if i am wrong in such noticing , my measuring makes no sense .
indeed , in spite of their neglect in physics education , recognitions are essential to logic , without which physics collapses . going the other way , recognitions basic to logical communication turn out to take place in rhythms that require maintenance adjustments to clock rates guided by measurements . with the distinction between recognizing and measuring in mind , we return to the issue of determining a departure from a desired clock hand position within a phase at the arrival of ( the center of ) a symbol carried by a signal . in its use to recognize a symbol
the mechanism of an inverted pendulum must be insensitive to the very timing variations of interest within the leeway of the phase of reception .
when finer - grained distinctions necessary to determining the clock hand position are implemented , straddling of boundaries is unavoidable , and the distinctions can not be _ recognitions _ but depend on _ measurements_. altogether we arrive at : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ fine - grained `` local clock readings''necessary to maintaining rhythms essential to the communication of logical symbols constitute measurements , idiosyncratic in that no exact agreement can be expected between any two measuring parties .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ only in special situations can a receiving party recognize in a signal the symbol intended by a transmitting party .
for example , when two people converse : each person s ear hears the symbols what the other s mouth puts into a spoken signal . for communication
the two parties have to share not only concepts but a rhythm , and the establishment of that rhythm requires reaching beyond logical recognitions to rely on necessarily idiosyncratic measurements that guide the adjustments needed to maintain the rhythm .
the conditions of shared concepts and a shared rhythm necessary to communication can reasonably be called _ intimacy_. without this intimacy of communication in which symbols are conveyed , there can be no logic , mo mathematics , and no physics .
in working back and forth between experiments on the optics bench and writing quantum states on the blackboard , we saw lens holders and lenses and lasers on the optics bench , but _ no _ quantum state vectors .
but _ must _ state vectors be invisible on the bench ?
nobody can lay formulas on top of an optics bench to see if they fit
. to be compared with mathematically expressed explanations raw experience with lenses and mirrors has to be first reflected into _ evidence written in symbols of a mathematical system based on axioms _ , recorded in a memory .
so our question became : can mathematically expressed evidence in a record ever determine its own explanation ?
the answer hinges on a striking property of quantum theory . in pre - quantum physics , including general relativity , the mathematical system available for expressing evidence involves the same axioms as that for expressing explanations .
quantum theory differs by invoking two distinct mathematical systems , hilbert - space constructions for _ explanations _ , and a distinct other system of probability measures for assertions about _ evidence _ implied by an explanation : [ cols="^,^,^,^ " , ] such graph fragments can be pasted together by condensing a signal edge of the graph for party a for transmission to another party b and the signal edge for reception of a s transmission by a party b into a single edge from the transmission move of a to the reception move of b , as follows .
a vertex at the head of a signal edge from a move of a at count @xmath9 is overlaid on a vertex colored a@xmath9 at the tail of an edge to reception at a move of party b , the vertex is removed and the signal arrows joined head to tail into a single directed edge .
this is illustrated in fig .
[ fig:3 ] .
such graphs are essentially _ occurrence graphs _
@xcite , specialized to exhibit a distinct trail for each party , with edges for signals linking parties .
when `` analog '' measurements with their idiosyncrasies that color the occurrence graphs are forgotten , the occurrence graph for a network of communicating parties can exhibit symmetry , illustrated in fig . [ fig:4 ] . in some interesting cases , forgetting the coloring by fine - grained clock readings and rate settings , an occurrence graph can be `` wrapped around '' to form a marked graph @xcite , as in figs .
[ fig:5 ] and [ fig:6 ] .
figure [ fig:7 ] shows an example of an occurrence graph for a network in which one set of parties is in motion relative to another set of parties .
occurrence graphs , marked graphs , and , more general petri nets @xcite form categories with interesting graph morphisms . going the other way , one studies morphisms among network histories , aided by the functor from network histories to occurrence graphs .
redrawn as a role - activity graph , with a vertical trail for each of four parties .
a circle at the top of a trail is identified with the circle at the bottom of the trail , and vertices ( square boxes ) connected by a horizontal line are identified .
vertical edges are understood to be downward - directed.,width=264 ] redrawn as a role - activity graph , with a vertical trail for each of four parties .
a circle at the top of a trail is identified with the circle at the bottom of the trail , and vertices ( square boxes ) connected by a horizontal line are identified .
vertical edges are understood to be downward - directed.,width=240 ] moving past one another .
solid boxes indicate a meeting between a party of one set and a party of another set .
all edges are directed downward.,width=305 ] the graphs are objects of respective categories in which morphisms include ( 1 ) isomorphisms from one induced subgraph to another ; ( 2 ) inclusions ; and ( 3 ) epimorphisms in which certain stretches of clock image over several vertices for moves along with neighbor - to - neighbor signals map to a single vertex .
example : view main memory as one party and view auxiliary memory as a second party ; then map the two parties into a single `` turing - machine '' party . by another such map , illustrated in fig .
[ fig:8 ] , a vertex at which two signal edges meet a party can be seen as a condensation of a pattern involving two parties , each with a vertex involving only one signal edge .
occurrence graphs of this form of `` no more than one signal per party vertex '' map to virtual braid diagrams @xcite , and it will be interesting to see what interpretation , if any , to make of virtual - braid isotopies .
a nice path to study more complex synchronization methods , including those used in gps , employs the two - step procedure of choosing a sensible form for records of timing recordable in the memories of communicating parties , and then translating those records to graphs .
we expect different synchronization methods to produce different formats for records , which in turn will imply different special properties of the occurrence graphs to which they map .
for that reason the basic starting point is the notion of the records recordable in the memories of communicating parties .
a noteworthy property that can be defined by a network history and read from the corresponding occurrence graphs is what we call _ echo count _ , which is an integer - valued measure relevant to communications , defined to be the difference between the cycle count of a transmitting clock a at the transmission to clock b and the cycle count at a at which an echo from b is possible .
let ec@xmath10 be the echo count for transmission from a during cycle count @xmath9 to echo back from b to a. note that : 1 .
the echo count can vary along a history in which one party receives a sequence of echoes from another party .
2 . except in special cases the echo count _ not _ symmetric . for instance , clock of a can run twice as fast as clock of b , resulting in ec@xmath11 2ec@xmath12 .
so far we have concentrated on colored directed graphs as reference systems for evidence . here
we put in a word about explanations of such evidence , by looking at what happens if one chooses to introduce the additional assumptions , not required for expressing the evidence , but needed for explanations .
we start with general relativity . in order to explain synchronous communication of symbols from one clock to another in the language of general relativity we follow convention by modeling a clock as a smooth embedding @xmath13 from a real interval @xmath14 into 4-dimensional manifold @xmath15 with a smooth metric tensor field @xmath16 of lorentzian signature and time orientation , such that the tangent vector @xmath17 is everywhere timelike with respect to @xmath16 and future - pointing @xcite . to express the positive duration of phases of moves , recall the distinction between an embedding @xmath18 as a curve that is , a function from @xmath19 to @xmath15 , and the image of this curve as a 1-dimensional submanifold of @xmath15 , denoted @xmath20 .
for lack of a better word , we call such an image a _ thread_. think of a dial position attached to each point of the thread for a party , and picture the thread for a clock that takes part in synchronous communication as striped by the 12 oclock phases in which transmission and reception are allowable .
the form of a general relativistic explanation of evidence presented in the colored occurrence graphs is then a corresponding network of threads , with timelike threads for parties and lightlike threads for signals from thread to thread .
such a network of threads in a manifold with metric maps to an assertion of evidence ; however , as in the case of the trace as a map from evidence to an assertion of evidence in quantum theory : the map from a network of threads to a colored occurrence graph is _ not _ injective : for given any given colored occurrence graph displaying evidence , there is a freedom to change the metric tensor and make a corresponding change in the convention for relating physical clocks to proper clocks leaving unchanged the assertion of evidence implied by the explanation .
indeed , in applications such as gps , one needs to invoke non - gravitational forces , and these forces have to be estimated from their effects on evidence , bringing in a much larger realm for free choice of explanation .
consider a spacetime manifold and two parties @xmath1 and @xmath2 as non - intersecting , threads colored by their respective clock readings . a change in clock rate is expressed by a change in the coloring along the thread .
if the manifold is flat , it is always possible to adjust the clock rates in such a way that : 1 .
synchronous communication can take place from @xmath1 to @xmath2 and from @xmath2 to @xmath1 ; 2 .
an event of @xmath1 can be chosen freely as a transmission event , provided the clock is reset , as represented by re - coloring the thread for @xmath1 so that the event corresponds to an integral clock reading ; 3 .
given a clock as thread @xmath1 colored by its reading , along with an integer @xmath21 , there exists a clock @xmath2 allowing for synchronous communication at echo distance ec@xmath22 , independent of cycle count @xmath9 .
the same holds in a curved spacetime if the clocks are not too far apart , which is the case if for each event @xmath23 of the thread for a there be an event of the thread for b within a radar neighborhood of @xmath23 with respect to the thread for a , and _ vice versa _ @xcite . in 1967 , the 13th general conference on weights and measures specified the international system ( si ) unit of time , the second , in terms of a cesium atomic clock rather than the motion of the earth .
specifically , a second was defined as the duration of 9,192,631,770 cycles of microwave light absorbed or emitted by the hyperfine transition of cesium-133 atoms in their ground state , supposing the atoms are undisturbed by external fields .
two commercially available cesium clocks functioning well can vary in rate by about 1 part in @xmath24 , and primary cesium standards approach 1 part in @xmath25 .
nonetheless , as clock improve in their reproducibility , the size of discrepancies that matter keeps shrinking . for example
the national institutes of science and technology ( nist ) detects a rate shift between two optical clocks of @xmath26 when one clock is lifted against the earth s gravity by 33 cm and this shift is proposed as a basis for mapping the earths gravitational field@xcite .
because of size of discrepancies that matter keeps shrinking in step with improvements of precision , we continue to experience the circumstance that `` no two clocks tick alike . '' for this reason the choice of @xmath27cs or any other clock design can only be a partial specification of the clocks used to generate coordinated universal time ( utc ) .
utc actually makes use of a global system of signaling between clocks , comparing clock readings at the arrival of signals , deciding what these readings `` would be '' if the clocks were proper clocks and the general relativistic metric tensor were that assumed , and issuing ex - post corrections to readings of clocks reported by national laboratories .
a big part in this inter - clock communication is played by the global positioning system ( gps).thus in practice , a `` standard clock '' in not local to any single physical clock , but instead is a creature of a network of communicating clocks governed by a scheme of comparisons of signal arrivals that guide adjustments of clock rates , or , what is the same in its effect on recorded times , corrections .
the second depends on ( a ) a network of clock - driven communications , and ( b ) the assumption of general relativity and of some particular choices of metric tensor field within that theory . by sorting out a reference system for evidence distinct from assumptions of geometry
a choice of metric tensor field
we offer the opportunity to put the choice - making aspect of utc up on the table where it can be considered more clearly , by virtue of a reference system for evidence independent of the dynamical and indeed chaotic nature of the metric tensor field of general relativity .
it follows from perlick s work @xcite that spacetime curvature imposes a lower bound on the duration of phases of moves in a network of synchronous communication by use of signals that propagate at the speed of light .
we note however , as follows from the remarks above on paired clocks , that there is no such bound if only two clocks constitute the network .
the tightness of synchronization in a network of communicating parties is indicated by the greatest required phase duration : the less the phase duration , the tighter the synchronization .
the tightness of synchronization possible when the network operation is restricted to a subregion is apt to be greater than for the network over the region .
for this reason there can be no network that is universally tightest over all subregions . applied to coordinate - generating networks such as gps
, the implication is that for the highest precision over a limited spacetime region , the scheme of synchronization must be specially adapted to the limited region of interest . for the future
is will be interesting to study the possibility of adapting clock networks to achieve the tightest synchronization possible for particular uses in which a limited region of spacetime is at issue . in a curved spacetime such as that appropriate to explain the global positioning system , there are no killing vector fields and indeed no exact isometries linking two disjoint spacetime regions .
yet there can be occurrence graphs for a clock network that , once idiosyncratic clock readings and rate settings are forgotten , exhibit exact isomorphisms from one graph fragment to another .
one can make an analogy to isomorphisms among square tiles laid over a region of a `` potatoid , '' where variations in the thickness of the grout take up the slack , as illustrated in fig .
[ fig:9 ] .
such isomorphisms come as close as one can get to resolving the need in quantum theory to speak of repeated occurrences of the preparation of an experiment .
quantum mechanical explanations imply probability distributions for clock readings of a receiving party at the arrival of a signal within a receptive phase .
( indeed the distributions can not be confined to a receptive phase , leading to occasional logical failures that can be reduced in their disruptive effects by well - known error - correction techniques , based on redundancy , but never reduced to the vanishing point . ) but what to make of a probability measure for clock readings ?
experimentally , one compares an asserted probability with relative frequencies of clock readings at signal arrivals .
for this one has to identify many disjoint fragments of an occurrence graph as pertaining to repetitions of a single quantum `` preparation . ''
assuming a flat spacetime , this identification perhaps presents no problem .
in contrast , when one wants to work with quantum theory adjoined to a curved spacetime of general relativity , the situation becomes more interesting .
in particular in a spacetime appropriate for gps , lacking any exact isometries from one spacetime region to another , the `` uncertainty in clock readings '' picks up a component from the general - relativistic curvature , in addition to any uncertainty asserted by quantum theory .
experimenting with pendulums and balances in experiments done with our own hands made us aware of a gap between a frequency `` @xmath28 '' on the blackboard and the rate of swinging of one pendulum compared to another that we could experience on the workbench .
the key in learning to navigate between the bench and the blackboard was to see the physical device , the paired - pendulum mechanism of the flip - flop , that both recognizes and records a symbol . by burying the flip - flop under the abstract notions of spacetime and of quantum states ,
theoretical physics has lost track of the rhythms and their maintenance essential to extracting information from the bench and using that information to control experiments . in this report
we take a first step toward bringing the rhythms and their maintenance back into physics as a background against which all else in physics takes place .
this background applies regardless of the mode of explanation , and in particular regardless of whether one explains the evidence extracted from experiments by invoking quantum theory or by invoking general relativity . because quantum mechanical explanations put planck s constant @xmath29 into limits of behavior of the flip - flop , and the flip - flop works also in the acquisition of evidence to be explained by general relativity , one glimpses as a question for the future the a possible role of @xmath29 in general relativity .
the flip - flop mimics gdel coding by coding whatever symbols it recognizes in a system of numerics endowed with axioms of arithmetic . grasping that symbols expressed are necessarily physical prepares one to trace the influence of physical symbols on the statements possible in physics .
so far what has been uncovered includes the following : 1 . among other things ,
a symbol can express a pattern of other symbols , so that any description in physics , whether evidence or explanation , involves making a choice of level of detail .
because of the separation of axioms needed to symbolize evidence from axioms needed to symbolize explanations , no quantum state can be determined from evidence without reaching beyond the evidence to exercise an irreducible element of free choice , _ i.e. _ to make a _ guess_. 3 .
the communication of recognizable symbols requires a rhythm , and the rhythm requires maintenance guided not by recognitions , but by measurements idiosyncratic to the party making them , on which other parties in a communications network must rely . seeing a physical mechanism for recognizing and recording a symbol opens avenues of exploration .
questions of `` who can know what and when can they know it ? '' become colored by clock phases imposed by the pair - pendulum mechanism on which the background of symbol exchange depends . we offer a restructuring of _ clock , signal _ , and _ time _ , incorporating attention to the recognition and recording of symbols .
this structure differs from that invoked by the iau by bringing concepts into alignment with practice .
recall that einstein defined spacetime in terms of light signals exchanged among clocks @xcite .
we see spacetime coordinates as implemented by devices based on the paired - pendulum mechanism for symbol recognition ( as is the implemented turing machine ) . _ time _ amounts to relations between the ordering by one clock of a communications network to ordering by another clock , with the result that no isolated clock can `` tell time . ''
the ticking of clock a is influenced by the ticking of other clocks with which clock a communicates .
our graph pictures of evidence formalize this structure , in which records of ` digital ' symbols are made in rhythms guided by idiosyncratic ` analog ' measurements .
the concept of a physical basis for recognition invites application to biology .
we note that in an organism the propagation of signals goes very slowly relative to that in electronics , so that the single oscillator that drives the clocking throughout a digital computer likely has no biological analog ; instead , we conjecture that a nervous system , whether that of a worm or of a person , involves rhythms in which independently adjusted oscillators take part . recalling the impossibility of a `` universally tightest '' communication network in the context of general relativity , we would be interested to join other in inquiring into constraints on coordination of such rhythms .
questions abound concerning the role of quantum explanations in biology . to this topic
we contribute a suggestion that dna can be viewed as a classical code for setting up situations , for example involving photosynthesis , describable quantum mechanically .
this paper grew out of a talk given at the workshop on topological quantum information , organized by l. h. kauffman and s. j. lomonaco , jr .
, on may 16 and 17 , 2011 , held at the centro di ricerca matematica ennio de giorgi ( crm ) in pisa , italy .
we thank the crm for hosting the workshop , at which several conversations took place that stimulated the work reported here ; we also thank the crm for financial help .
we thank prof .
kauffman for recognizing some condensed occurrence diagrams as virtual knot diagrams .
99[references ] t. l. booth , _ digital networks and computer systems _ , 2nd ed .
( wiley , new york , 1978 ) , p. 223 .
a. m. turing , `` on computable numbers with an application to the entscheidungsproblem , '' proc .
london math .
soc . , series 2 , * 42 * , 230265 ( 193637 ) .
c. e. shannon , `` a mathematical theory of communication , '' the bell system technical journal , vol . 27 , pp .
379423 , 623656 , july , october , 1948 . t. j. chaney and c. e. molnar , `` anomalous behavior of synchronizer and arbiter circuits , '' ieee trans .
computers * c-22 * , no . 4 , 421422 ( 1973 ) . h. j. gray , _ digital computer engineering _
( prentice hall , englewood cliffs , nj , 1963 ) , pp . 198201 . j. h. anderson and m. g. gouda , `` a new explanation of the glitch phenomenon , '' acta informatica * 28 * , no . 4 , 297309 ( 1991 ) . f. h. madjid and j. m. myers , `` matched detectors as definers of force , '' ann .
physics * 319 * , 251273 ( 2005 ) .
b. cheney and r. savara , `` metastability in scfl , '' in _ technical digest of the 17th annual ieee gallium arsenide integrated circuits symposium , 1995 _ , ( ieee , san diego , ca , 1995 ) , pp .
. h. meyr and g. ascheid , _ synchronization in digital communications_(wiley , new york , 1990 ) .
h. meyr , m. moeneclaey , and s. a. fechtel , _ digital communication receivers : synchronization , channel estimation , and signal processing_(wiley , new york , 1998 ) .
j. m. myers and f. h. madjid , `` a proof that measured data and equations of quantum mechanics can be linked only by guesswork , '' in s. j. lomonaco , jr . and h. e. brandt , eds . , _ quantum computation and information _ , contemporary mathematics series , vol . 305 ( american mathematical society , providence , ri , 2002 ) , pp . 221244 .
j. m. myers and f. h. madjid , `` ambiguity in quantum - theoretical descriptions of experiments , '' in k. mahdavi and d. koslover , eds .
, _ advances in quantum computation _ , contemporary mathematics series , vol . 482 ( american mathematical society , providence , ri , 2009 ) , pp . 107123
. m. soffel , s. a. klioner , g. petit , p. wolf , s. m. kopeikin , p. bretagnon , v. a. brumberg , n. capitaine , t. damour , t. fukushima , b. guinot , t .- y .
huang , l. lindegren , c. ma , k. nordtvedt , j. c. ries , p. k. seidelmann , d. vokrouhlicky , c. m. will , and c. xu , `` the iau 2000 resolutions for astrometry , celestial mechanics , and metrology in the relativistic framework : explanatory supplement , '' the astronomical journal * 126 * , 26872706 ( 2003 ) .
a. w. holt , `` introduction to occurrence systems , '' in e. l. jacks , ed . , _ associative information techniques _ , proceedings of the symposium held at general motors research laboratories , 1968 ( american elsevier , new york , 1971 ) , pp . 175203
. f. commoner , a. w. holt , s. even , and a. pnueli , `` marked directed graphs , '' journal of computer and system sciences * 5*(5 ) , 511523 ( 1971 ) . j. l. peterson , _ petri net theory and the modeling of systems _
( prentice - hall , englewood cliffs , nj , 1981 ) .
a. w. holt , role / activity models and petri nets , " second semi - annual technical report , project on development of theoretical foundations for description and analysis of discrete information systems , ntis ad - a008 385 ( massachusetts computers associates , wakefield , ma , 1975 ) . l. kauffman , `` introduction to virtual knot theory , '' arxiv:1101.0665v1 ( 2011 ) . v. perlick , `` on the radar method in general - relativistic spacetimes , '' arxiv:0708.0170v1 ( 2007 ) . c. w. chou , d. b. hume , t. rosenband , and d. j. wineland , `` optical clocks and relativity , '' science * 329 * , 16301633 ( 2010 ) .
a. einstein , _ ber die spezielle und die allgemeine relativittstheorie _ ,
24th ed .
( springer , berlin , 2008 ) ; english trans . by r. w. lawson as _ relativity , the special and general theory : a popular exposition _ ( bonanza books , new york , 1961 ) | preoccupied with measurement , physics has neglected the need , before anything can be measured , to _ recognize _ what it is that is to be measured .
the recognition of symbols employs a known physical mechanism . the elemental mechanism
a damped inverted pendulum joined by a driven adjustable pendulum ( in effect a clock)both recognizes a binary distinction and records a single bit . referred to by engineers as a `` clocked flip - flop , '' this paired - pendulum mechanism pervades scientific investigation .
it shapes evidence by imposing discrete phases of allowable _ leeway _ in clock readings ; and it generates a mathematical form of evidence that neither assumes a geometry nor assumes quantum states , and so separates _ statements of evidence _ from further assumptions required to _ explain _ that evidence , whether the explanations are made in quantum terms or in terms of general relativity .
cleansed of unnecessary assumptions , these expressions of evidence form a platform on which to consider the working together of general relativity and quantum theory as explanatory language for evidence from clock networks , such as the global positioning system .
quantum theory puts planck s constant into explanations of the required timing leeway , while explanations of leeway also draw on the theory of general relativity , prompting the question : does planck s constant in the timing leeway put the long known tension between quantum theory and general relativity in a new light ? |
a fundamental question in the theory of evolution has been how cooperation can emerge between selfish members @xcite .
another question is why evolution takes place in terms of intermittent bursts of activities , which are the characteristics of dynamical systems in a ` critical ' state @xcite . here
, we propose a generalized bak - sneppen ( bs ) model @xcite , which may solve the above two puzzles simultaneously .
we take an approach of evolutionary game theory and use the prisoner s dilemma ( pd ) games to mimic the interactions among members .
each member is identified by its stochastic strategy , specified by its ( history independent ) cooperation probability ( cp ) . here
, a ` member ' can represent an individual in a species , an agent in an economical system or a species in an ecological system .
the fitness of a member is given by the payoffs of the games with its neighbors .
we then apply bs dynamics and replace the least fit member and its neighbors by new members with random cps .
the neighbors of the non - cooperator are likely to vanish due to its low payoff , but the non - cooperator itself can also be removed through the bs mechanism . as the non - cooperators disappear , the overall cp increases , and a new comer ( with a random cp ) will have a lower cp than the increased average .
therefore , the new comer tends to cause its neighbor to be the least fit , and the replacement activity likely occurs at or near the new comer s site .
this invokes the spatio - temporal correlation between the least fit sites and can explain why replacements are episodic as well as how cooperation emerges .
evolutionary game theory has been one of the most powerful tools in studying the dynamics of evolution @xcite .
however , a simple straightforward application of game theory can not explain the strong cooperation between selfish " replicators observed in nature and society .
for the evolution to construct a new , upper level of organization , cooperation amongst the majority of the population is needed .
however , the game theoretical description of interactions between members usually leads to defections as evolutionarily stable strategies .
natural selection , which has been a fundamental principle of evolution , prefers the species that beat off the others and oppose cooperation .
there have been numerous studies looking for natural mechanisms for the evolution of cooperation among competitive members @xcite .
recently , nowak presented a state - of - art review on the evolution of cooperation and discussed five known mechanisms : kin selection , direct reciprocity , indirect reciprocity , network reciprocity , and group selection @xcite .
extensive studies provide the exact conditions for the emergence of cooperation for each of the five mechanisms .
however , such conditions do not seem to be general enough to explain the cooperative phenomena observed everywhere .
for example , for network reciprocity , the benefit - to - cost ratio of a cooperative behavior should be larger than the average degree @xcite , but this seems to be a rather strong assumption because the degrees are quite large in most cases in real population structures .
also , there have been a great deal of studies on self - organized criticality in game theory @xcite , but their dynamics leading to the critical states are not directly connected to the emergence of cooperation . here , we consider an evolutionary game on networks and show that cooperation can emerge when the benefit to cost ratio is larger than just 1 if we use the bs process .
when cooperators interact with defectors , they tend to disappear , giving rise to an assortment of cooperators @xcite .
furthermore , this behavior emerges in the long run even with a small `` chain - death '' rate , @xmath0 , where the number of neighbors that get replaced is less than one . for a uniform or random arrangement of cooperators and defectors , more cooperators than defectors disappear for small @xmath0 , but in the long run
, the bs process builds a self - organized structure so that the number of cooperators in the population increases .
an influential model aimed to mimic the interactions between competitive members in a population is the pd game .
it is one of the matrix games between two players who have two possible decisions , cooperation ( @xmath1 ) or defection ( @xmath2 ) .
we consider a case in which the payoffs are calculated by the cost @xmath3 and the benefit @xmath4 of a cooperative behavior . if one player defects while the other cooperates , the defector receives benefit @xmath4 without any cost whereas the cooperator pay cost @xmath3 and its payoff becomes @xmath5
. for mutual cooperation , both get benefit @xmath4 , but pay cost @xmath3 , and their payoffs become @xmath6 while the payoffs for mutual defection are 0 .
when we add @xmath3 to all elements so that payoff can be directly interpreted as ( non - negative ) fitness , the payoff matrix becomes @xmath7 where we set @xmath8 without loss of generality . with conventional competition processes , the matrix game shown above
does not , in general , predict the evolution of cooperation .
the birth - death process always predicts an evolution of defection .
cooperation can emerge for death - birth or imitation processes in a structured population , but only with a ( unrealistic ) large value of the benefit - to - cost ratio @xmath4 for real populations @xcite . here , we consider the pd game interaction , but introduce the bs mechanism @xcite as the competition process , and assume that the least fit member and its neighbors are prone to disappear .
each member is characterized by its strategy that determines when to choose the ` decisions ' @xmath1 or @xmath2 .
we consider the history - independent stochastic strategies , and the phenotype of a member , say the @xmath9th member , is represented by its cp @xmath10 .
the history independent pure ( deterministic ) strategies , the `` always @xmath1 '' and the `` always @xmath2 '' , correspond to the limits of @xmath11 and @xmath12 , respectively .
the fitness of a member is given by the sum of payoffs from its neighbors , and the member dies out if its total payoff is the minimum .
the died - out site is occupied by a new member with a new cp , which is drawn randomly from 0 to 1 .
neighbors of the least fit site may also be harmed in the process of establishing the steady interaction with the new comer .
hence , we replace the neighbors of the least fit site by new members with the `` chain - death '' probability @xmath13 . and ( b ) the rf @xmath14 for @xmath15 $ ] with @xmath16 and @xmath17 .
they are represented by colors , red for 0 and green for 1 , as indicated by the top panel .
the black dots in ( b ) represent the least fit sites .
, width=264 ]
we study the strategy evolution of a simple structured population from the initial state of random strategies .
initially , members in the population have cooperation probabilities that are drawn randomly from the uniform distribution of the interval [ 0 1 ] .
they play pd games with their nearest neighbors .
we assume that each member plays sufficiently many games prior to the reproduction process and use the payoff expectation value as its fitness .
the least fit member with the minimum payoff expectation is replaced by a new member with a new random cp .
in addition to the least fit member , the neighbors of the least fit member are also replaced by new members with the probability @xmath0 .
then , we recalculate the payoff expectations , and replacements occur at the new least fit member and its neighbors .
we continue these processes until the system reaches a steady state and calculate the statistical properties of the population , such as the mean cooperation probability , fitness distribution , avalanche size ( defined later ) distribution and etc .
for simplicity , we present our model and results in a one - dimensional ( 1d ) structure , but our main results hold in other population structures . initially ( @xmath18 )
, we assign a random cp , @xmath19 , to the site @xmath9 for @xmath20 . then , we calculate the payoff expectation , @xmath21 , at time @xmath18 , f_i(0 ) = b + 2 , [ e.fi ] of site @xmath9 with a periodic boundary condition and find the minimum payoff site , @xmath22 . except this minimum site , @xmath22 and its neighbors , @xmath23 , the cps
are not changed at @xmath24 , so we set @xmath25 unless @xmath26 or @xmath23 .
the cp at the @xmath22 site , @xmath27 , is given by a new random number between 0 and 1 .
for its neighbor sites , @xmath28 is given by a new independent random number with the probability @xmath0 , but remains as @xmath29 with the probability @xmath30 .
now , we recalculate the payoff @xmath31 of eq .
( [ e.fi ] ) with @xmath32 instead of @xmath33 .
we find the new minimum payoff site , @xmath34 , of @xmath24 and apply the same replacement dynamics to get @xmath35 configurations and so on .
figure 1 shows typical real space configurations of cp , @xmath10 , and the reduced fitness ( rf ) , @xmath36 $ ] .
we show the configurations for initial 6000 time steps of a @xmath17 system with @xmath37 and @xmath16 .
both the cp and the rf are represented by colors , 0 by red and 1 by green .
the least fit sites ( black sites in ( b ) ) and their two neighbors are where the replacement activity occurs .
comparison between the configurations in ( a ) and their equivalents in ( b ) reveals that the least fit sites are located where their neighbors are less cooperative [ relatively red in ( a ) ] . the disappearance of the `` red '' neighbors beside the least fit site by the bs - mechanism shifts the overall system to green ( more cooperative ) with time . , for ( a , c ) @xmath38 and ( b , d ) @xmath39 for systems with @xmath40 , 64 , 128 , and 256 . in ( a ) and ( b )
, the overall behaviors of mcp are shown while the initial transient characteristics are shown in ( c ) and ( d ) . for @xmath38 , mcp decreases first and then increases while it monotonically increases for @xmath16 . , width=302 ] for a quantitative analysis , we measure the mean cp ( mcp ) , @xmath41 , of the populations and show the results in fig . 2 . here , @xmath42 represents the ensemble average over many different realizations of random initial configurations .
note that the mcp also represents the overall fitness @xmath43 of the population because it is linearly related to mcp : f(t ) = = 2 + 2 ( b-1 ) c(t ) . in fig .
[ f.2 ] , the mcps for four different system sizes , @xmath40 , 64 , 126 , and 256 , are shown for two different values of @xmath0 , @xmath44 , and 1 .
we use @xmath37 for all figures in this paper , and all data are obtained from numerical simulations .
because we have assigned a random cp initially , the mcp starts from 0.5 at @xmath18 . for @xmath38 ,
the mcp decreases at the beginning and then increases to the steady values while it monotonically increases from the beginning for @xmath16 , as shown in figs .
2(c ) and ( d ) .
note that we have two different elements in mcp changes .
replacement of the least fit member ( which is likely to have a high cp ) tends to cause the mcp to decrease while the replacement of its neighbors ( which probably have low cps ) likely results in an increased mcp .
the competition between these two elements governs the early dynamics of the mcp .
it can decrease initially when @xmath45 , where @xmath46 is the number of neighbors .
for a sufficiently large system , there would be a site , @xmath47 , whose cp , @xmath48 , is arbitrarily close to one while those of its neighbors , @xmath49 , are almost zero .
hence , the expectation of mcp changes , @xmath50 would be @xmath51}$ ] and becomes negative for @xmath52 at the beginning . however , as time proceeds , the cp values develop spatio - temporal correlations , and they govern the long - time dynamics . initially , the isolated high - cp cooperators are likely to be the least fit member , and they are removed as time proceeds .
then , surviving cooperators remain in the groups , and , thus , have high fitness .
now , low - cp defectors can be the least fit member , especially when they are next to a very low - cp member .
the replacement of these low - cp member by new members with random cps causes the mcp to increase .
therefore , at a later time , the mcp easily becomes larger than the initial 0.5 even for @xmath53 .
now , a new comer with a random cp will have a lower cp than the increased average of mcp .
this in turn causes the least fit site to be likely located next to the new comer s site , resulting in avalanches of replacement activities .
, their neighbors , @xmath54 , and members that are replaced , @xmath55 , are shown together with mcp , @xmath1 , for ( a , c ) @xmath38 , and ( b , d ) @xmath39 . in ( a ) and ( b )
, the initial transient behaviors are shown while overall behaviors are shown in ( c ) and ( d ) .
the system size @xmath56 is used for all cases .
, width=302 ]
we start from the population with random strategies .
hence , there is no correlation between the cps initially , and we may understand the initial dynamics through the mean - field calculation .
we first define the mean cp of the replacement sites ( before the replacement ) , c_rep = ( c_min + 2c_nei ) , where mean - field dynamics can be easily analyzed . here , @xmath57 is the average of the cps for the least fit members , and @xmath54 is that for the neighbors of the least fit members . on average , cps of @xmath58 sites
are updated each time . since the average of the newly assigned random cooperation rate is 0.5 , @xmath55 satisfies , = = 0.5 - c_rep .
[ e.crep ] we measure @xmath55 and present them in fig .
[ f.crep.t ] , together with the cps of the least fit members , @xmath57 , that of the replaced members , @xmath55 , and the mcp , @xmath1 , for @xmath38 and @xmath39 .
the @xmath55 curves are , indeed , well described by eq .
( [ e.crep ] ) . if we represent the numerical solutions of eq .
( [ e.crep ] ) in the figure , they can not be distinguished from the @xmath55 curves from the simulations because they are almost identical . from fig .
[ f.crep.t ] , we also see that @xmath55 enters its steady value in a relatively short period of time compared to @xmath1 and rapidly converges to its steady - state value of 0.5 . for @xmath38 , the initial
@xmath55 is more than half and hence decreases to the steady value of 0.5 while it increases from the value below 0.5 for @xmath59 . for a sufficiently large system
, the initial value of @xmath57 would be 1 while @xmath54 is 0 .
hence , the initial value of @xmath55 would be @xmath60 , which is more than 0.5 for @xmath61 . in this transient time of @xmath55 , the dynamics of mcp , @xmath1 , would be mainly determined by the dynamics of @xmath55 .
therefore , @xmath1 initially decreases for @xmath62 as does @xmath55 .
however , after @xmath55 reaches a steady value , the correlation of the replacement sites mainly governs the dynamics , and @xmath1 begins to increase .
let @xmath47 be the least fit member at time @xmath63 ; then , at time @xmath64 , @xmath65 is always updated , and @xmath66 are updated with probability @xmath0 . after replacement ,
if the sum of the cps at these three sites , @xmath67 ( at the time @xmath64 ) , is small , at least one of @xmath68 , @xmath47 or @xmath69 sites , is likely to have small fitness .
therefore , they will be easily replaced in a relatively short time . in other words , a new born member with small
@xmath70 has a short lifetime and contributes less to the @xmath1 than those with large @xmath70 .
this mechanism makes @xmath1 increase up to ( almost ) @xmath57 , and hence , the system becomes cooperative overall .
thus , according to our model , the emergence of cooperation is intrinsically related to the dynamics leading to self - organized criticality ( soc ) .
in the steady states for five different system sizes of @xmath71 , 32 , 64 , 128 and 256 with ( a ) @xmath38 and ( b ) @xmath39 . the system size dependence of the effective lower and upper thresholds @xmath72 and @xmath73 ( defined in the text ) are shown in ( c ) .
legends of ( a ) are also applied to ( b ) . ]
we now show that our model , in fact , drives the population into a soc state as in the original bs model .
we measure the distributions of avalanche sizes and distances between successive least fit sites in the steady states and show that they follow power - law distributions .
following bak and sneppen @xcite , we would like to define the size of an avalanche as the number of subsequent replacements at the least fit sites below the lower threshold @xmath74 in its fitness value .
the fitness distributions @xmath75 share some characteristics of the bs model @xcite although their overall shapes are quite different .
a crucial similarity is that the fitness distribution @xmath75 in the steady state becomes zero for fitness @xmath76 smaller than a lower threshold @xmath77 as the system size goes to infinity .
the fitness distributions in the steady states for five different system sizes are shown in figs .
[ f.4](a ) and ( b ) for @xmath38 and @xmath39 .
as the system sizes increase , the peak positions of the fitness distribution move to the right to high values , and the peak widths become narrow . to estimate the threshold values @xmath74 and @xmath78
, we define the effective lower [ upper ] threshold @xmath79 [ @xmath80 as the @xmath76 value below [ above ] which the integrated distribution is 5 percent .
we plot them against @xmath81 in fig .
[ f.4](c ) for two different chain - death rates , @xmath38 and @xmath39 .
there are no noticeable differences in the thermodynamic values for the two @xmath0 values . using linear fitting
, we get rough estimates of the threshold values , @xmath82 and @xmath83 , for both @xmath0 values . of avalanche sizes @xmath84 .
distributions with three different values of @xmath74 , @xmath85 , @xmath86 , and @xmath87 are measured in systems of @xmath88 in their steady states .
data with @xmath86 show a most persistent straight line in the log - log scale fit , indicating the lower threshold @xmath86 for the @xmath88 system with @xmath16 .
the black line is the least - squares fit of the data for @xmath86 and is given in a form of @xmath89 with @xmath90 .
( b ) a distribution @xmath91 of the distances between successive minimum fitness sites in the steady states for the system of @xmath92 with @xmath16 . the black line is the least squares fit of the data in the form of @xmath93 with @xmath94 . ] for the avalanche size distribution @xmath95 , we need a more precise value of @xmath74 .
we measure @xmath95 with several different values of @xmath74 around the estimated value .
if the system is really in a soc state , we expect the avalanche size distribution @xmath95 to show a power - law distribution , for the exact value of @xmath74 for the given system .
figure [ f.5](a ) shows the distribution of avalanche sizes in a system of size @xmath88 .
we plot @xmath95 against @xmath84 on a log - log scale with three different values of @xmath74 around the value estimated from fig .
[ f.4](c ) to pinpoint the threshold @xmath74 . for @xmath39 shown in fig .
[ f.5](a ) , the avalanche size distribution is well fit by a power - law with @xmath86 .
it remains as a line in the log - log plot up to an avalanche size about 20000 , indicating power - law distributions @xmath89 .
the exponent obtained from a least - square fit of the form @xmath96 is @xmath97 .
this value is consistent with the known exponent of the 1d bs model @xcite .
the power law indicates that the evolution occurs in a dynamical criticality @xcite .
we measure the avalanche distributions for other @xmath0 and @xmath4 and found the critical exponent @xmath98 to be independent of the benefit - to - cost ratio @xmath4 or the chain - death probability @xmath0 .
we also measured the distance distribution between successive least fit sites . denoting the distance between successive minimum fitness sites by @xmath99 , we plot @xmath100 in fig . [ f.5](b ) .
the distance distribution is measured in the steady states for the system of @xmath92 with @xmath16 . when the distribution @xmath100 is plotted against @xmath99 on a log - log scale , it also becomes a line , indicating power - law distributions @xmath101 with the slop @xmath94 .
this exponent is also consistent with the known exponent of the 1d bs model @xcite .
it is notable that our model belongs to the same universality class as the bs model in spite of the complexity in computing the fitness of members and the non - trivial dynamics of the population - fitness changes .
we have considered the bs mechanism as a reproduction process with fitness given by a pd game payoff on a network structure .
our observation may have more natural implication in economical systems because the bs process with chain bankruptcy is a more feasible scenario .
it might be worthwhile analyzing weekly or monthly bankruptcy data and see if they follow a power - law distribution as our study suggests .
we have simulated our model with other values of the benefit - to - cost ratio @xmath4 and see that cooperation emerges in a wide range of chain - death rates @xmath0 , as long as @xmath4 is larger than 1 .
in contrast to a common belief , cooperation can emerge even with parameters that a population with random strategies decreases cooperation .
this is possible because the bs mechanism builds dynamical correlations that suppress the long - term survival of non - cooperators even in the region where mean - field calculation predicts a decrease in cooperators .
the same dynamical correlation leads to soc in replacement activities with the same exponents as the original bs model .
the strategy space presented here is rather small .
mixed but only history independent strategies are considered on a very simple population structure , a 1d lattice . however , we speculate that our main results , the emergence of cooperation and soc , are robust under variations in the population structure or the strategy space extension .
in fact , the preliminary results with the extended strategy space show that the emergence of cooperation appears more easily and rapidly when the reactive strategies are included .
this work was supported by the national research foundation of korea grant funded by the korean government(mest ) ( nrf-2010 - 0022474 ) .
j. would like to thank kias for the support during the visit . | cooperation and self - organized criticality are two main keywords in current studies of evolution .
we propose a generalized bak - sneppen model and provide a natural mechanism which accounts for both phenomena simultaneously .
we use the prisoner s dilemma games to mimic the interactions among the members in the population .
each member is identified by its cooperation probability , and its fitness is given by the payoffs from neighbors .
the least fit member with the minimum payoff is replaced by a new member with a random cooperation probability .
when the neighbors of the least fit one are also replaced with a non - zero probability , a strong cooperation emerges .
the bak - sneppen process builds a self - organized structure so that the cooperation can emerge even in the parameter region where a uniform or random population decreases the number of cooperators .
the emergence of cooperation is due to the same dynamical correlation that leads to self - organized criticality in replacement activities . |
millions of people edit wikipedia pages , however , in average we find that only @xmath0 contribute to @xmath1 of their content . such heterogeneous level of activity is reminiscent of the well - known and widely applicable law postulated by pareto @xcite , which states that @xmath1 of the effects are induced by @xmath2 of the causes .
the example of wikipedia users reported here highlights how heterogeneous the activity of their users are , with both , activity as well as degree following a power - law distribution.indeed , heavy - tailed distributions following a power - law have been observed in variety of social systems ever since pareto reported his observation of the extreme inequality of wealth distribution in italy back in 1896 @xcite . in recent years , due to ubiquitous computerization , networking and obsessive data collection , reports of heavy - tailed distributions have almost become a routine @xcite .
following simple distributions such as those of wealth , and income @xcite , certain structural properties of social systems were also found to be heavy - tailed distributed .
more specifically , distribution of the number of ties of a person ( degree ) has been shown to fall in this group for vast and still growing number of social networks @xcite .
power - law degree distributions , called scale - free @xcite , represent one of the three general properties of social networks ( short distances and high clustering being the other two @xcite ) .
a power - law degree distribution is not only the least intuitive and surprising property , but also is the most well - studied and debated feature of networks since extensively found in the late 90s @xcite . immediately following the empirical measurements ,
a number of plausible models aiming at explaining the emergence of these distributions have been proposed @xcite .
many models reproduce heterogeneous connectivity by amplifying small differences in connectivity frequently stochastically emerging using some kind of multiplicative process or `` preferential attachment '' @xcite .
other models propose different optimization strategies leading to scale - free @xcite .
a common attribute of all these models is that fat - tailed distributions emerge out of some kind of interaction between the basic system s elements .
in fact , the question is not whether there exists a mechanism that could produce scale - free networks similar to the ones observed , but which of the many mechanisms suggested are more likely to actually play a significant role in each network formation .
the data presented here suggests that there is a different underlying cause for heavy - tailed degree distributions which does not involve interactions between people .
we investigate distinct social networks focusing on the relationship between users activity and degree , specifically , the number of posts , messages , or actions of a user , i.e. _ activity _ and the number of user establishing a link with her / him , i.e. the incoming degree , or _ degree _ , for short . both , degree @xmath3 in the social network , and the activity @xmath4 of a user , exhibit power - law distributions @xmath5 , and @xmath6 , where @xmath7 and @xmath8 are the scale - free degree and activity exponents , respectively . positively skewed distributions of human activity
were recently reported in @xcite and we extend this result here for a number of datasets .
more importantly , in all instances we find that activity causally determine degree of the same user , suggesting that the broad distribution of one , could result from the broad distribution of the other .
it is important to note that the studied actions are not likely to be driven by interaction with other people .
activity and degree , as measured here , are taken from two different networks developed by the same pool of users , and so there is no reason to expect that they should depend on each other in some trivial fashion .
surprisingly , however , the number of potential followers of a user ( degree distribution ) appears to be entirely random except for its mean value , which is tightly controlled by the volume of activity of that user .
our observations convincingly point at the intrinsic activity of people as the driving force behind the evolution of the examined social systems and particularly the heterogeneity in user connectivity .
the observed degree distribution in social systems may merely be a manifestation of the similarly wide distribution of human activity related to the system construction .
these wide distributions in social collaborative networks can not be explained by interactive model since the observed actions are not likely to be caused by actions of other people .
we have analyzed activity of individuals over time collaboratively working on construction of extensive electronic data sets : wikipedia in four different languages ( _ http://www.wikipedia.org_ ) , and a collaborative news - sharing web - site ( _ http://www.news2.ru_ ) .
these datasets represent various domains of human activity and contain records of a vast number of individual user contributions to the collaboratively generated content ( see method ) . for each person , we analyze two properties defined in two independent layers : activity and degree . for instance , in wikipedia , the activity performed by users includes posting of new material and discussions about them .
this is the activity layer .
simultaneously , by tracing users contributing to other users personal or talk pages , we recover the underlying network of wikipedia contributors personal communication or social network .
the resulting network reliably represents actual interactions of wikipedia users @xcite and thus defines the social network layer . the number of incoming connections , i.e. others reaching out to the user in this network represents the degree . in principle , activity and degree as defined here are unrelated .
similarly , news2.ru posses the same two - layer structure of activity and degree ( see method ) .
we start by analyzing the distributions of various types of activities performed by users in these systems .
very few of the most active users perform the vast majority of work so that the activity levels frequently span five orders of magnitude ( fig .
[ fig1]a , b ) .
for instance , when analyzing the activity to a given wikipedia page , only 5% of users contribute @xmath1 of the edits ( fig .
[ fig5 ] in method ) .
this surprising result is similar to the @xmath9 rule postulated by pareto @xcite to describe the unequal distribution of wealth .
indeed , a power - law faithfully characterizes the activity distributions in fig .
the exponent of the activity distribution for spanish language wikipedia is @xmath10 ( fig . [ fig1]a ) , while the activity distribution for voting in stories in news2.ru is @xmath11 ( fig .
[ fig1]b , detailed fitting procedure in method @xcite ) the activity distributions in fig .
[ fig1]a represent the number of users as a function of the number of wikipedia edits in four languages .
interestingly , different populations performing similar activity in separate instances of similarly - built social systems exhibit identical activity distributions .
figure [ fig1]b shows several different activities performed by the same population of users at the social news aggregator news2.ru .
these activities differ in their complexity .
we consider submission of posts to be the most difficult and time consuming of the four activity types because it typically requires the user to locate the content on - line , evaluate its quality and publish at the news2.ru web site by filling a form with multiple fields .
considering the task complexity , writing comments is arguably easier task than posting .
there are on average nearly three comments per every published post .
these two content - generating tasks are followed by ranking of posts and comments .
the differences in the underlying complexity of the task seem to explain the difference in the range and slope of the observed distributions plotted in fig .
[ fig1]b .
we further observe the social networks emerging in each of these systems .
these networks serve different functions . in wikipedia
they arise due to the direct interaction required to coordinate common tasks .
in particular , we derive social networks from the record of edits of personal user pages by other users - a common way of personal communication in wikipedia ( the web site rules forbid activity - related confidential communication between its editors ) . in
news2.ru the social network emerges through declaration of personal attitudes - a user may indicate that he / she likes , dislikes or is neutral to any other user .
another social network arises from a set of explicit ( directed ) declarations of friendship between news2.ru users .
figure [ fig1 ] c and d present the degree distributions in these networks .
broad distributions are measured and present in each system , suggesting a scale - free behavior in their degree distribution .
the exponent of the degree distribution for spanish wikipedia is @xmath12 ( fig . [ fig1]c ) , and for the degree distribution in news2.ru is @xmath13 ( fig .
[ fig1]d ) .
the present data suggest a simple explanation of the origin of degree distributions .
we first observe that the number of incoming links aggregated by a person in all these social networks is highly correlated to the individual s activity .
the correlation between the degree and the activity measurements is presented in table [ table1 ] .
it is measured here as the correlation of the log - values to capture the gross relationship of these two variables across different orders of magnitude .
more importantly , the dependence analysis below suggests that the broad distribution of activity is the driving force of scale - free degree as will be discussed next .
it is important to emphasize that in order to avoid direct and rather obvious correlation between different aspects of activity of the same person , we test the correlation of individual s activity to her degree determined by actions of his / her followers rather than his / her own .
it is possible that these actions are driven by reciprocity , i.e. , a person is simultaneously active in the community and in constructing her social network inspiring others to link back to her . to determine the precise nature of the @xmath14 relationship , we analyze the joint distribution of degree and activity , @xmath15 ( fig .
[ fig2]a ) .
we find that the mean degree @xmath16 for a given level of activity follows a smooth monotonic function of @xmath4 ( fig .
[ fig2]b ) , whereas the opposite is not true , i.e. , the mean activity @xmath17 does not seem to be tightly determined by degree ( fig .
[ fig2]c ) . a similarly tight relationship exists for the standard deviation of the degree distribution @xmath18 for specific values of the activity ( fig .
[ fig2]d ) , but , again , the reverse is not true ( fig .
[ fig2]e ) .
the conditional mean and standard deviation of degree ( conditioned on activity ) show a tight relationship with approximately unit slope @xmath19 ( fig .
[ fig2]f ) .
however , the @xmath20 , @xmath17 values conditioned on degree are more variable ( fig . [ fig2]g ) . based on these observations we hypothesize that the conditional degree distribution @xmath21 may be scale invariant with scale @xmath16 entirely determined by activity : @xmath22 . here
, this functional dependence of scale can be estimated as the mean activity for a given @xmath4 : @xmath23 .
indeed , we observe that the conditional degree distribution appears to follow a geometric distribution for all @xmath16 : @xmath24 this theoretical distribution provides a remarkably accurate fit to the first two sample moments of degree for a given level of activity as shown in fig .
we plot the standard deviation @xmath18 versus mean degree @xmath16 for given activity for four wikipedia databases .
the curves follow a smooth , monotonically increasing functional form which is almost identical for all datasets ( as one would expect for activity conditioning degree ) .
when the analysis is repeated for activity conditioned on degree the variables do not appear to follow a tight relationship .
the tight relationship between @xmath18 versus @xmath16 conditioned on activity follows asymptotically a straight line with unit slope , which follows exactly the geometric distribution eq .
( [ geo ] ) . in fig .
[ fig3 ] , we compare the data to the analytic relationship between mean and standard deviation for geometric distribution eq .
( [ geo ] ) : @xmath25 and @xmath26 , where @xmath27 is the parameter of geometric distribution .
the data fit this theoretical curve surprisingly well for the four displayed languages of wikipedia ( @xmath28 in average ) .
the previous findings can be understood with the following hypothesis h1 : @xmath29 , activity deterministically affects the mean degree , but degree is otherwise random ( fig .
[ fig4]a ) .
note that for positive discrete variables like the degree with a given mean ,
the highest entropy or least informative and most random distribution is achieved by the geometric distribution as we find above @xcite .
the geometric distribution is analogous to exponential distribution in statistical mechanics , which maximizes entropy for continuum variables with fix mean .
we also tested the inverse hypothesis h2 : @xmath30 , degree deterministically affects mean activity , @xmath31 , and activity is otherwise random . the goodness - of - fit of these two analytic models to histograms of h1 : activity @xmath32 degree or h2 : degree @xmath32 activity was measured with the @xmath33-square statistics averaged over activity or degree respectively .
the likelihood that the observed distributions match h1 or h2 was assessed using surrogate data generated with monte - carlo sampling to estimate the chance occurrence of these averaged @xmath33-square values .
the results for the spanish language wikipedia data indicate that we can not dismiss the correctness of h1 ( fig .
[ fig4]b ) with a confidence of higher than 95% ( @xmath34 ) but that h2 can be soundly dismissed ( the chance of the corresponding @xmath33-square value occurring at random is @xmath35 ) .
the same is true for all other datasets ( see table [ table1 ] ) . in all datasets
the likelihood of h1 is several orders of magnitudes larger than h2 and thus we accept model h1 , which states that activity determines degree . given the explicit model of a geometric distribution for @xmath36 of hypothesis h1 , and the observed power - law distribution for activity , @xmath37 , one can explicitly derive the expected degree distribution.the conditional degree distribution closely matches a geometric distribution ( fig .
[ fig3 ] ) . for large mean values , say @xmath38 , it can be very well approximated by its continuous equivalent , the exponential distribution i.e. @xmath39 .
therefore : @xmath40 thus the exponent is predicted to be @xmath41 where @xmath42 defines @xmath43 for large @xmath4 as shown in figure [ fig2]b .
the observed exponents @xmath7 closely follow these predicted exponents for all datasets ( table [ table1 ] ) .
the causal inference argument provided here is borrowed from ideas recently developed in causal inference @xcite . there , a deterministic functional dependence of cause on mean effect is hypothesized and deviations from this mean effect are assumed to have fixed standard deviation but to be otherwise random . with two variables for which one wishes to establish causal direction ,
the model is evaluated in both directions and the more likely one is postulated to indicate the correct causal dependence , as we have done here .
this approach has been demonstrated to give the correct causal dependence for a large number of known causal relationships @xcite , and theoretical results indicate that there is only an exceedingly small class of functional relationships and distributions for which this procedure would give the incorrect answer .
such an identifiability proof does not yet exist for the present case where the standard deviation is not constant .
nevertheless , our explicit model of a deterministic effect of human activity on the success of establishing social links is the simplest possible explanation for the data available to us . for a different dataset
a different probabilistic model may be better suited .
the individual activity of people deterministically affects the mean success at establishing links in a social network , and the specific degree of a given user is otherwise random following a maximum entropy attachment ( mea ) model .
the mea model is exemplified in fig .
[ fig4]a and consists of the following steps : introduce a node @xmath44 with @xmath45 links , where @xmath45 is drawn from a probability given by the activity of the node .
the activity has an intrinsic power - law distribution .
then , link the @xmath45 links at random following maximum entropy principle with the concomitant geometric distribution @xmath46 .
this mechanism contrasts with the preferential attachment mechanism @xcite where each link attaches to a node with a probability proportional to the number of links of that node .
a possible mechanism by which a geometric distribution could arise is based on the notion of `` success '' . in this model ,
the activity of users aims to achieve a specific outcome ( a wikipedia project ) , and each new incoming link can aid in achieving this desired outcome ; once the goal is achieved the user stops collecting links .
the probability of the desired event in this model is @xmath47 .
hence , those users working so very hard may have an exceedingly unlikely event they are aiming for .
but eventually , they too will succeed , and will turn their attention away from the on - line social network .
the present data indicates that degree distribution is maximally random except for what can be determined solely from the volume of a user s activity .
does this mean that the precise content of a user s actions ( the meaning and quality of the edits in wikipedia , messages , etc ) is immaterial in determining his / her success in establishing relationships ?
one can only hope that small deviations from this maximum entropy attachment model will become more pronounced with increasing data - set sizes , which can then point us to the benefits of well thought out and carefully executed actions , specially in specialized large - scale collaborative projects like wikipedia .
whether the dynamics of preferential attachment is consistent with the maximum entropy distribution of degree remains to be established .
what is certain is that distributions of levels of activities in all tested populations are heavily heavy - tailed indicating highly varying level of involvement of users in collaborative efforts .
we showed here that this fact alone is sufficient to produce the heavy - tailed distribution of degree observed throughout social networks .
therefore , previous interactive models may not be necessary .
the present result shifts the burden of proof to explaining the origin to the incredible diversity in human effort observed here spanning five orders of magnitude .
the number of actions contained in the datasets range from hundreds of thousands to hundreds of millions of user actions . from the editing on wikipedia , to the votes , to commentaries on news2.ru , these actions represents different and natural underlying dynamics of social networks , since they range from collaborative interaction ( wikipedia ) to discussions about different interesting of human behavior ( new2.ru ) , which are intrinsic properties of the social nature of the web .
we have collected details about user activity in the wikipedia project and reconstructed the underlying social network .
in addition to the widely used term and category pages , wikipedia provides special pages associated with specific contributing authors and discussion ( talk ) pages maintained alongside each of these pages .
these user pages are widely used by wikipedia contributors for coordination behind the scenes of the project .
in fact , interaction via user and discussion pages dominates all other communication methods .
however , communication via personal user pages ( and the corresponding discussion pages ) differs from the topic - associated talk pages in that it is explicit person - to - person communication rather than general topic specific , usually impersonal communication . by tracing users contributing to other user s personal or talk pages ,
we recover the underlying network of wikipedia contributor s personal communication .
not surprisingly , as presented in the next section , the obtained social networks show a scale - free degree distribution , typically observed in a variety of social networks analyzed so far .
the other data set is a de - identified record of activities of social news aggregator news2.ru .
the record contains all actions performed by the community members over more than three years of collaborative selection and discussion of news - related content .
these , user - related actions include such events as submission of news article , comments as well as preference - revealing actions such as voting for articles ( `` dig '' or `` bury '' , using digg.com language ) and other users comments .
in addition to the trace of user activity , the data contains explicit social network layer . each user may publicly declare his / her ( positive , neutral or negative ) attitude to any other user .
considering the personal flavor of the rather emotional way people interact through commentary threads , this list of attitudes when aggregated can be perceived as social network .
in addition , users maintain list of friends , usually including users most favorable on them . these networks are directional , which allows to focus on the incoming links , since they can not be controlled by the target individual , but by his / her friends .
each of these systems represents different approaches to collaborative content creation .
the wikipedia editors interact to create the same content collaboratively so that the content contributed by one user can be complemented , altered or completely removed by others .
the news2.ru represents a mixed case in which the content is contributed individually , but collaboratively ranked .
given these fundamental differences in user activity and network dynamics , the similarities between these systems reported below are particularly revealing . to get the exponents @xmath7 and @xmath8 of power - law distribution , we present a rigorous statistical test based on maximum likelihood methods @xcite .
take the degree distribution as an example .
we fit degree distribution assuming a power law within a given interval . for this
, we use a generalized power - law form @xmath48 where @xmath49 and @xmath50 are the boundaries of the fitting interval , and the hurwitz @xmath51 function is given by @xmath52 .
we use the maximum likelihood method , following the rigorous analysis of clauset et al .
the fit was done in an interval where the lower boundary was @xmath49 . for each @xmath49 value
we fix the upper boundary to @xmath53 , where @xmath54 is the maximal degree .
we calculate the slopes in successive intervals by continuously increasing @xmath49 and varying the value of @xmath55 . in this way
, we sample a large number of possible intervals . for each one of them , we calculate the maximum likelihood estimator through the numerical solution of @xmath56),\ ] ] where @xmath57 are all the degrees that fall within the fitting interval , and @xmath58 is the total number of nodes with degrees in this interval . the optimum interval was determined through the kolmogorov - smirnov ( ks ) test . for the goodness - of - fit test
, we use the monte carlo method described in @xcite . for each possible fitting interval
, we calculate the kolmogorov - smirnov statistics @xmath59 for the obtained cumulative distribution function .
then we choose the interval with the minimal @xmath59 as the best fitting interval and take the @xmath60 in this interval as the final result .
as to the standard error estimation , we adopt the method in @xcite . the standard error on @xmath60 , which is derived from the width of the likelihood maximum , is @xmath61 , where @xmath62 is the number of data .
although the fitting method mentioned above is rigorous , it is suitable for fitting probability density distributions .
when we fit the data @xmath63 , we use another fitting method @xcite .
the procedure for determining fitting interval is similar . in each fitting intervals ,
the fittings were done using ordinary least squares methods .
the goodness of fitting was estimated through the coefficient of determination , @xmath64 , where @xmath65 . the value of @xmath64 is used as a measure of how reliably the fitted line describes the observed points , and is often described as the ratio of variation that can be explained by the fitted curve over the total variation .
we assume that any value above @xmath66 represents an accepted fitting .
the final result is the average of the accepted exponent . in fig.[fig5 ]
, each dot represents a distinct wikipedia project page .
horizontal axis measures the total number of edits for each project .
vertical axis represents the fraction of contributors to that project who performed @xmath1 of edits on that project .
this fraction drops fast ( with power law ) as the number of edits grows .
this suggests that the largest projects are dominated by a few very dedicated users .
perhaps more representative are the mean values ; the vertical line indicates the average edits and the horizontal line marks the fraction of users contributing @xmath1 if the work in the average across projects ( approximately @xmath0 ) the accuracy of fit of the data to the theoretical geometric distribution is measured as the @xmath67 goodness - of - fit to the conditional histogram . as an example , consider h1 for the spanish wikipedia data : for the theoretical distribution we use for each activity the mean degree @xmath16 as shown in fig .
[ fig2]b . the @xmath67 value
is then averaged over all activity bins shown in that figure . to test
if this observed average @xmath67 is consistent with chance assuming h1 we generate surrogate data following h1 : for each given activity , we generate the same amount of random numbers from a geometric distribution with the same mean values , calculate the @xmath67 values and again , average across activities .
we draw @xmath68 such samples and obtain a distribution of average @xmath67 ( fig .
[ fig4]b ) .
the chance that the @xmath67 for the spanish wikipedia data occurred by chance ( p - value ) is the fraction of times the surrogate data provided a value larger than the one observed ( red line in fig.[fig4]b ) .
the analysis for h2 is analogous using the data as shown in fig .
the resulting p - values for all datasets can be found in table i. 99 v. pareto , _
cours deconomie politique _
( f. rouge , luzanne , 1896 ) . b. a. huberman and l. a. adamic , internet : growth dynamics of the world - wide web , _ nature _ * 401 * , 131 ( 1999 ) .
r. l. axtell , zipf distribution of u.s .
firm sizes , _ science _ * 293 * , 1818 ( 2001 ) . c. castellano , s. fortunato , and v. loreto , statistical physics of social dynamics , _ rev . mod .
phys_. * 81 * , 591 ( 2009 ) .
barabsi , the origin of bursts and heavy tails in human dynamics , _ nature _ * 435 * , 227 ( 2005 ) .
d. rybski , s. v. buldyrev , s. havlin , f. liljeros , and h. a. makse , scaling laws of human interaction activity , _ proc .
natl . acad .
usa _ * 106 * , 12640 ( 2009 ) .
v. m. yakovenko and j. b. rosser , statistical mechanics of money , wealth , and income , _ rev .
phys . _ * 81 * , 1703 ( 2009 ) .
barabsi and r. albert , emergence of scaling in random networks , _ science _ * 286 * , 509 ( 1999 ) .
g. caldarelli , a. capocci , p. de los rios , and m.a .
muoz , scale - free networks from varying vertex intrinsic fitness , _ phys .
_ * 89 * , 258702 ( 2002 ) .
a. vzquez , a. flammini , a. maritan , and a. vespignani , modeling of protein interaction networks , _ complexus _ * 1 * , 38 ( 2003 ) . g. bianconi and a .-
barabsi , competition and multiscaling in evolving networks , _ europhys .
lett . _ * 54 * , 436 ( 2001 ) .
g. caldarelli , _ scale - free networks : complex webs in nature and technology _ ( oxford univ press , oxford , 2007 ) .
d. watts and s. strogatz , collective dynamics of small - world networks , _ nature _ * 393 * , 440 ( 1998 ) .
m. faloutsos , p. faloutsos , and c. faloutsos , on power - law relationships of the internet topology , _ proceedings of the conference on applications , technologies , architectures , and protocols for computer communication _
( acm , new york , 1999 ) , pp 251262 .
m. e. j. newman , the structure and function of complex networks , _ siam rev .
_ * 45 * , 167 ( 2003 ) .
m. mitzenmacher , a brief history of generative models for power - law and lognormal distributions , _ internet mathematics _ * 1 * , 226 ( 2004 )
. g. u. yule , a mathematical theory of evolution , based on the conclusions of dr .
j. c. willis , f.r.s .
, _ philos .
b _ * 213 * , 21 ( 1925 ) . h. a. simon , on a class of skew distribution functions , _ biometrika _ * 42 * , 425 ( 1955 ) .
b. mandelbrot , _ communication theory , ed .
w. jackson _ ( butterworth , london , 1953 ) , pp .
r. m. dsouza , c. borgs , j. t. chayes , n. berger , and r. d. kleinberg , emergence of tempered preferential attachment from optimization , _ proc .
usa _ * 104 * , 6112 ( 2007 ) .
f. papadopoulos , m. kitsak , m. .
serrano , m. bogu , and d. krioukov , popularity versus similarity in growing networks , _ nature _ * 489 * , 537 ( 2012 ) .
j. leskovec and e. horvitz , planetary - scale views on a large instant - messaging network , _ proceedings of the 17th international conference on world wide web _ , pp .
915924 ( 2008 ) .
perra , n. , gonalves , b. , pastor - satorras , r. & vespignani , a. activity driven modeling of time varying networks .
_ scientific reports _ * 2 * , 469 ( 2012 ) .
r. kumar , j. novak , p. raghavan , and a. tomkins , structure and evolution of blogspace , _ communications of the acm - the blogosphere _ * 47 * , 35 ( 2004 ) .
w. h. hsu , j. lancaster , m. s. r. paradesi , and t. weninger , structural link analysis from user profiles and friends networks : a feature construction approach , _ proceedings of the international conference on weblogs and social media _ , pp 7580 ( 2007 ) .
d. liben - nowell and j. kleinberg , tracing information flow on a global scale using internet chain - letter data , _ proc .
usa _ * 105 * , 4633 ( 2008 ) . f. topse , information theoretical optimization technique , _ kybernetika _ * 15 * , 8 ( 1979 ) .
j. pearl , causal inference in statistics : an overview , _ statistics surveys _ * 3 * , 96 ( 2009 ) .
p. hoyer , d. janzing , j. mooij , j. peters , and b. schlkopf , nonlinear causal discovery with additive noise models , _ proceedings of the conference neural information processing systems _ , ( 2009 ) .
k. zhang and a. hyvrinen , on the identifiability of the post - nonlinear causal model , _ proceedings of the 25th conference on uncertainty in artificial intelligence _
( auai press , arlington , 2009 ) , pp .
d. janzing , j. mooij , k. zhang , j. lemeire , j. zscheischler , p. daniuis , b. steudel , and b. schlkopf , information - geometric approach to inferring causal directions , _ artificial intelligence _ * 182 * , 1 ( 2012 ) .
a. clauset , c. r. shaliz , and m. e. j. newman , power - law distribution in empirical data , _ siam rev .
_ * 51 * , 661 ( 2009 ) .
l. k. gallos , p. barttfeld , s. havlin , m. sigman , and h. a. makse , collective behavior in the spatial spreading of obesity , _ sci .
* 2 * , 454 ( 2012 ) .
we thank g. khazankin , research institute of physiology sb rams for kindly providing access to invaluable data on news2.ru user activity . the research is supported by nsf emerging frontiers , arl , fp7 project socionical and multiplex , cnpq , capes , and funcap .
h.a.m . , s.h . and j.s.a .
designed research . l.m .
prepared data . l.m .
, l.c.p . and s.d.s.r .
analyzed the data .
all authors wrote , reviewed and approved the manuscript .
competing financial interests : the authors declare no competing financial interests . fig .
1 . probability distribution of activities and degree . ( a ) probability density function of wikipedia contributors as a function of the number of performed page edits in four languages .
( b ) probability density function of news2.ru for five different activities.lines indicate power - law fitting for spanish and stories with the maximum likelihood methods .
( c ) probability distribution of degree for social networks as a function of number of links between wikipedia contributors .
degree represents the number of links other users establish with a given user .
( d ) distribution for networks of relationship ( positive / negative ) between users of news2.ru web portal and users friendships .
analysis of joint distribution of activity and degree .
( a ) scatter plot of degree and activity for each user in wikipedia spanish dataset .
( b ) mean degree @xmath16 for given activity .
( c ) mean activity @xmath17 for given degrees .
( d ) standard deviation of degree @xmath18 for given activity .
( e ) @xmath20 for given degree .
( f ) relationship between standard deviation of degree @xmath18 and the mean value @xmath16 for given activity .
inset is the theoretical fit of geometric distributions for spanish wikipedia .
( g ) @xmath20 versus @xmath17 for given degree .
3 . test of `` maximum entropy attachment model '' via the geometric distribution . theoretical relationship of mean and standard deviation for geometric distribution ( solid curve ) and data points for wikipedia in four languages .
4 . causal hypotheses and test result . ( a )
schematic diagrams for hypotheses h1 and h2 .
h1 : mean degree is determined by activity through function @xmath22 .
then degree is random distributed according to the conditional probability distribution @xmath46 .
h2 is the other way around .
( b ) and ( c ) results of monte - carlo simulation with @xmath68 samples following h1 and h2 for the spanish wikipedia data .
the vertical red lines show the goodness - of - fit @xmath67 of the actual data to h1 and h2 , respectively .
the empirical analysis clearly favors h1 over h2 . | the probability distribution of number of ties of an individual in a social network follows a scale - free power - law .
however , how this distribution arises has not been conclusively demonstrated in direct analyses of people s actions in social networks . here
, we perform a causal inference analysis and find an underlying cause for this phenomenon .
our analysis indicates that heavy - tailed degree distribution is causally determined by similarly skewed distribution of human activity .
specifically , the degree of an individual is entirely random - following a `` maximum entropy attachment '' model - except for its mean value which depends deterministically on the volume of the users activity .
this relation can not be explained by interactive models , like preferential attachment , since the observed actions are not likely to be caused by interactions with other people . |
broad absorption lines ( bals ; weymann et al .
1991 ) are the most readily apparent signature of outflowing gas .
strong outflows are believed to be one of the most important feedback processes connecting central supermassive black holes ( smbhs ) to their host galaxies .
however , the detailed physics of these outflows is not well understood .
studies of bals can provide a useful approach to placing constraints on the physical properties of outflows to further understand the overall picture of galaxy evolution and the connection between smbhs and their surrounding hosts in quasars .
bals are observed in the blueward of emission lines with a wide range of blueshifted velocities ( up to @xmath3 ) and broad widths ( at least 2000 km s@xmath4 ) ( weymann et al .
based on the species of absorption , bals are divided into three subclasses , i.e. , high - ionization bals ( hibals ) , low - ionization bals ( lobals ) and iron low - ionization bals ( felobals ) ( e.g. , hall et al .
in addition to the bals of the high - ionization species ( e.g. , ) in hibals , and those of the low - ionization species ( e.g. , and ) in lobals , felobals also show bals in ground and excited of states feii and/or feiii ( hazard et al . 1987 ; becker et al .
statistical studies indicate that approximately 10 - 20% of quasars exhibit bals , and @xmath5 15% of bal quasars are detected as lobal quasars ; by contrast , felobals are the rarest subclass , appearing in at most @xmath5 1% of bal quasars ( e.g. , weymann et al . 1991 ; reichard et al . 2003 ; trump et al . 2006 ; gibson et al . 2009 ; zhang et al . 2010 ) .
bal variability studies provide one method of assessing bal structures , locations and dynamics , and the results of such studies can potentially constrain the physical mechanisms responsible for outflows . in principle , the observed bal variability could be explained by variations in the ionization parameter ( e.g. , trevese et al .
2013 ) , clouds transiting across the line of sight ( e.g. , hall et al . 2011 ) or a combination of processes driven by these two factors .
recently , time variability in individual sources or samples with multi - epoch observations has been reported for hibals ( e.g. , lundgren et al .
2007 ; capellupo et al . 2012 , 2013 ; filiz ak et al . 2013 ; joshi et al . 2014 ) , for lobals ( zhang et al .
2011 ; vivek et al .
2014 ) and for felobals ( vivek et al . 2012 ; mcgraw et al .
absorption - line variability is perhaps more common within shallower and higher - velocity troughs ( e.g. , capellupo et al . 2011 ) on longer time scales ( e.g. , gibson et al . 2008 , 2010 ;
capellupo et al .
2011 ) ; occasionally , absorption lines are even observed to disappear completely ( e.g. , hall et al .
2011 ; filiz ak et al .
however , there has been no study to date that has compared variability among different bal subclasses .
radio observations could be used to constrain bal orientations through investigations of radio variability , spectral index and morphology ( zhou et al .
2006 ; dipompeo et al .
2012 and references therein ) . to date , only one bal variability investigation of radio - loud ( rl ) bal quasars has been performed by welling et al .
they found that the bal variability did not significantly depend on either radio luminosity or radio - loudness ; however , there was tentative evidence of greater fractional bal variability within lobe - dominated rl quasars .
the first bright quasar survey ( fbqs ; white et al .
2000 ) provided a uniform radio - selected sample of 29 definite and tentative bal quasars ( becker et al .
2000 ) . using the data collected in this survey , it is possible to perform a statistical comparison of bal variability among radio - selected bal subclasses .
this paper is organized as follows : 2 describes the multi - epoch observations of the sample , 3 presents the bal variability , and 4 summarizes and discusses the implications . throughout this paper ,
we adopt the cdm ` concordance ' cosmology with h0 = 70 km s@xmath6mpc@xmath6 , @xmath7 , and @xmath8 .
white et al . ( 2000 ) presented the optical spectra of 636 quasars in the fbqs distributed over 2682 deg@xmath9 in their table 2 and figure 16 .
these quasar data form a heterogeneous set , collected at 5 different observatories between 1996 and 1998 .
the maximum coverage of the spectroscopic instruments is from @xmath5 3600 to 10000 @xmath10with a @xmath54 - 10 @xmath10resolution .
we checked these spectra and adjusted the bal quasar sample identified by white et al .
( 2000 ) and becker et al .
two tentative bals were rejected , one quasar ( fbqs j105528.808 + 312411.65 ) was reclassified as a felobal quasar , and 4 lobals were reclassified as hibals and felobals .
the final working sample contained 28 bals , including 15 hibals , 4 lobals and 9 felobals ( table [ tab1 ] ) .
of these , 5 felobals were discovered to exhibit abrupt dropes in flux , yielding almost no continuum windows below and causing overlapping troughs ; these were classified as ` overlapping - trough ' felobals ( ofelobals ) .
we searched the repeated spectroscopic observations from the sloan digital sky survey ( sdss ; york et al .
2000 ) databases that have been obtained over the past twelve years , from may 2000 to july 2012 . in the sdss seventh data release ( dr7 ; abazajian et al .
2009 ) , 27 cases have spectra of higher spectral resolution .
the dr7 spectrographs cover the wavelength range from 3800 to 9200 @xmath10with a resolution of @xmath5 1.5 @xmath10at 6500 @xmath10 , which is nearly coincident with the wavelength coverage of the fbqs .
in addition , 16 cases were observed with the baryon oscillation spectroscopic survey ( boss ) spectrographs , whose spectral resolution varies from @xmath5 1300 at 3600 @xmath10to 2500 at 10,000 @xmath10(smee et al .
these new repeated observations were released as part of the sdss tenth data release ( dr10 ; ahn et al .
2014 ) . meanwhile , we performed follow - up observations of fbqs j072831.6 + 402615 and j105528.8 + 312411 using the opto - mechanics research ( omr ) spectrograph of the xinglong 2.16 m telescope of the national astronomical observatories of china ( naoc ) on march 24 , 2009 .
a 200 @xmath10mm@xmath6 dispersion grating and a pi @xmath11 ccd detector were used to cover a wavelength range of 3800 - 8500 @xmath10at a 4 @xmath10resolution .
two exposures of 45 and 60 minutes each were acquired for both cases .
a slit width of 2.5 was chosen to match the typical seeing disk . for fbqs j072831.6 + 402615
, an additional spectroscopic observation was performed on dec .
12 , 2013 , using the yunnan faint object spectrograph and camera ( yfosc ) instrument of the lijiang 2.4 m telescope of the yunnan astronomical observatories ( ynao ) .
the blue - band grism , g14 , provided a wavelength coverage from @xmath5 3500 to 7500 @xmath10with a spectral resolving power of 1337 .
the typical seeing was around 1.5 , the slit width was 2.0 , and one exposure of 30 minutes was acquired .
the follow - up spectroscopic data were reduced using the iraf longslit and echelle packages .
the repeated spectra , which were converted to the quasar rest frame using the redshifts reported in white et al .
( 2000 ) , can not be directly compared with the fbqs spectra because of the absolute flux calibration or continuum variability .
thus , we used a power - law form ( @xmath12 ) to scale the repeated spectra to match the fbqs spectra , thereby masking the obvious absorption troughs .
however , we found that the power - law indexes were equal to zero in most cases .
the large time intervals between the fbqs and repeated spectroscopic observations are comparable to the expected time scale for outflow rearrangement , thus making the detection of variability more likely .
we restricted our focus to bal variability , while neglecting both emission - line variability and residual - continuum variability . through spectral comparisons among two or three epochs
, one can clearly identify variable absorption troughs in 12 bal quasars , including 5 `` strong '' variations ( figure 1 , 2 and 3 ) and 7 `` typical '' variations ( figure 4 ) . in this work
, a typical variation refers to a small - amplitude velocity structure and/or optical - depth change of the absorption trough ; by contrast , a strong variation is defined as an extremely large and continuous change . in the figures ,
the fbqs spectra are plotted in black , and the repeated spectra are plotted in red and green .
table 1 suggests that bal variability does not tend to increase with greater radio - loudness or luminosity .
this is consistent with the study of the bal variability of rl quasars performed by welling et al .
( 2014 ) , in which no evidence of any correlation of the radio - loudness and luminosity with the bal variability was found ; the bal variations in rl bal quasars appear to occur in a similar manner as those in radio - quiet bal quasars .
when we further explore the subtypes and structures of the bals , we find that the trends of bal variability are completely different for hibals , lobals , felobals and ofelobals .
typical bal variations are detected in the first three subtypes ; however , intriguingly , all 5 instances of strong variations are detected in ofelobals .
thus , we focus on strong variability in this work . in figures 1 , 2 and 3 ,
we present the difference spectra for strong variations .
the difference spectra are the observed spectra for the corresponding mjds normalized to the highest fluxes among the multi - epoch observations ; these spectra reveal the absorption - line variability more clearly .
next , we analyze the properties of these cases . as shown in figure 1 , fbqs j072831.6 + 402615 and j140806.2 + 305448 ( first reported in hall et al . 2011 ) exhibit the strong weakening and even disappearance of the absorption lines .
we providentially observed the process of transformation from an ofelobal quasar to a non - bal quasar in the repeated spectroscopic observations of these quasars . for fbqs j072831.6 + 402615 ,
its spectrum from the fbqs epoch contains strong absorption lines from and multiplets , with a blueshifted velocity of @xmath5 16,000 km s@xmath4 . in the omr spectrum
acquired 6.85 rest - frame years later , the absorption troughs of and uv1 and uv2 + 3 had decreased in absorption depth by a factor of four .
after an additional 2.88 rest - frame years , the residual overlapping troughs had completely disappeared in the yfos spectrum .
fbqs j140806.2 + 305448 was a spectacular ofelobal quasar with a covering fraction of absorption of @xmath5 75 percent , but now , it is only a modestly absorbing lobal
quasar . hall et al .
( 2011 ) have found there was no significant absorption variability in early observations , one rest - frame year prior to the keck / esi observation in march 2000 ; however , the trough outflowing at 12,000 km s@xmath4decreased by a factor of two , and troughs at the same velocity disappeared over the next @xmath5 7 rest - frame years .
the sdss dr10 spectrum shows a complete lack of troughs ( see also the het spectrum in hall et al .
2011 ) .
the top panels of figure 2 show that the three spectra of fbqs j152350.4 + 391404 are similar to the intermediate epochs of fbqs j140806.2 + 305448 , especially the keck / esi spectrum ( mjd 53377 ; hall et al .
troughs of multiplets are clearly visible in all three epochs .
the difference spectra reveal strong variability in the ultraviolet multiplets ( uv1 and uv2 + 3 ) and very slight variations in the optical multiplets ( opt8 and opt6 + 7 ) .
meanwhile , the red - band spectrographs of the sdss and boss captured a highly blueshifted velocity and an extremely broad absorption trough of h@xmath13 at approximately 4700 @xmath10 in the sdss dr7 and dr10 spectra .
furthermore , a near - infrared spectrum that is quasi - simultaneous with the sdss dr10 spectrum , which was acquired by the triplespec spectrograph of the hale 200-inch telescope at palomar observatory , suggests the existence of h@xmath14 bals .
the bottom panels present the observed spectra of the h@xmath13 and h@xmath14 regimes overlaid with the best - fit models and the normalized spectra of the blueshifted balmer absorption troughs .
the fitting procedure has been described in detail by dong et al .
( 2008 ) ; therefore , we outline it only briefly here . the optical continuum from 3600
@xmath10to 6900 @xmath10is approximated by a single power law .
emission multiplets , both broad and narrow , are modeled using the i zw 1 template provided by vron - cetty et al .
the emission lines are modeled as multiple gaussians : two gaussians for broad balmer lines , and one gaussian for [ ] .
as shown in figure 3 , the bals of multiplets are ambiguous or severely overlapped in fbqs j105528.8 + 312411 and j140800.4 + 345125 .
the spectral variability in these two cases is deepening overall below and blueward .
it is difficult to resolve the absorption multiplets based on their single observations , but it is possible to identify strong variations of the iron multiplets in their difference spectra .
the triple - epoch spectra of fbqs j105528.8 + 312411 reveal that the bals are nearly saturated and exhibit no shift .
the absorption structures are somewhat similar to those of fbqs j140806.2 + 305448 and j152350.4 + 391404 .
we present the identification of the ultraviolet and optical multiplets in the difference spectra . for fbqs j140800.4 + 345125
, we identify the absorption troughs of , and uv1 ( perhaps higher - velocity ) in the fbqs spectrum and mark them by blue diagonal line regions . however , the residual fluxes of the three troughs are constant in all epochs .
the normalized fluxes of these troughs in the difference spectra are @xmath5 1 .
this broad absorption line system is invariant .
in addition , there are other broad - band absorption troughs in the difference spectra blueward of the invariant absorption troughs noted above , which are surmised to represent the variable absorption of uv62 , uv1 and uv2 + 3 with the same blueshifted velocity as the invariant absorption system ; the iron absorption should be completely overlapping .
another possible explanation is that the absorption may be undetected in the later spectra of fbqs j105528.8 + 312411 and j140800.4 + 345125 .
the absorption variability may be an illusion caused by a variation in the strength of the emission , or the apparent overlapping - trough absorption may instead be attributable to reddening changes in the unusual reddening curves , which may cause a rapid dropoff in the flux below a certain wavelength ( e.g. , leighly et al .
in this work , we constructed a radio - selected bal quasar sample from the fbqs , including 28 reclassified bals , and also considered subsequent spectroscopic observations of these quasars from the sdss and two other instruments to explore the bal variability among their dual-/triple - epoch spectroscopic observations . of the investigated sample , 27 bal quasars were observed in at least one epoch in the sdss , and follow - up observations of 2 additional quasars were obtained . when we scaled the subsequently acquired spectra to match the continua of the fbqs spectra using a power - law form , we detected variable absorption troughs in 12 bal quasars through the spectral comparison , including 5 bal quasars with strong variations in the bals of multiplets .
intriguingly , all strong bal variations were detected in ofelobals ; however , other bals , either with typical variations or without variations , were non - ofelobals .
the apparently opposite trend was observed by vivek et al .
( 2012 , 2014 ) ; in that study , felobals were found to be less variable than hibals and lobals .
vivek et al .
( 2012 ) probed the time variability of 5 felobals spanning an interval of up to 10 years but only detected strong variations of uv34 and uv48 in the spectra of sdss j221511.93 - 004549.9 .
based on photoionization models , de kool et al .
( 2002b ) showed that an column density that is higher than that of can be easily produced in a high - density outflow ( @xmath15 @xmath2 for an ionization parameter of @xmath16 ) .
however , the remaining 4 felobals did not exhibit any significant changes either in optical depth or in velocity structure .
two of them , according to studies of their high - resolution spectra performed with vlt and keck / hires , suggested an electron density of @xmath17 @xmath2 with a distance from the continuum emission source of @xmath18 kpc for sdss j031856.62 - 060037.7 ( dunn et al .
2010 ) and @xmath19 @xmath2 with @xmath20 230 pc for sdss j084044.41 + 363327.8 ( de kool et al .
2002b , also included in this work . ) . for the 5 ofelobals with strong variations studied in this work , the physical properties of the outflow gas are different .
in the rotating outflow model , the trough variability arises from a structure in the bal outflow moving out of our line of sight to the ultraviolet - continuum - emitting region of the quasar s accretion disk . based on the size of that region and the time scale over which the absorption changes , hall et al .
( 2011 ) constrained the bal structure of fbqs j140806.2 + 305448 to a transverse velocity between 2600 and 22,000 km s@xmath4 and a line - of - sight velocity of 12,000 km s@xmath4 ; thus , the moving bal structure was considered to be located approximately 1.7 to 14 pc from the black hole , nearly ten to ninety times farther than the h@xmath13 broad - line region . meanwhile , the application of the large - scale spectral synthesis code cloudy ( c10.00 ; ferland et al .
1998 ) in the extensive parameter space suggests that the absorbers of all ofelobals considered in this work are ionized gases with a density of @xmath21 @xmath2 and a column density of @xmath22 @xmath23 .
the absorbers in fbqs j140806.2 + 305448 and j152350.4 + 391404 are located at distances from the subparsec scale to dozens of parsecs . for fbqs j072831.6 + 402615
, we can provide only an upper limit of @xmath5 447 pc , with considerable uncertainty .
more details can be found in appendix a. similarly , mcgraw et al .
( 2014 ) detected the bal variability of low - ionization species ( and ) in 4 objects with a representative upper limit on the distance of the absorber from the center engines of @xmath5 20 pc , with crossing speeds of @xmath24 km s@xmath4 .
for the remaining felobals considered in this work , previous works have reported different locations for the outflow gas .
de kool et al .
( 2002b ) have suggested that the bal absorbers in fbqs j084044.4 + 363328 consist of a high - velocity , low - density gas and that the scale of the distance from the black hole is two hundred parsecs .
analyses of the keck / hires spectra of fbqs j104459.5 + 365605 and j121442.3 + 280329 have placed the outflow gas of the absorbers at distances from the nucleus of 700 pc , with ne @xmath25 @xmath2 , and 130 pc , with @xmath1 @xmath2 , respectively ( de kool et al .
2001 , 2002a ) . for other felobals , specifically non - overlapping felobals ( non - ofelobals ) , in the literature
, the typical distances between the outflow gas and the nucleus have been derived to be approximately 3 to 28 kpc ( 3c 191 , hamann et al .
2001 ; qso j2359 - 1241 , korista et al . 2008
; sdss j0838 + 2955 , moe et al .
2009 ; sdss j0318 - 0600 , dunn et al . 2010
; collected in table 4 of lucy et al .
2014 ) .
above all , felobals with strong variations are constrained to be located in regions of high density ( @xmath26 @xmath2 ) and close to the continuum ionization sources ( the upper limit is generally dozens of times farther than the size of the broad - emission - line region )
. however , the absorbers of other felobals consist of low - density gas ( @xmath27 @xmath2 ) and are located at distances of hundreds to thousands of parsecs from the central black holes . in other words , the former are located in active galactic nucleus ( agn ) environments rather than in the host galaxy .
two subgroups , i.e. , felobals with and without strong bal variations , may represent the bimodality of absorption observed by de kool et al .
( 2002b ) .
our findings regarding the different bal variability behaviors of these two felobal subgroups suggest that they also correspond to overlapping and non - overlapping felobals .
if our hypothesis is correct , then ofelobals correspond to densities that are `` higher by several orders of magnitude than in the absorbers in the ` low - density gas at a large distance ' group '' , and `` their properties are best explained if they are 1000 times closer to the nucleus '' ( de kool et al .
2002b ) . on the other hand , we suggest that a high density of the absorbing gas and a large width of the absorption lines are the necessary conditions for ofelobals ( hall et al .
a higher density implies that the absorbing gas should be located closer to the central engine , and higher densities and broader absorption lines certainly require smaller bals ; these conditions are certainly consistent with the necessary conditions for variable bals .
the above estimations of the gas density and distance of strongly variable felobals are in good agreement with the requirements for ofelobals .
this is the reason why strong bal variations are detected only in ofelobals .
the possible geometries and origins ( arav et al . 1994 ; de kool & begelman 1995 ; murray et al .
1995 and elvis 2000 ) imply that the variability of the bals is explained by variations in the ionizing flux originating in the inner part of the accretion disk ( e.g. , netzer et al . 2002 ; trevese et al .
2013 ) and a transiting structure in the bal outflow ( e.g. , risaliti et al . 2002 ; hall et al .
the very optically thick gas that is required for ofelobals is very unlikely to undergo any change in ionization that is sufficiently large for the gas to become only slightly optically thick and thus exhibit a detectable response to ionizing - continuum variability .
meanwhile , the nine years of monitoring performed by the catalina surveys ( drake et al . 2014 ) , beginning in april 2005 , shows there is no apparent change in the optical continuum , at least in the @xmath28 band , for these ofelobal quasars ( table 1 ) .
however , we can not definitively rule out a change in the ionizing flux as the explanation for the strong variability of the absorption .
observations in the v band do not reflect the change in extreme uv and soft x - ray photons at @xmath29 @xmath10 ; moreover , the ionizing and observed continua along our line of sight may not always track each other ( kaastra et al .
if the bal variability is related to the transverse motions of flows , then a large transverse velocity is required to produce such variability on a relatively short time scale , and the absorbing gas must be close to the black hole in a rotating disk wind model . this analysis agrees with hall et al .
( 2011 ) that the strong bal variability is possibly attributable to variability in the covering factor of bal regions caused by clouds transiting across the line of sight rather than to ionization variations .
we also note that 3 of the 4 ofelobal quasars for which there are measurements of the radio spectral index are steep - spectrum radio sources , and the remaining one ( fbqs j152350.4 + 391404 ) has a radio spectral index of @xmath30 , falling close to the dividing line , @xmath31 ( becker et al . 2000 ) .
the steep radio spectra suggest that these quasars should be regarded as lobe - dominated radio sources that are viewed at larger angles .
this is also consistent with the interpretation of the strong bal variations based on the rotating outflow model .
the authors are grateful to the anonymous referee for the helpful suggestions and the staff at the lijiang 2.4 m telescope and the xinglong 2.16 m telescope for the support during the observations .
this work is supported by chinese natural science foundation ( nsfc-11203021 ) , national basic research program of china ( the 973 " program , 2013cb834905 ) and the soc program ( chinare2014 - 02 - 03 ) .
tw acknowledge financial support from chinese natural science foundation ( nsfc-11421303 ) and the strategic priority research of sciences ( xdb09000000 ) .
we acknowledge the use of the lijiang 2.4 m telescope of the yunnan astronomical observatories , the xinglong 2.16 m telescope of the national astronomical observatories of china , and the hale telescope at palomar observatory through the telescope access program ( tap ) , as well as the archive data from the fbqs , catalina surveys and sdss .
funding for the lijiang 2.4 m telescope has been provided by chinese academy of sciences and the people s government of yunnan province .
the xinglong 2.16 m telescope is supported by the open project program of the key laboratory of optical astronomy , national astronomical observatories , chinese academy of sciences .
tap is funded by the strategic priority research program the emergence of cosmological structures " ( xdb09000000 ) , national astronomical observatories , chinese academy of sciences , and the special fund for astronomy from the ministry of finance .
observations obtained with the hale telescope at palomar observatory were obtained as part of an agreement between the national astronomical observatories , chinese academy of sciences , and the california institute of technology .
funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , and the u.s .
department of energy office of science .
the sdss - iii web site is http://www.sdss3.org/. abazajian , k. n. , et al .
2009 , , 182 , 543 ahn , c. p. , et al .
2014 , , 211 , 17 ai , y. l. , et al .
2010 , , 716 , l31 arav , n. , li , z .- y . , & begelman , m. c. 1994 , , 432 , 62 baldwin , j. a. , et al .
2004 , , 615 , 610 becker , r. h. , et al .
1997 , , 479 , l93 becker , r. h. , et al .
2000 , , 538 , 72 capellupo , d. m. , et al .
2011 , , 413 , 908 capellupo , d. m. , et al .
2012 , , 422 , 3249 capellupo , d. m. , et al .
2013 , , 429 , 1872 de kool , m. , & begelman , m. c. 1995 , , 455 , 448 de kool , m. , et al .
2001 , , 548 , 609 de kool , m. , et al . 2002a , , 567 , 58 de kool , m. , et al .
2002b , , 570 , 514 dipompeo , m. a. , et al .
2012 , , 752 , 6 dong , x. , et al .
2008 , , 383 , 581 drake , a. j. , et al . 2014 , , 213 , 9 dunn , j. p. , et al .
2010 , , 709 , 611 elvis , m. 2000 , , 545 , 63 ferland , g. j. , et al .
1998 , , 110 , 761 filiz ak , n. , et al . 2012 , , 757 , 114 filiz ak , n. , et al . 2013 , , 777 , 168 gibson , r. r. , et al .
2008 , , 675 , 985 gibson , r. r. , et al .
2009 , , 692 , 758 gibson , r. r. , et al .
2010 , , 713 , 220 hall , p. b. , et al .
2002 , , 141 , 267 hall , p. b. , et al .
2011 , , 411 , 2653 hamann , f. w. , et al .
2001 , , 550 , 142 hazard , c. , et al .
1987 , , 323 , 263 kaastra , j. s. , et al .
2014 , science , 345 , 64 korista , k. t. , et al .
2008 , , 688 , 108 leighly , k. m. , et al .
2014 , , 788 , 123 lucy , a. b. , et al .
2014 , , 783 , 58 lundgren , b. f. , et al .
2007 , , 656 , 73 mcgraw , s. m. , et al .
2014 , iau symposium , 304 , 417 moe , m. , et al .
2009 , , 706 , 525 murray , n. , et al .
1995 , , 451 , 498 netzer , h. , et al .
2002 , , 571 , 256 reichard , et al .
2003 , , 126 , 2594 risaliti , g. , et al .
2002 , , 571 , 234 smee , s. a. , et al .
2013 , , 146 , 32 trevese , d. , et al .
2013 , , 557 , a91 trump , et al . 2006 , , 165 , 1 vron - cetty , m .-
p . , joly , m. , & vron , p. 2004
, , 417 , 515 vivek , m. , et al .
2012 , , 423 , 2879 vivek , m. , et al .
2014 , , 440 , 799 welling , c. a. , et al .
2014 , , 440 , 2474 weymann , r. j. , et al .
1991 , , 373 , 23 white , r. l. , et al .
2000 , , 126 , 133 york , d. g. , et al .
2000 , , 120 , 1579 zhang , s. , et al .
2010 , , 714 , 367 zhang , s .- h . , et al .
2011 , ra , 11 , 1163 zhou , h. , et al .
2006 , , 639 , 716
the ofelobals considered in this work are severely overlapping ; thus , the geometry and physical conditions of the outflow winds can not be accurately obtained through analysis of the combination of the absorption lines .
fortunately , we can approximately evaluate the bals by employing the large - scale spectral synthesis code cloudy ( c10.00 ; ferland et al .
1998 ) in the extensive parameter space . in the photoionization simulation ,
the geometry is assumed as a slab - shaped absorbing medium exposed to the ionizing continuum from the central engine with a uniform density , metallicity and abundance pattern .
a typical active galactic nucleus multi - component continuum from mathews & ferland ( 1987 ) is specified as the incident ionizing radiation .
the solar elemental abundance is adopted , and the gas is assumed to be free of dust .
each individual simulation model is customized in terms of the ionization parameter ( @xmath32 ) , electron density ( @xmath33 ) and hydrogen column density ( @xmath34 ) .
the variation ranges in @xmath35 space are @xmath36 , @xmath37 and @xmath38 , with a dex step of 1.0 . the full 371-level fe@xmath39 model implemented in cloudy , which includes all levels up to 11.6 ev ,
is used to reproduce the absorption .
the employed model can predict the populations of various levels of fe@xmath39 and the strengths of the absorption lines arising from these levels .
figure a1 presents a series of models in @xmath35 space . to clearly illustrate the trend , one gaussian profile , with its full width at half maximum ( @xmath40 ) set to 2000 km s@xmath4 ,
is used to visualize the models .
the calculations suggest that the iron multiplets raised from the ground state ( e.g. , uv2 + 3 at approximately 2400 @xmath10and uv1 at approximately 2600 @xmath10 ) are the most pervasive and that they are even generally saturated when the outflow winds have high column densities
. however , the iron multiplets that are raised from the excited state ( e.g. , uv144 - 149 and uv158 - 164 in the wavelength range of 2462 - 2530 @xmath10as well as opt6 + 7 and opt8 ) are more sensitive to the selected parameters .
the iron absorption troughs increase with increasing fe@xmath39 density , as determined by the combination of the ionization parameter , the electron density and the hydrogen column density .
moreover , for larger @xmath32 or @xmath33 , more fe@xmath39 will become ionized to fe@xmath41 , and the absorption troughs of the high multiplets uv144 - 149 and uv158 - 164 as well as opt6 + 7 and opt8 will become even weaker .
we can approximately limit the physical parameters of the outflow winds based on the absorption strength / depth ratios of the uv1 and uv2 + 3 multiplets , the uv144 - 149 and uv158 - 164 multiplets , the uv1 and uv2 + 3 multiplets , and the opt8 multiplets as is done for the estimation of the strength ratios of emission multiplets in baldwin et al .
the uv144 - 149 and uv158 - 164 multiplets in four ofelobals ( with the exception of fbqs j105528.8 + 312411 ) that are considered in this work have depths that are nearly identical to those of the troughs of uv1 and uv2 + 3 , and in three cases , considerable absorption troughs of opt6 + 7 and opt8 are evident .
these characteristics suggest that the outflow winds of ofelobals should be of high density ( @xmath42 ) and high column density ( @xmath43 ) . for individual sources with relatively isolated absorption multiplets , we attempt to match the observed maximal difference spectra using photoionization models with appropriate broadening widths and blueshifted velocities .
the covering factor is set to the average depths of uv1 and uv2 + 3 . in the search for optimal photoionization models , the reduced 2
is evaluated using data covering absorption rest frames of 2300 to 3400 @xmath10 , which contain the overall absorption features of the multiplets mentioned above . because the photoionization simulations are performed on a large grid , the physical parameters of the optimal models are determined through approximation , and a simulation using incremental grids will accurately evaluate the observations . in the right - hand panels of figure 1 and the top - right panel of figure 2 ,
the best - matched photoionization models for the observed difference spectra are represented by blue curves . for fbqs j140806.2 + 305448 ,
a model with @xmath44 , @xmath45 , @xmath46 and @xmath47 km s@xmath4exhibits absorption features that are quite consistent with the observations , with the exception of the absorption between the troughs of and opt8 .
the underestimation of the absorption at approximately 2850 @xmath10is attributed to @xmath482835 , 2840 , 2849 , 2860 , 2867 and @xmath482851 ( shi et al . , submitted ) .
similarly , a model with @xmath49 , @xmath50 , @xmath46 and @xmath51 km s@xmath4is the optimal photoionization model for fbqs j152350.4 + 391404 , and there are also obvious absorption lines present at approximately 2850 @xmath10 in the difference spectra . using the ionization parameter definitions and the inferred values of the ionization photon emission rate , ionization parameter and density , one can determine the distance of the outflow winds from the central ionization source , which is found to be @xmath52 0.16 pc for fbqs j140806.2 + 305448 and 34 pc for fbqs j152350.4 + 391404 .
based on the absorption depths of uv1 and uv2 + 3 and possibly uv62 , we obtain a model spectrum using @xmath53 , @xmath54 and @xmath55 to match the difference spectrum of fbqs j072831.6 + 402615 ; the distance of the outflow winds is then found to be @xmath5 447 pc . however , the difference spectra of fbqs j152350.4 + 391404 and j072831.6 + 402615 indicate stronger absorption of the uv144 - 149 and uv158 - 164 multiplets than is present in the optimal models , which most likely suggests a higher density and a higher ionization parameter or some microturbulence in the outflow winds ( shi et al . ,
submitted ) .
if the density and/or ionization parameters of these two cases are underestimated , then the absorbing gas should be located closer to the ionization continuum sources than our estimates indicate .
cc cc cc cc cc cc cl name ( fbqs)&@xmath56&@xmath57&log @xmath58 & log @xmath59&@xmath14&mjd&mjd - plate - fiber&mjd - plate - fiber&@xmath60&@xmath61&@xmath62&@xmath63&type + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8 ) & ( 9 ) & ( 10 ) & ( 11 ) & ( 12 ) & ( 13 ) & ( 14 ) + j072418.4 + 415914&1.552&-26.6&32.6&1.52&+0.0&51146&53312 - 1865 - 261 & & 2.31 & & & & lobal@xmath64 + j072831.6 + 402615&0.656&-28.0&32.2&0.57&-1.1&50774&54915&56654&6.85&9.73 & 0.09&0.06&ofelobal@xmath65 + j080901.3 + 275341&1.511&-27.7&31.9&0.40 & ... & 50851&52618 - 0930 - 135&55858 - 4457 - 0512 & 1.93 & 5.46 & & & hibal + j084044.4 + 363328&1.230&-27.2&31.7&0.39&-0.2&50838&52320 - 0864 - 149 & & 1.82 & & & & felobal + j091044.9 + 261253&2.920&-27.6&33.1&1.65&-0.5&51160&53415 - 2087 - 352 & & 1.58 & & & & hibal + j091328.2 + 394444&1.580&-27.4&32.1&0.65&-0.6&50094&52707 - 0937 - 569 & & 2.77 & & & & hibal + j093403.9 + 315331&2.419&-28.5&32.8&0.88&-0.2&50401&53386 - 1941 - 600 & & 2.39 & & & & hibal@xmath64 + j094602.2 + 274407&1.748&-27.9&32.4&0.74&@xmath66 - 1.5&50466&53385 - 1944 - 026 & & 2.91 & & & & hibal + j095707.3 + 235625&1.995&-27.4&34.1&2.64&-0.6&51172&53737 - 2298 - 409 & & 2.35 & & & & hibal + j103110.6 + 395322&1.082&-25.8&31.8&1.04&-0.2&50465&52998 - 1428 - 608 & & 3.33 & & & & lobal + j104459.5 + 365605&0.701&-26.2&32.2&1.30&-0.5&50486&53463 - 2090 - 329&55615 - 4635 - 0704 & 4.79 & 8.26 & & & felobal + j105427.1 + 253600&2.400&-28.0&32.6&0.91&-0.5&50486&53793 - 2357 - 227 & & 2.66 & & & &
lobal@xmath64 + j105528.8 + 312411&0.558&-23.9&31.7&1.68 & ... & 50597&53711 - 2026 - 085&54915 & 5.48&7.59&0.07&0.00&ofelobal@xmath65 + j120051.5 + 350831&1.700&-29.1&32.1&0.01&-0.8&50586&53469 - 2099 - 606&55648 - 4650 - 0052 & 2.93 & 5.14 & & & hibal + j121442.3 + 280329&0.698&-26.3&31.5&2.08&-0.8&50549&53823 - 2229 - 557 & & 5.28 & & & & felobal + j130425.5 + 421009&1.916&-28.7&32.1&1.62&+0.7&50495&53119 - 1458 - 579&55681 - 4704 - 0881 & 2.47 & 4.87 & & & hibal@xmath64 + j131213.5 + 231958&1.508&-27.4&33.3&1.30&-0.8&50585&54507 - 2651 - 380&56098 - 5977 - 0326 & 4.28 & 6.02 & & & hibal + j132422.5 + 245222&2.357&-27.4&32.8&1.32&-0.7&51009&54524 - 2664 - 537&56096 - 0186 - 0186 & 2.87 & 4.15 & & & hibal + j140800.4 + 345125&1.215&-26.3&32.0&1.01&-0.6&50486&53471 - 1839 - 377&55268 - 3855 - 0386 & 3.69 & 5.91&0.08&0.00&ofelobal@xmath65 + j140806.2 + 305448&0.842&-24.8&31.7&1.33&-0.7&50601&53795 - 2125 - 236&55276 - 3862 - 0124 & 4.75 & 6.95&0.07&0.00&ofelobal@xmath65 + j141334.4 + 421201&2.810&-28.3&33.5&1.69&-0.2&50586&52823 - 1347 - 539&56093 - 6059 - 0438 & 1.61 & 3.96 & & & hibal + j142013.0 + 253403&2.240&-27.9&32.1&0.49&-1.1&50596&53859 - 2127 - 011&56096 - 6015 - 0432 & 2.76 & 4.65 & & & hibal@xmath64 + j142703.6 + 270940&1.170&-25.6&32.0&1.27&-0.7&50616&53876 - 2134 - 256&56067 - 6018 - 0412 & 4.12 & 6.88 & & & felobal@xmath64 + j152314.4 + 375928&1.344&-26.5&31.9&0.78&-0.6&50245&53470 - 1400 - 450&56045 - 4979 - 0218 & 3.77 & 6.78 & & & hibal + j152350.4 + 391404&0.657&-26.3&31.6&0.61&-0.4&50507&52765 - 1293 - 234&56045 - 4979 - 0804 & 3.73 & 9.12&0.06&0.00&ofelobal@xmath65 + j160354.2 + 300208&2.026&-27.9&33.7&2.04&-0.6&50507&53496 - 1578 - 534&55707 - 5009 - 0569 & 2.71 & 4.71 & & & hibal + j164152.2 + 305851&2.000&-27.2&32.4&1.01&+0.5&51009&52781 - 1340 - 073&55832 - 5201 - 0914 & 1.62 & 4.40 & & & lobal + j165543.2 + 394519&1.747&-27.5&32.8&1.38&-0.2&51009&52079 - 0633 - 353&56098 - 6063 - 0438 & 1.07 & 5.08 & & & hibal@xmath64 | we present a study of the variability of broad absorption lines ( bals ) in a uniformly radio - selected sample of 28 bal quasars using the archival data from the first bright quasar survey ( fbqs ) and the sloan digital sky survey ( sdss ) as well as data we obtained ourselves , covering time scales of @xmath0 years in the quasar rest frame . to our surprise , 5 quasars exhibiting strong variations all belong to a special subclass of ` overlapping - trough ' iron low - ionization bal ( ofelobal ) quasars ; however , 4 other non - overlapping felobals ( non - ofelobals ) are invariant except in one case with a weak change in optical depth .
meanwhile , we also identify 6 typical variations of high - ionization and low - ionization bals in this bal quasar sample .
photoionization models suggest that ofelobals are formed in relatively dense ( @xmath1 @xmath2 ) outflows at distances from the continuum source that range from the subparsec scale to dozens of parsecs .
these formation conditions differ from those of non - ofelobals , which are likely produced by low - density gas located at distances of hundreds to thousands parsecs .
thus , ofelobals and non - ofelobals , i.e. , felobals with / without strong bal variations , may represent the bimodality of absorption ; the former are located in active galactic nucleus environments rather than in the host galaxy .
we suggest that a high density and a small distance are the necessary conditions for the creation of ofelobals .
as suggested in the literature , strong bal variability is possibly attributable to variability in the covering factor of the bal regions caused by clouds transiting across the line of sight rather than to variations in ionization . |
this paper is about _ leaky dynamical systems _ or dynamical systems with _
a generic setup consists of a triple @xmath0 where @xmath1 is the phase space of a map or flow denoted by @xmath2 , and @xmath3 is an open set .
we refer to @xmath4 as a _ closed system _ and @xmath5 as the hole through which mass is allowed to escape from the system .
more precisely , we follow trajectories in @xmath1 until they enter @xmath5 . once a point enters @xmath5 , it leaves the system forever , i.e. we stop considering it . holes can be large or small .
small holes are often used to model small ( unintended ) leaks in physical systems ; proximity of normalized surviving distributions to the physical measure of the closed system is a form of stability .
more generally , the study of @xmath0 can be viewed as the study of dynamics on _ non - invariant domains_. as an example of why such studies are relevant , consider the following .
it is well known that attractors are important because they capture the large - time behavior of dynamical systems , but invariant sets that are not attracting can substantially impact the qualitative behavior of a system as well : let @xmath6 be such a set , and @xmath7 a neighborhood of @xmath8
. then we may regard @xmath9 as the hole .
slow escape rates from such holes are known to impact the speed of correlation decay of the closed system .
escape dynamics have been studied by many authors .
we refer the reader to the part - review article @xcite , which contains many references , and will mention explicitly works that are closer to the present paper as we go along .
most previous works have focused on specific systems , such as anosov diffeomorphisms , interval and billiard maps . in this paper
, we seek a general understanding for as large a class of dynamical systems as we can .
specifically , we seek to relate escape rate to a dynamical invariant called _ pressure _ , which roughly speaking measures the discrepancy between metric entropy and sum of positive lyapunov exponents .
we now proceed to a discussion of what this paper is about . *
setting and questions * we begin with the simpler setting of a compact riemannian manifold @xmath1 without boundary and a diffeomorphism @xmath2 which is at least @xmath10 for some @xmath11 . in order to include applications to systems such as billiards , which are very important examples of dynamical systems of physical origin
, we also allow @xmath1 to be the union of a ( possibly open ) riemannian manifold and a singularity set @xmath12 , and @xmath2 to be piecewise smooth .
precise conditions on @xmath12 and the behavior of @xmath2 near it will be introduced in section 2 .
riemannian measure on @xmath1 ( or @xmath13 ) is denoted by @xmath14 throughout . unless otherwise stated , the hole @xmath5 is an arbitrary open set in @xmath1 .
let @xmath15 be a reference measure on @xmath1 .
we think of @xmath15 as the initial distribution of mass in the phase space before any escape takes place , and take the view that initial distributions related to @xmath14 are of particular physical interest .
notice that @xmath15 need not be @xmath2-invariant . indeed one can interpret the situation as follows : the escape of mass can begin before or after the closed system @xmath16 reaches a steady state . in the first case , @xmath15 is usually not invariant , and we assume it has a density with respect to @xmath14 . in the second case ,
we take @xmath15 to be an srb measure , which may be singular with respect to @xmath14 .
a basic quantity of interest is the _ escape rate _ ,
defined to be @xmath17 where @xmath18 when the limit exists . here
@xmath19 is the set of points which has not escaped by time @xmath20 . in general , the limit in ( [ eq : escape def ] ) may not exist , and we write @xmath21 and @xmath22 for the @xmath23 and @xmath24 of the quantity on the right hand side .
notice that while @xmath25 depends on @xmath15 , all initial distributions uniformly equivalent to @xmath15 have the same escape rate , i.e. if @xmath26 is a function with @xmath27 for some @xmath28 , then @xmath29 , and the same is true for @xmath21 and @xmath22
. for an @xmath2-invariant borel probability measure @xmath30 on @xmath1 , the _ pressure _ of @xmath30 , denoted @xmath31 , is defined to be @xmath32 where @xmath33 is the metric entropy of @xmath34 and @xmath35 is the the sum of the positive lyapunov exponents counted with multiplicity .
we will write @xmath36 where @xmath37 is a collection of invariant measures .
given an open system @xmath38 , we define the _ survivor set _ to be the @xmath2-invariant set @xmath39 .
is not invertible , we take @xmath40 in the definition of @xmath41 . ]
let @xmath42 denote the set of @xmath2-invariant borel probability measures supported on @xmath41 , and let @xmath43 be the subset of @xmath44 consisting of ergodic measures .
assuming @xmath25 is well defined , we say @xmath25 satisfies a _ variational principle _ if @xmath45 of interest also is whether the supremum in @xmath46 is attained , i.e. if there is a measure @xmath47 for which @xmath48 . obviously , one can also ask if @xmath49 for some @xmath30 without mentioning any variational principles .
the ideas in the last paragraph were suggested by a number of previously known results some of which are recalled below , but let us first summarize the questions to be addressed . this paper seeks to address for as large a class of dynamical systems as possible the following three questions for natural initial distributions @xmath15 : _ _ * * ( escape rate ) * is the escape rate @xmath50 well defined ? * * ( formula for escape rate ) * is @xmath51 for some @xmath52 ? + the same question can be posed for @xmath53 and @xmath54 . * * ( variational principle ) * does @xmath25 satisfy a variational principle ?
partial answers are given for very general dynamical systems , and complete answers for a more restricted class which includes many known examples .
a concrete application to the leaky periodic lorentz gas is mentioned explicitly .
* earlier works * @xcite consider a @xmath10 axiom a diffeomorphism @xmath55 of a compact riemannian manifold @xmath1 .
let @xmath6 be a basic set , and let @xmath56 .
then @xmath57 , and @xmath58 if and only if @xmath8 is an attractor .
this is the first result that systematically relates the escape of mass to pressure : in the case where @xmath8 is an axiom a attractor , no mass can escape from a neighborhood of @xmath8 , and @xmath59 ; for non - attracting basic sets such as horseshoes , mass escapes at exponential rates and @xmath60 .
the number @xmath61 has been shown to be equal to the _ topological pressure _ of @xmath2 with respect to the potential @xmath62 on @xmath8 ; see @xcite or @xcite for more detail .
the next result gives conditions under which the numerical value of @xmath61 is explicitly related to the rate of escape .
* ; * ? ? ?
* ; * ? ? ?
* theorem 4 ) on a closed set @xmath63 and @xmath64 on @xmath65 where @xmath66 , and @xmath67 on @xmath41 .
] let @xmath55 be a @xmath10 diffeomorphism of a compact riemannian manifold @xmath1 , and let @xmath3 be an open set .
we assume \(i ) @xmath41 is compact with @xmath68 , and \(ii ) @xmath69 is uniformly hyperbolic .
then @xmath70 is well defined and equals @xmath61 . in both of the settings above , @xmath71 , and @xmath72 for some @xmath52 .
( the latter follows from the continuity of @xmath73 and upper semicontinuity of @xmath74 ; see [ b ] ) .
thus for uniformly hyperbolic survivor sets @xmath41 with @xmath75 , * q1**q3 * have all been answered in the affirmative .
several works went beyond theorem 2 to give positive answers to * q1 * and * q2 * in a number of situations , including anosov diffeomorphisms with markov or small holes ( with no requirement on @xmath76 ) @xcite , uniformly expanding maps admitting markov partitions @xcite , piecewise expanding maps , and collet - eckmann maps of the interval with singularities @xcite .
* q3 * was partially addressed in @xcite : a variational principle was proved for an associated dynamical system , namely the symbolic dynamics of the original map ( but not for the map itself ) .
three sets of results are stated : * sect .
2.1 contains partial answers to * q3 * : lower bounds for @xmath53 are proved for very general dynamical systems ; no results on upper bounds are reported . *
2.2 and 2.3 provide complete answers to * q1**q3 * for systems admitting markov tower extensions with some additional conditions . * these results
are applied to the periodic lorentz gas with small holes ( theorem f ) .
our results in this subsection will assert , in essence , that _ for very general dynamical systems , @xmath77 for reasonable choices of @xmath78 . _ since @xmath79 decreases with @xmath78 , this inequality is not meaningful for @xmath78 too small . thus the selection of a suitable @xmath78 is an important part of the consideration .
we start with @xmath80 , the set of ergodic invariant measures supported on the survivor set @xmath41 . to obtain @xmath78 ,
restrictions will be placed on @xmath80 on account of i. the hole @xmath5 , \ii . the initial distribution @xmath15 , and \iii .
singularities of the map @xmath2 , if present .
we discuss these 3 types of restrictions separately .
the conditions we impose are admittedly motivated by our proofs , but the fact that they lead to a full variational principle for a large class of dynamical systems ( see sect . 2.3 ) suggests that these choices of @xmath78 are reasonable .
_ one should keep in mind that the escape rate is defined by @xmath17 when interpreting the inequality @xmath77 .
thus a lower bound for @xmath53 provides an upper bound of @xmath81 for the escape rate . in paragraphs
i and ii below , @xmath55 is a @xmath10 diffeomorphism ; systems with singularities are discussed in paragraph iii . throughout the paper
, @xmath82 denotes the ball of radius @xmath83 in @xmath1 centered at @xmath84 , and @xmath85 denotes the @xmath86-neighborhood of a set in @xmath1 . * i. restrictions on @xmath78 due to the hole @xmath5 * the following definition gives a sense of which @xmath87 we think impact the escape rate .
define @xmath88 notice that if @xmath87 has the property that for some @xmath89 , @xmath90 for all @xmath91 , then @xmath30 is in @xmath92 ( see sect .
4.2 , paragraph 4 ) .
the definition of @xmath92 can be relaxed in many ways ; in particular , it is not necessary for the entire ball @xmath93 to be in @xmath94 .
we mention one formulation , leaving the reader to contemplate others : given @xmath84 , let @xmath95 denote the local stable manifold of @xmath96 of radius @xmath86 .
we call an open set @xmath97 a _ @xmath98-neighborhood of @xmath96 _ if @xmath99 for every @xmath91 .
all of our results remain valid if * ii .
restrictions on @xmath78 due to the initial distribution @xmath15 * two types of initial distributions are considered . *
( a ) initial distributions with densities , possibly localized * let @xmath100 where @xmath101 is in @xmath102 .
for such an initial distribution , we consider @xmath103 * theorem a.
* _ let @xmath104 be as above . then _
\(i ) @xmath105 ; \(ii ) more generally , @xmath106 .
_ clearly , @xmath107 if @xmath108 for some @xmath28 ; thus ( ii ) reduces to ( i ) .
here we permit @xmath26 to vanish on parts of @xmath1 provided it is measurable with ess inf@xmath109 on an open set of @xmath1 .
we do not claim that the restrictions imposed on @xmath110 are necessary , but if the support of @xmath26 is localized in the phase space , invariant measures supported elsewhere are clearly irrelevant since they can not be seen " by the initial distribution @xmath111 . *
( b ) srb measures as initial distributions * in ( a ) , @xmath112 is not necessarily an invariant measure .
if , however , a steady state is reached before the leak begins , then it would be natural to take @xmath15 to be an srb measure @xmath113 , as we now do . for simplicity
, we assume @xmath113 has no zero lyapunov exponents . the challenge here is to identify a class of invariant measures @xmath114 that can be seen " by the srb measure @xmath113 , which is often singular .
we call @xmath115 a _ @xmath113-hyperbolic product set _ if the following hold .
1 . @xmath116 where @xmath117 and @xmath118 are two sets of relatively open local unstable and stable manifolds such that each @xmath119 intersects every @xmath120 in precisely one point .
in addition , there exist constants @xmath121 , @xmath122 such that @xmath123 where diam@xmath124 denotes the diameter of the unstable or stable manifold .
2 . @xmath125 for every relatively open @xmath126 .
3 . there exists a constant @xmath127 such that for @xmath113-a.e .
@xmath119 , the conditional probability of @xmath113 on @xmath128 has density @xmath129 .
we remark that ( w.3 ) is a general property of srb measures [ ly ] ; we have listed it separately only for emphasis .
define @xmath130 * theorem b. * _ under the conditions above , @xmath131 . _ _ remark 1 .
_ observe that if @xmath2 has an axiom a attractor @xmath8 and @xmath113 is the srb measure on the attractor , then @xmath114 imposes no restriction whatsoever on @xmath87 , i.e. @xmath132 . _ remark 2 .
_ in the case where the pushforward of lebesgue measure @xmath14 tends to @xmath113 , one might be tempted to conclude that @xmath133 .
this is _ not necessarily true _ , and the reason is as follows
: suppose @xmath2 has a lebesgue measure zero invariant set @xmath8 ( such as a horseshoe ) away from the support of the srb measure .
the rate at which points escape from a neighborhood of @xmath8 will be reflected in @xmath70 but not in @xmath134 ; this can easily lead to @xmath135 .
restrictions on @xmath78 due to the singularities of the map @xmath2 * we state here a version of our results that can be applied to planar billiards ; see theorem f below . following @xcite , we let @xmath136 be an open smooth ( at least @xmath137 ) finite dimensional riemannian manifold , and assume that @xmath138 is a compact metric space of finite capacity , such that @xmath139 where @xmath140 is the minimum cardinality of a covering of @xmath1 by open balls of radius @xmath83 . for billiards with corners
the set @xmath141 is technically not a manifold with boundary but a union of such glued together along some boundaries . ] where @xmath141 denotes the closure of @xmath136 .
let @xmath142 be the radius of injectivity of the exponential map exp@xmath143 .
we assume that there exist constants @xmath144 such that for each @xmath145 such that @xmath146 and @xmath147 , we have @xmath148 let @xmath149 be an open subset of @xmath136 and let @xmath150 be a mapping which is a @xmath151 diffeomorphism of @xmath149 onto its image .
let @xmath152 .
we think of @xmath12 as the singularity set of @xmath2 .
we assume that there exist constants @xmath153 such that for all @xmath154 , @xmath155 let @xmath156 denote the induced map on @xmath157 wherever it is defined .
we assume that there exists @xmath158 such that if @xmath154 , @xmath159 and @xmath160 is well - defined , then @xmath161 notice that for billiards with finite horizon , @xmath162 and @xmath163 ( see @xcite ) . in what follows
, we will assume without loss of generality that @xmath164 .
into such a system we introduce a hole @xmath3 . with regard to the choice of @xmath78 , in addition to the considerations above
, we must also restrict to invariant measures that respect the singularities ( see [ ks ] ) .
define @xmath165 * theorem c. * _ let @xmath104 be as above . then _ 1 . for @xmath166 , @xmath167 , and 2
if @xmath2 has an srb measure @xmath113 with no zero lyapunov exponents , then + @xmath168 we finish with the following . * remarks on upper bounds and the attractor case : * no general results are known for upper bounds on @xmath25 , not even for @xmath169 .
consider the special case where @xmath170 is an attractor .
assume there is a neighborhood @xmath97 of @xmath41 such that @xmath171 and @xmath172 .
let @xmath173 and @xmath169 , so that @xmath174 by definition .
since @xmath175 for all @xmath87 @xcite , showing that @xmath176 in this case is equivalent to proving @xmath177 .
the latter is known to be false in general , an example being the figure 8 attractor ( see fig . 1 ) ,
so one must rephrase the question to include some notion of typicality " .
still , @xmath178 means either @xmath179 or @xmath30 is an srb measure @xcite , and whether attractors with nonuniform expansion admit srb measures is well known to be a very difficult question ; see e.g. @xcite .
since any result on upper bounds for @xmath54 must include this attractor case , we conclude that in complete generality the question for upper bounds for @xmath54 ( and lower bounds for escape rates ) is intractable at the present time .
we will , however , identify a large class of dynamical systems for which @xmath180 for some @xmath78 .
this is the content of sects .
2.2 and 2.3 .
fig . 1 .
figure 8 attractor .
the only invariant measure is @xmath181 where @xmath182 is the saddle point . in this section ,
we assert for a class of dynamical systems the existence of @xmath183 the pressure of which is equal to @xmath25 , thereby answering * q1 * and * q2 * in the affirmative .
let @xmath55 be a @xmath184 diffeomorphism or a piecewise smooth diffeomorphism as in the setting of theorem c , and fix a hole @xmath3 .
we assume @xmath4 has a _ markov tower extension _ @xmath185 ; @xmath185 has an _ exponential tail _ ; @xmath185 _ respects the hole _ @xmath5 ; the transfer operator on the tower with holes " has a _
spectral gap_. while * ( a.1 ) * and * ( a.2 ) * are by now quite standard , and * ( a.3 ) * and * ( a.4 ) * have also appeared elsewhere , it will take a few pages to make precise this entire formal setting ; we postpone that to sect .
[ tower review ] .
let @xmath113 denote the ( unique ) ergodic srb measure on @xmath186 where @xmath187 is the projection , and let @xmath188 be the leading eigenvalue of the transfer operator on the tower with holes .
we will use the following notation : let @xmath189 denote a probability measure on @xmath1 . for @xmath190 ,
let @xmath191 denote the normalized surviving distribution at time @xmath20 , i.e. @xmath192 , assuming @xmath193 .
we call a measure @xmath15 conditionally invariant with eigenvalue @xmath194 if @xmath15 is supported on @xmath195 and @xmath196 .
* theorem d. * _ assume @xmath104 satisfies * ( a.1)* * ( a.4)*. then _ 1 .
@xmath134 is well defined and equals @xmath197 ; 2 .
@xmath198 converges weakly to a conditionally invariant measure @xmath199 with eigenvalue @xmath200 ; 3 .
there exists @xmath201 such that @xmath202 4 .
@xmath203 is defined by @xmath204 in addition , @xmath203 enjoys exponential decay of correlations on hlder observables .
our construction of @xmath203 generalizes that in @xcite , which assume the maps in question admit finite markov partitions .
see @xcite for the first generalization in this direction regarding pressure for one - dimensional maps with holes .
parts ( a ) and ( b ) of theorem d are also known for the periodic lorentz gas @xcite .
we assert here that these results hold generally for _ any _ dynamical system admitting a tower with the stated conditions . combining the results of the previous two sections , we are able to state a full variational principle ( answering * q1**q3 * in section 1 ) for maps admitting towers with a spectral gap as described in sect . [ tower results ] .
let @xmath6 be the reference hyperbolic product set which forms the base of the tower @xmath205 .
* theorem e. * _ assume @xmath104 satisfies * ( a.1)* * ( a.4 ) * , and let @xmath203 be as in theorem d. _ 1 . if @xmath206 where @xmath207 on a neighborhood of @xmath8 , then @xmath208 and @xmath209 2 .
if @xmath8 is contained in a @xmath113-hyperbolic product set , then @xmath210 and @xmath211 to our knowledge the condition in part ( b ) of theorem e can be arranged in all known tower constructions .
* remark on results for tower maps . *
we will , as an intermediate step to proving theorems d and e , prove the corresponding results for tower maps with markov holes .
these results are stated as theorems 4 and 5 in sect .
an illustrative example : the 2d periodic lorentz gas * we conclude this section by stating an application of our results to a concrete example .
the setting here is as in @xcite : let @xmath55 be a billiard map associated with a two dimensional periodic lorentz gas with finite horizon whose scatterers are bounded by @xmath212 curves with strictly positive curvature .
the holes we introduce into @xmath1 are derived from two types of holes in the billiard table @xmath213 .
we say @xmath214 is a hole of type i if @xmath215 is an open segment of an arc in the boundary of one of the scatterers in @xmath213 .
we say @xmath215 is a hole of type ii if it is an open convex set in @xmath213 whose closure is disjoint from any of the scatterers .
the hole @xmath214 induces a hole @xmath216 which we also call a hole of type i or type ii .
see @xcite for more general holes and details on the geometry they induce in @xmath1 .
* theorem f. * _ let @xmath2 be the billiard map in the last paragraph .
let @xmath217 be a hole of type i or type ii , and assume it is small enough in the sense of @xcite . then _
\(a ) @xmath218 ; \(b ) there exists @xmath201 such that @xmath219 .
theorem f is an immediate consequence of theorems d and e together with @xcite : in @xcite , towers with exponential tails respecting arbitrary holes of types i and ii are constructed , and for small enough holes the spectral gap property is guaranteed .
thus the conditions for theorem d are satisfied ; however , @xcite does not address variational principles or pressure so that theorem d , parts ( c ) and ( d ) , as well as theorem f are new results for this class of billiards . for the lorentz gas , @xmath206 where @xmath220 so that we are in the setting of theorem e(a ) ; however , @xmath221 only when @xmath222 so that @xmath107 since the set @xmath223 does not contain any invariant sets by the finite horizon condition and so can not contain the support of any invariant measure .
in this section , we first give the ideas common to the proofs of theorems a
let @xmath2 be the mapping in question , let @xmath15 be the reference measure ( i.e. @xmath224 in theorem a , @xmath225 in theorem b , and so on ) , and let @xmath78 be the relevant set of ergodic invariant measures with respect to which the pressure term is defined ( i.e. @xmath226 in theorem a , and so on ) .
generic " notation is used throughout sect .
[ proof a - c ] . if @xmath227 , then @xmath228 and the theorem is vacuously true .
consider @xmath229 .
leaving precision for later , our proof will proceed as follows : for @xmath230 , we introduce dynamical balls in @xmath231 of the form @xmath232 where @xmath233 and @xmath234 is a suitable function to be specified ( think of it as @xmath235 for the moment )
. we will prove _ i. relation to entropy : _
@xmath236 .
volume estimate : _
@xmath237 where @xmath238 is the sum of positive lyapunov exponents for @xmath30-a.e .
@xmath96 . from estimate
i , we deduce that @xmath231 contains @xmath239 disjoint sets of the type @xmath240 .
this together with estimate ii gives @xmath241 taking @xmath242 , dividing by @xmath20 and letting @xmath243 , gives @xmath244 , which is what we need .
we now proceed to make these ideas precise .
* i. relation to entropy . * for this part we cite the following very general result . [ prop : local entropy ] let @xmath245 be a measurable transformation of a compact metric space of finite capacity , and let @xmath246 be an ergodic invariant measure for @xmath247 .
let @xmath248 be a family of functions satisfying @xmath249 and @xmath250 , and define @xmath251 .
then for @xmath246-a.e .
@xmath96 , @xmath252 proposition [ prop : local entropy ] follows from ( * ? ? ?
* lemma 2 ) and ( * ? ? ?
* ; * ? ? ?
* main theorem ) .
note that although @xcite is phrased in terms of a continuous map , the proof does not use this fact . in the proofs of theorems
a c , proposition [ prop : local entropy ] will be applied with @xmath253 , @xmath254 and @xmath255 ( @xmath256 in theorems a and b ) .
observe that intersecting @xmath257 with @xmath231 does not affect its @xmath30-measure since @xmath30 is supported on the survivor set . from @xmath258 , we have @xmath259 so our @xmath260 satisfies the hypotheses of proposition [ prop : local entropy ] . *
ii . volume estimate .
* let @xmath261 . continuing to let @xmath15 denote the initial distribution and @xmath229
, we state the following desired volume estimate : [ prop : volume est ] there exists a measurable set @xmath262 with @xmath263 such that for @xmath30-a.e .
@xmath264 , @xmath265 * proof of theorems a
c assuming proposition [ prop : volume est ] : * let @xmath229 be given .
we fix @xmath266 , and let @xmath267 where @xmath268 is as in proposition [ prop : volume est ] . using propositions [ prop : local entropy ] and [ prop : volume est ] , we may choose first @xmath91 sufficiently small , and then @xmath269 sufficiently large and a measurable set @xmath270 with @xmath271 such that for every @xmath272 , 1 .
@xmath273 for all @xmath274 ; 2 .
@xmath275 for all @xmath274 . for @xmath274 ,
let @xmath276 be a maximal set of points such that @xmath277 whenever @xmath278 , @xmath279 .
by the maximality of @xmath280 , for every @xmath281 , there exists @xmath282 such that @xmath283 .
we will show momentarily that @xmath284 .
this will imply @xmath285 , and hence @xmath286 by ( i ) .
to show @xmath284 , it suffices to show @xmath287 , since @xmath288 .
now @xmath283 means there exists @xmath289 such that @xmath290 and @xmath291 for all @xmath292 .
thus the assertion above boils down to the following lemma .
[ lem : distance_estimates ] for any @xmath293 , if there exists @xmath289 with @xmath294 and @xmath295 , then @xmath296 .
it suffices to show @xmath297 , for that will imply @xmath298 , proving the lemma .
observe that @xmath299 the last inequality following from @xmath300 .
altogether , this gives @xmath301 .
to finish , consider the following two cases : + case 1 : @xmath302 . with @xmath303
, @xmath304 is automatically @xmath305 since it is @xmath306 .
+ case 2 : @xmath307 . in this case @xmath308 . for each @xmath309
, we have @xmath310 by definition .
since the @xmath311 are disjoint , we may estimate @xmath312 by @xmath313 this yields @xmath314 the theorem is proved since @xmath315 was chosen arbitrarily .
@xmath316 to complete the proofs of theorems a c , it remains only to prove the volume estimate in proposition [ prop : volume est ] .
in this section we prove proposition [ prop : volume est ] in the various settings of interest . the basic argument , which treats the case @xmath256 , @xmath317 , and @xmath318 is presented in sect .
proofs of other cases in theorems a c are presented as modifications of this one .
we consider here the most basic setup , namely where @xmath319 , @xmath169 , and @xmath318 ( as defined ) , and give a proof of proposition [ prop : volume est ] . _
_ from the pointwise nature of the result and the fact that the quantity on the left of ( [ vol ] ) increases as @xmath320 , it suffices to show that given @xmath321 , for @xmath30-a.e .
@xmath96 and arbitrarily small @xmath91 , there exists @xmath322 such that @xmath323 here , @xmath324 ; remember that @xmath325 is a dynamical ball _ in _ @xmath231 ( and not in @xmath1 ) .
such an object is cumbersome to work with since it involves both the dynamics and the hole . to remove the hole from consideration ,
we introduce @xmath326 by definition of @xmath92 , for any @xmath327 and @xmath30-a.e .
@xmath96 , @xmath328 for small enough @xmath86 .
thus it suffices to prove , for a suitably chosen @xmath329 and arbitrarily small @xmath91 , @xmath330 this is what we will do .
our strategy is to make these volume estimates in lyapunov charts and pass them back to the manifold . _
lyapunov charts and hyperbolic estimates .
_ let @xmath331 be the distinct lyapunov exponents of @xmath332 , with multiplicities @xmath333 respectively , and let @xmath334 be the subspace of @xmath335 corresponding to @xmath336 .
for each @xmath337 , we let @xmath338 denote the ball of radius @xmath83 centered at @xmath339 in @xmath340 , and let @xmath341 .
we recall below the following facts about lyapunov charts , following the exposition in @xcite .
* ; * ? ? ?
3.1 ) [ prop : lyapunov ] let @xmath342 be fixed .
then there is a measurable set @xmath343 , @xmath344 , a measurable function @xmath345 satisfying @xmath346 , and a family of charts @xmath347 with the following properties : * \(i ) @xmath348 ; + \(ii ) @xmath349 ; + \(iii ) for all @xmath350 , @xmath351 where @xmath63 is a constant depending only on the dimension of @xmath1 . *
let @xmath352 be defined where it makes sense .
then + \(i ) @xmath353 for @xmath354 ; + \(ii ) lip@xmath355 ; + \(iii ) lip@xmath356 .
the following notation is used : let @xmath357 where @xmath358 and @xmath359 .
we will estimate the volume of the sets in question by looking at slices parallel to @xmath360 , and will do so in lyapunov charts .
let @xmath361 and @xmath362 be the subspaces in the charts corresponding to @xmath360 and @xmath363 , and let @xmath364 and @xmath365 denote disks of radius @xmath83 centered at @xmath339 in @xmath361 and @xmath362 respectively .
we will work with compositions of chart maps , writing @xmath366 and study graph transforms by @xmath367 of functions from @xmath364 to @xmath365 .
the precise assertions are as follows : \(a ) for all @xmath327 sufficiently small , there exist @xmath368 small enough and a chart system ( with @xmath315 as in proposition 4.1 ) such that the following holds for @xmath30-a.e .
@xmath96 : let @xmath369 , and let @xmath370 be a @xmath371 function with @xmath372 and @xmath373 . then for @xmath374 , there exists @xmath375 defined on exponentially shrinking domains and with @xmath376 for all @xmath337 such that inductively @xmath377 that is to say , if @xmath378 is the graph transform by @xmath379 , then for each @xmath337 , @xmath380 .
\(b ) for @xmath381 and @xmath382 graph@xmath383 , let @xmath384 denote the tangent space to the graph of @xmath385 at @xmath386 .
if @xmath315 and @xmath215 in ( a ) are small enough , then for @xmath382 graph@xmath387 such that @xmath388 for all @xmath389 , @xmath390| < e^{n(\lambda^+ + 3k \delta)}\ ] ] where @xmath391 and @xmath392dim@xmath393 .
notice first that with @xmath394 , we are assured that @xmath395 lies in the chart at @xmath396 ; this is because @xmath397 ; see proposition 4.1 . since most of the other assertions in ( a ) and ( b ) follow from standard ( uniformly hyperbolic ) graph transform estimates
, we will only sketch the arguments for a few key points .
( a version of these estimates can be found in ( * ? ? ?
* ; * ? ? ?
* sect . 3.1 ) ; see also ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
b ) for similar results . )
the overflowing property " of the graph transforms can be justified as follows .
consider first the case where @xmath398 . by proposition 4.1(b)(i ) , @xmath399 for @xmath400 . by proposition 4.1(b)(iii ) together with chart size , we have , for all @xmath401 , @xmath402 this gives @xmath403 for @xmath404 with a small enough component in @xmath362 .
thus with @xmath315 and @xmath215 sufficiently small relative to @xmath329 , the overflowing property is assured from step to step for @xmath405 with @xmath398 . for graphs that do not pass through @xmath339
, we pivot them at @xmath406graph@xmath387 where @xmath407 is the stable manifold of @xmath96 in its chart . since @xmath408 where @xmath409 , movements of @xmath410 in the @xmath361-direction are negligible assuming @xmath411 .
the assertion in ( b ) is proved similarly : we view @xmath412 as a product of determinants . at each step , @xmath413 , and we may assume that approximations of the type in the last paragraph increase the error by a factor @xmath414 .
_ iii . completing the proof .
_ putting assertions ( a ) and ( b ) in ii together , we arrive at the following : define @xmath415 we foliate @xmath416 with planes @xmath417 parallel to @xmath418 , and view them as graphs of constant functions . by the overflowing property of the graph
transform at each step , @xmath419 for all @xmath420 is equivalent to @xmath421 . pulling back @xmath422
, we use the bound in assertion ( b ) to estimate the area of @xmath423 .
we then integrate over @xmath424 to obtain @xmath425 where @xmath426dim@xmath427 .
we now return to the argument outlined at the beginning of the proof .
let @xmath321 be given .
assuming always @xmath411 , we now take it small enough that @xmath428 , and let @xmath315 be small enough ( with respect to @xmath329 ) for assertions ( a ) and ( b ) to hold in the chart system @xmath429 associated with @xmath315 . for @xmath30-typical @xmath96 , we consider @xmath86 small enough that @xmath430 for all @xmath431 . choosing @xmath432 where @xmath63 is as in proposition 4.1(a)(iii )
, we define @xmath433 in the chart at @xmath96 as above , and observe that @xmath434 .
to finish , it remains to pass the estimate in ( [ bound2 ] ) back to @xmath1 .
proposition 4.1(a)(iii ) gives a bound on the jacobian of @xmath435 , allowing us to conclude @xmath436 for some constant @xmath437 depending on @xmath438 and @xmath329 .
@xmath316 we now explain how each of the other results in theorems a
c is deduced from the proof in sect .
* 1 . the @xmath98-neighborhood condition ( o ) :
* we continue to assume @xmath319 and @xmath169 . to relax the condition from the original definition of @xmath92 in sect .
[ upper bound ] to the one given by * ( o ) * , the proof in sect .
4.1 is modified as follows : given @xmath439 , we fix @xmath440 , a chart system @xmath429 , and a @xmath30-typical @xmath84 .
let @xmath97 , a @xmath98-neighborhood , and @xmath86 be such that @xmath441 for all @xmath337 .
we need to show @xmath442 .
let @xmath432 .
the following notation is used : for @xmath443 and small @xmath444 , let @xmath445 ; if @xmath446 are the coordinates of @xmath386 with respect to @xmath361 and @xmath362 , we write @xmath447 , and so on . to define the analog of @xmath433 in sect .
4.1 , let @xmath448 be sufficiently close to @xmath96 , let @xmath449 , and let @xmath450 be small enough that @xmath451 . define @xmath452 @xmath453 since @xmath454 , @xmath455 as @xmath456 .
it is straightforward to check that modulo a constant , leb@xmath457 is bounded below by the quantity on the right side of ( [ bound2 ] ) , and that @xmath458 .
@xmath316 in the settings below , we will revert back to @xmath92 as defined , leaving it to the reader to extend the proof to include the condition * ( o ) * if they so choose .
* 2 . initial distributions with densities : * continuing to assume @xmath256
, we let @xmath112 for some @xmath166 .
let @xmath459 , and let @xmath460 and @xmath461 have the meaning in the definition of @xmath110 .
observe that for @xmath30-a.e .
@xmath462 and small enough @xmath86 , one has @xmath463 .
an argument identical to that in sect .
4.1 proves proposition [ prop : volume est ] with @xmath464 . @xmath316 * 3 .
srb measures as initial distributions : * continuing to assume @xmath256 , we let @xmath225 as in theorem b. given @xmath465 , we fix a @xmath113 hyperbolic product set @xmath116 with @xmath466 , and show that the volume estimate for @xmath467 in sect .
4.1 holds for @xmath30-a.e .
@xmath468 .
let @xmath469 be a @xmath30-typical point , where @xmath470 and @xmath471 .
note that due to the uniform contraction of @xmath472 and @xmath473 required by ( w.1 ) of sect .
[ upper bound].b , @xmath96 can have no zero lyapunov exponents .
let @xmath474 denote the image of the local unstable manifold through @xmath96 in its chart . since local unstable manifolds are unique ( up to size ) , @xmath475 , which has the dimension of @xmath476 and is tangent to it at @xmath339 .
( since no zero lyapunov exponents is an assumption for theorem b , we have @xmath476 instead of @xmath361 . ) by conditions ( w.1 ) and ( w.2 ) , for all small enough @xmath477 , there exists @xmath478 such that ( i ) @xmath479 and ( ii ) for every @xmath480 , @xmath481 is the graph of a function from @xmath482 to @xmath365 with the properties of @xmath405 in sect .
define @xmath483 with @xmath83 small enough relative to @xmath86 , clearly @xmath484 . to estimate the measure of this set
, it is more convenient to bring @xmath113 to the chart ( instead of doing it on @xmath1 ) : let @xmath485 be the measure @xmath486 restricted to @xmath487 . by ( i ) above together with ( w.3 ) , @xmath488 .
we disintegrate @xmath485 into conditional probability measures on the leaves @xmath489 , letting @xmath490 denote the measure in the transverse direction . to estimate the @xmath485-measure of @xmath491 , we do it one @xmath492-leaf at a time , integrating with respect to @xmath490 afterwards .
condition ( w.3 ) ensures uniform lower bounds of the type in ( [ bound2 ] ) for @xmath490-almost all leaves .
@xmath316 * 4 . maps with singularities : * we discuss the case @xmath169 , leaving the others to the reader .
let @xmath493 , and observe the following lemma .
[ lem : approach ] let @xmath494 for all @xmath495 . then for any fixed @xmath327 , @xmath496 .
this follows from the simple estimate , @xmath497 = \sum_{i\ge 0 } \nu [ n_{{\varepsilon}e^{-\gamma i}}({\mathcal{s } } ) ] \le \sum_{i \ge 0 } c { \varepsilon}^\alpha e^{-\gamma \alpha i } \le c ' { \varepsilon}^\alpha\ , \ ] ] the first inequality coming from the definition of @xmath498 .
( see also ( * ? ? ?
* part i , lemma 3.1 ) . )
this means that for @xmath499 , we again have @xmath500 , for @xmath501 . continuing to follow the proof in sect .
4.1 , we note that the definition of @xmath498 together with implies that @xmath502 where @xmath503 , so lyapunov exponents are well defined @xmath30-a.e .
in addition , the lyapunov charts described in proposition 4.1 exist for this class of maps with some modifications due to the presence of singularities .
observe first that @xmath504 , @xmath505 and @xmath506 of proposition 4.1 hold as stated since these quantities depend only on @xmath507 at a typical point @xmath96 ( see ( * ? ? ?
* part i , theorem 2.2 ) ) .
, one can just as easily split the tangent space into @xmath508 , one for each lyapunov exponent , to obtain proposition 4.1@xmath506 using an argument identical to that in ( * ? ? ?
* ; * ? ? ?
3.1 ) . ] the other items of proposition [ prop : lyapunov ] are modified as follows .
fix @xmath315 as in proposition [ prop : lyapunov ] .
then there exist a set @xmath509 with @xmath344 and a measurable function @xmath510 , with @xmath511 , such that for all @xmath91 sufficiently small , the charts @xmath435 are defined on @xmath512 , where @xmath513 is the exponent from , and satisfy * @xmath514 for all @xmath515 , @xmath516 where @xmath63 is a constant depending only on the dimension of @xmath1 and @xmath517 from .
* let @xmath352 be defined where it makes sense .
then + @xmath518 lip@xmath519 ; + @xmath514 lip@xmath520 .
although the construction of these charts is similar to that found in @xcite , we include the necessary arguments in the appendix since the statements we need are somewhat different from those found in the literature . with the charts @xmath429 in place , the proof follows a similar line to that given in section 4.1 , with slight modifications due to the singularities .
for example , assertion ( a ) is no longer a uniform statement for all @xmath521 ; rather , we need to choose @xmath522 , but only after @xmath86 is fixed depending on the rate of approach of @xmath96 to the singularities .
we state precisely these changes below .
fix @xmath523 and choose @xmath411 such that @xmath524 . using lemma [ lem : approach ] , we choose @xmath91 such that @xmath525 .
next we choose @xmath526 with @xmath394 , so that there exists a chart system @xmath527 with the modified properties as listed in @xmath528-@xmath529 above .
note that @xmath530 .
we now choose @xmath531 and prove the estimate .
note that @xmath532 . finally , choosing @xmath533 guarantees that the assertions ( a ) and ( b ) of sect .
4.1 hold along the orbit of @xmath96 with @xmath329 replaced by @xmath534 , for then @xmath535 lies in the chart at @xmath396 by definition of @xmath536 and choice of @xmath83 .
in particular , @xmath537 is defined on @xmath535 .
we shrink @xmath83 further if necessary so that @xmath538 and define @xmath539 as in sect .
4.1 . then by item @xmath528 above , @xmath540 and the rest of the proof follows line by line with only minor changes to constants
for example , has the factor @xmath541 as indicated above .
this proves proposition [ prop : volume est ] for all @xmath499 .
but since @xmath321 was chosen arbitrarily , by lemma [ lem : approach ] we conclude that proposition [ prop : volume est ] holds for @xmath30-a.e .
this section is exclusively about escape dynamics on towers .
[ tower review ] reviews basic facts and notation for towers making precise * ( a.1)**(a.4 ) * in sect . [ tower results ] .
in sect . [ tower theorems ] we formulate results analogous to theorems d and e for _ towers with markov holes_. proofs are given in sects . 5.3 and 5.4 . * i. closed systems ( without holes ) * let @xmath55 is a ( piecewise ) @xmath10 diffeomorphism .
the material below is taken from @xcite .
we recall only essential definitions , referring the reader to @xcite for detail .
* generalized horseshoes : * the idea of a _ generalized horseshoe with infinitely many branches and variable return times _ , denoted @xmath542 , is as follows : @xmath6 is a compact subset with a hyperbolic product structure , _
i.e. _ , @xmath543 where @xmath544 and @xmath545 are continuous families of local stable and unstable manifolds , and @xmath546 for every @xmath119 where @xmath547 is the riemannian measure on the unstable manifold @xmath128 .
we say @xmath548 is an @xmath549-subset of @xmath8 if @xmath550 for some @xmath551 , and @xmath552-subsets are defined similarly .
modulo a set the restriction of which to each @xmath119 has @xmath547-measure zero , @xmath8 is a countable disjoint union of ( closed ) @xmath549-subsets @xmath553 with the property that for each @xmath554 , there exists @xmath555 such that @xmath556 is a @xmath552-subset of @xmath8 . the function @xmath557 given by @xmath558
is called the _ return time function _ to @xmath8 .
the definition of a generalized horseshoe includes conditions on hyperbolicity formulated as * ( p1)**(p5 ) * in @xcite .
we will omit them and focus instead on the estimates derived from these conditions that we will need .
let @xmath559 and @xmath560 denote respectively the elements of @xmath544 and @xmath545 containing @xmath96 .
* there is a _ separation time _
@xmath561 with the property that ( i ) @xmath562 for @xmath563 ; ( ii ) for @xmath564 , @xmath565 , and ( iii ) for @xmath566 , @xmath567 , @xmath568 , @xmath569 .
* there are constants @xmath121 and @xmath570 , related to the hyperbolicity and distortion of @xmath2 , such that if @xmath571 , then @xmath572 for all @xmath230 .
the following facts about the jacobian in the unstable direction are useful . for @xmath573 , the holonomy map @xmath574 is obtained by sliding along stable curves , i.e. @xmath575 . fix an arbitrary leaf @xmath576 .
we let @xmath577 be the unique point in @xmath578 , and define @xmath579 , where @xmath580 is the unstable jacobian of @xmath2 .
this function is used to define a family of reference measures @xmath581 , where @xmath582 is the measure on @xmath128 whose density with respect to @xmath547 is @xmath583 . for @xmath584 , let @xmath585 be such that @xmath586 .
we define @xmath587 , the jacobian of @xmath588 with respect to the measures @xmath582 and @xmath589 .
it is convenient in this section to follow the notation in [ y3 ] , some of which conflicts , however , with earlier notation .
for example , @xmath15 in the last paragraph is not intended to signify any relation to initial distributions in escape dynamics , and @xmath590 below is not related to the same notation in sect .
2.1 , paragraph iii .
we do not believe this will lead to problems as the contexts are quite different .
( ( * ? ? ? * ; * ? ? ?
* lemma 1 ) ) [ lemma : jacobian ] ( 1 ) for all @xmath128 , @xmath591 , @xmath592 . * for each @xmath119 and @xmath593 , @xmath594 for all @xmath571 . * @xmath595 ( depending on @xmath596 and @xmath485 ) such that for each @xmath119 , @xmath597 and all @xmath598 , @xmath599 * @xmath600 and @xmath601 on each @xmath119 .
we say @xmath542 has _ exponential return times _ if there exist @xmath602 and @xmath603 such that for all @xmath119 , @xmath604 for all @xmath230 . this property ( in fact
, integrability of @xmath605 is sufficient ) plus the requirement that g.c.d.@xmath606 guarantees that @xmath2 has a unique srb measure @xmath113 with @xmath607 ( ( * ? ? ?
* ; * ? ? ?
* theorem 1 ) ) .
* `` hyperbolic '' markov towers : * given @xmath2 with a generalized horseshoe @xmath542 , it is shown in @xcite that one can associate a _ markov extension _ @xmath608 which focuses on the return dynamics to @xmath8 . the set @xmath205 is the disjoint union @xmath609 where @xmath610 , the @xmath611 level of the tower , is defined to be @xmath612 , and @xmath613 is defined by @xmath614 for @xmath615 and @xmath616 when @xmath617 ; that is to say , @xmath613 maps @xmath618 successively up the tower until the return time for @xmath96 is reached .
a projection @xmath187 with @xmath619 is uniquely defined assuming the natural identification of @xmath620 with @xmath8 . for notational simplicity
, we will often refer to a point in @xmath205 as @xmath96 when the level @xmath621 is made clear by context .
the separation function @xmath622 above defines a countable partition @xmath623 on @xmath205 : for @xmath624 , @xmath625 .
it is easy to see that @xmath623 is a markov partition for @xmath613 with @xmath620 as a single element .
let @xmath626 .
note that @xmath627 maps @xmath628 bijectively onto a @xmath552-subset of @xmath620 , and if we rename the collection @xmath629 as @xmath630 , then @xmath630 is a countable collection of closed subsets of @xmath620 the @xmath631-images of which are precisely the @xmath632 in the paragraph on generalized horseshoes . stable and unstable sets for @xmath633 are defined as follows : let @xmath634 and @xmath635 be the stable and unstable families defining the hyperbolic product set @xmath636 .
we say @xmath637 is an _
unstable set _ of @xmath633 if @xmath638 for some @xmath639 .
since there can be no ambiguity , we will use @xmath640 to denote the set of all such @xmath641 , and let @xmath642 . _ stable sets _ of @xmath633 and @xmath643
are defined similarly .
two reference measures @xmath644 and @xmath645 are defined on @xmath646 as follows : on @xmath620 , identifying @xmath647 with @xmath648 for @xmath649 , @xmath644 is simply @xmath650 and @xmath645 is @xmath589 .
once these measures are defined on @xmath647 , there is exactly one way to extend them to @xmath651 so that if @xmath652 and @xmath653 denote the jacobians of @xmath613 on unstable sets with respect to @xmath644 and @xmath645 respectively , then @xmath654 on @xmath655 .
notice also that if we extend @xmath656 to @xmath657 by @xmath658 , then @xmath659 on all @xmath646 .
* quotient expanding " towers : * associated with @xmath660 is a quotient tower @xmath661 obtained by collapsing stable sets to points , i.e. , @xmath662 where for @xmath663 , @xmath664 if and only if @xmath665 for some @xmath666 .
let @xmath667 be the projection defined by @xmath668 .
we will use the notation @xmath669 , and so on .
lemma [ lemma : jacobian](1 ) and ( 2 ) together imply that there is a natural measure @xmath670 on @xmath671 with respect to which the jacobian of @xmath672 , @xmath673 , is well defined : specifically , we have @xmath674 on @xmath675 , and for @xmath676 , @xmath677 for any @xmath571 .
finally , with the definition of separation time inherited from @xmath620 , the distortion bound in lemma 5.1(3 ) holds for @xmath678 on @xmath679 .
* ii . systems with holes * the setting is as in paragraph i. we fix an open set @xmath3 and call it the hole . "
* towers with markov holes * ( following [ dwy ] ) : let @xmath185 be the tower arising from the horseshoe @xmath542 .
we say @xmath185 _ respects the hole _
@xmath5 if the following conditions are satisfied : 1 .
@xmath680 is the union of countably many elements of @xmath623 .
2 . @xmath681 , and there exist @xmath266 , @xmath682 such that all @xmath683 satisfy @xmath684 for all @xmath685 .
because of * ( h.1 ) * , we refer to @xmath686 , the hole on @xmath205 , as a markov hole . "
this implies in particular that for every @xmath337 and @xmath621 with @xmath687 , @xmath688 either does not meet @xmath5 or it is completely contained in @xmath5 .
equivalently , on the tower @xmath185 , each @xmath689 either falls into the hole completely on its way up the tower or returns to @xmath620 intact .
earlier on we have used @xmath104 to denote an open system .
observe that @xmath690 and @xmath691 where @xmath692 are open systems of the same type .
as before , we write @xmath693 and @xmath694
. in particular , @xmath695 . the notation @xmath696 for @xmath697
is sometimes used to distinguish between the system with and without holes .
corresponding objects for @xmath698 are denoted by @xmath699 and @xmath700 etc .
abstract towers and a notion of spectral gap * in paragraphs i and ii , we considered towers that arise from generalized horseshoes .
towers can , in fact , be defined in the abstract . leaving details to the reader
, an abstract expanding tower is a dynamical system @xmath661 where @xmath679 is a compact set , @xmath701 has a tower structure , @xmath672 moves points up the tower until their return time @xmath605 ; there is a countable markov partition @xmath702 on @xmath671 which is a generator and a reference measure @xmath670 with respect to which we have ( i ) @xmath703 on @xmath675 and ( ii ) modulo a set of @xmath670-measure zero , @xmath704 where @xmath705 maps each closed set @xmath706 homeomorphically onto @xmath679 with the distortion bound in lemma 5.1(3 ) .
abstract expanding towers with markov holes @xmath707 are defined in the obvious way , as are abstract hyperbolic towers .
given @xmath708 with @xmath709 for some @xmath710 and @xmath711 , and @xmath672 when no ambiguity can arise given context .
thus @xmath605 is the name of the return time function on @xmath712 and @xmath679 . ]
we fix @xmath713 with @xmath714 where @xmath485 is as in lemma [ lemma : jacobian](3 ) , and define a symbolic metric on @xmath671 by @xmath715 . since @xmath716 , lemma [ lemma : jacobian](3 ) implies that @xmath673 is log - lipshitz with respect to this metric .
let @xmath717 where @xmath718 and @xmath719 lip@xmath720 in the last displayed formula is with respect to the symbolic metric @xmath721 , and @xmath722 so defined is a banach space .
now consider the open system @xmath698 where @xmath707 is a markov hole .
following [ bdm ] , we let @xmath723 denote the transfer operator associated with @xmath724 defined on @xmath725 , i.e. , for @xmath726 and @xmath727 , @xmath728 we say @xmath698 has a _ spectral gap _ if * @xmath723 is quasi - compact with a unique eigenvalue @xmath200 of maximum modulus , and * @xmath200 is real with @xmath729 ; it is simple , with a one - dimensional eigenspace .
notice that if @xmath730 satisfies @xmath731 , then @xmath732 defines a conditionally invariant measure for @xmath672 with eigenvalue @xmath200 , i.e. @xmath733 . finally , if @xmath185 is an abstract hyperbolic tower that projects onto @xmath708 , and @xmath734 is a markov hole which projects onto @xmath707 , then we say @xmath735 has a spectral gap if @xmath698 does . the conditions * ( a.1)**(a.4 ) * in sect .
[ tower results ] have now been made precise .
as noted earlier , our aim in this section is to prove , as an intermediate step for theorems d and e , a version of the corresponding results for the open system @xmath690 .
these results are deduced from some previously known results for @xmath698 , which we first recall . * i. results for expanding towers * we consider here an abstract expanding tower @xmath698 with markov holes .
the following notation is used : let @xmath725 be the function space above , and define @xmath736 to be the set of bounded functions in @xmath725 whose lipschitz constant is also bounded , i.e. the definition of @xmath736 is the same as that of @xmath725 , but with the weights @xmath737 removed .
let @xmath738 denote the set of invariant measures on @xmath700 , and define @xmath739 ( mostly @xcite ; see remark below ) [ thm : exp conv ] assume @xmath709 , and @xmath698 has a spectral gap with largest eigenvalue @xmath200 .
let @xmath740 be the unique eigenfunction of @xmath200 with @xmath741 .
then : 1 .
there exist constants @xmath742 and @xmath743 such that for all @xmath726 , @xmath744 assume additionally ( * ) : @xmath745 and @xmath746 such that @xmath747 for all @xmath230 .
1 . @xmath748 2 .
let @xmath749 be defined by @xmath750 then @xmath751 and attains the supremum in ( b ) .
other properties of @xmath749 are that @xmath752 is ergodic , and enjoys exponential decay of correlations between @xmath26 and @xmath753 for @xmath754 and @xmath755 .
_ the restriction @xmath756 , which appears in the definition of @xmath757 , is omitted in [ bdm ] , as is the condition ( * ) , which is extremely mild , ; indeed , in that setting , @xmath758 and all measures @xmath759 satisfy @xmath760 . ] but a condition of this type is needed to ensure that @xmath751 . since a main novelty of theorem 3 is the noncompactness of the phase space @xmath671 , and these conditions are directly connected to the finiteness of various quantities , we will provide sketches of corrected proofs of theorem 3(b ) and ( c ) in sect .
the proofs of parts ( a ) and ( d ) in [ bdm ] are unaffected .
.2 in * ii .
results for hyperbolic towers arising from @xmath104 * we now return to the setting of sect .
2.1 , where @xmath55 is a @xmath10 diffeomorphism with or without singularities .
let @xmath3 , and assume that the open system @xmath104 satisfies * ( a.1)**(a.4 ) * in sect .
we first recall the following result proved in @xcite as part of our study of billiard systems with holes .
let @xmath761 be the class of measures @xmath215 on @xmath205 with the following properties : ( i ) @xmath215 has absolutely continuous conditional measures on unstable leaves ; and ( ii ) @xmath762 for some @xmath763 .
[ thm : hyp conv ] under the conditions above , the following hold for @xmath690 : * for all @xmath764 with @xmath765 , where @xmath766 is as in theorem 3(a ) , @xmath767 * there exists a conditionally invariant distribution @xmath768 , such that @xmath769 , @xmath770 , and for which the following hold : for all @xmath764 , @xmath771 where the convergence is in the weak * topology .
the measure @xmath772 can be thought of as the _ physical measure for the leaky system _ @xmath773 .
we formulate in theorem 5 the results which , along with theorem 4 , will give the analogs of theorems d and e for @xmath690 .
let @xmath774 denote the set of invariant probability measures supported on @xmath775 , and define @xmath776 furthermore , let @xmath777 be the set of bounded functions on @xmath205 which are continuous on each @xmath778 .
we postpone the definitions of lip@xmath779 and lip@xmath780 ( other function spaces that will appear ) until after the theorem .
[ thm : variational ] let @xmath690 be as above . then the following hold .
1 . @xmath781 2 .
let @xmath782 be defined by @xmath783 then @xmath784 and it attains the supremum in ( a ) .
3 . other properties of @xmath782 are that @xmath785 is ergodic , and exhibits exponential decay of correlations between @xmath26 and @xmath786 for @xmath787 and @xmath788 .
the function spaces @xmath789 and @xmath790 are defined as follows .
for @xmath791 and @xmath792 , we denote by @xmath793 the distance between @xmath794 and @xmath795 according to the riemannian metric on @xmath1 , and extend @xmath796 to @xmath797 by setting @xmath798 for all @xmath799 and @xmath571 .
it then follows from sect .
[ tower review].i that @xmath800 for all @xmath685 and @xmath663 , @xmath571 .
for @xmath801 , let @xmath802 be the supremum of lipshitz constants of @xmath803 with respect to @xmath796 , as @xmath804 ranges over all stable sets in @xmath643 . then lip@xmath805 .
the function space lip@xmath780 is defined similarly using @xmath806 where @xmath806 is the lipschitz constant of @xmath26 restricted to unstable sets in the metric @xmath721 .
we assume part ( a ) of theorem 3 has been proved , and proceed to the proofs of parts ( b ) and ( c ) , following mostly @xcite and highlighting several finiteness issues .
_ 1 . return map to @xmath807 and the full shift @xmath808 _
since @xmath672 is not defined everywhere on @xmath671 , let us first make precise the definition of the survivor set @xmath700 .
recall from sect .
5.1 that modulo a set of @xmath670-measure 0 , @xmath679 is the disjoint union of a countable number of closed subsets @xmath809 with the property that \(i ) _ in the absence of @xmath707 _
, @xmath705 maps each @xmath809 homeomorphically onto @xmath679 , and \(ii ) _ with @xmath707 present _ , each @xmath809 either falls entirely into @xmath707 on its way up the tower or returns to @xmath679 intact .
we rename the subcollection @xmath810 that return to @xmath679 in ( ii ) as @xmath811 , and define @xmath812 it is easy to see that there is a bijection @xmath813 such that @xmath814 where @xmath808 the full shift . moreover , with @xmath815 given its relative topology as a subset of @xmath679 , and
@xmath816 given the topology defined by cylinder sets , @xmath817 is a conjugating homeomorphism .
let @xmath818 denote the set of cylinders in @xmath819 defined by coordinates @xmath820 , and write @xmath821 .
we introduce a metric @xmath822 on @xmath819 compatible with its topology defined by @xmath823 : for @xmath824 , define @xmath825 lie in different @xmath826 , and let @xmath827 ( where @xmath713 is as in sect .
we say a function @xmath828 is locally hlder continuous if @xmath829 _ 2 .
sarig s abstract results on the pressure of @xmath808 _
we recall here a few relevant results for @xmath808 .
these results were proved in @xcite in more general settings of topologically mixing countable markov shifts . given @xmath828 ,
let @xmath830 .
the _ gurevic pressure _ of @xmath831 is defined to be @xmath832 where @xmath460 is any fixed element of @xmath833 . for @xmath831 locally hlder continuous , it is shown in @xcite , theorem 1 , that the limit above exists and is independent of @xmath460 .
this number is @xmath834 in general , and is equal to @xmath835 for many @xmath831 given that @xmath836 is an infinite shift .
we will also need the following definitions : the transfer operator associated with @xmath831 is given by @xmath837 let @xmath838 be the set of @xmath836-invariant borel probability measures on @xmath819 . given a potential @xmath828
, we say @xmath839 is a _
gibbs measure _ for @xmath831 if there exist constants @xmath840 and @xmath841 such that for any @xmath190 , @xmath842 and @xmath843 , @xmath844 the following version of results from @xcite are adequate for our purposes : let @xmath808 be as above , and let @xmath845 be locally hlder continuous .
assume @xmath846 .
then : * ( * ? ? ?
* theorem 1 ) @xmath847 ; * ( * ? ? ?
* theorem 3 ) @xmath848 * ( * ? ? ?
* theorem 8) suppose @xmath849 is a gibbs measure for @xmath831 , and @xmath850 .
then @xmath851 it follows from ( 1 ) and ( 2 ) above that for @xmath839 , @xmath852 provided @xmath846 and @xmath853 . _
notation : _ in what follows , we will identify @xmath854 with @xmath808 and use the two sets of notation interchangeably .
we also introduce the following notation : given @xmath759 , let @xmath855 denote the measure @xmath856 .
relating pressure on @xmath857 to that on @xmath858 _ let @xmath859 .
the aim of this step is to prove that for every @xmath759 with @xmath860 , @xmath861 the last two inequalities follow from theorem 6(a),(b ) once we check ( i ) @xmath831 is locally hlder continuous with respect to the metric @xmath822 , ( ii ) @xmath846 , and ( iii ) @xmath862 . for ( i )
, notice that by lemma [ lemma : jacobian ] , @xmath831 is locally hlder continuous with respect to the separation time metric @xmath721 , and @xmath863 . for ( ii ) ,
let @xmath864 denote the element of @xmath833 containing @xmath865 .
we fix @xmath866 and use the bounded distortion of @xmath678 given by lemma [ lemma : jacobian](3 ) to write @xmath867 here we have used @xmath868 and the fact that @xmath705 maps each @xmath869 bijectively onto @xmath807 . for ( iii ) , we will show @xmath860 implies @xmath862 : since @xmath674 on @xmath870 , @xmath871 thus if @xmath860 , then , noting @xmath872 , we have @xmath873 this completes the verification of ( i)(iii ) .
the equality in ( [ ineq ] ) follows from ( [ eq : finite ] ) together with the general formula of abramov @xcite , which says that @xmath874 . in all the references we know of ( e.g. @xcite ) ,
this equality is proved assuming the invertibility of the transformation . in the situation above
, @xmath672 is clearly not invertible , but the same result is easily deduced by passing to natural extensions ; see appendix b. _ 4 .
existence of a pressure - maximizing invariant measure @xmath749 _ let @xmath749 be the linear functional on @xmath875 defined by @xmath876 we refer the reader to [ bdm ] for verification that @xmath749 is a well defined , @xmath672-invariant probability measure on @xmath700 .
the aim of this step is to show that plugging @xmath877 into ( [ ineq ] ) , we get @xmath878 observe from the definition of @xmath831 in step 3 that @xmath879 .
the following lemma shows that @xmath880 is a gibbs measure for the potential @xmath831 , with @xmath881 .
* lemma 5.3 ) [ lem : nu bounds ] there exists a constant @xmath840 such that for any @xmath882 , any @xmath20-cylinder @xmath883 , and any @xmath884 , @xmath885 it remains only to check that @xmath886 , for this bound implies @xmath887 ( see step 3 above ) , and once we have that , theorem 6(c ) gives the first equation in ( [ eq ] ) .
the second equation follows from ( [ ineq ] ) and the first . in what follows
, @xmath596 will be used as a generic constant the value of which is permitted to vary from line to line . to prove @xmath886 , we first estimate @xmath888 where @xmath889 is an arbitrary point in @xmath460 .
the first inequality comes from lemma 5.2 , the second from lemma 5.1(3 ) , and the third from the tail bound for @xmath890 . using the invariance of @xmath749 and the fact that @xmath674 on @xmath891 , we obtain @xmath892 @xmath893 the inequalities above come from condition ( * ) in theorem 3 ; this is the only place in the entire proof that uses this condition .
we have also used the fact that @xmath894 is bounded by @xmath895 times the last quantity in ( [ eq : nu decay ] ) .
parts ( b ) and ( c ) of theorem 3 follow immediately from steps 3 and 4 .
@xmath316 we will prove this theorem by leveraging the corresponding results for expanding towers . _ variational principle _ ( theorem 5(a),(b ) ) : first , we show @xmath896 which follows immediately from the following lemma : [ lem : var tnu ] let @xmath897 and define @xmath898 .
then @xmath899 and 1 .
@xmath900 ; 2 .
@xmath901 .
let @xmath897 .
the fact that @xmath902 is clear .
that @xmath899 will follow once we prove assertion ( i ) of the lemma : from sect .
5.1i , we see that @xmath903 and @xmath673 are related by @xmath904 for a bounded function @xmath656 ( lemma 5.1(4 ) ) .
it follows that @xmath905 the invariance of @xmath849 being used in the second equality .
assertion ( ii ) follows from ( a ) the entropy of a transformation is equal to that of its natural extension , and ( b ) the natural extension of @xmath906 is isomorphic to that of @xmath907 .
see appendix b for more detail on ( b ) . to complete the proof
, we will show that ( i ) the results of theorem 3 are applicable to the quotient tower , and ( ii ) @xmath782 as defined in part ( b ) is in @xmath908 and projects to @xmath749 .
these two steps together will show that ( [ comp ] ) is in fact an equality , and the quantity on the right is @xmath909 . to apply theorem 3
, it suffices to show that condition ( * ) holds in the present setting , i.e. for the quotient tower of a hyperbolic tower arising from @xmath104 and satisfying * ( a.1)**(a.4)*. notice first that ( * ) holds if @xmath910 is bounded , for @xmath911 can grow at most exponentially and @xmath912 on the corresponding set is @xmath913 where @xmath656 is as in lemma 5.3 .
thus there is a potential problem only in the setting of theorem c , where @xmath910 may become arbitrarily large as one approaches the singularity set @xmath12 . here
it is * ( h.2 ) * of section [ tower review].ii and in sect .
2.1.iii that give what we need : since @xmath914 , we have @xmath915 for some @xmath526 and @xmath916 by * ( h.2)*. this together with implies that on @xmath917 , @xmath918 where @xmath182 is the dimension of @xmath919 . thus on @xmath920 , we have @xmath921 which , as explained above , gives ( * ) .
it remains to produce @xmath782 with the properties in ( ii ) .
let @xmath772 be the physical conditionally invariant distribution from theorem [ thm : hyp conv ] . for @xmath787
, define @xmath922 to be the measure such that @xmath923 .
notice that since @xmath924 and @xmath925 , we also have @xmath926 .
let @xmath927 denote the density of @xmath928 with respect to @xmath670 . now using theorem [ thm : hyp conv](b ) , @xmath929 let @xmath930 .
then @xmath931 is clearly linear in @xmath26 , positive and satisfies @xmath932 . also , @xmath933 so that @xmath931 extends to a bounded linear functional on @xmath777 . by the riesz representation theorem
, there exists a unique borel probability measure @xmath782 satisfying @xmath934 for each @xmath935 . since @xmath936 , the invariance of @xmath782 follows from @xmath937 by the conditional invariance of @xmath772 . since @xmath938 for every @xmath20
, it follows that @xmath939 .
to place @xmath784 , we need to show @xmath940 . this is true by ( [ comp2 ] ) with @xmath941 and the fact that the integral on the right is known to be finite .
_ other properties of @xmath782 .
_ the ergodicity of @xmath782 follows from that of @xmath749 . to show that @xmath782 enjoys exponential decay of correlations , we begin by decomposing @xmath782 into conditional measures @xmath942 on @xmath804-leaves and a transverse measure @xmath943 on the set of stable leaves in each @xmath778 . for @xmath801 ,
define @xmath944 .
since each @xmath942 is a probability measure , we have @xmath945 . by definition
, @xmath946 is constant on @xmath804-leaves and @xmath947 .
also if @xmath787 , then @xmath948 so that we may consider @xmath949 as a function on @xmath671
. now let @xmath787 and @xmath788 with @xmath950 .
define @xmath946 as above and let @xmath951 .
note that @xmath952 . then setting @xmath953 , we write @xmath954 since @xmath946 and @xmath955 are constant on @xmath804-leaves
, we have @xmath956 and @xmath957 .
then since @xmath949 and @xmath958 , the last term in is @xmath959 for some @xmath960 by theorem [ thm : exp conv](d ) ( see also ( * ? ? ?
2.8 ) ) .
the second term of is identically 0 since , @xmath961 to estimate the first term in , notice that @xmath962
. then since @xmath963 is continuous on each @xmath804 , there must exist @xmath964 such that @xmath965 .
thus latexmath:[\[\label{eq : stable contract } by definition of @xmath796 . taking both @xmath967 and @xmath621 to be
approximately @xmath968 completes the proof .
we now return to the original open system @xmath38 , where @xmath2 is any dynamical system admitting a tower with the properties in sect . [ tower results ] ( see sect .
5.1 for detail ) .
let @xmath969 be the srb measure for @xmath613 on @xmath205 before the removal of the hole .
note that @xmath970 , the unique srb measure for @xmath2 with @xmath607 .
it follows from ( * ? ? ?
* ; * ? ? ?
* section 2 ) that @xmath971 , so that @xmath972 by theorem [ thm : hyp conv](a ) .
since @xmath973 for each @xmath230 , we have @xmath974 and part ( a ) of theorem d is proved . to prove part ( b ) ,
define @xmath975 where @xmath772 is the conditionally invariant measure from theorem 4 .
we use @xmath976 to describe the surviving dynamics at time @xmath20 .
it follows from the relation @xmath977 that for any borel subset @xmath978 of @xmath195 , we have @xmath979 so that @xmath199 is a conditionally invariant measure for @xmath980 with eigenvalue @xmath200 . by theorem 4(b ) , @xmath981 proving part ( b ) . to prove part ( c ) ,
define @xmath982 where @xmath782 is from theorem 5 .
arguing analogously to , we see that @xmath203 is an invariant measure for @xmath2 supported on @xmath983 .
. we will show \(i ) @xmath985 , and \(ii ) @xmath986 .
integrating over sets of the form @xmath987 before summing over @xmath20 , we see that the left side of ( i ) is equal to @xmath988 and the right side is equal to @xmath989 , the latter using the invariance of @xmath782 and relation @xmath990 .
these two integrals are easily seen to be equal : let @xmath991 denote the jacobian with respect to @xmath992 for @xmath646 and @xmath993 where @xmath994 . then on @xmath620 , @xmath995 as @xmath620 is an isometric copy of @xmath8 , so we have @xmath996 for ( ii ) , that @xmath997 is obvious .
the reverse inequality follows from ( * ? ? ?
* proposition 2.8 ) since @xmath631 is at most countable - to - one . combining ( i ) and
( ii ) and using theorem 5(b ) , @xmath998 the following lemma completes the proof of part ( c ) .
[ lem : gsrb ] @xmath201 that @xmath203 is ergodic follows immediately from the fact that @xmath782 is ergodic . in order to show that @xmath201
, we will show that there exist @xmath89 such that for each @xmath91 , @xmath999 . once this is established , we conclude by an argument similar to lemma [ lem : approach ] that @xmath203-a.e .
point approaches @xmath1000 at an arbitrarily slow exponential rate . to establish this bound , we need estimates on how @xmath782 decays up the levels of the tower .
recall @xmath1001 . in the proof of theorem [ thm :
variational ] , we established that @xmath1002 , @xmath1003 , by using * ( h.2 ) * of section [ tower review].ii .
thus we have @xmath1004 using and @xmath1005 . finally , we prove part ( d ) .
if @xmath26 is a continuous function on @xmath1 , we define its lift to @xmath205 by @xmath1006 .
this lift is continuous on each @xmath778 and @xmath1007 so that @xmath1008 . using theorem 5(b ) , we have @xmath1009 since @xmath975 . to complete the proof of theorem d , it remains to show that @xmath203 enjoys exponential decay of correlations .
let @xmath1010 denote the hlder continuous functions on @xmath1 with exponent @xmath182 .
if @xmath1011 and @xmath1012 , then @xmath1013 .
this can be proved as in ( * ? ? ?
* ; * ? ? ?
* section 6 ) . also , taking @xmath1014 , for @xmath1015 , @xmath571 and @xmath1016 , @xmath1017 , we have @xmath1018 so taking @xmath1019 , we may apply to @xmath1020 .
we follow and note that @xmath1021 to conclude that the exponential decay of correlations for @xmath203 follows from that for @xmath782 . as an immediate corollary of theorem d , we have @xmath1022 since we have identified a measure , namely @xmath203 , in @xmath1023 with @xmath219
. we will call upon the results in sect .
2.1 to provide the reverse inequality
once we put ourselves in a viable setup . notice that @xmath982 necessarily gives positive measure to @xmath1024 . _
( a ) @xmath206 with @xmath1025 on a neighborhood of @xmath8 . _ in this case , @xmath1026 since we can simply take the set @xmath460 in the definition of @xmath110 to be this neighborhood . by theorem
c , @xmath1027 .
this together with ( [ halfeq ] ) gives the desired result .
_ ( b ) @xmath8 is contained in a @xmath113-hyperbolic product set .
_ taking this set to be @xmath1028 in the definition of @xmath114 , it is immediate that @xmath1029 .
theorem c and ( [ halfeq ] ) then give the two halves of the desired equality .
we begin with @xmath1032 and the set @xmath509 of regular points in the sense of oseledec .
each @xmath521 has @xmath182 distinct lyapunov exponents @xmath1033 with corresponding subspaces @xmath1034 such that @xmath1035 .
let @xmath1036 .
fix @xmath1037 .
it follows by standard arguments ( see ( * ? ? ?
* ; * ? ? ?
3.1 ) ) that for @xmath30-typical @xmath96 , one can define an inner product , @xmath1038 , on the tangent space @xmath335 such that item @xmath506 of proposition 4.1 holds .
denote by @xmath1039 the norm induced by @xmath1038 and by @xmath1040 the euclidean norm on @xmath335 .
it follows from the same construction that there exists a measurable function @xmath1041 , with @xmath1042 for @xmath1043 and @xmath1044 define a linear map @xmath1045 which takes @xmath334 to @xmath1046 for each @xmath337 and such that @xmath1047 .
then @xmath1048 is a lyapunov chart satisfying properties @xmath504 and @xmath505 of proposition 4.1 .
the construction outlined thus far is standard and is not affected by the presence of singularities ( see ( * ? ? ?
* part i , theorem 2.2 ) ) .
we now proceed to prove the statements of section 4.2.4 which are affected by the singularities .
we drop the subscript @xmath96 for simplicity of notation and write @xmath1049 and @xmath1050 in what follows .
notice that in the notation of sect . 4 , @xmath1051 denotes the ball of radius @xmath83 in the ( lyapunov ) norm @xmath1050 since that is the norm of the lyapunov charts @xmath435 . to distinguish between norms
, we use @xmath1052 to denote the ball of radius @xmath83 in the euclidean norm on @xmath335 .
we identify @xmath335 and @xmath1053 and view @xmath1054 formally as a change of norm .
_ proof of @xmath528 .
_ recall the injectivity radius from section [ upper bound ] , @xmath1055 , given by equation .
since we have assumed @xmath1056 , we have @xmath1057 for @xmath1058 .
thus again using , for @xmath1059 and @xmath1060exp@xmath1061 , we have @xmath1062 this implies that exp@xmath1063 maps @xmath1064 injectively into @xmath1065 .
thus for @xmath1066 , we use to estimate , @xmath1067 which establishes @xmath528 with @xmath1068 and @xmath1069 . note that by , @xmath1070 , with room to spare .
_ proof of @xmath529 .
_ recall that @xmath1071 taking @xmath1072 , we use to estimate @xmath1073 for some @xmath1074 . by , @xmath1075 .
since exp@xmath1076 , we have @xmath1077 , so that @xmath1078 . finally , since @xmath1079
, we conclude that @xmath1080 the statement follows by taking @xmath1081 to be the larger of @xmath1082 and @xmath1069 .
let @xmath1087 be a measure - preserving transformation ( mpt ) of a probability space .
recall that the _ natural extension _ of @xmath1087 , denoted here by @xmath1088 , is defined as follows : @xmath1089 @xmath1090 @xmath1091 is generated by cylinder sets with @xmath1092 in each coordinate , and @xmath1093 \(1 ) consider @xmath1094 where @xmath849 is any @xmath613-invariant borel probability measure , and let @xmath1095 be the corresponding quotient system .
we claim that the natural extensions of these two mpt s are isomorphic .
suppose @xmath1101 .
letting @xmath1102 denote the stable set of @xmath1103 , we have , by definition , @xmath1104 .
the uniform contraction of @xmath613 along stable sets implies that this intersection consists of a single point .
likewise , @xmath1105 . since @xmath1106 is equivalent to @xmath1107 , we have proved @xmath1108 . applying the same argument to the sequences @xmath1109 and @xmath1110 ,
we conclude that @xmath1111 for all @xmath1112 .
@xmath316 \(2 ) next given @xmath1095 and @xmath1113 , we call @xmath1114 with @xmath1115 normalized its _ induced map _ on @xmath679 , and claim that the induced map of @xmath1116 on @xmath1117 is the natural extension of @xmath705
. the proof is easy .
a. katok and j .- m .
strelcyn , with the collaboration of f. ledrappier and f. przytycki , _ invariant manifolds , entropy and billiards ; smooth maps with singularities _ , lecture notes in math .
* 1222 * , springer - verlag : berlin , 283 pages , 1986 .
young , _ ergodic theory of differentiable dynamical systems _ , real and complex dynamical systems , " eds . b. branner and p. hjorth , nato asi series c464 , kluwer academic publishers ( 1995 ) , 293 - 336 . | we study the relation between escape rates and pressure in general dynamical systems with holes , where pressure is defined to be the difference between entropy and the sum of positive lyapunov exponents .
central to the discussion is the formulation of a class of invariant measures supported on the survivor set over which we take the supremum to measure the pressure .
upper bounds for escape rates are proved for general diffeomorphisms of manifolds , possibly with singularities , for arbitrary holes and natural initial distributions including lebesgue and srb measures .
lower bounds do not hold in such generality , but for systems admitting markov tower extensions with spectral gaps , we prove the equality of the escape rate with the absolute value of the pressure and the existence of an invariant measure realizing the escape rate , i.e. we prove a full variational principle . as an application of our results , we prove a variational principle for the billiard map associated with a planar lorentz gas of finite horizon with holes . |
let @xmath1 be a positive definite riemannian metric on @xmath2 with the corresponding laplace - beltrami operator , @xmath3 and let @xmath4 be real valued .
we prove the following general bound which was already established ( under an additional necessary condition ) in higher dimensions in @xcite , but which was open in dimension two : * theorem . *
_ suppose that @xmath5 , and that @xmath6 with @xmath7 .
suppose that @xmath8 satisfies @xmath9 then @xmath10 where the constant @xmath11 depends only on @xmath12 , @xmath13 , and @xmath14 .
_ a function @xmath15 satisfying is sometimes called a weak quasimode .
it is a local object in the sense that if @xmath15 is a weak quasimode then @xmath16 , @xmath17 is also one .
the localization is also valid in phase space : for instance if @xmath18 then @xmath19 is also a weak quasimode see ( * ? ? ?
* chapter 7 ) or ( * ? ? ? * chapter 4 ) for the review of the weyl quantization @xmath20 .
if @xmath21 , then @xmath22 ( defined on @xmath23 ) is essentially self - adjoint and the spectrum of @xmath22 is discrete in a neighbourhood of @xmath24 see for instance ( * ? ? ?
* chapter 4 ) . in this case
weak quasimodes arise as _
spectral clusters _ : @xmath25 then @xmath26 , for any @xmath27 , is a weak quasimode in the sense of . since @xmath28 for @xmath29 , agmon estimates
( see for instance ( * ? ? ?
* chapter 6 ) ) and sobolev embedding show that @xmath30 for @xmath31 .
hence we get global bounds @xmath32 it should be stressed however that a weak quasimode is a more general notion than a spectral cluster .
the result also holds when @xmath33 is replaced by a two dimensional manifold and , as in the example above , gives global bounds on spectral clusters when the manifold is compact . if @xmath34 this is also a by - product of the bound of avakumovic - levitan - hrmander on the spectral function
see @xcite , and for a simple proof of a semiclassical generalization see @xcite or @xcite . in higher dimensions
the theorem requires an additional phase space localization assumption and is a special case of ( * ? ? ?
* theorem 6 ) : suppose @xmath35 is a function on @xmath36 satisfying @xmath37 for some @xmath38 .
suppose also that @xmath39 , and that for @xmath40 @xmath41 then for @xmath42 such that @xmath43 we have @xmath44 when @xmath45 the bound holds with @xmath46 , which is optimal in general if @xmath47 is not positive definite see ( * ? ? ?
* , 6 ) and 3 below for examples . a small bonus in dimension two is the fact that the frequency localization condition needed for is not necessary see below .
the proof of the theorem is reduced to a local result presented in proposition [ theorem2 ] .
that result follows in turn from a rescaling argument involving several cases , some of which use the most technically involved result of @xcite : if @xmath48 and for @xmath49 @xmath50 then for @xmath51 satisfying , the bound holds for any @xmath52 see ( * ? ? ?
* corollary 1 ) for the original results and propositions [ proposition1 ] and [ proposition2 ] for rescaled versions used in our proof .
we do not know of any simpler way to obtain .
by compactness of @xmath53 , it suffices to prove uniform @xmath54 bounds on @xmath8 over each small ball intersecting @xmath14 , where in our case the diameter of the ball can be taken to depend only on @xmath55 estimates for @xmath12 and @xmath13 over a unit sized neighborhood of @xmath14 , for some large @xmath56 . without loss of generality
we consider a ball centered at the origin in @xmath2 .
let @xmath57 after a linear change of coordinates , we may assume that @xmath58 next , by replacing @xmath59 by @xmath60 and @xmath61 by @xmath62 , for some constant @xmath63 depending on the @xmath64 norm of @xmath12 and @xmath13 over a unit neighborhood of @xmath14 , we may assume that @xmath65 this has the effect of multiplying @xmath66 by a constant in the equation , which can be absorbed into the constant @xmath11 in . in general , we let @xmath67 and will deduce the main theorem as a corollary of the following [ theorem2 ] suppose @xmath5 , that @xmath68 satisfy and , and that @xmath8 satisfies @xmath69 then @xmath70 where the constant @xmath11 depends only on @xmath71 in for some fixed @xmath56 .
we start the proof of proposition [ theorem2 ] by recording the following two propositions , which are consequences of ( * ? ? ?
* corollary 1 ) .
[ proposition1 ] suppose that - hold , and that @xmath72 for @xmath73 .
if the following holds , and @xmath5 , @xmath74 then @xmath75 , where @xmath11 depends only on @xmath71 in for some fixed @xmath56 .
[ proposition2 ] suppose that - hold , and that @xmath76 and @xmath77 . if the following holds , and @xmath5 , @xmath74 then @xmath75 , where @xmath11 depends only on @xmath71 in for some fixed @xmath56 . to see that these follow from ( * ? ? ?
* corollary 1 ) , we first note that in proposition [ proposition2 ] above , since @xmath78 , we have @xmath79 for @xmath73 , so since @xmath12 is positive definite the conditions on @xmath12 and @xmath13 in ( * ? ? ?
* corollary 1 ) are met .
we remark that the condition @xmath76 guarantees the the zero set of @xmath13 is a nearly - flat curve through the origin , although this is not strictly needed to apply the results of @xcite .
that the constants in the estimates of @xcite depend only on the above bounds on @xmath12 and @xmath13 follow from their proofs .
next , the estimates above can be localized , as remarked before , so we may assume that @xmath8 is compactly supported in @xmath80 , after which we may extend @xmath12 and @xmath13 globally without affecting the application of ( * ? ? ?
* corollary 1 ) .
indeed , in both propositions above the assumptions imply @xmath81 , so that one may cut off @xmath8 by a smooth function which is supported in @xmath80 and equals 1 for @xmath82 .
finally , the condition ( 1.4 ) of @xcite that @xmath83 for some @xmath84 is not needed for the @xmath54 results of that paper to hold in dimension two . to see this
, we note that since @xmath85 and @xmath86 on the ball @xmath87 , then if @xmath8 is supported in @xmath80 and @xmath88 for @xmath89 , then @xmath90 this follows by the semiclassical pseudodifferential calculus ( see ( * ? ? ?
* theorem 4.29 ) ) , since for @xmath91 with @xmath92 , @xmath93 .
hence , writing @xmath94 for the standard fourier transform of @xmath8 , @xmath95 an even better estimate than required .
we supplement propositions [ proposition1 ] and [ proposition2 ] with the following two lemmas .
[ lemma3 ] suppose that - hold , and that @xmath96 for @xmath97 . if the following holds , and @xmath5 , @xmath98 then @xmath99 , where @xmath11 depends only on @xmath71 in for some fixed @xmath56 .
consider the function @xmath100 , and @xmath101 .
then , since @xmath102 , we have @xmath103 since the spatial dimension equals 2 , interior sobolev estimates yield @xmath104 , where we note that the conditions and hold for @xmath105 since @xmath106 .
[ lemma4 ] suppose that - hold , and that @xmath107 for @xmath108 . if the following holds , and @xmath109 , @xmath110 then @xmath111 , where @xmath11 depends only @xmath71 in for some fixed @xmath56 .
let @xmath112 , @xmath113 , and @xmath114 .
note that the assumptions on @xmath59 in the statement and in imply that @xmath115 for @xmath116 , so that @xmath117 satisfies , and the constants @xmath71 in can only decrease for @xmath63 . then with @xmath118 , @xmath119 by proposition [ proposition1
] , we have @xmath120 , giving the desired result .
it suffices to prove that for each @xmath121 there is some @xmath122 so that @xmath123 , with a global constant @xmath11 . without loss of generality
we take @xmath124 .
we will split consideration up into four cases , depending on the relative size of @xmath125 and @xmath126 . since for @xmath66 bounded away from @xmath127 the result follows by elliptic estimates , we will assume @xmath128 so that @xmath129 below is at most @xmath130 .
* case 1 : * @xmath131 .
since @xmath132 , then lemma [ lemma3 ] applies to give the result with @xmath133 .
* case 2 : * @xmath134 .
since we may add a constant of size @xmath66 to @xmath13 without affecting , we may assume @xmath76 . by rotating we
may then assume @xmath135 where @xmath136 .
dividing @xmath13 by 4 if necessary we may assume @xmath137 .
let @xmath138 , @xmath139 , and @xmath140 with @xmath141 we have @xmath142 proposition [ proposition2 ] applies , since @xmath143 and @xmath117 satisfy - , and the constants @xmath71 in for @xmath143 and @xmath117 are bounded by those for @xmath12 and @xmath13 .
thus @xmath144 , giving the desired result on @xmath8 with @xmath145 .
* case 3 : * @xmath146 . in this case , with @xmath147 , it follows that @xmath107 for @xmath148 .
we may apply lemma [ lemma4 ] with @xmath13 replaced by @xmath149 to get the desired result with @xmath150 .
* case 4 : * @xmath151 . since @xmath132
, it follows that there is a point @xmath152 with @xmath153 where @xmath154 . since @xmath155
, we may translate and apply case 2 to get @xmath54 bounds on @xmath8 over a neighborhood of radius @xmath156 about @xmath152 .
this neighborhood contains the neighborhood about @xmath127 of radius @xmath157 .
in ( * ? ? ? * section 5 ) , it was shown that there exist @xmath158 for which @xmath159 for which @xmath160 showing that the assumption of definiteness of @xmath12 can not be relaxed to non - degeneracy in the main theorem . in ( * ? ? ? *
theorem 6 ) the positive result was established showing that this growth of @xmath161 for indefinite , non - degenerate @xmath12 in two dimensions is in fact worst case .
the example of @xcite was produced using harmonic oscillator eigenstates .
here we present a different construction of such a @xmath158 with similar @xmath54 growth to help illustrate the role played by the degeneracy of @xmath12 .
the idea is to produce a collection @xmath162 of functions satisfying ( or equivalent ) , for which @xmath163 , and where @xmath164 runs over @xmath165 different values .
the examples will have disjoint frequency support , hence are orthogonal in @xmath166 .
upon summation over @xmath164 the @xmath166 norm then grows as @xmath167 , whereas the @xmath54 norm grows as @xmath168 , yielding an example with worst case growth after normalization .
we start by considering the form @xmath169 with @xmath170 . to assure that @xmath171 , we will take the fourier transform of @xmath158 to be contained in the set @xmath172 , as well as @xmath173 to satisfy the frequency localization condition ( * ? ? ?
* ( 1.4 ) ) .
our example is then based on the fact that one can find @xmath165 disjoint rectangles , each of volume @xmath174 , within this region , as illustrated in the diagram .
each @xmath162 will be an appropriately scaled schwartz function with fourier transform localized to one of the rectangles .
we then form @xmath188 since there are @xmath165 terms in the sum , and the terms are orthogonal in @xmath166 , it follows that @xmath189 ) , and one may smoothly cutoff to a bounded set without changing the estimates .
we observe that for this example it also holds that @xmath190 hence , @xmath158 is also a counterexample for the form @xmath191 .
rotating by @xmath192 gives the form @xmath193 , including in particular the form considered in ( * ? ? ?
* section 6 ) .
we also observe that @xmath194 will be @xmath195 if one restricts the sum in @xmath158 to @xmath196 , which still has @xmath197 values of @xmath164 , and thus exhibits the same @xmath54 growth as @xmath158 .
this idea does not however work to yield a counterexample for the form @xmath198 . | we prove optimal pointwise bounds on quasimodes of semiclassical schrdinger operators with arbitrary smooth real potentials in dimension two .
this end - point estimate was left open in the general study of semiclassical @xmath0 bounds conducted by koch - tataru - zworski @xcite .
however , we show that the results of @xcite imply the two dimensional end - point estimate by scaling and localization . |
water can remain liquid below its melting point and stays in its metastable phase known as supercooled water @xcite .
several thermodynamic ( e.g. isothermal compressibility , isobaric heat capacity and thermal expansion coefficient ) and dynamic ( e.g. viscosity , diffusivity , structural relaxation ) observables of liquid water show an anomalous behaviour @xcite . in the supercooled state
the values of these physical observables clearly diverge , with a singularity temperature @xmath0 estimated at @xmath1 k. despite the wide interest and the numerous research efforts on this subject , the origin of these critical phenomena remained unclear @xcite and lively debated @xcite .
it is widely accepted that the water anomalies are connected with the formation of various structural aggregates whose features are modified by the temperature / pressure conditions @xcite . a crucial point is the physical origin of these structural aggregations . in their pioneering work poole , sciortino , essmann , and
stanley @xcite postulated the existence of a liquid - liquid critical point ( llcp ) based on a simulation study and on the experimental observation of polyamorphism in glassy water ( i.e. the existence of two forms of water glass : low - density and high - density amorphous ice ) . in the llcp hypothesis
a first order phase transition takes place involving two supercooled water liquid forms : the low - density and the high - density phases .
the low - density phase is characterized by tetrahedral intermolecular coordination , while the high - density one shows more packed structures with distorted networks .
the existence of a llcp is compatible with a series of computational and experimental results @xcite .
recent new simulation studies confirmed the llcp hyphotesis @xcite .
nevertheless , the matter is still under debate : in their recent simulation study , limmer and chandler interpreted the water anomalies as non - equilibrium phenomena associated to the liquid - crystal transition @xcite .
moreover , the relevance of ice nucleation phenomena and their relationship with the water anomalies has been reported @xcite . a fundamental question in order to explain the origin of water anomalies is if two distinct intermolecular local and transient configurations , whose densities are different , persist at temperature higher than @xmath0
. the fluctuations between this two water species could explain the thermodynamic and dynamic anomalies .
unfortunately , the most evident critical phenomena are expected to occur in the supercooled phase at temperature / pressure values not directly accessible in bulk water samples , and this prevents a direct experimental solution of the problem .
in fact , in several experimental and simulation works the water singularity temperature @xmath0 at atmospheric pressure has been extrapolated to be around 223 - 228 k , while homogeneous nucleation fixes the lowest reachable temperature at about 231 k for liquid water .
actually , in a macroscopic bulk sample the crystallization phenomena and several experimental difficulties limit the minimum temperature around 243 k. nano - confinement has the advantage of preventing water freezing and of extending the experimental investigation of the supercooled phase down to very low temperatures ; it is however still unclear to what extent the interactions with the pore surfaces modify the water properties @xcite .
the hydrogen bond network largely determines the intra- and inter - molecular vibrational spectrum of supercooled water ; thus the experimental investigation of the spectral features gives precious information on the local intermolecular structures and on their variation with the temperature / pressure conditions
. the investigation of high frequency spectrum , 1000 - 4000 @xmath2 , gives access to the intra - molecular vibrational dynamics that has been recently related to the llcp hypothesis @xcite .
the low frequency spectrum of water , @xmath3 1000 @xmath2 , reflects the inter - molecular dynamics and so it is particularly sensitive to the local molecular structures .
this frequency range of the water spectrum has been investigated in several light scattering studies , see for example @xcite and references at http://www.lsbu.ac.uk / water/. two main broad peaks are observed at 50 and 200 @xmath2 , generally attributed to the hydrogen bond `` stretching '' and `` bending '' vibrations .
the spectra show also a very low frequency wing , @xmath4 20 @xmath2 , that has been attributed to `` relaxation '' processes .
the light scattering data were measured at relatively high temperatures and have rather poor signal - to - noise ratio , due to the very weak water signal : these drawbacks did not permit to identify all the possible vibrational modes possibly contributing to the water spectrum in this region .
moreover , the reported attempts of fitting the light scattering data have been based on a simple superposition of independent gaussian and lorentzian functions , neglecting any possible coupling between the vibrational dynamics and the structural relaxation processes .
an accurate analysis of the low frequency spectral features of supercooled water , including a quantitative comparison to the predictions of recent models for water dynamics is still missing ; it actually appears as a required step towards the full understanding of the structural and dynamical properties of water in its low temperature metastable state . in recent years
, time - resolved non - linear spectroscopy demonstrated to be a very useful tool for the investigation of complex liquid dynamics @xcite .
several studies of this type have been performed on liquid water ; only in few cases have they been extended to its supercooled phase @xcite .
time - resolved optical kerr effect ( oke ) investigations of liquid water @xcite show the signature of fast vibrational dynamics at short times , followed by a slower monotonic relaxation .
the initial oscillatory component provides information about intermolecular hydrogen - bond dynamics , and can be directly compared with the low - frequency raman spectrum @xcite .
actually , the oke experiments on supercooled water @xcite have been focused so far on the investigation of the slower decay , which is interpreted as a structural relaxation phenomenon .
no detailed oke investigation of the vibrational dynamics in supercooled water has been carried out , due to the intrinsic experimental difficulties , so that a detailed picture of this dynamics is still missing @xcite . using an improved experimental set - up @xcite ,
we measured , for the first time according to our knowledge , the fast vibrational dynamics and the slow structural relaxation in a single data set characterized by low noise and wide dynamic range .
moreover , we reached temperatures lower than in previous experiments , approaching the homogeneous nucleation limit .
we succeeded in measuring the oke signal relaxation over a large time interval of about 25 ps with a high time resolution , better than 20 fs .
we can thus provide the unambiguous measurement of the entire correlation function in supercooled water , enabling a very stringent comparison with the available theoretical models .
time - resolved oke spectroscopy is a non - linear technique exploiting the transient birefringence induced by the electric field of radiation in an optically transparent and isotropic medium .
the method has been described in many papers and review articles @xcite . in short , the oscillating electric field of a polarized short laser pulse ( the pump )
induces optical birefringence in the sample .
in addition to the instantaneous electronic contribution ( which does not carry any dynamic information ) , the induced birefringence reflects the relaxation and vibrational response of the molecules in the sample . a second pulse ( the probe ) , spatially superimposed within the sample with the pump pulse and characterized by a different polarization , probes the induced birefringence .
the intensity of probe light leaking through a polarizer placed after the sample measures the material response as a function of the time delay between pump and probe .
if femtosecond laser pulses are employed , the method enables to measure the relaxation processes and the low frequency vibrational correlation functions of the system . in this sense
it is the time - domain counterpart of depolarized light scattering technique @xcite .
the wide accessible time window , extending from tens of femtoseconds to hundreds of nanoseconds , makes oke a very flexible technique , capable of revealing very different dynamic regimes , from simple liquids to supercooled liquids and glass formers @xcite . in the case of optical heterodyne detection ,
the delay time dependent signal is directly proportional to the material response ( relaxation ) function , @xmath5 , convoluted with the instrumental function , @xmath6 , see methods .
the response @xmath5 is directly connected to the time derivative of the time - dependent correlation function of the anisotropic component of the collective susceptibility@xcite : @xmath7 the imaginary part of the fourier transform of @xmath5 corresponds to the frequency - dependent response , the observable measured in depolarized light scattering ( dls ) experiments @xcite .
we stress here that measuring the correct instrumental function @xmath6 is a crucial step for achieving a reliable and accurate fitting of the experimental oke signal .
a detailed description of the method adopted is given in the methods section .
supercooling bulk water can be achieved only with particularly pure sample and using specific attention for the cooling procedure .
we performed the measurements on a sealed vial of cylindrical shape , prepared for pharmaceutical purposes by the angelini company , which allowed us to reach the temperature of 247 k. the sample temperature was controlled by means of a home - made cryostat with a stability of 0.1 k , using a peltier cooler .
the heterodyne - detected oke ( hd - oke ) data are reported in figure [ fig_1 ] : the data clearly show the complex dynamic features presented by liquid and supercooled water .
the vibrational dynamics , extending up to 1 psec , displays an articulate pattern that becomes progressively more defined going deep in the supercooled phase , and that at longer times merges into the monotonic decay due to structural relaxation .
the time - resolved hd - oke experiment measures the time - dependent correlation function of the water susceptibility ( i.e. collective polarizability ) , see equation .
the time evolution of this function reflects the water dynamics through the electronic polarizability .
numerical investigations @xcite show that the oke signal of water is the result of a complex interplay between intrinsic molecular terms and interaction - induced contributions .
these calculations reveal how the fluctuations of the water susceptibility , including the anisotropic components , are largely determined by the intermolecular translational dynamics .
this behavior is expected , given the essentially isotropic water molecular polarizability .
the density fluctuations ( i.e. translational dynamics ) dominate the relatively low frequency range of the density - of - states@xcite , @xmath3 400 @xmath2 , whereas the rotational / librational dynamics affect the high frequency spectrum@xcite , @xmath8 400 @xmath2 . moreover , the structure and dynamics of the h - bond network are fundamental properties that affect the susceptibility fluctuations at any frequency .
the relaxation features presented by supercooled water are very similar to those of glass - forming liquids @xcite , and the slow dynamics is found in agreement with the main mode - coupling theory predictions @xcite . in figure
[ fig_2 ] we report the master plot analysis of the slow component of the hd - oke signal decay .
the new data confirm , over an extended temperature range , the power law temperature dependence of the relaxation time .
other experimental investigations reported similar power law temperature dependence of different water observables @xcite , both static and dynamic ; recently a small angle x - ray scattering investigation @xcite found that also the correlation length of the density fluctuations follow a power law with similar value of @xmath9 . the analysis of the whole time - dependent correlation function measured by the optical experiments requires the utilization of a model that enables an operative parametrization of the vibrational modes present in the water dynamics .
this turns out to be a complex task , and different attempts have been pursued @xcite in that direction .
none of these water studies was extended into the supercooled state .
moreover , the data analysis and the fitting models adopted so far did not include any coupling between the fast vibrational dynamics and the slow structural relaxation .
we tested some of these fitting procedures on our hd - oke data and found that they fail to reproduce correctly the hd - oke data in the supercooled temperature range . in order to analyze the vibrational dynamics and the structural relaxation including their coupling processes
, we utilized the schematic solution of mode - coupling theory @xcite .
the formalism of mode - coupling theory ( mc ) stems from a generalization of the mori and zwanzig hydrodynamic theory @xcite . in mc theory
the retardation effects are taken into account by memory functions @xmath10 , which play the role of the hydrodynamic friction coefficients .
of course the task of any microscopic theory is that of providing the form of @xmath10 . in its simplest form , developed for simple atomic liquids ,
only density fluctuations contribute to memory ; in order to overcome the limitations of this approach and to extend the mc theory description to the dynamics of more complex systems , several versions of the theory have been developed . for molecular liquids in fact
orientational and intra - molecular degrees of freedom contribute to the overall dynamics . in our analysis
we adopt the multi - component schematic formulation of the mc theory ( smc)@xcite . briefly , in this approach the density - density time correlator , @xmath11 , termed the master correlator , is described by the differential equation @xcite : @xmath12 where @xmath13 is the mode frequency , @xmath14 is the friction coefficient and the memory function @xmath15 is expressed as a series expansion of the master correlator itself . the role of other observables , @xmath16 , linked to the density correlator dynamics ( e.g. the molecular orientation , the dynamics of a target molecule , etc . )
is described by a set of differential equations of the same form as equation , written in terms of slave correlators @xmath17 and of memory functions of the form @xmath18 , which implies the coupling of the master and slave correlators .
the key point in the smc analysis of the oke experimental data is that the time correlation function of the sample polarizability can be expressed as the sum of the time derivatives of the slave correlators : @xmath19 we underline here that the individual slave correlators describe `` average collective modes '' , so that they can not be associated to specific physical variables . in summary ,
the smc approach represents a robust physical model capable of describing a complex dynamics , with no need of unjustified assumptions concerning the decoupling of slow and fast dynamics .
we solved the smc equations numerically , taking the frequencies , friction and coupling coefficients as parameters to be adjusted in order to reproduce the hd - oke response by a fitting procedure .
the details on the smc equations and fitting procedures are reported in the methods section . in figure [ fig_3 ]
we show some of the measured hd - oke signals with the fits obtained using the smc model .
the model reproduces correctly the experimental data over the whole time range at all temperatures .
most remarkably , and differently from other fitting models widely used , this result is achieved without imposing any non - physical decoupling of vibrational and relaxation dynamics .
an important result of our work is that the fitting at temperatures @xmath20 313 k requires three slave correlators , whereas for higher temperatures only two correlators are needed .
we would like to stress out that the master correlator alone can not reproduce the data suggesting that the density fluctuations are not enough to describe completely the dynamics of collective water polarizability ; as expected the hydrogen bound dynamics adds not trivial dynamical components that determine the water susceptibility fluctuations .
the time domain data , collected in the time - resolved hd - oke experiments , are transformed to yield the corresponding spectra in the frequency domain .
in fact , the fourier transform of the hd - oke response function , @xmath21 , is directly connected with the spectrum @xmath22 measured in the depolarized light scattering ( dls ) experiments : the imaginary part of @xmath21 is proportional to @xmath22 corrected for the bose factor , i.e @xmath23 \propto s_\mathrm{dls}(\omega)/[n(\omega)+1]$ ] @xcite .
our results then can be directly compared to the dls , rayleigh and raman data , found in the literature .
the imaginary part of the frequency domain response function , @xmath23 $ ] , was obtained as the fourier transform of the hd - oke signal deconvoluted from the instrumental function : @xmath23 \propto im\lbrace ft\left [ s(t ) \right ] /ft\left [ g(t)\right ] \rbrace$ ] .
the fourier transform of the measured hd - oke response and of the calculated smc response are reported in figure [ fig_4 ] .
the experimental spectra clearly show the characteristic vibrational features of liquid water : the two broad bands around 50 and 200 @xmath2 , and the low frequency wing .
this wing corresponds to the slow decay of the oke signal measured in the time domain ; as discussed in our previous work @xcite , it is attributed to structural relaxation .
the other two bands are generally attributed to translational intermolecular modes ( i.e. the oscillatory dynamics of water molecules around their instantaneous center of mass ) hindered by the first neighbour molecular cage @xcite . in the liquid phase they are largely affected by the hydrogen bonds . a comparison with the calculated vibrational density - of - states reported in ref .
@xcite suggests specific assignment of the 50 and the 200 @xmath2 bands .
the first one is dominated by the transverse translational motion , while the second has essentially longitudinal character @xcite .
the physical interpretation is that the two intermolecular vibrations are mainly involving `` bending '' and `` stretching '' distortion of the hydrogen bonds , respectively @xcite .
this picture even if intuitively useful , is clearly oversimplified : it does nt take into account the coupling of the two vibrations imposed by the complex hydrogen bond network present in the liquid water .
the presence of two different local structures , the one corresponding to highly tetrahedral intermolecular structures , and that consisting of strongly distorted network , as well as their continuous reorganization induced by the structural relaxation processes , are expected to largely affect the vibrational dynamics .
the smc fit enables to disentangle the vibrational modes contributing to this dynamics and to point out their coupling to the structural relaxation processes .
the stretching mode , around 200 @xmath2 , needs two correlators to be properly reproduced .
one correlator , peaked at about 180 @xmath2 ( blue line ) , shows an almost temperature independent peak amplitude and a large spectral component extending to very low frequency , indicative of strong coupling to the structural dynamics .
the other correlator , peaked at about 225 @xmath2 ( orange line ) , appears to be substantially uncoupled to the structural dynamics ; its amplitude increases with decreasing temperature .
our hd - oke measurement represents the first clear experimental determination of this feature , even if it was inferred in a previous raman study@xcite .
the bending mode at about 50 @xmath2 can be described by a single correlator ( magenta line in figure [ fig_4 ] ) that turns out to be decoupled from the structural relaxation ( the peak close to 0 frequency ) .
it is possible that also the bending peak develops at low temperature a second component , but very likely the frequency separation is not sufficient to be resolved from our data analysis .
a comparison between these results and the recent debate on the nature of water anomalies is due .
our data and analysis show the presence of two distinct modes responsible for the part of the water spectrum normally attributed to the inter - molecular stretching vibrations .
raman spectra of low and high - density amorphous ice ( lda and hda ) in this frequency region differ from that of liquid water @xcite .
lda has a broad peak centred at higher frequency , about 220 @xmath2 , hda shows an even broader peak centered at about 185 @xmath2 ( see for example figure 1 in ref .
@xcite ) whereas the crystalline ice phases typically have a narrow peak at 230 @xmath2 .
this suggests that the modes observed in our experimental investigations can be associated with two fluctuating water species with different local structures ; a low - density form ( ld ) characterized by tetrahedral network and four - coordinated molecules similar to the lda structure , and a high - density ( hd ) form characterized by closely packed aggregates with lower coordination and high network distortions .
the mode at 225 @xmath2 can be attributed to the ld form and the mode at 180 @xmath2 to the hd one .
the present data strongly support the existence of the ld and hd water species .
nevertheless , as the observed water dynamic is measured at temperature / pressure relatively far from the supposed llcp , some ambiguity remains about the nature of the critical phenomena taking place into the supercooled phase .
the spectroscopic features attributed to the ld and hd forms could be attributed either to the coexistence of two extended liquid phases leading to the liquid - liquid phase transition@xcite , or to the continuous transition from a normal , unstructured liquid to a tetrahedrally structured liquid@xcite .
our new experimental data and analysis clearly show the presence of the high - frequency satellite mode contributing to the intermolecular stretching band of water , whose amplitude increases noticeably when the sample is cooled into the supercooled metastable phase .
the nature of this mode , peaked at 225 @xmath2 , appears clearly different from that of the nearby mode at 180 @xmath2 .
in fact , the former is strongly temperature dependent and only weakly coupled to the structural relaxation processes , while the latter is almost temperature independent and strongly coupled to structural rearrangements .
the surprising similarities of the spectra of lda and hda glass forms with those of the two high frequency distinct modes , reported in this work , suggest attributing them to the ld and hd water structures , respectively .
the ld water forms are expected to consist of aggregates of a small number of molecules ; being mostly localized in space ( at least in the temperature range investigated ) , their dynamics show negligible coupling with the structural relaxation .
the two local configurations of liquid water coexist in the entire temperature interval considered , above and below the melting temperature .
the number / lifetime of the ld species must increase with decreasing temperature in order to account for the amplitude increase of the corresponding correlator revealed by our analysis . on the other hand , the vibrational dynamics of the fraction of hd water
appears dominated by the bare caging effect that merges into the structural relaxation , a behaviour typical of fragile glass - former liquids .
we end by remarking that our findings are consistent with the conclusion drawn recently by nilsson et al.@xcite on the basis of x - ray emission and absorption experiments , and by overduin and patey@xcite from the analysis of their computer simulations . in both papers
large density fluctuations in liquid water are attributed to a short range bimodal aggregation of molecules with disordered and tetrahedral local environments .
supercooling bulk water is not an easy task because of crystallization .
temperatures as low as 20 degrees below melting can be achieved only with particularly pure sample and using specific attention for the cooling procedure . in our work
we performed the measurements on bi - distilled water prepared by the angelini company in a sealed vial of cylindrical shape , which enables us to reach a temperature as low as 247 k. the sample temperature is controlled with a stability of 0.1 k by means of a home - made cryostat .
the vial is inserted into a parallelepiped - shaped aluminum holder , with a central cylindrical housing whose diameter is close to that of the vial .
two fused silica windows on opposite sides of the aluminum holder allow the beams to cross the sample . a thin film of glycerol between the vial and the housing assures an efficient heat transfer .
the holder is fixed to the cold plate of a peltier cooler and the temperature is controlled by a platinum thermoresistance in thermal contact with the holder itself .
we used the optical configuration introduced by giraud et al.@xcite , which enables a reliable and efficient heterodyne - detection of the hd - oke signal ; a scheme of the optical set - up is shown in figure [ fig_5 ] .
the laser system utilized is a self - mode - locked ti : sapphire laser ( by femtolasers , mod .
fusion ) producing pulses of 20 fs duration with an energy of 3 nj ; 77.47 mhz repetition rate , 800 nm wavelength . other details on the experimental set - up are reported in ref .
@xcite .
the signal measured in an hd - oke experiment is @xcite : @xmath24 g(t ' ) dt ' \label{signalfit}\\ ] ] where @xmath25 is the instantaneous electronic response and and @xmath5 is the nuclear response , see eq.[signaleq ] .
@xmath6 is the instrumental function accounting for the finite time resolution determined by the time width of the pump and probe pulses . under specific conditions
@xcite , @xmath6 corresponds to the temporal cross - correlation of the laser pulse intensities .
the measurement of the real instrumental function @xmath6 is mandatory to extract the true oke response of water but it is far from trivial . by definition ,
the instrumental function in a hd - oke experiment is the signal measured in a system with response function @xmath26 , as directly inferred from eq.[signalfit ] . in practice
, this condition can be fulfilled by choosing a sample with negligible or instantaneous nuclear response . in most of the papers published on this subject in recent years a fused silica plate has been used to this purpose , as its nuclear contribution is considered negligible . in reality
, silica presents a fast and weak , but not null , nuclear contribution that prevents from measuring the real instrumental function .
alternatively , the instrumental function has been obtained by measuring the second harmonic cross - correlation of the laser pulses@xcite .
this procedure , however , implicates some changes and adjustments in the oke set - up that potentially modify the response function .
actually , any even small changes of the experimental conditions ( e.g. a different beam alignment or a modified angle of superposition of the beams inside the sample ) affect the instrumental function .
therefore , any method for obtaining the instrumental function which involves changes of the set - up is unsafe and can yield an incorrect time profile .
in other hd - oke investigations the response function has been obtained simply by fitting the electronic contribution to the hd - oke signal with an analytic peak function ( like gaussian , hyperbolic secant , etc . ) .
none of these methods enable the measurement of response function with the accuracy required by hd - oke investigations in the water sample , which is characterized by a fast and low level signal .
we extracted the instrumental function by performing hd - oke measurements in a reference sample .
this procedure allows us to accurately preserve the experimental conditions used during the water measurements . as the reference sample we chose a plate of calcium fluoride ( caf@xmath27 ) .
this a cubic ionic crystal which is optically isotropic and , in the frequency range probed in our experiment , has only one raman active band at 322 @xmath2 , due to the optical phonon t@xmath28 symmetry .
the calcium fluoride plate was dipped inside a water vial , identical to that used for water measurements , supported by the same sample holder .
the measurements of the instrumental function were done by just replacing the water vial with the water - caf@xmath27 vial , leaving the rest of the set - up unchanged .
the thickness of the caf@xmath27 plate , 3 mm , was enough to fully contain the probe and pump overlap volume , avoiding any spurious signal contribution by the outer water .
we took a reference measurement for each set of water data .
maximum attention was taken to preserve the experimental conditions between the water and the reference measurement .
figure [ fig_6 ] reports a typical hd - oke signal obtained in the caf@xmath27 reference sample .
we performed a least square fitting of the hd - oke data of the reference sample with the simulated signal , according to eq.[signalfit ] , where the nuclear response is a single harmonic damped oscillator .
we found that the instrumental function requires the sum of several analytical functions , typically a combination of gaussian , lorentzian and/or hyperbolic secant functions . in figure [ fig_6 ]
we report the instrumental function obtained by the iterative fitting procedure ( continuous line ) and the one obtained by fitting the electronic peak with a simple hyperbolic secant function ( dashed line ) .
apparently these two instrumental functions are very similar however the small differences can not be neglected for an accurate investigation of water fast hd - oke response .
this is mostly evident when the data are fourier transformed to the frequency domain . in fact
obtaining the imaginary part of the frequency dependent response from the time domain data involves the division by the fourier transform of the instrumental function .
figure [ fig_7 ] shows the comparison between the response calculated using the instrumental function obtained by the iterative fitting procedure ( continuous line ) and the one obtained by simply fitting the electronic peak with a hyperbolic secant function ( dashed line ) .
we clearly see that only an accurate measurement of the real time domain instrumental function yields the correct response . to our knowledge , this aspect of the oke data analysis has never been addressed in such a quantitative way , and the present level of accuracy in analyzing the role of the instrumental function in hd - oke experiments is unprecedented in recent literature . in particular , we want to stress that the knowledge of the correct instrumental function is fundamental in order to extract the true oke response , especially for the short delay time range .
this is particularly critical for weak signals characterized by complex relaxation dynamics , as in the case of water . according to eq.[mct - response ] , in our analysis of the hd - oke data we are decomposing the electronic polarizability correlator , @xmath29 , into the sum of @xmath30 correlators .
it is implicit that the individual slave correlators can not be directly associated to specific physical variables .
each of these correlators in fact describes an average collective mode " whose dynamics is described by the smc equations . the vibrational and relaxation features and coupling processes
are included in the smc equations by definition . in this respect
the present smc equations represent a robust physical model capable of describing a complex dynamics including vibrational and structural relaxations , implicitly accounting for their mutual coupling . differently from other approaches ,
the method does not require any decoupling or dynamic separation between the fast ( vibrational ) and the slow ( relaxation ) processes .
thus , once the temporal relaxation function of the master correlator is known , it can be used to calculate the relaxation of the slave correlators and , eventually , that of the oke signal .
we found that the oke data can be described in the entire temperature range by employing three slave correlators at most .
in particular , the weight of the highest frequency contribution decreases monotonically with increasing temperature , becoming negligible at the two highest temperatures , where only two slave correlators are sufficient to reproduce the data .
the parameters of the model are sixteen : the master equation parameters @xmath14 , @xmath13 , @xmath31 and @xmath32 , the slave equations parameters @xmath33 , @xmath34 , @xmath35 with @xmath36 and finally the three amplitudes @xmath37 , @xmath38 , and @xmath39 . the integro - differential equations were solved numerically with a step by step second order runge - kutta algorithm .
preliminary series of fits provided a qualitative estimate of the temperature behavior of the parameters . in agreement with similar analyses reported in the literature @xcite , in the subsequent step we forced some of the master parameters either to assume fixed values or to follow simple temperature trends , leaving all the slave parameters free . in figure [ fig_8 ]
we show the temperature dependence of the slave frequencies @xmath34 and the friction parameters @xmath14 and @xmath33 , and of the three vertices @xmath40 .
the temperature behavior of the fitting parameters is consistent with the results obtained by the previous studies in glass - forming liquids @xcite .
the contribution from all the authors was fundamental to realize this study .
more specifically : a.t .
, r.e . and r.t .
planned the experiments ; a.t .
, r.e . performed the experiments and analysed data ; r.r . and r.t .
interpreted the data and write the paper in collaboration with the other authors .
the research has been performed at lens .
this work was supported by regione toscana por - cro - fse 2007 - 2013 by ec cost action mp0902-coinapo and by prin 2010/2011 n.2010erfkxl .
we would like to thank a. marcelli for helpful discussion .
we acknowledge m. de pas , a. montori and m. giuntini for providing their continuous assistance in the set - up of the electronics .
r. ballerini and a. hajeb for the accurate mechanical realizations .
andrea taschin , paolo bartolini , agnese marcelli , roberto righini , and renato torre . a comparative study on bulk and nanoconfined water by time - resolved optical kerr effect spectroscopy .
, page doi:10.1039/c3fd00060e , 2013 .
francesco mallamace , matteo broccio , carmelo corsaro , antoniofaraone , domenico majolino , valentina venuti , li liu , chung - yuan mou , and sow - hsin chen .
evidence of the existence of the low - density liquid phase in supercooled , confined water .
, 104:424428 , 2007 .
f. aliotta , c. vasi , g. maisano , d. majolino , f. mallamace , and p. migliardo .
role of h bond and cooperative effects in normal and supercooled water studied by anisotropic low frequency light scattering .
, 84:47314738 , 1986 .
s. palese , s. mukamel , r.j.d .
miller , and w.t .
interrogation of vibrational structure and line broadening of liquid water by raman - induced kerr effect measurements within multimode brownian oscillator model .
, 100:1038010388 , 1996 .
k. winkler , j. lindner , and p. vohringer .
low - frequency depolarized raman - spectral density of liquid water from femtosecond optical kerr - effect measurements : lineshape analysis of restricted translational modes .
, 4:21442155 , 2002 .
neil t. hunt , andrew a. jaye , and stephen r. meech .
ultrafast dynamics in complex fluids observed through the ultrafast optically - heterodyne - detected optical - kerr - effect ( ohd - oke ) .
, 9:21672180 , 2007 .
d. prevosto , p. bartolini , r. torre , m. ricci , a. taschin , s. capaccioli , m. lucchesi . , and p.rolla .
relaxation processes in a epoxy resin studied by time resolved optical kerr effect . , 66:11502 , 2002 .
congcong huang , t.m .
weiss , d. nordlund , k.t .
wikfeldt , l.g.m .
pettersson , and a. nilsson .
increasing correlation length in bulk supercooled h2o , d2o , and nacl solution determined from small angle x - ray scattering .
, 133(13):134504 , october 2010 .
g. giraud , c.m .
gordon , i.r .
dunkin , and k. wynne .
the effects of anion and cation substitution on the ultrafast solvent dynamics of ionic liquids : a time - resolved optical kerr - effect spectroscopic study . ,
119:464477 , 2003 . 1 ps ) , the heterodyne - detected optical kerr effect ( hd - oke ) signal shows oscillations due to vibrational dynamics , that become more defined as the temperature is lowered below the melting point . at longer delay times
, the signal is characterised by a monotonous decay , which is strongly temperature dependent and clearly non - exponential . for a readable display ,
only 6 temperatures ( 247 , 258 , 273 , 293 , 313 , 353 k ) out of the 11 measured ( the previous ones plus 253 , 263 , 268 , 278 , 333 k ) are reported ; the intensities are normalized at 0.7 ps .
the upper and lower insets show , in a log - linear plot , the 247 k hd - oke data at short and long delay times , respectively . ] with stretching parameter @xmath41 .
the relaxation times obtained with the rescaling procedure are shown in the panel * b * of the figure .
the new data ( red circles ) extend and confirm the previous data ( empty circles ) @xcite .
the temperature dependence of these data are in agreement with the power law : @xmath42 , with critical temperature @xmath43 k and critical exponent @xmath44 .
we report the relaxation times and the power law in a linearized plot , in order to visualize the scaling proprieties and the critical temperature . ] , is directly related to the frequency - dependent spectrum measured in the dynamic light scattering experiments .
the imaginary part of the fourier transform of the heterodyne - detected oke data ( open circle ) , after that the contribution of the instrumental function has been removed by de - convolution , and of schematic mode - coupling ( smc ) fits ( red line ) at 247 , 258 , 273 , 293 , 313 k are reported .
the smc model enables to disentangle the dynamical features that generate the simulated response function .
the contributions of the three slave correlators are reported in the figure as magenta - blu - orange lines . moreover
, a simple damped harmonic oscillator ( dark yellow ) has been added to the smc response to reproduce the very high frequency vibrational contributions ( @xmath45 400 @xmath2 ) ; this does not affect the response function in the lower frequency range , where the dynamics relevant to our investigation takes place .
the simulation of hd - oke data by smc model requires two vibrational modes in order to fit the intermolecular stretching band of water ( blu and orange lines ) .
the characteristics of these two modes are clearly different in terms of spectrum shape and temperature dependence . ] , and hence the accurate deconvolution of the hd - oke signal . in our experiments @xmath6 was obtained by measuring the hd - oke signal of a caf@xmath27 plate placed in a purposely designed holder which enables switching to the water sample without even touching the rest of optical set - up . in the figure
we show the hd - oke signal of caf@xmath27 ( circles ) and the instrumental function ( red line ) obtained with the fitting procedure described in the methods .
the nuclear response , @xmath46 , is modelled as the time - derivative of a single damped oscillator ; and the instrumental function , @xmath47 , is simulated as the sum of gaussian , lorentzian , and hyperbolic secant functions .
the comparison with the fit of the instantaneous electronic part performed using just a single hyperbolic secant ( blue line ) is reported too . ]
was obtained as the fourier transform of the hd - oke signal deconvoluted from the instrumental function : @xmath23 \propto im\lbrace ft\left [ s(t ) \right ] /ft\left [ g(t)\right ] \rbrace$ ] . as reference sample we used a calcium fluoride , caf@xmath27 , crystal plate
whose spectrum is well known and characterized by a single intermolecular vibration .
the simplicity of the caf@xmath27 response enables a reliable test of the instrumental function extracted from the time domain data .
the figure compares the caf@xmath27 frequency dependent response calculated using the instrumental function obtained by the iterative fitting procedure ( red line ) to the one obtained by fitting the electronic peak with a hyperbolic secant function ( blue line ) .
only the knowledge of the real instrumental function allows measuring the correct response . ] and the vertex @xmath31 of the master oscillator are almost constant in the whole temperature range , and we locked their values to 66 @xmath2 and @xmath48 , respectively .
the second vertex @xmath32 , instead , was found to increase almost linearly with decreasing temperature , and it was forced to obey the linear dependence @xmath49 .
the friction parameter @xmath14 and the remaining parameters of the slave oscillators were left free of varying . in the figure we report
the best fit values of the parameters .
panel * a * reports the frequencies of the slave oscillators ( circles , squares , and triangles refer to @xmath50 , @xmath51 , and @xmath52 , respectively ) . in panel
* b * we show the temperature behavior of the master and slave friction parameters ( diamond refers to the master oscillator , circles , squares , and triangles refer to @xmath50 , @xmath51 , and @xmath52 , respectively ) .
the temperature dependence of the vertices @xmath40 is reported in panel * c * ( circles , squares , and triangles refer to @xmath50 , @xmath51 , and @xmath52 , respectively ) . ] | the liquid and supercooled states of water show a series of anomalies whose nature is lively debated .
a key role is attributed to the formation of structural aggregates induced by critical phenomena occurring deep in the supercooled region ; the nature of the water anomalies and of the hidden critical processes remains elusive .
here we report a time - resolved optical kerr effect investigation of the vibrational dynamics and relaxation processes in supercooled bulk water .
the experiment measures the water intermolecular vibrations and the structural relaxation process in an extended temperature range , and with unprecedented data quality .
a mode - coupling analysis of the experimental data enables to characterize the intermolecular vibrational modes and their interplay with the structural relaxation process .
the results bring evidence of the coexistence of two local configurations , which are interpreted as high density and low density water forms , with an increasing weight of the latter at low temperatures . published in : * nature communication vol.4 , pag.2401 , 2013 ; doi : 10.1038/ncomms3401*. |
in recent times many models have been proposed to study the dynamics of granular materials like sand in different practical situations ( see e.g. @xcite for an overview ) . in many applications
one has to store this kind of materials in view of their later use , so that filling and emptying a container are crucial processes .
granular materials adapt their shape to the container ( like a fluid does ) , but in general the free surface of a heap strongly depends on its formation process , for example on intensity and dislocation of the source .
moreover , the pressure on the bottom of the structure does not grow linearly with the height of the pile , since part of it is released against the walls through arcs of grains , a fact that can even produce silos explosion and collapse .
in this paper we deal with the simple problem of pouring at low intensity granular matter into a silo of given cross - section @xmath0 : it is known from the experiments ( see e.g. @xcite ) that if the source is independent of time , the free surface of the growing heap evolves towards a well - defined profile which then retains its shape while growing with a constant velocity . the two - layer model of hadeler and kuttler @xcite , which basically describes the formation of sandpiles over an open bounded plane table , is a system of two partial differential equations for a standing layer @xmath1 and a small rolling layer @xmath2 of grains running down the slope .
it can be adapted in a natural way to the case of the silo problem ( see again @xcite and more specifically @xcite ) by adding a suitable boundary condition on the silo walls . if @xmath3 denotes the vertical source of material and @xmath4 the final time , then the model takes the form @xmath5 the nonlinear term which appears in both the equations with opposite sign expresses the exchange term between the two layers , @xmath6 being the maximal ( _ critical _ ) slope that the material can support without flowing down , @xmath7 and @xmath8 respectively the mobility and the collision rate parameters .
the boundary condition comes from the total mass conservation law ( when @xmath9 ) , which suggests @xmath10 but the first equation in ( [ hk ] ) is an advection equation for @xmath2 in the direction of @xmath11 , so that boundary conditions can not be imposed for the outgoing direction of its flow and condition ( [ bc ] ) reduces to a pure homogeneous neumann condition on @xmath1 .
existence and uniqueness results for the solution of a system similar to ( [ hk ] ) under general assumptions on the data has been recently discussed in @xcite .
if the source is constant in time the free profile is expected to evolve towards a _
similarity solution _ , according to the following definition .
we call a pair of functions @xmath12 a similarity solution of system ( [ hk ] ) in @xmath13 if there exists a positive constant @xmath14 such that the functions @xmath15 solve the system ( disregarding the initial condition ) .
it can be considered as a sort of equilibrium for the model : the rolling layer is constant in time , while the free surface keeps growing by a rigid traslation of its shape at a constant rate @xmath14 . in the toy one - dimensional ( 1d )
case for the cross - section , these quasi - stationary profiles @xmath16 can be expressed by closed integral formulas ( see @xcite and next section ) in terms of the source and of the other problem parameters . in two dimensions ( 2d )
it can be proved that similarity solutions exist , but their expressions are known only in special cases .
that is why in section 3 we discuss a finite element ( fe ) characterization of such solutions in the general case .
finite difference ( fd ) numerical schemes for the model of hadeler and kuttler have been studied in @xcite and @xcite in the case of growing sandpiles on a bounded open table , and in @xcite on a table partially bounded by vertical walls . in section 4 we will adapt such schemes to the present problem of silos in order to show through the experiments of section 5 that the growing heaps generated by the evolving model perfectly match the similarity solutions .
we recall the basic theorem of existence for similarity solutions in the case of a constant in time source term .
( @xcite ) assume @xmath17 ; then there exists a similarity solution @xmath16 for problem ( [ hk ] ) ( in the sense of definition 1 ) , with @xmath18 unique up to an additive constant and @xmath19 the main idea in the proof is to consider the basic properties of the flux function @xmath20 . if one looks for it in the form of a gradient ( @xmath21 ) , then its potential @xmath22 should solve the semidefinite neumann problem for the laplacian @xmath23 with @xmath24 , which has a solution ( unique up to an additive constant ) due to the zero - mean property of @xmath25
. then we can derive @xmath26 from @xmath22 , and also deduce : @xmath27 so that @xmath18 can be determined up to an additive constant .
formula ( [ vel ] ) says that the growth velocity of the similarity profile coincides with the average precipitation , that is with the mean value of the source intensity and is independent from the other parameters .
theorem 1 does not say anything about uniqueness : in principle other solution pairs @xmath16 could exist such that @xmath28 is not a gradient . anyway , numerical experiments of section 5 show that the solutions given by the previous theorem are the only significant ( physical ) ones , since the evolving profiles tend asymptotically to them .
+ in 1d the previous result yields explicit expressions for similarity solutions .
if for example @xmath13 coincides with the interval @xmath29 , one finds ( see @xcite for details ) : @xmath30 where @xmath31 such expressions give several informations about solutions : * if @xmath32 for any @xmath33 , then @xmath34 , @xmath35 and @xmath36 , that is the free surface grows remaining flat , as expected ; * if the source is not identically zero , then @xmath37 everywhere , even at the boundary , confirming what already stated about condition ( [ bc ] ) ;
* @xmath38 , that is the standing layer never exceeds the critical slope ; * the rolling layer thickness is directly proportional to the source intensity and inversely proportional to @xmath6 . in higher dimensions
explicit formulas for similarity solutions can not be deduced in general , and we will see in the next section how to detect them numerically . here
we just report the two special cases of a central point source , for 1d and 2d radial cross - sections respectively , where similarity solutions can be explicitly computed .
* example 1.*(@xcite ) assume @xmath39 and @xmath40 ( where @xmath41 denotes the usual dirac function centered in @xmath42 ) , that is there is a point source placed over the middle of the silo .
similarity solutions then take the form ( see figure [ sol1ddelta ] ) : @xmath43 @xmath44 * example 2.*(@xcite ) let @xmath45 be the ball of radius @xmath46 centered at the origin , and @xmath47 a point source over its center ; then similarity solutions are radial functions , and radial symmetry arguments yields ( see figure [ sol3d ] ) : @xmath48 previous examples show that the typical profile of a growing symmetric heap of grains in a silo under the effect of a central source is different from the classical conical shape which would emerge on an open table without walls . in both dimensions
the standing layer now assumes a strictly convex ( logarithmic ) profile
. the maximal slope of the pile changes instead with the dimension : in 2d it is reached right in the center , and coincides with the critical slope @xmath6 , whereas in 1d it depends on the parameters @xmath49 and on the size @xmath50 of the container .
in particular , when the ratio @xmath51 ( grains roll slowly and are easily trapped ) the slope of the pile always remains very close to the critical angle @xmath6 . on the contrary , when @xmath52 is large the grains move very fast from the beginning , and larger variations of the slope can emerge . for what concerns the rolling layer in 2d , figure [ sol3d ] shows the emergence of a singularity in the center , in accordance with the fact that its expression comes from the solution of a potential problem with a central dirac source .
for the sake of simplicity from now on we assume @xmath53 .
the proof of theorem 1 in the previous section shows that in order to characterize the similarity profiles one needs to solve in @xmath13 the elliptic neumann problem ( [ neum ] ) . from a numerical point of view , this can be done for example by using a finite element approach .
if @xmath54 denotes a regular triangulation of @xmath13 of size @xmath55 , and @xmath56 and @xmath57 are the finite element spaces of respectively piecewise linear and piecewise constant functions on @xmath54 , galerkin method requires to solve the discrete variational problem@xmath58 however , if @xmath13 is not a polygonal domain it has to be replaced in ( [ discr_neum ] ) by a suitable set @xmath59 defined as the union of the triangular elements of @xmath54 ; @xmath59 will be closed to @xmath13 , but in general the right - hand side @xmath25 will not retain its zero - mean property on @xmath59 , and the discrete semidefinite problem will not be solvable at all . a way to overcome
this difficulty is to replace @xmath25 in ( [ discr_neum ] ) by the function @xmath60 , with @xmath61 by construction , @xmath62 , and @xmath63 for any @xmath64 , so that ( [ discr_neum ] ) becomes solvable ( see @xcite for details ) .
then by definition @xmath65 will be a piecewise constant vector on @xmath54 , and its norm an element of the discrete space @xmath66 .
hence , from ( [ vfromw ] ) , @xmath67 now , since @xmath68 , the gradient of @xmath69 on any triangle @xmath70 is given by @xmath71 it remains to compute @xmath72 . in 1d its value at any node @xmath73 can be determined by direct integration of the piecewise constant function @xmath74 from 0 to @xmath75 ( which corresponds to choose the particular solution vanishing at the origin ) : @xmath76 in higher dimension a different strategy can be used : plugging @xmath77 into ( [ discr_neum ] ) , @xmath69 becomes the solution of the discrete variational neumann problem ( with @xmath78 given ) @xmath79
in this section we want to show that the previously characterized similarity solutions asymptotically arise as profiles of the growing heaps in the dynamic process of filling the silo . in order to do that we implemented a numerical scheme for the complete system ( [ hk ] ) , adapting to this case the finite difference scheme used in @xcite for the growing sandpiles . in 1d , if @xmath39 and @xmath80 denotes the space discretization step , a uniform mesh is described by the nodes @xmath81 , for @xmath82 .
if @xmath83 is the time step , our explicit scheme reads for the internal nodes @xmath84 as @xmath85 where @xmath86 denote respectively the approximate values at time @xmath87 in @xmath75 of the solutions @xmath88 and of the upwind flux derivative @xmath89 in the direction determined by the sign of @xmath90 , that is @xmath91 ( in each node the spatial derivative @xmath90 is defined as the term of maximal absolute value between the backward and the forward first differences ) . to complete the scheme we added initial conditions ( @xmath92 ) and boundary terms induced by the neumann condition on @xmath1 .
the extension of this approach to the 2d case is straightforward if we restrict ourselves to square or rectangular cross - sections for the silo .
it is enough to decompose the flux term as @xmath93 , and to repeat the 1d approach in each direction . in order to study the asymptotic behavior of the growing heaps , in the numerical tests the scheme was stopped when the relative growth per iteration of the standing layer resulted approximately the same at each node , revealing the emergence of a similarity profile .
such profile was then translated to the base of the silo and compared with the computed similarity solution .
for the 1d cross - section case we assumed @xmath94 and tested different choices of the source term @xmath3 . in each example
, for a given uniform partition ( of step @xmath64 ) of @xmath13 , we compared the exact similarity solutions given by formulas ( [ sim1 ] ) ( the couple @xmath95 , with @xmath96 ) , the discrete similarity solutions computed by the fe approach of section 3 ( @xmath97 , with @xmath98 ) , and the stabilized profiles determined by the fd scheme for the evolutive problem described in section 4 ( @xmath99 , with the first one shifted towards zero , that is @xmath100 , where @xmath101 is the iterate selected by the stopping criterion ) .
figure [ grow1d ] shows three examples of growing heaps according to different source supports @xmath102 ( centered symmetric , close to the boundary , disconnected ) . in all the cases
it can be seen the formation of a similarity solution ( in dot lines ) .
we found first order convergence of the approximate similarity solutions to the exact ones for the fe method in uniform norm , and approximately the same order for the asymptotic convergence of the growing profiles computed by the fd scheme to the quasi - stationary ones . in table 1
we report the values found for the symmetric centered support case of figure [ sol1dcentral ] .
other tests gave similar results .
note that the stabilized rolling layer is everywhere positive , with a small depression in the central region corresponding to the source support .
when its length tends to zero one recovers the situation of the point source of example 1 , that is @xmath1 assumes the known logarithmic profile and the depression region of @xmath2 disappears . in the more realistic case of a spatial silo , that is when the cross - section @xmath13 is a 2d domain ,
the similarity solutions can only be approximated , so that we just estimated the quantities @xmath104 and @xmath105 , where as before @xmath97 denote the fe solutions of the stationary model and @xmath99 the fd solutions of the evolutive model after the stopping criterion applies , computed over the same mesh introduced in @xmath13 .
we restricted our tests to the case of a rectangular domain @xmath13 decomposed through a uniform mesh , in order to use the same set of nodes for the two schemes .
if for example @xmath106 , a mesh with @xmath107 equispaced nodes ( with @xmath108 for the fd scheme can be used as well as a base for a uniform courant fe triangulation over @xmath13 .
the experiments gave results similar to those of the 1d case .
figures [ cen2d ] and [ scon2d ] illustrate the results corresponding to a source supported in a small ball in the center of the silo or in the union of two disconnected balls , showing the profiles of the standing layers and the level lines of the rolling layers .
the correspondence of the similarity solutions ( above ) to the evolving profiles ( below ) appears evident .
figure [ riemp ] shows the growing heap in the square silo for the first example at four successive time steps . | the problem of filling a silo of given bounded cross - section with granular matter can be described by the two - layer model of hadeler and kuttler @xcite . in this paper
we discuss how similarity quasi - static solutions for this model can be numerically characterized by the direct finite element solution of a semidefinite elliptic neumann problem .
we also discuss a finite difference scheme for the dynamical model through which we can show that the growing profiles of the heaps in the silo evolve in finite time towards such similarity solutions .
_ keywords _ : granular matter , finite difference schemes , finite element schemes |
study of passage through an isolated resonance in a multi - frequency quasi - linear hamiltonian system can be reduced to the case of one - frequency system ( see , e.g. , @xcite ) .
the corresponding hamiltonian has the form @xmath3 here @xmath4 , @xmath5 mod @xmath6 are conjugate canonical variables , @xmath7 is a slow time , @xmath8 , and @xmath1 is a small parameter , @xmath9 .
equations of motion are @xmath10 for @xmath11 we get an unperturbed system with the hamiltonian @xmath12 and action - angle variables @xmath4 , @xmath5 .
the function @xmath13 is the frequency of the unperturbed motion .
for some value of the slow time @xmath14 , where there is a resonance , @xmath13 vanishes : @xmath15 .
we assume that the resonance is non - degenerate : @xmath16 . here
`` prime '' denotes the derivative with respect to @xmath7 .
let , for definiteness , @xmath17 .
we assume that @xmath14 is the only resonant moment of the slow time : @xmath18 is different from 0 at @xmath19 .
action @xmath4 is an adiabatic invariant : its changes along trajectory of ( [ motion ] ) are small over long time intervals . for motion
far from the resonance value @xmath4 oscillates with an amplitude @xmath0 .
passage through a narrow neighbourhood of the resonance leads to a change in @xmath4 of order @xmath20 ( so called jump of the adiabatic invariant ) .
there is an asymptotic formula for this jump ( @xcite , @xcite ) .
let @xmath21 and @xmath22 be values of @xmath4 along a trajectory of ( [ motion ] ) at moments of slow time @xmath23 and @xmath24 , where @xmath25 .
then @xmath26 here @xmath27 is the value of @xmath5 on the considered trajectory at @xmath28 .
there are formulas for change of the angle ( phase ) @xmath5 due to passage through the resonance as well @xcite .
one can replace in the left hand side of ( [ difference ] ) @xmath29 with values of the improved adiabatic invariant @xmath30 , but the error estimate in ( [ difference ] ) still will be @xmath31 .
it was suggested in @xcite to eliminate an asymmetry in ( [ difference ] ) by replacing in the right hand side @xmath21 with @xmath32 , where @xmath32 is the value of @xmath4 on the considered trajectory at @xmath28 .
a numerical simulation in @xcite shows that this symmetrization indeed improves the accuracy of formula ( [ difference ] ) for @xmath29 replaced with @xmath30 considerably .
it is conjectured in @xcite on the basis of the numerical simulation that the error term in the modified formula is @xmath2 . in the current paper
we prove this conjecture by means of a hamiltonian adiabatic perturbation theory .
we show that this improvement of accuracy occurs due to cancellations of many terms in formulas of the perturbation theory considered up to terms of 4th order in @xmath1 .
we obtain also formulas which describe change of phase @xmath5 due to passage through resonance with the same accuracy @xmath2 . as a result
, we obtain formulas which allow to predict motion in post - resonance region with accuracy @xmath2 , provided that the motion in the pre - resonance region is known . in the last section we provide a numerical verification of these formulas .
we consider hamiltonian system ( [ motion ] ) with hamilton s function ( [ ham ] ) .
we assume that the function @xmath33 is of class @xmath34 for @xmath35 , where @xmath36 and @xmath37 are some open intervals in @xmath38 .
we assume that @xmath39 is @xmath6periodic in @xmath5 and that the frequency @xmath13 does not vanish in @xmath37 other than at @xmath28 . at the resonance state ,
@xmath15 , but @xmath40 .
let @xmath41 , @xmath42 be a solution of ( [ motion ] ) on a time interval @xmath43 $ ] , where @xmath44 , and @xmath45 are some constants .
let @xmath25 .
we denote @xmath46 , @xmath47 , and @xmath48 , @xmath49 . here @xmath50 .
[ thm1 ] @xmath51 [ thm2 ] @xmath52 in the above theorems , @xmath53 and angular brackets denote averaging with respect to @xmath5 : @xmath54 .
moreover , @xmath55 the new results here are : * estimate @xmath2 in the first formula of theorem [ thm1 ] , * the last two terms and the estimate @xmath2 in the second formula in theorem [ thm1 ] , * the last term and the estimate @xmath31 in the second formula in theorem [ thm2 ] .
combining results of theorems [ thm1 ] , [ thm2 ] , we can obtain prediction of motion in the post - resonance region with accuracy @xmath2 as follows : [ cor1 ] @xmath56 where @xmath57
in order to get the above estimates , four steps of adiabatic perturbation theory are performed in both original hamiltonian system ( in subsection [ s3.1 ] ) and an approximate hamiltonian system ( in subsection [ s3.2 ] ) .
we follow an approach of @xcite here .
consider dynamics described by the hamiltonian @xmath58 the frequency @xmath18 can be expanded near the resonance as @xmath59 with @xmath60 .
so we can get formula for @xmath5 as @xmath61 let @xmath62 where @xmath63 is @xmath6-periodic in @xmath5 and @xmath64 . here
we have not defined @xmath65 yet .
make the canonical transformation of variables @xmath66 with the generating function @xmath67 .
old and new variables are related as follows : @xmath68 the new hamiltonian , which describes dynamics of variables @xmath69 , @xmath70 , is @xmath71 we would like to find @xmath72 such that there is no dependence on the new phase @xmath70 in the new hamiltonian @xmath73 in terms up to 4th order in @xmath1 .
thus the new hamiltonian should have a form @xmath74 here we have not defined functions @xmath75 yet . equating terms of the same order in @xmath1 in ( [ hnew1 ] ) and ( [ hnew2 ] )
we find that functions @xmath76 have forms : @xmath77 where @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 are smooth functions
. denote @xmath85 we can find the explicit form of @xmath65 : @xmath86 satisfying @xmath87 , @xmath88 .
so we have the expressions @xmath89 also @xmath90 the new hamiltonian is @xmath91 for @xmath92 , the motion is described by differential equations : @xmath93 here @xmath94 , @xmath95 , @xmath96 are smooth functions
. * remark .
* by differentiating @xmath97 with respect to @xmath69 on both sides , we obtain @xmath98
. therefore , @xmath99 .
now let us consider the approximate hamiltonian @xmath100 equations of motion are @xmath101 here @xmath102 , @xmath60 .
we will consider the solution of these equations with initial conditions at resonance , i.e. when @xmath28 : @xmath103 , @xmath104 .
we get the formula for @xmath105 as @xmath106 let @xmath107 where @xmath108 is @xmath6-periodic in @xmath105 and @xmath109 . here
we have not defined @xmath110 yet .
make the canonical transformation of variables @xmath111 with a generating function @xmath112 .
the old and new variables are related as follows : @xmath113 the new hamiltonian , which describes dynamics of variables @xmath114 , @xmath105 , is @xmath115 we would like to find @xmath116 such that there is no dependence on the phase @xmath105 in the new hamiltonian @xmath117 in terms up to 4th order in @xmath1 .
thus the new hamiltonian should have a form @xmath118 equating terms of the same order in @xmath1 in ( [ hhnew1 ] ) and ( [ hhnew2 ] ) we find that functions @xmath119 have forms @xmath120 the new hamiltonian is @xmath121 the motion is described by differential equations @xmath122 here @xmath123 is a smooth function .
we will prove asymptotic formulas for the action variable @xmath4 in both theorems [ thm1 ] and [ thm2 ] first , and then asymptotic formulas for the angle variable @xmath5 in these theorems .
denote @xmath124 , @xmath125 , @xmath126 .
denote @xmath127 and @xmath128 . in the following text
, we use the notations @xmath129 , where @xmath130 and @xmath131 , @xmath132 . for simplicity of the exposition we will assume that @xmath24 and @xmath23 are symmetric with respect to @xmath14 : @xmath133 .
general case can be easily reduced to this one .
the following lemmas will be used in the proof .
[ err_omega ] @xmath134 where @xmath135 .
[ est ] @xmath136 [ est_comb ] @xmath137 @xmath138 for @xmath139 $ ] , substitute with formulas ( [ j ] ) and ( [ dsdphi ] ) : @xmath140 @xmath138 for @xmath141 $ ] , substitute with formulas ( [ phi ] ) and ( [ phia ] ) : @xmath142+\frac1{\varepsilon}o(\tau_l-\tau_*)^4+\int\limits_{\tau_*}^{\tau_l}\frac{\partial h_1(i,{\varphi},\tau_1)}{\partial i}\,{\mathrm{d}}\tau_1+\int\limits_{\tau_*}^{\tau_r}\frac{\partial h_1(i,{\varphi},\tau_1)}{\partial i}\,{\mathrm{d}}\tau_1\\ & & = \int\limits_{\tau_*}^{\tau_l}\frac{\partial h_1(i_*,{\varphi}_a,\tau_*)}{\partial i}\,{\mathrm{d}}\tau_1+\int\limits_{\tau_*}^{\tau_r}\frac{\partial h_1(i_*,{\varphi}_a,\tau_*)}{\partial i}\,{\mathrm{d}}\tau_1+o({\varepsilon})\\ & & = o({\varepsilon})\end{aligned}\ ] ] [ est_frac ] @xmath143 if @xmath144 .
here @xmath145 , @xmath146 , and @xmath147 is a smooth function .
the same estimate is valid if @xmath18 is replaced with @xmath148 .
[ est_int ] @xmath149 if @xmath150 $ ] .
here @xmath145 , @xmath151 , @xmath152 , and @xmath147 is a smooth function .
the same estimate is valid if @xmath153 is replaced with @xmath154 and @xmath155 $ ] .
the same estimate is valid if @xmath18 is replaced with @xmath148 .
[ cancnear ] @xmath156 where @xmath157 and @xmath158 is a smooth function .
[ cancfar ] @xmath159 where @xmath160 and @xmath161 is a smooth function .
[ varbi ] let @xmath162 be a twice continuously differentiable function . then @xmath163 @xmath164 [ varbj ] let @xmath165 be a twice continuously differentiable function . then for @xmath166\cup[\tau_r,\tau_+]$ ] , @xmath167 where @xmath168 , @xmath169 , @xmath170 , @xmath171 are smooth functions .
@xmath172 then @xmath173 also @xmath174 this implies the result of the lemma .
let @xmath175 we have the identity : @xmath176 we should prove that @xmath177 , @xmath178 , and @xmath179 .
here we use the sum @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath132 . for convenience ,
we consider @xmath185 as the main part of expression of @xmath186 , then consider the others .
@xmath187\alpha^1(j_l,{\varphi}_l,\tau_l ) -\omega(\tau_l)\alpha^1(i_*,{\varphi}_{a_l},\tau_*)}{\omega(\tau_l)\omega(\tau_l)}\\ & & -{\varepsilon}\frac{\big[\omega(\tau_r)-\frac12\omega''_*(\tau_r-\tau_*)^2 + o(\tau_r-\tau_*)^3\big]\alpha^1(j_r,{\varphi}_r,\tau_r)-\omega(\tau_r)\alpha^1(i_*,{\varphi}_{a_r},\tau_*)}{\omega(\tau_r)\omega(\tau_r)}\\ & = & { \varepsilon}\frac{\omega(\tau_l)\big[\alpha^1(j_l,{\varphi}_l,\tau_l)-\alpha^1(i_*,{\varphi}_{a_l},\tau_*)\big]-\frac{\omega''_*(\tau_l-\tau_*)^2}2\alpha^1(j_l,{\varphi}_l,\tau_l)+o(\tau_l-\tau_*)^3}{\omega(\tau_l)\omega(\tau_l)}\\ & & -{\varepsilon}\frac{\omega(\tau_r)\big[\alpha^1(j_r,{\varphi}_r,\tau_r)-\alpha^1(i_*,{\varphi}_{a_r},\tau_*)\big]-\frac{\omega''_*(\tau_r-\tau_*)^2}2\alpha^1(j_r,{\varphi}_r,\tau_r)+o(\tau_r-\tau_*)^3}{\omega(\tau_r)\omega(\tau_r)}\\ & = & { \varepsilon}\frac{\alpha^1(j_l,{\varphi}_l,\tau_l)-\alpha^1(i_*,{\varphi}_{a_l},\tau_*)}{\omega(\tau_l ) } -{\varepsilon}\frac{\omega''_*(\tau_l-\tau_*)^2\alpha^1(j_l,{\varphi}_l,\tau_l)}{2\omega(\tau_l)\omega(\tau_l)}\\ & & -{\varepsilon}\frac{\alpha^1(j_r,{\varphi}_r,\tau_r)-\alpha^1(i_*,{\varphi}_{a_r},\tau_*)}{\omega(\tau_r ) } + { \varepsilon}\frac{\omega''_*(\tau_r-\tau_*)^2\alpha^1(j_r,{\varphi}_r,\tau_r)}{2\omega(\tau_r)\omega(\tau_r)}+o({\varepsilon}^{\frac32})\\ & = & \frac{{\varepsilon}}{\omega(\tau_l)}\big[\big(\alpha^1(j_l,{\varphi}_l,\tau_l)-\alpha^1(i_*,{\varphi}_{a_l},\tau_*)\big)+\big(\alpha^1(j_r,{\varphi}_r,\tau_r)-\alpha^1(i_*,{\varphi}_{a_r},\tau_*)\big)\big]\\ & & -\frac{\varepsilon}2\frac{\omega''_*(\tau_l-\tau_*)^2}{\omega(\tau_l)}\left(\frac{\alpha^1(j_l,{\varphi}_l,\tau_l)}{\omega(\tau_l)}+\frac{\alpha^1(j_r,{\varphi}_r,\tau_r)}{\omega(\tau_r)}\right)+o({\varepsilon}^{\frac32})\\ & = & o(\sqrt{\varepsilon})e_{1+}+o({\varepsilon}^{\frac32})\left(\frac{\alpha^1(j_l,{\varphi}_l,\tau_l)}{\omega(\tau_l)}+\frac{\alpha^1(j_l,{\varphi}_l,\tau_l)+o(\sqrt{\varepsilon})}{\omega(\tau_r)}\right)+o({\varepsilon}^{\frac32})\\ & = & o(\sqrt{\varepsilon})e_{1+}+o({\varepsilon}^{\frac32})\left(\alpha^1(j_l,{\varphi}_l,\tau_l)\cdot\frac{\omega(\tau_l)+\omega(\tau_r)}{\omega(\tau_l)\omega(\tau_r)}\right)+o({\varepsilon}^{\frac32})\\ & = & o(\sqrt{\varepsilon})e_{1+}+o({\varepsilon}^{\frac32})\end{aligned}\ ] ] here @xmath188+\big[\alpha^1(j_r,{\varphi}_r,\tau_r)-\alpha^1(i_*,{\varphi}_{a_r},\tau_*)\big]$ ] , and we are going to estimate @xmath189 using lemmas [ est ] and [ est_comb ] : @xmath190+\frac{\partial\alpha^1(i_*,{\varphi}_{a_l},\tau_*)}{\partial { \varphi}}\big[({\varphi}_l-{\varphi}_{a_l})+({\varphi}_r-{\varphi}_{a_r})\big]\\ & & { } + \frac{\partial\alpha^1(i_*,{\varphi}_{a_l},\tau_*)}{\partial \tau}\big[(\tau_l-\tau_*)+(\tau_r-\tau_*)\big]+o({\varphi}_{a_l}-{\varphi}_{a_r})^2+o({\varepsilon})\\ & = & o({\varepsilon})\end{aligned}\ ] ] therefore , @xmath191 .
then we consider the term @xmath192 in expression of @xmath186 . here
@xmath193 @xmath194 from lemmas [ est ] and [ est_frac ] , @xmath195+o({\varepsilon}^{\frac32})\\ & = & \frac{{\varepsilon}^2}{\omega^2(\tau_l)}\left[\frac{\partial\beta^2}{\partial j}(j_l - j_r)+\frac{\partial\beta^2}{\partial{\varphi}}({\varphi}_l-{\varphi}_r)+\frac{\partial\beta^2}{\partial\tau}(\tau_l-\tau_r)\right]+o({\varepsilon}^{\frac32})\\ & = & o({\varepsilon}^{\frac32})\end{aligned}\ ] ] also we know that @xmath196 ^ 3\\ & = & \omega^3(\tau)-\frac32\omega''_*\omega^2(\tau)(\tau-\tau_*)^2+o(\tau-\tau_*)^5.\end{aligned}\ ] ] so @xmath197\omega'(\tau_l)\alpha^2(j_l,{\varphi}_l,\tau_l)}{\omega^3(\tau_l)\omega^3(\tau_l)}\\ & & { } -{\varepsilon}^2\frac{\omega^3(\tau_l)\big[\omega'(\tau_l)-\omega''_*(\tau_l-\tau _ * ) + o(\tau_l-\tau_*)^2\big ] \alpha^2(i_*,{\varphi}_{a_l},\tau_*)}{\omega^3(\tau_l)\omega^3(\tau_l)}\\ & & { } -{\varepsilon}^2\frac{\big[\omega^3(\tau_r)-\frac32\omega''_*\omega^2(\tau_r)(\tau_r-\tau_*)^2+o(\tau_r-\tau_*)^5\big]\omega'(\tau_r)\alpha^2(j_r,{\varphi}_r,\tau_r)}{\omega^3(\tau_r)\omega^3(\tau_r)}\\ & & { } + { \varepsilon}^2\frac{\omega^3(\tau_r)\big[\omega'(\tau_r)-\omega''_*(\tau_r-\tau _ * ) + o(\tau_r-\tau_*)^2\big ] \alpha^2(i_*,{\varphi}_{a_r},\tau_*)}{\omega^3(\tau_r)\omega^3(\tau_r)}+o({\varepsilon}^{\frac32})\\ & = & { \varepsilon}^2\frac{\omega^3(\tau_l)\omega'(\tau_l)\big[\alpha^2(j_l,{\varphi}_l,\tau_l)-\alpha^2(i_*,{\varphi}_{a_l},\tau_*)\big]}{\omega^3(\tau_l)\omega^3(\tau_l)}\\ & & { } -{\varepsilon}^2\frac{\frac32\omega''_*\omega'(\tau_l)\omega^2(\tau_l)(\tau_l-\tau_*)^2\alpha^2(j_l,{\varphi}_l,\tau_l)}{\omega^3(\tau_l)\omega^3(\tau_l ) } + { \varepsilon}^2\frac{\omega''_*\omega^3(\tau_l)\alpha^2(i_*,{\varphi}_{a_l},\tau_*)(\tau_l-\tau_*)}{\omega^3(\tau_l)\omega^3(\tau_l)}\\ & & { } -{\varepsilon}^2\frac{\omega^3(\tau_r)\omega'(\tau_r)\big[\alpha^2(j_r,{\varphi}_r,\tau_r)-\alpha^2(i_*,{\varphi}_{a_r},\tau_*)\big]}{\omega^3(\tau_r)\omega^3(\tau_r)}\\ & & { } + { \varepsilon}^2\frac{\frac32\omega''_*\omega'(\tau_r)\omega^2(\tau_r)(\tau_r-\tau_*)^2\alpha^2(j_r,{\varphi}_r,\tau_r)}{\omega^3(\tau_r)\omega^3(\tau_r ) } -{\varepsilon}^2\frac{\omega''_*\omega^3(\tau_r)\alpha^2(i_*,{\varphi}_{a_r},\tau_*)(\tau_r-\tau_*)}{\omega^3(\tau_r)\omega^3(\tau_r)}+o({\varepsilon}^{\frac32})\\ & = & { \varepsilon}^2\frac{\omega'_*}{\omega^3(\tau_l)}\big[\alpha^2(j_l,{\varphi}_l,\tau_l)-\alpha^2(i_*,{\varphi}_{a_l},\tau_*)\big]-{\varepsilon}^2\frac{\omega'_*}{\omega^3(\tau_r)}\big[\alpha^2(j_r,{\varphi}_r,\tau_r)-\alpha^2(i_*,{\varphi}_{a_r},\tau_*)\big ] \\ & & { } -\frac32{\varepsilon}^2\frac{\omega''_*\omega'(\tau_l)}{\omega(\tau_l)\omega^3(\tau_l)}(\tau_l-\tau_*)^2\alpha^2(j_l,{\varphi}_l,\tau_l ) + \frac32{\varepsilon}^2\frac{\omega''_*\omega'(\tau_r)}{\omega(\tau_r)\omega^3(\tau_r)}(\tau_r-\tau_*)^2\alpha^2(j_r,{\varphi}_r,\tau_r ) \\ & & { } + { \varepsilon}^2\frac{\omega''_*}{\omega^3(\tau_l)}(\tau_l-\tau_*)\alpha^2(i_*,{\varphi}_{a_l},\tau_*)-{\varepsilon}^2\frac{\omega''_*}{\omega^3(\tau_r)}(\tau_r-\tau_*)\alpha^2(i_*,{\varphi}_{a_r},\tau_*)+o({\varepsilon}^{\frac32})\\\end{aligned}\ ] ] @xmath198\\ & & { } + { \varepsilon}^2\frac{\omega''_*(\tau_l-\tau_*)}{\omega^3(\tau_l)}\big[\alpha^2(j_l,{\varphi}_l,\tau_l)-\alpha^2(i_*,{\varphi}_{a_l},\tau_*)\big]-{\varepsilon}^2\frac{\omega''_*(\tau_r-\tau_*)}{\omega^3(\tau_r)}\big[\alpha^2(j_r,{\varphi}_r,\tau_r)-\alpha^2(i_*,{\varphi}_{a_r},\tau_*)\big]\\ & & { } -\frac32{\varepsilon}^2\frac{\omega''_*\omega'_*(\tau_l-\tau_*)^2}{\omega(\tau_l)\omega^3(\tau_l)}\big[\alpha^2(j_l,{\varphi}_l,\tau_l)-\alpha^2(j_r,{\varphi}_r,\tau_r)\big ] \\ & & { } + { \varepsilon}^2\frac{\omega''_*(\tau_l-\tau_*)}{\omega^3(\tau_l)}\big[\alpha^2(i_*,{\varphi}_{a_l},\tau_*)-\alpha^2(i_*,{\varphi}_{a_r},\tau_*)\big ] + o({\varepsilon}^{\frac32})\\ & = & o(\sqrt{\varepsilon})e_{2+}+o({\varepsilon})\big[\alpha^2(j_l,{\varphi}_l,\tau_l)-\alpha^2(i_*,{\varphi}_{a_l},\tau_*)\big]+o({\varepsilon})\big[\alpha^2(j_r,{\varphi}_r,\tau_r)-\alpha^2(i_*,{\varphi}_{a_r},\tau_*)\big]\\ & & { } + o({\varepsilon})\big[\alpha^2(j_l,{\varphi}_l,\tau_l)-\alpha^2(j_r,{\varphi}_r,\tau_r)\big]+o({\varepsilon})\big[\alpha^2(i_*,{\varphi}_{a_l},\tau_*)-\alpha^2(i_*,{\varphi}_{a_r},\tau_*)\big]+o({\varepsilon}^{\frac32})\\ & = & o(\sqrt{\varepsilon})e_{2+}+o({\varepsilon}^{\frac32}),\end{aligned}\ ] ] where @xmath199+\big[\alpha^2(j_r,{\varphi}_r,\tau_r)-\alpha^2(i_*,{\varphi}_{a_r},\tau_*)\big]$ ] .
similarly to @xmath200 , we can obtain @xmath201 from lemma [ est_comb ] .
therefore , it is true that @xmath202 .
similarly , we can derive that @xmath203 therefore , @xmath204 we simply use lemma [ varbi ] and lemma [ cancnear ] in order to get estimate : @xmath206 for combined term @xmath207 , we consider @xmath208 , where @xmath209 and also @xmath210 for @xmath211 , we apply lemma [ err_omega ] , then lemma [ varbj ] : @xmath212+{\varepsilon}^5\frac{\check\gamma(j,{\varphi},\tau)}{(\tau-\tau_*)^7}-{\varepsilon}^5\frac{\widetilde\gamma(i_*,{\varphi}_a,\tau_*)}{\omega^7(\tau)}+o\left(\frac{{\varepsilon}^5}{(\tau-\tau_*)^6}\right)\\ & = & \frac{{\varepsilon}^5\delta_1(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^8}\int\limits_{\tau_*}^{\tau}\frac{\partial h_1(i_*,{\varphi}_a,\tau_*)}{\partial{\varphi}}\,{\mathrm{d}}\tau_1 \\ & & { } + \frac{{\varepsilon}^4\delta_2(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^5 } + \frac{{\varepsilon}^5\delta_3(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^8}\int\limits_{\tau_*}^{\tau}\frac{\partial h_1(i_*,{\varphi}_a,\tau_*)}{\partial i}\,{\mathrm{d}}\tau_1 + \frac{{\varepsilon}^5\delta_4(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^7 } \\ & & { } + o\left(\frac{{\varepsilon}^7}{(\tau-\tau_*)^{10}}\right ) + o\left(\frac{{\varepsilon}^6}{(\tau-\tau_*)^8}\right ) + o\left(\frac{{\varepsilon}^5}{(\tau-\tau_*)^6}\right ) + o\left(\frac{{\varepsilon}^4}{(\tau-\tau_*)^4}\right ) + o\left(\frac{{\varepsilon}^3}{(\tau-\tau_*)^2}\right)\\ & & { } + \frac{{\varepsilon}^6\delta_5(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^9 } + \frac{{\varepsilon}^7\delta_6(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^{11 } } + \frac{{\varepsilon}^8\delta_7(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^{13 } } + \frac{{\varepsilon}^9\delta_8(i_*,{\varphi}_a,\tau_*)}{(\tau-\tau_*)^{15}}.\end{aligned}\ ] ] here @xmath213 is a smooth function . also @xmath214 using lemma [ est_int ] for estimate of integrals and lemma [ cancfar ] for cancellation in symetric intervals , we obtain : @xmath215 for term @xmath216 , we apply an integration by parts : @xmath217 and similarly @xmath218 therefore , the estimate @xmath219 is obtained . by joining estimate of terms @xmath220 together , and taking into account
that @xmath221 , as well as the identity @xmath222 where @xmath223 , we have finished the proof of the first formula of theorem [ thm1 ] .
let @xmath224 we have @xmath225 making use of lemmas [ err_omega ] , [ est ] , [ est_frac ] , [ est_int ] , [ varbi ] , [ varbj ] , one can show that @xmath226 , @xmath227 , @xmath228 , @xmath229 , @xmath230 .
this leads to the first formula of theorem [ thm2 ] with the sign @xmath231 " .
the proof of the formula with the sign
@xmath232 " is completely analogous .
[ replacenear ] for @xmath233 and smooth function @xmath234 , @xmath235 for @xmath144 , lemma [ varbj ] can be simplified as @xmath236 . thus with lemmas [ err_omega ] and [ est_frac ] , @xmath237 applying the result of ( a ) and lemma [ varbj ]
, we obtain @xmath238 here @xmath239 and @xmath240 is smooth functions .
[ replacefar ] let @xmath165 be a twice continuously differentiable function .
then for @xmath233 , @xmath241 , @xmath242 for @xmath150\cup[t_r , t_+]$ ] , we can simplify lemma [ varbj ] as follows : @xmath243 thus with lemmas [ err_omega ] , [ est_int ] and [ cancfar ] , @xmath244 applying the result of ( a ) and lemma [ varbj ] , we obtain @xmath245 here @xmath239 , @xmath240 and @xmath246 are smooth functions .
[ j - jpm ] for @xmath150 $ ] , @xmath247 for @xmath155 $ ] , similarly we have @xmath248 making use of of formula ( [ dotj ] ) and lemma [ varbj ] , for @xmath150 $ ] , we get @xmath249 similarly for @xmath155 $ ] , we obtain the second formula .
the relation between @xmath250 and @xmath251 is : @xmath252 let @xmath253 the identity becomes @xmath254 we will discuss the estimate term by term .
here @xmath256 , @xmath257 , @xmath132 .
making use of equations ( [ dsdj ] ) and lemma [ replacenear ] , we get @xmath258 using lemmas [ cancnear ] and [ varbi ] , we have @xmath260 [ cancij+- ] @xmath261 making use of lemmas [ varbi ] , [ varbj ] and [ j - jpm ] , we get @xmath262\,{\mathrm{d}}t \\ & & { } + { \varepsilon}\int\limits_{t_*}^{t_r}\left[{\varepsilon}\int\limits_{t}^{t_r}\frac{\partial h_1(i_*,{\varphi}_a,\tau_*)}{\partial{\varphi}}\,{\mathrm{d}}t_1+{\varepsilon}\frac{\alpha^1(j_r,{\varphi}_r,\tau_r)}{\omega(\tau_r)}+{\varepsilon}^2\frac{\omega'(\tau_r)\alpha^2(j_r,{\varphi}_r,\tau_r)}{\omega^3(\tau_r)}\right.\\ & & \qquad{}+{\varepsilon}^3\frac{\big(\omega'(\tau_r)\big)^2\alpha^3(j_r,{\varphi}_r,\tau_r)}{\omega^5(\tau_r)}+{\varepsilon}^4\frac{\big(\omega'(\tau_r)\big)^3\alpha^4(j_r,{\varphi}_r,\tau_r)}{\omega^7(\tau_r)}+\left.\int\limits_{t_+}^{t_r}\dot j\,{\mathrm{d}}t\right]\,{\mathrm{d}}t + o({\varepsilon}^{\frac32})\\ & = & { \varepsilon}\int\limits_{t_l}^{t_*}\bigg[{\varepsilon}\frac{\alpha^1(i_*,{\varphi}_{a_l},\tau_*)}{\omega(\tau_l)}+{\varepsilon}^2\frac{\omega'_*\alpha^2(i_*,{\varphi}_{a_l},\tau_*)}{\omega^3(\tau_l)}+{\varepsilon}^3\frac{{\omega'_*}^2\alpha^3(i_*,{\varphi}_{a_l},\tau_*)}{\omega^5(\tau_l)}+{\varepsilon}^4\frac{{\omega'_*}^3\alpha^4(i_*,{\varphi}_{a_l},\tau_*)}{\omega^7(\tau_l)}\\ & & \qquad{}+\int\limits_{t_-}^{t_l}{\varepsilon}^5\frac{7{\omega'_*}^4\alpha^4(i_*,{\varphi}_a,\tau_*)}{\omega^8(\tau_1)}\,{\mathrm{d}}t_1 + { \varepsilon}\int\limits_{t_*}^{t_r}\bigg[{\varepsilon}\frac{\alpha^1(i_*,{\varphi}_{a_r},\tau_*)}{\omega(\tau_r)}+{\varepsilon}^2\frac{\omega'_*\alpha^2(i_*,{\varphi}_{a_r},\tau_*)}{\omega^3(\tau_r)}\\ & & \qquad{}+{\varepsilon}^3\frac{{\omega'_*}^2\alpha^3(i_*,{\varphi}_{a_r},\tau_*)}{\omega^5(\tau_r)}+{\varepsilon}^4\frac{{\omega'_*}^3\alpha^4(i_*,{\varphi}_{a_r},\tau_*)}{\omega^7(\tau_r)}+\int\limits_{t_+}^{t_r}{\varepsilon}^5\frac{7{\omega'_*}^4\alpha^4(i_*,{\varphi}_a,\tau_*)}{\omega^8(\tau_1)}\,{\mathrm{d}}t_1\,\bigg]\,{\mathrm{d}}t+o({\varepsilon}^{\frac32})\\ & = & o({\varepsilon}^{\frac32}).\end{aligned}\ ] ] we also have ( cf . section [ e4 ] ) : @xmath263 using this result , formula ( [ e1phi ] ) and lemma [ cancij+- ] , we obtain @xmath264 we know that @xmath266 .
thus with lemma [ j - jpm ] , we have @xmath267 from lemma [ varbj ] , we get the error between @xmath69 and @xmath32 : @xmath269 as @xmath270 we have @xmath271 @xmath272\\ & & { } + { \varepsilon}^2\int\limits_{t_r}^{t_+}\left(-\frac1{2\omega}\right ) \frac{\partial^3\big\langle\widetilde h_1 ^
2\big\rangle^{\varphi}(j_-,\tau_*)}{\partial i^3 } \left(\,\int\limits_{-\infty}^{+\infty}\dot i_a\,{\mathrm{d}}t+o({\varepsilon}^{\frac32})\right)\,{\mathrm{d}}t + o({\varepsilon}^{\frac32})\\ & = & { \rm p.v.}\,{\varepsilon}^2\int\limits_{t_-}^{t_+}\frac{\partial{{{\cal r}}}_2(j_-,\tau)}{\partial i}\,{\mathrm{d}}t+{\varepsilon}^3\frac{\partial^3\big\langle\widetilde h_1 ^ 2\big\rangle^{\varphi}(j_-,\tau_*)}{\partial i^3}\int\limits_{-\infty}^{+\infty}\frac{\partial h_1(i_*,{\varphi}_a,\tau_*)}{\partial{\varphi}}\,{\mathrm{d}}t\int\limits_{t_r}^{t_+}\frac1{2\omega}\,{\mathrm{d}}t + o({\varepsilon}^{\frac32})\\ & = & { \rm p.v.}\,{\varepsilon}^2\int\limits_{t_-}^{t_+}\frac{\partial{{{\cal r}}}_2(i_*,\tau)}{\partial i}\,{\mathrm{d}}t-\frac{{\varepsilon}^2\ln{\varepsilon}}{4\omega'_*}\frac{\partial^3\big\langle\widetilde h_1 ^ 2\big\rangle^{\varphi}(i_*,\tau_*)}{\partial i^3 } \int\limits_{-\infty}^{+\infty}\frac{\partial h_1(i_*,{\varphi}_a,\tau_*)}{\partial{\varphi}}\,{\mathrm{d}}t+o({\varepsilon}^{\frac32}).\end{aligned}\ ] ] @xmath274 so @xmath275 @xmath277 so @xmath278 there exists a function @xmath280 , such that @xmath281 . here
+ @xmath282 with @xmath283 .
thus @xmath284 there exists a function @xmath285 , such that @xmath286 . here
+ @xmath287 with @xmath288 .
thus @xmath289 there exists a function @xmath290 , such that @xmath291 . here
+ @xmath292 with @xmath293 .
thus @xmath294 there exists a function @xmath295 , such that @xmath296 . here
+ @xmath297 with @xmath298 .
thus @xmath299 thus we have @xmath300 therefore , applying equation ( [ s ] ) and lemma [ replacefar ] , we obtain @xmath301 where @xmath302 and @xmath303 are smooth functions .
applying lemma [ replacefar ] , we easily get the estimate @xmath305 combining obtained estimates and the identity ( [ phiiden ] ) , and taking into account that @xmath306 , we get the second formula in theorem [ thm1 ] . with results of four steps of perturbation theory
, we have @xmath307 from this relation , similarly to the above proof , we obtain the second formula of theorem [ thm2 ] .
3@xmath318 . value of @xmath323 . @xmath324 here @xmath325\,{\mathrm{d}}\theta\\ & = & \sqrt{\varepsilon}\cdot\frac5{2\sqrt6}\left(\cos\check{\check{\varphi } } _ * \int\limits_0^{+\infty } \sin\tfrac{\theta^2}2\,{\mathrm{d}}\theta + \sin\check{\check{\varphi } } _ * \int\limits_0^{+\infty } \cos\tfrac{\theta^2}2\,{\mathrm{d}}\theta\right)\\ & = & \sqrt{\varepsilon}\cdot\frac{5\sqrt\pi}{4\sqrt6}\,(\cos\check{\check{\varphi}}_*+\sin\check{\check{\varphi}}_*).\end{aligned}\ ] ] and @xmath326 4@xmath318 . value of @xmath328 .
@xmath329 here @xmath330\,{\mathrm{d}}\theta\\ & = & \sqrt{\varepsilon}\cdot\frac{\sqrt3}{\sqrt2 } \left(\sqrt2\cos\check{\varphi}_*\cdot\sqrt\frac\pi8 -\sqrt2\sin\check{\varphi}_*\cdot\sqrt\frac\pi8\right)\\ & = & \sqrt{\varepsilon}\cdot\frac{\sqrt{3\pi}}{2\sqrt2}\,(\cos\check{\varphi}_*-\sin\check{\varphi}_*).\end{aligned}\ ] ] our goal here is to check numerically , if the obtained values @xmath350 indeed approximate actual values of @xmath351 on solutions of system ( [ example_phi ] ) with the accuracy @xmath352 , as it is predicted by corollary [ cor1 ] . to this end
we introduce the variable @xmath353 and integrate numerically by 4th order runge - kutta algorithm the system @xmath354 on the slow time interval @xmath355 . at @xmath356
we have values @xmath357 and @xmath358 .
these calculations are performed for 48 values of @xmath251 equally spaced on @xmath359 $ ] , and 11 values of @xmath1 , \{0.02 , 0.015 , 0.01 , 0.007 , 0.005 , 0.003 , 0.002 , 0.0015 , 0.001 , 0.0007 , 0.0005}. we calculate @xmath360}|i_+^{\rm{numer}}-i_+^{\rm{theor}}|,\quad e_{{\varphi}_+}({\varepsilon})=\max_{{\varphi}_-\in[0,2\pi]}|{\varphi}_+^{\rm{numer}}-{\varphi}_+^{\rm{theor}}|\,,\ ] ] and plot values @xmath361 and @xmath362 as functions of @xmath363 in figures [ figurei ] and [ figurephi ] .
linear least squares fit of the data in figures [ figurei ] and [ figurephi ] gives slopes @xmath364 and @xmath365 , respectively . the ideal results for accuracy @xmath366 would be @xmath367 and @xmath368 .
thus the numerical simulation indicates that the accuracy is @xmath366 , as expected .
kevorkian , j. and cole , j. d. : _ multiple scale and singular perturbation methods_. applied mathematical sciences , vol . 114 .
springer - verlag , 1996 .
chirikov , b. v. : _ the passage of a nonlinear oscillatory system through resonance_. sov .
, dokl . 4 , 1959 , pp .
390 - 394 .
bosley , d. l. : _ an improved matching procedure for transient resonance layers in weakly nonlinear oscillatory systems_. siam j. appl .
math , vol .
56 , no . 2 , 1996 , pp .
420 - 445 .
alekseev , p. a. : _ on change of action at passage through a resonance in a quasilinear hamiltonian system_. m.sc .
thesis , moscow university , 2007 . | we consider a quasi - linear hamiltonian system with one and a half degrees of freedom .
the hamiltonian of this system differs by a small , @xmath0 , perturbing term from the hamiltonian of a linear oscillatory system .
we consider passage through a resonance : the frequency of the latter system slowly changes with time and passes through 0 .
the speed of this passage is of order of @xmath1 .
we provide asymptotic formulas that describe effects of passage through a resonance with an accuracy @xmath2 .
this is an improvement of known results by chirikov ( 1959 ) , kevorkian ( 1971 , 1974 ) and bosley ( 1996 ) .
the problem under consideration is a model problem that describes passage through an isolated resonance in multi - frequency quasi - linear hamiltonian systems . |
the prevailing interpretation of the ly@xmath0 forest is that it arises naturally from absorption by trace amounts of neutral hydrogen in a photo - ionized , inhomogenous intergalactic medium ( igm ) . to model the ly@xmath0 forest within this picture no speculative assumptions about the physical properties of individual ly@xmath0 absorbers have to be made .
one merely considers a cosmological simulation of structure formation , including dark matter and a baryonic component as well as a photo - ionizing background radiation field , and evaluates the absorption properties along lines of sight through the simulation box at desired redshifts . during the past few years
it has been realized that the artificial absorption spectra resulting from this approach bear close resemblance to the ly@xmath0 forest seen in qso absorption spectra , and these cosmological models of the ly@xmath0 forest have been able to quantitatively account for the observed distribution functions in hi column density and b - parameters to a reasonable accuracy ( @xcite @xcite @xcite @xcite ; for related semianalytic modeling see , e.g. , @xcite @xcite @xcite ) . within recent years , observations have demonstrated that strong ly@xmath0 forest absorbers generally show associated metal line absorption ( @xcite @xcite @xcite @xcite ) .
selected metal absorption lines can be readily incorporated into cosmological models of the ly@xmath0 forest from a knowledge of densities , temperature , and uv radiation field along the lines of sight , if a metal enrichment pattern of the baryonic igm is specified .
such models account fairly well for the observed properties of the civ(1548 , 1550 ) doublet and a handful of other lines , for an igm metallicity @xmath7 , if a scatter of about an order of magnitude is assumed ( @xcite @xcite @xcite ) . instead of making an a priori selection of a few metal lines to include in a model , it is useful to make the models _ predict _ which metal lines , out of hundreds of candidates , should be observable ( in spectra of a given resolution and s / n ) associated with ly@xmath0 absorbers of given metallicities and hi column densities .
such an approach allows a comprehensive screening for lines that deserve a more detailed treatment , dependent on the specific purpose of the modeling , and allows for the sharpest possible test of the models .
we describe the results from an implementation of such a technique in the following .
let us denote a metal absorption line produced by an element @xmath8 in ionization stage @xmath9 with rest transition wavelength @xmath10 and oscillator strength @xmath11 as @xmath12 .
for such a line associated with with a ly@xmath0 absorber of neutral hydrogen column density @xmath5 and metallicity @xmath13 \equiv \log{(n_z / n_h)}-\log{(n_z / n_h)_{\odot}}$ ] we can define the following _ line observability index _ : @xmath14,n_{\rm hi } ) & \equiv & -17.05 \ , + \
, \log{n_{\rm hi } } \ , + \ , [ { \rm z / h } ] \ , + \log{(f\ , { \lambda}^2 ) } \\ \nonumber & & + \log{({n_z / n_{\rm h}})_{\odot } } \ , + \ , \log{(x_z^i / x_h^0)}\end{aligned}\ ] ] ( see @xcite for more details ) .
this expression assumes units of @xmath15 for @xmath5 and for @xmath10 .
the choice of additive constant then implies @xmath16 , where @xmath17 is the rest equivalent width , for weak lines .
hence , this quantity can be used to rank metal lines in terms of strength , and it can be compared to detection limits in spectra of a given quality .
the model - dependent ionization corrections are contained in the last term , the ratio of ionization fractions of @xmath18 and @xmath19 within the absorber . as a specific application of the lox technique we use the standard cdm , @xmath20 , @xmath21 simulation described in @xcite and @xcite .
we assume a haardt & madau background radiation spectrum of slope -1.8 at the high - energy end @xcite normalized to match the observed mean flux decrement @xmath22 @xcite , and we focus on redshifts @xmath23 .
the simulations exhibit rather tight correlations between @xmath5 of an absorbing region and the typical temperature and total gas density within the region , so the ionization correction term is essentially a function only of @xmath5 , and we calculate this term using the photoionization code cloudy 90 @xcite . the lox has been evaluated for 199 metal lines from the database contained in the voigt profile fitting software vpfit @xcite .
figure 1 shows the results as a function on @xmath5 for nine of the strongest lines , assuming a carbon abundance of [ c / h]=-2.5 and a relative abundance pattern similar to that observed in population ii stars .
a more exhaustive list of lines is presented in @xcite , which also discusses the dependence of the lox on @xmath24 , @xmath25 , and the normalization of the radiation field @xmath26 .
this dependence is weak , and the following predictions are believed to hold rather generally in cosmological models of the ly@xmath0 forest : * the civ(1548 , 1550 ) doublet is the strongest line with rest wavelength @xmath27 regardless of the hi column density of the absorber . * ovi(1032 , 1038 ) is the only detectable line in low column density ( @xmath28 ) absorbers , and hence the best probe of metallicity in the low density igm . * the ciii(977 ) and siiii(1206 ) lines are the potentially strongest metal lines , but they are only expected to be seen in the relatively uncommon absorbers with @xmath29 .
the first prediction is easily verifiable from available data and is found to hold .
the other predictions , featuring lines that are embedded in the ly@xmath0 forest , remain to be verified . if few or no ovi lines are found at @xmath23 in real spectra then the metallicity of the low - density regions of the igm may actually be less than -2.5 , or alternatively there may be a problem with this picture of the ly@xmath0 forest . in absorbers with higher values of @xmath5 ,
the higher number of observable metal lines makes a more thorough comparison between model and observations possible .
we are currently initiating a detailed systematic study of the distributions and relative strenghts of metal lines associated with such systems .
99 bi , h.g .
1993 , , 405 , 479 bi , h.g . &
davidsen , a.f .
1997 , , 479 , 523 carswell , r.f . ,
webb , j.k . ,
baldwin , j.a . , & atwood , b. 1987 , , 319 , 709 cen , r. , miralda - escud , j. , ostriker , j.p .
, & rauch m. 1994 , , 427 , l9 cowie , l.l . , songaila , a. , kim , t .- s . , &
1995 , aj , 109 , 1522 dav , r. , hernquist , l. , weinberg , d.h . , &
katz , n. 1997 , , 477 , 21 ferland , g.j .
, 1996 , university of kentucky , department of astronomy , internal report haardt , f. & madau , p. 1997 , in prep .
haehnelt , m.g . ,
steinmetz , m. , & rauch , m. 1996 , , 465 , l65 hellsten , u. , dav , r. , hernquist , l. , weinberg , d. & katz , n. 1997 , , vol .
487 hellsten , u. , hernquist , l. , katz , n. , & weinberg , d. 1997 , , submitted .
astro - ph/9708090 hernquist , l. , katz , n. , weinberg , d.h . , & miralda - escud , j. 1996 , , 457 , l51 hui , l. , gnedin , n.y . ,
& zhang , y. 1997 , astro - ph/9608157 , 486 .
katz , n. , weinberg , d.h . , & hernquist , l. 1996a , apjs , 105 , 19 meyer , d.m . & york , d.g .
1987 , , 315 , l5 petitjean , p. , mcket , j.p . & kates , r.e .
1995 , a&a , 295 , l9 press , w.h .
, rybicki , g.b . , & schneider , d.p .
1993 , , 414 , 64 rauch , m. , haehnelt , m.g . , & steinmetz , m. 1996 , , 481 , 601 .
songaila , a. & cowie , l. l. 1996 , aj , 112 , 335 [ sc ] zhang , y. , anninos , p. , & norman , m.l .
1995 , , 453 , l57 womble , d.s . ,
sargent , w.l.w .
, & lyons , r.s .
1995 , in cold gas at high redshift , eds .
m. bremer et al . , kluwer 1996 , ( astro - ph/9511035 ) . | the metal absorption lines found in association with ly@xmath0 absorbers of moderate to high hi column density contain valuable information about the metallicity and ionization conditions within the absorbers and offer a stronger test of models of the intergalactic medium at @xmath1 than hi absorption lines alone .
we have developed a method to predict the strengths of metal absorption lines within the framework of cosmological models for the ly@xmath0 forest .
the method consists of evaluating a quantity , the _ line observability index _
, for a database of hundreds of candidate metal lines , allowing a comprehensive identification of the lines the model predicts to be detectable associated with a ly@xmath0 absorber of a given hi column density and metallicity . applying this technique to a particular class of models at @xmath2
, we predict that the ovi(1032 , 1038 ) doublet is the only practical probe of the metallicity of low column density absorbers ( @xmath3 ) , that civ ( 1548 ) is the strongest line with rest wavelength @xmath4 regardless of @xmath5 , and that the strongest metal lines should be ciii(977 ) and siiii(1206.5 ) , which peak at @xmath6 .
= = = 1=1=0pt = 2=2=0pt |
entangled quantum state is one of essential key ingredients in the implementation of quantum communication and quantum computation @xcite . among kinds of schemes proposed to generate entangled states @xcite , cavity quantum electrodynamic systems(cqeds ) , which marry atomic and photonic quantum bits together ,
are paid more attentions because of both its low decoherent rate and promising feasibility to scale - up . in recent years , several elegant works based on cqeds have been done . in @xcite , a pair of momentum- and polarization- entangled photons
is sent to two spatially separated cavities , each of which contains a v - type atom .
after the atoms absorb the photons , the photon entanglement is transferred to the atoms . in @xcite ,
entangled states of two @xmath0-type atoms individually trapped in spatially separated cavities are generated through the interference of the polarized photons leaking out of the cavities on a beam splitter and co - instantaneous single - photon detections . in these schemes , in order to guarantee the interference effect , several conditions are required .
first , the two atoms have to be simultaneously excited or driven .
second , spatial shapes of photon pulses leaking from the cavities should be similar .
third , the two photon paths from the cavities to the beam splitter must be symmetrical .
in addition , the success of generating entangled atomic states is probabilistic since the photon - state - projection detection behind the beam splitter is performed .
we also notice that the previous schemes are based on either photon emission or photon absorption process . in this paper
, we consider a cqed system in which a @xmath0-type atom and a v - type atom are trapped individually in two spatially separated cavities that are connected by an optical fiber .
this setup is closely related to two previous schemes . in @xcite , pellizzari proposed a cavity - fibre - cavity system to realize reliable transfer of a quantum state . in that scheme ,
two three - level @xmath0 atoms that are adiabatically and simultaneously driven by a laser beam are trapped individually in two single - mode cavities connected via an optical fibre . in @xcite ,
serafini et al employed the similar setup in which two - level atoms are trapped in fibre - connected cavities to realize highly reliable swap and entangling gates .
we notice that quantum information processes realized in the fibre - connected cavity setup depends on atomic level configurations and laser driving ways . in the present scheme , without using any laser driving , only the @xmath0-type atom is initially to be pumped in the excited state . instead of using either photon absorption or photon emission as done in the previous scheme ,
both the photon emission of the @xmath0-type atom and the photon absorption of the v - type atom are involved . in this way , an entangled state of the two atoms
can deterministically be generated through the atom - field interaction without the photon interference and the photon - state projection detection .
type atom and a v - type atom are trapped in two spatially separated cavities a and b , respectively . the two cavities are linked through an optical fibre.,width=340 ] as shown in fig.[fig : model ] , a @xmath0-type atom with two degenerate states @xmath2 and @xmath3 and a v - type atom with two degenerate states @xmath4 and @xmath5 are trapped in two spatially separated cavities that are linked by an optical fibre . the two atoms interact with local cavity fields , respectively . in recent years , such fiber - connected cavity systems are often empolyed for realizing quantum information processes @xcite .
the hamiltonian of the whole system can be written as @xcite @xmath6 where @xmath7,\\ h_{b}&=&\sum_{j=\pm1}\big[\omega^f_{b , j}a^+_{b , j}a_{b , j}+\omega^a_{b } \sigma^z_{b , j}+\lambda_{b , j}a_{b , j}\sigma^{+}_{b , j}+\mathrm{h.c.}\big],\\ h_f&=&\sum_{k=1}^{\infty}\sum_{j=\pm1}\big[\omega_{j , k } b^+_{j , k } b_{j , k}+\nu_{j , k}b_{j , k}[a^+_{a , j}+(-1)^ke^{i\phi}a^+_{b , j}]+\mathrm{h.c.}\big],\end{aligned}\ ] ] where @xmath8 is the energy of the system consisting of atom a / b and the corresponding local cavity fields , and @xmath9 represents the free energy of fibre modes and the interaction between cavity and fibre modes . in ( 2)-(4 ) , @xmath10 and @xmath11 are annihilation and creation operators for photons of frequency @xmath12 and polarization j(=
-1,+1 corresponding to right and left circular polarizations , respectively ) in cavity a / b , and @xmath13 , @xmath14 are annihilation and creation operators for photons of frequency @xmath15 and polarization j in mode k of the fibre field , @xmath16 , @xmath17 and @xmath18 , @xmath19 is the frequency of the transition @xmath20 , @xmath21 is the coupling constant of atom a / b with cavity mode @xmath22 , and @xmath23 the coupling constant of cavity mode @xmath22 with fibre mode @xmath24 . here
, we assume that the right and left circular polarization modes in both the cavities have the same frequency @xmath25 and the interaction between the atoms and the cavity fields is on resonance .
that is , @xmath26 . in the short fibre limit @xmath27 , where @xmath28 is the length of fibre and @xmath29 is the decay rate of the cavity fields into a continuum of fibre modes , only resonant modes b of the fibre interacts with the cavity modes @xcite .
for this case , the hamiltonian @xmath30 may be approximated to @xmath31,\ ] ] where the phase @xmath32 in ( 4 ) has been absorbed into the annihilation and creation operators of the modes of the second cavity field . in the interaction picture ,
the total hamiltonian now becomes @xmath33.\ ] ]
in this section , we show that an extremely entangled states of the atoms can be deterministically generated in an ideal situation . to explain the scheme of the generation of entangled states , first , let us see how the entangling process works .
suppose that at the initial time the atom a is pumped in the excited state @xmath34 and the atom b is in the ground state @xmath35 . through interacting with local cavity fields ,
the atom a emits either a @xmath36 or @xmath37 polarized photon and then the atom - field system is in the state @xmath38 , where @xmath39 is a cavity field state with one photon in mode @xmath40 .
assume that the coupling of the cavity modes with fibre modes is so strong that the emitted photon can be transferred into the fibre modes before reabsorbed by the atom a. through the fibre , the photon enters the cavity b. after absorbing the photon , the atom b is excited in either @xmath41 or @xmath42 . in this way , the entangled state @xmath43 can be created . in order to protect the entangled state from spontaneous emission , as shown in fig.[fig : model ] , one may apply a @xmath44-polarized light to the transition between @xmath45 and @xmath46 , and generate a stable entangled state : @xmath47 , where @xmath48 and @xmath49 are stable states .
now , let us go to investigate the generation of the entangled state in detail .
the time evolution of the whole system state is governed by the schrdinger equation @xmath50 suppose that at the initial time the system is in the state @xmath51 .
the state vector @xmath52 at time t can be expanded as @xmath53 where @xmath54 represents either a photon number state with one right(-1 ) or left(+1 ) circular polarized photon . upon substituting ( [ psi1 ] ) in ( 7 ) with the conditions @xmath55 , and @xmath56 , one has @xmath57+\frac{\lambda^2}{2(\lambda^2+\nu^2)}\big[cos(\sqrt{1+\frac{\nu^2}{\lambda^2}}\sqrt{2}\lambda t)-1\big],\\ & & d_2=d_6=-\frac{i}{2\sqrt2}\big[sin(\sqrt2\lambda t)+\frac{\lambda}{\sqrt{\lambda^2+\nu^2}}sin(\sqrt{1+\frac{\nu^2}{\lambda^2}}\sqrt{2}\lambda t)\big],\\ & & d_3=d_7=\frac{\lambda\nu}{2(\lambda^2+\nu^2)}\big[cos(\sqrt{1+\frac{\nu^2}{\lambda^2}}\sqrt{2}\lambda t)-1\big],\\ & & d_4=d_8=\frac{i}{2\sqrt2}\big[sin(\sqrt2\lambda t)-\frac{\lambda}{\sqrt{\lambda^2+\nu^2}}sin(\sqrt{1+\frac{\nu^2}{\lambda^2}}\sqrt{2}\lambda t)\big],\\ & & d_5=d_9=\frac{1}{2\sqrt2(\lambda^2+\nu^2)}\big[\nu^2+\lambda^2cos(\sqrt{1+\frac{\nu^2}{\lambda^2}}\sqrt{2}\lambda t)-(\lambda^2+\nu^2)cos(\sqrt2\lambda t)\big].\end{aligned}\ ] ] we notice that @xmath58 , @xmath59 , and the system is in the state @xmath60 and @xmath61 . in this state ,
the atoms are completely separated from the cavity fields and in the extremely entangled state @xmath62 . in eqs .
( 9)-(13 ) , one may also find that if @xmath63 the conditions @xmath58 and @xmath59 can also be satisfied at the time @xmath64 .
it means that the requirement @xmath65 can be loosed in the limit @xmath63 . in order to make the thing more clear ,
let us introduce the normal modes @xmath66 in terms of these normal modes , the hamiltonian ( [ hi1 ] ) can be rewritten as @xmath67 .
\end{aligned}\ ] ] as noted in ( 16 ) , the normal mode @xmath68 is resonant with the atomic transitions from both @xmath69 to @xmath70 of the atom a and @xmath71 to @xmath72 of the atom b , but the normal modes @xmath73 off - resonant . in the limit @xmath74 , the off - resonant modes can be safely neglected . in this case
, the hamiltonian ( 16 ) becomes @xmath75.\ ] ] in the hamiltonian ( 17 ) , the fibre modes are completely depressed since the resonant modes @xmath76 do not contain the fibre modes and the original system is modeled by a system consisting of two atoms interacting with two resonant modes in one cavity . for the initial condition @xmath77 , where @xmath78 is the vacuum state of the normal mode @xmath79 , the solution of the schrdinger equation ( 7 ) with the hamiltonian ( 17 ) can be found to be @xmath80 where @xmath81 is a state in which there is either one right(-1 ) or left(+1 ) circular polarized photon in the normal mode @xmath76 . under the condition @xmath82 , and @xmath83 , the expansion coefficients in ( 18 ) are given by @xmath84 from eqs.(18 ) and ( [ coeff2 ] ) , one can see that at the moments @xmath85 the system is in the state @xmath86 in which the two atoms are separated from the normal modes and the extremely entangled state of the atoms @xmath62 is achieved .
this result depends on the condition @xmath87 . in order to check the approximation validity , in fig.[nodis ] , we show the fidelities @xmath88 , where @xmath89 is the vacuum state for photons in both the cavities and the fibre , and @xmath90 that corresponds to the limit @xmath91 .
+ + as shown in fig.[nodis ] , the results related to the hamiltonians ( 6 ) and ( 17 ) become nearly same and the approximation condition @xmath87 holds very well when @xmath92 . from the results above ,
we come to the conclusion that to generate the extremely entangled state of the atoms it is needed to choose the right interaction time and the right ratio of @xmath93 to @xmath94 when @xmath93 is comparable to @xmath94 , but it is needed only to control the interaction time when @xmath87 .
in this section , we investigate the influence of spontaneous emission and photon leakage on the generation of atomic entangled states .
the master equation for the density matrix of the whole system is @xmath95 & -\sum_{j=-1,1}\frac{\gamma_{a , j}}{2}(a^+_{a , j}a_{a , j}\rho-2a_{a , j}\rho a^+_{a , j}+\rho a^+_{a , j}a_{a , j})\\ & -\sum_{j=-1,1}\frac{\gamma_{b , j}}{2}(a^+_{b , j}a_{b , j}\rho-2a_{b , j}\rho a^+_{b , j}+\rho a^+_{b , j}a_{b , j})\\ & -\sum_{j=-1,1}\frac{\gamma_{f , j}}{2}(b^+_{j}b_{j}\rho-2b_{j}\rho b^+_{j}+\rho b^+_{j}b_{j})\\ & -\sum_{j=-1,1}\frac{\kappa_{a , j}}{2}(\sigma^+_{a , j}\sigma_{a , j}\rho-2\sigma_{a , j}\rho \sigma^+_{a , j}+\rho \sigma^+_{a , j}\sigma_{a , j})\\ & -\sum_{j=-1,1}\frac{\kappa_{b , j}}{2}(\sigma^+_{b , j}\sigma_{b , j}\rho-2\sigma_{b , j}\rho \sigma^+_{b , j}+\rho \sigma^+_{b , j}\sigma_{b , j } ) , \end{aligned}\ ] ] where @xmath96 is given by ( 6 ) , and @xmath97 and @xmath98 are the decay rates for photons in mode @xmath22 of cavities a and b , and fibre , respectively , and @xmath99 and @xmath100 are spontaneous emission rates that are related to the decay channel of atom a : @xmath101 and the decay channel of atom b : @xmath102 , respectively . in solving the master equation
, we always choose @xmath103 , and @xmath104 , @xmath105 , and @xmath106 . by use of the solution of eq .
( 20 ) , the fidelity @xmath107 can be calculated .
+ + in fig.[rfdis ] , the fidelity @xmath108 is plotted for the cases with @xmath109 but different values of @xmath93 and @xmath110 . as shown in fig .
3(c ) and ( d ) , even for the case with @xmath111 , the effect of photon leakage out of the fibre can be greatly depressed when the cavity - fiber coupling strength @xmath93 is much larger than the cavity - atom coupling strength @xmath94
. this result can be easily understood . from the discussion in the preceding section ,
there is only the resonant normal mode @xmath76 plays role when @xmath87 . according to ( 14 ) , the resonant normal mode is not involved with the fibre mode . in the limit , therefore , the fibre mode is not excited and is always kept in the vacuum state . on the other hand , the condition @xmath87 means that the transmission time of photons through the fibre becomes much shorter than the interaction time of the atoms with the cavity fields and then the fibre photon is equivalently kept in the vacuum state in the whole process .
this reason clearly explains why the influence of photon loss can be diminished in the limit @xmath87 . from eqs .
( 9)-(13 ) , @xmath112 at @xmath113 if the condition @xmath61 is satisfied .
it means that all the terms associated with the basic vectors with one fibre photon in ( 8) disappear under this condition . in this case , therefore , the fibre field in the state vector ( 8) is in the vacuum state .
therefore , the appropriate choice of the ratio @xmath114 may be useful to get a higher fidelity of the entangled state even if the condition @xmath87 is not well satisfied . in fig .
3(a ) and ( b ) , the results with @xmath115 and @xmath116 are shown .
it is observed that the former result is obviously better than the latter one when @xmath117 .
next , we investigate the influence of spontaneous emission and cavity leakage . in figs.[rkdis ] , the fidelity @xmath108 is shown for the cases with @xmath118 and @xmath119 , and different values of @xmath120 and @xmath121 . comparing these figures , we find that the atomic spontaneous emission has a stronger influence on the fidelity than the cavity photon leakage does . to obtain the fidelity higher than 0.97
, one should keep the decay rates @xmath122 and @xmath123 . in recent experiments @xcite , for an optical cavity with the wavelength region @xmath124
, the condition @xmath125 is achievable .
therefore , the present scheme with a high fidelity larger than 0.98 is feasible in current experiments .
in this paper , we propose a scheme to generate an entangled state of two atoms trapped in spatially separated two cavities that are connected by an optical fibre .
we show that an extremely entangled state of the atoms can be deterministically generated if the ratio of the coupling constant of the cavity field with the atoms to the coupling constant of the cavity field with the fibre mode satisfies a certain condition .
the scheme has three features .
first , it is based on both the photon emission of a @xmath0-type atom and the photon absorption of a v - type atom .
therefore , in contrast to the previous schemes @xcite , either a pair of entangled photons or the photon interference process is not required .
second , since the present scheme is based on the interaction dynamics of atoms with cavity fields and then the co - instantaneous single photon detection is not required , the generation of entangled states is deterministic .
third , in contrast to the previous schemes in which manipulations such as excitation , photon emission or absorption have to be performed simultaneously on all of atoms under consideration , such manipulations are performed respectively on each of atoms in the present scheme .
this may reduce difficulties resulting from manipulations performed at the same time on two atoms .
the influence of various decoherence processes such as spontaneous emission and photon loss on the generation of entangled states is also investigated .
we find that the effect of photon leakage out of the fibre can be greatly depressed and even ignored if the coupling of the fibre mode with the cavity fields is much stronger than the coupling of the atoms with the cavity fields .
if the condition is not well satisfied , the appropriate ratio of the coupling constant of the cavity field with the atoms to the coupling constant of the cavity field with the fibre mode can also ensure a high fidelity of generating entangled states . as regards the effect of photon loss out of the cavities and spontaneous emission of the atoms , we find that the implementation of the present scheme is within the scope of the current techniques .
bennett , g. brassard , c. crepeau , r. jozsa , a. peres , and w. k. wooters , phys .
lett . * 70 * 1895 ( 1993 ) .
a. barenco , d. deutsch , a. ekert , and r. jozsa , phys .
* 74 * 4083 ( 1995 ) .
bennett , d.p .
divincenzo , nature(london ) * 404 * , 247 ( 2000 ) .
nielsen and i.l .
chuang , quantum computation and quantum information ( cambridge university press , cambridge , england , 2000 ) .
reina , l. quiroga , and n.f .
johnson , phys .
a * 62 * , 012305 ( 2000 ) .
t. pellizzari , s.a .
gardiner , j.i .
cirac , and p. zoller , phys .
lett . * 75 * 3788 ( 1995 ) .
i.l . chuang and y. yamamoto , phys .
a * 52 * , 3489 ( 1995 ) .
cirac and p. zoller , phys .
lett . * 74 * , 4091 ( 1995 ) .
turchette , c.s .
wood , b.e .
king , c.j.myatt , d. leifried , w.m .
itano , c. monroe , and d.j .
wineland , phys .
. lett . * 81 * , 3631 ( 1998 ) .
s.lloyd , m.s .
shahriar , j.h .
shapiro , and p.r .
hemmer , phys .
lett . * 87 * 167903 ( 2001 ) .
x.l . feng , z.m .
zhang , x.d .
gong , and z.z .
xu , phys .
90 * , 217902 ( 2003 ) .
duan , h.j .
kimber , phys .
lett . * 90 * , 253601 ( 2003 ) .
browne , m.b .
plenio , and s.f .
huelga , phys .
lett . * 91 * , 067901 ( 2003 ) .
duan , m.j .
madsen , d.l .
moehring , p. maunz , r.n .
kohn , jr . and c. monroe , phys .
a * 73 * , 062324(2006 ) .
t. pellizzari , phys .
lett . * 79 * , 5242 ( 1997 ) .
a. serafini , s. mancini , and s. bose , phys .
* 96 * , 010503 ( 2006 ) .
zhang - qi yin and fu - li li , quant - ph/0606152 [ phys .
a ( to be published ) ]
. s. m. spillane , t. j. kippenberg , k. j. bahala , k. w. goh , e. wilcut , and h. j. kimble , phys .
a * 71 * 013817 ( 2005 ) and references therein .
j. r. buck and h. j. kimble , phys .
a**67 * * , 033806(2003 ) . | we consider a system consisting of a @xmath0-type atom and a v - type atom , which are individually trapped in two spatially separated cavities that are connected by an optical fibre .
we show that an extremely entangled state of the two atoms can be deterministically generated through both photon emission of the @xmath0-type atom and photon absorption of the v - type atom in an ideal situation .
the influence of various decoherence processes such as spontaneous emission and photon loss on the fidelity of the entangled state is also investigated .
we find that the effect of photon leakage out of the fibre on the fidelity can be greatly diminished in some special cases . as regards the effect of spontaneous emission and photon loss from the cavities , we find that the present scheme with a fidelity higher than @xmath1 may be realized under current experiment conditions . |
the structure , strength and stability of ligand - receptor bonds have been investigated extensively over several decades @xcite .
the concepts ( and definitions ) of ligand and receptor have also evolved during this period . in the words of martin karplus @xcite , `` the ligand can be as small as an electron , an atom or diatomic molecule and as large as a protein '' . in principle , the definition of a ligand can be extended even further to a stiff filament formed by a hierarchical organization of proteins ; the microtubule ( mt ) @xcite , a nano - tube in eukaryotic cells , is an example of such filaments .
thus , if such a stiff straight filament is tethered to a flat rigid wall by wall - anchored proteins that bind specifically to the filament , the attachment can be viewed as a ` ligand - receptor bond ' where the filament is the analog of a ligand .
inspired by this generalized definition of ligands and receptors , we study here a very special type of _ non - covalent _ filament - wall attachment from the perspective of ligand - receptor bonds .
the tether protein that we consider here are ` active ' in the sense that these consume chemical fuel for their mechanical function .
the theoretical model developed in this paper is motivated by the typical attachment formed by a mt with the cortex of a living eukaryotic cell where the two are linked by dynein molecules @xcite ; dynein is a motor protein powered by input chemical energy extracted from atp hydrolysis @xcite .
there are equispaced dynein - binding sites on the surface of a mt where the head of a dynein can bind specifically ; the tail of the dyneins are anchored on the cortex .
thus , dyneins function as tethers linking the mt with the wall ; unbinding of _ all _ the dynein heads from the mt would rupture the mt - cortex attachment .
however , the model developed in this paper captures only some of the key features of the real mt - cortex attachments like , for example , the polymerization - depolymerization of mt @xcite .
the wall in our model mimics the cell cortex .
it is worth pointing out that the mt - wall model developed here differs fundamentally from another mt - wall model reported earlier in ref.@xcite ( from now onwards referred to as the ssc model ) .
unlike the active ( i , e . , energy consuming ) linkers , which here mimic dynein motors , the dominant tethers in the ssc model are passive . moreover , the wall in the ssc model represents a kinetochore @xcite , a proteinous complex on the surface of a chromosome , whereas the wall in the model developed in this paper represents the cell cortex .
although the formation of the mt - wall attachment itself is an interesting phenomenon , we do not study it here . instead ,
in this paper we consider a pre - formed attachment to investigate its strength and stability using a kinetic model that mimics the protocols of dynamic force spectroscopy @xcite .
two distinct protocols are routinely used in force spectroscopy for measuring the strength and stability of ligand - receptor bonds @xcite . in the _ force - clamp _ protocol
, a time - independent load tension is applied against the bond ; the time duration after which the bond is just broken is the life time of the bond . since the underlying physical process is dominated by thermal fluctuations different values of life time
are observed upon repetition of the experiment .
therefore , the _ stability _ of the bond is characterized by the life time distribution ( ltd ) .
on the other hand , in the _ force - ramp _ protocol the magnitude of the time - dependent load tension is ramped up at a pre - decided rate till the bond just gets ruptured ; the rupture force distribution ( rfd ) characterizes the _ strength _ of the bond . for a _ slip - bond _ the mean life time ( mlt ) decreases monotonically with the increasing magnitude of the tension in the force - clamp experiment . in contrast , a nonmotonic variation ( an initial increase followed by decrease ) of the mlt with increasing load tension in the force - clamp experiments is a characteristic feature of _ catch - bonds _ @xcite . although the force - ramp is a more natural protocol for dynamic force spectroscopy @xcite the results provide somewhat indirect evidence for catch - bonds .
we mimic both the force - clamp and force - ramp protocols in our computer simulations of the theoretical model to compute the ltd and rfd for the motor - linked mt - wall attachment . the results
establish that this attachment is , effectively , like a slip - bond .
this is in sharp contrast to results obtained from the ssc model ; the latter exhibits a catch - bond - like behavior as observed also in experiments performed _ in - vitro_. to our knowledge , similar dynamic force spectroscopic experiments have not been attempted so far on mt - cell cortex attachments . nevertheless , our generic theoretical model and the results are likely to motivate such experiments in near future .
the model is quite general . in principle , the same model can serve , after appropriate minor adaptation , as a minimal model for many other attachments formed by a polymerizing - depolymerizing filament with a wall where the specific interaction of the two arises from a set of active linkers .
we model the mt as a strictly one - dimensional stiff filament . the plus end of the mt is oriented along the + x - direction of the one - dimensional coordinate system chosen for the model .
the rigid wall facing the mt filament is perpendicular to the x - axis .
as stated earlier , our model is motivated by the cortex - mt attachment where dynein motors tether the mt to the cortex . since a dynein motor , fuelled by atp ,
has a natural tendency to walk towards the minus end of the mt , the molecular motors in our model are also assumed to be minus - end directed , while consuming input energy , in the absence of external load tension .
we assume that each motor is permanently anchored onto the rigid wall and can not detach from it while the head of a motor can attach to the mt and a motor already attached to the mt can also detach from it .
we model the linkage between a motor and the rigid wall by an elastic element that is approximated as a hookean spring .
the wall is assumed to execute one - dimensional diffusion along the x - coordinate with a diffusion constant @xmath0 .
and @xmath1 represents the distance of mt tip and dynein motor from cell cortex . in the inset
mt interact with cell cortex through the cortical dynein motor and dynein motor is attached with cell cortex using different protein .
dynein are attached with cell cortex through a spring with spring constant @xmath2 .
external force @xmath3 is applied on axoneme ( orange ) from which mts ( green ) are generated.,scaledwidth=50.0% ] although each dynein motor has two heads , each capable of binding specific sites on the mt , we denote the position of a motor by the location of its midpoint .
the two heads are , effectively , connected at the midpoint by a hinge ; the midpoint is also linked to the point of anchoring on the wall by an elastic element that is assumed to be a hookean spring . in the one - dimensional coordinate system the origin is fixed on the wall . with respect to this origin
@xmath4 denotes the position of the _ midpoint _ of an arbitrary molecular motor at time @xmath5 while @xmath6 denotes the corresponding position the mt tip . in this model ,
@xmath7 can change because of the following processes : + ( i ) stepping of a motor while the wall remains frozen in its current position , and + ( ii ) diffusion of the wall while motor does not step out of its current binding site on the mt .
we assume that @xmath8 is the maximum distance upto which a motor can bind to the mt .
so the motor can bind only if its head is located between @xmath9 and @xmath10 ( i.e. , @xmath11 ) .
@xmath12 is the total number of motors ( the subscript ` d ' indicates minus - ended directed motor like , for example , dynein ) that can simultaneously attach to the mt whereas @xmath13 denotes the number of motors actually attached at any arbitrary instant of time ( i.e. , @xmath14 . for any given unbound motor
, @xmath15 denotes the rate of binding of its head to the mt .
therefore , at any arbitrary instant of time @xmath5 , the rate of a binding event , i.e. , the rate at which any of the unbound motor binds to the mt is @xcite @xmath16 because of being anchored on the fixed rigid wall , the motors bound to the mt can not walk freely along the mt track . moreover , because of the upper cutoff imposed on the maximum possible elongation of the spring , a motor can move only upto a maximum distance @xmath10 .
if the rest position of motor head is @xmath17 , the spring force acting on the motor is given by @xmath18 here we assume external load @xmath3 to be equally shared by @xmath19 motors so that each motor feels the load @xmath20 thus , the effective force felt by single motor head that is bound to mt and located at a position @xmath7 is @xmath21 where @xmath22 is given by ( [ eq - spring ] ) .
let @xmath23 denote the rate of unbinding of the head of a motor in the absence of load force .
when a load force @xmath3 tends to rupture the attachment , the unbinding rate increases .
following kramers ( or bell ) theory we assume an approximate exponential dependence @xcite @xmath24 the characteristic ` detachment force ' @xmath25 can be expressed as @xmath26 where @xmath27 is the extension of the energy barrier between the bound and unbound state of a motor
. the effective rate of unbinding of a single motor head from the mt is given by @xcite @xmath28 suppose the rate of forward hopping of a motor towards the minus end of of the mt is @xmath29 .
but the forward stepping of motor will be opposed by the spring force @xmath22 .
therefore , the effective rate of forward stepping of a motor is given by @xmath30 here @xmath31 is a constant parameter ( @xmath32 ) .
the characteristic force @xmath33 can be expressed as @xmath34 where @xmath35 is the length of a single subunit of mt .
based on the well known experimentally observed facts , we assume that a motor can also step towards the positive end of mt under sufficiently high load force @xcite ; the rate of stepping in that ` reverse ' direction is given by @xmath36 where the rate @xmath37 of stepping of a motor towards the plus end of the mt in the absence of load force is very small @xmath38 because the natural direction of these motors is the minus end of mt .
the form ( [ k_r ] ) captures the intuitive expectation that @xmath39 would increase with increasing spring force . the probability that ( the midpoint of ) a motor is located at @xmath40 and mt tip is at @xmath41 , while the total number @xmath13 of motors are bound to the mt simultaneously at that instant of time , is given by @xmath42 .
note that @xmath43 is a continuous variable whereas @xmath44 can take only non - negative integer values .
let @xmath45 be the conditional probability that , given a mt - bound motor located at site @xmath7 , there is another mt - bound motor at site @xmath46 on the mt @xmath47 .
then @xmath48 is the conditional probability that , given a motor at site @xmath49 , the site @xmath50 is empty .
let @xmath51 be the probability that site @xmath7 is not occupied by any motor , irrespective of the state of occupation of any other site . under mfa ,
the equations governing the time evolution of @xmath52 is given by @xmath53 where @xmath54 and @xmath55 , the length of each subunit of mt is also the spacing between the successive motor - binding sites on the mt .
the rates of polymerization and de - polymerization of a mt tip are given by @xmath56 and @xmath57 , respectively .
the rate of depolymerization of mt is suppressed by externally applied tension @xcite .
we assume that the mt - bound minus - end directed motors at the tip ( plus - end ) of the mt prevents mt protofilaments from curling outwards , thereby slowing down depolymerization rate @xcite : @xmath58/f_{\star}\biggr ) .\ ] ] where @xmath59 is the characteristic load force at which the mt depolymerization rate is an exponentially small fraction of @xmath60 .
the kronecker delta function ensures that the external force affects the depolymerization rate @xmath61 only if the motor is bound to the tip of the mt .
the force balance equation is given by @xmath62 where @xmath63 is a gaussian white noise .
let @xmath64 be the probability of finding mt tip at position @xmath41 at time @xmath65 .
time evolution of the @xmath64 govern by the fokker - planck equation @xmath66 this fokker - planck equation can also be re - cast as an equation of continuity for the probability density @xmath67 with the probability current density @xmath68 which is given by @xmath69 .
\label{flux}\ ] ] for the calculation of the lifetime of the mt - wall attachment , we have assumed that initially there is no gap between mt tip and cell wall i.e @xmath70 .
we have imposed reflecting boundary at the cell wall because mt tip can not penetrate the cell wall .
so @xmath71 we have placed an absorbing boundary at @xmath72 so that @xmath73 ; this boundary condition is motivated by our calculation of the life time of the mt - wall attachment where the lifetime is essentially a first - passage time @xcite .
l*2cr parameter & values + spacing between binding sites on mt @xmath74 & 8/13 @xmath75 + maximal distance dynein can bind @xmath10 & 32 @xmath75 + rate of polymerization of mt @xmath56 & 30 @xmath76 + rate of load - free depolymerization of mt @xmath60 & 350 @xmath76 + rate of binding of motor to mt @xmath77 & 3 @xmath76 + rate of unbinding of motor from mt@xmath78 & 3 @xmath76 + rate of forward stepping of motor @xmath79 & 6 @xmath76 + rate of backward stepping of motor @xmath80 & 0.1 @xmath76 + characteristic depolymerization force @xmath81 & 1 @xmath82 + characterstic spring force of motor @xmath83 & 1 @xmath82 + rest length of the spring @xmath17 & 10 @xmath75 + spring constant @xmath2 & 1000 @xmath84 + diffusion constant @xmath85 & 700 @xmath86 + effective drag coefficient @xmath87 & 6@xmath88 + by using a method of discretization proposed originally by wang , peskin and elston @xcite , we discretize space into contiguous discrete cells , each of length @xmath89 , where @xmath90 denotes the position of the center of the @xmath91-th cell . the potential @xmath92 is thus replaced by the discretized counterpart @xmath93x_{j } \label{discrete}\ ] ] because of this discretization , the fokker - planck equation ( [ fokker_eq_tip ] ) is replaced a master equation where the discrete jumps from the center of one cell to those of its adjacent cells in the forward or backward directions are given by the transition rates @xmath94 and @xmath95 respectively , where @xmath96 .
we have carried out simulations of this discretized version of the model using gillespie algorithm @xcite . in each time step @xmath97 six types of event are possible , namely , binding/ unbinding , forward / backward hopping of each motor and forward / backward movement of the mt tip . in the simulation
a motor can attach to a site @xmath90 only if it it empty .
similarly , a motor can step forward or backward provided the target site is empty .
we have generated trajectories up to @xmath98 time steps and , after averaging over the trajectories we get the results of our interest .
the common parameter values used in the simulation are listed in table [ parameter_values ] .
in the two subsections we present the results of our simulations under force - clamp and force - ramp conditions .
the main quantities of interest in these two cases are the ltd and rfd , respectively .
\(a ) ( b ) in the fig.[fig_dyn_clamp](a ) the probability distributions of the lifetimes are plotted for three different fixed values of the externally applied tension @xmath99 .
the corresponding values of the mean lifetimes @xmath100 are plotted against @xmath99 in fig.[fig_dyn_clamp](b ) .
each of the distributions of lifetimes shown in the three panels of fig.[fig_dyn_clamp](a ) , for which @xmath99 has a fixed value , has been fitted to the function ( see the lines fitted to the data ) @xcite @xmath101 to extract the numerical value of the fitting parameter @xmath102 which has a dimension of time@xmath103 . by repeating this fitting process also for several other values of @xmath99 ( for which the distributions of the lifetimes have not been presented in this paper ) , we have extracted the @xmath99-dependence of the fitting parameter @xmath102 .
the `` best fit function '' @xmath104 is plotted in the inset in the lower left corner of fig.[fig_dyn_clamp](b ) .
the function @xmath104 , in turn , is well approximated by the functional form @xcite @xmath105 with @xmath106 min@xmath107 and @xmath108 pn .
the monotonic decrease of the mean lifetime @xmath100 with @xmath99 indicates a slip - bond - like behaviour of the mt - motor attachment .
moreover , the best fits ( [ prob_clamp ] ) and ( [ kappa(f ) ] would suggest a functional form @xcite @xmath109 the simulation data for @xmath110 corresponding to @xmath111 pn , indeed , fit well with the form ( [ lifetime ] ) , if the values @xmath106 min@xmath107 and @xmath112 pn extracted above from fig.[fig_dyn_clamp](a ) are used ( see the blue line fit in fig.[fig_dyn_clamp](b ) ) .
but , for @xmath113 pn , this fit shows increasing deviation from the simulation data with decreasing @xmath99 .
however , in the regime @xmath113 pn , the functional form ( [ lifetime ] ) would still be consistent with the numerical data for @xmath100 if one uses @xmath114 min@xmath115 and @xmath116 4.32 pn for the two fitting parameters ( see the green dashed line in fig.[fig_dyn_clamp](b ) ] .
thus , the simulation data indicate two different regimes in both of which @xmath100 decreases exponentially with @xmath99 , but the decrease is sharper in the higher tension regime that corresponds to @xmath111 pn .
a possible interpretation of these two apparent regimes is that in the exponential of the expression for @xmath104 there is a correction term proportional to @xmath117 ; such corrections emerge naturally from systematic taylor expansion in the extended bell models @xcite .
from the log - log plot of @xmath100 against @xmath12 in the inset at the upper right corner of the fig.[fig_dyn_clamp](b ) we conclude that the mean lifetime increases nonlinearly with the number of motors following @xmath118 with @xmath119 .
the effects of the motors on the lifetime of the mt - surface attachment is not simply additive .
recall that , at any arbitrary instant of time @xmath5 , the load force is equally shared by the @xmath13 number of motors bound to the mt .
therefore , unbinding of a motor from the mt increases the load on those still attached to it .
if the redistributed share of the load is too high to resist , more motors are likely to unbind from the mt thereby causing further increase in the load share of the remaining motors still bound to the mt .
the larger is the total number @xmath12 , less catastrophic is the effect of such `` avalanche '' of unbinding of the motors and hence the nonlinear increase in the mean lifetime of the mt - wall attachment .
\(a ) ( b ) is applied on a suitable handle ( orange cylinder ) attached to the minus end of the same mt .
, scaledwidth=40.0% ] the distribution of the rupture force obtained from our simulation using the ramp force @xmath120 is shown in the fig.[fig_dyn_ramp](a ) . at any constant loading rate ,
the distribution exhibits a single peak ; the asymmetric bell - shape is the well known typical shape of rfd for common ligand - receptor bonds@xcite .
at very small forces the bonds are unlikely to get enough time to rupture .
similarly , at very large forces the likelihood of bond rupture is also small because the bond would have ruptured already at some intermediate value of the force .
therefore , @xmath121 is expected to be small at both the extremes ; naturally , it would exhibit a peak corresponding to some intermediate value of force .
these intuitive arguments not only explain the qualitative shape of @xmath121 in fig.[fig_dyn_ramp](a ) but also the observation that most probable rupture force ( i.e. , the force that corresponds to the peak in the distribution of the rupture forces ) is larger in case of a faster loading rate .
the simulation data fits well with the function @xcite @xmath122 \label{prob_ramp}\ ] ] with a dimensionless fitting parameter @xmath123 and @xmath124 , given by ( [ kappa(f ) ] ) .
the survival probability @xmath125 is defined as the probability that till time @xmath5 the mt tip has not reached the absorbing boundary at @xmath10 .
as is well known @xcite , the survival probability @xmath126 is related to the detachment rate @xmath104 by @xcite @xmath127 \label{surv_ramp}\ ] ] in the fig.[fig_dyn_ramp](b ) the simulation data for the survival probability are compared with the corresponding function @xmath128 obtained using ( [ surv_ramp ] ) .
the agreement is excellent thereby establishing consistency of the data obtained for the different quantities from our computer simulations . as expected @xmath129 at @xmath130 ( i.e. , @xmath131 ) and @xmath132 as @xmath133 ( i.e. , @xmath134 ) .
the survival probability remains high upto certain force after which it falls rapidly .
it is worth pointing out that one can account for the skewed shape of the curves @xmath121 in fig.[fig_dyn_ramp](a ) mathematically @xcite exploiting increase of @xmath104 and sharp decrease of @xmath128 in the relation @xmath135 .
extending the earlier generalizations @xcite of the concept of a ligand , we have treated a microtubule ( mt ) as a ` ligand ' that is tethered to a ` receptor ' wall by a group of minus - end directed molecular motors @xcite .
the tails of the motors are permanently anchored on the wall while their motor heads can bind to- and unbind from the mt .
this model of mt - wall attachment captures only a few key ingredients of the mt - cortex attachments in eukaryotic cells , particularly those formed during chromosome segregation .
this minimal model incorporates the polymerization and depolymerization kinetics of mt .
but , for the sake of simplicity , it does not include the processes of ` catastrophe ' and ` rescue ' that are caused by the
dynamic instability of mt filaments citedesai97 although these can be captured in an extended version of this model @xcite .
we consider a pre - formed mt - wall attachment and carry out computer simulations to study statistical properties of its rupture under conditions that mimic the protocols of force - clamp and force - ramp experiments _ in - vitro _ @xcite .
the simulation results that we report are interpreted in the light of the theory of single - molecule force spectrocopy , popularized by bell @xcite and some of its later generalizations @xcite . to our knowledge ,
the _ in - vitro
_ experimental set up ( schematically depicted in fig.[fig_mtdynein ] ) that comes closest to our model is that used by laan et al .
@xcite ; the wall in our model corresponds to the microfabricated vertical barrier that mimics the cell cortex in their experiment .
dynein motors were anchored on the barrier just as we have depicted in fig.[fig_xm ] for our model . in the experiment these dyneins captured the mt that grew from a centrosome which was fixed on a horizontal glass surface . a slightly different _ in - vitro _ experimental set us was used by hendricks et al .
@xcite ; in this experiment a dynein coated bead was used to mimic the cell cortex .
however , within the broader context of the role of mt - cortex interaction in positioning of the mitotic spindle , the authors of refs.@xcite , demonstrated only the stabilization of the mt by the dynein tethers .
in contrast , the main aim of our study is to focus on the ltd and rfd for the pre - formed attachment of a mt with a wall tethered by minus - end directed motors .
this work has been supported by a j.c .
bose national fellowship ( dc ) and by the `` prof .
s. sampath chair '' professorship ( dc ) . | a microtubule ( mt ) is a tubular stiff filament formed by a hierarchical organization of tubulin proteins .
we develop a stochastic kinetic model for studying the strength and stability of a pre - formed attachment of a mt with a rigid wall where the mt is tethered to the wall by a group of motor proteins .
such an attachment , formed by the specific interactions between the mt and the motors , is an analog of ligand - receptor bonds ; the mt and the motors anchored on the wall being the counterparts of the ligand and receptors , respectively . however , unlike other ligands ,
the length of a mt can change with time because of its polymerization - depolymerization kinetics .
the simple model developed here is motivated by the mts linked to the cell cortex by dynein motors .
we present the theory for both force - ramp and force - clamp conditions . in the force - ramp protocol
we investigate the strength of the attachment by assuming imposition of a time - dependent external load tension that increases linearly with time till the attachment gets ruptured ; we calculate the distribution of the rupture forces that fluctuates from one loading to another . in the force - clamp protocol , to test the stability , we compute the distribution of the lifetimes of the attachments under externally applied time - independent load tension ; the results establish the mt - wall attachment to be an analog of a slip - bond . |
iron - based superconductors are currently at the focus of condensed - matter research . discovered less than two years ago @xcite ,
these novel materials have attracted attention not only because of their high critical temperatures , but also because of their similarities with cuprates .
their phase diagram resembles that of the cuprates , most notably , the superconducting state seems to develop from an unconventional normal state , after magnetic order is destroyed by doping .
they are also layered materials , consisting of feas , fete , fes or fese planes , separated by spacer layers . however , there are also some important differences .
most notably , the parent compounds of cuprates are antiferromagnetic insulators , whereas the parent compounds of iron - based superconductors are antiferromagnetic spin density wave ( sdw ) metals @xcite .
optical spectroscopy is a powerful probe of electrodynamic response of high - t@xmath3 superconductors @xcite .
optical constants provide insight into low - energy excitations and charge dynamics , critical for understanding physics of strongly correlated systems .
the information obtained from optical constants can be used to test existing theories and/or stimulate development of new theoretical models .
optical spectroscopy is also a crucial experimental method for electronic band structure determination .
several families of iron - based superconductors have been discovered and they are conveniently refereed to as the `` 11 '' , `` 1111 '' or `` 122 '' families @xcite .
the 11 family is peculiar because the spacer layers are absent , and it is believed that this family will allow the intrinsic properties of iron - containing planes to be isolated . however , most of optical studies so far have focused on the 122 family @xcite ( and to a lesser extent 1111 family @xcite ) for which large single crystals can be grown . in this work , we have investigated the electrodynamic response of a member of the 11 ( fete ) family . to the best of our knowledge
this is the first optical study of a superconducting member of this family
. the only previous ir study on 11 family was on a non - superconducting fe@xmath5te ( ref . ) .
structural analysis has shown that the exact chemical composition of the analyzed sample is fe@xmath0te@xmath1s@xmath2 ( ref . ) .
note that in addition to being doped with s , this sample also has 6 @xmath6 of excess iron , which might play important role in charge dynamics @xcite .
magnetization measurements have revealed that the studied system undergoes structural and magnetic transitions , with transition temperature around 23 k ( ref . ) .
transport measurements on the other hand do not display any signatures of these transitions , and dc resistivity monotonically increases as temperature decreases down to 8 k , when the system undergoes superconducting transition and the resistivity abruptly drops to zero .
infrared ( ir ) reflectance measurements were performed at the university of akron on a bruker ifs 66v / s , whereas uv - visible experiments were conducted using varian / cary 300 .
an overfilling technique was used to obtain the absolute values of reflectance from the sample with surface area of approximately 1 mm @xmath7 1 mm ( ref . ) .
electrodynamic response was probed in the frequency range 50 - 50 000 @xmath4 ( 6 mev - 6.2 ev ) and as a function of temperature in the range 10 k - 300 k , all in the normal state .
the optical constants were extracted from reflectance data using kramers - kronig ( kk ) analysis .
in addition , we have performed magneto - optical measurements at 4.2 k ( in the superconducting state ) with magnetic fields up to 18 tesla .
figure [ fig : ref ] displays raw reflectance data r(@xmath8 ) and the real part of optical conductivity @xmath9 .
the absolute value of reflectance gradually decreases with frequency from about 0.9 at 50 @xmath4 to about 0.25 at 50000 @xmath4 , which can be interpreted as metallic behavior .
however the temperature dependence is opposite from expected @xcite , as the reflectivity decreases with decreasing temperature .
this anomalous behavior is even more obvious in @xmath9 : zero - energy ( drude - like ) peak is absent from the data ( except maybe at 300 k ) and the conductivity decreases monotonically with decreasing temperature at the lowest measured frequencies .
this result indicates that at low temperatures quasiparticles are not well defined , which will become even more obvious from the extended drude analysis below .
fe@xmath0te@xmath1s@xmath2 therefore can be considered to be an `` incoherent metal '' @xcite .
although the overall behavior of reflectance is metallic , the plasma edge can not be clearly resolved in the spectra , as reflectance gradually decreases with frequency . at higher frequencies the response is dominated by interband transitions , but they also are not easily discerned in the spectra . a shoulder in optical conductivity around 14000 @xmath4 ( 1.7 ev ) may originate from transitions involving iron _
3d _ states , similar to what was predicted by haule et al . in a theoretical study of another iron - based superconductor @xcite .
we also do not observe any phonon peaks in the far - ir , similar to fe@xmath5te ( ref . ) .
structural and magnetic transitions at 23 k do not seem to have any significant effect on optical spectra .
theoretical calculations of fese , based on lda+dmft , predicted `` incoherent metal '' with a pseudogap at low frequencies @xcite , in accord with our results .
these calculations also predict a smooth crossover of optical conductivity to a power law behavior @xmath10 at higher frequencies @xcite .
dashed black line in fig .
[ fig : ref](b ) represents the best fit to the optical conductivity and the obtained power law @xmath11 .
te@xmath1s@xmath2 at several different temperatures ranging from 10 k to room temperature .
the temperature dependence is restricted to the region below about 4000 @xmath4 .
( b ) the optical conductivity @xmath9 extracted directly from reflectance using kk analysis .
the total dl fit , as well as the individual components of the fit , are shown with gray lines .
the dashed black line is the fit to the power - law behavior at higher frequencies.,width=359 ] [ fig : ref ] to gain further insight into the electronic properties of fe@xmath0te@xmath1s@xmath2 we fit the data using a standard drude - lorentz ( dl ) model @xcite . the minimal model to achieve a good fit consisted of a drude and three lorentzian modes , centered at around 300 @xmath4 , 3000 @xmath4 and 14800 @xmath4 .
the total fits at all temperatures , as well as the three individual contributions at 10 k , are shown in fig .
[ fig : ref](b ) with gray lines .
the lowest lying oscillator displays most prominent temperature dependence .
its energy and intensity grow significantly as temperature decreases .
the mid - ir peak at 3000 @xmath4 ( 372 mev ) might be a generic feature of iron - based superconductors ; similar peaks have been observed in other families @xcite .
the oscillator at 14800 @xmath4 simulates the effect of interband transitions , presumably involving iron _
3d _ states , as discussed above . in the one - component approach
one assumes that only a single type of carriers are present in the system , but their scattering rate acquires frequency dependence @xcite . within
the so - called `` extended '' drude model one calculates the optical scattering rate @xmath12 and effective mass @xmath13 from the complex optical conductivity @xmath14 as : @xmath15 \label{eq : tau}%\ ] ] @xmath16 \frac{1}{\omega } \label{eq : mass}%\ ] ] where the plasma frequency @xmath17 ( @xmath18 is the carrier density and @xmath19 their band mass ) can be estimated from the integration of @xmath20 up to the frequency of the onset of interband absorption .
however , as pointed out above , the plasma edge is not very prominent in the spectra of fe@xmath0te@xmath1s@xmath2 , rendering the value of plasma frequency ambiguous .
instead , we fit the value of @xmath21 in eq .
[ eq : mass ] so that the effective mass at frequencies around 3500 @xmath4 is equal to unity ( dashed line in fig .
[ fig : tau](b ) ) .
the best fit is achieved for @xmath21= 26,000 @xmath4 ( 3.2 ev ) and the results of the analysis are shown in fig .
[ fig : tau ] .
the features of an `` incoherent metal '' are now obvious .
the scattering rate is relatively flat and featureless at room temperature , but at low temperatures it develops a peak below 180 @xmath4 .
the effective mass , on the other hand , becomes negative .
these feature of the spectra are indications that the quasiparticle concept , on which the extended drude model is based , is not applicable at low temperatures . and effective mass @xmath13 obtained from eqs .
[ eq : tau ] and [ eq : mass ] .
both optical functions reveal the failure of the quasipaticle concept at low temperatures.,width=359 ] [ fig : tau ]
model - independent sum rules are important tools in condensed - matter physics @xcite .
the so - called effective spectral weight function n(@xmath8 ) is frequently used for the analysis of ir spectra .
it is defined as @xcite : @xmath22 and for @xmath23 it becomes the global oscillator strength sum rule : @xmath24 which is a statement on the conservation of electric charge @xcite .
eq . [ eq : n ] is often used to quantify spectral weight redistribution between temperatures t@xmath25 and t@xmath26 in the form : @xmath27 fig .
[ fig : gsw ] shows the results of this analysis applied to fe@xmath0te@xmath1s@xmath2 at t@xmath25= 77 k and t@xmath26= 10 k. @xmath28 has a characteristic shape , which indicates that the spectral weight is removed from the low frequency region , below 1000 @xmath4 and is transferred to higher frequencies , in the region around 2000 - 4000 @xmath4 . within the error bars ,
the spectral weigh is recovered by the mid - ir region .
however , @xmath29 spectrum can not reveal the energy scale at which the transfer of spectral weight occurs . in order to address this question
we introduce `` generalized spectral weight '' function @xmath30 : @xmath31 the idea comes from the correlation functions frequently used in signal processing @xcite .
more recently ( auto)correlation function was used for the analysis of arpes spectra , where it is directly related to the quasiparticle density of states @xcite . note that @xmath30 is a function of the frequency shift @xmath32 , not the upper integration limit @xmath8 .
function @xmath30 is expected to display characteristic features at the values of energy shifts @xmath33 that connect regions between which a large amount of spectral weight is transferred . in practical applications ,
function @xmath30 is usually dominated by the spectral weight which does not participate in redistribution , and these characteristic features might not be obvious . in those cases
the first derivative of @xmath30 is useful .
examples of these calculations will be presented in a separate publication @xcite . in fig .
[ fig : gsw ] we show the results of generalized spectral weight analysis ( eq . [ eq : aleph ] ) applied to fe@xmath0te@xmath1s@xmath2 at t@xmath25 = 77 k and t@xmath26 = 10 k. the @xmath30 is dominated by the spectral weight that does not participate in transfer , so instead we display the first derivative @xmath34 .
the @xmath34 spectrum reveals a broad peak centered around 2250 @xmath4 ( 280 mev ) .
we take this as the characteristic energy scale over which the majority of spectral weight is transferred between 10 and 77 k. ) and generalized spectral weight ( eq . [ eq : aleph ] ) analysis for fe@xmath0te@xmath1s@xmath2 at t@xmath25 = 10 k and t@xmath26 = 77 k.,width=359 ] [ fig : gsw ]
in a density functional study of fes , fese and fete it was reported that the fermi surface and electronic structure of these compounds is similar to those of iron pnictides ( 1111 and 122 families ) @xcite .
therefore , the 11 family was suppose to be a model system in which to study the intrinsic properties of iron - containing planes .
however , our results , combined with previous ir studies , indicate that there might be some important differences between the electronic structure of 11 and the other families .
the reflectance of fe@xmath0te@xmath1s@xmath2 is similar to the reflectance of fe@xmath5te ( ref . ) . in the far - ir region
the reflectance of fe@xmath5te decreases with decreasing temperature , resulting in a reduction of conductivity in far - ir region , similar to what we observe in fe@xmath0te@xmath1s@xmath2 .
chen et al .
speculate that this incoherent transport in fe@xmath5te is caused by strong scattering from excess iron @xcite . however , there are also some important differences between fe@xmath0te@xmath1s@xmath2 and fe@xmath5te .
we observe no rapid increase in conductivity at low frequencies for 10 k measurements . on the other hand in fe@xmath5te a narrow drude - like peak develops in the optical conductivity at low temperatures .
this coherent behavior appears below structural and magnetic phase transition at 65 k , which implies that it is related to sdw order . in fe@xmath5te
this phase transition has stronger influence on charge transport than in fe@xmath0te@xmath1s@xmath2 : the dc resistivity changes character from insulating to metallic below the transition .
incoherent charge transport in the 11 family should be contrasted with a coherent response which has been observed in 1111 @xcite and 122 families @xcite .
infrared spectra of both undoped ( parent ) and doped phases of these families display well defined drude - like modes .
similar to fe@xmath5te sdw transition has a dramatic effects on their optical properties .
the response of parent compounds bafe@xmath26as@xmath26 and srfe@xmath26as@xmath26 becomes even more coherent below the sdw transitions @xcite as the width of drude mode is reduced by an order of magnitude .
the infrared spectra of these parent compounds are dominated by the mid - ir peak , which may have the same origin as the peak we observe in fe@xmath0te@xmath1s@xmath2 around 3000 @xmath4 ( 372 mev ) . the absence of sdw gap from ir spectra of both fe@xmath0te@xmath1s@xmath2 and fe@xmath5te ( ref . )
is also interesting .
recent arpes study of a parent compound fe@xmath35te ( ref . ) has also revealed that the sdw gap is absent .
optical spectra reveal that spectral weight is shifting with temperature , and the generalized spectral weight analysis we introduced indicates that a typical energy scale for the shift is about 280 mev .
the spectral weight of fe@xmath0te@xmath1s@xmath2 is removed from the low - energy region , which can be interpreted as a pseudogap feature , however we point out that this behavior starts already at room temperature ( fig .
[ fig : ref ] ) and therefore is unlikely to be related to sdw transition .
all this indicates that the electronic structure of the 11 family might be different from the 1111 and 122 families , for which clear signatures of sdw gaps have been observed @xcite .
ir studies on these two families have found a gap ( or even several gaps ) in the excitation spectra . on the other hand
both the parent compound @xcite and a doped sample studied in this work did not reveal the presence of a gap in their excitation spectra .
finally we address what happens below 8 k , when the system becomes superconducting .
we have performed magneto - optical studies in 18 tesla superconducting magnet at the national high magnetic field lab .
[ fig : magnet ] displays the results of these measurements .
the magneto - reflection ratio r(18 t)/r(0 t ) is shown as a function of frequency . apart from the vertical offset @xcite , within the noise level
the ratio is a straight line , which indicates the absence of field - induced effects in fe@xmath0te@xmath1s@xmath2 .
this is in contrast with bafe@xmath36co@xmath37as@xmath26 with t@xmath3= 22 k ( optimally doped member of the 122 family ) , where clear field - induced changes have been observed caused by the suppression of the superconducting gap @xcite .
we speculate that field - induced changes in fe@xmath0te@xmath1s@xmath2 are not observed because either they are below the detection limit of our experiment , or because the superconductiong gap is outside of our frequency window . , and this part of the spectrum is not shown @xcite.,width=359 ] [ fig : magnet ]
in summary , we have presented the results of infrared and optical spectroscopy studies of novel iron - based superconductor fe@xmath0te@xmath1s@xmath2 .
the results indicate incoherent normal state charge transport and absence of well defined quasiparticles at all temperatures down to t@xmath3 .
we have introduced `` generalized spectral weight analysis '' and used it to track redistribution with temperature .
the analysis reveals that the characteristic energy scale for the spectral weight shifts is approximately 280 mev .
our results , combined with previous reports , indicate that there are important differences between 11 and other families of iron - based superconductors .
we thank d.n .
basov and c.c .
homes for critical reading of the manuscript .
special thanks to r.d .
ramsier for the use of his equipment .
research at brookhaven national laboratory ( rongwei hu and c. p. ) is supported by the u.s .
department of energy , office of basic energy sciences , division of materials sciences and engineering under award @xmath38 ( de - ac02 - 98ch10886 ) . w.z .
hu , q.m . zhang and n.l .
wang , physica c * 469 * , 545 ( 2009 ) j.yang , d. huvonen , u. nagel , t. room , n. ni , p.c .
canfield , s.l .
budko , j.p .
carbotte and t. timusk , physical review letters * 102 * , 187003 ( 2009 ) .
hu , j. dong , g. li , z. li , p. zheng , g.f .
chen , j.l .
luo and n.l .
wang , physical review letters * 101 * , 257005 ( 2008 ) .
f. pfuner , j.g .
analytis , j .- h .
chu , i.r .
fisher and l. degiorgi , the european physical journal b * 67 * 513 ( 2009 ) .
a. akrap , j.j .
cao , z.a .
xu and c.c .
homes , physical review b * 80 * , 180502(r ) ( 2009 ) k.w .
kim , m. rossle , a. dubroka , v.k .
malik , t. wolf and c. bernhard , arxiv:0912:0140 ( 2009 ) .
e. van heumen , y. huang , s. de jong , a.b .
kuzmenko , m.s .
golden and d. van der marel arxiv:0912.0636v1 ( 2009 ) .
b. gorshunov , d. wu , a. a. voronkov , p. kallina , k. iida , s. haindl , f. kurth , l. schultz , b. holzapfel and m. dressel , arxiv:0912.1256v1 ( 2009 ) .
d. wu , n. barisic , p. kallina , a. faridian , b. gorshunov , n. drichko , l. j. li , x. lin , g. h. cao , z. a. xu n. l. wang and m. dressel , arxiv:0912.3334v1 .
qazilbash , j.j .
hamlin , r.e .
baumbach , m.b .
maple and d.n .
basov , arxiv:0808:3748 ( 2009 ) .
chen , r.h .
yuan , t. dong and n.l .
wang , arxiv:0910:1318v1 ( 2009 ) .
chen , z.g .
chen , j. dong , w.z .
hu , g. li , x.d .
zhang , p. zheng , j.l . luo and n.l .
wang , physical review b. * 79 * , 140509(r ) ( 2009 ) .
rongwei hu , emil s. bozin , j.b .
warren and c. petrovic , physical review b * 80 * , 214514 ( 2009 ) .
u. chatterjee , m. shi , a. kaminski , a. kanigel , h. m. fretwell , k. terashima , t. takahashi , s. rosenkranz , z. z. li , h. raffy , a. santander - syro , k. kadowaki , m. r. norman , m. randeria and j. c. campuzano , physical review letters * 96 * , 107006 ( 2006 ) .
y. xia , d. qian , l. wray , d. hsieh , g.f .
chen , j.l .
luo , n.l .
wang and m.z .
hasan , physical review letters * 103 * , 037002 ( 2009 ) .
dordevic , l.w .
kohlman , l. c. tung , y .- j .
wang , a. gozar , g. logvenov and i. bozovic , physical review b * 79 * , 134503 ( 2009 ) . | we have used optical spectroscopy to probe the normal state electrodynamic response of fe@xmath0te@xmath1s@xmath2 , a member of the 11 family of iron - based superconductors with t@xmath3= 8 k. measurements have been conducted over a wide frequency range ( 50 - 50000 @xmath4 ) at selected temperatures between 10 and 300 k. at low temperatures the material behaves as an `` incoherent metal '' : a drude - like peak is absent from the optical conductivity , and all optical functions reveal that quasiparticles are not well defined down to the lowest measured temperature .
we introduce `` generalized spectral weight '' analysis and use it to track temperature induced redistribution of spectral weight .
our results , combined with previous reports , indicate that the 11 family of iron - based superconductors might be different from other families . |
over many years the study of phase transitions in fluid mixtures has become an important field because of their omnipresent occurrence in nature and more importantly because of their technological meaning in creating new materials@xcite : blending of different species can reduce cost , improve processibility , provide synergy between components and allow for recycling .
sophisticated analytical techniques
e.g. , self consistent field theory@xcite or p - rism theory@xcite exist for calculating the phase behaviour and detailed interfacial properties of polymer blends . because of the large extension of polymer chains equilibrium properties
are well described ; however , the analytical description is much less satisfactory for the dynamics of multi component systems .
the transition from a completely homogeneous mixture to an equilibrated two phase system is a very long process consisting of a sequence of highly inhomogeneous states .
many different methods to analyse this process have been applied including experiments , theory , and computational simulations .
after quenching a polymer blend from the one phase region deep into the two phase region spinodal decomposition takes place .
different time regimes are recognised during spinodal decomposition . during early stages
the amplitudes of the concentration fluctuations , that are amplified , are still so small that they do not interact with each other . for later times
the local composition reaches the equilibrium value of the coexisting phases , and nonlinear interactions between the fluctuation modes become more and more important . at even later stages hydrodynamics dominates the coarsening@xcite .
spinodal decomposition can be approximately described through time dependent ginzburg
landau theory , also known as cahn hilliard
cook theory@xcite .
many efforts have been made to numerically calculate the time evolution of a mixture in the framework of this theory@xcite .
these calculations are appropriate to give a first insight into how phase separation takes place , but are far from describing the system quantitatively well .
even though binary polymer blends are an ideal testing bed for these approaches , earlier monte carlo simulations@xcite found rather pronounced deviations from their predictions . to find a quantitatively better description ,
it is necessary to go beyond ginzburg
landau theory .
self consistent field theory@xcite ( scft ) has proven to be one of the most successful descriptions of equilibrium properties of polymer mixtures on a mean field level .
the idea to use scft to develop a dynamical mean field theory is not new@xcite , but , although scft describes equilibrium properties well , usually the influence of the single chain dynamics is neglected due to computational expenses , again leading to a more qualitative than quantitative description of the phase separation . in most calculations
a simple constant onsager coefficient is used which would be appropriate if the movement of the polymers were comparable to the movement of point particles . remembering the connectivity of polymers it is not likely that this local coupling is sufficient to lead to a quantitatively correct description of the dynamics and other nonlocal onsager coefficients have been proposed@xcite . in this study
we are interested in the early stages of spinodal decomposition in an incompressible symmetric binary polymer mixture after a quench from the one phase region to the two phase region paying special attention to the differences in the collective dynamics when the chains obey local and rouse dynamics .
we therefore employ two dynamical versions of self consistent field theory for polymer mixtures . the first method , referred to as dynamic self consistent field theory@xcite ( dscft ) , uses the free energy functional of the scft for an incompressible mixture in terms of the local composition .
this free energy functional leads to a kinetic equation for the time dependent local composition which is integrated numerically .
this method has been applied for studying the ordering kinetics in block copolymer and surfactant systems@xcite .
we use both a local and a nonlocal onsager transport coefficient , as is expected for polymers obeying rouse dynamics .
the underlying idea of the second method is to express the free energy of the investigated system only in terms of the effective external fields , which are thermodynamically conjugated to the composition .
this description leads to an equation of motion for the effective external fields , which is integrated in time . following maurits _
et al._@xcite who derived a similar equation of motion neglecting random fluctuations we refer to this method as the external potential dynamics ( epd ) , which has been shown to automatically incorporate rouse dynamics .
the aim of our study is to explore both methods and to investigate the role of the onsager coefficient . to decide whether our results resemble any ` real ' dynamics a comparison with results obtained through other methods is preferable
although many experiments have analysed spinodal decomposition in binary polymer mixtures only few@xcite have looked at early stages of demixing and the influence of the onsager coefficient .
another difficulty in comparing experimental results with our calculations lies in the fact that actual demixing is influenced by many factors , for example , preparation of the probe , polydispersity , or strong dynamic asymmetries due to different glass transition temperatures of the two species@xcite .
therefore a quantitative mapping between theory and experimental systems is difficult . to gain a quantitative insight
we compare our mean field calculations with results obtained through monte carlo simulations .
no parameters have to be adjusted for this comparison . in the simulations we employ the bond fluctuation model@xcite which is well established for studying properties of polymer melts .
it has been shown that the equilibrium properties of a polymer mixture as given by the self consistent field theory are almost quantitatively reproduced through monte carlo simulations using the bond fluctuation model@xcite .
our paper is organised in the following way . in section
[ sec : scft ] we introduce the self consistent field theory for a binary polymer mixture , explaining how equilibrium properties can be derived with this method .
then we introduce general aspects of the dynamics in a polymer mixture and show how they can be incorporated in the scft leading to dynamic self consistent field theory .
the following section presents another approach to including dynamics in scft regarding the effective external fields as the time dependent variable .
implications regarding density fluctuations in this description are discussed in detail .
section [ sec : mc ] serves as a brief introduction to the performed monte carlo simulations and explains how a direct comparison with scft calculations is possible . in section [ sec : results ] results of our calculations and simulations are presented first showing the difference between the use of local and nonlocal coupling , then comparing the mean field results with the simulations and finally analysing the influence of random fluctuations .
the paper finishes with a summary .
we consider an incompressible mixture of a and b polymers consisting of @xmath0 polymers in a volume @xmath1 with periodic boundary conditions .
there are @xmath2 polymers of kind a in the system , for b polymers respectively @xmath3 . in scft@xcite polymers
are modelled as gaussian chains with the end to
end distance @xmath4 .
in addition we choose the a and b polymers to both have the same number of monomers @xmath5 and to be of the same architecture .
the overall particle density in the system is denoted as @xmath6 .
the microscopic density @xmath7 of the a monomers can be defined through the polymer conformations @xmath8 : @xmath9 @xmath10 parameterises the contour of a chain . for b monomers a similar equation holds .
( all the following equations regarding only a monomers , are equivalently valid for b monomers without mentioning this explicitly . ) regarding a repulsion between the two kinds of polymers that is expressed through the flory
huggins parameter @xmath11 , the canonical partition function has the following form : @xmath12{\mathcal{d}}[{\mathbf{r}}_{i_b } ] { \mathcal{p}}_a[{\mathbf{r}}_{i_a}]{\mathcal{p}}_b[{\mathbf{r}}_{i_b}]\right)\cdot\\ & \cdot\exp\left[-\rho\int_v d^3{\mathbf{r}}\chi\hat{\phi}_a\hat{\phi}_b\right ] \delta(\hat{\phi}_a+\hat{\phi}_b-1 ) \end{split } \label{eq : part_bulk}\ ] ] the functional integral @xmath13 sums over all possible conformations of the chains .
@xmath14 $ ] is the so called wiener measure @xmath15\sim\exp\left[-\frac{3}{2r_e^2 }
\int_{0}^{1}d\!\tau\left(\frac{d { \mathbf{r}}}{d\tau}\right)^2\right ] \label{eq : wien_mes}\ ] ] which represents the statistical weight of a noninteracting gaussian chain .
repulsive interactions at short interparticle distances reduce fluctuations of the total density .
this is incorporated effectively via the incompressibility constraint which is expressed through the @xmath16function in eq.([eq : part_bulk])@xcite .
this defines a system with many interacting polymer chains ; its partition function obviously can not be analytically solved . inserting new auxiliary field variables @xmath17 , @xmath18 , @xmath19 , @xmath20 via a hubbard
stratonovich transformation , we can reformulate the many polymer problem in terms of a single polymer in external fields .
@xmath21/k_bt\right ] \label{eq : part_new}\ ] ] thus we have found an expression for the canonical free energy depending on the new variables and the single chain partition function @xmath22 : @xmath23}{k_bt}&= -\frac{\bar{\phi_a}\rho v}{n}\ln\frac{q_a}{n_a}-\frac{\bar{\phi_b}\rho v}{n}\ln\frac{q_b}{n_b}\\ & -\frac{\rho}{n}\int_v d^3{\mathbf{r}}(w_a\phi_a+w_b\phi_b)+ \frac{\rho}{n}\int_v d^3{\mathbf{r}}\chi n\phi_a\phi_b , \end{split } \label{eq : free_en}\ ] ] @xmath24 denoting the average density of the a polymers in the system and @xmath25{\mathcal{p}}_a[{\mathbf{r } } ] \exp\left[-\int_0^{1}d\tau w_a({\mathbf{r}}(\tau))\right ] \label{eq : si_ch_part}\ ] ] being the partition function of a single a chain in the external field @xmath19 .
the functional integral in eq.([eq : part_new ] ) also can not be calculated explicitly .
therefore we employ a saddle point approximation .
this means that only the largest contribution to the integrand is considered and the integration does not have to be carried out .
the saddle point approximation of eq.([eq : part_new ] ) is equivalent to the minimisation of the free energy ( [ eq : free_en ] ) with respect to the auxiliary variables .
the values of the fields and densities at the saddle point are denoted by lower case letters and are given by the set of equations that has to be solved self consistently : @xmath26 \label{eq : saddle_phi_a}\\ \frac{\delta f}{\delta w_b}\arrowvert_{w_b}=0 & \quad:\quad & \phi_b=-\frac{\bar{\phi}_b v}{q_b}\frac{\delta q_b}{\delta w_b } \equiv \phi^*_b[w_b ] \label{eq : saddle_phi_b}\\ \frac{\delta f}{\delta \phi_a}\arrowvert_{\phi_a}-\frac{\delta f}{\delta \phi_b}\arrowvert_{\phi_b}=0 & \quad:\quad & w_a - w_b=\chi n(\phi_b-\phi_a)\label{eq : eq_w_a}\\ & \phi_a+\phi_b=1 & \label{eq : eq_w_b}\end{aligned}\ ] ] the averages of the microscopic densities are given by : @xmath27 and @xmath28 .
eqs.([eq : saddle_phi_a ] ) and ( [ eq : saddle_phi_b ] ) make clear that in equilibrium the average microscopic density @xmath29 actually equals the thermal average monomer density of a single a chain in an external field @xmath30 . in other words after starting with a description where we would have to take all possible interactions of every polymer chain into account , see eq.([eq : part_bulk ] ) , we have now found a mean field description of the system in which it is sufficient to look at a single chain in an effective external field . to calculate the monomer densities it is useful to define the end segment distribution @xmath31 , which gives the probability to find the end of a chain with length @xmath32 at position @xmath33 when exposed to a field @xmath30 : @xmath34{\mathcal{p}}_a[{\mathbf{r}}(t ) ] \delta({\mathbf{r}}(t)-{\mathbf{r}})\exp\left[-\int^t_0 d\tau w_a({\mathbf{r}}(\tau))\right ] , \label{eq : end_seg_dis}\ ] ] for this end segment distribution the following diffusion equation holds@xcite @xmath35 with the boundary condition @xmath36 .
after finding a solution to this equation the monomer density is immediately given through the following expression : @xmath37 the single chain partition function can be calculated with : @xmath38 after having presented all necessary equations to calculate equilibrium properties of polymer mixtures , we are now interested in using this description for calculating the dynamics in a polymer mixture .
two ways to achieve this are introduced in the following sections . because of the fact that the concentrations of the polymers are conserved the continuity equation is valid : @xmath39 @xmath40 denoting the current density of the monomers at position @xmath33 at time @xmath32 .
one now assumes@xcite a linear relation between the current density and the gradient of the exchange potential @xmath41 ( note there is only one independent chemical potential in the system because of the incompressibility constraint ) @xmath42 with the kinetic coefficient @xmath43 describing the connection between the force acting on the monomers through the gradient of the chemical potential at position @xmath44 and the resulting current density at position @xmath33 .
this describes a purely relaxational dynamics , effects due to hydrodynamic flow are not captured .
the onsager coefficient @xmath45 can be modelled in different ways : the simplest approach would be local coupling which results in the onsager coefficient being proportional to the local density .
bearing in mind that polymers have a certain extension it is clear that nonlocal coupling should lead to a better description , although local coupling is often used in calculations of dynamic models based on ginzburg landau type energy functionals for simplicity reasons@xcite . in the rouse model forces acting on a monomer caused by the other monomers
are also taken into account@xcite .
this leads to a kinetic factor that is proportional to the pair
correlation function@xcite .
these two approaches lead to the following onsager coefficients : @xmath46 @xmath47 denoting the single chain diffusion constant and @xmath48 the pair correlation function .
the rouse onsager coefficient written in eq .
is only approximately valid , when the pair
correlation function is the same for both polymer species .
a more general expression is found in reference@xcite .
another model for nonlocal coupling is the reptation model@xcite which is appropriate for polymer melts with very long chains , i.e. very entangled chains .
hereby the idea is that a polymer chain is constrained by the other polymers and is forced to move along the polymer tube axis .
this dynamics is expected@xcite to also lead to an onsager coefficient that is proportional to the pair correlation function .
therefore the influence of single chain rouse and reptation dynamics on the collective dynamics of the system is qualitatively comparable .
reptation shall not be regarded in the further study .
eqs.([eq : cont_eq ] ) and lead to the following diffusion equation the last term representing noise that obeys the fluctuation dissipation theorem .
@xmath49 after the fourier transformation this diffusion equation has a simple form : @xmath50 these diffusion equations implicitly assume that the relaxation time of the chain conformations is smaller than the time scale on which composition fluctuations evolve .
the chain conformations are expected to be ` in equilibrium ' with respect to the instantaneous spatially varying composition . to use the introduced diffusion equation ( [ eq : diff_eq ] ) in the frame of scft , let us return to the general expression ( [ eq : part_new ] ) that describes the canonical partition function depending upon the variables @xmath17 , @xmath18 , @xmath19 , @xmath20 that are independent of each other .
if we now employ a saddle point approximation in the variables @xmath19 and @xmath20 we find eqs.([eq : saddle_phi_a ] ) and ( [ eq : saddle_phi_b ] ) , that describe a unique relation between the fields @xmath30 , @xmath51 and the densities @xmath52 , @xmath53 .
this means we can calculate the densities explicitely with eqs.([eq : diff_eq_end ] ) , ( [ eq : dens_calc ] ) , and ( [ eq : si_ch_part_calc ] ) , if the fields @xmath30 and @xmath51 are known .
therefore with this approximation the free energy @xmath54 becomes a function that depends on the densities ( or fields ) only : @xmath55,w_b[\phi_b]\right]=f\left[\phi_a,\phi_b\right]$ ] . with this free energy
it is now possible to calculate the exchange potential @xmath56 : @xmath57}{\delta \phi_a({\mathbf{r}})}-\frac{\delta f[\phi_a({\mathbf{r}}),\phi_b({\mathbf{r}})]}{\delta \phi_b({\mathbf{r } } ) } = \frac{1}{n}\left\{\chi n(\phi_b({\mathbf{r}})-\phi_a({\mathbf{r}}))-(w_a[\phi_a({\mathbf{r}})]-w_b[\phi_b({\mathbf{r}})])\right\ } \label{eq : chem_pot}\ ] ] the simple form of diffusion equation ( [ eq : diff_eq_q ] ) and the diffusion equation for the end segment distribution of a polymer chain ( [ eq : diff_eq_end ] ) suggests to use a fourier expansion of all spatially dependent variables for actual calculations because the fourier functions are eigenfunctions of the squared gradient @xmath58 .
the following set of orthonormal functions is used in all our scft calculations : @xmath59 we have now found all necessary equations to numerically calculate the time evolution of the densities in a binary polymer mixture , leading us the following procedure we refer to as the dynamic self consistent field theory ( dscft ) method : first we have a given density profile at time @xmath60 . as mentioned before eqs.([eq : diff_eq_end ] ) , ( [ eq : dens_calc ] ) and ( [ eq : si_ch_part_calc ] ) give us the possibility to explicitly calculate the single chain densities for known external fields . unfortunately the inversion is needed because we have given densities and have to find the matching external fields .
this leads to a set of nonlinear equations , that is numerically solved through the newton
broyden method@xcite .
after the external fields , that ` create ' the given density profile , have been found the exchange potential @xmath61 is calculated via eq.([eq : chem_pot ] ) and then inserted into the diffusion equation ( [ eq : diff_eq_q ] )
. the diffusion equation is subsequently integrated using a simplified runge
kutta method .
this leads to a new given density after a discrete time step and the whole procedure starts anew .
apart from regarding local dynamics in our dscft calculations it is favourable to consider nonlocal coupling because this leads to a better description of the dynamics in a polymer mixture .
however in dscft the difficulty in using rouse dynamics lies in the computational expense of calculating the pair
correlation function for each time step . as an approximation during early stages of demixing the pair
correlation function of a homogeneous melt , as it is given through the random phase approximation ( rpa)@xcite , is used leading to the following onsager coefficient : @xmath62 @xmath63 is defined as @xmath64 with @xmath4 denoting the end to
end distance of a polymer . in the previously introduced dscft method we needed a way to reduce the number of independent variables in the partition function . through the saddle point approximation in the fields we obtained a free energy functional in terms of the densities .
this , in turn , yields a langevin equation for the dynamics of the densities . for the external potential dynamics ( epd )
method@xcite we are looking for a way to express the dynamics of the binary polymer mixture through an equation of motion for the external fields @xmath19 and @xmath20 .
our starting point is again the canonical partition function ( [ eq : part_bulk ] ) . via a hubbard stratonovich transformation we introduce the the field variables @xmath65 and @xmath66 and obtain : @xmath67{\mathcal{d}}[{\mathbf{r}}_{i_b } ] { \mathcal{p}}_a[{\mathbf{r}}_{i_a}]{\mathcal{p}}_b[{\mathbf{r}}_{i_b}]\right)\cdot \nonumber \\ & & \cdot \exp\left[-\rho\int_v d^3{\mathbf{r}}\frac{\chi}{4 } \left\ { ( \hat{\phi}_a+\hat{\phi}_b)^2 -(\hat{\phi}_a-\hat{\phi}_b)^2 \right\}\right ] \delta(\hat{\phi}_a+\hat{\phi}_b-1 ) \nonumber \\ & \sim & \frac{1}{n_a!n_b!}\int\left(\prod^{n_a}_{i_a=1}\prod^{n_b}_{i_b=1 } { \mathcal{d}}[{\mathbf{r}}_{i_a}]{\mathcal{d}}[{\mathbf{r}}_{i_b } ] { \mathcal{p}}_a[{\mathbf{r}}_{i_a}]{\mathcal{p}}_b[{\mathbf{r}}_{i_b}]\right)\cdot \nonumber \\ & & \cdot \int { \mathcal{d}}u { \mathcal{d}}w \exp\left[-\frac{\rho v \chi}{4 } \right ] \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r } } \left\{\frac{w}{2}(\hat\phi_a-\hat\phi_b)+\frac{w^2}{4\chi n } \right\}\right ] \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r}}\ ; \frac{u}{2}(\hat{\phi}_a+\hat{\phi}_b-1)\right ] \label{eq : zuw}\end{aligned}\ ] ] this defines a free energy function @xmath68 in terms of the fields @xmath69 and @xmath70 : @xmath71/k_bt\right ] \label{eq : part_uw}\ ] ] with @xmath72}{k_bt}= -\frac{\bar{\phi_a}\rho v}{n}\ln\frac{q_a[(u\!+\!w)/2]}{n_a}-\frac{\bar{\phi_b}\rho v}{n}\ln\frac{q_b[(u\!-\!w)/2]}{n_b } + \frac{\rho}{n}\int_v d^3{\mathbf{r}}\left [ \frac{w^2}{4\chi n } -\frac{1}{2}(u-\frac{\chi n}{2})\right ] \label{eq : free_en_uw}\ ] ] alternatively , we could have started with the free energy functional ( [ eq : part_new ] ) and integrate out the gaussian variables @xmath17 and @xmath18 .
so far no approximations have been used , but to find an energy functional that only depends upon @xmath70 we now employ a saddle point approximation with respect to @xmath69 : @xmath73}{\delta u}\arrowvert_{u^*}=0\quad:\quad\phi_a^*({\mathbf{r}})+\phi_b^*({\mathbf{r}})=1 \label{eq : saddle_u}\ ] ] here we use the definition , see eqs . and : @xmath74 . for @xmath75
the equivalent definition applies . in equilibrium
the field variable @xmath66 is conjugated to the overall density of the system , which is constant in an incompressible mixture .
we therefore believe that the influence of this approximation on the description of the system through the field @xmath70 is very small .
we shall discuss this in detail below . if we replace @xmath69 with the @xmath76 $ ] that fulfils this constraint we end up with a free energy functional that only depends on the field variable @xmath70 : @xmath77}{k_b t}=\frac{\rho v \chi}{4}+\frac{\rho}{n}\int_v d^3{\mathbf{r}}\frac{w^2}{4\chi n } -\frac{\bar{\phi_a}\rho v}{n}\ln\frac{q_a[(u^*\!+\!w)/2]}{n_a}-\frac{\bar{\phi_b}\rho v}{n}\ln\frac{q_b[(u^*\!-\!w)/2]}{n_b } \label{eq : free_en_w}\ ] ] where we have used the fact that adding a constant field @xmath78 to @xmath79 does not change the value of @xmath80 $ ] .
we chose @xmath78 in such a way that @xmath81 .
a difficulty in describing the system with the order parameter @xmath70 is the interpretation of the field fluctuations in terms of the physical density fluctuations .
we can calculate the averages of the microscopic densities _
after _ the saddle point integration over @xmath69 .
these averages are marked by the subscript epd . to this end
, we introduce a local exchange potential @xmath82 , which couples to the microscopic density difference @xmath83 : @xmath84 & \sim & \frac{1}{n_a!n_b!}\int\left(\prod^{n_a}_{i_a=1}\prod^{n_b}_{i_b=1 } { \mathcal{d}}[{\mathbf{r}}_{i_a}]{\mathcal{d}}[{\mathbf{r}}_{i_b } ] { \mathcal{p}}_a[{\mathbf{r}}_{i_a}]{\mathcal{p}}_b[{\mathbf{r}}_{i_b}]\right ) \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r } } \left\{\frac{\delta \mu}{2}(\hat\phi_a-\hat\phi_b ) \right\}\right ] \cdot \nonumber \\ & & \cdot \int { \mathcal{d}}w \exp\left[-\frac{\rho v \chi}{4 } \right ] \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r } } \left\{\frac{w}{2}(\hat\phi_a-\hat\phi_b)+\frac{w^2}{4\chi n } \right\}\right ] \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r}}\ ; \frac{u^*[w]}{2}(\hat{\phi}_a+\hat{\phi}_b-1)\right ] \nonumber \\ & \sim & \int { \mathcal{d}}w \exp\left [ -\tilde{g}[w,\delta \mu ] \right ] \qquad \mbox{with } \nonumber \\ \tilde{g}[w,\delta \mu ] & = & \frac{\rho v \chi}{4 } + \frac{\rho}{n } \int_v
{ d}^3{\bf r}\ ; \frac{w^2}{4\chi n } -\frac{\bar\phi_a\rho v}{n}\ln\frac{q_a[(u^*+w+\delta\mu)/2]}{n_a}-\frac{\bar\phi_b\rho v}{n}\ln\frac{q_b[(u^*-w-\delta\mu)/2]}{n_b}\label{eq : av_1}\end{aligned}\ ] ] thermodynamic averages of the microscopic density difference are obtained via functional derivatives : @xmath85 } \frac{\delta\tilde{\cal z}[\delta \mu ] } { \delta \delta \mu({\bf r})}\arrowvert_{\delta \mu = 0 } \nonumber \\ & = & \langle \phi^*_a({\bf r})- \phi^*_b({\bf r } ) \rangle \label{eq : phi_av_ok } \\
\langle [ \hat\phi_a({\bf r})-\hat\phi_b({\bf r})][\hat\phi_a({\bf r'})-\hat\phi_b({\bf r ' } ) ] \rangle_{\rm epd } & = & \left(\frac{2n}{\rho}\right)^2 \frac{1}{\tilde{\cal z}[\delta \mu ] } \frac{\delta^2\tilde{\cal z}[\delta \mu ] } { \delta \delta \mu({\bf r } ) \delta\delta \mu({\bf r'})}\arrowvert_{\delta \mu = 0 } \nonumber \\ & = & \left\langle [ \phi^*_a({\bf r})-\phi^*_b({\bf r})][\phi^*_a({\bf r'})-\phi^*_b({\bf r ' } ) ] \right\rangle -\frac{n}{\rho}\left\langle\frac{\delta \phi_a^*({\bf r})}{\delta w_a({\bf r'})}+\frac{\delta \phi_b^*({\bf r})}{\delta w_b({\bf r ' } ) } \right\rangle \label{eq : phi^2_av_ok}\end{aligned}\ ] ] similarly , we can calculate the fluctuations of the total density , which are induced by the saddle point approximation .
@xmath86 & \sim & \frac{1}{n_a!n_b!}\int\left(\prod^{n_a}_{i_a=1}\prod^{n_b}_{i_b=1 } { \mathcal{d}}[{\mathbf{r}}_{i_a}]{\mathcal{d}}[{\mathbf{r}}_{i_b } ] { \mathcal{p}}_a[{\mathbf{r}}_{i_a}]{\mathcal{p}}_b[{\mathbf{r}}_{i_b}]\right ) \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r } }
\left\{\frac{\mu}{2}(\hat\phi_a+\hat\phi_b ) \right\}\right ] \cdot \nonumber \\ & & \cdot \int { \mathcal{d}}w \exp\left[-\frac{\rho v \chi}{4 } \right ] \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r } } \left\{\frac{w}{2}(\hat\phi_a-\hat\phi_b)+\frac{w^2}{4\chi n } \right\}\right ] \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r}}\ ; \frac{u^*[w]}{2}(\hat{\phi}_a+\hat{\phi}_b-1)\right ] \nonumber \\ & \sim & \int { \mathcal{d}}w \exp\left [ -\tilde{g}[w,\mu \right ] \qquad \mbox{with } \nonumber \\ \tilde{g}[w,\mu ] & = & \frac{\rho v \chi}{4 } + \frac{\rho}{n } \int_v { d}^3{\bf r}\ ; \frac{w^2}{4\chi n } -\frac{\bar\phi_a\rho v}{n_a}\ln\frac{q_a[(u^*+w+\mu)/2]}{n_a}-\frac{\bar\phi_b\rho v}{n_b}\ln\frac{q_b[(u^*-w+\mu)/2]}{n_b}\end{aligned}\ ] ] moments of the total density averaged over the field configurations of @xmath70 are given by : @xmath87 } \frac{\delta\tilde{\cal z}[\mu ] } { \delta \mu({\bf r})}\arrowvert _ { \mu = 0 } \nonumber \\ \rightarrow \qquad \qquad \langle \hat\phi_a({\bf r})+\hat\phi_b({\bf r } ) \rangle_{\rm epd } & = & \langle \phi^*_a({\bf r } ) + \phi^*_b({\bf r } ) \rangle = 1\\ \left(\frac{\rho}{2n}\right)^2 \langle ( \hat\phi_a({\bf r})+\hat\phi_b({\bf r})(\hat\phi_a({\bf r'})+\hat\phi_b({\bf r ' } ) \rangle_{\rm epd } & = & \frac{1}{\tilde{\cal z}[\mu ] } \frac{\delta^2\tilde{\cal z } [ \mu ] } { \delta\mu({\bf r } ) \delta \mu({\bf r'})}\arrowvert _ { \mu = 0 } \nonumber
\\ \rightarrow \qquad \qquad \langle ( \hat\phi_a({\bf r})+\hat\phi_b({\bf r})(\hat\phi_a({\bf r'})+\hat\phi_b({\bf r ' } ) \rangle_{\rm epd } & = & 1 - \frac{n}{\rho}\left\langle \frac{\delta \phi_a^*({\bf r})}{\delta w_a({\bf r'})}+\frac{\delta \phi_b^*({\bf r})}{\delta w_b({\bf r ' } ) } \right\rangle \label{eq : dens}\end{aligned}\ ] ] these equations describe the actual fluctuations of the microscopic composition of the system _ after _ the saddle point approximation , i.e. , of the epd method .
having performed a saddle point integration in @xmath69 , we have ignored fluctuations , and we can not expect eqs.([eq : phi^2_av_ok ] ) or ( [ eq : dens ] ) to be accurate . indeed , while the incompressibility constraint is fulfilled _ on average _ eq.([eq : dens ] ) demonstrates that the saddle point approximation leads to spurious fluctuations of the total density . in appendix [ app : rpa_epd ] we use the random phase approximation ( rpa ) to evaluate eq.([eq : phi^2_av_ok ] ) and show explicitely the deviations between eq.([eq : phi^2_av_ok ] ) and the well known rpa structure factor .
alternatively , we can deduce the exact averages from the full free energy functional @xmath88 $ ] in eq.([eq : free_en_uw ] ) by introducing a local exchange potential @xmath82 like in eq .. after a variable substitution @xmath89 this leads to : @xmath90 = g[u , w ] - \frac{\rho}{n } \int_v d^3{\mathbf{r}}\ ; \frac{-\delta \mu^2 + 2 \delta \mu w}{4\chi n}\ ] ] with this free energy functional we obtain the exact averages of the microscopic densities : @xmath91[\hat\phi_a({\bf r'})-\hat\phi_b({\bf r ' } ) ] \rangle_{uw } & = & \frac{-2 \delta({\bf r}-{\bf r'})}{\rho \chi } + \frac{1}{(\chi n)^2 } \langle w({\bf r})w({\bf r'})\rangle_{uw } \label{eq : phi^2_av}\end{aligned}\ ] ] recently , ganesan and fredrickson@xcite have used a complex langevin method to sample the fluctuations of both fields @xmath69 and @xmath70 and have obtained the average @xmath92monomer density as @xmath93 . since @xmath69 has to be complex to make the last term in eq.([eq : zuw ] ) a proper representation of @xmath94 individual contributions to this average also have an imaginary part and the numerical procedure is quite involved .
we expect the saddle point integration over @xmath69 to be accurate whenever @xmath88 $ ] can be well approximated by a parabola in @xmath95 . in this case the fluctuations of @xmath70 are only very little affected by the saddle point approximation in @xmath69 and the fluctuations of @xmath70 in the epd method will closely mimic the fluctuations of @xmath70 of the exact partition function ( [ eq : zuw ] ) .
hence , we can use eq.([eq : phi^2_av ] ) with @xmath96 ( i.e. , _
after _ the saddle point approximation ) to obtain a very good approximation for the structure factor . in appendix
[ app : rpa_epd ] we confirm that in rpa the fluctuations of @xmath70 are , of course , not affected by the saddle point integration over @xmath69 .
therefore we use the fourier transform of eqs.([eq : phi_av ] ) and ( [ eq : phi^2_av ] ) in our calculations : @xmath97 because the composition is conserved and @xmath98 , we expect the order parameter @xmath70 with which we are now describing our system to also be a conserved quantity
. therefore the dynamics of @xmath70 are given through the relaxational dynamics of a model b system , referring to the classification introduced by hohenberg and halperin@xcite .
@xmath99 with the chemical potential being the first derivative of the free energy with respect to the order parameter : @xmath100}{\delta w({\mathbf{r } } ) } = \frac{1}{n2\chi n}\left(w+\chi n\left[\phi_a^*({\mathbf{r}})-\phi_b^*({\mathbf{r}})\right]\right ) \label{eq : chem_pot_epd}\ ] ] the fourier transform of this new diffusion equation is : @xmath101+\eta(q ) \label{eq :
diff_eq_epd_q}\ ] ] @xmath102 is white noise that obeys the fluctuation dissipation theorem .
the method using this diffusion equation we refer to as the epd method@xcite . we have found a diffusion equation that describes the dynamics in terms of the external field @xmath103 and leads to the right physical equilibrium .
a similar equation without noise has been derived by maurits _
et al._@xcite .
the question to be asked is whether this dynamics represents any actual physical dynamics and how the choice of the onsager coefficient influences the dynamics of the densities .
it can be shown@xcite , see also appendix [ app : rouse_epd ] , that using local coupling in the epd method is a good approximation for rouse dynamics .
the onsager coefficient that we would have to use in dscft to reproduce rouse dynamics is given in eq.([eq : rouse_dyn ] ) .
the equivalent ( local ) kinetic coefficient in the epd method for the same dynamics of the densities is : @xmath104 for the epd calculations again the fourier expansion of eq.([eq : four_exp ] ) is used .
after having found the initial fields that create the given densities with the methods used for the dscft the chemical potential @xmath105 is calculated according to eq.([eq : chem_pot_epd ] ) .
@xmath105 is then plugged into eq.([eq : diff_eq_epd_q ] ) to find the time derivative of the difference in the fields .
thereafter @xmath106 is integrated via the simplified runge kutta method .
after we have found the new @xmath103 , we need to find the variable @xmath79 to make sure the incompressibility constraint @xmath107 given through the saddle point approximation ( [ eq : saddle_u ] ) is fulfilled again using the newton broyden method . the new fields lead us to the new chemical potential to calculate @xmath106 and so forth .
the method of the epd has two main advantages compared to dscft : first of all it incorporates nonlocal coupling and secondly proves to be up to an order of magnitude computationally faster .
there are two main reasons for this huge speed up : in epd the number of equations that have to be solved via the newton
broyden method to fulfil incompressibility is just the number of fourier functions used .
the number of equations in dscft that have to be solved to find the new fields after integrating the densities is twice as large .
on the other hand , comparing diffusion equation ( [ eq : diff_eq ] ) used in the dscft method with equation ( [ eq : diff_eq_epd ] ) in epd it is easily seen , that the right hand side of the latter is a simple multiplication with the squared wave vector of the relevant mode , whereas the right hand side of eq.([eq : diff_eq ] ) is a complicated multiplication of three spatially dependent variables .
the monte carlo simulations presented in this study make use of the bond fluctuation model ( bfm)@xcite , which is a coarse grained lattice model , that incorporates the relevant features of polymers . these are connectivity of the monomers along a chain , excluded volume of the segments and thermal interaction between monomers . in this model
each effective monomer occupies a cube of the lattice and blocks the eight sites at the cube corners for other monomers .
monomers of a chain are connected by one of 108 possible bond vectors of length 2 , @xmath108 , @xmath109 , 3 , or @xmath110 measured in units of the lattice spacing .
( all following lengths are assumed to be given in these units unless an explicit unit is given . )
these bond vectors are chosen to ensure the excluded volume condition , that makes sure that they do not cross each other during their movement .
this large number of possible bond vectors allows 87 different bond angles , that provide a good approximation for continuous connectivity between the monomers of the chain .
each of the effective monomers represents 35 real chemical repeat units@xcite .
the number density of the occupied sites is chosen to be @xmath111 , which reproduces the properties of a polymer melt well .
interactions between the monomers are modelled through a square well potential with monomers of the same kind attracting and monomers of different kinds repelling each other .
the interactions are chosen to be symmetric and to act inside a radius that extends over the first peak of the pair correlation function .
this means the interactions act up to a distance of @xmath109 which is equivalent to the 54 neighbouring cubes of a monomer .
@xmath112 the moves used to simulate a purely diffusive movement of the monomers are local random monomer hopping moves , where one tries to move a randomly picked monomer to a neighbouring lattice site .
the monte carlo simulations were carried out with chains of 64 effective segments in a box with length @xmath113 . with an overall number density of @xmath111
there are 256 000 particles in the system . in the monte carlo simulations
however the choice of @xmath114 , which is equivalent to a polymerisation of 200300 in real polymers , is a compromise between the possibility to compare the simulation results with mean field results and the largest still sensibly manageable amount of computational resources , because an increase in the polymerisation leads to both an increase in the length scale and a slowing down in the kinetics of phase separation . to actually compare monte carlo simulations with scft calculations the parameters of both models have to mapped onto each other : in scft the only present length scale is the end to
end distance @xmath4 of the polymers .
it can be measured directly in the monte carlo simulations @xmath115 .
this gives us the length of the system to be @xmath116 . in the scft calculations only the combination @xmath117 of flory huggins parameter and
chain length enters .
this sets the temperature scale .
the flory huggins parameter @xmath11 can be calculated from the interaction parameter @xmath118 of the square well potential defining the interaction between monomers in the monte carlo simulations with the following relation@xcite : @xmath119 = 2z_{\text{eff}}\epsilon\ ] ] @xmath120 is the effective coordination number in the bulk , i.e. the average number of intermolecular contacts per monomer .
we hereby speak of contacts , if the distance between the monomers is smaller than @xmath109 .
the average composition @xmath121 is a parameter of the scft , the total number density @xmath122 of monomers is only required if fluctuations are considered .
the single chain diffusion constant @xmath47 , which can be extracted from the monte carlo simulations by measuring the mean square displacements of the chains , gives the time scale @xmath123 monte carlo steps ( mcs ) .
@xmath124 is constant , because @xmath47 and @xmath4 are almost independent of time and composition@xcite .
the length of the chains used in the monte carlo simulations is @xmath114 .
this is somewhat larger than the entanglement length @xmath125@xcite , therefore , we are in a cross over regime between rouse dynamics and reptation@xcite .
this means , when we are comparing dynamic mean field results with monte carlo results we expect to find a reasonable agreement if we regard rouse dynamics .
this identifies all parameters of the scft calculations ( without noise ) .
if we neglect fluctuations , systems with identical @xmath117 , @xmath4 , and composition but different degree of interdigitation @xmath126@xcite and statistical segment length @xmath127 give identical results .
the degree of interdigitation controls the strength of fluctuations and mean field theory is believed to be correct in the limit @xmath128 .
the statistical segment length sets the smallest length scale the gaussian description of polymers is valid .
if we were interested in the structure on smaller length scales we would have to use a different chain model ( e.g. , worm like chain@xcite ) .
in the following sections we regard the early stages of spinodal decomposition after a quench from the one phase region with @xmath129 into the miscibility gap with @xmath130 for a symmetric binary polymer mixture . for much larger
incompatibility the width of the interface becomes of the order of the statistical segment length , and properties on this length scale can not be described by the gaussian chain model . for smaller incompatibilities in the vicinity of the critical point @xmath131 composition fluctuations
are very strong ( i.e. , non gaussian ) and the mean field approximation becomes worse .
we will first show some general results for the dynamical mean field theory .
we will then carry on to compare these results with monte carlo simulations showing what role the onsager coefficient plays .
after having neglected random statistical fluctuations in scft so far we will explain how fluctuations are implemented into the two dynamic mean field theories and discuss their influence on the dynamics .
if a system is quenched from the one into the two phase region the linearized theory of spinodal decomposition@xcite predicts that fluctuations with wave lengths @xmath132 larger than a critical value @xmath133 , i.e wave vectors @xmath134 below a critical value @xmath135 , start growing spontaneously .
this is illustrated in fig.[fig : fig_1 ] showing results obtained through dscft with local coupling in a one dimensional system . before the quench at time @xmath136 we have a homogeneous mixture with random statistical density fluctuations .
after the quench these fluctuations are amplified .
it is apparent that the resulting density profile is well describable with plane waves , so that the set of basis functions we have chosen proves to be not only technically convenient for the calculations but also describes the physical phenomena well . at later times
we see that the mode with the wavelength about a third of the system size is amplified most until at even later stages a saturation inside domains takes place until these domains are separated by sharp interfaces .
later stages of demixing are not appropriately described within our purely diffusive model because hydrodynamic mechanisms and random fluctuations are neglected but play a dominant role .
linearised cahn
cook theory@xcite predicts an exponential behaviour for the density modes @xmath137 as long as the difference @xmath138 between the actual density and the average density is small .
starting with a homogeneous mixture before the quench this requirement is fulfilled during early demixing .
our density coefficients @xmath139 are equivalent to the actual , but discretised , density modes . in fig.[fig : phi_an_log_t ] some of the coefficients @xmath139 are displayed versus time .
these results were obtained for a three dimensional system using dscft with a local onsager coefficient .
the exponential behaviour is obviously well reproduced for early times . for larger values of @xmath134 the exponential behaviour changes earlier .
this is also in agreement with experimental results@xcite .
the faster growth of density modes with smaller wave vectors leads to the creation of small domains on a length scale @xmath140 .
these domains then cause density modes with wave lengths smaller than the extension of the domain to be damped .
fig.[fig : relax_w_s ] shows the corresponding relaxation rate @xmath141 versus @xmath134 .
we see that for wave vectors with a value below @xmath135 an exponential growth of the density modes sets in whereas for @xmath142 initial fluctuations are damped . in the growth region
there is a maximum .
the two dashed lines framing our results are the relaxation rates for temperatures just below the critical temperature ( weak segregation limit wsl ) and for very low temperatures ( strong segregation limit ssl ) as they are given by the cahn hilliard
cook theory : @xmath143\left(1-\frac{q^2}{q_c^2}\right)\ ] ] with @xmath144^{1/2}\\ & = \frac{\sqrt{k}}{r_e}(\chi n/2 - 1)^{1/2 } \end{split}\ ] ] the last line is only valid for symmetric mixtures with @xmath145 when the spinodal point lies at @xmath146 .
@xmath147 is found to be 18 in the wsl@xcite and 12 in the ssl@xcite .
the maximum growth rate is found for @xmath148 . with @xmath130 lying between the ssl and the wsl our result was to be expected .
the influence of rouse dynamics on the relaxation rate is displayed in fig.[fig : epd ] .
here we have presented results for a two dimensional system using dscft with onsager coefficients describing both local kinetics and rouse kinetics for an approximately homogeneous mixture and epd .
the epd and the dscft results using rouse dynamics are virtually identical , but remembering that the epd method is computationally much more favourable , this method should be used when rouse kinetics is considered . local and nonlocal coupling lead to distinct differences in the relaxation rates : in the range @xmath149 nonlocal coupling leads to a reduction of the relaxation rate and a shift of the maximum to smaller values of the wave vector . also in the region @xmath142 the initial density fluctuations are not as strongly damped as for local dynamics .
note that the difference in growth rate between the local dynamics in the wsl and ssl is very similar in magnitude to the difference in growth rate between the full scft calculation using a local or a rouse onsager coefficient .
this demonstrates that the square gradient expression for the free energy is not sufficiently accurate : quantitative deviations from experiments or simulations might be either due to the additional approximations of the square gradient approach or the wave vector dependence of the onsager coefficient . to compare simulations and theory quantitatively and to extract the interplay between single chain dynamics and the kinetics of collective composition fluctuations a quantitative self consistent field calculation
is required .
the monte carlo simulations were performed on a cray t3e using a trivial parallelisation scheme running 64 configurations in parallel to achieve good statistics .
5 400 000 mcs were performed which is equivalent to 45 days of cpu time per processor .
the epd and dscft calculations , the results of which are presented in this section , are for a three dimensional system using @xmath150 basis functions .
the equivalent to 12 700 000 mcs were performed taking approximately 65 days on a cray j90 .
the global structure factor defined by @xmath151 e^{i\mathbf{qr}}\right|^2\right\rangle\ ] ] is an important experimentally measurable quantity for the description of the phase separation process . in fig.[fig : glostr_scf_mc ] the global structure factor is plotted versus wave vector @xmath134 for different times , thin lines denoting scft results , wide lines with the same symbols corresponding monte carlo simulation results .
( because the scft calculations could only be performed for a single starting configuration instead of 64 like in the monte carlo simulations , for the initial time the global structure factor of a homogeneous mixture given by rpa@xcite was used .
global structure factors for later times were extracted through the exponential time dependence of the density modes in the scft calculations . )
part a ) compares monte carlo with local kinetics and part b ) monte carlo with rouse kinetics .
fig.[fig : glostr_scf_mc]a ) clearly shows no quantitative agreement between dscft with local kinetics and the monte carlo simulations : the peak in scft grows much quicker and also the position of the peak is too far right . as has been mentioned before , including rouse dynamics leads to a reduction of the relaxation rate in the growth region and the wave vector with the maximum growth rate is shifted towards a smaller value .
fig.[fig : glostr_scf_mc]b ) satisfies these expectations : the position of the peaks for scft and monte carlo almost coincide and the growth rates of the peak are much closer to each other although scft still overestimates the growth rate .
a more detailed comparison is possible if we plot the corresponding relaxation rates versus wave vector .
as we have seen before the modes of the density calculated with the dynamic mean field theory follow a clear exponential behaviour . to derive the relaxation rates of the global structure factor , we need to plot the modes of the global structure factor on a logarithmic scale versus time .
this is done in fig.[fig : mc_sq_t_log ] for some randomly chosen values of @xmath134 .
the modes with a q value smaller than the critical value @xmath135 derived through dscft show a clear exponential behaviour .
this behaviour changes with time , especially the bigger the value of @xmath134 of the growing mode the sooner the change in the exponential behaviour takes place .
this is qualitatively in agreement with the results we obtained through dynamic mean field theory , see section [ sec : gen_asp ] .
modes with @xmath152 are more or less constant in our figure .
therefore an estimation of @xmath135 from the monte carlo results alone would suffer from some ambiguity : of course , this is not only because the accuracy of the monte carlo results is limited , but is a matter of principle@xcite .
the interplay of fluctuations and nonlinear effects has the consequence that a well defined @xmath135 does not exist . for times earlier than the displayed interval random fluctuations influence the behaviour of the modes so strongly , that an exponential behaviour is not visible .
the relaxation rates resulting from the fits to the points in fig.[fig : mc_sq_t_log ] for the two different time intervals indicated through the solid lines and the corresponding mean field relaxation rates are displayed in fig.[fig : relaxationsrate_local_neu ] . only the exponential behaviour of the earlier of the two marked time intervals may be interpreted as the expected behaviour of early spinodal decomposition .
for later times a gradual change away from the exponential behaviour sets in .
the second fit shows an apparent exponential behaviour , because the time interval is too small to resolve the change .
hence the resulting ` relaxation rate ' for the later time interval may only be treated as an indication for the deviation from the earlier exponential behaviour .
part a ) shows the relaxation rate for dscft with local dynamics in the time interval @xmath153 and the monte carlo results for the two time intervals @xmath154 and @xmath155 ; part b ) shows the same results but regarding rouse dynamics through the use of epd and dscft with rouse dynamics .
the inset figures are the corresponding cahn plots displaying @xmath156 versus @xmath157 .
as is obvious from the behaviour of the global structure factor , local dynamics gives a relaxation rate that is much too large and the wave vector of the maximal growth rate is also too big .
the cahn plot of the mean field relaxation rate shows the linear behaviour , which is expected for local coupling .
this is in strong disagreement with the monte carlo results . for rouse dynamics
we actually find an almost quantitative agreement for earlier times in the region of the positive growth rate .
for later times , however , the relaxation rate decreases in the monte carlo simulations this has also been observed in experimental studies@xcite , while scft still shows the same exponential behaviour as is found for earlier times . in the cahn plot
we see that rouse dynamics leads to a nonlinear run of @xmath156 versus @xmath157 .
this is obviously also the case for the simulation results and is in agreement with earlier simulations@xcite and experimental observations@xcite .
this nonlinear behaviour of the cahn plot is related to the fact that we consider a deep quench , for which @xmath158@xcite . for a shallow quench ,
for which @xmath117 exceeds the critical value @xmath159@xcite only slightly , one has @xmath160@xcite , and then the theory would yield a linear cahn plot , also consistent with corresponding observations@xcite .
the latter case is less interesting , however , because then the polymer mixture is to a large extent equivalent in behaviour to a fluid mixture of small molecules , and there is no longer an effect of internal rouse relaxation modes on the phase separation dynamics in this limit .
if we now compare our results for larger wave vectors we see big discrepancies independent of the chosen dynamics in scft : at @xmath135 the structure factor is independent of time in the early stages of the scft calculations , i.e. structure factors at different times exhibit a common crossing point at @xmath135 , see the inset graphs of fig.[fig : glostr_scf_mc ] .
no such intersection occurs in the simulation data .
mean field theory damps the density modes with @xmath142 while monte carlo simulations lead to a relaxation rate fluctuating around zero .
this behaviour is also seen in the global structure factor , see fig.[fig : glostr_scf_mc ] : the right side of the scft peak decays fast , while the monte carlo peak is much broader with a slower decay .
both the earlier change of the exponential behaviour and the form of the relaxation rate for larger wave vectors are the result of random fluctuations .
as was expected , the influence of fluctuations during the very early stages of spinodal decomposition on the growth of the density modes is rather small , but they are crucial for smaller wave lengths and determine the change in the exponential behaviour , because random fluctuations cause some modes to reach an amplitude where the non linear regime sets in early .
the diffusion equations we have used so far are completely deterministic , but obviously in all dynamic processes random statistical fluctuations are present . to regard these fluctuations we have to add a random force @xmath102 to our diffusion equations ( [ eq : diff_eq ] ) and ( [ eq : diff_eq_epd ] ) that is linked to the onsager coefficient through the fluctuation
dissipation theorem : @xmath161 or fourier transformed : @xmath162 in our calculations the diffusion equation is integrated through discrete time intervals @xmath163 that are determined through the runge
kutta scheme . in dscft , for example
, we use the langevin equation : @xmath164 with @xmath165 expressing random fluctuations that obey the fluctuation dissipation theorem@xcite : @xmath166 @xmath167 is a random number with the properties @xmath168 , @xmath169 . the analogue @xmath165 is used to perform epd calculations with fluctuations .
the difficulty in this case is to interpret the resulting fluctuations of the external field @xmath170 in terms of the density @xmath171 .
as we have seen in section [ sec : epd ] the field and the density fluctuations are linked to each other through equations and . to ensure that the way the fluctuations are included is correct and the validity of these eqs . and is given , we consider a homogeneous , @xmath172 , one dimensional system with length @xmath173 using both dscft and epd . from the dscft calculations we derive @xmath174 for each wave vector while the epd method leads to @xmath175 . in both cases we averaged over 10000 snapshots of the density or the field made during their time evolution . in fig.[fig
: phi - w_rpa ] @xmath174 and @xmath175 are plotted versus @xmath134 .
the solid line is the expected result from rpa . for larger values of @xmath134
we find good agreement , but the amount of used configurations was not enough to find reliable results for smaller values of @xmath134 because the correlation times are much longer due to the diffusive dynamics .
if we plot @xmath174 versus @xmath175 , as is done in fig.[fig : epd_dscft_fluct ] , we find good agreement of our data with the expected linear behaviour . after having proven that the above treatment of random fluctuations leads to the expected behaviour we focus on the influence of fluctuations on spinodal decomposition .
monte carlo results show that the exponential behaviour changes earlier than in scft calculations without fluctuations resulting in a reduced relaxation rate .
the relaxation rates for different time intervals obtained through the monte carlo simulations and the epd method are presented in fig.[fig : relaxationsrate_dw_dt_fluct ] .
the epd results were obtained by averaging the time evolution of the fields of 64 two dimensional configurations . for earlier time
intervals the relaxation rate is quantitatively very similar to the rate without fluctuations in the region below @xmath135 . for later times
however when calculations without fluctuations still show the same exponential behaviour the relaxation rate is reduced as is also seen in the monte carlo simulations . in the range above @xmath135 the modes of the fields or densities
do not follow an exponential behaviour but fluctuate around zero .
consequently the relaxation rate is not well defined in this region leading to strong fluctuations of the relaxation rate in fig.[fig : relaxationsrate_dw_dt_fluct ] .
for very early times when the density modes are still of the order of the fluctuations of the homogeneous system it is also not possible to see an exponential behaviour because the density changes caused through random fluctuations conceal the growth of the modes during very early spinodal decomposition .
in this study we analysed the influence of single chain dynamics on the collective diffusion during early stages of spinodal decomposition in a symmetric binary polymer blend .
we used the scft for polymer mixtures to explore two versions of dynamical mean field theory .
the single chain dynamics enters these descriptions through an onsager coefficient .
the first method we call dscft propagates the densities in time and gives us the possibility to model both local dynamics and approximately the nonlocal dynamics we expect for the rouse model .
in dscft the correct treatment of rouse dynamics would involve the calculation of the pair
correlation function at every time step , which is a computationally rather expensive task . on the other hand during early stages of demixing the mixture is only weakly inhomogeneous so that the use of the pair
correlation function of a homogeneous mixture , which is analytically known , serves as a sufficient approximation . in the second method
the instantaneous configuration is not described by the densities but by the effective external field ( epd ) .
we find a langevin equation for the external field which using ` local ' kinetics in the fields is found to describe a polymer mixture with rouse dynamics .
apart from ` automatically ' including rouse dynamics this epd method has the big advantage of being up to an order of magnitude computationally faster than the dscft method . first numerical calculations with these methods for a quench from the one phase , @xmath129 , to the two phase , @xmath130 , region neglecting random fluctuations show a clear exponential behaviour of the density modes , as was expected for a mean field description
. the relaxation rate of the density mode is strongly influenced by the choice of the onsager coefficient : in the growth region rouse dynamics reduces the relaxation rate compared to local dynamics and the position of the maximum growth rate is shifted to smaller values of @xmath134 . for @xmath142 however rouse dynamics causes the modes to be damped less quickly .
we also find good agreement for early stages between the dscft method using the pair
correlation function for the homogeneous system and the epd method . to quantitatively test these mean field predictions
we compare them with results obtained through monte carlo simulations employing the bond fluctuation model .
the chains used in these simulations are expected to show rouse behaviour because of the chosen chain length @xmath114 .
the comparison is possible without any adjustable parameter .
we compare the global structure factor , which is the experimentally accessible quantity , and the relaxation rates .
local dynamics in dscft overestimates the growth of the global structure factor by far , but the agreement is better for rouse dynamics especially for earlier times . neglecting random fluctuations in our mean field calculations proves to be justified for wave vectors with positive relaxation rates and earlier times , but should be included to investigate later times .
fluctuations can easily be included in dscft and epd .
the difficulty in the epd method is the fact that the field fluctuations have to be interpreted in terms of the physical density fluctuations .
we can find a relation between the field and the density fluctuations .
epd calculations with fluctuations lead to an earlier change in the exponential behaviour of the density modes as was also the case for the monte carlo simulations .
the missing dampening of the modes with @xmath142 as is found in the simulations is also reproduced .
we have seen that the single chain dynamics has a pronounced influence on the collective dynamics of a polymer mixture . comparing quantitatively monte carlo simulations and dynamical mean field theory we have validated the mean field calculations .
note , however , that we have considered a deep quench far below the critical point ; for shallow quenches close to the critical point mean field theories are not expected to be accurate .
the later stages of spinodal decomposition are not accessible with either method . during later times
hydrodynamical interactions become important .
lattice model monte carlo simulations lack a hydrodynamical mechanism . in dynamical mean field calculations hydrodynamic coupling
can be included@xcite , and , hence , they can be extended beyond the validity of the lattice model .
we have benefitted from discussion with d. dchs , f. schmid , and v. ganesan .
financial support was provided by the graduierten kolleg `` supramolecular systems '' of the university of mainz and the dfg under grant bi314/17 in the priority program `` wetting and structure formation at interfaces '' .
generous access to computers at the nic jlich and the hlr stuttgart are gratefully acknowledged .
as mentioned before it is possible to show that the epd method using local coupling is a good approximation for reproducing rouse dynamics of the physical densities .
the derivation of the epd method we present in this section was introduced by n.m .
_ and can be found in reference@xcite . following the method we used to include dynamics in the scft to achieve the dscft method
we employ a saddle point approximation in the external fields which leads to a bijective relation between the external fields @xmath30 , @xmath51 and the densities @xmath52 , @xmath53 .
this means we can choose with which of the two variable sets we would like to calculate . on the other hand the pair
correlation function which is part of the rouse onsager coefficient can be expressed as the functional derivative of the density with respect to the external potential , see also eq.([eq : pair - corr ] ) .
@xmath176 to calculate in @xmath177 space we have to transform the time derivative of the densities according to the chain rule : @xmath178 combining this equation with the diffusion equation for rouse dynamics ( [ eq : rouse_dyn ] ) this leads us to @xmath179 using the approximation @xmath180 one easily arrives at @xmath181 leading to an equation of motion for the external fields : @xmath182 approximation ( [ eq : approx_epd ] ) is obviously exactly valid for a homogeneous mixture , because the pair correlation function only depends on the distance @xmath183 between two points . in the inhomogeneous case , if @xmath33 and @xmath184 are in different phases and neither of them in the interface , @xmath185 and @xmath186 are of different sign making this approximation justifiable even if the actual values differ . because of the incompressibility constraint @xmath187 and @xmath188 are not independent of each other .
there is only one independent chemical potential @xmath189 so the equation of motion for the external fields which is to be used in our case has the form : @xmath190 if we compare this equation with equation ( [ eq : diff_eq_epd ] ) we see that densities evolving in time according to rouse dynamics are well described through the epd method if a local kinetic coefficient as given in eq.([eq : onsager_epd ] ) is used .
the single chain partition function @xmath191 is defined through ( compare with eq.([eq : si_ch_part ] ) ) : @xmath192{\mathcal{p}}_1[{\mathbf{r } } ] \exp\left[-\frac{\rho}{n}\int_v d^3{\mathbf{r}}\;w({\mathbf{r } } ) \hat{\phi}_1({\mathbf{r}})\right ] \label{eq : si_ch_part_2}\ ] ] where @xmath193 denotes the single chain density .
we now expect the system to be only weakly inhomogeneous , i.e. , meaning the density and the external field only differ a little from the average value : @xmath194 ; @xmath195 . the density and the external field are presented as a fourier expansion : @xmath196 this fourier expansion is now inserted in eq.([eq : si_ch_part_2 ] ) : @xmath197{\mathcal{p}}_1[{\mathbf{r } } ] \exp\left[-\bar{w}-\frac{\rho v}{n}\sum_{{\mathbf{q}}\neq 0 } w_{{\mathbf{q}}}\hat\phi_{-\mathbf{q}}\right]\\ & = \exp\left[-\bar{w}_a\right]\int{\mathcal{d}}[{\mathbf{r}}_1 ] { \mathcal{p}}_1[{\mathbf{r}}_1 ] \left\{1-\frac{\rho v}{n}\sum_{{\mathbf{q}}\neq 0}w_{{\mathbf{q } } } \hat\phi_{-\mathbf{q}}+\frac{\rho^2v^2}{2n^2}\sum_{{\mathbf{q , q'}}\neq 0 } w_{{\mathbf{q}}}w_{{\mathbf{q'}}}\hat\phi_{-\mathbf{q } } \hat\phi_{-\mathbf{q'}}+\cdots\right\}\\ & = \exp\left[-\bar{w}_a\right]q_{0}\left\langle 1- \frac{\rho v}{n}\sum_{{\mathbf{q}}\neq 0}w_{{\mathbf{q } } } \hat\phi_{-\mathbf{q}}+\frac{\rho^2v^2}{2n^2}\sum_{{\mathbf{q , q'}}\neq 0 } w_{{\mathbf{q}}}w_{{\mathbf{q'}}}\hat\phi_{-\mathbf{q } } \hat\phi_{-\mathbf{q'}}+\cdots\right\rangle_0 \end{split}\ ] ] @xmath198 denotes the partition function of a single chain without an external field .
the average in the last line is to be taken over all chain configurations that are possible when there is no external field present .
because the average deviation of the density @xmath199 from the average value is zero if there is no external field , @xmath200 is valid .
the average @xmath201 is given through the single chain structure factor @xmath202 of a gaussian chain : @xmath203 .
neglecting higher terms , we obtain the rpa - result for the single chain partition function : @xmath204q_{0 } \exp\left[\frac{1}{2n}\sum_{{\mathbf{q}}\neq 0}s_0({\mathbf{q } } ) \label{eq : q_rpa}\ ] ] obviously a corresponding expression is valid for a b polymer .
these rpa single chain partition functions are plugged into eq.([eq : free_en_uw ] ) leading to : @xmath205}{k_bt}= -\frac{\bar{\phi_a}\rho v}{n}\ln\frac{q_{a,0}}{n_a } -\frac{\bar{\phi_b}\rho v}{n}\ln\frac{q_{b,0}}{n_b } + \frac{\rho v \chi}{4 } + \frac{(\bar{\phi_a}-\bar\phi_b)\rho v \bar w}{2n } + \frac{\rho v}{4n\chi n}\bar{w}^2+\\ + \frac{\rho v}{4n\chi n}\sum_{{\mathbf{q}}\neq 0}|w_{{\mathbf{q}}}|^2 -\frac{\bar{\phi_a}\rho v}{2n^2 } \sum_{{\mathbf{q}}\neq 0}s_0({\mathbf{q}})\left|\frac{(w_{{\mathbf{q}}}+u_{{\mathbf{q}}})}{2}\right|^2 -\frac{\bar{\phi_b}\rho v}{2n^2 } \sum_{{\mathbf{q}}\neq 0}s_0({\mathbf{q}})\left|\frac{(u_{{\mathbf{q}}}-w_{{\mathbf{q}}})}{2}\right|^2\end{gathered}\ ] ] regarding only the wave vector dependent parts of the free energy we find : @xmath206}{k_bt}=&-\frac{\rho v}{2n^2}\sum_{{\mathbf{q}}\neq 0 } \left\ { \left(\frac{s_0({\mathbf{q}})}{4}-\frac{1}{2\chi}\right)|w_{{\mathbf{q}}}|^2 + \left(\frac{s_0({\mathbf{q}})}{4}\right)|u_{{\mathbf{q}}}|^2\right\}+\\ & \qquad-\frac{\rho v}{2n^2}\sum_{{\mathbf{q}}\neq 0}\left\ { \frac{\bar{\phi}_a-\bar{\phi}_b}{2}s_0({\bf q})(u_{{\mathbf{q}}}w_{-{\mathbf{q } } } ) \right\}+g_{\text{hom}}=\\ = & -\frac{\rho v}{2n^2}\sum_{{\mathbf{q}}\neq 0 } \left\{\left(\frac{s_0({\mathbf{q}})}{4}-\frac{1}{2\chi}-\frac{1}{4}(\bar{\phi}_a-\bar{\phi}_b)^2s_0({\mathbf{q } } ) \right)|w_{\mathbf{q}}|^2\right\}+\\ & \qquad-\frac{\rho v}{2n^2}\sum_{{\mathbf{q}}\neq 0 } \left\{\frac{s_0({\bf q})}{4}\left|u_{{\mathbf{q}}}+(\bar{\phi}_a-\bar{\phi}_b)w_{{\mathbf{q}}}\right|^2\right\}+g_{\text{hom } } \end{split}\ ] ] we use this free energy to evaluate the partition function of eq.([eq : part_uw ] ) . following the procedure we used before ,
we employ a saddle point approximation with respect to the field @xmath69 : @xmath207}{\delta u}|_{u^*}=0\qquad:\qquad u^*_{{\mathbf{q}}}= - ( \bar{\phi}_a-\bar{\phi}_b)w_{{\mathbf{q}}}\ ] ] since the free energy is quadratic in the rpa approximation the saddle point integration is equivalent to the functional integration over @xmath69 .
this again leads us to an expression for the free energy @xmath208 $ ] only depending on the external field variable @xmath70 : @xmath209}{k_bt}=\frac{\rho v}{2n^2}\sum_{{\mathbf{q}}\neq 0 } \left(\bar{\phi}_a\bar{\phi}_b s_0({\bf q } ) -\frac{1}{2\chi } \right)|w_{{\mathbf{q}}}|^2+g_{\text{hom}}\ ] ] now the average @xmath210 can be calculated : @xmath211 using our result ( [ eq : phi^2_av_four ] ) @xmath212 we express the density fluctuations in terms of the field fluctuations and recover the well known rpa expression@xcite : @xmath213^{-1 } \equiv 4 s_{\rm rpa}({\bf q})\ ] ] using the rpa single chain partition function ( [ eq : q_rpa ] ) , we calculate @xmath214 the last equation is equivalent to eq.[eq : pair - corr_epd ] but due to the particle conservation the @xmath215 contribution has to be taken out of the sum .
@xmath216 is just the average overall density @xmath217 that can not change if the external field is altered . with these expressions
we obtain for the `` literal '' fluctuations in the epd method according to eq.([eq : phi^2_av_ok ] ) : @xmath218 generally , the deviation for the rpa result is of similar magnitude as the rpa structure factor itself . for a symmetric quench @xmath219 , however , we accidentally recover the rpa result .
this example also illustrates that one can obtain the average of the composition by sampling the average of @xmath220 , i.e. , the densities of single chains in the field configuration @xmath70 , but one should not use this to calculate fluctuations . for introductory reviews on spinodal decomposition , see k. binder , in _ phase transformations in materials .
_ p. 405 , p. hansen ed .
, vch weinheim ( 1991 ) and t. hashimoto , in _ structure and properties of polymers .
kramer ed .
, vch weinheim ( 1993 ) .
this is common practice in self consistent field calculations .
it is understood that the incompressibility constraint is enforced only after coarse graining the microscopic densities over a small but finite volume .
on smaller length scales the specific spatial dependence of the repulsive interactions matters and leads , e.g. , to packing effects .
this is beyond our treatment . | we investigate how the dynamics of a single chain influences the kinetics of early stage phase separation in a symmetric binary polymer mixture .
we consider quenches from the disordered phase into the region of spinodal instability . on a mean field level we approach this problem with two methods : a dynamical extension of the self consistent field theory for gaussian chains , with the density variables evolving in time , and the method of the external potential dynamics where the effective external fields are propagated in time .
different wave vector dependencies of the kinetic coefficient are taken into account .
these early stages of spinodal decomposition are also studied through monte carlo simulations employing the bond fluctuation model that maps the chains in our case with 64 effective segments on a coarse grained lattice .
the results obtained through self consistent field calculations and monte carlo simulations can be compared because the time , length , and temperature scales are mapped onto each other through the diffusion constant , the chain extension , and the energy of mixing .
the quantitative comparison of the relaxation rate of the global structure factor shows that a kinetic coefficient according to the rouse model gives a much better agreement than a local , i.e. wave vector independent , kinetic factor .
including fluctuations in the self consistent field calculations leads to a shorter time span of spinodal behaviour and a reduction of the relaxation rate for smaller wave vectors and prevents the relaxation rate from becoming negative for larger values of the wave vector .
this is also in agreement with the simulation results . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.