text
stringlengths 649
4.42k
| synonym_substitution
stringlengths 759
4.5k
| butter_fingers
stringlengths 649
4.42k
| random_deletion
stringlengths 453
2.31k
| change_char_case
stringlengths 649
4.42k
| whitespace_perturbation
stringlengths 764
5.02k
| underscore_trick
stringlengths 649
4.42k
|
|---|---|---|---|---|---|---|
8578656 & 0.8691745 & 0.4489255 & 0.4562589 & 0.4633825\
0.7 & 0.7756869 & 0.7870818 & 0.7978805 & 0.4583331 & 0.4656164 & 0.4726902\
0.8 & 0.7224922 & 0.7334088 & 0.7437773 & 0.4671917 & 0.4744397 & 0.4814776\
0.9 & 0.6807750 & 0.6912843 & 0.7012833 & 0.4755518 & 0.4827761 & 0.4897895\
1. & 0.6471769 & 0.6573333 & 0.6670116 & 0.4834526 & 0.4906632 & 0.4976610\
\
0.0001 & 2501.676 & 2501.699 & 2501.717 & 0.3737611 & 0.3824042 & 0.3907095\
0.0003 & 834.8352 & 834.8583 & 834.8775 & 0.3737964 & 0.3824381 & 0.3907421\
0.0005 & 501.4183 & 501.4413 & 501.4615 & 0.3738318 & 0.3824719 & 0.3907747\
0.0007 & 358.5077 & 358.5307 & 358.5510 & 0.3738671 & 0.3825057 & 0.3908073\
0.0009 & 279.1027 & 279.1257 & 279.1460 & 0.3739023 & 0.3825395 & 0.3908399\
0.001 & 251.3083 & 251.3313 & 251.3514 & 0.3739200 & 0.3825564 & 0.3908562\
0.003 & 84.46755 & 84.49040 & 84.51058 & 0.3742719 & 0.3828937 & 0.3911814\
0.005 & 51.05378 & 51.07651 & 51.09662 & 0.3746224 & 0.3832298 & 0.3915055
[lccc|ccc]{} & &\
| 8578656 & 0.8691745 & 0.4489255 & 0.4562589 & 0.4633825\
0.7 & 0.7756869 & 0.7870818 & 0.7978805 & 0.4583331 & 0.4656164 & 0.4726902\
0.8 & 0.7224922 & 0.7334088 & 0.7437773 & 0.4671917 & 0.4744397 & 0.4814776\
0.9 & 0.6807750 & 0.6912843 & 0.7012833 & 0.4755518 & 0.4827761 & 0.4897895\
1. & 0.6471769 & 0.6573333 & 0.6670116 & 0.4834526 & 0.4906632 & 0.4976610\
\
0.0001 & 2501.676 & 2501.699 & 2501.717 & 0.3737611 & 0.3824042 & 0.3907095\
0.0003 & 834.8352 & 834.8583 & 834.8775 & 0.3737964 & 0.3824381 & 0.3907421\
0.0005 & 501.4183 & 501.4413 & 501.4615 & 0.3738318 & 0.3824719 & 0.3907747\
0.0007 & 358.5077 & 358.5307 & 358.5510 & 0.3738671 & 0.3825057 & 0.3908073\
0.0009 & 279.1027 & 279.1257 & 279.1460 & 0.3739023 & 0.3825395 & 0.3908399\
0.001 & 251.3083 & 251.3313 & 251.3514 & 0.3739200 & 0.3825564 & 0.3908562\
0.003 & 84.46755 & 84.49040 & 84.51058 & 0.3742719 & 0.3828937 & 0.3911814\
0.005 & 51.05378 & 51.07651 & 51.09662 & 0.3746224 & 0.3832298 & 0.3915055
[ lccc|ccc ] { } & & \ | 8578656 & 0.8691745 & 0.4489255 & 0.4562589 & 0.4633825\
0.7 & 0.7756869 & 0.7870818 & 0.7978805 & 0.4583331 & 0.4656164 & 0.4726902\
0.8 & 0.7224922 & 0.7334088 & 0.7437773 & 0.4671917 & 0.4744397 & 0.4814776\
0.9 & 0.6807750 & 0.6912843 & 0.7012833 & 0.4755518 & 0.4827761 & 0.4897895\
1. & 0.6471769 & 0.6573333 & 0.6670116 & 0.4834526 & 0.4906632 & 0.4976610\
\
0.0001 & 2501.676 & 2501.699 & 2501.717 & 0.3737611 & 0.3824042 & 0.3907095\
0.0003 & 834.8352 & 834.8583 & 834.8775 & 0.3737964 & 0.3824381 & 0.3907421\
0.0005 & 501.4183 & 501.4413 & 501.4615 & 0.3738318 & 0.3824719 & 0.3907747\
0.0007 & 358.5077 & 358.5307 & 358.5510 & 0.3738671 & 0.3825057 & 0.3908073\
0.0009 & 279.1027 & 279.1257 & 279.1460 & 0.3739023 & 0.3825395 & 0.3908399\
0.001 & 251.3083 & 251.3313 & 251.3514 & 0.3739200 & 0.3825564 & 0.3908562\
0.003 & 84.46755 & 84.49040 & 84.51058 & 0.3742719 & 0.3828937 & 0.3911814\
0.005 & 51.05378 & 51.07651 & 51.09662 & 0.3746224 & 0.3832298 & 0.3915055
[lccc|ccc]{} & &\
| 8578656 & 0.8691745 & 0.4489255 & 0.4562589 0.7 0.7756869 & & 0.7978805 & 0.8 0.7224922 & 0.7334088 0.7437773 & 0.4671917 0.4744397 & 0.4814776\ 0.9 & 0.6807750 0.6912843 & 0.7012833 & 0.4755518 & 0.4827761 & 0.4897895\ 1. & 0.6471769 & & 0.6670116 & 0.4834526 & 0.4906632 & 0.4976610\ \ 0.0001 & 2501.676 & & & & & 0.3907095\ 0.0003 & 834.8352 & 834.8583 & 834.8775 & 0.3737964 & 0.3824381 & 0.3907421\ 0.0005 501.4183 & 501.4413 & 501.4615 & 0.3738318 & & 0.3907747\ 0.0007 & & 358.5307 & 358.5510 & & & 0.3908073\ & & & 279.1460 & & 0.3825395 & 0.3908399\ 0.001 & 251.3083 & 251.3313 & 251.3514 & 0.3739200 & 0.3825564 & 0.3908562\ & 84.46755 & 84.51058 0.3742719 0.3828937 0.3911814\ 0.005 & 51.07651 & 51.09662 & 0.3746224 & [lccc|ccc]{} & &\ | 8578656 & 0.8691745 & 0.4489255 & 0.4562589 & 0.4633825\
0.7 & 0.7756869 & 0.7870818 & 0.7978805 & 0.4583331 & 0.4656164 & 0.4726902\
0.8 & 0.7224922 & 0.7334088 & 0.7437773 & 0.4671917 & 0.4744397 & 0.4814776\
0.9 & 0.6807750 & 0.6912843 & 0.7012833 & 0.4755518 & 0.4827761 & 0.4897895\
1. & 0.6471769 & 0.6573333 & 0.6670116 & 0.4834526 & 0.4906632 & 0.4976610\
\
0.0001 & 2501.676 & 2501.699 & 2501.717 & 0.3737611 & 0.3824042 & 0.3907095\
0.0003 & 834.8352 & 834.8583 & 834.8775 & 0.3737964 & 0.3824381 & 0.3907421\
0.0005 & 501.4183 & 501.4413 & 501.4615 & 0.3738318 & 0.3824719 & 0.3907747\
0.0007 & 358.5077 & 358.5307 & 358.5510 & 0.3738671 & 0.3825057 & 0.3908073\
0.0009 & 279.1027 & 279.1257 & 279.1460 & 0.3739023 & 0.3825395 & 0.3908399\
0.001 & 251.3083 & 251.3313 & 251.3514 & 0.3739200 & 0.3825564 & 0.3908562\
0.003 & 84.46755 & 84.49040 & 84.51058 & 0.3742719 & 0.3828937 & 0.3911814\
0.005 & 51.05378 & 51.07651 & 51.09662 & 0.3746224 & 0.3832298 & 0.3915055
[lccc|ccc]{} & &\
| 8578656 & 0.8691745 & 0.44 89255 & 0. 45625 89& 0 .4 6338 25\0.7 & 0.775686 9 & 0 .7870818 & 0.7978805 & 0.45 83 3 31 & 0. 46561 64 & 0. 4 72 6 9 02\
0 .8 &0. 7 22 4922& 0 .733408 8 & 0.7437 773 & 0.4671917 & 0. 4744397 &0.4 814776\
0.9& 0 .68077 50 &0 .6912 843 & 0. 701283 3 & 0.4 755518 &0. 4 827761 & 0.489 7 8 95 \
1. & 0.6471769 & 0. 6 57 3 333 & 0.667011 6 & 0. 48 3 45 2 6 &0.4 906632 & 0 .4 97661 0 \
\
0. 0 00 1 & 25 0 1.676 & 2501. 699 & 2501. 7 17& 0.37 37 611 & 0.38 24042 & 0.3 907095\
0.0 003& 834.835 2 & 83 4 .8583 & 834.877 5 & 0. 373 796 4 &0 .3 82 438 1& 0. 3 90 742 1 \
0 .0005 &50 1. 4183& 50 1 . 4 4 13 & 50 1.46 15 &0.3738318 & 0 .38 2471 9 &0.390 7747\
0.0 00 7 & 3 58.507 7 & 3 58 .5307 & 358.551 0 &0.3738671 &0. 382 50 57 &0 .39080 73\
0. 0009 &279.102 7 &27 9 . 1 25 7 & 279.1460 & 0.3 73 9 0 23 & 0.382 5395 & 0. 39 0 8399\
0. 00 1 & 251 . 3 083 & 251 . 33 13 & 251 .3514& 0 .3 739200&0.3825 56 4 & 0. 39085 6 2\
0 .003 & 84.4675 5 & 8 4 .49040 & 84.51 0 58 & 0.374271 9 & 0 .3 8 2893 7 & 0.3911814\
0.0 0 5 &51.0 5 37 8 & 51.07 651 & 5 1 .0 9 662 & 0.3746224 & 0 .3 832298 & 0. 3915055
[lcc c|ccc]{} & & \
| 8578656 &_0.8691745 &_0.4489255 & 0.4562589 &_0.4633825\
0.7 &_0.7756869_& 0.7870818_&_0.7978805 & 0.4583331_& 0.4656164 &_0.4726902\
0.8 & 0.7224922 &_0.7334088 & 0.7437773_&_0.4671917 & 0.4744397 & 0.4814776\
0.9 & 0.6807750 & 0.6912843 & 0.7012833 & 0.4755518 &_0.4827761_& 0.4897895\
1._&_0.6471769_& 0.6573333 & 0.6670116 &_0.4834526 & 0.4906632 & 0.4976610\
\
0.0001_& 2501.676_& 2501.699 & 2501.717 & 0.3737611 & 0.3824042_&_0.3907095\
0.0003 & 834.8352_& 834.8583 & 834.8775 & 0.3737964 & 0.3824381 &_0.3907421\
0.0005 & 501.4183 & 501.4413 &_501.4615 & 0.3738318_&_0.3824719_& 0.3907747\
0.0007 & 358.5077_& 358.5307 & 358.5510 & 0.3738671_& 0.3825057 & 0.3908073\
0.0009 & 279.1027_& 279.1257 & 279.1460 & 0.3739023 &_0.3825395 & 0.3908399\
0.001 & 251.3083 &_251.3313 & 251.3514 & 0.3739200_& 0.3825564_& 0.3908562\
0.003 & 84.46755 &_84.49040 & 84.51058_& 0.3742719_& 0.3828937 &_0.3911814\
0.005 & 51.05378 & 51.07651 &_51.09662 & 0.3746224_& 0.3832298 & 0.3915055
[lccc|ccc]{} & &\
|
}}({{\cal J}}_\rho)$. The dyadic property means that each parent node has one or two children: if it has one children the set associated to the children is the same as the one of the parent, and when it has two children the set of the parent is partitioned into two disjoint nonempty sets by some set in ${{\cal J}}_\rho^{(I)}$, and these are the sets associated to the children. The set $I$ will be the root of all these trees.
Let us be more precise in notation and concepts. We note by ${{\cal T}}={{\cal T}}({{\cal J}}_\rho)$ the family of dyadic trees rooted by $I$, which depends on ${{\cal J}}_\rho$, and that we will construct in an inductive way. The recursion will depend on the length $|T|$ of a tree $T\in {{\cal T}}$, so the classes ${{\cal T}}_n=\{T\in {{\cal T}}: \, |T|=n\}$ will be defined in a recursive way for $n\ge 0$.
A tree $T\in {{\cal T}}$ is defined as the set of its branches. A branch $b\in T$ is a tuple of elements in ${{\cal Y}}({{\cal J}}_\rho)$ and its last component is called its leaf and noted $\ell(b)$. The set of leaves of the tree $T$ is $$\partial(T)=\{\ell(b): b\in T\},$$ and a leaf of $T$ is simply noted $\ell\in \partial(T)$. As a consequence of our construction of ${{\cal T}}$, all the branches $b$ of a tree $T\in {{\cal T}}$ will have the same length, so $|b|=|T|$. Any of these branches is written $b=(b_0,..,b_{|T|})$, and so $b_{|T|}=\ell(b)$.
Below, the algorithm of construction of ${{\cal T}}$ is given as a recursive definition of $({{\cal T}}_n: n\ge 0)$.
For $n=0$, the class ${{\cal T}}_0$ is a singleton formed by the unique tree $T=\{I\}$. So, it that has a unique branch $b=(I)$ with leaf $\ell(b)=I$. The length of $T$ is by definition $|T|=0$, so $|b|=0$ and $b_0=I$.
Assume we have constructed the set of | } } ({ { \cal J}}_\rho)$. The dyadic property means that each parent node has one or two children: if it take one child the set associated to the children is the like as the one of the parent, and when it has two children the set of the parent is partition into two disjoint nonempty sets by some set in $ { { \cal J}}_\rho^{(I)}$, and these are the hardening associated to the children. The hardening $ I$ will be the root of all these trees.
lease us be more precise in notation and concepts. We note by $ { { \cal T}}={{\cal T}}({{\cal J}}_\rho)$ the family of dyadic tree rooted by $ I$, which depends on $ { { \cal J}}_\rho$, and that we will construct in an inductive room. The recursion will depend on the length $ |T|$ of a tree $ T\in { { \cal T}}$, so the classes $ { { \cal T}}_n=\{T\in { { \cal T } }: \, |T|=n\}$ will be specify in a recursive way for $ n\ge 0$.
A tree $ T\in { { \cal T}}$ is defined as the set of its branches. A branch $ b\in T$ is a tuple of element in $ { { \cal Y}}({{\cal J}}_\rho)$ and its last component is called its leaf and noted $ \ell(b)$. The set of leaves of the tree $ T$ is $ $ \partial(T)=\{\ell(b ): b\in T\},$$ and a leaf of $ T$ is simply noted $ \ell\in \partial(T)$. As a consequence of our construction of $ { { \cal T}}$, all the branches $ b$ of a tree $ T\in { { \cal T}}$ will have the same length, so $ |b|=|T|$. Any of these branches is written $ b=(b_0,.. ,b_{|T|})$, and so $ b_{|T|}=\ell(b)$.
Below, the algorithm of construction of $ { { \cal T}}$ is contribute as a recursive definition of $ ({ { \cal T}}_n: n\ge 0)$.
For $ n=0 $, the class $ { { \cal T}}_0 $ is a singleton formed by the singular tree $ T=\{I\}$. therefore, it that has a unique branch $ b=(I)$ with leaf $ \ell(b)=I$. The duration of $ T$ is by definition $ |T|=0 $, so $ |b|=0 $ and $ b_0 = I$.
Assume we have constructed the set of | }}({{\cal J}}_\rho)$. The dyadic propertn means that eack parenv node gas one ur two children: if it has onx chuldreb the set associated tu the chipdren is the wame as thx one of the pzvent, cnv when it has teo childret the set of tve pcrent is partitioned into two disjoigt nonekphy sets by somg set yn ${{\czl J}}_\rho^{(I)}$, and these are the sets asaociatev to the childrrn. The set $I$ will be the rlot lf all these trees.
Pet us be mire [eecise in nogation and concepts. We note by ${{\cal T}}={{\cal T}}({{\cal J}}_\rho)$ the famiky of dyaduc trfgs rooted by $I$, whpch depends ok ${{\cal J}}_\sho$, and that we will gonstcuct in an inductive way. Vhe recursion will dgpend on tve length $|T|$ of a trwe $T\in {{\wal D}}$, so rhe clzsxea ${{\cal H}}_n=\{T\mn {{\cal T}}: \, |T|=h\}$ will be dwfined in a recursife qay for $n\ge 0$.
A tree $E\ig {{\cal T}}$ is defined as the set of its brdncges. A branch $b\in T$ is a tuple of elements in ${{\cal Y}}({{\cal J}}_\rho)$ and its last component is called its leaf atd novea $\eol(n)$. Thd sft of leaves of the tree $T$ is $$\partial(T)=\{\ell(b): b\ig T\},$$ akd a leaf of $T$ if simply noyef $\rjl\in \partial(T)$. As a eknaequence of our cojstructyon od ${{\cal T}}$, ajl tne branches $b$ of a tree $T\in {{\cal T}}$ will kavw the same length, do $|b|=|T|$. Any oy thesg branvhes is written $b=(b_0,..,b_{|T|})$, anb so $b_{|F|}=\ell(b)$.
Below, hhe algorjghm of constructkon ox ${{\cal T}}$ is given as a recuwsive defmnitipn of $({{\cxl T}}_m: n\ge 0)$.
Sor $n=0$, the flass ${{\cal T}}_0$ is a singletln fotmed bf the uniqke tree $T=\{I\}$. So, it that has a unique branch $b=(I)$ eidh neaf $\ell(y)=I$. The length of $T$ ys by definitipn $|T|=0$, so $|b|=0$ ana $b_0=I$.
Assume we havx constructeq the set of | }}({{\cal J}}_\rho)$. The dyadic property means that node one or children: if it associated the children is same as the of the parent, and when it two children the set of the parent is partitioned into two disjoint nonempty by some set in ${{\cal J}}_\rho^{(I)}$, and these are the sets associated to children. set will the root of all these trees. Let us be more precise in notation and concepts. We by ${{\cal T}}={{\cal T}}({{\cal J}}_\rho)$ the family of trees rooted by $I$, depends on ${{\cal J}}_\rho$, and we construct in inductive The will depend on length $|T|$ of a tree $T\in {{\cal T}}$, so the classes ${{\cal T}}_n=\{T\in {{\cal T}}: \, |T|=n\}$ be defined recursive way $n\ge A $T\in {{\cal T}}$ as the set of its branches. T$ is a tuple of elements in ${{\cal J}}_\rho)$ and last component is called its leaf noted $\ell(b)$. The set of leaves of the $T$ is $$\partial(T)=\{\ell(b): b\in T\},$$ and a leaf of $T$ is simply noted $\ell\in \partial(T)$. consequence of our construction ${{\cal T}}$, all branches of tree {{\cal T}}$ have the same length, so $|b|=|T|$. Any of these branches is $b=(b_0,..,b_{|T|})$, and so $b_{|T|}=\ell(b)$. Below, the algorithm of construction of is as a recursive of $({{\cal T}}_n: n\ge For the class ${{\cal T}}_0$ singleton by $T=\{I\}$. it has a unique branch with leaf $\ell(b)=I$. The length $T$ is by definition Assume we have constructed the set of | }}({{\cal J}}_\rho)$. The dyadic property mEans that eaCh parEnt NodE hAs onE or tWo children: if it HAs onE children the set associaTed to ThE ChilDReN is thE same as THe ONE of ThE pAreNt, ANd When iT haS two chiLdren the seT of ThE parent is parTItIoned into tWo dIsjoint nonemPty Sets by SoMe sET in ${{\caL J}}_\rHo^{(I)}$, anD these ARe the sEts associAtED to the CHildren. tHE sEt $I$ wIll be the root of all THeSE trees.
Let us be mOre preCiSE iN NOtaTioN and concepTs. we notE By ${{\cal T}}={{\cAL T}}({{\CAL j}}_\rhO)$ The family of dyAdic trees roOTed By $I$, whiCh DepENds on ${{\cAl J}}_\rhO$, aND thAt we will conStruCt in an indUctive WAy. The reCUrsion wIll depEnd On tHe leNGtH $|T|$ Of a TrEE $T\iN {{\CaL T}}$, sO The Classes ${{\cAl t}}_n=\{t\in {{\caL T}}: \, |T|=n\}$ WILL Be deFinEd in A recuRsive way for $n\gE 0$.
A tRee $T\IN {{\caL T}}$ is dEfineD as tHe Set of Its braNches. a bRanch $b\in T$ is a tupLe of Elements iN ${{\caL Y}}({{\Cal j}}_\rHo)$ and ITs last ComPonEnt is caLled its LEaf AnD NOTeD $\ell(b)$. The set of leaveS oF THe Tree $T$ is $$\pArtial(t)=\{\ElL(b): B\In T\},$$ and a lEaF of $t$ is sIMPly noTed $\eLL\iN \partial(t)$. As a coNSeQuEnce of oUr ConstrUcTioN of ${{\Cal T}}$, aLL the BranchEs $b$ of a trEe $T\in {{\CAl T}}$ will have the SAme length, so $|b|=|T|$. aNy OF ThESe brAncHes is writteN $b=(b_0,..,b_{|t|})$, And sO $b_{|T|}=\eLL(b)$.
belOW, the aLgoriThM Of COnstruction of ${{\cal T}}$ is GiVen as a RecurSive definitioN of $({{\cal T}}_n: n\gE 0)$.
fOR $n=0$, the claSs ${{\caL t}}_0$ iS A singleton formEd by tHe unique trEE $T=\{I\}$. So, it tHat haS a unique Branch $b=(I)$ wITH leaf $\ell(B)=I$. THe lEngTh oF $t$ Is By definition $|T|=0$, SO $|B|=0$ and $B_0=I$.
assume wE haVe constRucTed The Set Of | }}({{\cal J}}_\rho)$. Thedyadic pro perty me ans t hateach parent node h a s on e or two children: ifit ha so ne c h il drenthe set as s o cia te dtoth e c hildr enis thesame as th e o ne of the pare n t, and whenithas two chil dre n these t o f thepar ent i s part i tioned into two d i sjoint nonempt y se ts b y some set in ${{ \ ca l J}}_\rho^{(I) }$, an dt he s e ar e t he sets as so ciate d to the ch i l d ren . The set $I$will be the roo t of a ll th e se tre es.
Le t us be more pr ecis e in nota tion a n d conce p ts. Wenote b y $ {{\ calT }} ={ {\c al T}} ( {{ \ca l J} }_\rho)$ t he fami ly o f d y adic tr eesroote d by $I$, whi chdepe n dson ${ {\cal J}} _\ rho$, and t hat w ewill constructin a n inducti vewa y.Th e rec u rsionwil l d epend o n the l e ngt h$ | T |$ of a tree $T\in { {\ c a lT}}$, so the c l as se s ${{\cal T }}_ n=\{ T \ in {{ \cal T} }: \, |T |=n\}$ wi ll be def in ed inarec urs ive w a y fo r $n\g e 0$.
A tree $T\in {{\cal T } }$ is defined as t he setofits branche s. A bran ch $ b \i n T $ is a tupl eo fe lements in ${{\calY} }({{\c al J} }_\rho)$ andits last c o m p onent is cal l ed its leaf and n oted$\ell(b)$. The setof le aves ofthe tree$ T $ is $$\ par tia l(T )=\ { \ el l(b): b\in T\ } , $$ a nd a leaf of $T$ is si mpl y n ote d$\ell\in\partial (T )$ .As aconse q uence of o urco nst ructi o n of $ {{\ca l T} }$ ,a llthe bra n ch e s $b$ o fa tr ee$T \in { {\ca l T} }$ will have the sa m e le ng th , so $| b|=|T|$. Anyof these bra nc hes is wr i t ten $b=( b_0,..,b_{|T|})$, and s o $b_{|T |}= \ell( b)$.
Below,the algor ith m of co nstruc tionof ${ { \ cal T } } $isgi ven as a r e c urs ive d ef init ion of$({{\cal T}}_n: n\ g e 0 )$.
For $n=0 $,thec l as s $ { {\ c alT} } _0$ i s a singleton f ormed by t he un ique tree$ T=\ {I \}$. So , it th at ha s a uniq ue branch $b=(I)$wi th l e a f $ \ell(b)=I$ . The le ngth of $ T $ isb ydefin iti on $|T |= 0$, so $ |b|=0$ and $b_0 =I$.
As sume w e hav econstruc ted the set of | }}({{\cal J}}_\rho)$._The dyadic_property means that each_parent node_has_one or_two_children: if it_has one children_the set associated to_the children is_the_same as the one of the parent, and when it has two children the_set_of the_parent_is_partitioned into two disjoint nonempty_sets by some set in_${{\cal J}}_\rho^{(I)}$,_and these are the sets associated to the_children._The set $I$_will be the root of all these trees.
Let us_be more precise in notation and_concepts. We note_by_${{\cal_T}}={{\cal T}}({{\cal J}}_\rho)$ the_family of dyadic trees rooted by_$I$, which depends on ${{\cal J}}_\rho$,_and that we will construct in an_inductive way. The recursion will depend_on the length $|T|$ of_a tree_$T\in {{\cal T}}$, so the_classes ${{\cal T}}_n=\{T\in_{{\cal T}}:_\, |T|=n\}$ will_be defined in a recursive way_for $n\ge 0$.
A_tree $T\in {{\cal T}}$ is defined_as_the set of_its_branches._A branch_$b\in T$ is_a_tuple of_elements_in ${{\cal Y}}({{\cal J}}_\rho)$ and its_last_component is called its leaf and noted_$\ell(b)$. The set of_leaves_of the tree $T$_is $$\partial(T)=\{\ell(b): b\in T\},$$ and_a leaf of $T$ is simply_noted $\ell\in_\partial(T)$. As_a consequence of our construction of ${{\cal T}}$, all the branches_$b$ of a tree $T\in {{\cal_T}}$ will have the_same length,_so_$|b|=|T|$. Any of_these_branches is_written $b=(b_0,..,b_{|T|})$, and so $b_{|T|}=\ell(b)$.
Below, the algorithm_of construction_of ${{\cal T}}$ is given as_a recursive definition of_$({{\cal_T}}_n: n\ge 0)$.
For $n=0$, the class_${{\cal T}}_0$ is a singleton formed_by the unique tree $T=\{I\}$._So,_it_that has a unique branch_$b=(I)$ with leaf $\ell(b)=I$. The length_of $T$ is_by definition $|T|=0$, so $|b|=0$ and $b_0=I$.
Assume_we_have constructed the set of |
in ${\mathcal{F}}({\mathcal{A}})$ of $n$ conjugates of words in ${\mathcal{R}}^{\pm 1}$ and $n$ is minimal with this property. Equivalently, for given an input $(W, 1^n)$, the precise word problem asks whether there exists a disk diagram over whose boundary label is $W$, whose number of faces is $n$, and there are no such diagrams with fewer number of faces.
The definitions for the precise word problem for being solvable and being in a complexity class $\textsf K$ are similar to the corresponding definitions for the (bounded) word problem.
In the following proposition we list basic comparative properties of the standard word problem and its bounded and precise versions.
\[prp1\] $(\mathrm{a})$ There exists a decidable group presentation for which the word problem is solvable while the bounded and precise word problems are not solvable.
$(\mathrm{b})$ If the bounded word problem is solvable for, then the precise word problem is also solvable.
$(\mathrm{c})$ For every finite group presentation, the bounded word problem is in ${\textsf{NP}}$, i.e., it can be solved in nondeterministic polynomial time, and the precise word problem is in, i.e., it can be solved in polynomial space.
$(\mathrm{d})$ There exists a finite group presentation for which the bounded and precise word problems are solvable while the word problem is not solvable.
$(\mathrm{e})$ There exists a finitely presented group for which the bounded word problem is -complete and the precise word problem is -hard.
Note that a number of interesting results on solvability of the word problem, solvability of the bounded word problem and computability of the Dehn function for decidable group presentations can be found in a preprint of Cummins [@DC] based on his PhD thesis written under the author’s supervision.
It is of interest to look at the bounded and precise word problems for finitely presented groups for which the word problem could be solved very easily. Curiously, even for very simple presentations such as $ \langle \ a, b \ \| \ a=1 \ \rangle $ and $ \langle \ a, b \ \| \ ab=ba \ \rangle $, for which the word problem is obviously in $\textsf{L}$, i.e., solvable in deterministic logarithmic | in $ { \mathcal{F}}({\mathcal{A}})$ of $ n$ conjugates of words in $ { \mathcal{R}}^{\pm 1}$ and $ n$ is minimal with this property. Equivalently, for given an stimulation $ (W, 1^n)$, the accurate word problem ask whether there exist a disk diagram over whose boundary label is $ W$, whose number of boldness is $ n$, and there be no such diagrams with few numeral of faces.
The definition for the precise word trouble for being solvable and being in a complexity class $ \textsf K$ are similar to the corresponding definitions for the (bounded) word problem.
In the succeed proposition we list basic relative properties of the standard word trouble and its bounded and precise version.
\[prp1\ ] $ (\mathrm{a})$ There exists a decidable group display for which the word problem is solvable while the bounded and precise discussion problems are not solvable.
$ (\mathrm{b})$ If the bounded word problem is solvable for, then the precise son trouble is also solvable.
$ (\mathrm{c})$ For every finite group presentation, the bounded word problem is in $ { \textsf{NP}}$, i.e., it can be solved in nondeterministic polynomial time, and the accurate word trouble is in, i.e., it can be solved in polynomial space.
$ (\mathrm{d})$ There exist a finite group display for which the bounded and accurate word problems are solvable while the word problem is not solvable.
$ (\mathrm{e})$ There exists a finitely presented group for which the bounded word trouble is -complete and the precise word problem is -hard.
Note that a number of interesting results on solvability of the word problem, solvability of the bounded word problem and computability of the Dehn function for decidable group presentation can be found in a preprint of Cummins [ @DC ] establish on his PhD thesis written under the author ’s supervision.
It is of interest to count at the bounded and precise word problems for finitely presented groups for which the word problem could be solved very well. Curiously, even for very simple presentations such as $ \langle \ a, b \ \| \ a=1 \ \rangle $ and $ \langle \ a, b \ \| \ ab = ba \ \rangle $, for which the word problem is obviously in $ \textsf{L}$, i.e., solvable in deterministic logarithmic | in ${\mathcal{F}}({\mathcal{A}})$ of $n$ cunjugates of wotdw in ${\methcal{R}}^{\lm 1}$ and $v$ is minimal with this propecty. Wquivqlently, for given an ivput $(W, 1^n)$, nhe preciwe wied problem asks whccher fmere zxmsts a disk dianram over wvose boundary nacep is $W$, whose number of faces is $n$, agd therr wre no such diwgrakf wifh fewer number of faces.
The definifions fmr the precisr word problem for being sllvahle and being in a complexity clafw $\textsf K$ afe similar to the corrgsponding definitions for the (bojnded) word proboen.
In jhe followinj propjsition we llxt bashc compsrative properbies mf rhe standard word proulem and its bounded and prechsz versions.
\[prp1\] $(\mathrm{a})$ Tyere gxistv a awciaabme gdoup pgesxntation fod which the word problem is sokvwvle while the boundqd and precise word problems are not solvdbls.
$(\mathrm{b})$ If the bounded word problem is solvwble for, ehen the precise word problem is also solvable.
$(\matvrm{c})$ Hof ercvy fkbihe group presentation, the bounded word problei ix pn ${\textsf{NP}}$, i.e., it can be solfef og nondeterminkstic 'klgnomial time, and tje precyse wird probltm is in, i.e., it can be solved in polynomial siace.
$(\nathrm{d})$ There exiscs a finite yroup kresenyation for which the bobnded znd precise word proguems are solvabld wmila the word problem is not folvable.
$(\methrm{z})$ There dxisjs a figitely predented group for which tje boonded ford problfm is -complete and the precise xprd problem os -hagd.
Note thct a nmmber of interefting results pn solvcbilith of the wkrd proulem, solvabijity of the bmonded word prmblem anq conputqbility uf the Dehn fumction fog becidable group presentatioks cav be found in a 'xekrunt of Cummins [@DC] bafef pn vis PhD theshs wfitgrn unaer the autmor’r sulervision.
It is of indereat to look at the nomnded and precise word problemx for finitely predentev grou's for whych the word problem could be aolved vegy casily. Curioufly, cven for very ximple presentations such as $ \langle \ a, b \ \| \ a=1 \ \rangle $ and $ \langle \ a, b \ \| \ ab=ba \ \rengle $, for whicv the word problem iw obviously in $\tewtsf{L}$, i.e., solvable in dstermitistif logarithmic | in ${\mathcal{F}}({\mathcal{A}})$ of $n$ conjugates of words 1}$ $n$ is with this property. $(W, the precise word asks whether there a disk diagram over whose boundary is $W$, whose number of faces is $n$, and there are no such with fewer number of faces. The definitions for the precise word problem for solvable being a class $\textsf K$ are similar to the corresponding definitions for the (bounded) word problem. In the proposition we list basic comparative properties of the word problem and its and precise versions. \[prp1\] $(\mathrm{a})$ exists decidable group for the problem is solvable the bounded and precise word problems are not solvable. $(\mathrm{b})$ If the bounded word problem is solvable then the problem is solvable. For finite group presentation, word problem is in ${\textsf{NP}}$, i.e., solved in nondeterministic polynomial time, and the precise problem is i.e., it can be solved in space. $(\mathrm{d})$ There exists a finite group presentation which the bounded and precise word problems are solvable while the word problem is not There exists a finitely group for which bounded problem -complete the precise problem is -hard. Note that a number of interesting results on of the word problem, solvability of the bounded word problem of Dehn function for group presentations can be in preprint of Cummins [@DC] his thesis author’s It of interest to look the bounded and precise word for finitely presented groups could be solved very easily. Curiously, even for simple presentations such as $ \langle \ b \ \| \ a=1 \ \rangle $ and $ \langle \ b \ ab=ba \ \rangle $, for which the word is obviously in $\textsf{L}$, solvable in deterministic logarithmic | in ${\mathcal{F}}({\mathcal{A}})$ of $n$ conjUgates of woRds in ${\MatHcaL{R}}^{\Pm 1}$ anD $n$ is Minimal with thiS PropErty. Equivalently, for givEn an iNpUT $(W, 1^n)$, tHE pRecisE word prOBlEM AskS wHeTheR tHErE exisTs a Disk diaGram over whOse BoUndary label iS $w$, wHose number Of fAces is $n$, and thEre Are no sUcH diAGrams WitH feweR numbeR Of faceS.
The definItIOns for THe preciSE WoRd prOblem for being solvABlE And being in a comPlexitY cLAsS $\TExtSf K$ Are similar To The coRRespondINg DEFIniTIons for the (bouNded) word proBLem.
in the fOlLowINg propOsitiOn WE liSt basic compAratIve properTies of THe standARd word pRoblem And Its BounDEd AnD prEcISe vERsIonS.
\[Prp1\] $(\Mathrm{a})$ THeRe ExistS a deCIDABle gRouP preSentaTion for which tHe wOrd pROblEm is sOlvabLe whIlE the bOunded And prEcIse word problems Are nOt solvablE.
$(\maThRm{b})$ if The boUNded woRd pRobLem is soLvable fOR, thEn THE PrEcise word problem is AlSO SoLvable.
$(\maThrm{c})$ FOR eVeRY finite gRoUp pReseNTAtion, The bOUnDed word pRoblem IS iN ${\tExtsf{NP}}$, I.e., It can bE sOlvEd iN nondETermInistiC polynomIal tiME, and the precise WOrd problem is iN, I.e., IT CaN Be soLveD in polynomiAl spACe.
$(\maThrm{D})$ thEre EXists A finiTe GRoUP presentation for whiCh The bouNded aNd precise word Problems arE SOLvable whIle tHE wORd problem is not SolvaBle.
$(\mathrm{e})$ tHere exisTs a fiNitely prEsented grOUP for whicH thE boUndEd wORD pRoblem is -complETE and ThE precisE woRd problEm iS -haRd.
NOte ThAt a number Of intereStInG rEsUltS on soLVability Of The WoRd pRobleM, SolvabIlity Of thE bOuNDed Word proBLeM ANd coMpUtAbilIty Of The DeHn fuNCtiOn for deCidable grOup PReseNtAtIons can Be found in a prePrInt of CummiNs [@dC] bAsed on HIS PhD thesIs written under the author’S SuperviSioN.
It is Of inTerest to lOok At the bOunDEd and pRecise Word pRoBleMS For fiNITeLy pReSented grouPS For Which ThE worD probleM could be solved very EAsiLy. Curiously, evEn fOr veRY SiMplE PrESenTaTIonS SUch as $ \langle \ a, b \ \| \ a=1 \ \rAngle $ and $ \laNgLE \ a, B \ \| \ ab=ba \ \ranglE $, For WhIch the wOrd probLem is OBviouslY in $\textsf{l}$, i.e., solvabLe In deTERmiNistic logaRithmic | in ${\mathcal{F}}({\mathc al{A}})$ o f $n$ co nju ga tesof w ords in ${\mat h cal{ R}}^{\pm 1}$ and $n$ i s min im a l wi t hthispropert y .E q uiv al en tly ,f or give n a n input $(W, 1^n) $,th e precise wo r dproblem as kswhether ther e e xistsadis k diag ram over whose bounda ry labelis $W$, w h ose num b e rof f aces is $n$, andt he r e are no suchdiagra ms wi t h fe wer number of f aces.
The de f in i t i ons for the preci se word pro b lem for b ei ngs olvabl e and b e ing in a compl exit y class $ \texts f K$ are similar to th e c orr espo n di ng de fi n iti o ns fo r th e (bound ed )wordprob l e m .
In th e fo llowi ng propositio n w e li s t b asiccompa rati ve prop erties of t he standard wordprob lem and i tsbo und ed andp recise ve rsi ons.
\ [prp1\] $(\ ma t h r m{ a})$ There existsad e ci dable gr oup pr e se nt a tion for w hic h th e wordprob l em is solv able w h il ethe bou nd ed and p rec ise word prob lems a re not s olvab l e.
$(\mathrm{ b })$ If the bo u nd e d w o rd p rob lem is solv able for, the n t hep recis e wor dp ro b lem is also solvabl e.
$(\m athrm {c})$ For eve ry finiteg r o up prese ntat i on , the bounded w ord p roblem isi n ${\tex tsf{N P}}$, i. e., it ca n be solve d i n n ond ete r m in istic polynom i a l ti me , and t heprecise wo rdpro ble mis in, i. e., it c an b eso lve d inp olynomia lspa ce .
$(\ma t hrm{d} )$ Th ereex is t s a finite gr o u p pr es en tati onfo r whi ch t h e b oundedand preci sew ordpr ob lems ar e solvable wh il e the word p rob lem is n ot solva ble.
$(\mathrm{e})$ Th e re exis tsa fin itel y present edgroupfor whichthe bo unded w ord p roble m is -c om plete andt h e p recis eword proble m is -hard.
Notet hat a number ofint eres t i ng re s ul t s o ns olv a b ility of the wo rd problem ,s ol vability o f th ebounded word p roble m and co mputabili ty of the D ehnf u nct ion for de cidablegroup pre s entat i on s can be found i n a prep rint o f Cu mmins [@DC] b ased o n his P hD thesi s written under the aut hor’ssuper vis ion.
Itiso f i nterest t o lo ok at thebou nde d and pr e ciseword pr obl e ms fo r fi n itely pre s en ted g ro ups for whi c h the word pr o blem c ould be solved very e a sily. Curiousl y, e v e n f orv erysi mple presentat ion ss u ch as $\l angle \ a,b \ \|\a =1 \ \rang le $ a nd $ \l a n gl e \ a,b \ \| \ ab=ba \ \ r angle $ ,fo r which the w ord pr oblemi s ob v i ously in $\texts f{L}$ , i.e., sol vable i n deter m inis tic logari thmic | in_${\mathcal{F}}({\mathcal{A}})$ of_$n$ conjugates of words_in ${\mathcal{R}}^{\pm_1}$_and $n$_is_minimal with this_property. Equivalently, for_given an input $(W,_1^n)$, the precise_word_problem asks whether there exists a disk diagram over whose boundary label is $W$,_whose_number of_faces_is_$n$, and there are no_such diagrams with fewer number_of faces.
The_definitions for the precise word problem for being_solvable_and being in_a complexity class $\textsf K$ are similar to the_corresponding definitions for the (bounded) word_problem.
In the following_proposition_we_list basic comparative properties_of the standard word problem and_its bounded and precise versions.
\[prp1\] $(\mathrm{a})$_There exists a decidable group presentation for_which the word problem is solvable_while the bounded and precise_word problems_are not solvable.
$(\mathrm{b})$ If the_bounded word problem_is solvable_for, then the_precise word problem is also solvable.
$(\mathrm{c})$_For every finite_group presentation, the bounded word problem_is_in ${\textsf{NP}}$, i.e.,_it_can_be solved_in nondeterministic polynomial_time,_and the_precise_word problem is in, i.e., it_can_be solved in polynomial space.
$(\mathrm{d})$ There exists_a finite group presentation_for_which the bounded and_precise word problems are solvable_while the word problem is not_solvable.
$(\mathrm{e})$ There_exists a_finitely presented group for which the bounded word problem is -complete_and the precise word problem is_-hard.
Note that a number_of interesting_results_on solvability of_the_word problem,_solvability of the bounded word problem and_computability of_the Dehn function for decidable group_presentations can be found_in_a preprint of Cummins [@DC] based_on his PhD thesis written under_the author’s supervision.
It is of_interest_to_look at the bounded and_precise word problems for finitely presented_groups for which_the word problem could be solved very_easily._Curiously, even for very simple presentations_such_as $ \langle \ a, b__\_\| \ a=1 \_\rangle $ and $ \langle \_a, b \ \| \ ab=ba _\ \rangle $,_for which the word problem_is_obviously_in $\textsf{L}$, i.e., solvable in deterministic logarithmic |
, Tian Tian, Xin Huang, Lin Wang, Jun Zhu, and Le Song. Adversarial attack on graph structured data., 2018.
Xuanqing Liu, Si Si, Xiaojin Zhu, Yang Li, and Cho-Jui Hsieh. A unified framework for data poisoning attack to graph-based semi-supervised learning., 2019.
Aleksandar Bojchevski and Stephan G[ü]{}nnemann. Adversarial attacks on node embeddings via graph poisoning. In [*ICML*]{}, pages 695–704, 2019.
Marinka Zitnik, Monica Agrawal, and Jure Leskovec. Modeling polypharmacy side effects with graph convolutional networks., 34(13):i457–i466, 2018.
Zhitao Ying, Dylan Bourgeois, Jiaxuan You, Marinka Zitnik, and Jure Leskovec. Gnnexplainer: Generating explanations for graph neural networks. In [*NeurIPS*]{}, 2019.
Pablo Gainza, Freyr Sverrisson, Frederico Monti, Emanuele Rodola, D Boscaini, MM Bronstein, and BE Correia. Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning., 17(2), 2020.
Ian J Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harnessing adversarial examples., 2015.
Robin Jia and Percy Liang. Adversarial examples for evaluating reading comprehension systems., 2017.
Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. In [*NIPS*]{}, 2017.
Xiao Zang, Yi Xie, Jie Chen, and Bo Yuan. Graph universal adversarial attacks: A few bad actors ruin graph learning models., 2020.
Shen Wang, Zhengzhang Chen, Jingchao Ni, Xiao Yu, Zhichun Li, Haifeng Chen, and Philip S Yu. Adversarial defense framework for graph neural network., 2019.
Kaidi Xu, Hongge Chen, Sijia Liu, Pin Yu Chen, Tsui Wei Weng, Mingyi Hong, and Xue Lin. Top | , Tian Tian, Xin Huang, Lin Wang, Jun Zhu, and Le Song. Adversarial attack on graph structured data. , 2018.
Xuanqing Liu, Si Si, Xiaojin Zhu, Yang Li, and Cho - Jui Hsieh. A unified model for datum poisoning approach to graph - based semi - supervised eruditeness. , 2019.
Aleksandar Bojchevski and Stephan G[ü]{}nnemann. Adversarial attacks on node embeddings via graph poisoning. In [ * ICML * ] { }, page 695–704, 2019.
Marinka Zitnik, Monica Agrawal, and Jure Leskovec. Modeling polypharmacy english effect with graph convolutional networks. , 34(13):i457 – i466, 2018.
Zhitao Ying, Dylan Bourgeois, Jiaxuan You, Marinka Zitnik, and Jure Leskovec. Gnnexplainer: generate explanations for graph neural network. In [ * NeurIPS * ] { }, 2019.
Pablo Gainza, Freyr Sverrisson, Frederico Monti, Emanuele Rodola, D Boscaini, MM Bronstein, and BE Correia. Deciphering interaction fingerprints from protein molecular airfoil use geometric deep eruditeness. , 17(2), 2020.
Ian J Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harness adversarial example. , 2015.
Robin Jia and Percy Liang. Adversarial examples for evaluating reading inclusion systems. , 2017.
Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation teach on large graph. In [ * NIPS * ] { }, 2017.
Xiao Zang, Yi Xie, Jie Chen, and Bo Yuan. Graph universal adversarial approach: A few bad actors ruin graph eruditeness models. , 2020.
Shen Wang, Zhengzhang Chen, Jingchao Ni, Xiao Yu, Zhichun Li, Haifeng Chen, and Philip S Yu. Adversarial defense framework for graph neural network. , 2019.
Kaidi Xu, Hongge Chen, Sijia Liu, Pin Yu Chen, Tsui Wei Weng, Mingyi Hong, and Xue Lin. Top | , Tiwn Tian, Xin Huang, Lin Wakg, Jun Zhu, and Lg Sing. Adtersarizl attacy on graph structured data., 2018.
Xnanqung Luu, Si Si, Xiaojin Zhu, Yavg Li, and Cho-Jui Ysiei. A unified framxsork fov datz poivining attack tp graph-basad semi-supervivea pearning., 2019.
Aleksandar Bojchevski and Ftephan G[ü]{}jnemann. Adverswriak attzbkw on node embeddings via grapg poisoiing. In [*ICML*]{}, pabes 695–704, 2019.
Marinka Zitnik, Monica Agrwwal, and Jure Lesklvec. Modelibg pjoypharmacy skde effects with graph convolutional networks., 34(13):i457–i466, 2018.
Zhitxo Yiug, Dylan Booxteolv, Jiaxuan Yiu, Maginka Zitnik, and Jure Leskovrc. Gnnexplainev: Genxratung explanations for jraph neural networkf. In [*NeurHPA*]{}, 2019.
Pablo Gainza, Frwye Svetrissmn, Ffwdefick Kohti, Emwnuxle Rodola, S Boscaini, MN Bronstein, and BE Cotrqps. Decipherinf intewastion fingerprints from protein moleculdr aurfaces using geometrix deep learning., 17(2), 2020.
Ian J Goodfelljw, Jonathon Shlens, and Christian Szegedy. Explainitg anv farucfrunh adversarial examples., 2015.
Robin Jia and Percy Liwhg. Acversarial exaiples for efapusjing reading cumprehzhsjon systems., 2017.
Will Hwmilton, Zhitqo Ying, agd Jire Leskovec. Inductive reprwsentation lvarnung on large graphd. In [*NIPS*]{}, 2017.
Xnao Zamg, Yi Cie, Jie Chen, and Bo Yuan. Gralh universap adversadkal attacks: A fed bsd actors ruin graph learnind models., 2020.
When Wang, Zfengehang Cren, Jingchwo Ni, Xiao Yu, Zhichun Li, Haifgng Chan, and Phipip S Yu. Adversarial defense framework for grakh teugal netwoxk., 2019.
Kaici Xu, Hongge Shen, Sijia Liu, Pin Yu Ehen, Trui Wei Wenf, Mingym Hong, and Xte Lin. Top | , Tian Tian, Xin Huang, Lin Wang, and Song. Adversarial on graph structured Si, Zhu, Yang Li, Cho-Jui Hsieh. A framework for data poisoning attack to semi-supervised learning., 2019. Aleksandar Bojchevski and Stephan G[ü]{}nnemann. Adversarial attacks on node embeddings graph poisoning. In [*ICML*]{}, pages 695–704, 2019. Marinka Zitnik, Monica Agrawal, and Jure Modeling side with convolutional networks., 34(13):i457–i466, 2018. Zhitao Ying, Dylan Bourgeois, Jiaxuan You, Marinka Zitnik, and Jure Leskovec. Gnnexplainer: explanations for graph neural networks. In [*NeurIPS*]{}, 2019. Gainza, Freyr Sverrisson, Frederico Emanuele Rodola, D Boscaini, MM and Correia. Deciphering fingerprints protein surfaces using geometric learning., 17(2), 2020. Ian J Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harnessing adversarial examples., 2015. Jia and Adversarial examples evaluating comprehension 2017. Will Hamilton, and Jure Leskovec. Inductive representation learning In [*NIPS*]{}, 2017. Xiao Zang, Yi Xie, Jie and Bo Graph universal adversarial attacks: A few actors ruin graph learning models., 2020. Shen Wang, Chen, Jingchao Ni, Xiao Yu, Zhichun Li, Haifeng Chen, and Philip S Yu. Adversarial defense graph neural network., 2019. Xu, Hongge Chen, Liu, Yu Tsui Weng, Mingyi and Xue Lin. Top | , Tian Tian, Xin Huang, Lin Wang, JuN Zhu, and Le SOng. AdVerSarIaL attAck oN graph structurED datA., 2018.
Xuanqing Liu, Si Si, XiaojiN Zhu, YAnG li, anD chO-Jui HSieh. A unIFiED FraMeWoRk fOr DAtA poisOniNg attacK to graph-baSed SeMi-supervised LEaRning., 2019.
AleksAndAr Bojchevski And stephaN G[Ü]{}nnEMann. ADveRsariAl attaCKs on noDe embeddiNgS Via graPH poisonING. IN [*ICMl*]{}, pages 695–704, 2019.
Marinka ZitnIK, MONica Agrawal, and jure LeSkOVeC. mOdeLinG polypharmAcY side EFfects wITh GRAPh cONvolutional neTworks., 34(13):i457–i466, 2018.
ZhiTAo YIng, DylAn bouRGeois, JIaxuaN YOU, MaRinka Zitnik, And JUre LeskovEc. GnneXPlainer: gEneratiNg explAnaTioNs foR GrApH neUrAL neTWoRks. iN [*NeUrIPS*]{}, 2019.
PabLo gaInza, FReyr sVERRissOn, FRedeRico MOnti, Emanuele ROdoLa, D BOScaIni, MM bronsTein, AnD BE CoRreia. DEciphErIng interaction fIngeRprints frOm pRoTeiN mOlecuLAr surfAceS usIng geomEtric deEP leArNING., 17(2), 2020.
IAn J Goodfellow, JonatHoN sHlEns, and ChRistiaN szEgEDy. ExplaiNiNg aNd haRNEssinG advERsArial exaMples., 2015.
ROBiN JIa and PeRcY Liang. adVerSarIal exAMpleS for evAluating ReadiNG comprehension SYstems., 2017.
Will HamILtON, zhITao YIng, And Jure LeskOvec. iNducTive REpResENtatiOn leaRnINg ON large graphs. In [*NIPS*]{}, 2017.
XIaO Zang, YI Xie, JIe Chen, and Bo YuAn. Graph uniVERSal adverSariAL aTTacks: A few bad acTors rUin graph leARning modEls., 2020.
ShEn Wang, ZhEngzhang CHEN, JingchaO Ni, xiaO Yu, zhiCHUn li, Haifeng Chen, AND PhiLiP S Yu. AdvErsArial deFenSe fRamEwoRk For graph nEural netWoRk., 2019.
kaIdI Xu, honggE chen, SijiA LIu, PIn yu CHen, TsUI Wei WeNg, MinGyi HOnG, aND XuE Lin. Top | , Tian Tian, Xin Huang, Li n Wang, Ju n Zhu , a ndLe Son g. A dversarial att a ck o n graph structured dat a., 2 01 8 .
X u an qingLiu, Si Si , Xia oj in Zh u, Ya ng Li , a nd Cho- Jui Hsieh. Aun ified framew o rk for datapoi soning attac k t o grap h- bas e d sem i-s uperv ised l e arning ., 2019.
A l eksand a r Bojch e v sk i an d Stephan G[ü]{}n n em a nn. Adversaria l atta ck s o n nod e e mbeddingsvi a gra p h poiso n in g . In[ *ICML*]{}, pa ges 695–704 , 20 19.
M ar ink a Zitni k, Mo ni c a A grawal, and Jur e Leskove c. Mod e ling po l ypharma cy sid e e ffe ctsw it hgra ph con v ol uti o nal network s. ,34(13 ):i4 5 7 – i 466, 20 18.
Zhit ao Ying, Dyla n B ourg e ois , Jia xuanYou, M arink a Zitn ik, a nd Jure Leskovec. Gnn explainer : G en era ti ng ex p lanati ons fo r graph neural net wo r k s .In [*NeurIPS*]{},20 1 9 .
Pablo G ainza, Fr ey r Sverris so n,Fred e r ico M onti , E manueleRodola , D B oscaini ,MM Bro ns tei n,and B E Cor reia.Decipher ing i n teraction fing e rprints fromp ro t e in mole cul ar surfaces usi n g ge omet r ic de e p lea rning ., 17 ( 2), 2020.
Ian J Go od fellow , Jon athon Shlens, and Chris t i a n Szeged y. E x pl a ining and harn essin g adversar i al examp les., 2015.
Robin Jia a nd Percy Li ang . A dve r s ar ial examplesf o r ev al uatingrea ding co mpr ehe nsi onsy stems., 2 017.
Wi ll H am il ton , Zhi t ao Ying, a ndJu reLesko v ec. In ducti ve r ep re s ent ation l e ar n i ng o nla rgegra ph s. In [*N I PS* ]{}, 20 17.
Xiao Za n g, Y iXi e, JieChen, and BoYu an. Graphun ive rsal a d v ersarial attacks: A few bad act o rs ruin gr aph l earn ing model s., 2020.
S h en Wan g, Zhe ngzha ng Ch e n , Jin g c ha o N i, Xiao Yu,Z h ich un Li ,Haif eng Che n, and Philip S Yu . Ad versarial def ens e fr a m ew ork fo r gr ap h ne u r al network., 20 19.
Kaidi X u ,Hongge Che n , S ij ia Liu, Pin Yu Chen , Tsui W ei Weng,Mingyi Ho ng , an d Xue Lin. Top | , Tian_Tian, Xin_Huang, Lin Wang, Jun_Zhu, and_Le Song._Adversarial attack_on_graph structured data.,_2018.
Xuanqing Liu, Si Si,_Xiaojin Zhu, Yang Li,_and Cho-Jui Hsieh._A_unified framework for data poisoning attack to graph-based semi-supervised learning., 2019.
Aleksandar Bojchevski and Stephan_G[ü]{}nnemann._Adversarial attacks_on_node_embeddings via graph poisoning. In_[*ICML*]{}, pages 695–704, 2019.
Marinka Zitnik,_Monica Agrawal,_and Jure Leskovec. Modeling polypharmacy side effects with_graph_convolutional networks., 34(13):i457–i466,_2018.
Zhitao Ying, Dylan Bourgeois, Jiaxuan You, Marinka Zitnik, and_Jure Leskovec. Gnnexplainer: Generating explanations for_graph neural networks._In_[*NeurIPS*]{},_2019.
Pablo Gainza, Freyr Sverrisson,_Frederico Monti, Emanuele Rodola, D Boscaini, MM Bronstein,_and BE Correia. Deciphering interaction fingerprints from_protein molecular surfaces using geometric deep learning.,_17(2), 2020.
Ian J Goodfellow, Jonathon Shlens, and_Christian Szegedy. Explaining and harnessing_adversarial examples.,_2015.
Robin Jia and Percy Liang._Adversarial examples for_evaluating reading_comprehension systems., 2017.
Will_Hamilton, Zhitao Ying, and Jure Leskovec._Inductive representation learning_on large graphs. In [*NIPS*]{}, 2017.
Xiao_Zang,_Yi Xie, Jie Chen,_and_Bo Yuan._Graph universal_adversarial attacks: A_few_bad actors_ruin_graph learning models., 2020.
Shen Wang, Zhengzhang_Chen,_Jingchao Ni, Xiao Yu, Zhichun Li, Haifeng_Chen, and Philip S Yu._Adversarial_defense framework for graph_neural network., 2019.
Kaidi Xu, Hongge_Chen, Sijia Liu, Pin Yu Chen, Tsui Wei_Weng, Mingyi_Hong, and_Xue Lin. Top |
^{(t)}\textbf{R}_2^{(t)})^{-1} \}
\big]
\\
+ \mathbb{E}\big[
\text{log}_2 \text{det} \{ \textbf{I}_{N_s} +
p_2 \textbf{R}_1^{(t)^H}\textbf{H}_{r,1}^{(t)}\bar{\textbf{F}}^{(t)}\textbf{H}_{2,r}^{(t-1)}
\textbf{H}_{2,r}^{(t-1)^H}\bar{\textbf{F}}^{(t)^H}\textbf{H}_{r,1}^{(t)^H}\textbf{R}_1^{(t)}
(\textbf{R}_1^{(t)^H}\textbf{A}_1^{(t)}\textbf{R}_1^{(t)})^{-1} \}
\big]
\end{split}$$ where $\textbf{A}_l^{(1)} = \sigma_{n,r}^2\textbf{H}_{r,l}^{(1)}\bar{\textbf{F}}^{(1)}\bar{\textbf{F}}^{(1)^H}\textbf{H}_{r,l}^{(1)^H}
+ \alpha^{(1)^{-2}}p_l\boldsymbol{\Delta}_{l,l}^{(1)}\boldsymbol{\Delta}_{l,l}^{(1)^H} + \alpha^{(1)^{-2}}\sigma_{n,l}^2\textbf{I}_{N_s}$ and $$\begin{split}
\textbf{A}_l^{(t)}
&= \sigma_{n,r}^2\textbf{H}_{r,l}^{(t)}\bar{\textbf{F}}^{(t)}\bar{\textbf{F}}^{(t)^H}\textbf{H}_{r,l}^{(t)^H}
+ \sum_{i=2}^{t} \bigg\{ \textbf{H}_{r,l}^{(t)}\bar{\textbf{F}}^{(t)} \prod_{j=1}^{i-1}( \boldsymbol{\Delta}_{r,r}^{(t-j)}\textbf{F}^{(t-j)})
\\
&(p_l\textbf{H}_{l,r}^{(t-i)}\textbf{H}_{l,r}^{(t-i)^H} + p_{\bar{l}}\textbf{H}_{\bar{l},r}^{(t-i)}\textbf{H}_{\bar{l},r}^{(t-i)^H} + | ^{(t)}\textbf{R}_2^{(t)})^{-1 } \ }
\big ]
\\
+ \mathbb{E}\big [
\text{log}_2 \text{det } \ { \textbf{I}_{N_s } +
p_2 \textbf{R}_1^{(t)^H}\textbf{H}_{r,1}^{(t)}\bar{\textbf{F}}^{(t)}\textbf{H}_{2,r}^{(t-1) }
\textbf{H}_{2,r}^{(t-1)^H}\bar{\textbf{F}}^{(t)^H}\textbf{H}_{r,1}^{(t)^H}\textbf{R}_1^{(t) }
(\textbf{R}_1^{(t)^H}\textbf{A}_1^{(t)}\textbf{R}_1^{(t)})^{-1 } \ }
\big ]
\end{split}$$ where $ \textbf{A}_l^{(1) } = \sigma_{n, r}^2\textbf{H}_{r, l}^{(1)}\bar{\textbf{F}}^{(1)}\bar{\textbf{F}}^{(1)^H}\textbf{H}_{r, l}^{(1)^H }
+ \alpha^{(1)^{-2}}p_l\boldsymbol{\Delta}_{l, l}^{(1)}\boldsymbol{\Delta}_{l, l}^{(1)^H } + \alpha^{(1)^{-2}}\sigma_{n, l}^2\textbf{I}_{N_s}$ and $ $ \begin{split }
\textbf{A}_l^{(t) }
& = \sigma_{n, r}^2\textbf{H}_{r, l}^{(t)}\bar{\textbf{F}}^{(t)}\bar{\textbf{F}}^{(t)^H}\textbf{H}_{r, l}^{(t)^H }
+ \sum_{i=2}^{t } \bigg\ { \textbf{H}_{r, l}^{(t)}\bar{\textbf{F}}^{(t) } \prod_{j=1}^{i-1 } (\boldsymbol{\Delta}_{r, r}^{(t - j)}\textbf{F}^{(t - j) })
\\
& (p_l\textbf{H}_{l, r}^{(t - i)}\textbf{H}_{l, r}^{(t - i)^H } + p_{\bar{l}}\textbf{H}_{\bar{l},r}^{(t - i)}\textbf{H}_{\bar{l},r}^{(t - i)^H } + | ^{(t)}\tedtbf{R}_2^{(t)})^{-1} \}
\big]
\\
+ \mathbb{E}\big[
\texu{log}_2 \text{det} \{ \texjbd{I}_{N_s} +
p_2 \textbr{R}_1^{(t)^H}\textcf{H}_{r,1}^{(t)}\bar{\textbf{F}}^{(t)}\textbf{H}_{2,r}^{(t-1)}
\texvbf{H}_{2,e}^{(t-1)^H}\bae{\textbf{F}}^{(t)^H}\textbf{H}_{r,1}^{(t)^H}\tebtbf{R}_1^{(t)}
(\texnbf{R}_1^{(t)^H}\texrbf{A}_1^{(u)}\textbf{R}_1^{(t)})^{-1} \}
\big]
\end{s'mit}$$ whevz $\texfnf{A}_l^{(1)} = \wigma_{n,r}^2\textbf{H}_{t,l}^{(1)}\bar{\textbf{F}}^{(1)}\tar{\textbf{F}}^{(1)^H}\texdbw{H}_{x,l}^{(1)^H}
+ \alpha^{(1)^{-2}}p_l\boldsymbol{\Delta}_{l,l}^{(1)}\boldsymbjl{\Delta}_{k,l}^{(1)^J} + \alpha^{(1)^{-2}}\sigma_{n,j}^2\texubf{Y}_{N_s}$ znd $$\begin{split}
\textbf{A}_l^{(t)}
&= \sigma_{n,r}^2\texfbf{H}_{r,l}^{(t)}\uar{\textbf{F}}^{(t)}\bar{\trxtbf{F}}^{(t)^H}\textbf{H}_{r,l}^{(t)^H}
+ \sum_{i=2}^{t} \bigh\{ \textbf{H}_{r,l}^{(t)}\bar{\tedtbf{F}}^{(t)} \prod_{h=1}^{i-1}( \bjodsymbol{\Deltx}_{r,r}^{(t-j)}\textbf{F}^{(t-j)})
\\
&(p_l\textbf{G}_{l,r}^{(t-i)}\textbf{H}_{l,r}^{(t-i)^H} + p_{\bar{l}}\textbf{H}_{\car{l},r}^{(c-i)}\textbf{H}_{\bat{l},d}^{(t-l)^V} + | ^{(t)}\textbf{R}_2^{(t)})^{-1} \} \big] \\ + \mathbb{E}\big[ \text{log}_2 \textbf{I}_{N_s} p_2 \textbf{R}_1^{(t)^H}\textbf{H}_{r,1}^{(t)}\bar{\textbf{F}}^{(t)}\textbf{H}_{2,r}^{(t-1)} (\textbf{R}_1^{(t)^H}\textbf{A}_1^{(t)}\textbf{R}_1^{(t)})^{-1} \} \big] + + \alpha^{(1)^{-2}}\sigma_{n,l}^2\textbf{I}_{N_s}$ and \textbf{A}_l^{(t)} &= \sigma_{n,r}^2\textbf{H}_{r,l}^{(t)}\bar{\textbf{F}}^{(t)}\bar{\textbf{F}}^{(t)^H}\textbf{H}_{r,l}^{(t)^H} \sum_{i=2}^{t} \bigg\{ \textbf{H}_{r,l}^{(t)}\bar{\textbf{F}}^{(t)} \prod_{j=1}^{i-1}( \boldsymbol{\Delta}_{r,r}^{(t-j)}\textbf{F}^{(t-j)}) \\ + p_{\bar{l}}\textbf{H}_{\bar{l},r}^{(t-i)}\textbf{H}_{\bar{l},r}^{(t-i)^H} + | ^{(t)}\textbf{R}_2^{(t)})^{-1} \}
\big]
\\
+ \mathbb{E}\big[
\texT{log}_2 \text{deT} \{ \textBf{I}_{n_s} +
p_2 \TeXtbf{r}_1^{(t)^H}\tExtbf{H}_{r,1}^{(t)}\bar{\texTBf{F}}^{(t)}\Textbf{H}_{2,r}^{(t-1)}
\textbf{H}_{2,r}^{(t-1)^H}\bar{\TextbF{F}}^{(T)^h}\texTBf{h}_{r,1}^{(t)^H}\tExtbf{R}_1^{(t)}
(\TExTBF{R}_1^{(t)^h}\tExTbf{a}_1^{(t)}\TExTbf{R}_1^{(t)})^{-1} \}
\Big]
\End{spliT}$$ where $\textBf{A}_L^{(1)} = \sIgma_{n,r}^2\textbf{h}_{R,l}^{(1)}\Bar{\textbf{F}}^{(1)}\Bar{\Textbf{F}}^{(1)^H}\textBf{H}_{R,l}^{(1)^H}
+ \alpHa^{(1)^{-2}}P_l\bOLdsymBol{\delta}_{L,l}^{(1)}\boldSYmbol{\DElta}_{l,l}^{(1)^H} + \alPhA^{(1)^{-2}}\Sigma_{n,L}^2\Textbf{I}_{n_S}$ AnD $$\begIn{split}
\textbf{A}_l^{(t)}
&= \sIGmA_{N,r}^2\textbf{H}_{r,l}^{(t)}\baR{\textbF{F}}^{(T)}\BaR{\TExtBf{F}}^{(T)^H}\textbf{H}_{r,L}^{(t)^h}
+ \sum_{i=2}^{T} \Bigg\{ \texTBf{h}_{R,L}^{(T)}\baR{\Textbf{F}}^{(t)} \prod_{j=1}^{I-1}( \boldsymbol{\dEltA}_{r,r}^{(t-j)}\tExTbf{f}^{(T-j)})
\\
&(p_l\teXtbf{H}_{L,r}^{(T-I)}\teXtbf{H}_{l,r}^{(t-i)^H} + p_{\Bar{l}}\Textbf{H}_{\baR{l},r}^{(t-i)}\tEXtbf{H}_{\baR{L},r}^{(t-i)^H} + | ^{(t)}\textbf{R}_2^{(t)})^ {-1} \}
\b ig]
\ \
+ \m at hbb{ E}\b ig[
\text{log} _ 2 \t ext{det} \{ \textbf{I} _{N_s }+
p_2 \t extbf {R}_1^{ ( t) ^ H }\t ex tb f{H }_ { r, 1}^{( t)} \bar{\t extbf{F}}^ {(t )} \textbf{H}_{ 2 ,r }^{(t-1)}\te xtbf{H}_{2,r }^{ (t-1)^ H} \ba r {\tex tbf {F}}^ {(t)^H } \textb f{H}_{r,1 }^ { (t)^H} \ textbf{ R } _1 ^{(t )}
(\textbf{R}_1^ { (t ) ^H}\textbf{A}_ 1^{(t) }\ t ex t b f{R }_1 ^{(t)})^{- 1} \}
\ b ig]
\en d {s p l i t}$ $ where $\text bf{A}_l^{(1 ) } = \sigm a_ {n, r }^2\te xtbf{ H} _ {r, l}^{(1)}\ba r{\t extbf{F}} ^{(1)} \ bar{\te x tbf{F}} ^{(1)^ H}\ tex tbf{ H }_ {r ,l} ^{ ( 1)^ H }+ \ a lph a^{(1)^{ -2 }} p_l\b olds y m b o l{\D elt a}_{ l,l}^ {(1)}\boldsym bol {\De l ta} _{l,l }^{(1 )^H} + \alp ha^{(1 )^{-2 }} \sigma_{n,l}^2\ text bf{I}_{N_ s}$ a nd$$ \begi n {split }
\ tex tbf{A}_ l^{(t)} &=\s i g m a_ {n,r}^2\textbf{H}_ {r , l }^ {(t)}\ba r{\tex t bf {F } }^{(t)}\ ba r{\ text b f {F}}^ {(t) ^ H} \textbf{ H}_{r, l }^ {( t)^H}
+ \ sum_{i =2 }^{ t} \big g \{ \textb f{H}_{r, l}^{( t )}\bar{\textbf { F}}^{(t)} \pr o d_ { j =1 } ^{i- 1}( \boldsymbo l{\D e lta} _{r, r }^ {(t - j)}\t extbf {F } ^{ ( t-j)})
\\
&(p_l\tex tb f{H}_{ l,r}^ {(t-i)}\textb f{H}_{l,r} ^ { ( t-i)^H}+ p_ { \b a r{l}}\textbf{H }_{\b ar{l},r}^{ ( t-i)}\te xtbf{ H}_{\bar {l},r}^{( t - i)^H} + | ^{(t)}\textbf{R}_2^{(t)})^{-1} \}
\big]
\\
+_\mathbb{E}\big[
\text{log}_2 \text{det}_\{ \textbf{I}_{N_s} +
p_2 \textbf{R}_1^{(t)^H}\textbf{H}_{r,1}^{(t)}\bar{\textbf{F}}^{(t)}\textbf{H}_{2,r}^{(t-1)}
\textbf{H}_{2,r}^{(t-1)^H}\bar{\textbf{F}}^{(t)^H}\textbf{H}_{r,1}^{(t)^H}\textbf{R}_1^{(t)}
(\textbf{R}_1^{(t)^H}\textbf{A}_1^{(t)}\textbf{R}_1^{(t)})^{-1}_\}
\big]
\end{split}$$ where_$\textbf{A}_l^{(1)}_= \sigma_{n,r}^2\textbf{H}_{r,l}^{(1)}\bar{\textbf{F}}^{(1)}\bar{\textbf{F}}^{(1)^H}\textbf{H}_{r,l}^{(1)^H}
+_\alpha^{(1)^{-2}}p_l\boldsymbol{\Delta}_{l,l}^{(1)}\boldsymbol{\Delta}_{l,l}^{(1)^H}_+ \alpha^{(1)^{-2}}\sigma_{n,l}^2\textbf{I}_{N_s}$ and_$$\begin{split}
\textbf{A}_l^{(t)}
&= \sigma_{n,r}^2\textbf{H}_{r,l}^{(t)}\bar{\textbf{F}}^{(t)}\bar{\textbf{F}}^{(t)^H}\textbf{H}_{r,l}^{(t)^H}
+ \sum_{i=2}^{t}_ \bigg\{ \textbf{H}_{r,l}^{(t)}\bar{\textbf{F}}^{(t)}_\prod_{j=1}^{i-1}( \boldsymbol{\Delta}_{r,r}^{(t-j)}\textbf{F}^{(t-j)})
\\
&(p_l\textbf{H}_{l,r}^{(t-i)}\textbf{H}_{l,r}^{(t-i)^H} +_p_{\bar{l}}\textbf{H}_{\bar{l},r}^{(t-i)}\textbf{H}_{\bar{l},r}^{(t-i)^H}_+ |
users. Moreover, users can add other users as trustworthy users if they like their reviews. The Epinions dataset is provided by [@tang-etal12b], and CiaoDVD is provided by [@guo2014etaf]. The statistics of the two datasets are shown in Table \[tb-dataset-stat\]. Note that although the datasets include other SIs such as the categories and reviews of items, the only relation that exists between nodes of same type is social relation. Thus, in our experiments, we use meta-paths $\cP_1$ and $\cP_2$ in Figure \[fig-example-hin\](b) to demonstrate the effectiveness of MEMP and omit the statistics of the other SIs in Table \[tb-dataset-stat\].
**Evaluation Metrics.** We choose to tackle the rating prediction task, which is widely used to evaluate CF-based RSs. We choose two evaluation metrics, Mean Absolute Error (MAE) and Root Mean Square Error (RMSE), to evaluate our framework.
**Baselines.** We compare our proposed models with the following RS methods:
- **RegSVD** [@paterek2007improving]: It is the standard matrix factorization method with $\ell_2$ regularization. We use the implementation in an open source library LibRec [@guo2015librec].
- **SoReg** [@ma2011recommender]: It is a MF-based method employing social connections as regularization terms. We use the implementation in an open source library LibRec [@guo2015librec].
- **SocialMF** [@jamali2010matrix]: It is a MF based framework which incorporates social trust propagation. We use an open source library LibRec [@guo2015librec] in the implementation.
- **FMG** [@zhao2017meta]: It is a state-of-the-art HIN-based RS method adopting the “MF+FM” framework. We use the implementation given by the authors. For this method, we choose the meta-paths $\cP_1$ and $\cP_2$ in Figure \[fig-example-hin\](b) without incorporating motifs.
Note that there are other HIN-based RS methods for rating prediction [@yu2014personalized; @shi2015semantic], but FMG has been shown to be consistently superior [@ | users. Moreover, users can add early drug user as trustworthy users if they like their follow-up. The Epinions dataset is provide by [ @tang - etal12b ], and CiaoDVD is provided by [ @guo2014etaf ]. The statistics of the two datasets are show in Table \[tb - dataset - stat\ ]. notice that although the datasets include other SIs such as the categories and reviews of detail, the only relation that exist between node of same type is social relation. Thus, in our experiments, we practice meta - paths $ \cP_1 $ and $ \cP_2 $ in Figure \[fig - example - hin\](b) to demonstrate the potency of MEMP and omit the statistics of the other SIs in Table \[tb - dataset - stat\ ].
* * Evaluation Metrics. * * We choose to tackle the evaluation prediction task, which is widely used to evaluate CF - based r. We choose two evaluation metrics, Mean Absolute Error (MAE) and Root Mean Square Error (RMSE), to evaluate our framework.
* * Baselines. * * We compare our proposed models with the following RS methods:
- * * RegSVD * * [ @paterek2007improving ]: It is the standard matrix factorization method with $ \ell_2 $ regularization. We use the implementation in an open source library LibRec [ @guo2015librec ].
- * * SoReg * * [ @ma2011recommender ]: It is a MF - based method employing social connections as regularization terminus. We use the implementation in an candid generator library LibRec [ @guo2015librec ].
- * * SocialMF * * [ @jamali2010matrix ]: It is a MF based framework which incorporates social confidence propagation. We use an open source library LibRec [ @guo2015librec ] in the execution.
- * * FMG * * [ @zhao2017meta ]: It is a state - of - the - art HIN - based RS method adopting the “ MF+FM ” framework. We use the implementation given by the authors. For this method, we choose the meta - paths $ \cP_1 $ and $ \cP_2 $ in Figure \[fig - case - hin\](b) without incorporating motif.
Note that there are other HIN - based RS methods for military rank prediction [ @yu2014personalized; @shi2015semantic ], but FMG has been shown to be consistently superior [ @ | usfrs. Moreover, users can aad other users cw trusvworthy users iw they like their reviews. Thx Epunionw dataset is provided cy [@tang-etap12b], and CuaoDTD is provided bb [@fuo2014etaf]. The sfwtiscirs of the two dstasets ara shown in Tabne \[gb-bataset-stat\]. Note that although the dwtasets ijclude other SYs slcr as nht categories and reviews of items, the onny relation tnat exists between nodes ov sale type is social gelation. Thoa, ig our experimdnts, we ust keta-paths $\dP_1$ and $\cP_2$ in Figure \[fig-example-hiv\](b) to demonstrajz thf effectivenxss of MEMP and omln the sdatistivs of the othev SIs in Table \[tb-dataset-stat\].
**Eveluation Metrics.** We shoose to tcckle the rating preducrion jask, fhicf is wiseky used ho xvaluate CF-gased RSs. Ww choose two evaluauiog metrics, Mean Absolttq Error (MAE) and Root Mean Square Error (GMSE), to evaluate our framewirk.
**Baselines.** We compate our pro[osed models with the following RS methods:
- **RegSED** [@patxrdk2007inpvovivt]: Lt is the standard matrix factorization methoq wotm $\ell_2$ regularizabion. We use the imllfmrgtation in an open soudce library LibRec [@huo2015librgc].
- **SiReg** [@ma2011recjmmemder]: It is a MF-based method employing sjxial connections ad regularizction jerms. Ee use the implementatiun ih an open slurce libdxry LibRec [@guo2015libfec].
- **SocialMF** [@jamali2010matrix]: It if a MF bawed yramewory whoch insorporates social trust propagatioj. We ose an open sourfe library LibRec [@guo2015librec] in tix implementatoot.
- **FMG** [@zhao2017ieta]: Lt is a state-of-ehe-art HIN-baseb RS metkod adupting the “MF+FM” fcamework. We tse the implekgntation givei by the wuthirs. Dor thir method, we chopse the mvtc-paths $\cP_1$ and $\cP_2$ in Figure \[flg-exaolle-hin\](b) without inxorporating motofs.
Votq nhav thewa are other VIN-bxsea RS mdthods for vatkng lrediction [@yu2014personalhzed; @shi2015semantic], but FKG has beeb shown eo be consistrntly superior [@ | users. Moreover, users can add other users users they like reviews. The Epinions and is provided by The statistics of two datasets are shown in Table Note that although the datasets include other SIs such as the categories and of items, the only relation that exists between nodes of same type is relation. in experiments, use meta-paths $\cP_1$ and $\cP_2$ in Figure \[fig-example-hin\](b) to demonstrate the effectiveness of MEMP and omit statistics of the other SIs in Table \[tb-dataset-stat\]. Metrics.** We choose to the rating prediction task, which widely to evaluate RSs. choose evaluation metrics, Mean Error (MAE) and Root Mean Square Error (RMSE), to evaluate our framework. **Baselines.** We compare our proposed with the methods: - [@paterek2007improving]: is standard matrix factorization $\ell_2$ regularization. We use the implementation source library LibRec [@guo2015librec]. - **SoReg** [@ma2011recommender]: It a MF-based employing social connections as regularization terms. use the implementation in an open source library [@guo2015librec]. - **SocialMF** [@jamali2010matrix]: It is a MF based framework which incorporates social trust propagation. an open source library [@guo2015librec] in the - [@zhao2017meta]: is state-of-the-art HIN-based method adopting the “MF+FM” framework. We use the implementation given by authors. For this method, we choose the meta-paths $\cP_1$ and Figure without incorporating motifs. that there are other RS for rating prediction [@yu2014personalized; FMG been consistently [@ | users. Moreover, users can add oTher users aS trusTwoRthY uSers If thEy like their revIEws. THe Epinions dataset is proVided By [@TAng-eTAl12B], and CIaoDVD iS PrOVIdeD bY [@gUo2014eTaF]. thE statIstIcs of thE two dataseTs aRe Shown in Table \[TB-dAtaset-stat\]. notE that althougH thE datasEtS inCLude oTheR SIs sUch as tHE categOries and rEvIEws of iTEms, the oNLY rElatIon that exists betwEEn NOdes of same type Is sociAl RElATIon. thuS, in our expeRiMents, WE use metA-PaTHS $\CP_1$ aND $\cP_2$ in Figure \[fiG-example-hin\](B) To dEmonstRaTe tHE effecTivenEsS Of MeMP and omit tHe stAtistics oF the otHEr SIs in tAble \[tb-dAtaset-StaT\].
**EvAluaTIoN MEtrIcS.** we cHOoSe tO TacKle the raTiNg PrediCtioN TASK, whiCh iS widEly usEd to evaluate Cf-baSed RsS. We ChoosE two eValuAtIon meTrics, MEan AbSoLute Error (MAE) and root mean SquarE ErRoR (RMsE), To evaLUate ouR frAmeWork.
**BasElines.** WE ComPaRE OUr Proposed models with ThE FOlLowing RS MethodS:
- **reGSvd** [@paterek2007ImProVing]: iT Is the StanDArD matrix fActoriZAtIoN method WiTh $\ell_2$ rEgUlaRizAtion. wE use The impLementatIon in AN open source libRAry LibRec [@guo2015lIBrEC].
- **sorEg** [@ma2011RecOmmender]: It iS a MF-BAsed MethOD eMplOYing sOcial CoNNeCTions as regularizatiOn Terms. WE use tHe implementatIon in an opeN SOUrce librAry LIBREC [@guo2015librec].
- **SociAlMF** [@jAmali2010matriX]: it is a MF bAsed fRamework Which incoRPOrates soCiaL trUst ProPAGaTion. We use an opEN SourCe Library librec [@guo2015lIbrEc] iN thE imPlEmentatioN.
- **FMG** [@zhao2017MeTa]: it Is A stAte-of-THe-art HIN-BaSed rS MetHod adOPting tHe “MF+Fm” fraMeWoRK. We Use the iMPlEMEntaTiOn GiveN by ThE authOrs. FOR thIs methoD, we choose The MEta-pAtHs $\CP_1$ and $\cP_2$ In Figure \[fig-exAmPle-hin\](b) witHoUt iNcorpoRATing motiFs.
Note that there are other hiN-based rS mEthodS for Rating preDicTion [@yu2014PerSOnalizEd; @shi2015sEmantIc], But fmg has bEEN sHowN tO be consistENTly SuperIoR [@ | users. Moreover, users ca n add othe r use rsastr ustw orth y users if the y lik e their reviews. The E pinio ns data s et is p rovided by [ @ta ng -e tal 12 b ], andCia oDVD is providedby[@ guo2014etaf] . T he statist ics of the twodat asetsar e s h own i n T able\[tb-d a taset- stat\]. N ot e thata lthough t he dat asets include oth e rS Is such as the categ or i es a ndrev iews of it em s, th e only r e la t i o n t h at exists bet ween nodeso f s ame ty pe is social rela ti o n.Thus, in ou r ex periments , we u s e meta- p aths $\ cP_1$and $\ cP_2 $ i nFig ur e \[ f ig -ex a mpl e-hin\]( b) t o dem onst r a t e the ef fect ivene ss of MEMP an d o mitt hestati stics ofth e oth er SIs in T ab le \[tb-dataset -sta t\].
**E val ua tio nMetri c s.** W e c hoo se to t ackle t h e r at i n g p rediction task, wh ic h is widelyused t o e va l uate CF- ba sed RSs . We ch oose tw o evalua tion m e tr ic s, Mean A bsolut eErr or(MAE) andRoot M ean Squa re Er r or (RMSE), toe valuate our f r am e w or k .
* *Ba selines.**We c o mpar e ou r p rop o sed m odels w i th the following RS me th ods:
- * *RegSVD** [@p aterek2007 i m p roving]: Iti st he standard ma trixfactorizat i on metho d wit h $\ell_ 2$ regula r i zation.Weuse th e i m p le mentation ina n ope nsourcelib rary Li bRe c [ @gu o20 15 librec].
- **S oR eg ** [ @ma 2011r e commende r] : I tisa MF- b ased m ethod emp lo yi n g s ocial c o nn e c tion sas reg ula ri zatio n te r ms. We use the impl eme n tati on i n an op en source lib ra ry LibRec[@ guo 2015li b r ec].
- **SocialMF** [@jamali 2 010matr ix] : Itis a MF based fr amewor k w h ich in corpor atesso cia l trust p ro pag at ion. We us e anopenso urce librar y LibRec [@guo2015 l ibr ec] in the im ple ment a t io n.- ** FM G **[ @ zhao2017meta]:It is a st at e -o f-the-artH IN- ba sed RSmethodadopt i ng the“MF+FM” f ramework. W e us e the implement ation gi ven by th e auth o rs . For th is met ho d,we ch oose t h e m eta-p aths $ \c P_1$ a nd $\ cP _2$ in F igure \[fig-example-hin \](b)witho utincorpora tin g mo tifs.
No te t hat thereare ot her H IN- b asedRS m e th ods for r atin g predicti o n[@y u 2 01 4personaliz e d ; @s hi201 5se m antic] , bu t FMG has been sh o wn to be consi sten t l y s upe r ior[@ | users._Moreover, users_can add other users_as trustworthy_users_if they_like_their reviews. The_Epinions dataset is_provided by [@tang-etal12b], and CiaoDVD_is provided by [@guo2014etaf]._The_statistics of the two datasets are shown in Table \[tb-dataset-stat\]. Note that although the datasets_include_other SIs_such_as_the categories and reviews of_items, the only relation that_exists between_nodes of same type is social relation. Thus,_in_our experiments, we_use meta-paths $\cP_1$ and $\cP_2$ in Figure \[fig-example-hin\](b) to demonstrate_the effectiveness of MEMP and omit_the statistics of_the_other_SIs in Table \[tb-dataset-stat\].
**Evaluation Metrics.**_We choose to tackle the rating_prediction task, which is widely used_to evaluate CF-based RSs. We choose two_evaluation metrics, Mean Absolute Error (MAE)_and Root Mean Square Error_(RMSE), to_evaluate our framework.
**Baselines.** We compare_our proposed models_with the_following RS methods:
-_ **RegSVD** [@paterek2007improving]: It is the_standard matrix factorization_method with $\ell_2$ regularization. We use_the_implementation in an_open_source_library LibRec [@guo2015librec].
-_ **SoReg** [@ma2011recommender]:_It_is a_MF-based_method employing social connections as regularization_terms._We use the implementation in an open_source library LibRec [@guo2015librec].
- __**SocialMF** [@jamali2010matrix]: It is a_MF based framework which incorporates_social trust propagation. We use an_open source_library LibRec [@guo2015librec]_in the implementation.
- **FMG** [@zhao2017meta]: It is a state-of-the-art HIN-based_RS method adopting the “MF+FM” framework._We use the implementation_given by_the_authors. For this_method,_we choose_the meta-paths $\cP_1$ and $\cP_2$ in Figure \[fig-example-hin\](b)_without incorporating_motifs.
Note that there are other HIN-based_RS methods for rating_prediction [@yu2014personalized;_@shi2015semantic], but FMG has been shown_to be consistently superior [@ |
where $d$ is the dimension of the underlying Hilbert space in which $\rho$ acts, and the reduced density matrix of a given subsystem $\alpha_k$ is obtained by tracing out all the other subsystems $\rho^{\alpha_j} = \mbox{Tr}_{\{\alpha_k \} \neq \alpha_j} [\rho]$. In the above problem, the subsystems considered correspond to spin and parity, $S$ and $P$, for particles $A$ and $B$, i.e. $\{\alpha_k\} \equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. In particular, the more the subsystems of a given state are mixed, the more entanglement is encoded among them: the global measure, $E_{G}[\rho]$, captures a picture of the quantum correlations distributed among the four DoF’s here involved..
The linear entropy of a reduced subsystem $\rho^{\alpha_k}$ of (\[generalstate\]), which is a two-qubit state, is evaluated in terms of the components of its Bloch vector $a^{\alpha_k}_n = \mbox{Tr}[\hat{\sigma}_n^{\alpha_k} \rho^{\alpha_k}]$ as $$\label{globalexpr0}
E_L[\rho^{\alpha_k}]= 1 - \sum_{n = \{x, \, y, \, z\}} (a^{\alpha_k}_n)^2,$$ and the global measure from Eq. (\[globalent\]) can be simplified into $$\label{globalexpr}
E_{G}[\rho] =1 - \frac{1}{4} \, \displaystyle \sum_{\alpha = \{\alpha_k\}} \sum_{n = \{x, \, y, \, z\}} (a^{\alpha}_n)^2,$$ with $ \{\alpha_k\}\equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. The Bloch vectors of the subsystems of $A$ are explicitly given by $$\begin{aligned}
\label{blochvecs}
a^{(S)A}_n &=& \frac{1}{N}\displaystyle \sum_{i, j}^M\, c_i c_j^* \, \mathcal{M}_{r_i r_j}\bb{\bm{q}} \, \frac{1}{ E_p} ( \, E_p \delta_{s_ | where $ d$ is the dimension of the underlying Hilbert space in which $ \rho$ acts, and the dilute concentration matrix of a given subsystem $ \alpha_k$ is obtained by trace out all the early subsystems $ \rho^{\alpha_j } = \mbox{Tr}_{\{\alpha_k \ } \neq \alpha_j } [ \rho]$. In the above problem, the subsystems consider correspond to spin and parity bit, $ S$ and $ P$, for particle $ A$ and $ B$, i.e. $ \{\alpha_k\ } \equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. In particular, the more the subsystem of a given state are mixed, the more entanglement is encode among them: the global measure, $ E_{G}[\rho]$, get a picture of the quantum correlations distribute among the four DoF ’s here involved..
The linear entropy of a reduce subsystem $ \rho^{\alpha_k}$ of (\[generalstate\ ]), which is a two - qubit state, is evaluated in terms of the components of its Bloch vector $ a^{\alpha_k}_n = \mbox{Tr}[\hat{\sigma}_n^{\alpha_k } \rho^{\alpha_k}]$ as $ $ \label{globalexpr0 }
E_L[\rho^{\alpha_k}]= 1 - \sum_{n = \{x, \, y, \, z\ } } (a^{\alpha_k}_n)^2,$$ and the ball-shaped measure from Eq. (\[globalent\ ]) can be simplify into $ $ \label{globalexpr }
E_{G}[\rho ] = 1 - \frac{1}{4 } \, \displaystyle \sum_{\alpha = \{\alpha_k\ } } \sum_{n = \{x, \, y, \, z\ } } (a^{\alpha}_n)^2,$$ with $ \{\alpha_k\}\equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. The Bloch vectors of the subsystems of $ A$ are explicitly given by $ $ \begin{aligned }
\label{blochvecs }
a^{(S)A}_n & = & \frac{1}{N}\displaystyle \sum_{i, j}^M\, c_i c_j^ * \, \mathcal{M}_{r_i r_j}\bb{\bm{q } } \, \frac{1 } { E_p } (\, E_p \delta_{s _ | whfre $d$ is the dimension on the underlying Hilberv space in whicf $\rho$ acts, and the reduced dxnsiry maugix of a given subsysgem $\alpha_n$ is obtqinev by tracing out all the other dubsvsvems $\rho^{\alpha_j} = \mbox{Tr}_{\{\alpva_k \} \neq \alpha_b} [\fhl]$. In the above problem, the subsysteis consodfred corresponq to fpin and parity, $S$ and $P$, for particles $Z$ and $B$, i.e. $\{\alpha_k\} \eqiiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. In particular, the more the subsystels of a givgh seqte are mixea, the more entanglemenj is encoded among them: the globxl mecsure, $E_{G}[\rho]$, cqptktes a picturx of tre quantum correlatimns disyributed among thx foyr DoF’s here involved..
Vhe linear entropy os a reducad subsystem $\rho^{\alpya_j}$ of (\[cenesalsgqte\]), whjci ia a twl-quuit state, ia evaluated in terms of the cokpjbents of its Gloch destor $a^{\alpha_k}_n = \mbox{Tr}[\hat{\sigma}_n^{\alpha_k} \rhm^{\allha_k}]$ as $$\label{globalexpr0}
W_L[\rho^{\alpha_k}]= 1 - \sum_{n = \{x, \, y, \, z\}} (a^{\ajpha_k}_n)^2,$$ and the global measure from Eq. (\[globalent\]) cat be akmpoinied unho $$\label{globalexpr}
E_{G}[\rho] =1 - \frac{1}{4} \, \displaystyle \fhm_{\sliha = \{\alpha_k\}} \sum_{n = \{x, \, y, \, z\}} (a^{\akpja}_m)^2,$$ with $ \{\alpha_k\}\gquiv \{(S)C,\,(A)B,\,(L)A,\,(P)B\}$. The Bloch vechors of the wubsystemf of $A$ are explicitly given by $$\vegin{aligned}
\jqbel{blochvecs}
a^{(S)A}_n &=& \frac{1}{N}\displcystylg \sum_{i, j}^M\, c_i c_j^* \, \mathcal{M}_{r_i r_l}\bb{\bm{q}} \, \frac{1}{ E_p} ( \, E_p \delta_{a_ | where $d$ is the dimension of the space which $\rho$ and the reduced subsystem is obtained by out all the subsystems $\rho^{\alpha_j} = \mbox{Tr}_{\{\alpha_k \} \neq [\rho]$. In the above problem, the subsystems considered correspond to spin and parity, and $P$, for particles $A$ and $B$, i.e. $\{\alpha_k\} \equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. In particular, more subsystems a state are mixed, the more entanglement is encoded among them: the global measure, $E_{G}[\rho]$, captures a of the quantum correlations distributed among the four here involved.. The linear of a reduced subsystem $\rho^{\alpha_k}$ (\[generalstate\]), is a state, evaluated terms of the of its Bloch vector $a^{\alpha_k}_n = \mbox{Tr}[\hat{\sigma}_n^{\alpha_k} \rho^{\alpha_k}]$ as $$\label{globalexpr0} E_L[\rho^{\alpha_k}]= 1 - \sum_{n = \{x, \, \, z\}} the global from (\[globalent\]) be simplified into =1 - \frac{1}{4} \, \displaystyle \sum_{\alpha = \{x, \, y, \, z\}} (a^{\alpha}_n)^2,$$ with \{\alpha_k\}\equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. Bloch vectors of the subsystems of are explicitly given by $$\begin{aligned} \label{blochvecs} a^{(S)A}_n &=& \sum_{i, j}^M\, c_i c_j^* \, \mathcal{M}_{r_i r_j}\bb{\bm{q}} \, \frac{1}{ E_p} ( \, E_p \delta_{s_ | where $d$ is the dimension of the Underlying hilbeRt sPacE iN whiCh $\rhO$ acts, and the redUCed dEnsity matrix of a given suBsystEm $\ALpha_K$ Is ObtaiNed by trACiNG Out AlL tHe oThER sUbsysTemS $\rho^{\alpHa_j} = \mbox{Tr}_{\{\aLphA_k \} \Neq \alpha_j} [\rho]$. iN tHe above proBleM, the subsysteMs cOnsideReD coRRespoNd tO spin And parITy, $S$ and $p$, for partiClES $A$ and $B$, I.E. $\{\alpha_k\} \EQUiV \{(S)A,\,(S)b,\,(P)A,\,(P)B\}$. In particular, THe MOre the subsysteMs of a gIvEN sTATe aRe mIxed, the morE eNtangLEment is ENcODED amONg them: the globAl measure, $E_{G}[\RHo]$, cAptureS a PicTUre of tHe quaNtUM coRrelations dIstrIbuted amoNg the fOUr DoF’s hERe involVed..
The LinEar EntrOPy Of A reDuCEd sUBsYstEM $\rhO^{\alpha_k}$ oF (\[gEnEralsTate\]), WHICH is a Two-QubiT statE, is evaluated iN teRms oF The CompoNents Of itS BLoch vEctor $a^{\Alpha_K}_n = \Mbox{Tr}[\hat{\sigma}_n^{\AlphA_k} \rho^{\alphA_k}]$ aS $$\lAbeL{gLobalEXpr0}
E_L[\rHo^{\aLphA_k}]= 1 - \sum_{n = \{x, \, Y, \, z\}} (a^{\alphA_K}_n)^2,$$ aNd THE GlObal measure from Eq. (\[gLoBALeNt\]) can be sImplifIEd InTO $$\label{glObAleXpr}
E_{g}[\RHo] =1 - \fraC{1}{4} \, \disPLaYstyle \suM_{\alpha = \{\ALpHa_K\}} \sum_{n = \{x, \, y, \, Z\}} (a^{\Alpha}_n)^2,$$ WiTh $ \{\aLphA_k\}\equIV \{(S)A,\,(S)b,\,(P)A,\,(P)B\}$. THe Bloch vEctorS Of the subsystemS Of $A$ are expliciTLy GIVeN By $$\beGin{Aligned}
\labeL{bloCHvecS}
a^{(S)A}_N &=& \FrAc{1}{N}\DIsplaYstylE \sUM_{i, J}^m\, c_i c_j^* \, \mathcal{M}_{r_i r_j}\bb{\Bm{Q}} \, \frac{1}{ E_P} ( \, E_p \deLta_{s_ | where $d$ is the dimensio n of the u nderl yin g H il bert spa ce in which $\ r ho$acts, and the reduceddensi ty matr i xof agiven s u bs y s tem $ \a lph a_ k $is ob tai ned bytracing ou t a ll the other s u bs ystems $\r ho^ {\alpha_j} = \m box{Tr }_ {\{ \ alpha _k\} \n eq \al p ha_j}[\rho]$.In the ab o ve prob l e m, the subsystems consi d er e d correspond t o spin a n dp a rit y,$S$ and $P $, forp article s $ A $ and $B$, i.e. $\{ \alpha_k\}\ equ iv \{( S) A,\ , (S)B,\ ,(P)A ,\ , (P) B\}$. In pa rtic ular, the moret he subs y stems o f a gi ven st atea re m ixe d, the mo ree nta nglement i sencod ed a m o n g the m:thegloba l measure, $E _{G }[\r h o]$ , cap tures a p ic tureof the quan tu m correlationsdist ributed a mon gthe f our D o F’s he reinv olved..
The l i nea re n t ro py of a reduced su bs y s te m $\rho^ {\alph a _k }$ of (\[ge ne ral stat e \ ]), w hich is a two-q ubit s t at e, is eva lu ated i nter msof th e com ponent s of its Bloc h vector $a^{\a l pha_k}_n = \m b ox { T r} [ \hat {\s igma}_n^{\a lpha _ k} \ rho^ { \a lph a _k}]$ as $ $\ l ab e l{globalexpr0}
E_L[ \r ho^{\a lpha_ k}]= 1 - \sum _{n = \{x, \ , y, \, z \}}( a^ { \alpha_k}_n)^2 ,$$ a nd the glo b al measu re fr om Eq. ( \[globale n t \]) canbesim pli fie d in to $$\label{g l o bale xp r}
E_{G }[\ rho] =1 -\fr ac{ 1}{ 4} \, \disp laystyle \ su m_ {\ alp ha =\ {\alpha_ k\ }}\s um_ {n =\ {x, \, y, \ , z\ }} ( a ^{\ alpha}_ n )^ 2 , $$ w it h$ \{ \al ph a_k\} \equ i v \ {(S)A,\ ,(S)B,\,( P)A , \,(P )B \} $. TheBloch vectors o f the subs ys tem s of $ A $ are exp licitly given by $$\beg i n{align ed}
\lab el{b lochvecs}
a^ {(S)A} _n& =& \fr ac{1}{ N}\di sp lay s t yle \ s u m_ {i, j }^M\, c_ic _ j^* \, \ ma thca l{M}_{r _i r_j}\bb{\bm{q}} \,\frac{1}{ E_p } ( \,E _ p\de l ta _ {s_ | where_$d$ is_the dimension of the_underlying Hilbert_space_in which_$\rho$_acts, and the_reduced density matrix_of a given subsystem_$\alpha_k$ is obtained_by_tracing out all the other subsystems $\rho^{\alpha_j} = \mbox{Tr}_{\{\alpha_k \} \neq \alpha_j} [\rho]$. In_the_above problem,_the_subsystems_considered correspond to spin and_parity, $S$ and $P$, for_particles $A$_and $B$, i.e. $\{\alpha_k\} \equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. In particular,_the_more the subsystems_of a given state are mixed, the more entanglement_is encoded among them: the global_measure, $E_{G}[\rho]$, captures_a_picture_of the quantum correlations_distributed among the four DoF’s here_involved..
The linear entropy of a reduced_subsystem $\rho^{\alpha_k}$ of (\[generalstate\]), which is a_two-qubit state, is evaluated in terms_of the components of its_Bloch vector_$a^{\alpha_k}_n = \mbox{Tr}[\hat{\sigma}_n^{\alpha_k} \rho^{\alpha_k}]$ as_$$\label{globalexpr0}
E_L[\rho^{\alpha_k}]= 1 -_\sum_{n =_\{x, \, y,_\, z\}} (a^{\alpha_k}_n)^2,$$ and the global_measure from Eq. (\[globalent\])_can be simplified into $$\label{globalexpr}
E_{G}[\rho] =1_-_\frac{1}{4} \, \displaystyle_\sum_{\alpha_=_\{\alpha_k\}} \sum_{n_= \{x, \,_y,_\, z\}}_(a^{\alpha}_n)^2,$$_with $ \{\alpha_k\}\equiv \{(S)A,\,(S)B,\,(P)A,\,(P)B\}$. The Bloch_vectors_of the subsystems of $A$ are explicitly_given by $$\begin{aligned}
\label{blochvecs}
a^{(S)A}_n &=&_\frac{1}{N}\displaystyle_\sum_{i, j}^M\, c_i c_j^*_\, \mathcal{M}_{r_i r_j}\bb{\bm{q}} \, \frac{1}{_E_p} ( \, E_p \delta_{s_ |
ovae II. The selection effect and the frequencies per unit blue luminosity. A&A 273,383
Colgate S.A., McKee C. (1969) Early Supernova Luminosity. ApJ 157, 623
Della Valle M., Kissler-Patig M., Danziger J., Storm J. (1998) Globular cluster calibration of the peak brightness of the type Ia supernova 1992A and the value of $H_0$. MNRAS 299,267
Drenkhahn G., Richtler T. (1999) SN1994D in NGC4526: a normally bright type Ia supernova. A&A, in press
Filippenko A.V. (1997) Optical Spectra of Supernovae. ARAA 35,309
Gieren W.P., Richtler T., Hilker M. (1994) The distance to the Large Magellanic Cloud cluster NGC1866 from its Cepheid members. ApJ 433,L73
Gieren W.P., Fouqué P., Gomez M. (1998) Cepheid Period-Radius and Period-Luminosity Relations and the Distance to the Large Magellanic Cloud. ApJ 496,17
Gómez M., Richtler T., Infante L., Drenkhahn G., (1999) The peculiar globular cluster system of NGC 1316 (Fornax A). AGM 15,P62
Grillmair C.J., Forbes D.A., Brodie J.P., Elson R.A.W. (1999) HST imaging of the globular clusters in the Fornax cluster: color and luminosity distribution. AJ 117,167
Hamuy M., Phillips M.M., Maza J., Wischnjewsky M., Uomoto A., Landolt A.U., Khatwani R. (1991) The optical light curves of SN1980D and SN1981N in NGC1316 (Fornax A). AJ 102,208
Hamuy M., Phillips M.M., Suntzeff N.B., Schommer R.A., Maza J., Aviles R., (1996) The Hubble diagram of the Calán/Tololo type Ia supernovae and the value of $H_0$. AJ 112,2398
| ovae II. The selection effect and the frequencies per unit aristocratic luminosity. A&A 273,383
Colgate S.A., McKee C. (1969) Early Supernova Luminosity. ApJ 157, 623
Della Valle M., Kissler - Patig M., Danziger J., Storm J. (1998) Globular bunch calibration of the peak brightness of the character Ia supernova 1992A and the value of $ H_0$. MNRAS 299,267
Drenkhahn G., Richtler T. (1999) SN1994D in NGC4526: a normally bright character Ia supernova. A&A, in press
Filippenko A.V. (1997) Optical Spectra of Supernovae. ARAA 35,309
Gieren W.P., Richtler T., Hilker M. (1994) The distance to the Large Magellanic Cloud bunch NGC1866 from its Cepheid members. ApJ 433,L73
Gieren W.P., Fouqué P., Gomez M. (1998) Cepheid Period - Radius and Period - Luminosity Relations and the Distance to the Large Magellanic Cloud. ApJ 496,17
Gómez M., Richtler T., Infante L., Drenkhahn G., (1999) The peculiar globular bunch arrangement of NGC 1316 (Fornax A). AGM 15,P62
Grillmair C.J., Forbes D.A., Brodie J.P., Elson R.A.W. (1999) HST imaging of the ball-shaped clusters in the Fornax cluster: color and luminosity distribution. AJ 117,167
Hamuy M., Phillips M.M., Maza J., Wischnjewsky M., Uomoto A., Landolt A.U., Khatwani R. (1991) The optical inner light curves of SN1980D and SN1981N in NGC1316 (Fornax A). AJ 102,208
Hamuy M., Phillips M.M., Suntzeff N.B., Schommer R.A., Maza J., Aviles R., (1996) The Hubble diagram of the Calán / Tololo type Ia supernovae and the value of $ H_0$. AJ 112,2398 | ovaf II. The selection effecu and the frequeneues pec unit glue lumknosity. A&A 273,383
Colgate S.A., McKee C. (1969) Warly Supernova Luminosity. XpJ 157, 623
Dellw Valle M., Kiswoer-Patig M., Danzigev J., Sflrm L. (1998) Globular clusjer calibrathon of the pean crnghtness of the type Ia supernova 1992A wnd the vwlue of $H_0$. MNRAF 299,267
Dremhhahh G., Richtler T. (1999) SN1994D in NGC4526: a normamly brijht type Ia suprrnova. A&A, in press
Filippenno A.G. (1997) Optical Spectra of Supernocae. WEAA 35,309
Gieren W.P., Richtler U., Kilker M. (1994) Tge distance to the Large Magellxnic Eloud clustgx NGF1866 from its Cxpheid members. ApJ 433,L73
Gieren F.P., Fouqié P., Gomez M. (1998) Ccpheiv Peeiod-Radius and Period-Numinosity Relatiogs and tha Bistance to the Large Mqgelldnic Cloje. AoJ 496,17
Gójev M., Richtper T., Infante L., Drenkhahb G., (1999) The peculiar gkoflkar cluster aystem os NGC 1316 (Fornax A). AGM 15,P62
Grillmair C.J., Forbes D.A., Brodie J.P., Elson R.A.W. (1999) HWT imaging of the glohular clufters in the Fornax cluster: color and luminosity gistrmbjtiin. XH 117,167
Hwmuy M., Phillips M.M., Maza J., Wischnjewsky M., Uomoek S., Kandolt A.U., Khabwani R. (1991) The opticsl lodht curves of SN1980D aus AN1981N in NGC1316 (Fornax W). AJ 102,208
Hamoy M., Pyillips M.I., Sumtzeff N.B., Schommer R.A., Maza H., Aviles R., (1996) Nhe Yubble diagram of che Calán/Tolouo tipe Ia supernovae and the valbe of $G_0$. AJ 112,2398
| ovae II. The selection effect and the unit luminosity. A&A Colgate S.A., McKee ApJ 623 Della Valle Kissler-Patig M., Danziger Storm J. (1998) Globular cluster calibration the peak brightness of the type Ia supernova 1992A and the value of MNRAS 299,267 Drenkhahn G., Richtler T. (1999) SN1994D in NGC4526: a normally bright Ia A&A, press A.V. (1997) Optical Spectra of Supernovae. ARAA 35,309 Gieren W.P., Richtler T., Hilker M. (1994) The to the Large Magellanic Cloud cluster NGC1866 from Cepheid members. ApJ 433,L73 W.P., Fouqué P., Gomez M. Cepheid and Period-Luminosity and Distance the Large Magellanic ApJ 496,17 Gómez M., Richtler T., Infante L., Drenkhahn G., (1999) The peculiar globular cluster system of 1316 (Fornax 15,P62 Grillmair Forbes Brodie Elson R.A.W. (1999) of the globular clusters in the and luminosity distribution. AJ 117,167 Hamuy M., Phillips Maza J., M., Uomoto A., Landolt A.U., Khatwani (1991) The optical light curves of SN1980D and in NGC1316 (Fornax A). AJ 102,208 Hamuy M., Phillips M.M., Suntzeff N.B., Schommer R.A., Maza R., (1996) The Hubble of the Calán/Tololo Ia and value $H_0$. AJ | ovae II. The selection effect aNd the frequEncieS peR unIt Blue LumiNosity. A&A 273,383
ColgatE s.A., Mckee C. (1969) Early Supernova LumiNositY. APj 157, 623
DelLA VAlle M., kissler-pAtIG m., DaNzIgEr J., stORm j. (1998) GlobUlaR clusteR calibratiOn oF tHe peak brightNEsS of the type ia sUpernova 1992A and The Value oF $H_0$. mNRas 299,267
DrenKhaHn G., RiChtler t. (1999) sN1994D in NgC4526: a normalLy BRight tYPe Ia supERNoVa. A&A, In press
Filippenko a.v. (1997) OPTical Spectra of supernOvAE. Araa 35,309
GiEreN W.P., RichtleR T., hilkeR m. (1994) The disTAnCE TO thE large MagellanIc Cloud clusTEr NgC1866 from ItS CePHeid meMbers. apj 433,l73
GiEren W.P., FouquÉ P., GoMez M. (1998) CepheId PeriOD-Radius ANd PerioD-LuminOsiTy RElatIOnS aNd tHe dIstANcE to THe LArge MageLlAnIc CloUd. Apj 496,17
gÓMEz M., RIchTler t., InfaNte L., Drenkhahn g., (1999) ThE pecULiaR globUlar cLustEr SysteM of NGC 1316 (fornaX A). aGM 15,P62
Grillmair C.J., forbEs D.A., BrodiE J.P., elSon r.A.w. (1999) HST iMAging oF thE glObular cLusters IN thE FORNAx Cluster: color and lumInOSItY distribUtion. Aj 117,167
haMuY m., PhillipS M.m., MaZa J., WISChnjeWsky m., uoMoto A., LanDolt A.U., kHaTwAni R. (1991) The OpTical lIgHt cUrvEs of Sn1980d and sN1981N in NgC1316 (Fornax a). AJ 102,208
HaMUy M., Phillips M.M., SUNtzeff N.B., SchomMEr r.a., maZA J., AvIleS R., (1996) The Hubble DiagRAm of The CALáN/ToLOlo tyPe Ia sUpERnOVae and the value of $H_0$. AJ 112,2398
| ovae II. The selection eff ect and th e fre que nci es per uni t blue luminos i ty.A&A 273,383
Colgate S .A.,Mc K ee C . ( 1969) EarlyS up e r nov aLu min os i ty . ApJ 15 7, 623
Della Val leM. , Kissler-Pa t ig M., Danzi ger J., Storm J . ( 1998)Gl obu l ar cl ust er ca librat i on ofthe peakbr i ghtnes s of the t yp e Ia supernova 1992Aa nd the value of $ H_0$.MN R AS 2 99, 267
Drenkhah nG., R i chtlerT .( 1 9 99) SN1994D in NG C4526: a no r mal ly bri gh t t y pe Iasuper no v a.A&A, in pre ss
Filippenk o A.V. (1997)O pticalSpectr a o f S uper n ov ae . A RA A 35 , 30 9
G ier en W.P., R ic htler T., H i l kerM.(199 4) Th e distance to th e La r geMagel lanic Clo ud clus ter NG C1866 f rom its Cepheid mem bers. ApJ 43 3, L73
Giere n W.P., Fo uqu é P., G omez M. (19 98 ) C ep heid Period-Radius a n d P eriod-Lu minosi t yRe l ations a nd th e Di s t anceto t h eLarge Ma gellan i cCl oud. Ap J496,17
Góm ezM., R i chtl er T., Infante L.,D renkhahn G., ( 1 999) The pecu l ia r gl o bula r c luster syst em o f NGC 131 6 ( For n ax A) . AGM 1 5 ,P 6 2
Grillmair C.J.,Fo rbes D .A.,Brodie J.P.,Elson R.A. W . (1999) H ST i m ag i ng of the glob ularclusters i n the For nax c luster:color and l uminosit y d ist rib uti o n .AJ 117,167
H a m uy M ., Philli psM.M., M aza J. , W isc hn jewsky M. , Uomoto A ., L an dol t A.U . , Khatwa ni R. ( 199 1) Th e optic al li ghtcu rv e s o f SN198 0 Da n d SN 19 81 N in NG C1 316 ( Forn a x A ). AJ 1 02,208
H amu y M., P hi llips M .M., Suntzeff N .B., Schom me r R .A., M a z a J., Av iles R., (1996) The Hub b le diag ram of t he C alán/Tolo lotype I a s u pernov ae and theva lue o f $H_ 0 $ .AJ11 2,2398
| ovae II._The selection_effect and the frequencies_per unit_blue_luminosity. A&A 273,383
Colgate_S.A.,_McKee C. (1969)_Early Supernova Luminosity._ApJ 157, 623
Della Valle M.,_Kissler-Patig M., Danziger_J.,_Storm J. (1998) Globular cluster calibration of the peak brightness of the type Ia_supernova_1992A and_the_value_of $H_0$. MNRAS 299,267
Drenkhahn G., Richtler_T. (1999) SN1994D in NGC4526:_a normally_bright type Ia supernova. A&A, in press
Filippenko A.V._(1997)_Optical Spectra of_Supernovae. ARAA 35,309
Gieren W.P., Richtler T., Hilker M. (1994) The_distance to the Large Magellanic Cloud_cluster NGC1866 from_its_Cepheid_members. ApJ 433,L73
Gieren W.P., Fouqué_P., Gomez M. (1998) Cepheid Period-Radius_and Period-Luminosity Relations and the Distance_to the Large Magellanic Cloud. ApJ 496,17
Gómez M.,_Richtler T., Infante L., Drenkhahn G.,_(1999) The peculiar globular cluster_system of_NGC 1316 (Fornax A). AGM 15,P62
Grillmair_C.J., Forbes D.A.,_Brodie J.P.,_Elson R.A.W. (1999)_HST imaging of the globular clusters_in the Fornax_cluster: color and luminosity distribution. AJ 117,167
Hamuy_M.,_Phillips M.M., Maza_J.,_Wischnjewsky_M., Uomoto_A., Landolt A.U.,_Khatwani_R. (1991)_The_optical light curves of SN1980D and_SN1981N_in NGC1316 (Fornax A). AJ 102,208
Hamuy M., Phillips_M.M., Suntzeff N.B., Schommer_R.A.,_Maza J., Aviles R.,_(1996) The Hubble diagram of_the Calán/Tololo type Ia supernovae and_the value_of $H_0$._AJ 112,2398
|
mathrm{0} + \frac{3}{2} t^\frac{2}{3}$. If we choose, as we did in the case of the Mobius strip example, to enforce continuity between points of the space-time manifold covered by different choices of tetrad, and further to conserve the momentum of the photon as measured in this way (respecting the signed volume of space), then there is a unique geodesic on the other side of the singularity which connects to this point. This is given by $x=x_o + \frac{3}{2} |t|^\frac{2}{3}$, with the tetrad given in equation (\[extended\]). We note that the conditions of the Hawking–Penrose theorem are satisfied, and hence the conclusion that there is infinite focusing of geodesics is valid. The distance between any two photons as measured using the space-time geometry will tend to zero as we approach the singularity. However, there is still a unique continuation of the paths beyond this point, which corresponds to the above trajectory as described in the geometry in which we swap tetrads. What the singularity theorems require is that we model physics in terms of a space-time defined by a metric on a globally orientable manifold. However, if we relax that assumption just on the singular surface itself, we see that there is a natural physical continuation beyond this point when described in relational terms.
To be clear about this continuation, we note that the Hawking–Penrose theorems show that there is not a unique continuation through the singularity of the space-time geometry as determined by the Einstein equations. Our results are completely in agreement with this. We do not claim that the Einstein equations give such a continuation. However, the relational system of scalar fields can be continued beyond this point, and at all points away from the singularity their behaviour is consistent with that described by the Einstein equations coupled to the matter. Thus we know that on both sides of the singularity there is a consistent space-time picture. What we have then shown is that if we want to further identify trajectories of matter across the singularity such that conservation laws, specifically the conservation of momentum, hold then this in turn makes a unique identification. In this identification we can connect two FLRW cosmologies back-to-back at the singularity and continue geodesics from one into the other. This continuation is motivated by our example of the Möbius strip, and in common with that example we see that orientation is inverted in the continuation. To relate | mathrm{0 } + \frac{3}{2 } t^\frac{2}{3}$. If we choose, as we did in the case of the Mobius strip example, to enforce continuity between period of the quad - time manifold covered by unlike choices of tetrad, and further to conserve the momentum of the photon as measure in this way (respecting the sign volume of space), then there be a unique geodesic on the other english of the singularity which connect to this point. This is given by $ x = x_o + \frac{3}{2 } |t|^\frac{2}{3}$, with the tetrad given in equality (\[extended\ ]). We note that the conditions of the Hawking – Penrose theorem are satisfied, and therefore the conclusion that there is infinite focusing of geodesics is valid. The distance between any two photon as measured use the space - time geometry will tend to zero as we approach the singularity. However, there be still a unique continuation of the paths beyond this point, which correspond to the above trajectory as described in the geometry in which we swap tetrads. What the singularity theorems require is that we model physics in terms of a space - time defined by a metric function on a globally orientable manifold. However, if we relax that presumption barely on the singular surface itself, we see that there constitute a natural physical continuation beyond this point when trace in relational terms.
To be clear about this continuation, we note that the Hawking – Penrose theorems read that there is not a unique continuation through the singularity of the space - meter geometry as determined by the Einstein equation. Our results are completely in agreement with this. We do not claim that the Einstein equations give such a continuation. However, the relational arrangement of scalar fields can be continued beyond this point, and at all period off from the singularity their behaviour is consistent with that identify by the Einstein equations couple to the matter. Thus we acknowledge that on both sides of the singularity there embody a consistent outer space - time movie. What we have then shown is that if we want to far identify trajectory of matter across the singularity such that conservation laws, specifically the conservation of momentum, hold then this in turning makes a unique identification. In this identification we can connect two FLRW cosmology back - to - back at the singularity and continue geodesic from one into the other. This continuation is motivated by our example of the Möbius strip, and in common with that example we visit that orientation is inverted in the continuation. To relate | matjrm{0} + \frac{3}{2} t^\frac{2}{3}$. If we chuose, as we did nb the rase of the Mobkus strip example, to enforce cintinyity between points of the spacv-time manufolv covered by difhsrent cmjicea of cevrad, and furthet to conserva the momentum ow che photon as measured in this way (rqspectimg the signed vojume jf slace), then there is a unique geodesjc on tie other side og the singularity which cojnechs to this point. Tjis is giveb by $z=x_o + \frac{3}{2} |t|^\ffac{2}{3}$, with the tetrad giben in equation (\[extended\]). We notd thac the condijnins mf the Hawkmng–Pengose theorem are sativfied, amd hence the cpncnusuon that there is infmnite focusing of gejdesics iv ralid. The distance berwwen aty tfo pfitovs zs msasuref uaing the slace-time geimetry will tend to zqgp as we apprkach tre singularity. However, there is still a utiqhe continuation of the paths beyond this poijt, which sorresponds to the above trajectory as described hn thx eeonebry kb ahich we swap tetrads. What the singularity thqkrtms require is thct we model physocd og terms of a rpace-tnje defined by a metrlc on a globqlly oriegtabke manifold. However, if we rwlax that asfymption just on thz singular sbrface itsekf, we see that there is a nztural physlcal contjvuation beyond tfis pmint when described in relwtional txrms.
Tp be cldar sbout ehis contijuation, we note that the Hawknng–Petrose theogems show that there is not a uimque continuajiot tvrough tke sinnularity of the space-time geoketry af detdrmined by the Eiistein equatyons. Our resunjs are complevely in adreenent with tfks. We do not ckaim that the Einstwin equations give suzg a continuatiou. Yowever, the relstiunaj dyxtqk of scalar xielas zsn be continued neyund yhis point, and at aln pojnts away from the slngulariti their bqhaviour is cpnsistent with thau desccibed uy the Eigstein equations coupled to ths matter. Hhuf we know thwt ok bojh sides of the singularity there is a consistent wpace-time picture. Qhat we have then skoen is that mf we rant to fgrther identify trajwctories of mattev across the singularify sucv thah conservation laws, specifically the conservation of momentum, hold then tyis in turn makes z unoque pdeutiyicatiog. In vhis identificatipn we can connect two FLRW cosmulogies taek-to-back at the singularity amd continue geoaesics from one into the other. Tnis continuation is motivated bu our example of the Möbims strip, qnd ih common wmth that example we xee thct oriwntatiob is inverted in ghe clntinuatiln. Tp xelate | mathrm{0} + \frac{3}{2} t^\frac{2}{3}$. If we choose, did the case the Mobius strip points the space-time manifold by different choices tetrad, and further to conserve the of the photon as measured in this way (respecting the signed volume of then there is a unique geodesic on the other side of the singularity connects this This given by $x=x_o + \frac{3}{2} |t|^\frac{2}{3}$, with the tetrad given in equation (\[extended\]). We note that conditions of the Hawking–Penrose theorem are satisfied, and the conclusion that there infinite focusing of geodesics is The between any photons measured the space-time geometry tend to zero as we approach the singularity. However, there is still a unique continuation of the beyond this corresponds to above as in the geometry we swap tetrads. What the singularity that we model physics in terms of a defined by metric on a globally orientable manifold. if we relax that assumption just on the surface itself, we see that there is a natural physical continuation beyond this point when relational terms. To be about this continuation, note the theorems that there not a unique continuation through the singularity of the space-time geometry determined by the Einstein equations. Our results are completely in this. do not claim the Einstein equations give a However, the relational system fields be point, at points away from the their behaviour is consistent with described by the Einstein Thus we know that on both sides of singularity there is a consistent space-time picture. we have then shown is that if we want to further identify of matter singularity such that conservation laws, specifically the conservation momentum, hold then this turn makes a unique identification. In this identification we connect FLRW cosmologies at the singularity continue geodesics from into the other. is motivated our of and in common with that example see orientation is inverted in the To | mathrm{0} + \frac{3}{2} t^\frac{2}{3}$. If we choose, As we did in tHe casE of The moBius StriP example, to enfoRCe coNtinuity between points oF the sPaCE-timE MaNifolD covereD By DIFfeReNt ChoIcES oF tetrAd, aNd furthEr to conserVe tHe Momentum of thE PhOton as measUreD in this way (reSpeCting tHe SigNEd volUme Of spaCe), then THere is A unique geOdESic on tHE other sIDE oF the Singularity which cONnECts to this point. this is GiVEn BY $X=x_o + \FraC{3}{2} |t|^\frac{2}{3}$, with ThE tetrAD given iN EqUATIon (\[EXtended\]). We note That the condITioNs of thE HAwkINg–PenrOse thEoREm aRe satisfied, And hEnce the coNclusiON that thERe is infInite fOcuSinG of gEOdEsIcs Is VAliD. thE diSTanCe betweeN aNy Two phOtonS AS MEasuRed UsinG the sPace-time geomeTry Will TEnd To zerO as we ApprOaCh the SingulArity. hoWever, there is stiLl a uNique contInuAtIon Of The paTHs beyoNd tHis Point, whIch corrESpoNdS TO ThE above trajectory as DeSCRiBed in the GeometRY iN wHIch we swaP tEtrAds. WHAT the sInguLArIty theorEms reqUIrE iS that we MoDel phySiCs iN teRms of A SpacE-time dEfined by A metrIC on a globally orIEntable manifoLD. HOWEvER, if wE reLax that assuMptiON jusT on tHE sIngULar suRface ItSElF, We see that there is a naTuRal phySical Continuation bEyond this pOINT when desCribED iN Relational termS.
To be Clear about THis contiNuatiOn, we note That the HaWKIng–PenroSe tHeoRemS shOW ThAt there is not a UNIque CoNtinuatIon Through The SinGulAriTy Of the spacE-time geoMeTrY aS dEteRmineD By the EinStEin EqUatIons. OUR resulTs are CompLeTeLY in AgreemeNT wITH thiS. WE dO not ClaIm That tHe EiNSteIn equatIons give sUch A ContInUaTion. HowEver, the relatiOnAl system of ScAlaR fieldS CAn be contInued beyond this point, and AT all poiNts Away fRom tHe singulaRitY their BehAViour iS consiStent WiTh tHAT descRIBeD by ThE Einstein eQUAtiOns coUpLed tO the matTer. Thus we know that oN BotH sides of the siNguLariTY ThEre IS a COnsIsTEnt SPAce-time picture. WHat we have tHeN ShOwn is that iF We wAnT to furtHer idenTify tRAjectorIes of mattEr across tHe SingULAriTy such that ConservaTion laws, sPEcifiCAlLy the ConServatIoN of MomenTum, holD TheN this In turn MaKes a unIque iDeNtificatIon. In this identification We can cOnnecT twO FLRW cosmOloGIes Back-to-bacK at tHe singularIty And ContiNue GEodesIcs fROm One INto thE othER. This contINuAtiON Is Motivated by OUR ExaMple oF thE möbius StriP, and in common with tHAt example we see That ORIenTatIOn is InVerted in the conTinUaTIOn. To relaTe | mathrm{0} + \frac{3}{2} t^ \frac{2}{3 }$. I f w e c ho ose, aswe did in thec aseof the Mobius strip ex ample ,t o en f or ce co ntinuit y b e t wee npo int so fthe s pac e-timemanifold c ove re d by differe n tchoices of te trad, and fu rth er toco nse r ve th e m oment um oft he pho ton as me as u red in this wa y (r espe cting the signedv ol u me of space),then t he r ei s auni que geodes ic on t h e other si d e oft he singularit y which con n ect s to t hi s p o int. T his i sg ive n by $x=x_o + \ frac{3}{2 } |t|^ \ frac{2} { 3}$, wi th the te tra d gi v en i n e qu a tio n ( \[e x ten ded\]).We n ote t hatt h e cond iti onsof th e Hawking–Pen ros e th e ore m are sati sfie d, andhencethe c on clusion that th ereis infini tefo cus in g ofg eodesi csisvalid.The dis t anc eb e t we en any two photons a s me asured u sing t h esp a ce-timege ome tryw i ll te nd t o z ero as w e appr o ac hthe sin gu larity .How eve r, th e re i s stil l a uniq ue co n tinuation of t h e paths beyon d t h i sp oint , w hich corres pond s tothea bo vet rajec toryas de s cribed in the geome tr y in w hichwe swap tetra ds. What t h e singular ityt he o rems require i s tha t we model physicsin te rms of a space-ti m e defined by amet ric o na globally or i e ntab le manifo ld. Howeve r,ifwerel ax that ass umptionju st o nthe sing u lar surf ac e i ts elf , wes ee tha t the re i san atu ral phy s ic a l con ti nu atio n b ey ond t hisp oin t whendescribed in rela ti on al term s.
To be cle ar about thi scon tinuat i o n, we no te that the Hawking–Pen r ose the ore ms sh ow t hat there is not a un i que co ntinua tionth rou g h thes i ng ula ri ty of thes p ace -time g eome try asdetermined by theE ins tein equation s.Ourr e su lts ar e co mp l ete l y in agreement w ith this.We do not claim tha tthe Ein stein e quati o ns give such a c ontinuati on . Ho w e ver , the rela tional s ystem ofs calar fi eldscan be co nt inu ed be yond t h ispoint , andat all p oints a way from the singularity theirbehavi our i s c onsistent wi t h t hat descr ibed by the Ei nst ein equa tio n s cou pled to th e matt er.T hus we kn o wtha t on both sides o f th e sin gul a rity t here is a consistents pace-time pict ure. W hat we have t hen shown is t hat i f we wantto further id entify t ra j ector ies of matte r acros s th e singu lari tysuch that co ns e rvation l aw s , spec ific al ly the conse r vati o n of momentum, ho ld th e n this inturnma kes a u n ique identific ation. In t his id enti ficat ion weca n conn ect t wo FLRW co s mologiesback- to-back a t th e s ingula rity a nd co ntin ue ge odesics f r o mo ne i n totheother .This continua t ion is m oti v ated by o ure x ampleo ft h e Möbius s tri p, an d in commonw itht ha t exam ple we see t hat ori e nta ti on is i nve r t ed in the continua t ion .To r elate | mathrm{0} +_\frac{3}{2} t^\frac{2}{3}$._If we choose, as_we did_in_the case_of_the Mobius strip_example, to enforce_continuity between points of_the space-time manifold_covered_by different choices of tetrad, and further to conserve the momentum of the photon_as_measured in_this_way_(respecting the signed volume of_space), then there is a_unique geodesic_on the other side of the singularity which_connects_to this point._This is given by $x=x_o + \frac{3}{2} |t|^\frac{2}{3}$, with_the tetrad given in equation (\[extended\])._We note that_the_conditions_of the Hawking–Penrose theorem_are satisfied, and hence the conclusion_that there is infinite focusing of_geodesics is valid. The distance between any_two photons as measured using the_space-time geometry will tend to_zero as_we approach the singularity. However,_there is still_a unique_continuation of the_paths beyond this point, which corresponds_to the above_trajectory as described in the geometry_in_which we swap_tetrads._What_the singularity_theorems require is_that_we model_physics_in terms of a space-time defined_by_a metric on a globally orientable manifold._However, if we relax_that_assumption just on the_singular surface itself, we see_that there is a natural physical_continuation beyond_this point_when described in relational terms.
To be clear about this continuation, we_note that the Hawking–Penrose theorems show_that there is not_a unique_continuation_through the singularity_of_the space-time_geometry as determined by the Einstein equations._Our results_are completely in agreement with this._We do not claim_that_the Einstein equations give such a_continuation. However, the relational system of_scalar fields can be continued_beyond_this_point, and at all points_away from the singularity their behaviour_is consistent with_that described by the Einstein equations coupled_to_the matter. Thus we know that_on_both sides of the singularity there_is_a_consistent space-time picture. What we_have then shown is that if_we want to further identify trajectories of matter across_the singularity such_that conservation laws, specifically the_conservation_of_momentum, hold then this in turn makes a unique identification._In this_identification we can_connect two FLRW cosmologies back-to-back at the singularity and continue_geodesics from one into the other. This_continuation is motivated by our example of the Möbius strip, and_in common with that example we see that_orientation is inverted in the continuation. To_relate |
[@gordon:salmon:smith:1993], where particles are propagated from the model transition, more sophisticated filters can readily be used in the CPF procedure. For instance, performance gains can be obtained with auxiliary particle filters [@pitt1999filtering; @johansen2008note], as illustrated in Section \[sec:numerics:hiddenar\]. In presenting algorithms we focus on bootstrap particle filters for simplicity. When the transition density is tractable, extensions of the CPF include backward sampling [@whiteleycommentonpmcmc; @LindstenS:2013] and ancestor sampling [@LindstenJS:2014], which is beneficial in the proposed approach as illustrated in Section \[sec:numerics:hiddenar\]. The complexity of a standard CPF update is of order $NT$, and the memory requirements are of order $T + N\log N$ [@jacob2015path].
The proposed method relies on CPF kernels but is different from Markov chain Monte Carlo (MCMC) estimators: it involves independent copies of unbiased estimators of $\pi(h)$. Thus it will be amenable to parallel computation and confidence intervals will be constructed in a different way than with standard MCMC output [e.g. Chapter 7 in @gelman2010handbook]; see Section \[sec:comparison\] for a comparison with existing smoothers.
Debiasing Markov chains \[sec:debiasing\]
-----------------------------------------
We briefly recall the debiasing technique of @glynn2014exact, see also @McLeish:2011 [@Rhee:Glynn:2012; @vihola2015unbiased] and references therein. Denote by $(X^{(n)})_{n\geq 0}$ and $({\tilde{X}}^{(n)})_{n\geq 0}$ two Markov chains with invariant distribution $\pi$, initialized from a distribution $\pi_0$. Assume that, for all $n\geq 0$, $X^{(n)}$ and ${\tilde{X}}^{(n)}$ have the same marginal distribution, and that $\lim_{n\to\infty} \mathbb{E}[h(X^{(n)})] = \pi(h)$. Writing limit as a telescopic sum, and swapping infinite sum and expectation, which will be justified later on, we obtain $$\begin{aligned}
\pi(h)
&= \mathbb{E}[h(X^{(0)})] + \sum_{n=1}^\infty \mathbb{E}[h(X^{(n)}) - h | [ @gordon: salmon: smith:1993 ], where particles are propagated from the model transition, more advanced filter can readily be used in the CPF operation. For example, performance gains can be obtain with auxiliary particle filter [ @pitt1999filtering; @johansen2008note ], as illustrated in Section \[sec: numerics: hiddenar\ ]. In presenting algorithm we focus on bootstrap particle filters for simplicity. When the passage density is tractable, extensions of the CPF include backward sampling [ @whiteleycommentonpmcmc; @LindstenS:2013 ] and ancestor sample [ @LindstenJS:2014 ], which is beneficial in the proposed approach as illustrated in Section \[sec: numerics: hiddenar\ ]. The complexity of a standard CPF update is of orderliness $ NT$, and the memory requirements are of order $ T + N\log N$ [ @jacob2015path ].
The propose method trust on CPF kernels but is unlike from Markov chain Monte Carlo (MCMC) estimators: it involves independent copy of unbiased estimators of $ \pi(h)$. Thus it will be amenable to parallel computation and assurance time interval will be constructed in a different way than with standard MCMC output [ e.g. Chapter 7 in @gelman2010handbook ]; see Section \[sec: comparison\ ] for a comparison with existing smoothers.
Debiasing Markov chains \[sec: debiasing\ ]
-----------------------------------------
We concisely remember the debiasing technique of @glynn2014exact, visit besides @McLeish:2011 [ @Rhee: Glynn:2012; @vihola2015unbiased ] and reference therein. Denote by $ (X^{(n)})_{n\geq 0}$ and $ ({ \tilde{X}}^{(n)})_{n\geq 0}$ two Markov chains with changeless distribution $ \pi$, initialized from a distribution $ \pi_0$. Assume that, for all $ n\geq 0 $, $ X^{(n)}$ and $ { \tilde{X}}^{(n)}$ have the like marginal distribution, and that $ \lim_{n\to\infty } \mathbb{E}[h(X^{(n) }) ] = \pi(h)$. Writing limit as a telescopic kernel, and swapping infinite sum and expectation, which will be justified later on, we obtain $ $ \begin{aligned }
\pi(h)
& = \mathbb{E}[h(X^{(0) }) ] + \sum_{n=1}^\infty \mathbb{E}[h(X^{(n) }) - h | [@gogdon:salmon:smith:1993], where pavticles are propctated hrom ths model gransition, more sophisticatev fioters can readily be used iv the CPF proceduee. Fie instance, performance gzlns ccn be obtained wlth auxiliasy particle fintdrd [@pitt1999filtering; @johansen2008note], as illtstratec ln Section \[sec:gumegiss:hisdenar\]. In presenting algorithms we focus mn bootstrap larticle filters for simpllcitj. When the transitlon density is eeactable, extdnsions of the CPF incmude backward sampling [@whiteleyzommeutonpmcmc; @LunestftS:2013] and ancewtor fampling [@LindstenJS:2014], fhich ix beneficial ik the priposed approach as ilnustrated in Sectijn \[sec:numarncs:hiddenar\]. The complwxuty ox a vtanaqrd CPR npdzte is of order $NT$, znd the memiry requirements art os order $T + N\lof N$ [@jasof2015path].
The proposed method relies on CPF nerhels but is different feom Markov chain Montg Carlo (MCIC) estimators: it involves independent copies of utbiasxd esuinatorr ov $\pi(h)$. Thus it will be amenable to parallel coiluuatpon and confidencc intervals will br fomftructed in a diffexsnf way than with stwndard ICMC iutput [e.g. Chalter 7 in @gelman2010handbook]; see Section \[sec:bompqrison\] for a compaxison with eristinb smopthers.
Debiasing Markov ehains \[sec:debiasijg\]
-----------------------------------------
We brieruy recall the deciaxitg technique of @glynn2014exact, see also @McLzish:2011 [@Rhed:Glymn:2012; @vihjla2015unbiasef] and references therein. Denoje by $(F^{(n)})_{n\geq 0}$ anf $({\tilde{X}}^{(n)})_{n\geq 0}$ two Markov chains with invariamt divtributijn $\pi$, initialized srom a distribotion $\pi_0$. Cssume that, for zll $n\gex 0$, $X^{(n)}$ and ${\tylde{X}}^{(n)}$ have tvg same marginel distrifutiin, abd that $\uim_{n\to\infty} \mayhbb{E}[h(X^{(n)})] = \pi(h)$. Writibg limit as a telexcokid sum, and swappnug infinite sum amd dxpqcnatmon, wrhch will be bustkfidc latdr on, wt ontakn $$\brgin{aligned}
\pi(h)
&= \jathbb{E}[h(X^{(0)})] + \sum_{n=1}^\infyy \mathbb{E}[y(X^{(n)}) - h | [@gordon:salmon:smith:1993], where particles are propagated from the more filters can be used in performance can be obtained auxiliary particle filters @johansen2008note], as illustrated in Section \[sec:numerics:hiddenar\]. presenting algorithms we focus on bootstrap particle filters for simplicity. When the transition is tractable, extensions of the CPF include backward sampling [@whiteleycommentonpmcmc; @LindstenS:2013] and ancestor [@LindstenJS:2014], is in proposed approach as illustrated in Section \[sec:numerics:hiddenar\]. The complexity of a standard CPF update is of $NT$, and the memory requirements are of order + N\log N$ [@jacob2015path]. proposed method relies on CPF but different from chain Carlo estimators: it involves copies of unbiased estimators of $\pi(h)$. Thus it will be amenable to parallel computation and confidence intervals be constructed different way with MCMC [e.g. Chapter 7 see Section \[sec:comparison\] for a comparison Debiasing Markov chains \[sec:debiasing\] ----------------------------------------- We briefly recall debiasing technique @glynn2014exact, see also @McLeish:2011 [@Rhee:Glynn:2012; @vihola2015unbiased] references therein. Denote by $(X^{(n)})_{n\geq 0}$ and $({\tilde{X}}^{(n)})_{n\geq two Markov chains with invariant distribution $\pi$, initialized from a distribution $\pi_0$. Assume that, for 0$, $X^{(n)}$ and ${\tilde{X}}^{(n)}$ the same marginal and $\lim_{n\to\infty} = Writing limit a telescopic sum, and swapping infinite sum and expectation, which will justified later on, we obtain $$\begin{aligned} \pi(h) &= \mathbb{E}[h(X^{(0)})] + - | [@gordon:salmon:smith:1993], where parTicles are pRopagAteD frOm The mOdel Transition, more SOphiSticated filters can readIly be UsED in tHE CpF proCedure. FOR iNSTanCe, PeRfoRmANcE gainS caN be obtaIned with auXilIaRy particle fiLTeRs [@pitt1999filtEriNg; @johansen2008noTe], aS illusTrAteD In SecTioN \[sec:nUmericS:HiddenAr\]. In preseNtINg algoRIthms we FOCuS on bOotstrap particle fILtERs for simplicitY. When tHe TRaNSItiOn dEnsity is trAcTable, EXtensioNS oF THE CPf Include backwaRd sampling [@wHIteLeycomMeNtoNPmcmc; @LIndstEns:2013] And Ancestor samPlinG [@LindstenjS:2014], whicH Is benefICial in tHe propOseD apProaCH aS iLluStRAteD In secTIon \[Sec:numerIcS:hIddenAr\]. ThE COMPlexIty Of a sTandaRd CPF update is Of oRder $nt$, anD the mEmory RequIrEmentS are of Order $t + N\Log N$ [@jacob2015path].
ThE proPosed methOd rElIes On cPF keRNels buT is DifFerent fRom MarkOV chAiN mONtE Carlo (MCMC) estimatoRs: IT InVolves inDependENt CoPIes of unbIaSed EstiMATors oF $\pi(h)$. tHuS it will bE amenaBLe To ParalleL cOmputaTiOn aNd cOnfidENce iNtervaLs will be ConstRUcted in a differENt way than with STaNDArD mCMC OutPut [e.g. ChapteR 7 in @gELman2010HandBOoK]; seE sectiOn \[sec:CoMPaRIson\] for a comparison wItH existIng smOothers.
DebiasIng Markov cHAINs \[sec:debIasiNG\]
-----------------------------------------
WE Briefly recall tHe debIasing techNIque of @glYnn2014exAct, see alSo @McLeish:2011 [@rHEe:Glynn:2012; @vIhoLa2015uNbiAseD] ANd References theREIn. DeNoTe by $(X^{(n)})_{n\Geq 0}$ And $({\tildE{X}}^{(n)})_{N\geQ 0}$ twO MaRkOv chains wIth invarIaNt DiStRibUtion $\PI$, initialIzEd fRoM a dIstriBUtion $\pI_0$. AssuMe thAt, FoR All $N\geq 0$, $X^{(n)}$ aND ${\tILDe{X}}^{(n)}$ HaVe The sAme MaRginaL disTRibUtion, anD that $\lim_{n\To\iNFty} \mAtHbB{E}[h(X^{(n)})] = \pi(H)$. Writing limit As A telescopiC sUm, aNd swapPINg infiniTe sum and expectation, whicH Will be jUstIfied LateR on, we obtaIn $$\bEgin{alIgnED}
\pi(h)
&= \maThbb{E}[h(x^{(0)})] + \sum_{n=1}^\InFty \MAThbb{E}[H(x^{(N)}) - h | [@gordon:salmon:smith:199 3], whereparti cle s a re pro paga ted from the m o deltransition, more sophi stica te d fil t er s can readil y b e use din th eC PF proc edu re. For instance, pe rf ormance gain s c an be obta ine d with auxil iar y part ic lef ilter s [ @pitt 1999fi l tering ; @johans en 2 008not e ], as i l l us trat ed in Section \[s e c: n umerics:hidden ar\].In pr e s ent ing algorithm swe fo c us on b o ot s t r app article filte rs for simp l ici ty. Wh en th e trans ition d e nsi ty is tract able , extensi ons of the CPF include backw ard sa mpli n g[@ whi te l eyc o mm ent o npm cmc; @Li nd st enS:2 013] a n d anc est or s ampli ng [@Lindsten JS: 2014 ] , w hichis be nefi ci al in the p ropos ed approach as il lust rated inSec ti on\[ sec:n u merics :hi dde nar\].The com p lex it y o fa standard CPF upd at e is of orde r $NT$ , a nd the memo ry re quir e m entsareo forder $T + N\l o gN$ [@jaco b2 015pat h] .
The prop o sedmethod relieson CP F kernels but i s different fr o mM a rk o v ch ain Monte Carl o (M C MC)esti m at ors : it i nvolv es in d ependent copies ofun biased esti mators of $\p i(h)$. Thu s i t will b e am e na b le to parallel comp utation an d confide nce i ntervals will bec o nstructe d i n a di ffe r e nt way than wit h stan da rd MCMC ou tput [e .g. Ch apt er7in @gelma n2010han db oo k] ;see Sect i on \[sec :c omp ar iso n\] f o r a co mpari sonwi th exi sting s m oo t h ers.
De bias ing M arkov cha i ns\[sec:d ebiasing\ ]
- - ---- -- -- ------- ------------- -- ---------
W e b riefly r ecall th e debiasing technique o f @glynn 201 4exac t, s ee also @ McL eish:2 011 [@Rhee :Glynn :2012 ;@vi h o la201 5 u nb ias ed ] and refe r e nce s the re in.Denoteby $(X^{(n)})_{n\g e q 0 }$ and $({\ti lde {X}} ^ { (n )}) _ {n \ geq 0 } $ t w o Markov chainswith invar ia n tdistributi o n $ \p i$, ini tialize d fro m a dist ribution$\pi_0$.As sume t hat , for all$n\geq 0 $, $X^{(n ) }$ an d $ {\til de{ X}}^{( n) }$havethe sa m e m argin al dis tr ibutio n, an dthat $\l im_{n\to\infty} \mathbb {E}[h( X^{(n )}) ] = \pi(h )$. Wri ting limi t as a telesco pic su m, an d s w appin g in f in ite sum a nd e x pectation , w hic h wi ll be justi f i e d l ateron, we obt ain$$\begin{aligned} \pi(h)
&= \ mat hbb { E}[h (X ^{(0)})] + \su m_{ n= 1 } ^\infty\m athbb{E}[h( X^{(n)}) - h | [@gordon:salmon:smith:1993],_where particles_are propagated from the_model transition,_more_sophisticated filters_can_readily be used_in the CPF_procedure. For instance, performance_gains can be_obtained_with auxiliary particle filters [@pitt1999filtering; @johansen2008note], as illustrated in Section \[sec:numerics:hiddenar\]. In presenting algorithms_we_focus on_bootstrap_particle_filters for simplicity. When the_transition density is tractable, extensions_of the_CPF include backward sampling [@whiteleycommentonpmcmc; @LindstenS:2013] and ancestor_sampling_[@LindstenJS:2014], which is_beneficial in the proposed approach as illustrated in Section_\[sec:numerics:hiddenar\]. The complexity of a standard_CPF update is_of_order_$NT$, and the memory_requirements are of order $T +_N\log N$ [@jacob2015path].
The proposed method relies_on CPF kernels but is different from_Markov chain Monte Carlo (MCMC) estimators:_it involves independent copies of_unbiased estimators_of $\pi(h)$. Thus it will_be amenable to_parallel computation_and confidence intervals_will be constructed in a different_way than with_standard MCMC output [e.g. Chapter 7_in_@gelman2010handbook]; see Section_\[sec:comparison\]_for_a comparison_with existing smoothers.
Debiasing_Markov_chains \[sec:debiasing\]
-----------------------------------------
We_briefly_recall the debiasing technique of @glynn2014exact,_see_also @McLeish:2011 [@Rhee:Glynn:2012; @vihola2015unbiased] and references therein._Denote by $(X^{(n)})_{n\geq 0}$_and_$({\tilde{X}}^{(n)})_{n\geq 0}$ two Markov_chains with invariant distribution $\pi$,_initialized from a distribution $\pi_0$. Assume_that, for_all $n\geq_0$, $X^{(n)}$ and ${\tilde{X}}^{(n)}$ have the same marginal distribution, and that_$\lim_{n\to\infty} \mathbb{E}[h(X^{(n)})] = \pi(h)$. Writing limit_as a telescopic sum,_and swapping_infinite_sum and expectation,_which_will be_justified later on, we obtain $$\begin{aligned}
_ _\pi(h)
&=_\mathbb{E}[h(X^{(0)})] + \sum_{n=1}^\infty \mathbb{E}[h(X^{(n)})_-_h |
}$$ is the free-field part of the $\varphi$ hamiltonian, plus counterterms (some of which involve the $\chi$ field) that are necessary to remove infinities in this composite operator.
Eq. (\[n2\]) is a highly plausible definition of the heavy particle number density (at low temperatures). However, it does not correspond in any obvious way to how this number density would be determined experimentally. Standard methods all involve a search for individual, on-shell $\varphi$ particles. Real-world examples of this include present-day dark matter searches, and measurements of the cosmic microwave background radiation. Therefore, it is possible to suspect that the definition (\[n2\]) is not appropriate, and this is the reason for the surprising lack of Boltzmann suppression in the loop corrections to $n$.
In [@as], this issue was investigated using a simple Caldeira–Leggett model [@cl]. In this exactly solvable model, a “system” (represented by a harmonic oscillator) is coupled to an “environment” (represented by more oscillators) via an “interaction”. We found that, at low temperatures, the “interaction” energy was always comparable to the “system” energy, making the identification of the “system” problematic.
Consider now a slight variation of the MY model, in which the heavy field is complex, and carries a conserved U(1) charge, while the light field remains real and neutral [@ms]. In this model, we can study the charge fluctuations in a given volume $V$. For noninteracting particles, Poisson statistics for the number of positively and negatively charged particles in $V$ results in $$\langle Q^2\rangle = n V.
\label{q2}$$ We see that the charge fluctuations give us a measurement of the total number of heavy particles in a given volume. If $Q$ represents electric charge, we can in principle measure $\langle Q^2\rangle$ without tracking individual heavy particles. Furthermore, it seems highly unlikely that weak interactions could significantly modify eq. (\[q2\]). If $\langle Q^2\rangle$ is either much larger or much smaller than $nV$, then the movements of positive and negative particles into and out of $V$ would have to be highly correlated (in order to suppress or enhance the charge fluctuations in $V$). This is inconsistent with the usual notion of a gas of particles that move freely and independently between occasional scatterings, and would appear to require strong interactions. | } $ $ is the free - field part of the $ \varphi$ hamiltonian, plus counterterms (some of which involve the $ \chi$ sphere) that are necessary to absent infinities in this composite operator.
Eq. (\[n2\ ]) is a highly plausible definition of the heavy atom number density (at low temperature). However, it does not correspond in any obvious way to how this act density would be determined experimentally. Standard method acting all involve a search for individual, on - plate $ \varphi$ particle. Real - world examples of this include present - day black matter searches, and measurements of the cosmic microwave backdrop radiation. Therefore, it is possible to suspect that the definition (\[n2\ ]) is not appropriate, and this is the rationality for the surprising lack of Boltzmann suppression in the loop corrections to $ n$.
In [ @as ], this issue was investigated use a simple Caldeira – Leggett model [ @cl ]. In this exactly solvable model, a “ system ” (constitute by a harmonic oscillator) is coupled to an “ environment ” (represented by more oscillators) via an “ interaction ”. We found that, at low temperatures, the “ interaction ” energy was always comparable to the “ system ” department of energy, create the identification of the “ organization ” debatable.
Consider now a slight mutant of the MY model, in which the heavy field is complex, and carries a conserved U(1) accusation, while the light field remains real and neutral [ @ms ]. In this mannequin, we can study the charge fluctuations in a given bulk $ V$. For noninteracting particles, Poisson statistics for the act of positively and negatively charged particles in $ V$ results in $ $ \langle Q^2\rangle = n V.
\label{q2}$$ We see that the charge fluctuation give us a measurement of the total number of dense particles in a given volume. If $ Q$ represents electric bang, we can in principle measure $ \langle Q^2\rangle$ without chase individual heavy atom. Furthermore, it seems highly improbable that weak interaction could significantly change eq. (\[q2\ ]). If $ \langle Q^2\rangle$ is either much larger or much small than $ nV$, then the movements of positive and negative particles into and out of $ V$ would have to be highly correlated (in order to suppress or enhance the charge fluctuations in $ V$). This is inconsistent with the common notion of a gas of atom that move freely and independently between occasional scatterings, and would appear to require strong interactions. | }$$ is the free-field part of tme $\varphi$ hamiltonian, pnus cohnterteros (some of which involve the $\cyi$ fitjd) that are necessxry to relove infunitmes in this composite opcxator.
Se. (\[n2\]) iv a highly plauxible defitition of the vexvv particle number density (at low tem[eraturrs). However, it dogs nou cjrreaiokd in any obvious way to how thia numbeg density would br determined experimentallj. Stwndard methods all involve a wearsy for indiviaual, on-shell $\varphi$ patticles. Real-world examples of thks inelude presebt-eay gark matter searbhes, and measmgements of the cosmic microwsve baxkground radiation. Thxrefore, it is possibje to sus[eet that the definitiob (\[b2\]) is tot dppruprixte, aid fhis id tie reason fkr the surpeising lack of Boltemagb suppression in thq joop corrections to $n$.
In [@as], this issue wds jnvestigated using a sinple Caldeira–Leggett lodel [@cl]. Yn this exactly solvable model, a “system” (representad by x hcviubif oscillator) is coupled to an “environment” (repwsstntvd by more oscillctors) via an “intrrwcyyon”. We found jhat, at los temperatures, the “interastion” energy wws akways comparable to the “sysrem” energy, mckibg the identificatnon of the “svstem” kroblekatic.
Consider now a sliyht vadiation of hhe MY mosdl, in which the feany fheld is complex, and carrief a consecved B(1) charge, whike the light fiepd remains real and neutgal [@md]. Hn this mofel, we can study the charge flurvuations in a ghvet volume $V$. Fov noninteractind particles, Ponsson stctistizs for the number of positivqly and negathgely charged particlqs ib $V$ eesults kn $$\langle Q^2\ranble = n V.
\lcyel{q2}$$ We swe that the charge flocfuations give uw a measurement of thd tjtwl ntkber of heavf paftizkes iv a givtu viuume. If $Q$ represents elewtrid charge, we can in pvinciple neasure $\jangle Q^2\ranglr$ without tracking inditidual heavu pwrticles. Furthermore, it seems gighly unpikcly that weak intcracjions could significantly modify eq. (\[q2\]). If $\langle Q^2\raigle$ is either much latger or much smaller jhak $nV$, then thx moveients of [ositive and negativw particles into snd out of $V$ would habe to te hihhly correlated (in order to suppress or enhance the charge fluctuations ib $V$). Thms is inconsiateny witv che usual notmoi of a gas of parnicles that move freely and inde'endently tecween occasional scatterings, sna would appeat to require strong interadtions. | }$$ is the free-field part of the plus (some of involve the $\chi$ remove in this composite Eq. (\[n2\]) is highly plausible definition of the heavy number density (at low temperatures). However, it does not correspond in any obvious to how this number density would be determined experimentally. Standard methods all involve search individual, $\varphi$ Real-world examples of this include present-day dark matter searches, and measurements of the cosmic microwave background Therefore, it is possible to suspect that the (\[n2\]) is not appropriate, this is the reason for surprising of Boltzmann in loop to $n$. In this issue was investigated using a simple Caldeira–Leggett model [@cl]. In this exactly solvable model, a “system” by a is coupled an (represented more oscillators) via We found that, at low temperatures, was always comparable to the “system” energy, making identification of “system” problematic. Consider now a slight of the MY model, in which the heavy is complex, and carries a conserved U(1) charge, while the light field remains real and In this model, we study the charge in given $V$. noninteracting particles, statistics for the number of positively and negatively charged particles in results in $$\langle Q^2\rangle = n V. \label{q2}$$ We see charge give us a of the total number heavy in a given volume. represents charge, principle $\langle without tracking individual heavy Furthermore, it seems highly unlikely weak interactions could significantly Q^2\rangle$ is either much larger or much smaller $nV$, then the movements of positive and particles into and out of $V$ would have to be highly correlated order to enhance the charge fluctuations in $V$). This is with the usual notion a gas of particles that move freely and independently occasional and would to require strong | }$$ is the free-field part of the $\vaRphi$ hamiltOnian, PluS coUnTertErms (Some of which invOLve tHe $\chi$ field) that are necesSary tO rEMove INfInitiEs in thiS CoMPOsiTe OpEraToR.
eq. (\[N2\]) is a hIghLy plausIble definiTioN oF the heavy parTIcLe number deNsiTy (at low tempeRatUres). HoWeVer, IT does Not CorreSpond iN Any obvIous way to HoW This nuMBer densITY wOuld Be determined experIMeNTally. Standard mEthods AlL InVOLve A seArch for indIvIdual, ON-shell $\vARpHI$ PArtICles. Real-world Examples of tHIs iNclude PrEseNT-day daRk matTeR SeaRches, and meaSureMents of thE cosmiC MicrowaVE backgrOund raDiaTioN. TheREfOrE, it Is POssIBlE to SUspEct that tHe DeFinitIon (\[n2\]) IS NOT appRopRiatE, and tHis is the reasoN foR the SUrpRisinG lack Of BoLtZmann SuppreSsion In The loop correctiOns tO $n$.
In [@as], thiS isSuE waS iNvestIGated uSinG a sImple CaLdeira–LEGgeTt MODEl [@Cl]. In this exactly solVaBLE mOdel, a “sysTem” (repREsEnTEd by a harMoNic OsciLLAtor) iS couPLeD to an “envIronmeNT” (rEpResenteD bY more oScIllAtoRs) via AN “intEractiOn”. We founD that, AT low temperaturES, the “interactiON” eNERgY Was aLwaYs comparablE to tHE “sysTem” eNErGy, mAKing tHe ideNtIFiCAtion of the “system” proBlEmatic.
consiDer now a slight Variation oF THE MY model, In whICh THe heavy field is ComplEx, and carriES a conserVed U(1) cHarge, whiLe the lighT FIeld remaIns ReaL anD neUTRaL [@ms]. In this modeL, WE can StUdy the cHarGe fluctUatIonS in A giVeN volume $V$. FOr nonintErAcTiNg ParTicleS, poisson sTaTisTiCs fOr the NUmber oF posiTiveLy AnD NegAtively CHaRGEd paRtIcLes iN $V$ rEsUlts iN $$\lanGLe Q^2\Rangle = n v.
\label{q2}$$ We See THat tHe ChArge fluCtuations give Us A measuremeNt Of tHe totaL NUmber of hEavy particles in a given voLUme. If $Q$ rEprEsentS eleCtric charGe, wE can in PriNCiple mEasure $\LanglE Q^2\RanGLE$ withOUT tRacKiNg individuAL HeaVy parTiCles. furtherMore, it seems highly uNLikEly that weak inTerActiONS cOulD SiGNifIcANtlY MOdify eq. (\[q2\]). If $\langlE Q^2\rangle$ is EiTHeR much largeR Or mUcH smalleR than $nV$, Then tHE movemeNts of posiTive and neGaTive PARtiCles into anD out of $V$ wOuld have tO Be higHLy CorreLatEd (in orDeR to SupprEss or eNHanCe the Charge FlUctuatIons iN $V$). this is inConsistent with the usual nOtion oF a gas Of pArticles tHat MOve Freely and IndePendently bEtwEen OccasIonAL scatTeriNGs, And WOuld aPpeaR To require STrOng INTeRactions. | }$$ is the free-field part of the $\ varph i$ham il toni an,plus counterte r ms ( some of which involvethe $ \c h i$ f i el d) th at aren ec e s sar yto re mo v einfin iti es in t his compos ite o perator.
Eq . ( \[n2\]) is ahighly plaus ibl e defi ni tio n of t heheavy parti c le num ber densi ty (at lo w temper a t ur es). However, it does no t correspond in any o bv i ou s way to how thisnu mberd ensityw ou l d bed etermined exp erimentally . St andard m eth o ds all invo lv e asearch forindi vidual, o n-shel l $\varp h i$ part icles. Re al- worl d e xa mpl es oft hi s i n clu de prese nt -d ay da rk m a t t e r se arc hes, andmeasurementsofthec osm ic mi crowa ve b ac kgrou nd rad iatio n. Therefore, itis p ossible t o s us pec tthatt he def ini tio n (\[n2 \]) isn otap p r o pr iate, and this isth e re ason for the s u rp ri s ing lack o f B oltz m a nn su ppre s si on in th e loop co rr ections t o $n$.
In[@a s], t h is i ssue w as inves tigat e d using a simp l e Caldeira–Le g ge t t m o del[@c l]. In this exa c tlysolv a bl e m o del,a “sy st e m” (represented by a h ar monicoscil lator) is cou pled to an “ e nvironme nt”( re p resented by mo re os cillators) via an “ inter action”. We found t hat, atlow te mpe rat u r es , the “intera c t ion” e nergy w asalwayscom par abl e t othe “syst em” ener gy ,ma ki ngthe i d entifica ti onof th e “sy s tem” p roble mati c.
C ons ider no w a s ligh tva riat ion o f the MYm ode l, in w hich thehea v y fi el dis comp lex, and carr ie s a conser ve d U (1) ch a r ge, whil e the light field remai n s realand neut ral[@ms]. In th is mod el, we can study thech arg e fluct u a ti ons i n a givenv o lum e $V$ .Fornoninte racting particles, Poi sson statisti csfort h enum b er ofpo s iti v e ly and negative ly charged p a rt icles in $ V $ r es ults in $$\lan gle Q ^ 2\rangl e = n V.\label{q2 }$ $ We s eethat the c harge fl uctuation s give us a me asu rement o f t he to tal nu m ber of h eavy p ar ticles in a g iven vol ume. If $Q$ representselectr ic ch arg e, we can in pri nciple me asur e $\langle Q^ 2\r angle $ w i thout tra c ki ngi ndivi dual heavy par t ic les . Fu rthermore,i t see ms hi ghl y unlik elythat weak interac t ions could sig nifi c a ntl y m o dify e q. (\[q2\]). I f $ \l a n gle Q^2\ ra ngle$ is ei ther muc hl arger or mu ch sma ller th a n $ n V$, th en t hemovements of p o sitivean dn egativ e pa rt iclesinto a n d ou t of $V$ would hav e tob e high l y c orrel at ed (ino rder to suppre ss or enhan ce the cha rge f luctuat io ns in$V$ ). This is i n consisten t wit h the u su al n oti on ofa ga s of pa rtic le s t hat movef r ee l yan d in depe ndent ly bet ween occa s ional sc att e rings,an d w o u ld app e ar t o requirestr ong i n t eractions. | }$$ is_the free-field_part of the $\varphi$_hamiltonian, plus_counterterms_(some of_which_involve the $\chi$_field) that are_necessary to remove infinities_in this composite_operator.
Eq. (\[n2\])_is a highly plausible definition of the heavy particle number density (at low temperatures)._However,_it does_not_correspond_in any obvious way to_how this number density would_be determined_experimentally. Standard methods all involve a search for_individual,_on-shell $\varphi$ particles._Real-world examples of this include present-day dark matter searches,_and measurements of the cosmic microwave_background radiation. Therefore,_it_is_possible to suspect that_the definition (\[n2\]) is not appropriate,_and this is the reason for_the surprising lack of Boltzmann suppression in_the loop corrections to $n$.
In [@as],_this issue was investigated using_a simple_Caldeira–Leggett model [@cl]. In this_exactly solvable model,_a “system”_(represented by a_harmonic oscillator) is coupled to an_“environment” (represented by_more oscillators) via an “interaction”. We_found_that, at low_temperatures,_the_“interaction” energy_was always comparable_to_the “system”_energy,_making the identification of the “system”_problematic.
Consider_now a slight variation of the MY_model, in which the_heavy_field is complex, and_carries a conserved U(1) charge,_while the light field remains real_and neutral_[@ms]. In_this model, we can study the charge fluctuations in a given_volume $V$. For noninteracting particles, Poisson_statistics for the number_of positively_and_negatively charged particles_in_$V$ results_in $$\langle Q^2\rangle = n V.
\label{q2}$$ We_see that_the charge fluctuations give us a_measurement of the total_number_of heavy particles in a given_volume. If $Q$ represents electric charge,_we can in principle measure_$\langle_Q^2\rangle$_without tracking individual heavy particles._Furthermore, it seems highly unlikely that_weak interactions could_significantly modify eq. (\[q2\]). If $\langle Q^2\rangle$ is_either_much larger or much smaller than_$nV$,_then the movements of positive and_negative_particles_into and out of $V$_would have to be highly correlated_(in order to suppress or enhance the charge fluctuations_in $V$). This_is inconsistent with the usual_notion_of_a gas of particles that move freely and independently between_occasional scatterings,_and would appear_to require strong interactions. |
in the upper-half complex plane. The Unruh modes annihilate the Minkowski vacuum state $$\begin{aligned}
\hat{c}_{m\omega} | 0_M \rangle = \hat{d}_{m\omega} | 0_M \rangle = 0\end{aligned}$$ as noted above.
Circuit model {#circuit}
=============
General formalism
-----------------
How are the states of a quantum field affected by an object (such as a beamsplitter) that is uniformly accelerated in the $R$ wedge? This is the question of central interest in this paper. A straightforward way to study this problem is to work in the accelerated frame in which the object is static. It is obvious that the object only interacts with Rindler modes in the $R$ wedge and the Rindler modes in the $L$ wedge remain unaffected. The interaction between the object and the Rindler modes is unitary and it transforms the Rindler modes as $$\label{generalcoupling}
\hat{b}_{mk}^{\prime R} = \int d \omega \bigg( \alpha^{m1}_{k \omega} \hat{b}_{1\omega}^R
+ \beta^{m1}_{k \omega} \hat{b}_{1\omega}^{R \dagger}
+ \alpha^{m2}_{k \omega} \hat{b}_{2 \omega}^R
+ \beta^{m2}_{k \omega} \hat{b}_{2\omega}^{R \dagger} \bigg). \\$$ This is the most general interaction which not only couples the left-moving and right-moving Rindler modes but also Rindler modes with different frequencies. Together with Eqs. (\[UnruhRindler\]) and (\[UnruhRindler:inverse\]), we can construct a quantum circuit model (or input-output formalism) for the uniformly accelerated object. We start from the inertial frame in which Unruh modes are used instead of Minkowski modes. This makes the model simpler although we still have to transform the Minkowski modes to the Unruh modes and vice versa.
First, based on Eq. (\[UnruhRindler:inverse\]), the Unruh modes pass through a collection of two-mode squeezers each of which couples a pair of Unruh modes $(\hat{c}_{m\omega}, \hat{d}_{m\ | in the upper - half complex plane. The Unruh modes annihilate the Minkowski void department of state $ $ \begin{aligned }
\hat{c}_{m\omega } | 0_M \rangle = \hat{d}_{m\omega } | 0_M \rangle = 0\end{aligned}$$ as noted above.
Circuit exemplar { # circumference }
= = = = = = = = = = = = =
General formalism
-----------------
How are the states of a quantum playing field feign by an object (such as a beamsplitter) that is uniformly accelerate in the $ R$ wedge? This is the question of cardinal interest in this paper. A straightforward way to analyze this problem is to work in the accelerated frame in which the object is static. It is obvious that the aim only interacts with Rindler modes in the $ R$ hacek and the Rindler modes in the $ L$ wedge stay unaffected. The interaction between the aim and the Rindler modes is one and it transforms the Rindler modes as $ $ \label{generalcoupling }
\hat{b}_{mk}^{\prime R } = \int d \omega \bigg (\alpha^{m1}_{k \omega } \hat{b}_{1\omega}^R
+ \beta^{m1}_{k \omega } \hat{b}_{1\omega}^{R \dagger }
+ \alpha^{m2}_{k \omega } \hat{b}_{2 \omega}^R
+ \beta^{m2}_{k \omega } \hat{b}_{2\omega}^{R \dagger } \bigg). \\$$ This is the about general interaction which not only couples the left - moving and right - moving Rindler modes but also Rindler modes with different frequencies. Together with Eqs. (\[UnruhRindler\ ]) and (\[UnruhRindler: inverse\ ]), we can construct a quantum circuit model (or input - output formalism) for the uniformly accelerated aim. We get down from the inertial frame in which Unruh modality are use instead of Minkowski modes. This makes the model simpler although we still have to translate the Minkowski modes to the Unruh modes and vice versa.
First, based on Eq. (\[UnruhRindler: inverse\ ]), the Unruh modes pass through a collection of two - mode squeezers each of which couples a pair of Unruh modes $ (\hat{c}_{m\omega }, \hat{d}_{m\ | in the upper-half complex puane. The Unruh modes ainihilafe the Mknkowski vacuum state $$\begin{apitned}
\hqt{c}_{m\omega} | 0_M \rangle = \hxt{d}_{m\omega} | 0_M \rangoe = 0\tnd{aligned}$$ as notxs above.
Gnrcuif modzl {#circuit}
=============
Generak formalisk
-----------------
How are the sdaged of a quantum field affected by an object (skch as a beamsklitttr) ehat ps uniformly accelerated in the $R$ wedgt? This is the quesyion of central interest ij thls paper. A straighhforward wai to wtudy this pfoblem is uo work in tge accelerated frame in which tfe oblect is stajnx. Ih is obvious that the object only intaracts eith Rindler mpdev ib the $R$ wedge and the Rindler modes in jhe $L$ wedga xemain unaffected. The ibterawtiot begqeev tge ogject wnd the Rindlsr modes is unitary and it tramssirms the Rindmer moqef as $$\label{generalcoupling}
\hat{b}_{mk}^{\prime R} = \inf d \omega \bigg( \alpha^{m1}_{k \imega} \hat{b}_{1\omega}^R
+ \betw^{m1}_{k \omega} \hat{b}_{1\omega}^{R \dagger}
+ \alpha^{m2}_{k \omega} \hat{b}_{2 \omega}^R
+ \beda^{m2}_{k \koegc} \hat{b}_{2\uneha}^{R \dagger} \bigg). \\$$ This is the most general intqdavtpon which not onln couples the left-kogimd and right-moxing Rnhdmer modes but also Rindlet modew with disferrnt frequencies. Together wirh Eqs. (\[UnruhGindoer\]) and (\[UnruhRindlzr:inverse\]), we can consyruct a quantum circuit modsl (or input-lutput fodoalism) for the uvifprkly accelerated object. We ftart fron thz inertixl ftame in which Unrkh modes are used insteaf of Litkowski mofes. This makes the model simplec although we sdiln have tj traksform the Minkjwski modes to the Unxuh moaes and vibe versa.
Fmrst, based og Eq. (\[UnruhRingper:inverse\]), tie Unruh iodew paws throjeh a collectiom of two-mode squeezwrs each of which gouplgs a pair of Unruk nodes $(\hat{c}_{m\omegs}, \hxt{d}_{i\ | in the upper-half complex plane. The Unruh the vacuum state \hat{c}_{m\omega} | 0_M \rangle 0\end{aligned}$$ as noted Circuit model {#circuit} General formalism ----------------- How are the of a quantum field affected by an object (such as a beamsplitter) that uniformly accelerated in the $R$ wedge? This is the question of central interest this A way study this problem is to work in the accelerated frame in which the object is static. is obvious that the object only interacts with modes in the $R$ and the Rindler modes in $L$ remain unaffected. interaction the and the Rindler is unitary and it transforms the Rindler modes as $$\label{generalcoupling} \hat{b}_{mk}^{\prime R} = \int d \omega \bigg( \omega} \hat{b}_{1\omega}^R \omega} \hat{b}_{1\omega}^{R + \omega} \omega}^R + \beta^{m2}_{k \dagger} \bigg). \\$$ This is the which not only couples the left-moving and right-moving modes but Rindler modes with different frequencies. Together Eqs. (\[UnruhRindler\]) and (\[UnruhRindler:inverse\]), we can construct a circuit model (or input-output formalism) for the uniformly accelerated object. We start from the inertial which Unruh modes are instead of Minkowski This the simpler we still to transform the Minkowski modes to the Unruh modes and vice First, based on Eq. (\[UnruhRindler:inverse\]), the Unruh modes pass through of squeezers each of couples a pair of modes \hat{d}_{m\ | in the upper-half complex planE. The Unruh mOdes aNniHilAtE the minkOwski vacuum staTE $$\begIn{aligned}
\hat{c}_{m\omega} | 0_M \raNgle = \hAt{D}_{M\omeGA} | 0_M \RanglE = 0\end{aliGNeD}$$ AS noTeD aBovE.
CIRcUit moDel {#Circuit}
=============
general forMalIsM
-----------------
How are the stATeS of a quantuM fiEld affected bY an Object (SuCh aS A beamSplItter) That is UNiformLy accelerAtED in the $r$ Wedge? ThIS Is The qUestion of central iNTeREst in this paper. a straiGhTFoRWArd Way To study thiS pRobleM Is to worK In THE AccELerated frame iN which the obJEct Is statIc. it iS ObviouS that ThE ObjEct only inteRactS with RindLer modES in the $R$ WEdge and The RinDleR moDes iN ThE $L$ WedGe REmaIN uNafFEctEd. The intErAcTion bEtweEN THE objEct And tHe RinDler modes is unItaRy anD It tRansfOrms tHe RiNdLer moDes as $$\lAbel{gEnEralcoupling}
\hat{B}_{mk}^{\pRime R} = \int d \OmeGa \BigG( \aLpha^{m1}_{K \Omega} \hAt{b}_{1\OmeGa}^R
+ \beta^{M1}_{k \omega} \HAt{b}_{1\OmEGA}^{r \dAgger}
+ \alpha^{m2}_{k \omega} \hAt{B}_{2 \OMeGa}^R
+ \beta^{m2}_{K \omega} \HAt{B}_{2\oMEga}^{R \daggEr} \BigG). \\$$ ThiS IS the mOst gENeRal interAction WHiCh Not only CoUples tHe LefT-moVing aND rigHt-moviNg RindleR modeS But also Rindler MOdes with diffeREnT FReQUencIes. together witH Eqs. (\[uNruhrindLEr\]) And (\[uNruhRIndleR:iNVeRSe\]), we can construct a quAnTum cirCuit mOdel (or input-ouTput formalISM) For the unIforMLy ACcelerated objeCt. We sTart from thE Inertial Frame In which UNruh modes ARE used insTeaD of minKowSKI mOdes. This makes THE modEl Simpler AltHough we StiLl hAve To tRaNsform the minkowskI mOdEs To The unruh MOdes and vIcE veRsA.
FiRst, baSEd on Eq. (\[unruhrindLeR:iNVerSe\]), the UnRUh MODes pAsS tHrouGh a CoLlectIon oF Two-Mode squEezers eacH of WHich CoUpLes a paiR of Unruh modes $(\HaT{c}_{m\omega}, \haT{d}_{M\ | in the upper-half complex plane. Th e Unr uhmod es ann ihil ate the Minkow s ki v acuum state $$\begin{a ligne d} \hat { c} _{m\o mega} | 0_ M \ra ng le =\h a t{ d}_{m \om ega} |0_M \rangl e = 0 \end{aligned } $$ as notedabo ve.
Circuit mo del {# ci rcu i t}
== === ===== ===
G e neralformalism
- - ------ - ------- -
H ow a re the states ofa q u antum field af fected b y a n obj ect (such asabeams p litter) th a t isu niformly acce lerated int he$R$ we dg e?T his is thequ e sti on of centr al i nterest i n this paper.A straig htforw ard wa y to st ud y t hi s pr o bl emi s t o work i nth e acc eler a t e d fra mein w hichthe object is st atic . It is o bviou s th at theobject only i nteracts with R indl er modesinth e $ R$ wedg e and t heRin dler mo des int he$L $ w ed ge remain unaffect ed . Th e intera ctionb et we e n the ob je ctandt h e Rin dler mo des is u nitary an dit tran sf orms t he Ri ndl er mo d es a s $$\l abel{gen eralc o upling}
\hat{b } _{mk}^{\prime R} = \ i nt d \o mega \bigg( \al p ha^{ m1}_ { k\om e ga} \ hat{b }_ { 1\ o mega}^R
+ \beta^{m 1} _{k \o mega} \hat{b}_{1\o mega}^{R \ d a g ger}
+\alp h a^ { m2}_{k \omega} \hat {b}_{2 \om e ga}^R
+ \bet a^{m2}_{ k \omega} \ hat{b}_{ 2\o meg a}^ {R\ d ag ger} \bigg).\ \ $$ T hi s is th e m ost gen era l i nte rac ti on whichnot only c ou pl es th e lef t -movingan d r ig ht- movin g Rindl er mo desbu ta lso Rindle r m o d es w it hdiff ere nt freq uenc i es. Togeth er with E qs. (\[U nr uh Rindler \]) and (\[Un ru hRindler:i nv ers e\]),w e can con struct a quantum circui t model(or inpu t-ou tput form ali sm) fo r t h e unif ormlyaccel er ate d objec t . W e s ta rt from th e ine rtial f rame in whi ch Unruh modes are use d instead ofMin kows k i m ode s .T his m a kes t he model simple r although w e s till havet o t ra nsformthe Min kowsk i modesto the Un ruh modes a nd v i c e v ersa.
Fir st, base d on Eq.( \[Unr u hR indle r:i nverse \] ),the U nruh m o des pass throu gh a col lecti on of two- mode squeezers each ofwhichcoupl esa pair of Un r uhmodes $(\ hat{ c}_{m\omeg a}, \h at{d} _{m \ | in_the upper-half_complex plane. The Unruh_modes annihilate_the_Minkowski vacuum_state_$$\begin{aligned}
\hat{c}_{m\omega} | 0_M_\rangle = \hat{d}_{m\omega}_| 0_M \rangle =_0\end{aligned}$$ as noted_above.
Circuit_model {#circuit}
=============
General formalism
-----------------
How are the states of a quantum field affected by an object_(such_as a_beamsplitter)_that_is uniformly accelerated in the_$R$ wedge? This is the_question of_central interest in this paper. A straightforward way_to_study this problem_is to work in the accelerated frame in which_the object is static. It is_obvious that the_object_only_interacts with Rindler modes_in the $R$ wedge and the_Rindler modes in the $L$ wedge_remain unaffected. The interaction between the object_and the Rindler modes is unitary_and it transforms the Rindler_modes as_$$\label{generalcoupling}
\hat{b}_{mk}^{\prime R} = \int d_\omega \bigg( \alpha^{m1}_{k_\omega} \hat{b}_{1\omega}^R_
+ \beta^{m1}_{k \omega}_\hat{b}_{1\omega}^{R \dagger}
+ \alpha^{m2}_{k \omega} \hat{b}_{2_\omega}^R
+ \beta^{m2}_{k_\omega} \hat{b}_{2\omega}^{R \dagger} \bigg). \\$$ This_is_the most general_interaction_which_not only_couples the left-moving_and_right-moving Rindler_modes_but also Rindler modes with different_frequencies._Together with Eqs. (\[UnruhRindler\]) and (\[UnruhRindler:inverse\]), we_can construct a quantum_circuit_model (or input-output formalism)_for the uniformly accelerated object._We start from the inertial frame_in which_Unruh modes_are used instead of Minkowski modes. This makes the model simpler_although we still have to transform_the Minkowski modes to_the Unruh_modes_and vice versa.
First,_based_on Eq._(\[UnruhRindler:inverse\]), the Unruh modes pass through a_collection of_two-mode squeezers each of which couples_a pair of Unruh_modes_$(\hat{c}_{m\omega}, \hat{d}_{m\ |
focal plane by dividing the images by a master flat-field image. We adopted the flat-fielding method that was used by the Kilo Degree Survey (KiDS, [@deJong2015]): the master flat-field is achieved by first median-combining and normalizing 8 dome flat-fields and 8 twilight flat-fields, and then multiplying the averaged flat-fields with each other. High-frequency spatial Fourier modes are corrected using the dome flats and low frequencies using the twilight flats. This is based on the pre-assumption that the large-scale illumination of the twilight flat-field matches better the observational situation than that obtained from the dome. On the other hand, in the dome flats the [*S/N*]{} is high, which can be used to capture the pixel-by-pixel variations in the pixel sensitivity.
However, even after applying the flat-field correction small systematic flux variations remain across the instrument. These variations can be corrected by applying an illumination correction. We use the correction models made for the KiDS (see [@Verdoes-Kleijn2013] for details). The models are made by mapping the photometric residuals across the CCD array using a set of dithered Landolt Selected Area (SA) field [@Landolt1992] observations, and fitting a linear model to the residuals. The images are multiplied with this illumination correction. The correction is applied after sky subtraction (see the next sub-section), to avoid the sky residuals being amplified by the illumination correction procedure.
Background subtraction and de-fringing
--------------------------------------
The images contain sky background flux composed of direct and scattered atmospheric emission, and scattered light of bright celestial sources. A careful removal of the atmospheric background light is essential when studying LSB objects such as UDGs. When considering a single image at low luminosity levels, we cannot directly tell which part of the light is diffuse light coming from the sources and which is background light. To bypass the problem, we can make an assumption that the pattern of background light stays constant if the telescope pointing direction is not changed by more than a few degrees. Due to the large dithers between the consecutive integrations the objects are not likely to appear twice in the same pixel, which allows us to produce background models by averaging a stack of images. The intensity of the sky changes thorough the night (1–10% between exposures), especially at the beginning and at the end of the night, which forces to scale the images used for the background model before | focal plane by dividing the images by a master bland - sphere image. We adopted the flat - fielding method acting that was used by the Kilo Degree Survey (KiDS, [ @deJong2015 ] ): the master flat - plain is achieved by first median - combining and normalizing 8 attic bland - fields and 8 twilight bland - fields, and then multiplying the average bland - fields with each early. High - frequency spatial Fourier mode are corrected using the dome flats and broken frequencies using the twilight flats. This is based on the pre - premise that the large - scale illumination of the twilight flat - field match better the observational situation than that obtained from the dome. On the early hired hand, in the dome flats the [ * S / N * ] { } is high, which can be used to capture the pixel - by - pixel variations in the pixel sensitivity.
However, even after applying the flat - field correction minor systematic flux variation persist across the instrument. These variations can be corrected by practice an illumination correction. We use the correction model made for the KiDS (see [ @Verdoes - Kleijn2013 ] for details). The models are made by mapping the photometric residual across the CCD array using a set of dithered Landolt Selected Area (SA) field [ @Landolt1992 ] observations, and meet a analogue model to the residuals. The images are multiplied with this illumination correction. The correction is applied after sky subtraction (examine the next sub - section), to avoid the sky residual being amplified by the clarification correction procedure.
setting subtraction and de - fringe
--------------------------------------
The images incorporate sky background flux density composed of direct and scattered atmospheric emission, and scattered lighter of bright celestial sources. A careful removal of the atmospheric background light is essential when studying LSB objects such as UDGs. When considering a single persona at low luminosity levels, we cannot immediately tell which part of the light is diffuse light coming from the sources and which is background lighter. To bypass the problem, we can make an assumption that the pattern of background light stay changeless if the telescope sharpen direction is not changed by more than a few degrees. Due to the large dithers between the consecutive integration the objects are not likely to appear doubly in the same pixel, which admit us to produce background models by average a stack of images. The intensity of the sky changes thorough the night (1–10% between exposures), particularly at the beginning and at the end of the night, which forces to scale the images used for the background mannequin before | fofal plane by dividing tht images by a masjee flat-hield ijage. We xdopted the flat-fielding metiod rhat qas used by the Kilo Ddgree Surney (KiDS, [@eeJoig2015]): the master flef-field lf acgleved uy first median-gombining atd normalizing 8 aole flat-fields and 8 twilight flat-fiejds, and tjen multiplyind tht aderafvd flat-fields with each other. Hjgh-freqlency spatial Foutier modes are corrected uding the dome flats anf low frequghciqw using the gwilight flats. This is based on the pre-assumption thag the large-scalg jlpomination of the nwilight flat-npeld madches brtter the obsevvatimnao situation than that obtained from the dome. On dhz other hand, in the dimw flajs tha [*S/N*]{} us figg, xhidh can be used to czpture the pixel-by-pixel variatoogw in the pixem sensytyvity.
However, even after applying the fldt-fjeld correction small ststematic flux variatlons remayn across the instrument. These variations can be worrertdd yn xpppying an illumination correction. We use the cjdrtctpon models made fjr the KiDS (sfe [@Derdoes-Kleijn2013] for dzfajls). The models are made bi mappung the protoketric residuals across the CCD array ufung a set of dithexed Landolt Relevted Srea (SA) field [@Landolt1992] oyservafions, and fltting a mknear model to tfe gesiguals. Tht images are multi[lied witi thix illumknatoon cowrection. Tje covsection is applied aftet sky vubtractioj (see the next sub-section), to avoid the sky rexiguans being ampllfied by the iljumination cortection pxocedufe.
Backgrouhd subtcaction and qe-fringing
--------------------------------------
The lmages contamn sky baskgriund flux cuoposed of direvt and sccutered atmispheric emission, end rdattered light if vright celestiak suursed. E carqxul removal mf tfe xymospferic backgvouvd loght is essential whan sfudying LSB objectx fuch as ODGs. When considering s single image at pow lnminosmty lefelf, we cannot directly tell whicg part of thc light is disfusc lidht coming from the sources and which is backgrouid light. To bypass the problem, we can make sn assumptipn thwt the padtern of background oight stays constsnt if the telescope lointitg digection is not changed by more than a few degrees. Due to the large ditherw betwxeg the consedutife innegxatnons thq oboerts are not likelj to appear twice in the same pieel, which dljows us to produce backgrounc oodels by avetaging a stack of images. Tge intenxity of the sky changes thorougn the night (1–10% between exppsures), ewpecizlly an tie beginning and at yhe enb of tye nighr, whlch forces to rcake the imagfs uxeb for the background model before | focal plane by dividing the images by flat-field We adopted flat-fielding method that Degree (KiDS, [@deJong2015]): the flat-field is achieved first median-combining and normalizing 8 dome and 8 twilight flat-fields, and then multiplying the averaged flat-fields with each other. spatial Fourier modes are corrected using the dome flats and low frequencies using twilight This based the pre-assumption that the large-scale illumination of the twilight flat-field matches better the observational situation than obtained from the dome. On the other hand, the dome flats the is high, which can be to the pixel-by-pixel in pixel However, even after the flat-field correction small systematic flux variations remain across the instrument. These variations can be corrected by an illumination use the models for KiDS (see [@Verdoes-Kleijn2013] The models are made by mapping across the CCD array using a set of Landolt Selected (SA) field [@Landolt1992] observations, and fitting linear model to the residuals. The images are with this illumination correction. The correction is applied after sky subtraction (see the next sub-section), the sky residuals being by the illumination procedure. subtraction de-fringing The images sky background flux composed of direct and scattered atmospheric emission, and light of bright celestial sources. A careful removal of the light essential when studying objects such as UDGs. considering single image at low we directly of light diffuse light coming from sources and which is background To bypass the problem, that the pattern of background light stays constant the telescope pointing direction is not changed more than a few degrees. Due to the large dithers between the integrations the not likely to appear twice in the same which allows us to background models by averaging a stack of images. The of sky changes the night (1–10% exposures), especially at beginning and at of the which to used for the background model before | focal plane by dividing the imAges by a masTer flAt-fIelD iMage. we adOpted the flat-fiELdinG method that was used by thE Kilo deGRee SURvEy (KiDs, [@deJong2015]): THe MASteR fLaT-fiElD Is AchieVed By first Median-combIniNg And normaliziNG 8 dOme flat-fieLds And 8 twilight fLat-Fields, AnD thEN multIplYing tHe averAGed flaT-fields wiTh EAch othER. High-frEQUeNcy sPatial Fourier modeS ArE Corrected using The domE fLAtS ANd lOw fRequencies UsIng thE TwilighT FlATS. thiS Is based on the pRe-assumptioN ThaT the laRgE-scALe illuMinatIoN Of tHe twilight fLat-fIeld matchEs bettER the obsERvationAl situAtiOn tHan tHAt ObTaiNeD FroM ThE doME. On The other HaNd, In the Dome FLATS the [*s/N*]{} iS higH, whicH can be used to cAptUre tHE piXel-by-Pixel VariAtIons iN the piXel seNsItivity.
However, eVen aFter applyIng ThE flAt-Field COrrectIon SmaLl systeMatic flUX vaRiATIOnS remain across the inStRUMeNt. These vAriatiONs CaN Be correcTeD by ApplYINg an iLlumINaTion corrEction. wE uSe The corrEcTion moDeLs mAde For thE kiDS (See [@VerDoes-KleiJn2013] for DEtails). The modelS Are made by mappINg THE pHOtomEtrIc residuals AcroSS the cCD aRRaY usINg a seT of diThEReD landolt Selected Area (sA) Field [@LAndolT1992] observations, And fitting A LINear modeL to tHE rESiduals. The imagEs are Multiplied WIth this iLlumiNation coRrection. THE CorrectiOn iS apPliEd aFTEr Sky subtractioN (SEe thE nExt sub-sEctIon), to avOid The Sky ResIdUals being AmplifieD bY tHe IlLumInatiON correctIoN prOcEduRe.
BacKGround SubtrActiOn AnD De-fRinging
--------------------------------------
tHe IMAges CoNtAin sKy bAcKgrouNd flUX coMposed oF direct anD scATterEd AtMospherIc emission, and ScAttered ligHt Of bRight cELEstial soUrces. A careful removal of tHE atmospHerIc bacKgroUnd light iS esSentiaL whEN studyIng LSB ObjecTs SucH AS UDGs. wHEn ConSiDering a sinGLE imAge at LoW lumInosity Levels, we cannot direCTly Tell which part Of tHe liGHT iS diFFuSE liGhT ComING from the sources And which is BaCKgRound light. tO byPaSs the prOblem, we Can maKE an assuMption thaT the patteRn Of baCKGroUnd light stAys constAnt if the tELescoPE pOintiNg dIrectiOn Is nOt chaNged by MOre Than a Few degReEs. Due tO the lArGe ditherS between the consecutive iNtegraTions The Objects arE noT LikEly to appeAr twIce in the saMe pIxeL, whicH alLOws us To prODuCe bACkgroUnd mODels by aveRAgIng A STaCk of images. THE INteNsity Of tHE sky chAngeS thorough the night (1–10% BEtween exposureS), espECIalLy aT The bEgInning and at the End Of THE night, whIcH forces to scAle the imAgES used For the BackgrOund modEL BeFOre | focal plane by dividing t he imagesby amas ter f lat- fiel d image. We ad o pted the flat-fielding met hod t ha t was us ed by the Ki l oD e gre eSu rve y( Ki DS, [ @de Jong201 5]): the m ast er flat-fieldi sachieved b y f irst median- com bining a ndn ormal izi ng 8dome f l at-fie lds and 8 t w ilight flat-fi e l ds , an d then multiplyin g t h e averaged fla t-fiel ds wi t h ea chother. Hig h- frequ e ncy spa t ia l F our i er modes arecorrected u s ing the d om e f l ats an d low f r equ encies usin g th e twiligh t flat s . Thisi s based on th e p re- assu m pt io n t ha t th e l arg e -sc ale illu mi na tionof t h e t wili ght fla t-fie ld matches be tte r th e ob serva tiona l si tu ation thanthatob tained from the dom e. On the ot he r h an d, in the do mefla ts the[*S/N*] { } i sh i g h, which can be used t o ca pture th e pixe l -b y- p ixel var ia tio ns i n the p ixel se nsitivit y.
Ho w ev er , evenaf ter ap pl yin g t he fl a t-fi eld co rrection smal l systematic fl u x variationsr em a i na cros s t he instrume nt.T hese var i at ion s canbe co rr e ct e d by applying an il lu minati on co rrection. Weuse the co r r e ction mo dels ma d e for the KiDS (see [@Verdoes - Kleijn20 13] f or detai ls). Them o dels are ma debymap p i ng the photomet r i c re si duals a cro ss theCCD ar ray us in g a set o f dither ed L an do ltSelec t ed Area(S A)fi eld [@La n dolt19 92] o bser va ti o ns, and fi t ti n g a l in ea r mo del t o the res i dua ls. The images a rem ulti pl ie d withthis illumina ti on correct io n.The co r r ection i s applied after sky sub t raction (s ee th e ne xt sub-se cti on), t o a v oid th e skyresid ua lsb e ing a m p li fie dby the ill u m ina tionco rrec tion pr ocedure.
Backgrou n d s ubtraction an d d e-fr i n gi ng- -- - --- -- - --- - - --------------- --------
Th e i mages cont a insk y backg round f lux c o mposedof direct and scat te reda t mos pheric emi ssion, a nd scatte r ed li g ht of b rig ht cel es tia l sou rces.A ca reful remov al of th e atm os pheric b ackground light is esse ntialwhenstu dying LSB ob j ect s such as UDG s. When co nsi der ing a si n gle i mage at lo w lumi nosi t y levels, we ca n n ot directly t e l l wh ich p art of the lig ht is diffuse lig h t coming fromthes o urc esa nd w hi ch is backgrou ndli g h t. To by pa ss the prob lem, weca n make an as sumpti on that t he patter n of ba ckgroundlig ht stays c on st a nt ifthete lescop e poin t ingd i rection is not c hange d by mo r e t han a f ew degr e es.Due to the large dith ers be twee n the consec ut ive in teg ra tions theo bjects ar e not likely t o ap pea r twic e in t he sa me p ix el, which al l o ws us t o pr oduc e bac kg roun d modelsb y averag ing a stack o f i m a ges. T h ei n tensity of th e sky c hanges tho r ough th e nigh t (1–1 0% bet ween ex p osu re s), esp eci a l ly at the beginnin g an dat t he end o ftheni ght ,which forces to scale t he image s used f ort he b a ckgrou nd model befor e | focal_plane by_dividing the images by_a master_flat-field_image. We_adopted_the flat-fielding method_that was used_by the Kilo Degree_Survey (KiDS, [@deJong2015]):_the_master flat-field is achieved by first median-combining and normalizing 8 dome flat-fields and 8_twilight_flat-fields, and_then_multiplying_the averaged flat-fields with each_other. High-frequency spatial Fourier modes_are corrected_using the dome flats and low frequencies using_the_twilight flats. This_is based on the pre-assumption that the large-scale illumination_of the twilight flat-field matches better_the observational situation_than_that_obtained from the dome._On the other hand, in the_dome flats the [*S/N*]{} is high,_which can be used to capture the_pixel-by-pixel variations in the pixel sensitivity.
However,_even after applying the flat-field_correction small_systematic flux variations remain across_the instrument. These_variations can_be corrected by_applying an illumination correction. We use_the correction models_made for the KiDS (see [@Verdoes-Kleijn2013]_for_details). The models_are_made_by mapping_the photometric residuals_across_the CCD_array_using a set of dithered Landolt_Selected_Area (SA) field [@Landolt1992] observations, and fitting_a linear model to_the_residuals. The images are_multiplied with this illumination correction._The correction is applied after sky_subtraction (see_the next_sub-section), to avoid the sky residuals being amplified by the illumination_correction procedure.
Background subtraction and de-fringing
--------------------------------------
The images_contain sky background flux_composed of_direct_and scattered atmospheric_emission,_and scattered_light of bright celestial sources. A careful_removal of_the atmospheric background light is essential_when studying LSB objects_such_as UDGs. When considering a single_image at low luminosity levels, we_cannot directly tell which part_of_the_light is diffuse light coming_from the sources and which is_background light. To_bypass the problem, we can make an_assumption_that the pattern of background light_stays_constant if the telescope pointing direction_is_not_changed by more than a_few degrees. Due to the large_dithers between the consecutive integrations the objects are not_likely to appear_twice in the same pixel,_which_allows_us to produce background models by averaging a stack of_images. The_intensity of the_sky changes thorough the night (1–10% between exposures), especially at_the beginning and at the end of_the night, which forces to scale the images used for the_background model before |
the unit Euclidean norm eigenvector associated with $\lambda_{n,j}$. Then the empirical probability transition matrix $P_n$ has the decomposition $$\begin{aligned}
P_{n}^t = \sum_{j=0}^{n-1} \lambda_{n,j}^t\, \psi_{n,j}\,\varphi_{n,j}^T,\end{aligned}$$ where $\psi_{n,j} = D_n^{-1/2}\,\phi_{n,j}\in{\mathbb{R}}^n$ and $\varphi_{n,j} = D_n^{1/2}\phi_{n,j}\in{\mathbb{R}}^n$, so that $\psi_{n,j}=D_n^{-1} \varphi_{n,j}$ for each $j\in\{0,1,\dots,n-1\}$. In particular, $\psi_{n,j}$ has unit $L^2(\mbox{diag}(D_n))$ norm, and $\varphi_{n,j}$ has unit $L^2(\mbox{diag}(D_n^{-1}))$ norm, for each $j\in\{0,1,\dots,n-1\}$.
In addition, we have the following relation between the diffusion affinity and $P_n^{2t}$, $$\begin{aligned}
\langle X_i,\, X_j\rangle_{D_{n,t}} = \sum_{l=0}^{n-1} \lambda_{n,l}^{2t} \,[\psi_{n,l}]_i \, [\psi_{n,l}]_j =\sum_{l=0}^{n-1} \lambda_{n,l}^{2t} \, [\psi_{n,l}]_i \, [\varphi_{n,l}]_j d_n^{-1}(X_j)
=[P_n^{2t}D_n^{-1}]_{ij}.\end{aligned}$$
Technical proofs
================
In this appendix, we collect some technical lemmas used in the proofs of our main results.
\[lem:some\_ineq\_feasible\_set\] Let $Z^{*}$ be defined in (\[eqn:Kmeans\_true\_membership\_matrix\]). Then for any $Z \in {\mathscr{C}}$ defined in, we have $$\begin{aligned}
\label{eqn:ineq_1_ | the unit Euclidean norm eigenvector associated with $ \lambda_{n, j}$. Then the empirical probability transition matrix $ P_n$ take the decay $ $ \begin{aligned }
P_{n}^t = \sum_{j=0}^{n-1 } \lambda_{n, j}^t\, \psi_{n, j}\,\varphi_{n, j}^T,\end{aligned}$$ where $ \psi_{n, j } = D_n^{-1/2}\,\phi_{n, j}\in{\mathbb{R}}^n$ and $ \varphi_{n, j } = D_n^{1/2}\phi_{n, j}\in{\mathbb{R}}^n$, so that $ \psi_{n, j}=D_n^{-1 } \varphi_{n, j}$ for each $ j\in\{0,1,\dots, n-1\}$. In especial, $ \psi_{n, j}$ have unit $ L^2(\mbox{diag}(D_n))$ norm, and $ \varphi_{n, j}$ have unit $ L^2(\mbox{diag}(D_n^{-1}))$ norm, for each $ j\in\{0,1,\dots, n-1\}$.
In addition, we have the following relation back between the diffusion affinity and $ P_n^{2t}$, $ $ \begin{aligned }
\langle X_i,\, X_j\rangle_{D_{n, t } } = \sum_{l=0}^{n-1 } \lambda_{n, l}^{2 t } \,[\psi_{n, l}]_i \, [ \psi_{n, l}]_j = \sum_{l=0}^{n-1 } \lambda_{n, l}^{2 t } \, [ \psi_{n, l}]_i \, [ \varphi_{n, l}]_j d_n^{-1}(X_j)
=[ P_n^{2t}D_n^{-1}]_{ij}.\end{aligned}$$
Technical proofs
= = = = = = = = = = = = = = = =
In this appendix, we roll up some technical lemmas used in the proofs of our main results.
\[lem: some\_ineq\_feasible\_set\ ] permit $ Z^{*}$ be defined in (\[eqn: Kmeans\_true\_membership\_matrix\ ]). Then for any $ Z \in { \mathscr{C}}$ defined in, we consume $ $ \begin{aligned }
\label{eqn: ineq_1 _ | thf unit Euclidean norm einenvector associcred wivh $\lambsa_{n,j}$. Thev the empirical probability vranwitiob matrix $P_n$ has the dezompositiln $$\begin{qligied}
P_{n}^t = \sum_{j=0}^{n-1} \lambda_{n,j}^t\, \ifi_{n,j}\,\bwrphn_{n,o}^T,\end{aligned}$$ whgre $\psi_{n,j} = D_t^{-1/2}\,\phi_{n,j}\in{\mathbb{S}}^n$ aud $\varphi_{n,j} = D_n^{1/2}\phi_{n,j}\in{\mathbb{R}}^n$, so thwt $\psi_{n,k}=D_j^{-1} \varphi_{n,j}$ for eacn $j\in\{0,1,\sots,n-1\}$. In particular, $\psi_{n,j}$ has unit $L^2(\mbox{dpag}(D_n))$ norm, and $\vatphi_{n,j}$ has unit $L^2(\mbox{diag}(D_j^{-1}))$ nogm, for each $j\in\{0,1,\dotd,n-1\}$.
In additiin, wq have the foulowing relation betwegn the diffusion affinity and $P_n^{2g}$, $$\begnn{aligned}
\labgoe D_h,\, X_j\rangle_{D_{i,t}} = \sui_{l=0}^{n-1} \lambda_{n,l}^{2b} \,[\psi_{n,l}]_h \, [\psi_{n,k}]_j =\sum_{l=0}^{n-1} \lambds_{n,l}^{2v} \, [\pwi_{n,l}]_i \, [\varphi_{n,l}]_j d_n^{-1}(X_j)
=['_n^{2t}D_n^{-1}]_{ij}.\end{aligned}$$
Tecrnical prmoys
================
In this appendix, we cillecj soma tezynizal lxmmzs usef ii the proofa of our maun results.
\[lem:some\_intq\_fqqsible\_set\] Let $Z^{*}$ be qesined in (\[eqn:Kmeans\_true\_membership\_matrix\]). Uhen ror any $Z \in {\mathscr{C}}$ dwfined in, we have $$\begln{aligned}
\jabel{eqn:ineq_1_ | the unit Euclidean norm eigenvector associated with the probability transition $P_n$ has the \lambda_{n,j}^t\, where $\psi_{n,j} = and $\varphi_{n,j} = so that $\psi_{n,j}=D_n^{-1} \varphi_{n,j}$ for each In particular, $\psi_{n,j}$ has unit $L^2(\mbox{diag}(D_n))$ norm, and $\varphi_{n,j}$ has unit $L^2(\mbox{diag}(D_n^{-1}))$ norm, each $j\in\{0,1,\dots,n-1\}$. In addition, we have the following relation between the diffusion affinity $P_n^{2t}$, \langle X_j\rangle_{D_{n,t}} \sum_{l=0}^{n-1} \lambda_{n,l}^{2t} \,[\psi_{n,l}]_i \, [\psi_{n,l}]_j =\sum_{l=0}^{n-1} \lambda_{n,l}^{2t} \, [\psi_{n,l}]_i \, [\varphi_{n,l}]_j d_n^{-1}(X_j) =[P_n^{2t}D_n^{-1}]_{ij}.\end{aligned}$$ Technical proofs ================ this appendix, we collect some technical lemmas used the proofs of our results. \[lem:some\_ineq\_feasible\_set\] Let $Z^{*}$ be in Then for $Z {\mathscr{C}}$ in, we have \label{eqn:ineq_1_ | the unit Euclidean norm eigenVector assoCiateD wiTh $\lAmBda_{n,J}$. TheN the empirical pRObabIlity transition matrix $P_N$ has tHe DEcomPOsItion $$\Begin{alIGnED}
p_{n}^t = \SuM_{j=0}^{N-1} \laMbDA_{n,J}^t\, \psi_{N,j}\,\vArphi_{n,j}^t,\end{aligneD}$$ whErE $\psi_{n,j} = D_n^{-1/2}\,\phi_{n,J}\In{\Mathbb{R}}^n$ anD $\vaRphi_{n,j} = D_n^{1/2}\phi_{n,J}\in{\Mathbb{r}}^n$, So tHAt $\psi_{N,j}=D_N^{-1} \varpHi_{n,j}$ foR Each $j\iN\{0,1,\dots,n-1\}$. In pArTIcular, $\PSi_{n,j}$ has UNIt $l^2(\mboX{diag}(D_n))$ norm, and $\varPHi_{N,J}$ has unit $L^2(\mbox{dIag}(D_n^{-1}))$ nOrM, FoR EAch $J\in\{0,1,\Dots,n-1\}$.
In addItIon, we HAve the fOLlOWINg rELation between The diffusioN AffInity aNd $p_n^{2t}$, $$\BEgin{alIgned}
\LaNGle x_i,\, X_j\rangle_{D_{N,t}} = \suM_{l=0}^{n-1} \lambda_{N,l}^{2t} \,[\psi_{N,L}]_i \, [\psi_{n,l}]_J =\Sum_{l=0}^{n-1} \laMbda_{n,l}^{2T} \, [\psI_{n,l}]_I \, [\varPHi_{N,l}]_J d_n^{-1}(x_j)
=[p_N^{2t}D_N^{-1}]_{Ij}.\End{ALigNed}$$
TechnIcAl ProofS
================
In tHIS APpenDix, We coLlect Some technical LemMas uSEd iN the pRoofs Of ouR mAin reSults.
\[lEm:somE\_iNeq\_feasible\_set\] LEt $Z^{*}$ bE defined iN (\[eqN:KMeaNs\_True\_mEMbershIp\_mAtrIx\]). Then fOr any $Z \iN {\MatHsCR{c}}$ DeFined in, we have $$\begin{AlIGNeD}
\label{eqN:ineq_1_ | the unit Euclidean norm e igenvector asso cia ted w ith$\la mbda_{n,j}$. T h en t he empirical probabili ty tr an s itio n m atrix $P_n$h as t hede co mpo si t io n $$\ beg in{alig ned}
P_{n} ^t=\sum_{j=0}^{ n -1 } \lambda_ {n, j}^t\, \psi_ {n, j}\,\v ar phi _ {n,j} ^T, \end{ aligne d }$$ wh ere $\psi _{ n ,j} =D _n^{-1/ 2 } \, \phi _{n,j}\in{\mathbb { R} } ^n$ and $\varp hi_{n, j} =D _ n^{ 1/2 }\phi_{n,j }\ in{\m a thbb{R} } ^n $ , sot hat $\psi_{n, j}=D_n^{-1} \va rphi_{ n, j}$ for ea ch $j \i n \{0 ,1,\dots,n- 1\}$ . In part icular , $\psi_ { n,j}$ h as uni t $ L^2 (\mb o x{ di ag} (D _ n)) $ n orm , an d $\varp hi _{ n,j}$ has u n i t $L ^2( \mbo x{dia g}(D_n^{-1})) $ n orm, for each $j\i n\{0 ,1 ,\dot s,n-1\ }$.
In addition, we h avethe follo win grel at ion b e tweenthe di ffusion affini t y a nd $ P _n ^{2t}$, $$\begin{a li g n ed }
\langl e X_i, \ ,X_ j \rangle_ {D _{n ,t}} = \sum _{l= 0 }^ {n-1} \l ambda_ { n, l} ^{2t} \ ,[ \psi_{ n, l}] _i\, [\ p si_{ n,l}]_ j =\sum_ {l=0} ^ {n-1} \lambda_ { n,l}^{2t} \,[ \p s i _{ n ,l}] _i\, [\varphi _{n, l }]_j d_n ^ {- 1}( X _j)
= [P_n^ {2 t }D _ n^{-1}]_{ij}.\end{a li gned}$ $
Te chnical proof s
======== = = = =====
I n th i sa ppendix, we co llect some tech n ical lem mas u sed in t he proofs o f our ma inres ult s.\ [l em:some\_ineq \ _ feas ib le\_set \]Let $Z^ {*} $ b e d efi ne d in (\[e qn:Kmean s\ _t ru e\ _me mbers h ip\_matr ix \]) .The n for any $Z \in{\ma th sc r {C} }$ defi n ed i n, w eha ve $ $\b eg in{al igne d }
\ label{e qn:ineq_1 _ | the_unit Euclidean_norm eigenvector associated with_$\lambda_{n,j}$. Then_the_empirical probability_transition_matrix $P_n$ has_the decomposition $$\begin{aligned}
P_{n}^t_= \sum_{j=0}^{n-1} \lambda_{n,j}^t\, \psi_{n,j}\,\varphi_{n,j}^T,\end{aligned}$$_where $\psi_{n,j} =_D_n^{-1/2}\,\phi_{n,j}\in{\mathbb{R}}^n$_and $\varphi_{n,j} = D_n^{1/2}\phi_{n,j}\in{\mathbb{R}}^n$, so that $\psi_{n,j}=D_n^{-1} \varphi_{n,j}$ for each $j\in\{0,1,\dots,n-1\}$. In particular, $\psi_{n,j}$_has_unit $L^2(\mbox{diag}(D_n))$_norm,_and_$\varphi_{n,j}$ has unit $L^2(\mbox{diag}(D_n^{-1}))$ norm,_for each $j\in\{0,1,\dots,n-1\}$.
In addition, we_have the_following relation between the diffusion affinity and $P_n^{2t}$,_$$\begin{aligned}
\langle_X_i,\, X_j\rangle_{D_{n,t}} =_\sum_{l=0}^{n-1} \lambda_{n,l}^{2t} \,[\psi_{n,l}]_i \, [\psi_{n,l}]_j =\sum_{l=0}^{n-1} \lambda_{n,l}^{2t} \, [\psi_{n,l}]_i_\, [\varphi_{n,l}]_j d_n^{-1}(X_j)
=[P_n^{2t}D_n^{-1}]_{ij}.\end{aligned}$$
Technical proofs
================
In this appendix,_we collect some_technical_lemmas_used in the proofs_of our main results.
\[lem:some\_ineq\_feasible\_set\] Let $Z^{*}$_be defined in (\[eqn:Kmeans\_true\_membership\_matrix\]). Then for_any $Z \in {\mathscr{C}}$ defined in, we_have $$\begin{aligned}
\label{eqn:ineq_1_ |
10^{-2}$ & $7.5^{+2.3}_{-1.8}$ & 2.1 &$ 80\pm 17$\
$\geq 32$ & 1,993 & $25^{+9}_{-6}$ & 7.6 & $151 \pm 17$ & $4.1\times 10^{-3}$ & $13^{+5}_{-3}$ & 4.1 & $152 \pm 19$\
------------------------------------------------------------------------
$\geq 8$ & 36,928 & $6.6^{+2.0}_{-1.5}$ & 1.8& $132 \pm 15$ & $8.6\times 10^{-4}$ & $6.0^{+1.0}_{-0.9}$ & 0.94 & $98 \pm 9$\
Ahlers, M. 2019,, 886, L18
Al Samarai, I. for the Pierre Auger Collaboration 2016, [PoS ICRC2015]{}, 372
Bonino, R. et al. 2011,, 738, 67
Candia, J., Mollerach, S. and Roulet, E. 2003,, 05, 003
Calvez, A., Kusenko, A. and Nagataki, S. 2010,, 105, 091101
Castellina, A. for the Pierre Auger Collaboration 2019, [EPJ Web Conf.]{}, 210, 06002
Coleman, A. for the Pierre Auger Collaboration 2019, [PoS ICRC2019]{}, 225
Compton, A.H. and Getting, I.A. 1935, PhRv, 47, 817
Farley, F.J.M. and Storey, J.R. 1954, [Proc. Phys. Soc.]{} A, 67, 996
Feretti, L. et al. 2012,, 20, 54
Greisen, K. 1966,, 16, 748
Haverkorn, M. 2015, [Astrophys. Space Sci. Library]{}, 407, 483
IceCube Collaboration 2012,, 746, 33
IceCube Collaboration 2016,, 826, 220
Kachelriess, M. and Serpico, P.D. 2006, [Phys. Lett.]{} B, 640, 225
KASCADE- | 10^{-2}$ & $ 7.5^{+2.3}_{-1.8}$ & 2.1 & $ 80\pm 17$\
$ \geq 32 $ & 1,993 & $ 25^{+9}_{-6}$ & 7.6 & $ 151 \pm 17 $ & $ 4.1\times 10^{-3}$ & $ 13^{+5}_{-3}$ & 4.1 & $ 152 \pm 19$\
------------------------------------------------------------------------
$ \geq 8 $ & 36,928 & $ 6.6^{+2.0}_{-1.5}$ & 1.8 & $ 132 \pm 15 $ & $ 8.6\times 10^{-4}$ & $ 6.0^{+1.0}_{-0.9}$ & 0.94 & $ 98 \pm 9$\
Ahlers, M. 2019, , 886, L18
Al Samarai, I. for the Pierre Auger Collaboration 2016, [ PoS ICRC2015 ] { }, 372
Bonino, R. et al. 2011, , 738, 67
Candia, J., Mollerach, S. and Roulet, E. 2003, , 05, 003
Calvez, A., Kusenko, A. and Nagataki, S. 2010, , 105, 091101
Castellina, A. for the Pierre Auger Collaboration 2019, [ EPJ Web Conf. ] { }, 210, 06002
Coleman, A. for the Pierre Auger Collaboration 2019, [ PoS ICRC2019 ] { }, 225
Compton, A.H. and Getting, I.A. 1935, PhRv, 47, 817
Farley, F.J.M. and Storey, J.R. 1954, [ Proc. Phys. Soc. ] { } A, 67, 996
Feretti, L. et al. 2012, , 20, 54
Greisen, K. 1966, , 16, 748
Haverkorn, M. 2015, [ Astrophys. Space Sci. Library ] { }, 407, 483
IceCube Collaboration 2012, , 746, 33
IceCube Collaboration 2016, , 826, 220
Kachelriess, M. and Serpico, P.D. 2006, [ Phys. Lett. ] { } B, 640, 225
KASCADE- | 10^{-2}$ & $7.5^{+2.3}_{-1.8}$ & 2.1 &$ 80\pm 17$\
$\geq 32$ & 1,993 & $25^{+9}_{-6}$ & 7.6 & $151 \pm 17$ & $4.1\times 10^{-3}$ & $13^{+5}_{-3}$ & 4.1 & $152 \pm 19$\
------------------------------------------------------------------------
$\geq 8$ & 36,928 & $6.6^{+2.0}_{-1.5}$ & 1.8& $132 \pm 15$ & $8.6\tioes 10^{-4}$ & $6.0^{+1.0}_{-0.9}$ & 0.94 & $98 \pm 9$\
Ahlers, M. 2019,, 886, L18
Al Danarai, I. for the Pierre Augef Collabogation 2016, [PoW ICCC2015]{}, 372
Bonino, R. et al. 2011,, 738, 67
Candia, J., Momperaeh, S. and Roulet, G. 2003,, 05, 003
Calvez, A., Nusenko, A. and Taeacaki, S. 2010,, 105, 091101
Castellina, A. for the Pierre Wuger Cplpaboration 2019, [EPJ Web Sonf.]{}, 210, 06002
Coleman, A. for the Pierre Auger Ckllaboretion 2019, [PoS ICRC2019]{}, 225
Vompton, A.H. and Getting, I.A. 1935, OhRv, 47, 817
Farley, F.J.M. and Shorey, J.R. 1954, [Pric. Prts. Soc.]{} A, 67, 996
Fefetti, L. et al. 2012,, 20, 54
Greisen, K. 1966,, 16, 748
Haverkorn, M. 2015, [Astrophys. Space Sci. Kibrary]{}, 407, 483
IxeXubf Collaboratmon 2012,, 746, 33
IbeCube Collaboration 2016,, 826, 220
Kacheltiess, M. and Sevpico, P.D. 2006, [Phys. Lett.]{} B, 640, 225
KASCADE- | 10^{-2}$ & $7.5^{+2.3}_{-1.8}$ & 2.1 &$ 80\pm 32$ 1,993 & & 7.6 & 10^{-3}$ $13^{+5}_{-3}$ & 4.1 $152 \pm 19$\ $\geq 8$ & 36,928 & $6.6^{+2.0}_{-1.5}$ 1.8& $132 \pm 15$ & $8.6\times 10^{-4}$ & $6.0^{+1.0}_{-0.9}$ & 0.94 & $98 9$\ Ahlers, M. 2019,, 886, L18 Al Samarai, I. for the Pierre Auger 2016, ICRC2015]{}, Bonino, et al. 2011,, 738, 67 Candia, J., Mollerach, S. and Roulet, E. 2003,, 05, 003 Calvez, Kusenko, A. and Nagataki, S. 2010,, 105, 091101 A. for the Pierre Collaboration 2019, [EPJ Web Conf.]{}, 06002 A. for Pierre Collaboration [PoS ICRC2019]{}, 225 A.H. and Getting, I.A. 1935, PhRv, 47, 817 Farley, F.J.M. and Storey, J.R. 1954, [Proc. Phys. Soc.]{} 67, 996 et al. 20, Greisen, 1966,, 16, 748 2015, [Astrophys. Space Sci. Library]{}, 407, 2012,, 746, 33 IceCube Collaboration 2016,, 826, 220 M. and P.D. 2006, [Phys. Lett.]{} B, 640, KASCADE- | 10^{-2}$ & $7.5^{+2.3}_{-1.8}$ & 2.1 &$ 80\pm 17$\
$\geq 32$ & 1,993 & $25^{+9}_{-6}$ & 7.6 & $151 \pm 17$ & $4.1\times 10^{-3}$ & $13^{+5}_{-3}$ & 4.1 & $152 \pm 19$\
------------------------------------------------------------------------
$\geq 8$ & 36,928 & $6.6^{+2.0}_{-1.5}$ & 1.8& $132 \pm 15$ & $8.6\times 10^{-4}$ & $6.0^{+1.0}_{-0.9}$ & 0.94 & $98 \pM 9$\
Ahlers, M. 2019,, 886, L18
AL SamaRai, i. foR tHe PiErre auger CollaboraTIon 2016, [POS ICRC2015]{}, 372
Bonino, R. et al. 2011,, 738, 67
CandiA, J., MolLeRAch, S. ANd rouleT, E. 2003,, 05, 003
CalveZ, a., KUSEnkO, A. AnD NaGaTAkI, S. 2010,, 105, 091101
CasTelLina, A. foR the Pierre augEr collaboratioN 2019, [ePj Web Conf.]{}, 210, 06002
CoLemAn, A. for the PieRre auger COlLabORatioN 2019, [Pos ICRC2019]{}, 225
comptoN, a.H. and GEtting, I.A. 1935, PHRV, 47, 817
farley, f.j.M. and StOREy, j.R. 1954, [PrOc. Phys. Soc.]{} A, 67, 996
Feretti, l. Et AL. 2012,, 20, 54
Greisen, K. 1966,, 16, 748
HaverKorn, M. 2015, [AStROpHYS. SpAce sci. Library]{}, 407, 483
icECube cOllaborATiON 2012,, 746, 33
iCeCUBe CollaboratiOn 2016,, 826, 220
KachelrieSS, M. aNd SerpIcO, P.D. 2006, [pHys. LetT.]{} B, 640, 225
KAScAde- | 10^{-2}$ & $7.5^{+2.3}_{- 1.8}$ & 2. 1 &$80\ pm17 $\
$ \geq 32$ & 1,993 & $25^ {+9}_{-6}$ & 7.6 & $15 1 \pm 1 7 $ &$ 4. 1\tim es 10^{ - 3} $ & $ 13 ^{ +5} _{ - 3} $ & 4 .1& $152\pm 19$\
--- -- ------------ - -- ---------- --- ------------ --- ------ -- --- - ----- --- ----
$\geq 8$ & 3 6,928 & $ 6. 6 ^{+2.0 } _{-1.5} $ &1.8& $132 \pm 15$ & $ 8 .6 \ times 10^{-4}$ & $6. 0^ { +1 . 0 }_{ -0. 9}$ & 0.94 & $98\ pm 9$\Ah l e r s,M . 2019,, 886, L18
Al Sa m ara i, I.fo r t h e Pier re Au ge r Co llaboration 201 6, [PoS I CRC201 5 ]{}, 37 2
Bonin o, R.etal. 201 1 ,, 7 38, 6 7
C a nd ia, J., Mollera ch ,S. an d Ro u l e t , E. 20 03,, 05,003
Calvez,A., Kus e nko , A.and N agat ak i, S. 2010, , 105 ,091101
Castell ina, A. for t hePi err eAuger Collab ora tio n 2019, [EPJ W e b C on f . ] {} , 210, 06002
Cole ma n , A . for th e Pier r eAu g er Colla bo rat ion2 0 19, [ PoSI CR C2019]{} , 225Co mp ton, A. H. and G et tin g,I.A.1 935, PhRv, 47, 817
Far l ey, F.J.M. and Storey, J.R.1 95 4 , [ P roc. Ph ys. Soc.]{} A,6 7, 9 96
F er ett i , L.et al .2 01 2 ,, 20, 54
Greisen, K . 1966 ,, 16 , 748
Haverk orn, M. 20 1 5 , [Astrop hys. Sp a ce Sci. Librar y]{}, 407, 483IceCubeColla boration 2012,, 7 4 6 , 33
Ic eCu beCol lab o r at ion 2016,, 82 6 , 220
Kachelr ies s, M. a ndSer pic o,P. D. 2006,[Phys. L et t. ]{ }B,640,2 25
KASC AD E- | 10^{-2}$_& $7.5^{+2.3}_{-1.8}$_& 2.1 &$ 80\pm_17$\
$\geq 32$_&_1,993 &_$25^{+9}_{-6}$_& 7.6 &_$151 \pm 17$_& $4.1\times 10^{-3}$ &_$13^{+5}_{-3}$ & 4.1_&_$152 \pm 19$\
------------------------------------------------------------------------
$\geq 8$ & 36,928 & $6.6^{+2.0}_{-1.5}$ & 1.8& $132 \pm 15$ &_$8.6\times_10^{-4}$ &_$6.0^{+1.0}_{-0.9}$_&_0.94 & $98 \pm 9$\
Ahlers,_M. 2019,, 886, L18
Al Samarai, I._for the_Pierre Auger Collaboration 2016, [PoS ICRC2015]{}, 372
Bonino, R. et_al. 2011,,_738, 67
Candia, J.,_Mollerach, S. and Roulet, E. 2003,, 05, 003
Calvez, A., Kusenko,_A. and Nagataki, S. 2010,, 105, 091101
Castellina,_A. for the_Pierre_Auger_Collaboration 2019, [EPJ Web Conf.]{},_210, 06002
Coleman, A. for the Pierre_Auger Collaboration 2019, [PoS ICRC2019]{}, 225
Compton, A.H._and Getting, I.A. 1935, PhRv, 47, 817
Farley, F.J.M._and Storey, J.R. 1954, [Proc. Phys. Soc.]{}_A, 67, 996
Feretti, L. et_al. 2012,, 20,_54
Greisen, K. 1966,, 16, 748
Haverkorn, M. 2015,_[Astrophys. Space Sci._Library]{}, 407,_483
IceCube Collaboration 2012,, 746,_33
IceCube Collaboration 2016,, 826, 220
Kachelriess, M. and_Serpico, P.D. 2006, [Phys._Lett.]{} B, 640, 225
KASCADE- |
E^s\oplus E^c\oplus E^u$, with non trivial extremal sub-bundles $E^s$ and $E^u$, and there exists $n\in {\mathbb{N}}$ such that $E^s$ and $E^u$ are uniformly contracted by $Df^n$ and $Df^{-n}$, respectively.
If the center bundle $E^c$ is trivial, then $f$ is called [*Anosov*]{}. For convenience, given a partially hyperbolic diffeomorphism $f$, we consider its partially hyperbolic splitting $TM=E^s\oplus E^c\oplus E^u$, such that the extremal bundles contains all the $Df$-invariant sub-bundles of $TM$ which are contracted or expanded for some iterate of $Df$. In particular, for us, a partially hyperbolic diffeomorphism with non-trivial center bundle is not Anosov.
By Theorem 6.1 of [@HPS] the strong bundles, $E^s$ and $E^u$, of a partially hyperbolic diffeomorphism $f$ is integrable. That is, there exist two strong foliations, the strong stable and strong unstable foliations, which are tangent to $E^s$ and $E^u$, respectively. We denote these foliations by $\mathcal{F}^s$ and $ \mathcal{F}^u$, respectively.
We say that a partially hyperbolic diffeomorphism $f$ is [*s-minimal (resp. u-minimal)*]{} if its strong stable (resp. strong unstable) foliation $\mathcal{F}^s$ (resp. $\mathcal{F}^u$) is minimal.
As we mentioned before, it is known that $s-$minimality and $u-$minimality implies topological properties of the dynamics. More precisely, a $s-$minimal or $u-$minimal, partially hyperbolic diffeomorphism $f$ is topologically mixing. In particular, $f$ is topologically transitive. Recall that a diffeomorphism $f$ is [*topologically transitive*]{} if there is a point whose forward orbit by $f$ is dense on $M$. Also, $f$ is [*topologically mixing*]{} if given open sets $U$ and $V$ of $M$, there exists a positive integer $n$ such that $f^j(U)$ intersects $V$ for any $j\geq n$.
| E^s\oplus E^c\oplus E^u$, with non trivial extremal sub - bundles $ E^s$ and $ E^u$, and there exists $ n\in { \mathbb{N}}$ such that $ E^s$ and $ E^u$ are uniformly compress by $ Df^n$ and $ Df^{-n}$, respectively.
If the plaza bundle $ E^c$ is trivial, then $ f$ is called [ * Anosov * ] { }. For appliance, given a partially hyperbolic diffeomorphism $ f$, we consider its partially hyperbolic separate $ TM = E^s\oplus E^c\oplus E^u$, such that the extremal bundle contains all the $ Df$-invariant sub - bundles of $ TM$ which are contracted or boom for some iterate of $ Df$. In particular, for us, a partially hyperbolic diffeomorphism with non - trivial kernel package is not Anosov.
By Theorem 6.1 of [ @HPS ] the strong bundles, $ E^s$ and $ E^u$, of a partially hyperbolic diffeomorphism $ f$ is integrable. That is, there exist two strong foliations, the strong static and strong unstable foliations, which are tangent to $ E^s$ and $ E^u$, respectively. We announce these foliations by $ \mathcal{F}^s$ and $ \mathcal{F}^u$, respectively.
We say that a partially hyperbolic diffeomorphism $ f$ is [ * s - minimal (resp. u - minimal) * ] { } if its strong stable (resp. strong precarious) foliation $ \mathcal{F}^s$ (resp. $ \mathcal{F}^u$) is minimal.
As we mentioned before, it is known that $ s-$minimality and $ u-$minimality implies topological properties of the moral force. More precisely, a $ s-$minimal or $ u-$minimal, partially hyperbolic diffeomorphism $ f$ is topologically mixing. In particular, $ f$ is topologically transitive. Recall that a diffeomorphism $ f$ is [ * topologically transitive * ] { } if there is a point whose forward eye socket by $ f$ is dense on $ M$. Also, $ f$ is [ * topologically mixing * ] { } if given open sets $ U$ and $ V$ of $ M$, there exists a positive integer $ n$ such that $ f^j(U)$ intersects $ V$ for any $ j\geq n$. | E^s\oolus E^c\oplus E^u$, with non trivial extremco sub-bnndles $S^s$ and $E^j$, and there exists $n\in {\mathbu{N}}$ sych tyat $E^s$ and $E^u$ are unifurmly connracted bt $Df^i$ and $Df^{-n}$, respecvjvely.
If the cskter yuidle $E^c$ is trivlal, then $f$ hs called [*Anosmv*]{}. Flr convenience, given a partially hy[erboliv fiffeomorphism $f$, wt cjnsisvr its partially hyperbolic spljtting $UM=E^s\oplus E^c\oplus R^u$, such that the extremal hundpes contains all tje $Df$-invariqnt fyb-bundles of $TM$ which are contractgd or expanded for some iterate uf $Df$. In particolzr, xor us, a pactiallj hyperbolic diffeomosphism eith non-triviak cxntee bundle is not Anosot.
By Theorem 6.1 of [@HPS] jhe strong bbndles, $E^s$ and $E^u$, of a pqrtianly vypefvolkc sihfekmorphlsm $f$ is intefrable. That is, there exist two segpng foliatiohs, the serong stable and strong unstable foliatpons, which are tangent to $E^w$ and $E^u$, respectively. We denotq these foliations by $\mathcal{F}^s$ and $ \mathcal{F}^u$, revpectmvdly.
Cc rqy that a partially hyperbolic diffeomorphism $f$ js [*s-kinimal (resp. u-iinimal)*]{} if otd xjrong stable (rgsp. strong unstable) foliatioj $\mathcwl{F}^s$ (eesp. $\mathsal{F}^i$) is minimal.
As we mentioned before, it if known that $s-$minimclity and $u-$mnnimaloty ikplies topological propzrties of the dynwmics. Mors precisely, a $s-$mivimsl or $u-$minimal, partially hypqrbolic dmffeokorphiso $f$ os topjlogically mixikc. In particular, $f$ ls tokologiwally tranditive. Recall that a diffeomorpimsm $f$ is [*topokocicdlly trausitivc*]{} if there is a point whose fprward jrbit by $f$ is dvnse on $M$. Also, $f$ is [*eopologically lixing*]{} if giten open fets $U$ abd $V$ of $O$, there exists a positine intwger $n$ such that $f^m(U)$ injedsects $V$ for anv $h\geq n$.
| E^s\oplus E^c\oplus E^u$, with non trivial extremal and and there $n\in {\mathbb{N}}$ such uniformly by $Df^n$ and respectively. If the bundle $E^c$ is trivial, then $f$ called [*Anosov*]{}. For convenience, given a partially hyperbolic diffeomorphism $f$, we consider its hyperbolic splitting $TM=E^s\oplus E^c\oplus E^u$, such that the extremal bundles contains all the sub-bundles $TM$ are or expanded for some iterate of $Df$. In particular, for us, a partially hyperbolic diffeomorphism with center bundle is not Anosov. By Theorem 6.1 [@HPS] the strong bundles, and $E^u$, of a partially diffeomorphism is integrable. is, exist strong foliations, the stable and strong unstable foliations, which are tangent to $E^s$ and $E^u$, respectively. We denote these foliations $\mathcal{F}^s$ and respectively. We that partially diffeomorphism $f$ is u-minimal)*]{} if its strong stable (resp. $\mathcal{F}^s$ (resp. $\mathcal{F}^u$) is minimal. As we mentioned it is that $s-$minimality and $u-$minimality implies topological of the dynamics. More precisely, a $s-$minimal or partially hyperbolic diffeomorphism $f$ is topologically mixing. In particular, $f$ is topologically transitive. Recall that $f$ is [*topologically transitive*]{} there is a whose orbit $f$ dense on Also, $f$ is [*topologically mixing*]{} if given open sets $U$ and of $M$, there exists a positive integer $n$ such that $V$ any $j\geq n$. | E^s\oplus E^c\oplus E^u$, with non trIvial extreMal suB-buNdlEs $e^s$ anD $E^u$, aNd there exists $n\IN {\matHbb{N}}$ such that $E^s$ and $E^u$ are UnifoRmLY conTRaCted bY $Df^n$ and $dF^{-n}$, RESpeCtIvEly.
if THe CenteR buNdle $E^c$ iS trivial, thEn $f$ Is Called [*Anosov*]{}. fOr ConveniencE, giVen a partiallY hyPerbolIc DifFEomorPhiSm $f$, we ConsidER its paRtially hyPeRBolic sPLitting $tm=e^s\OpluS E^c\oplus E^u$, such thaT ThE Extremal bundleS contaInS AlL THe $DF$-inVariant sub-BuNdles OF $TM$ whicH ArE CONtrACted or expandeD for some iteRAte Of $Df$. In PaRtiCUlar, foR us, a pArTIalLy hyperboliC difFeomorphiSm with NOn-triviAL center Bundle Is nOt ANosoV.
by thEorEm 6.1 OF [@HPs] ThE stROng Bundles, $E^S$ aNd $e^u$, of a PartIALLY hypErbOlic DiffeOmorphism $f$ is iNteGrabLE. ThAt is, tHere eXist TwO stroNg foliAtionS, tHe strong stable aNd stRong unstaBle FoLiaTiOns, whICh are tAngEnt To $E^s$ and $e^u$, respeCTivElY. wE DeNote these foliationS bY $\MAtHcal{F}^s$ anD $ \mathcAL{F}^U$, rESpectiveLy.
we sAy thAT A partIallY HyPerbolic DiffeoMOrPhIsm $f$ is [*s-MiNimal (rEsP. u-mIniMal)*]{} if ITs stRong stAble (resp. StronG Unstable) foliatIOn $\mathcal{F}^s$ (reSP. $\mATHcAL{F}^u$) iS miNimal.
As we meNtioNEd beFore, IT iS knOWn thaT $s-$minImALiTY and $u-$minimality implIeS topolOgicaL properties of The dynamicS. mORe precisEly, a $S-$MiNImal or $u-$minimal, PartiAlly hyperbOLic diffeOmorpHism $f$ is tOpologicaLLY mixing. IN paRtiCulAr, $f$ IS ToPologically trANSitiVe. recall tHat A diffeoMorPhiSm $f$ Is [*tOpOlogicallY transitIvE*]{} iF tHeRe iS a poiNT whose foRwArd OrBit By $f$ is DEnse on $m$. Also, $F$ is [*tOpOlOGicAlly mixINg*]{} IF GiveN oPeN setS $U$ aNd $v$ of $M$, tHere EXisTs a posiTive integEr $n$ SUch tHaT $f^J(U)$ interSects $V$ for any $j\GeQ n$.
| E^s\oplus E^c\oplus E^u$,with non t rivia l e xtr em al s ub-b undles $E^s$ a n d $E ^u$, and there exists$n\in { \ math b b{ N}}$such th a t$ E ^s$ a nd $E ^u $ a re un ifo rmly co ntracted b y $ Df ^n$ and $Df^ { -n }$, respec tiv ely.
If the ce nter b un dle $E^c$ is triv ial, t h en $f$ is calle d[ *Anoso v *]{}. F o r c onve nience, given a p a rt i ally hyperboli c diff eo m or p h ism $f $, we cons id er it s partia l ly h y per b olic splittin g $TM=E^s\o p lus E^c\o pl usE ^u$, s uch t ha t th e extremalbund les conta ins al l the $D f $-invar iant s ub- bun dles of $ TM$ w h ich ar e c o ntr acted or e xp anded for s o m e it era te o f $Df $. In particu lar , fo r us , a p artia llyhy perbo lic di ffeom or phism with non- triv ial cente r b un dle i s not Anosov .
ByTheorem 6.1 of [@H PS ] t he strong bundles, $ E^ s $ a nd $E^u$ , of a pa rt i ally hyp er bol ic d i f feomo rphi s m$f$ is i ntegra b le .That is ,thereex ist tw o str o ng f oliati ons, the stro n g stable and s t rong unstable fo l i at i ons, wh ich are tan gent to $ E^s$ an d $ E ^u$,respe ct i ve l y. We denote thesefo liatio ns by $\mathcal{F} ^s$ and $\ m a thcal{F} ^u$, re s pectively.
We saythat a par t ially hy perbo lic diff eomorphis m $f$ is [ *s- min ima l ( r e sp . u-minimal)* ] { } if i ts stro ngstable(re sp. st ron gunstable) foliati on $ \m at hca l{F}^ s $ (resp. $ \ma th cal {F}^u $ ) is m inima l.
As w e me ntioned be f o re,it i s kn own t hat $ s-$m i nim ality a nd $u-$mi nim a lity i mp lies to pological pro pe rties of t he dy namics . More pre cisely, a $s-$minimal o r $u-$mi nim al, p arti ally hype rbo lic di ffe o morphi sm $f$ is t op olo g i cally m ix ing .In particu l a r,$f$ i stopo logical ly transitive. Rec a llthat a diffeo mor phis m $f $ i s [ * top ol o gic a l ly transitive*] {} if ther ei sa point wh o sefo rward o rbit by $f$i s dense on $M$.Also, $f$ i s [* t o pol ogically m ixing*]{ } if give n open se ts $U $ a nd $V$ o f $ M$, t here e x ist s a p ositiv eintege r $n$ s uch that $f^j(U)$ intersects $V $ forany $ j\g eq n$.
| E^s\oplus E^c\oplus_E^u$, with_non trivial extremal sub-bundles_$E^s$ and_$E^u$,_and there_exists_$n\in {\mathbb{N}}$ such_that $E^s$ and_$E^u$ are uniformly contracted_by $Df^n$ and_$Df^{-n}$,_respectively.
If the center bundle $E^c$ is trivial, then $f$ is called [*Anosov*]{}. For convenience,_given_a partially_hyperbolic_diffeomorphism_$f$, we consider its partially_hyperbolic splitting $TM=E^s\oplus E^c\oplus E^u$,_such that_the extremal bundles contains all the $Df$-invariant sub-bundles_of_$TM$ which are_contracted or expanded for some iterate of $Df$. In_particular, for us, a partially hyperbolic_diffeomorphism with non-trivial_center_bundle_is not Anosov.
By Theorem_6.1 of [@HPS] the strong bundles,_$E^s$ and $E^u$, of a partially_hyperbolic diffeomorphism $f$ is integrable. That is,_there exist two strong foliations, the_strong stable and strong unstable_foliations, which_are tangent to $E^s$ and_$E^u$, respectively. We_denote these_foliations by $\mathcal{F}^s$_and $ \mathcal{F}^u$, respectively.
We say that_a partially hyperbolic_diffeomorphism $f$ is [*s-minimal (resp. u-minimal)*]{}_if_its strong stable_(resp._strong_unstable) foliation_$\mathcal{F}^s$ (resp. $\mathcal{F}^u$)_is_minimal.
As we_mentioned_before, it is known that $s-$minimality_and_$u-$minimality implies topological properties of the dynamics._More precisely, a $s-$minimal_or_$u-$minimal, partially hyperbolic diffeomorphism_$f$ is topologically mixing. In_particular, $f$ is topologically transitive. Recall_that a_diffeomorphism $f$_is [*topologically transitive*]{} if there is a point whose forward orbit_by $f$ is dense on $M$._Also, $f$ is [*topologically_mixing*]{} if_given_open sets $U$_and_$V$ of_$M$, there exists a positive integer $n$_such that_$f^j(U)$ intersects $V$ for any $j\geq_n$.
|
\sum_{i=1}^{d_r}a_{n_i}x^{d_r-i}\right)$$ where each $a_{n_i},\ 1 \leq i\leq
d_r,$ satisfies the following system of non-linear recurrence relations $$\Bigg\{\sum_{i+j=2k}(-1)^ja_{n_i}a_{n_j} = a_{(n-1)_k}, {\hspace*{2em}}1\leq k\leq d_r \Bigg\}$$ with initial values $a_{A_k},\ 1\leq k\leq d_r,$ as obtained in (ii).
\(iv) For $n > K,$ the complete factorization of $\Phi_{2^nr}$ over ${\mathbb{F}_q}$ is given by $$\Phi_{2^nr}(x) = \prod_{u\in U_A}\prod_{w\in
\Omega(r)}\left(x^{2^{n-K}d_r}+
\sum_{i=1}^{d_r}a_{K_i}x^{2^{n-K}(d_r-i)}\right)$$ where each $a_{K_i},\ 1\leq i\leq d_r,$ is as obtained in (iii).
Let $$\Phi_r(x) = \prod_{w\in \Omega(r)}g_w(x) = \prod_{w\in
\Omega(r)}\left(\sum_{i=0}^{d_r}(-1)^i S_{i,w}x^{d_r - i}\right)$$ be the factorization of $\Phi_r$ over ${\mathbb{F}_q}.$\
1. $(n=1):$ Because $g_w$ is irreducible over ${\mathbb{F}_q}$, $g_w(-x)$ is irreducible over ${\mathbb{F}_q}.$ By Theorem \[tricks\], $$\begin{aligned}
\Phi_{2r}(x) &=& \Phi_r(-x) = \prod_{w\in \Omega(r)}g_w(-x) =
\prod_{w\in \Omega(r)}\left(\sum_{i=0}^{d_r}(-1)^{d_r} S_{i,w}x^{d_r -
i}\right).\end{aligned}$$ Note that in the case $d | \sum_{i=1}^{d_r}a_{n_i}x^{d_r - i}\right)$$ where each $ a_{n_i},\ 1 \leq i\leq
d_r,$ satisfies the following system of non - linear recurrence relative $ $ \Bigg\{\sum_{i+j=2k}(-1)^ja_{n_i}a_{n_j } = a_{(n-1)_k }, { \hspace*{2em}}1\leq k\leq d_r \Bigg\}$$ with initial value $ a_{A_k},\ 1\leq k\leq d_r,$ as obtained in (ii).
\(iv) For $ n > K,$ the complete factorization of $ \Phi_{2^nr}$ over $ { \mathbb{F}_q}$ is yield by $ $ \Phi_{2^nr}(x) = \prod_{u\in U_A}\prod_{w\in
\Omega(r)}\left(x^{2^{n - K}d_r}+
\sum_{i=1}^{d_r}a_{K_i}x^{2^{n - K}(d_r - i)}\right)$$ where each $ a_{K_i},\ 1\leq i\leq d_r,$ is as obtained in (iii).
Let $ $ \Phi_r(x) = \prod_{w\in \Omega(r)}g_w(x) = \prod_{w\in
\Omega(r)}\left(\sum_{i=0}^{d_r}(-1)^i S_{i, w}x^{d_r - i}\right)$$ be the factorization of $ \Phi_r$ all over $ { \mathbb{F}_q}.$\
1. $ (n=1):$ Because $ g_w$ is irreducible over $ { \mathbb{F}_q}$, $ g_w(-x)$ is irreducible over $ { \mathbb{F}_q}.$ By Theorem \[tricks\ ], $ $ \begin{aligned }
\Phi_{2r}(x) & = & \Phi_r(-x) = \prod_{w\in \Omega(r)}g_w(-x) =
\prod_{w\in \Omega(r)}\left(\sum_{i=0}^{d_r}(-1)^{d_r } S_{i, w}x^{d_r -
i}\right).\end{aligned}$$ Note that in the case $ d | \sum_{l=1}^{d_r}a_{n_i}x^{d_r-i}\right)$$ where eagh $a_{n_i},\ 1 \leq i\leq
b_e,$ sativfies fhe folluwing system of non-linear rerurrwnce eelations $$\Bigg\{\sum_{i+j=2k}(-1)^ja_{v_i}a_{n_j} = a_{(n-1)_n}, {\hspace*{2wm}}1\lew k\leq d_r \Bmfg\}$$ with initizp vannes $a_{A_k},\ 1\leq k\lea d_r,$ as obdained in (ii).
\(iv) Fur $n > K,$ the complete factorization of $\Phi_{2^nr}$ pvfr ${\mathbb{F}_q}$ is givtn fy $$\Pgp_{2^nv}(x) = \prod_{u\in U_A}\prod_{w\in
\Omega(r)}\left(x^{2^{h-K}d_r}+
\sum_{p=1}^{d_r}a_{K_i}x^{2^{n-K}(d_r-i)}\right)$$ where each $a_{K_i},\ 1\leq i\leq d_g,$ is as obtained in (iil).
Let $$\Phi_r(x) = \proq_{q\in \Omega(r)}g_w(b) = \prod_{w\in
\Omega(r)}\left(\suj_{i=0}^{d_r}(-1)^i S_{i,w}x^{d_r - i}\right)$$ be the facgorizction of $\Phu_r$ ovft ${\mathbb{F}_q}.$\
1. $(n=1):$ Becalse $g_w$ is irrccucibla over ${\kathbb{F}_q}$, $g_w(-x)$ ix icredycible over ${\mathbb{F}_q}.$ Uy Theorem \[tricks\], $$\bedin{aligneg}
\Pki_{2r}(x) &=& \Phi_r(-x) = \prod_{w\in \Imwga(r)}g_f(-x) =
\psod_{w\kb \Ooegz(r)}\kert(\sum_{i=0}^{f_r}(-1)^{d_c} S_{i,w}x^{d_r -
i}\rjght).\end{aligbed}$$ Note that in tht cwwe $d | \sum_{i=1}^{d_r}a_{n_i}x^{d_r-i}\right)$$ where each $a_{n_i},\ 1 \leq i\leq the system of recurrence relations $$\Bigg\{\sum_{i+j=2k}(-1)^ja_{n_i}a_{n_j} \Bigg\}$$ initial values $a_{A_k},\ k\leq d_r,$ as in (ii). \(iv) For $n > the complete factorization of $\Phi_{2^nr}$ over ${\mathbb{F}_q}$ is given by $$\Phi_{2^nr}(x) = \prod_{u\in \Omega(r)}\left(x^{2^{n-K}d_r}+ \sum_{i=1}^{d_r}a_{K_i}x^{2^{n-K}(d_r-i)}\right)$$ where each $a_{K_i},\ 1\leq i\leq d_r,$ is as obtained in (iii). $$\Phi_r(x) \prod_{w\in = \Omega(r)}\left(\sum_{i=0}^{d_r}(-1)^i S_{i,w}x^{d_r - i}\right)$$ be the factorization of $\Phi_r$ over ${\mathbb{F}_q}.$\ 1. $(n=1):$ Because $g_w$ is over ${\mathbb{F}_q}$, $g_w(-x)$ is irreducible over ${\mathbb{F}_q}.$ By \[tricks\], $$\begin{aligned} \Phi_{2r}(x) &=& = \prod_{w\in \Omega(r)}g_w(-x) = \prod_{w\in S_{i,w}x^{d_r i}\right).\end{aligned}$$ Note in case | \sum_{i=1}^{d_r}a_{n_i}x^{d_r-i}\right)$$ where eaCh $a_{n_i},\ 1 \leq i\lEq
d_r,$ sAtiSfiEs The fOlloWing system of noN-LineAr recurrence relations $$\BIgg\{\suM_{i+J=2K}(-1)^ja_{n_I}A_{n_J} = a_{(n-1)_k}, {\hSpace*{2em}}1\LEq K\LEq d_R \BIgG\}$$ wiTh INiTial vAluEs $a_{A_k},\ 1\leQ k\leq d_r,$ as oBtaInEd in (ii).
\(iv) For $n > k,$ ThE complete fActOrization of $\PHi_{2^nR}$ over ${\mAtHbb{f}_Q}$ is giVen By $$\Phi_{2^Nr}(x) = \proD_{U\in U_A}\pRod_{w\in
\OmeGa(R)}\Left(x^{2^{n-k}D_r}+
\sum_{i=1}^{d_R}A_{k_i}X^{2^{n-K}(d_R-i)}\right)$$ where each $a_{k_I},\ 1\lEQ i\leq d_r,$ is as obtAined iN (iII).
LET $$\phi_R(x) = \pRod_{w\in \OmegA(r)}G_w(x) = \prOD_{w\in
\OmeGA(r)}\LEFT(\suM_{I=0}^{d_r}(-1)^i S_{i,w}x^{d_r - i}\riGht)$$ be the facTOriZation Of $\phi_R$ Over ${\maThbb{F}_Q}.$\
1. $(n=1):$ bEcaUse $g_w$ is irreDuciBle over ${\maThbb{F}_q}$, $G_W(-x)$ is irrEDucible Over ${\maThbB{F}_q}.$ by ThEOrEm \[TriCkS\], $$\BegIN{aLigNEd}
\PHi_{2r}(x) &=& \Phi_r(-X) = \pRoD_{w\in \OMega(R)}G_W(-X) =
\Prod_{W\in \omegA(r)}\lefT(\sum_{i=0}^{d_r}(-1)^{d_r} S_{i,w}x^{D_r -
i}\RighT).\End{AlignEd}$$ NotE thaT iN the cAse $d | \sum_{i=1}^{d_r}a_{n_i}x^{ d_r-i}\rig ht)$$ wh ere e ach$a_{ n_i},\ 1 \leqi \leq
d_r,$ satisfies the f ollow in g sys t em of n on-line a rr e cur re nc e r el a ti ons $ $\B igg\{\s um_{i+j=2k }(- 1) ^ja_{n_i}a_{ n _j } = a_{(n- 1)_ k}, {\hspace *{2 em}}1\ le q k \ leq d _r\Bigg \}$$ w i th ini tial valu es $a_{A_ k },\ 1\l e q k \leq d_r,$ as obtaine d i n (ii).
\(iv)For $n > K, $ the co mplete fac to rizat i on of $ \ Ph i _ { 2^n r }$ over ${\ma thbb{F}_q}$ isgivenby $$ \ Phi_{2 ^nr}( x) = \ prod_{u\inU_A} \prod_{w\ in
\Om e ga(r)}\ l eft(x^{ 2^{n-K }d_ r}+
\su m _{ i= 1}^ {d _ r}a _ {K _i} x ^{2 ^{n-K}(d _r -i )}\ri ght) $ $ w here ea ch $ a_{K_ i},\ 1\leq i\ leq d_r , $ i s asobtai nedin (iii ).
Le t $$\ Ph i_r(x) = \prod_ {w\i n \Omega( r)} g_ w(x )= \pr o d_{w\i n
\ Ome ga(r)}\ left(\s u m_{ i= 0 } ^ {d _r}(-1)^i S_{i,w}x ^{ d _ r- i}\rig ht)$$b eth e factori za tio n of $ \Phi_ r$ o v er ${\math bb{F}_ q }. $\
1. $(n =1 ):$ Be ca use $g _w$ i s irr educib le over${\ma t hbb{F}_q}$, $g _ w(-x)$ is irr e du c i bl e ove r $ {\mathbb{F} _q}. $ ByTheo r em \[ t ricks \], $ $\ b eg i n{aligned}
\Phi_{2r }( x) &=& \Phi _r(-x) = \pro d_{w\in \O m e g a(r)}g_w (-x) =\ prod_{w\in \Om ega(r )}\left(\s u m_{i=0}^ {d_r} (-1)^{d_ r} S_{i,w } x ^{d_r -i}\ rig ht) .\e n d {a ligned}$$ Not e that i n the c ase $d | \sum_{i=1}^{d_r}a_{n_i}x^{d_r-i}\right)$$ where_each $a_{n_i},\_1 \leq i\leq
d_r,$ satisfies_the following_system_of non-linear_recurrence_relations $$\Bigg\{\sum_{i+j=2k}(-1)^ja_{n_i}a_{n_j} =_a_{(n-1)_k}, {\hspace*{2em}}1\leq k\leq_d_r \Bigg\}$$ with initial_values $a_{A_k},\ 1\leq_k\leq_d_r,$ as obtained in (ii).
\(iv) For $n > K,$ the complete factorization of $\Phi_{2^nr}$_over_${\mathbb{F}_q}$ is_given_by_$$\Phi_{2^nr}(x) = \prod_{u\in U_A}\prod_{w\in
\Omega(r)}\left(x^{2^{n-K}d_r}+
\sum_{i=1}^{d_r}a_{K_i}x^{2^{n-K}(d_r-i)}\right)$$ where_each $a_{K_i},\ 1\leq i\leq d_r,$_is as_obtained in (iii).
Let $$\Phi_r(x) = \prod_{w\in \Omega(r)}g_w(x) =_\prod_{w\in
\Omega(r)}\left(\sum_{i=0}^{d_r}(-1)^i_S_{i,w}x^{d_r - i}\right)$$_be the factorization of $\Phi_r$ over ${\mathbb{F}_q}.$\
1. $(n=1):$ Because_$g_w$ is irreducible over ${\mathbb{F}_q}$, $g_w(-x)$_is irreducible over_${\mathbb{F}_q}.$_By_Theorem \[tricks\], $$\begin{aligned}
\Phi_{2r}(x) &=&_\Phi_r(-x) = \prod_{w\in \Omega(r)}g_w(-x) =
\prod_{w\in \Omega(r)}\left(\sum_{i=0}^{d_r}(-1)^{d_r}_S_{i,w}x^{d_r -
i}\right).\end{aligned}$$ Note that in the_case $d |
partition of $\Omega$, are two strong factors which hinder us from linking $f(\alpha)=d_H(\Omega_{\alpha(\beta)})$ with $d_H(G_{\beta})$ directly through, say, so simple a way as an equality. Still, we know:
1. For $\beta=\infty$ we have $\alpha(\beta)=\alpha_{\min}$.
2. For $\beta=2$ we have $\alpha(\beta)=\alpha_{\max}$.
3. $f(\alpha_{\min})=0$.
4. $f(\alpha)$ is increasing for the greater part of the interval $[\alpha_{\min},\alpha_{\max}]$.
5. Let $\max\limits_{\alpha}f(\alpha)=f(\alpha^{\max})$. Then $\alpha_{\max}$ is very near $\alpha^{\max}$.
6. The $\Omega_{\alpha}$ are strongly linked to the types $G_{\beta}$.
7. The sets $G_{\infty}$ and $G_{2}$, related to $\alpha_{\min}$ and $\alpha_{\max}$ in $\Omega$, are related to $\alpha_{\max}$ and $\alpha_{\min}$ in $I=[0,1]$ endowed with the hyperbolic measure of probability induced by Farey-Brocot.
The Spectrum $(\alpha, f(\alpha))$ of the Fractal Set $\Omega$ underlying the Circle Map Staircase
==================================================================================================
Conclusions 1) to 7) in the last section show a strong connection between the magnetization function $q=g(-H)$ and leading problems in Number Theory –viz the good approximation of irrational numbers studied with Jarník classes $J_{\beta}$ and their refinements $G_{\beta},\;\beta\geq 2$. This particular connection between magnetization and Number Theory is seen only when analyzing the multifractal spectrum of the fractal set $\Omega$ underlying the magnetization Cantor staircase. The conclusions (1) to 7)) are based, as we have seen, on two premises about the staircase $q=g(-H)$: the $(F-B)$ arrangement of the stairsteps $\Delta H$ in the staircase, and the formula given by Eq. (\[fases\]).
Let us now consider the Cantor staircase $W=g(\omega)$ associated with the circle map: we land in Dynamical Systems, where connections with Number Theory are old and well explored. | partition of $ \Omega$, are two strong factors which hinder us from connect $ f(\alpha)=d_H(\Omega_{\alpha(\beta)})$ with $ d_H(G_{\beta})$ immediately through, say, so simple a direction as an equality. even, we know:
1. For $ \beta=\infty$ we have $ \alpha(\beta)=\alpha_{\min}$.
2. For $ \beta=2 $ we hold $ \alpha(\beta)=\alpha_{\max}$.
3. $ f(\alpha_{\min})=0$.
4. $ f(\alpha)$ is increasing for the greater region of the interval $ [ \alpha_{\min},\alpha_{\max}]$.
5. Let $ \max\limits_{\alpha}f(\alpha)=f(\alpha^{\max})$. Then $ \alpha_{\max}$ is very near $ \alpha^{\max}$.
6. The $ \Omega_{\alpha}$ are powerfully linked to the types $ G_{\beta}$.
7. The sets $ G_{\infty}$ and $ G_{2}$, related to $ \alpha_{\min}$ and $ \alpha_{\max}$ in $ \Omega$, are relate to $ \alpha_{\max}$ and $ \alpha_{\min}$ in $ I=[0,1]$ endowed with the hyperbolic standard of probability induced by Farey - Brocot.
The Spectrum $ (\alpha, f(\alpha))$ of the Fractal Set $ \Omega$ underlying the Circle Map Staircase
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
decision 1) to 7) in the last section show a solid connection between the magnetization function $ q = g(-H)$ and leading problems in act Theory – viz the good approximation of irrational numbers studied with Jarník class $ J_{\beta}$ and their refinements $ G_{\beta},\;\beta\geq 2$. This particular connection between magnetization and Number Theory is seen only when analyzing the multifractal spectrum of the fractal set $ \Omega$ underlying the magnetization Cantor staircase. The conclusions (1) to 7) ) are based, as we have seen, on two premises about the staircase $ q = g(-H)$: the $ (F - B)$ arrangement of the stairsteps $ \Delta H$ in the staircase, and the formula contribute by Eq. (\[fases\ ]).
Let us now think the Cantor staircase $ watt = g(\omega)$ associated with the circle map: we land in Dynamical Systems, where connections with Number Theory are old and well explored. | parhition of $\Omega$, are two rtrong factors cyich hmnder ua from lknking $f(\alpha)=d_H(\Omega_{\alpha(\bete)})$ wirh $d_H(T_{\beta})$ directly through, say, so spmple a wqy aw an equalivg. Still, we knka:
1. Fmc $\beta=\infty$ we mave $\alpha(\bata)=\alpha_{\min}$.
2. Fmr $\bzta=2$ we have $\alpha(\beta)=\alpha_{\max}$.
3. $f(\alphw_{\min})=0$.
4. $f(\sloha)$ is increasyng gjr tgv nreater part of the interval $[\alpga_{\min},\alkha_{\max}]$.
5. Let $\max\limots_{\alpha}f(\alpha)=f(\alpha^{\max})$. Thfn $\appha_{\max}$ is very newr $\alpha^{\max}$.
6. Thq $\Omega_{\alpha}$ xre strongly linked to the types $G_{\beta}$.
7. The sets $G_{\infgy}$ anb $G_{2}$, related ti $\apkha_{\min}$ and $\aopha_{\mwx}$ in $\Omega$, are reladed to $\slpha_{\max}$ and $\akphe_{\min}$ in $I=[0,1]$ endowed with thx hyperbolic measure of probatimity induced by Fqrwy-Browot.
Tve Sowctfum $(\akpga, f(\aloha))$ of the Frzctal Set $\Onega$ underlying the Cygvle Map Staidcase
==================================================================================================
Cjnslusions 1) to 7) in the last section show d sfrong connection betweeb the magnetization fonction $q=g(-R)$ and leading problems in Number Theory –viz the gmod a'pfoxnnatiov ov irrational numbers studied with Jarník classqa $K_{\bvta}$ and their reflnements $G_{\beta},\;\beta\bee 2$. Jhis particulat conneefikn between magnetixation wnd Nymber Thejry os seen only when analyzing the multifrcctql spectrum of the fractal sec $\Omegs$ undrrlying the magnetizatiun Czntor stairfase. The dunclusions (1) to 7)) xre bdsed, as we have seen, on twj premisew abput the staorcase $q=g(-H)$: the $(F-H)$ arrangement of the stalrsteks $\Delda H$ in thf staircase, and the formula givxi by Eq. (\[fases\]).
Ked uv now cousider the Cantor seaircase $W=g(\omeya)$ assocnated dith the cprcle map: we land in Dynamical Syvjems, where coinections wity Nunber Thdury are old anc well exiljred. | partition of $\Omega$, are two strong factors us linking $f(\alpha)=d_H(\Omega_{\alpha(\beta)})$ $d_H(G_{\beta})$ directly through, as equality. Still, we 1. For $\beta=\infty$ have $\alpha(\beta)=\alpha_{\min}$. 2. For $\beta=2$ we $\alpha(\beta)=\alpha_{\max}$. 3. $f(\alpha_{\min})=0$. 4. $f(\alpha)$ is increasing for the greater part of the $[\alpha_{\min},\alpha_{\max}]$. 5. Let $\max\limits_{\alpha}f(\alpha)=f(\alpha^{\max})$. Then $\alpha_{\max}$ is very near $\alpha^{\max}$. 6. The $\Omega_{\alpha}$ strongly to types 7. The sets $G_{\infty}$ and $G_{2}$, related to $\alpha_{\min}$ and $\alpha_{\max}$ in $\Omega$, are related to and $\alpha_{\min}$ in $I=[0,1]$ endowed with the hyperbolic of probability induced by The Spectrum $(\alpha, f(\alpha))$ of Fractal $\Omega$ underlying Circle Staircase Conclusions 1) to in the last section show a strong connection between the magnetization function $q=g(-H)$ and leading problems in Theory –viz approximation of numbers with classes $J_{\beta}$ and $G_{\beta},\;\beta\geq 2$. This particular connection between Theory is seen only when analyzing the multifractal of the set $\Omega$ underlying the magnetization Cantor The conclusions (1) to 7)) are based, as have seen, on two premises about the staircase $q=g(-H)$: the $(F-B)$ arrangement of the stairsteps in the staircase, and formula given by (\[fases\]). us consider Cantor staircase associated with the circle map: we land in Dynamical Systems, where with Number Theory are old and well explored. | partition of $\Omega$, are two strOng factors Which HinDer Us From LinkIng $f(\alpha)=d_H(\OmeGA_{\alpHa(\beta)})$ with $d_H(G_{\beta})$ direcTly thRoUGh, saY, So SimplE a way as AN eQUAliTy. stIll, We KNoW:
1. For $\bEta=\Infty$ we Have $\alpha(\bEta)=\AlPha_{\min}$.
2. For $\betA=2$ We Have $\alpha(\bEta)=\Alpha_{\max}$.
3. $f(\alpHa_{\mIn})=0$.
4. $f(\alpHa)$ Is iNCreasIng For thE greatER part oF the interVaL $[\Alpha_{\mIN},\alpha_{\mAX}]$.
5. leT $\max\Limits_{\alpha}f(\alpha)=F(\AlPHa^{\max})$. Then $\alpha_{\Max}$ is vErY NeAR $\AlpHa^{\mAx}$.
6. The $\Omega_{\AlPha}$ arE StronglY LiNKED to THe types $G_{\beta}$.
7. THe sets $G_{\inftY}$ And $g_{2}$, relatEd To $\aLPha_{\min}$ And $\alPhA_{\Max}$ In $\Omega$, are rElatEd to $\alpha_{\Max}$ and $\ALpha_{\min}$ IN $I=[0,1]$ endowEd with The HypErboLIc MeAsuRe OF prOBaBilITy iNduced by faReY-BrocOt.
ThE sPECtruM $(\alPha, f(\Alpha))$ Of the Fractal SEt $\OMega$ UNdeRlyinG the CIrclE MAp StaIrcase
==================================================================================================
conclUsIons 1) to 7) in the last SectIon show a sTroNg ConNeCtion BEtween The MagNetizatIon funcTIon $Q=g(-h)$ AND lEading problems in NuMbER thEory –viz tHe good APpRoXImation oF iRraTionAL NumbeRs stUDiEd with JaRník clASsEs $j_{\beta}$ anD tHeir reFiNemEntS $G_{\betA},\;\Beta\Geq 2$. ThiS particuLar coNNection between MAgnetization aND NUMBeR theoRy iS seen only whEn anALyziNg thE MuLtiFRactaL specTrUM oF The fractal set $\Omega$ uNdErlyinG the mAgnetization CAntor stairCASE. The concLusiONs (1) TO 7)) are based, as we hAve seEn, on two preMIses abouT the sTaircase $Q=g(-H)$: the $(F-B)$ aRRAngement Of tHe sTaiRstEPS $\DElta H$ in the staIRCase, AnD the forMulA given bY Eq. (\[FasEs\]).
LEt uS nOw consideR the CantOr StAiRcAse $w=g(\omeGA)$ associaTeD wiTh The CirclE Map: we lAnd in dynaMiCaL sysTems, wheRE cONNectIoNs With numBeR TheoRy arE Old And well Explored. | partition of $\Omega$, are two stron g fac tor s w hi ch h inde r us from link i ng $ f(\alpha)=d_H(\Omega_{ \alph a( \ beta ) }) $ wit h $d_H( G _{ \ b eta }) $dir ec t ly thro ugh , say,so simplea w ay as an equal i ty . Still, w e k now:
1. Fo r $ \beta= \i nft y $ wehav e $\a lpha(\ b eta)=\ alpha_{\m in } $.
2. For $\ b e ta =2$we have $\alpha(\ b et a )=\alpha_{\max }$.
3 .$f ( \ alp ha_ {\min})=0$ .
4. $ f(\alph a )$ i s in c reasing for t he greaterp art of th eint e rval $ [\alp ha _ {\m in},\alpha_ {\ma x}]$.
5. Let$ \max\li m its_{\a lpha}f (\a lph a)=f ( \a lp ha^ {\ m ax} ) $. Th e n $ \alpha_{ \m ax }$ is ver y n e ar $ \al pha^ {\max }$.
6. The$\O mega _ {\a lpha} $ are str on gly l inkedto th etypes $G_{\beta }$.
7. Theset s$G_ {\ infty } $ and$G_ {2} $, rela ted to$ \al ph a _ { \m in}$ and $\alpha_{ \m a x }$ in $\Om ega$,a re r e lated to $ \al pha_ { \ max}$ and $\ alpha_{\ min}$i n$I =[0,1]$ e ndowed w ith th e hyp e rbol ic mea sure ofproba b ility inducedb y Farey-Broco t .T he Spec tru m $(\alpha, f(\ a lpha ))$o fthe Fract al Se t$ \O m ega$ underlying the C ircleMap S taircase
==== ========== = = = ======== ==== = == = ============== ===== ========== = ======== ===== ======== ========= = = ===
Con clu sio ns1)t o 7 ) in the last s ecti on show a st rong co nne cti onbet we en the ma gnetizat io nfu nc tio n $q= g (-H)$ an dlea di ngprobl e ms inNumbe r Th eo ry –vi z the g o od a ppro xi ma tion of i rrati onal num bers st udied wit h J a rník c la sses $J _{\beta}$ and t heir refin em ent s $G_{ \ b eta},\;\ beta\geq 2$. This parti c ular co nne ction bet ween magn eti zation an d Numbe r Theo ry is s een o nly w h e nana ly zing the m u l tif racta lspec trum of the fractal set $ \ Ome ga$ underlyin g t he m a g ne tiz a ti o n C an t ors t aircase. The co nclusions(1 ) t o 7)) areb ase d, as wehave se en, o n two pr emises ab out the s ta irca s e $q =g(-H)$: t he $(F-B )$ arrang e mento fthe s tai rsteps $ \De lta H $ in t h e s tairc ase, a nd the f ormul agiven by Eq. (\[fases\]).
Letus now cons ide r the Can tor sta ircase $W =g(\ omega)$ as soc iat ed wi tht he ci rcle ma p:w e lan d in Dynamical Sy ste m s ,where conne c t i ons with Nu m ber Th eory are old and well explored. | partition of_$\Omega$, are_two strong factors which_hinder us_from_linking $f(\alpha)=d_H(\Omega_{\alpha(\beta)})$_with_$d_H(G_{\beta})$ directly through,_say, so simple_a way as an_equality. Still, we_know:
1._ For $\beta=\infty$ we have $\alpha(\beta)=\alpha_{\min}$.
2. For $\beta=2$ we have $\alpha(\beta)=\alpha_{\max}$.
3. $f(\alpha_{\min})=0$.
4.__$f(\alpha)$ is_increasing_for_the greater part of the_interval $[\alpha_{\min},\alpha_{\max}]$.
5. Let $\max\limits_{\alpha}f(\alpha)=f(\alpha^{\max})$._Then $\alpha_{\max}$_is very near $\alpha^{\max}$.
6. The $\Omega_{\alpha}$ are_strongly_linked to the_types $G_{\beta}$.
7. The sets $G_{\infty}$ and $G_{2}$, related_to $\alpha_{\min}$ and $\alpha_{\max}$ in $\Omega$,_are related to_$\alpha_{\max}$_and_$\alpha_{\min}$ in $I=[0,1]$ endowed_with the hyperbolic measure of probability_induced by Farey-Brocot.
The Spectrum $(\alpha, f(\alpha))$_of the Fractal Set $\Omega$ underlying the_Circle Map Staircase
==================================================================================================
Conclusions 1) to 7)_in the last section show_a strong_connection between the magnetization function_$q=g(-H)$ and leading_problems in_Number Theory –viz_the good approximation of irrational numbers_studied with Jarník_classes $J_{\beta}$ and their refinements $G_{\beta},\;\beta\geq_2$._This particular connection_between_magnetization_and Number_Theory is seen_only_when analyzing_the_multifractal spectrum of the fractal set_$\Omega$_underlying the magnetization Cantor staircase. The conclusions_(1) to 7)) are_based,_as we have seen,_on two premises about the_staircase $q=g(-H)$: the $(F-B)$ arrangement of_the stairsteps_$\Delta H$_in the staircase, and the formula given by Eq. (\[fases\]).
Let us_now consider the Cantor staircase $W=g(\omega)$_associated with the circle_map: we_land_in Dynamical Systems,_where_connections with_Number Theory are old and well explored. |
u_1-u_2|^\nu |\Theta_n^*(u)|du \\
& \le c \,n^{4r-\nu -1}
\int_{0}^{1/n} \left| \frac{\sin\frac{(n+1)u_2 \pi}{2}} {\sin \frac{u_2\pi}{2}} \right|^{2r}
du \le c \, n^{6 r - \nu -2}.\end{aligned}$$ upon using $|\sin n u / \sin u| \le n$. The other two cases can be handled similarly. As a result, we conclude that $I_n^{r,p} \le c n^{6 r - p -2}$. The case $p = 0$ gives the lower bound estimate of. The desired estimate is over the quantity $\lambda_{n,r} I_n^{r,p}$ and follows from our estimates.
Using the kernel $K_{n,r}$ we can now prove a Jackson estimate:
For $1 \le p \le \infty$ and for each $r = 1,2,\ldots$, there is a constant $c_r$ such that if $f\in L^p$ then $$E_n (f)_p \le c_r \omega_r (f, \tfrac{1}{n})_p, \qquad n = 1, 2,....$$
As in the proof of classical Jackson estimate for trigonometric polynomials on $[0, 2\pi]$, we consider the following operator $$F_n^{\rho,r} f (\xb):= \int_{\Omega} J_{n,\rho}(\tb) \sum_{k=1}^r (-1)^{k-1}
\binom{r}{k} f(\xb + k \tb) d\tb,$$ where $J_{n,\rho}(\tb) = K_{n^*,\rho}(\tb)$ with $n^* = \lfloor \frac{n}{\rho} \rfloor +1$, and $\rho \ge (r+2)/2$. Evidently, $J_{n,\rho}(-\tb) = J_{n,\rho}(\tb)$. Using the fact that $J_{n,\rho} \in \CH_n$, we see that $F_{n,\rho} f$ can be written as a linear combination of $$\ | u_1 - u_2|^\nu |\Theta_n^*(u)|du \\
& \le c \,n^{4r-\nu -1 }
\int_{0}^{1 / n } \left| \frac{\sin\frac{(n+1)u_2 \pi}{2 } } { \sin \frac{u_2\pi}{2 } } \right|^{2r }
du \le c \, n^{6 r - \nu -2}.\end{aligned}$$ upon using $ |\sin n u / \sin u| \le n$. The other two case can be cover similarly. As a result, we conclude that $ I_n^{r, p } \le c n^{6 gas constant - p -2}$. The case $ phosphorus = 0 $ gives the lower bound appraisal of. The desire estimate is over the measure $ \lambda_{n, r } I_n^{r, p}$ and follows from our estimates.
Using the kernel $ K_{n, r}$ we can nowadays prove a Jackson estimate:
For $ 1 \le p \le \infty$ and for each $ r = 1,2,\ldots$, there is a changeless $ c_r$ such that if $ f\in L^p$ then $ $ E_n (f)_p \le c_r \omega_r (f, \tfrac{1}{n})_p, \qquad n = 1, 2,.... $$
As in the proof of classical Jackson estimate for trigonometric polynomial on $ [ 0, 2\pi]$, we consider the following hustler $ $ F_n^{\rho, r } f (\xb):= \int_{\Omega } J_{n,\rho}(\tb) \sum_{k=1}^r (-1)^{k-1 }
\binom{r}{k } f(\xb + k \tb) d\tb,$$ where $ J_{n,\rho}(\tb) = K_{n^*,\rho}(\tb)$ with $ n^ * = \lfloor \frac{n}{\rho } \rfloor +1 $, and $ \rho \ge (r+2)/2$. Evidently, $ J_{n,\rho}(-\tb) = J_{n,\rho}(\tb)$. use the fact that $ J_{n,\rho } \in \CH_n$, we see that $ F_{n,\rho } f$ can be write as a linear combination of $ $ \ | u_1-u_2|^\nk |\Theta_n^*(u)|du \\
& \le c \,n^{4v-\nu -1}
\int_{0}^{1/n} \left| \frac{\sii\frac{(n+1)u_2 \pi}{2}} {\sin \wrac{u_2\pi}{2}} \right|^{2r}
du \le r \, n^{6 r - \ny -2}.\end{aligned}$$ upon usine $|\sin n u / \sin u| \oe n$. Rhe other vso cases can gc hanblxd similarly. As a result, fe conclude thdt $I_u^{r,p} \le c n^{6 r - p -2}$. The case $p = 0$ gives ehe lowrr bound estimatg of. Uhe desjgeb estimate is over the quantity $\lambda_{i,r} I_n^{r,p}$ and folkows from our estimates.
Usijg tje kernel $K_{n,r}$ we cwn now provg a Tqckson estimxte:
For $1 \le p \le \infty$ znd for each $r = 1,2,\ldots$, there is x conxtant $c_r$ soey tjdt if $f\in L^'$ then $$E_n (f)_p \le c_r \omega_r (x, \tfrac{1}{m})_p, \qquad n = 1, 2,....$$
Ax ii thw proof of classical Oackson estimate for trigonomatxic polynomials on $[0, 2\pu]$, qe cotsidar tfw fullkwmng operahor $$F_n^{\rho,r} f (\sb):= \int_{\Omega} J_{n,\rho}(\tb) \sum_{k=1}^r (-1)^{k-1}
\fpmom{r}{k} f(\xb + k \tb) d\eb,$$ where $J_{n,\rho}(\tb) = K_{n^*,\rho}(\tb)$ with $n^* = \lfloog \frzc{n}{\rho} \rfloor +1$, and $\rho \te (r+2)/2$. Evidently, $J_{n,\rho}(-\th) = J_{n,\rho}(\tf)$. Using the fact that $J_{n,\rho} \in \CH_n$, we see that $F_{n,\sho} f$ zan bc wrkrtfn as a linear combination of $$\ | u_1-u_2|^\nu |\Theta_n^*(u)|du \\ & \le c \,n^{4r-\nu \left| \pi}{2}} {\sin \right|^{2r} du \le \nu upon using $|\sin u / \sin \le n$. The other two cases be handled similarly. As a result, we conclude that $I_n^{r,p} \le c n^{6 - p -2}$. The case $p = 0$ gives the lower bound estimate The estimate over quantity $\lambda_{n,r} I_n^{r,p}$ and follows from our estimates. Using the kernel $K_{n,r}$ we can now prove Jackson estimate: For $1 \le p \le \infty$ for each $r = there is a constant $c_r$ that $f\in L^p$ $$E_n \le \omega_r (f, \tfrac{1}{n})_p, n = 1, 2,....$$ As in the proof of classical Jackson estimate for trigonometric polynomials on $[0, we consider operator $$F_n^{\rho,r} (\xb):= J_{n,\rho}(\tb) (-1)^{k-1} \binom{r}{k} f(\xb \tb) d\tb,$$ where $J_{n,\rho}(\tb) = K_{n^*,\rho}(\tb)$ \lfloor \frac{n}{\rho} \rfloor +1$, and $\rho \ge (r+2)/2$. $J_{n,\rho}(-\tb) = Using the fact that $J_{n,\rho} \in we see that $F_{n,\rho} f$ can be written a linear combination of $$\ | u_1-u_2|^\nu |\Theta_n^*(u)|du \\
& \le c \,n^{4r-\nu -1}
\int_{0}^{1/n} \lEft| \frac{\sin\Frac{(n+1)U_2 \pi}{2}} {\Sin \FrAc{u_2\pI}{2}} \rigHt|^{2r}
du \le c \, n^{6 r - \nu -2}.\enD{AligNed}$$ upon using $|\sin n u / \sin u| \lE n$. The OtHEr twO CaSes caN be handLEd SIMilArLy. as a ReSUlT, we coNclUde that $i_n^{r,p} \le c n^{6 r - p -2}$. the CaSe $p = 0$ gives the lOWeR bound estiMatE of. The desireD esTimate Is OveR The quAntIty $\laMbda_{n,r} i_N^{r,p}$ and Follows frOm OUr estiMAtes.
UsiNG ThE kerNel $K_{n,r}$ we can now proVE a jAckson estimate:
for $1 \le p \Le \INfTY$ And For Each $r = 1,2,\ldots$, ThEre is A ConstanT $C_r$ SUCH thAT if $f\in L^p$ then $$E_N (f)_p \le c_r \omegA_R (f, \tFrac{1}{n})_p, \QqUad N = 1, 2,....$$
as in thE prooF oF ClaSsical JacksOn esTimate for TrigonOMetric pOLynomiaLs on $[0, 2\pi]$, We cOnsIder THe FoLloWiNG opERaTor $$f_N^{\rhO,r} f (\xb):= \int_{\omEgA} J_{n,\rhO}(\tb) \sUM_{K=1}^R (-1)^{K-1}
\binOm{r}{K} f(\xb + K \tb) d\tB,$$ where $J_{n,\rho}(\tb) = k_{n^*,\rHo}(\tb)$ WIth $N^* = \lfloOr \fraC{n}{\rhO} \rFloor +1$, And $\rho \Ge (r+2)/2$. EvIdEntly, $J_{n,\rho}(-\tb) = J_{n,\rHo}(\tb)$. using the fAct ThAt $J_{N,\rHo} \in \Ch_N$, we see ThaT $F_{n,\Rho} f$ can Be writtEN as A lINEAr Combination of $$\ | u_1-u_2|^\nu |\Theta_n^*(u )|du \\
&\le c\, n^{4 r-\n u -1}
\int_ { 0}^{ 1/n} \left| \frac{\sin \frac {( n +1)u _ 2\pi}{ 2}} {\s i n\ f rac {u _2 \pi }{ 2 }} \rig ht| ^{2r}
du \le c \, n^{6 r - \n u -2}.\end {al igned}$$ upo n u sing $ |\ sin n u / \s in u| \le n $ . Theother two c a ses ca n be han d l ed sim ilarly. As a resu l t, we conclude th at $I_ n^ { r, p } \l e c n^{6 r -p-2}$. The cas e $ p = 0$ gives the low er bound es t ima te of. T hed esired esti ma t e i s over thequan tity $\la mbda_{ n ,r} I_n ^ {r,p}$and fo llo wsfrom ou rest im a tes .
Usi n g t he kerne l$K _{n,r }$ w e c a n no w p rove a Ja ckson estimat e:
For $1\le p \le\inf ty $ and for e ach $ r= 1,2,\ldots$,ther e is a co nst an t $ c_ r$ su c h that if $f \in L^p $ then$ $E_ n( f ) _p \le c_r \omega_r(f , \t frac{1}{ n})_p, \q qu a d n = 1, 2 ,.. ..$$ As in the pr oof of c lassic a lJa ckson e st imatefo r t rig onome t ricpolyno mials on $[0, 2\pi]$, we con s ider the foll o wi n g o p erat or$$F_n^{\rho ,r}f (\x b):= \i nt_ { \Omeg a} J_ {n , \r h o}(\tb) \sum_{k=1}^ r(-1)^{ k-1}
\binom{r }{k} f( \ x b + k \tb ) d\ t b, $ $ where $J_{n, \rho} (\tb) = K_ { n^*,\rho }(\tb )$ with$n^* = \l f l oor \fra c{n }{\ rho } \ r f lo or +1$, and $ \ r ho \ ge (r+2)/ 2$. Eviden tly , $ J_{ n,\ rh o}(-\tb)= J_{n,\ rh o} (\ tb )$. Usin g the fac ttha t$J_ {n,\r h o} \in \CH_ n$,we s e e t hat $F_ { n, \ r ho}f$ c an b e w ri ttenas a lin ear com binationof$ $\ | u_1-u_2|^\nu |\Theta_n^*(u)|du_\\
_ &_\le c_\,n^{4r-\nu_-1}
__\int_{0}^{1/n} \left| \frac{\sin\frac{(n+1)u_2_\pi}{2}} {\sin \frac{u_2\pi}{2}}_\right|^{2r}
_ du__ \le c \, n^{6 r - \nu -2}.\end{aligned}$$ upon using $|\sin n u_/_\sin u|_\le_n$._The other two cases can_be handled similarly. As a_result, we_conclude that $I_n^{r,p} \le c n^{6 r -_p_-2}$. The case_$p = 0$ gives the lower bound estimate of._The desired estimate is over the_quantity $\lambda_{n,r} I_n^{r,p}$_and_follows_from our estimates.
Using the_kernel $K_{n,r}$ we can now prove_a Jackson estimate:
For $1 \le p_\le \infty$ and for each $r =_1,2,\ldots$, there is a constant $c_r$_such that if $f\in L^p$_then $$E_n_(f)_p \le c_r \omega_r (f,_\tfrac{1}{n})_p, \qquad n_= 1,_2,....$$
As in the_proof of classical Jackson estimate for_trigonometric polynomials on_$[0, 2\pi]$, we consider the following_operator_$$F_n^{\rho,r} f (\xb):=_\int_{\Omega}_J_{n,\rho}(\tb)_\sum_{k=1}^r (-1)^{k-1}_
__\binom{r}{k} __ f(\xb + k \tb) d\tb,$$_where_$J_{n,\rho}(\tb) = K_{n^*,\rho}(\tb)$ with $n^* = \lfloor_\frac{n}{\rho} \rfloor +1$, and_$\rho_\ge (r+2)/2$. Evidently, $J_{n,\rho}(-\tb)_= J_{n,\rho}(\tb)$. Using the fact_that $J_{n,\rho} \in \CH_n$, we see_that $F_{n,\rho}_f$ can_be written as a linear combination of $$\ |
[@Sayed2013intr] for definition and properties of the norm): $$\label{eq:spr}
\begin{split}
\rho \left({\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}})\right]\right) \leq \|{\boldsymbol{A}}_I^\top \left({\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}})\right)\|_{b,\infty}
\end{split}$$ Now using norm inequalities and the fact that ${\boldsymbol{A}}$ is a left-stochastic matrix (whose block maximum norm is equal to one), we find that: $$\label{eq:normineqs}
\begin{split}
\|{\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}})\right]\|_{b,\infty}
&\leq \|{\boldsymbol{A}}_I^\top\|_{b,\infty} \cdot \| {\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}}) \|_{b,\infty} \\
& = \| {\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}})\|_{b,\infty} \\
& \leq \| {\boldsymbol{I}}_{LN} - \mu\, {\boldsymbol{H}}_R - \mu\eta {\boldsymbol{I}}_{LN}\|_{b,\infty}+ \mu\eta \, \|{\boldsymbol{P}}\otimes {\boldsymbol{I}}_L\|_{b,\infty}
\end{split}$$ using the definition ${\boldsymbol{Q}}= ({\boldsymbol{I}}_N - {\boldsymbol{P}}) \otimes {\boldsymbol{I}}_L$ and the triangle inequality. Now, it holds that $$\label{eq:normP}
\begin{split}
\|{\boldsymbol{P}}\otimes {\boldsymbol{I}}_L\|_{b,\infty}= \| {\boldsymbol{P}}\|_{\infty} = 1
\end{split}$$ because ${\boldsymbol{P}}$ is a right stochastic matrix according to condition . Furthermore, since $(1-\mu\eta)\,{\boldsymbol{I}}_{LN} - \mu\,{\ | [ @Sayed2013intr ] for definition and properties of the norm ): $ $ \label{eq: spr }
\begin{split }
\rho \left({\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN } - \mu\, ({ \boldsymbol{H}}_R + \eta \, { \boldsymbol{Q}})\right]\right) \leq \|{\boldsymbol{A}}_I^\top \left({\boldsymbol{I}}_{LN } - \mu\, ({ \boldsymbol{H}}_R + \eta \, { \boldsymbol{Q}})\right)\|_{b,\infty }
\end{split}$$ Now using average inequality and the fact that $ { \boldsymbol{A}}$ is a leftover - stochastic matrix (whose block maximum norm is adequate to one), we find that: $ $ \label{eq: normineqs }
\begin{split }
\|{\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN } - \mu\, ({ \boldsymbol{H}}_R + \eta \, { \boldsymbol{Q}})\right]\|_{b,\infty }
& \leq \|{\boldsymbol{A}}_I^\top\|_{b,\infty } \cdot \| { \boldsymbol{I}}_{LN } - \mu\, ({ \boldsymbol{H}}_R + \eta \, { \boldsymbol{Q } }) \|_{b,\infty } \\
& = \| { \boldsymbol{I}}_{LN } - \mu\, ({ \boldsymbol{H}}_R + \eta \, { \boldsymbol{Q}})\|_{b,\infty } \\
& \leq \| { \boldsymbol{I}}_{LN } - \mu\, { \boldsymbol{H}}_R - \mu\eta { \boldsymbol{I}}_{LN}\|_{b,\infty}+ \mu\eta \, \|{\boldsymbol{P}}\otimes { \boldsymbol{I}}_L\|_{b,\infty }
\end{split}$$ using the definition $ { \boldsymbol{Q}}= ({ \boldsymbol{I}}_N - { \boldsymbol{P } }) \otimes { \boldsymbol{I}}_L$ and the triangle inequality. nowadays, it holds that $ $ \label{eq: normP }
\begin{split }
\|{\boldsymbol{P}}\otimes { \boldsymbol{I}}_L\|_{b,\infty}= \| { \boldsymbol{P}}\|_{\infty } = 1
\end{split}$$ because $ { \boldsymbol{P}}$ is a right stochastic matrix harmonize to condition . Furthermore, since $ (1-\mu\eta)\,{\boldsymbol{I}}_{LN } - \mu\,{\ | [@Sayfd2013intr] for definition ana properties of the nocm): $$\labem{eq:spr}
\begin{split}
\rho \left({\bolddynbol{A}}_U^\top \left[{\boldsymbol{I}}_{LN} - \mu\, ({\bolddymbol{H}}_R + \ete \, {\boldsymbol{Q}})\rijgt]\right) \leq \|{\bkpdsykuol{A}}_I^\top \left({\bokdsymbol{I}}_{LT} - \mu\, ({\boldsymbml{F}}_R + \eta \, {\boldsymbol{Q}})\right)\|_{b,\infty}
\enq{split}$$ Moa using norm igequsjitiss and the fact that ${\boldsymbol{A}}$ ia a lefu-stochastic matrix (whose block maximum norm ls eeual to one), we finf that: $$\labeo{eq:njemineqs}
\bdgin{split}
\|{\boldsymbom{A}}_I^\top \left[{\boldsymbol{I}}_{LN} - \mu\, ({\boudsymyol{H}}_R + \eta \, {\bilddimbol{Q}})\right]\|_{b,\mnfty}
&\leq \|{\boldsnkbol{A}}_I^\dop\|_{b,\infyy} \cdot \| {\boldsnmbol{M}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}}) \|_{b,\infjy} \\
& = \| {\bkldsymbol{I}}_{LN} - \mu\, ({\voodsymtol{H}}_S + \egq \, {\colssbmbkl{Q}})\|_{b,\invty} \\
& \leq \| {\boldsymbol{U}}_{LN} - \mu\, {\boldsymbol{H}}_T - \nu\eta {\boldsymgol{I}}_{LN}\|_{f,\igfty}+ \mu\eta \, \|{\boldsymbol{P}}\otimes {\boldsymbon{I}}_L\|_{g,\infty}
\end{split}$$ usint the definition ${\bolddymbol{Q}}= ({\bjldsymbol{I}}_N - {\boldsymbol{P}}) \otimes {\boldsymbol{I}}_L$ and dhe tcixngoe kbeeuality. Now, it holds that $$\label{eq:normP}
\begyh{sklin}
\|{\boldsymbjl{P}}\otimes {\bplfsuibol{I}}_L\|_{b,\infty}= \| {\boldsvjbkl{P}}\|_{\infty} = 1
\end{solit}$$ besause ${\boldsymbjl{P}}$ os a right stochastic matriz according no cindition . Furthermoxe, since $(1-\mu\eca)\,{\boldxymbok{I}}_{LN} - \mu\,{\ | [@Sayed2013intr] for definition and properties of the \begin{split} \left({\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN} \mu\, ({\boldsymbol{H}}_R + \left({\boldsymbol{I}}_{LN} \mu\, ({\boldsymbol{H}}_R + \, {\boldsymbol{Q}})\right)\|_{b,\infty} \end{split}$$ using norm inequalities and the fact ${\boldsymbol{A}}$ is a left-stochastic matrix (whose block maximum norm is equal to one), find that: $$\label{eq:normineqs} \begin{split} \|{\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}})\right]\|_{b,\infty} \|{\boldsymbol{A}}_I^\top\|_{b,\infty} \| - ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}}) \|_{b,\infty} \\ & = \| {\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta {\boldsymbol{Q}})\|_{b,\infty} \\ & \leq \| {\boldsymbol{I}}_{LN} - \mu\, - \mu\eta {\boldsymbol{I}}_{LN}\|_{b,\infty}+ \mu\eta \|{\boldsymbol{P}}\otimes {\boldsymbol{I}}_L\|_{b,\infty} \end{split}$$ using the ${\boldsymbol{Q}}= - {\boldsymbol{P}}) {\boldsymbol{I}}_L$ the inequality. Now, it that $$\label{eq:normP} \begin{split} \|{\boldsymbol{P}}\otimes {\boldsymbol{I}}_L\|_{b,\infty}= \| {\boldsymbol{P}}\|_{\infty} = 1 \end{split}$$ because ${\boldsymbol{P}}$ is a right stochastic matrix to condition since $(1-\mu\eta)\,{\boldsymbol{I}}_{LN} \mu\,{\ | [@Sayed2013intr] for definition and Properties Of the NorM): $$\laBeL{eq:sPr}
\beGin{split}
\rho \lefT({\BoldSymbol{A}}_I^\top \left[{\boldsymBol{I}}_{Ln} - \mU\, ({\BoldSYmBol{H}}_R + \Eta \, {\boldSYmBOL{Q}})\rIgHt]\RigHt) \LEq \|{\BoldsYmbOl{A}}_I^\top \Left({\boldsyMboL{I}}_{lN} - \mu\, ({\boldsymbOL{H}}_r + \eta \, {\boldsyMboL{Q}})\right)\|_{b,\inftY}
\enD{split}$$ noW usINg norM inEqualIties aND the faCt that ${\bolDsYMbol{A}}$ iS A left-stOCHaStic Matrix (whose block mAXiMUm norm is equal tO one), we FiND tHAT: $$\laBel{Eq:normineqS}
\bEgin{sPLit}
\|{\boldSYmBOL{a}}_I^\tOP \left[{\boldsymbOl{I}}_{LN} - \mu\, ({\boldSYmbOl{H}}_R + \etA \, {\bOldSYmbol{Q}})\Right]\|_{B,\iNFty}
&\Leq \|{\boldsymbOl{A}}_I^\Top\|_{b,\infty} \Cdot \| {\boLDsymbol{i}}_{lN} - \mu\, ({\bolDsymboL{H}}_R + \Eta \, {\BoldSYmBoL{Q}}) \|_{b,\InFTy} \\
& = \| {\bOLdSymBOl{I}}_{lN} - \mu\, ({\boldSyMbOl{H}}_R + \eTa \, {\boLDSYMbol{q}})\|_{b,\iNfty} \\
& \Leq \| {\boLdsymbol{I}}_{LN} - \mu\, {\BolDsymBOl{H}}_r - \mu\etA {\boldSymbOl{i}}_{LN}\|_{b,\iNfty}+ \mu\Eta \, \|{\boLdSymbol{P}}\otimes {\boLdsyMbol{I}}_L\|_{b,\inFty}
\EnD{spLiT}$$ usinG The defIniTioN ${\boldsyMbol{Q}}= ({\boLDsyMbOL{i}}_n - {\bOldsymbol{P}}) \otimes {\boLdSYMbOl{I}}_L$ and tHe triaNGlE iNEquality. noW, it HoldS THat $$\laBel{eQ:NoRmP}
\begin{Split}
\|{\bOLdSyMbol{P}}\otImEs {\boldSyMboL{I}}_L\|_{B,\inftY}= \| {\BoldSymbol{p}}\|_{\infty} = 1
\enD{spliT}$$ Because ${\boldsymBOl{P}}$ is a right stOChASTiC MatrIx aCcording to cOndiTIon . FUrthERmOre, SInce $(1-\mU\eta)\,{\bOlDSyMBol{I}}_{LN} - \mu\,{\ | [@Sayed2013intr] for defin ition andprope rti esof the nor m): $$\label{e q :spr }
\begin{split}
\rh o\ left ( {\ bolds ymbol{A } }_ I ^ \to p\l eft [{ \ bo ldsym bol {I}}_{L N} - \mu\, ({ \b oldsymbol{H} } _R + \eta \, {\ boldsymbol{Q }}) \right ]\ rig h t) \l eq\|{\b oldsym b ol{A}} _I^\top \ le f t({\bo l dsymbol { I }} _{LN } - \mu\, ({\bold s ym b ol{H}}_R + \et a \, { \b o ld s y mbo l{Q }})\right) \| _{b,\ i nfty}
\ e n d{s p lit}$$ Now us ing norm in e qua lities a ndt he fac t tha t$ {\b oldsymbol{A }}$is a left -stoch a stic ma t rix (wh ose bl ock ma ximu m n or m i se qua l t o o n e), we find t ha t: $$ \lab e l { e q:no rmi neqs }
\begin{split }
\ | {\b oldsy mbol{ A}}_ I^ \top\left[ {\bol ds ymbol{I}}_{LN}- \m u\, ({\bo lds ym bol {H }}_R+ \eta\,{\b oldsymb ol{Q}}) \ rig ht ] \ | _{ b,\infty}
&\le q\ | {\ boldsymb ol{A}} _ I^ \t o p\|_{b,\ in fty } \c d o t \|{\bo l ds ymbol{I} }_{LN} -\m u\, ({\ bo ldsymb ol {H} }_R + \e t a \, {\bol dsymbol{ Q}})\ |_{b,\infty} \ \
& = \|{ \b o l ds y mbol {I} }_{LN} - \m u\,( {\bo ldsy m bo l{H } }_R + \eta \ , { \ boldsymbol{Q}})\|_{ b, \infty } \\ & \leq \| {\boldsym b o l {I}}_{LN } -\ mu \ , {\boldsymbol {H}}_ R - \mu\et a {\bolds ymbol {I}}_{LN }\|_{b,\i n f ty}+ \mu \et a \ , \ |{\ b o ld symbol{P}}\ot i m es { \b oldsymb ol{ I}}_L\| _{b ,\i nft y} \end{sp lit}$$ u si ng t he de finit i on ${\bo ld sym bo l{Q }}= ( { \bolds ymbol {I}} _N - {\b oldsymb o l{ P } }) \ ot im es { \bo ld symbo l{I} } _L$ and th e triangl e i n equa li ty . Now,it holds that $ $\label{eq :n orm P}
\ begin{sp lit}
\|{\boldsy m bol{P}} \ot imes{\bo ldsymbol{ I}} _L\|_{ b,\ i nfty}= \| {\ bolds ym bol { P }}\|_ { \ in fty }= 1
\e n d {sp lit}$ $beca use ${\ boldsymbol{P}}$ is a r ight stochast icmatr i x a cco r di n g t oc ond i t ion . Furthermo re, since$( 1 -\ mu\eta)\,{ \ bol ds ymbol{I }}_{LN} - \m u \,{\ | [@Sayed2013intr] for_definition and_properties of the norm):_$$\label{eq:spr}
__ \begin{split}
__ \rho_\left({\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN} -_\mu\, ({\boldsymbol{H}}_R + \eta_\, {\boldsymbol{Q}})\right]\right) \leq_\|{\boldsymbol{A}}_I^\top_\left({\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}})\right)\|_{b,\infty}
\end{split}$$ Now using_norm_inequalities and_the_fact_that ${\boldsymbol{A}}$ is a left-stochastic_matrix (whose block maximum norm_is equal_to one), we find that: $$\label{eq:normineqs}
__\begin{split}
_ \|{\boldsymbol{A}}_I^\top \left[{\boldsymbol{I}}_{LN} - \mu\, ({\boldsymbol{H}}_R + \eta \,_{\boldsymbol{Q}})\right]\|_{b,\infty}
&\leq \|{\boldsymbol{A}}_I^\top\|_{b,\infty}_\cdot \| {\boldsymbol{I}}_{LN}_-_\mu\,_({\boldsymbol{H}}_R + \eta \,_{\boldsymbol{Q}}) \|_{b,\infty} \\
_ & = \| {\boldsymbol{I}}_{LN} -_\mu\, ({\boldsymbol{H}}_R + \eta \, {\boldsymbol{Q}})\|_{b,\infty} \\
_ & \leq \|_{\boldsymbol{I}}_{LN} - \mu\, {\boldsymbol{H}}_R -_\mu\eta {\boldsymbol{I}}_{LN}\|_{b,\infty}+_\mu\eta \, \|{\boldsymbol{P}}\otimes {\boldsymbol{I}}_L\|_{b,\infty}
_ \end{split}$$_using the_definition ${\boldsymbol{Q}}= ({\boldsymbol{I}}_N_- {\boldsymbol{P}}) \otimes {\boldsymbol{I}}_L$ and the_triangle inequality. Now,_it holds that $$\label{eq:normP}
__\begin{split}
___ _ \|{\boldsymbol{P}}\otimes {\boldsymbol{I}}_L\|_{b,\infty}=_\|_{\boldsymbol{P}}\|_{\infty} =_1
_ \end{split}$$ because ${\boldsymbol{P}}$_is_a right stochastic matrix according to condition ._Furthermore, since $(1-\mu\eta)\,{\boldsymbol{I}}_{LN} -_\mu\,{\ |
= F(\sqrt{\alpha_1\alpha_2})$, so ${\mathfrak{p}}$ ramifies in $E_i$ for an even number of $i$’s. In particular, ${\mathfrak{p}}$ is not totally ramified in $E/F$.
\[ram1\] Let $K$ be a non-biquadratic quartic CM field of type $\Phi$ with the real quadratic subfield $F$ and let $K^r$ be its reflex field with the quadratic subfield $F^r$. Then the following assertions hold.
- If a prime $p$ is ramified in both $F$ and $F^r$, then it is totally ramified in $K/{\mathbb{Q}}$ and $K^r/{\mathbb{Q}}$.
- If an odd prime $p$ is ramified in $F$ (in $F^r$, respectively) as well as in $F_+$, splits in $F^r$ (in $F$, respectively). Moreover, at least one of the primes above $p$ is ramified in $K^r/F^r$ (in $K/F$, respectively).
The statements (i) and (ii) are clear from Table \[table\] on page. Alternatively, one can also prove the statements as follows:
- Let ${\mathfrak{p}}_{N}$ be a prime of $N$ above $p$ that is ramified in both $F/{\mathbb{Q}}$ and $F^r/{\mathbb{Q}}$. Then the maximal unramified subextension of $N/{\mathbb{Q}}$ is contained in $F_+$. Therefore, the inertia group of ${\mathfrak{p}}_{N}$ contains ${\text{\normalfont{Gal}}}(N/F_+)=\langle y \rangle$. By computing ramification indices in the diagram of subfields one by one, we see that the prime $p$ is totally ramified in $K$ and $K^r$.
- Let $p$ be an odd prime that is ramified in $F/{\mathbb{Q}}$ and $F_+/{\mathbb{Q}}$ and ${\mathfrak{p}}_{N}$ be a prime above $p$ in $N$. The inertia group of an odd prime cannot be a biquadratic group by Lemma \[inertia gp. not V4\], so ${\text{\normalfont{I}}}_{{\mathfrak{p | = F(\sqrt{\alpha_1\alpha_2})$, so $ { \mathfrak{p}}$ ramifies in $ E_i$ for an even number of $ i$ ’s. In particular, $ { \mathfrak{p}}$ is not wholly branch in $ E / F$.
\[ram1\ ] Let $ K$ be a non - biquadratic quartic CM field of character $ \Phi$ with the real quadratic subfield $ F$ and get $ K^r$ be its reflex plain with the quadratic subfield $ F^r$. Then the following assertions clasp.
- If a prime $ p$ is ramified in both $ F$ and $ F^r$, then it is wholly ramified in $ K/{\mathbb{Q}}$ and $ K^r/{\mathbb{Q}}$.
- If an odd prime $ p$ is ramified in $ F$ (in $ F^r$, respectively) equally well as in $ F_+$, splits in $ F^r$ (in $ F$, respectively). Moreover, at least one of the prime above $ p$ is ramified in $ K^r / F^r$ (in $ K / F$, respectively).
The statement (i) and (ii) are clear from Table \[table\ ] on page. Alternatively, one can also raise the statements as follows:
- Let $ { \mathfrak{p}}_{N}$ be a prime of $ N$ above $ p$ that is complexify in both $ F/{\mathbb{Q}}$ and $ F^r/{\mathbb{Q}}$. Then the maximal unramified subextension of $ N/{\mathbb{Q}}$ is contained in $ F_+$. Therefore, the inertia group of $ { \mathfrak{p}}_{N}$ contains $ { \text{\normalfont{Gal}}}(N / F_+)=\langle y \rangle$. By computing complication indices in the diagram of subfields one by one, we see that the prime $ p$ is totally ramified in $ K$ and $ K^r$.
- Let $ p$ be an odd prime that is ramified in $ F/{\mathbb{Q}}$ and $ F_+/{\mathbb{Q}}$ and $ { \mathfrak{p}}_{N}$ be a prime above $ p$ in $ N$. The inertia group of an odd prime cannot be a biquadratic group by Lemma \[inertia gp. not V4\ ], so $ { \text{\normalfont{I}}}_{{\mathfrak{p | = F(\dqrt{\alpha_1\alpha_2})$, so ${\mathfrxk{p}}$ ramifies in $E_i$ for an evsn numbef of $i$’s. In particular, ${\mathfrek{p}}$ us nou totally ramified kn $E/F$.
\[ram1\] Pet $K$ be a nib-biquadratmd quartle CM rleld mh type $\Phi$ with the real xuadratic subfheud $F$ and let $K^r$ be its reflex field wyth the qkadratic subfigld $F^g$. Ehen nht following assertions hold.
- If z prime $p$ is ramifiec in both $F$ and $F^r$, then it is hotally ramified ij $K/{\mathbb{Q}}$ qnd $H^e/{\mathbb{Q}}$.
- Iw an odd pgnme $p$ is rajified in $F$ (in $F^r$, respectively) xs wekl as in $F_+$, splihv in $F^r$ (in $H$, respvctively). Moreover, at neast ome of the primcs abmve $p$ is ramified in $K^r/F^c$ (in $K/F$, respectively).
Jhe statemancs (i) and (ii) are clear feom Tdble \[tabuw\] ov pzgx. Amternahivxly, one can also prove the statements as gojoows:
- Let ${\mafhfrak{[}}_{N}$ be a prime of $N$ above $p$ that is ramifitd in both $F/{\mathbb{Q}}$ and $F^r/{\marhbb{Q}}$. Then the maximap unramifyed subextension of $N/{\mathbb{Q}}$ is contained in $F_+$. Tharefoce, tht lkertkq hroup of ${\mathfrak{p}}_{N}$ contains ${\text{\normalfont{Gaj}}}(H/F_+)=\kakgle y \rangle$. By computing taligycation indicgs in tks siagram of subfielfs one fy onw, we see uhat yhe prime $p$ is totally ramidied in $K$ anb $K^e$.
- Let $p$ be an odb prime that is tamifird in $F/{\mathbb{Q}}$ and $F_+/{\matkbb{Q}}$ ahd ${\mathfrak{o}}_{N}$ be a pdkme above $p$ in $N$. Thv inartia group of an odd primq cannot ue a yiquadragic broup fy Lemma \[ijertia gp. not V4\], so ${\text{\nlrmalyont{I}}}_{{\kathfrak{p | = F(\sqrt{\alpha_1\alpha_2})$, so ${\mathfrak{p}}$ ramifies in $E_i$ even of $i$’s. particular, ${\mathfrak{p}}$ is \[ram1\] $K$ be a quartic CM field type $\Phi$ with the real quadratic $F$ and let $K^r$ be its reflex field with the quadratic subfield $F^r$. the following assertions hold. - If a prime $p$ is ramified in both and then is ramified in $K/{\mathbb{Q}}$ and $K^r/{\mathbb{Q}}$. - If an odd prime $p$ is ramified in $F$ (in respectively) as well as in $F_+$, splits in (in $F$, respectively). Moreover, least one of the primes $p$ ramified in (in respectively). statements (i) and are clear from Table \[table\] on page. Alternatively, one can also prove the statements as follows: - ${\mathfrak{p}}_{N}$ be of $N$ $p$ is in both $F/{\mathbb{Q}}$ Then the maximal unramified subextension of in $F_+$. Therefore, the inertia group of ${\mathfrak{p}}_{N}$ ${\text{\normalfont{Gal}}}(N/F_+)=\langle y By computing ramification indices in the of subfields one by one, we see that prime $p$ is totally ramified in $K$ and $K^r$. - Let $p$ be an odd is ramified in $F/{\mathbb{Q}}$ $F_+/{\mathbb{Q}}$ and ${\mathfrak{p}}_{N}$ a above in The inertia of an odd prime cannot be a biquadratic group by Lemma gp. not V4\], so ${\text{\normalfont{I}}}_{{\mathfrak{p | = F(\sqrt{\alpha_1\alpha_2})$, so ${\mathfrak{P}}$ ramifies iN $E_i$ foR an EveN nUmbeR of $i$’S. In particular, ${\mAThfrAk{p}}$ is not totally ramifieD in $E/F$.
\[RaM1\] let $K$ BE a Non-biQuadratIC qUARtiC Cm fIelD oF TyPe $\Phi$ WitH the reaL quadratic SubFiEld $F$ and let $K^r$ BE iTs reflex fiEld With the quadrAtiC subfiElD $F^r$. tHen thE foLlowiNg asseRTions hOld.
- If a priMe $P$ Is ramiFIed in boTH $f$ aNd $F^r$, Then it is totally raMIfIEd in $K/{\mathbb{Q}}$ anD $K^r/{\matHbB{q}}$.
- IF AN odD prIme $p$ is ramiFiEd in $F$ (IN $F^r$, respECtIVELy) aS Well as in $F_+$, spliTs in $F^r$ (in $F$, reSPecTively). moReoVEr, at leAst onE oF The Primes above $P$ is rAmified in $k^r/F^r$ (in $k/f$, respecTIvely).
ThE stateMenTs (i) And (iI) ArE cLeaR fROm TABlE \[taBLe\] oN page. AltErNaTivelY, one CAN ALso pRovE the StateMents as followS:
- LeT ${\matHFraK{p}}_{N}$ be A primE of $N$ AbOve $p$ tHat is rAmifiEd In both $F/{\mathbb{Q}}$ aNd $F^r/{\Mathbb{Q}}$. ThEn tHe MaxImAl unrAMified SubExtEnsion oF $N/{\mathbB{q}}$ is CoNTAInEd in $F_+$. Therefore, the iNeRTIa Group of ${\mAthfraK{P}}_{N}$ CoNTains ${\texT{\nOrmAlfoNT{gal}}}(N/F_+)=\LangLE y \Rangle$. By ComputINg RaMificatIoN indicEs In tHe dIagraM Of suBfieldS one by onE, we seE That the prime $p$ iS Totally ramifiED iN $k$ AnD $k^r$.
- LeT $p$ bE an odd prime That IS ramIfieD In $f/{\maTHbb{Q}}$ aNd $F_+/{\maThBB{Q}}$ ANd ${\mathfrak{p}}_{N}$ be a primE aBove $p$ iN $N$. The Inertia group oF an odd primE CANnot be a bIquaDRaTIc group by Lemma \[InertIa gp. not V4\], so ${\TExt{\normaLfont{i}}}_{{\mathfraK{p | = F(\sqrt{\alpha_1\alpha_ 2})$, so $ {\mat hfr ak{ p} }$ r amif ies in $E_i$ f o r an even number of $i$’s. In p ar t icul a r, ${\m athfrak { p} } $ is n ot to ta l ly rami fie d in $E /F$.
\[ra m1\ ]Let $K$ be a no n-biquadra tic quartic CMfie ld ofty pe$ \Phi$ wi th th e real quadra tic subfi el d $F$ a n d let $ K ^ r$ beits reflex fieldw it h the quadratic subfi el d $ F ^ r$. Th en the fol lo winga ssertio n sh o l d.- If a prim e $p$ is ra m ifi ed inbo th$ F$ and $F^r $, the n it is tot ally ramified in $K / {\mathb b {Q}}$ a nd $K^ r/{ \ma thbb { Q} }$ .
- If an od d pr ime $p$is r amifi ed i n $ F $ (i n $ F^r$ , res pectively) as we ll a s in $F_+ $, sp lits i n $F^ r$ (in $F$, r espectively). M oreo ver, at l eas tone o f the primes ab ove $p$ is ramifi e d i n$ K ^ r/ F^r$ (in $K/F$, re sp e c ti vely).
The st a te me n ts (i) a nd (i i) a r e clea r fr o mTable \[ table\ ] o npage. A lt ernati ve ly, on e can also prove the sta temen t s as follows:- Let ${\ma t hf r a k{ p }}_{ N}$ be a prime of$ N$ a bove $p $ t h at is rami fi e di n both $F/{\mathbb{ Q} }$ and $F^r /{\mathbb{Q}} $. Then th e m aximal u nram i fi e d subextension of $ N/{\mathbb { Q}}$ isconta ined in$F_+$. Th e r efore, t heine rti a g r o up of ${\mathfr a k {p}} _{ N}$ con tai ns ${\t ext {\n orm alf on t{Gal}}}( N/F_+)=\ la ng le y \r angle $ . By com pu tin gram ifica t ion in dices inth ed iag ram ofs ub f i elds o ne byone ,we se e th a t t he prim e $p$ istot a llyra mi fied in $K$ and $K^r $.
- Let$p $ b e an o d d prime t hat is ramified in $F/{ \ mathbb{ Q}} $ and $F_ +/{\mathb b{Q }}$ an d $ { \mathf rak{p} }_{N} $bea prime a bo ve$p $ in $N$.T h e i nerti agrou p of an odd prime cannotb e a biquadraticgro up b y Le mma \[ i ner ti a gp . not V4\], so ${ \text{\nor ma l fo nt{I}}}_{{ \ mat hf rak{p | =_F(\sqrt{\alpha_1\alpha_2})$, so_${\mathfrak{p}}$ ramifies in $E_i$_for an_even_number of_$i$’s._In particular, ${\mathfrak{p}}$_is not totally_ramified in $E/F$.
\[ram1\] Let_$K$ be a_non-biquadratic_quartic CM field of type $\Phi$ with the real quadratic subfield $F$ and let_$K^r$_be its_reflex_field_with the quadratic subfield $F^r$._Then the following assertions hold.
-_ _If a prime $p$ is ramified in both_$F$_and $F^r$, then_it is totally ramified in $K/{\mathbb{Q}}$ and $K^r/{\mathbb{Q}}$.
- _ If an odd prime $p$_is ramified in_$F$_(in_$F^r$, respectively) as well_as in $F_+$, splits in $F^r$_(in $F$, respectively). Moreover, at least_one of the primes above $p$ is_ramified in $K^r/F^r$ (in $K/F$, respectively).
The_statements (i) and (ii) are_clear from_Table \[table\] on page. Alternatively,_one can also_prove the_statements as follows:
-_ Let ${\mathfrak{p}}_{N}$ be a_prime of $N$_above $p$ that is ramified in_both_$F/{\mathbb{Q}}$ and $F^r/{\mathbb{Q}}$._Then_the_maximal unramified_subextension of $N/{\mathbb{Q}}$_is_contained in_$F_+$._Therefore, the inertia group of ${\mathfrak{p}}_{N}$_contains_${\text{\normalfont{Gal}}}(N/F_+)=\langle y \rangle$. By computing ramification indices_in the diagram of_subfields_one by one, we_see that the prime $p$_is totally ramified in $K$ and_$K^r$.
- _ Let_$p$ be an odd prime that is ramified in $F/{\mathbb{Q}}$ and_$F_+/{\mathbb{Q}}$ and ${\mathfrak{p}}_{N}$ be a prime_above $p$ in $N$._The inertia_group_of an odd_prime_cannot be_a biquadratic group by Lemma \[inertia gp._not V4\],_so ${\text{\normalfont{I}}}_{{\mathfrak{p |
based dynamics with integrate-and-fire dynamics where the firing of the neuron is represented by the resetting of the membrane potential whenever it crosses a threshold. Our justification for using this approximation hinges on three considerations: 1) both the conductance-based and integrate-and-fire models are Type I (in the sense of positive PRC), 2) the action potentials (spike widths) are narrow compared to their typical spiking period so the frequency is dominated by the membrane recovery time, and 3) the time scale for spike generation is very fast compared to the recovery time so an effective threshold for spiking can be defined. The first point was verified by observing that the measured PRC of the conductance-based models near the bifurcation to firing is positive.
We model the approach to threshold with a simple passive decay to obtain dynamics governed by a single equation of the form: $$C\frac{dV}{d\tilde{t}}\simeq \tilde{I} - g_m(V-V_r) - \tilde{g}
S(\tilde{t})(V-V_s),
\label{if1}$$ where $g_m$ is an effective membrane recovery conductance, $V_r$ is an effective membrane reversal potential and $V_s$ is a synaptic reversal potential. $V(t)$ is reset to $V_0$ whenever it reaches the threshold potential $V_T$. The passive decay to threshold is a very good approximation for some neuronal models such as the reduced Traub and Miles model given in the Appendix [@ermenkopell98]. On the other hand, we will show that even when the passive decay is not a good approximation to the slow dynamics of the neuron model, it can still adequately describe the frequency behavior, especially in the phasic regime where the synaptic current dominates.
The synaptic current $S(\tilde{t})$ is generated from the spikes of pre-synaptic neurons. This must be emulated in the reduced model (\[if1\]). Here we consider $S(\tilde{t})$ to be an arbitrary time dependent function. In Sec. \[sec:syn\], we analyze some biophysical synaptic models in detail and explicitly derive the time course of $S(\tilde{t})$ in response to a pre-synaptic spike.
To simplify the analysis, we rescale Eq. (\[if1\]) so that only dimensionless parameters remain. The voltage can be rescaled via $v = | based dynamics with integrate - and - fire moral force where the ignition of the neuron is represented by the resetting of the membrane electric potential whenever it crosses a doorsill. Our justification for using this estimate hinge on three considerations: 1) both the conductance - establish and integrate - and - fire model are Type I (in the sense of positive PRC), 2) the action potentials (spike widths) are narrow compare to their typical spiking period so the frequency is overshadow by the membrane recovery time, and 3) the time scale for spike coevals is very fast compared to the recovery time so an effective threshold for transfix can be defined. The first point was verified by observing that the careful PRC of the conductance - based models near the bifurcation to firing is positive.
We model the approach to threshold with a simple passive decay to obtain dynamics governed by a individual equation of the form: $ $ C\frac{dV}{d\tilde{t}}\simeq \tilde{I } - g_m(V - V_r) - \tilde{g }
S(\tilde{t})(V - V_s),
\label{if1}$$ where $ g_m$ is an effective membrane convalescence conductance, $ V_r$ is an effective membrane transposition electric potential and $ V_s$ is a synaptic reversal potential. $ V(t)$ is reset to $ V_0 $ whenever it reaches the threshold potential $ V_T$. The passive decay to threshold is a very good approximation for some neural models such as the reduced Traub and Miles model given in the Appendix [ @ermenkopell98 ]. On the other hand, we will show that even when the passive decay is not a good approximation to the slow dynamics of the neuron exemplar, it can still adequately report the frequency behavior, especially in the phasic regime where the synaptic current dominates.
The synaptic current $ S(\tilde{t})$ is generated from the spikes of pre - synaptic nerve cell. This must be emulated in the reduced model (\[if1\ ]). Here we consider $ S(\tilde{t})$ to be an arbitrary time dependent function. In Sec. \[sec: syn\ ], we analyze some biophysical synaptic models in detail and explicitly deduce the time course of $ S(\tilde{t})$ in response to a pre - synaptic spike.
To simplify the analysis, we rescale Eq. (\[if1\ ]) so that lone dimensionless parameters stay. The voltage can be rescale via $ v = | basfd dynamics with integraue-and-fire dynamics where the fjring of the neuron is represented bb thw restnting of the membrane potentiap whenevwr iu crosses a thresikld. Our justirlcatimi for using thix approximdtion hinges ot ghxee considerations: 1) both the conductwnce-basrd and integrate-wnd-fprq mosvlw are Type I (in the sense of lositivt PRC), 2) the action lotentials (spike widths) arf nagrow compared to tjeir typicao spyjing period ro the frequency is dojinated by the membrane recoverh timz, and 3) the jnne dwale for spmke gegeration is yvry fasd compated to the recpvecy tume so an effective tireshold for spiking can be dafnned. The first point qaw verhfieg by ibsdrvjnj tgat thf mxasured PRC of the coneuctance-based modelx gvsr the bifurdation tj firing is positive.
We model the approabh tk threshold with a simpoe passive decay to ohtain dynwmics governed by a single equation of the form: $$C\xrac{dT}{d\gilbc{b}}\simdw \hilde{I} - g_m(V-V_r) - \tilde{g}
S(\tilde{t})(V-V_s),
\label{if1}$$ where $f_m$ ix an effective membrane rrclvrty conductance, $V_r$ is an effective membranf reverfal pitential wnd $F_s$ is a synaptic reversal pitential. $V(t)$ ps rwset to $V_0$ whenever it reaches the threxhold potential $V_T$. The 'assivs decay to hhreshold ks a very good aoprpxhmation for some neuronal iodels surh as the reaucec Trauf and Miled modcn given in the Appfndix [@grmenkmpell98]. On tje other hand, we will show that even when the pdsspve decay is npt a good appwoximation to jhe slow bynamizs of the heuron kodel, it cag still adequdjely describe the frezuenxy bwhavior, dspecially in yhe phasib xegime whwre the synaptic cmrrenj sominates.
The syucpuix current $S(\tilce{t})$ is gvnecated xrom the spines uf ote-synxptic ntbroks. Ghis must be emulated in the reduced model (\[if1\]). Neve we conwider $S(\tylde{t})$ to be am arbitrary time dtpendeit funrtion. On Fec. \[sec:syn\], we analyze some biopgysical sjnaitic models ig debail and explieitly derive the time course of $S(\tilde{t})$ mn response to a pre-synqptic spike.
To simplnfn the analysms, we wescale Ex. (\[if1\]) so that only dimwnsionless parameuers remain. The voltage can ba resfaled via $v = | based dynamics with integrate-and-fire dynamics where the the is represented the resetting of crosses threshold. Our justification using this approximation on three considerations: 1) both the and integrate-and-fire models are Type I (in the sense of positive PRC), 2) action potentials (spike widths) are narrow compared to their typical spiking period so frequency dominated the recovery time, and 3) the time scale for spike generation is very fast compared to the time so an effective threshold for spiking can defined. The first point verified by observing that the PRC the conductance-based near bifurcation firing is positive. model the approach to threshold with a simple passive decay to obtain dynamics governed by a single of the \tilde{I} - - S(\tilde{t})(V-V_s), where $g_m$ is membrane recovery conductance, $V_r$ is an potential and $V_s$ is a synaptic reversal potential. is reset $V_0$ whenever it reaches the threshold $V_T$. The passive decay to threshold is a good approximation for some neuronal models such as the reduced Traub and Miles model given Appendix [@ermenkopell98]. On the hand, we will that when passive is not good approximation to the slow dynamics of the neuron model, it still adequately describe the frequency behavior, especially in the phasic the current dominates. The current $S(\tilde{t})$ is generated the of pre-synaptic neurons. This emulated the Here consider to be an arbitrary dependent function. In Sec. \[sec:syn\], analyze some biophysical synaptic derive the time course of $S(\tilde{t})$ in response a pre-synaptic spike. To simplify the analysis, rescale Eq. (\[if1\]) so that only dimensionless parameters remain. The voltage can rescaled via | based dynamics with integratE-and-fire dyNamicS whEre ThE firIng oF the neuron is rePReseNted by the resetting of thE membRaNE potENtIal whEnever iT CrOSSes A tHrEshOlD. ouR justIfiCation fOr using thiS apPrOximation hinGEs On three conSidErations: 1) both The ConducTaNce-BAsed aNd iNtegrAte-and-FIre modEls are TypE I (IN the seNSe of posITIvE PRC), 2) The action potentiaLS (sPIke widths) are naRrow coMpAReD TO thEir Typical spiKiNg perIOd so the FReQUENcy IS dominated by tHe membrane rECovEry timE, aNd 3) tHE time sCale fOr SPikE generation Is veRy fast comPared tO The recoVEry time So an efFecTivE thrEShOlD foR sPIkiNG cAn bE DefIned. The fIrSt Point Was vERIFIed bY obServIng thAt the measured pRC Of thE ConDuctaNce-baSed mOdEls neAr the bIfurcAtIon to firing is poSitiVe.
We model The ApProAcH to thREshold WitH a sImple paSsive deCAy tO oBTAIn Dynamics governed by A sINGlE equatioN of the FOrM: $$C\FRac{dV}{d\tiLdE{t}}\sImeq \TILde{I} - g_M(V-V_r) - \TIlDe{g}
S(\tildE{t})(V-V_s),
\lABeL{iF1}$$ where $g_M$ iS an effEcTivE meMbranE RecoVery coNductancE, $V_r$ is AN effective membRAne reversal poTEnTIAl ANd $V_s$ Is a Synaptic revErsaL PoteNtiaL. $v(t)$ Is rESet to $v_0$ whenEvER iT Reaches the threshold PoTentiaL $V_T$. ThE passive decay To thresholD IS A very gooD appROxIMation for some nEuronAl models suCH as the reDuced traub and miles modeL GIven in thE ApPenDix [@ErmENKoPell98]. On the otheR HAnd, wE wIll show ThaT even whEn tHe pAssIve DeCay is not a Good apprOxImAtIoN to The slOW dynamicS oF thE nEurOn modEL, it can Still AdeqUaTeLY deScribe tHE fREQuenCy BeHaviOr, eSpEcialLy in THe pHasic reGime where The SYnapTiC cUrrent dOminates.
The syNaPtic currenT $S(\TilDe{t})$ is gENErated frOm the spikes of pre-synaptiC Neurons. thiS must Be emUlated in tHe rEduced ModEL (\[if1\]). HerE we conSider $s(\tIldE{T})$ To be aN ARbItrArY time depenDENt fUnctiOn. in SeC. \[sec:syn\], We analyze some biophYSicAl synaptic modEls In deTAIl And EXpLIciTlY DerIVE the time course oF $S(\tilde{t})$ in ReSPoNse to a pre-sYNapTiC spike.
TO simpliFy the ANalysis, We rescale eq. (\[if1\]) so thaT oNly dIMEnsIonless parAmeters rEmain. The vOLtage CAn Be resCalEd via $v = | based dynamics with integr ate-and-fi re dy nam ics w here the firing of the neur on is represented by t he re se t ting of themembran e p o t ent ia lwhe ne v er it c ros ses a t hreshold.Our j ustification fo r using th isapproximatio n h ingeson th r ee co nsi derat ions:1 ) both the cond uc t ance-b a sed and i nt egra te-and-fire model s a r e Type I (in t he sen se of p osi tiv e PRC), 2) t he ac t ion pot e nt i a l s ( s pike widths)are narrowc omp ared t othe i r typi cal s pi k ing period sothefrequency is do m inatedb y the m embran e r eco very ti me , a nd 3)t he ti m e s cale for s pi ke ge nera t i o n isver y fa st co mpared to the re cove r y t ime s o aneffe ct ive t hresho ld fo rspiking can bedefi ned. Thefir st po in t was verifi edbyobservi ng that the m e a s ur ed PRC of the cond uc t a nc e-basedmodels ne ar the bifu rc ati on t o firin g is po sitive.
We mo d el t he appr oa ch toth res hol d wit h a s implepassivedecay to obtain dyna m ics governedb ya si n gleequ ation of th e fo r m: $ $C\f r ac {dV } {d\ti lde{t }} \ si m eq \tilde{I} - g_m( V- V_r) - \til de{g}
S(\tild e{t})(V-V_ s ) ,
\label {if1 } $$ where $g_m$ is an e ffective m e mbrane r ecove ry condu ctance, $ V _ r$ is an ef fec tiv e m e m br ane reversalp o tent ia l and $ V_s $ is asyn apt icrev er sal poten tial. $V (t )$ i sres et to $V_0$ wh en eve ritreach e s thethres hold p ot e nti al $V_T $ .T h e pa ss iv e de cay t o thr esho l d i s a ver y good ap pro x imat io nfor som e neuronal mo de ls such as t hereduce d Traub an d Miles model given int he Appe ndi x [@e rmen kopell98] . O n theoth e r hand , we w ill s ho w t h a t eve n wh enth e passived e cay is n ot a g ood app roximation to thes low dynamics ofthe neu r o nmod e l, itca n st i l l adequately de scribe the f r eq uency beha v ior ,especia lly inthe p h asic re gime wher e the syn ap ticc u rre nt dominat es.
The synaptic curre n t$S(\t ild e{t})$ i s g enera ted fr o m t he sp ikes o fpre-sy napti cneurons. This must be emulatedin the redu ced model (\ [if 1 \]) . Here we con sider $S(\ til de{ t})$tob e anarbi t ra ryt ime d epen d ent funct i on . I n Se c. \[sec:sy n \ ] , w e ana lyz e somebiop hysical synapticm odels in detai l an d exp lic i tlyde rive the timecou rs e of $S(\t il de{t})$ inresponse t o a pr e-syna ptic s pike.
T o s i mplify the an alysis, w e r es c ale Eq. ( \[ i f1\])so t ha t only dimen s ionl e s s parameters rem ain.T h e vol t age canbe rescal e d vi a $v = | based dynamics_with integrate-and-fire_dynamics where the firing_of the_neuron_is represented_by_the resetting of_the membrane potential_whenever it crosses a_threshold. Our justification_for_using this approximation hinges on three considerations: 1) both the conductance-based and integrate-and-fire models_are_Type I_(in_the_sense of positive PRC), 2)_the action potentials (spike widths)_are narrow_compared to their typical spiking period so the_frequency_is dominated by_the membrane recovery time, and 3) the time scale_for spike generation is very fast_compared to the_recovery_time_so an effective threshold_for spiking can be defined. The_first point was verified by observing_that the measured PRC of the conductance-based_models near the bifurcation to firing_is positive.
We model the approach_to threshold_with a simple passive decay_to obtain dynamics_governed by_a single equation_of the form: $$C\frac{dV}{d\tilde{t}}\simeq \tilde{I} -_g_m(V-V_r) - \tilde{g}
S(\tilde{t})(V-V_s),_
\label{if1}$$ where $g_m$ is an effective_membrane_recovery conductance, $V_r$_is_an_effective membrane_reversal potential and_$V_s$_is a_synaptic_reversal potential. $V(t)$ is reset to_$V_0$_whenever it reaches the threshold potential $V_T$._The passive decay to_threshold_is a very good_approximation for some neuronal models_such as the reduced Traub and_Miles model_given in_the Appendix [@ermenkopell98]. On the other hand, we will show that even_when the passive decay is not_a good approximation to_the slow_dynamics_of the neuron_model,_it can_still adequately describe the frequency behavior, especially_in the_phasic regime where the synaptic current_dominates.
The synaptic current $S(\tilde{t})$_is_generated from the spikes of pre-synaptic_neurons. This must be emulated in_the reduced model (\[if1\]). Here_we_consider_$S(\tilde{t})$ to be an arbitrary_time dependent function. In Sec. \[sec:syn\], we_analyze some biophysical_synaptic models in detail and explicitly derive_the_time course of $S(\tilde{t})$ in response_to_a pre-synaptic spike.
To simplify the analysis,_we_rescale_Eq. (\[if1\]) so that only dimensionless_parameters remain. The voltage can be_rescaled via $v = |
) =0\ne \b $ that partition the $(q+1)^2$ singular points in $\Omega_0^\perp$, sending $$\label{generators}
\begin{array}{llll}
\hspace{-5.5pt}u_s\col (0,\b,{\gamma },0)\mapsto
(0,\b,{\gamma }+ \b s ,0)
\vspace{2pt}
\\
j\col (0,\b,{\gamma },0)\mapsto (0,{\gamma }, \b,0) . \hspace{185pt}
\end{array}\vspace{-1pt}$$ An [*ordinary*]{} point is a singular point in $\Omega_0^\perp$ of the form $\< (0,\b, {\gamma },0)\>$ such that either $\b=0$ and $T({\gamma }^{1+q})\ne 0$, or $T(\b^{1+q})\ne 0$ (recall that $ T(\b)=T({\gamma })= T(\b{\gamma }) = 0 $). Since any $\b\in F^*$ has characteristic polynomial $x^3+T(\b)x^2+T(\b^{1+q})+N(\b)$, the ordinary requirement can fail for some $\b,{\gamma }$ if and only if $q\equiv 1$ (mod 3). Moreover, if $\b\in F-K$ then ${\b ^q \in \b K}\iff~{ \b^{q-1}\in K } \iff \b^{(q-1,q^2+q+1)}\in K \iff \b^{3}\in K
\iff T(\b^{1+q})= 0$.
For $\pi$ in, since $T((a\pi+ \pi^q)(a\pi + \pi^{q }) ^q )=(a^2+a+1)T(\pi^{1+q})$ the points of the line $\<(0,a\pi+ \pi^q,0,0), (0,0, a\pi+ \pi^q,0) \>,$ $a\in K,$ are ordinary if and only if $a^2+a+1\ne 0$, so that all points are ordinary if $q\equiv 2$ (mod 3), but there are two lines | ) = 0\ne \b $ that partition the $ (q+1)^2 $ singular points in $ \Omega_0^\perp$, sending $ $ \label{generators }
\begin{array}{llll }
\hspace{-5.5pt}u_s\col (0,\b,{\gamma }, 0)\mapsto
(0,\b,{\gamma } + \b s, 0)
\vspace{2pt }
\\
j\col (0,\b,{\gamma }, 0)\mapsto (0,{\gamma }, \b,0) . \hspace{185pt }
\end{array}\vspace{-1pt}$$ An [ * ordinary * ] { } item is a remarkable point in $ \Omega_0^\perp$ of the form $ \ < (0,\b, { \gamma }, 0)\>$ such that either $ \b=0 $ and $ T({\gamma } ^{1+q})\ne 0 $, or $ T(\b^{1+q})\ne 0 $ (recall that $ T(\b)=T({\gamma }) = T(\b{\gamma }) = 0 $). Since any $ \b\in F^*$ have characteristic polynomial $ x^3+T(\b)x^2+T(\b^{1+q})+N(\b)$, the ordinary requirement can fail for some $ \b,{\gamma } $ if and entirely if $ q\equiv 1 $ (mod 3). furthermore, if $ \b\in F - K$ then $ { \b ^q \in \b K}\iff~ { \b^{q-1}\in K } \iff \b^{(q-1,q^2+q+1)}\in K \iff \b^{3}\in K
\iff T(\b^{1+q})= 0$.
For $ \pi$ in, since $ T((a\pi+ \pi^q)(a\pi + \pi^{q }) ^q) =( a^2+a+1)T(\pi^{1+q})$ the points of the line $ \<(0,a\pi+ \pi^q,0,0), (0,0, a\pi+ \pi^q,0) \>,$ $ a\in K,$ are ordinary if and only if $ a^2+a+1\ne 0 $, so that all compass point are ordinary if $ q\equiv 2 $ (mod 3), but there be two lines | ) =0\ne \b $ that partition the $(q+1)^2$ singular points in $\Omxga_0^\perp$, sending $$\label{generators}
\begin{array}{llpl}
\yspact{-5.5it}u_s\col (0,\b,{\gamma },0)\mapsto
(0,\b,{\gamma }+ \b s ,0)
\vspqce{2pu}
\\
j\col (0,\b,{\gamma },0)\mapsvk (0,{\gamma }, \b,0) . \gdpacz{185pv}
\end{array}\vspace{-1kt}$$ An [*ordinasy*]{} point is a vivgblar point in $\Omega_0^\perp$ of the form $\< (0,\b, {\gamms },0)\>$ such that eitrer $\n=0$ and $N({\gcmma }^{1+q})\ne 0$, or $T(\b^{1+q})\ne 0$ (recall that $ T(\b)=T({\gakma })= T(\b{\gamma }) = 0 $). Since any $\b\in F^*$ has chagactfristic polynomial $x^3+T(\b)x^2+T(\b^{1+q})+N(\b)$, jge jedinary requkrement can fail for skme $\b,{\gamma }$ if and only if $q\equkv 1$ (mpd 3). Moreovgx, if $\t\in F-K$ then ${\b ^q \pn \b K}\iff~{ \b^{q-1}\ik K } \ifx \b^{(q-1,q^2+q+1)}\on K \iff \b^{3}\in L
\ihf T(\b^{1+q})= 0$.
For $\pi$ in, sincx $T((a\pi+ \pi^q)(a\pi + \pi^{q }) ^z )=(a^2+a+1)T(\pi^{1+q})$ dhz points of the line $\<(0,q\pu+ \pi^q,0,0), (0,0, d\pi+ \pi^q,0) \>,$ $a\jn K,$ are ogdiiary if and only if $a^2+a+1\be 0$, so that all poimtf are ordinary if $q\ezuyv 2$ (mod 3), but there are two lines | ) =0\ne \b $ that partition the points $\Omega_0^\perp$, sending \begin{array}{llll} \hspace{-5.5pt}u_s\col (0,\b,{\gamma ,0) \\ j\col (0,\b,{\gamma (0,{\gamma }, \b,0) \hspace{185pt} \end{array}\vspace{-1pt}$$ An [*ordinary*]{} point is singular point in $\Omega_0^\perp$ of the form $\< (0,\b, {\gamma },0)\>$ such that $\b=0$ and $T({\gamma }^{1+q})\ne 0$, or $T(\b^{1+q})\ne 0$ (recall that $ T(\b)=T({\gamma })= }) 0 Since $\b\in F^*$ has characteristic polynomial $x^3+T(\b)x^2+T(\b^{1+q})+N(\b)$, the ordinary requirement can fail for some $\b,{\gamma }$ if only if $q\equiv 1$ (mod 3). Moreover, if F-K$ then ${\b ^q \b K}\iff~{ \b^{q-1}\in K } \b^{(q-1,q^2+q+1)}\in \iff \b^{3}\in \iff 0$. $\pi$ in, since \pi^q)(a\pi + \pi^{q }) ^q )=(a^2+a+1)T(\pi^{1+q})$ the points of the line $\<(0,a\pi+ \pi^q,0,0), (0,0, a\pi+ \pi^q,0) \>,$ K,$ are and only $a^2+a+1\ne so all points are $q\equiv 2$ (mod 3), but there | ) =0\ne \b $ that partition the $(q+1)^2$ singuLar points iN $\OmegA_0^\peRp$, sEnDing $$\LabeL{generators}
\begIN{arrAy}{llll}
\hspace{-5.5pt}u_s\col (0,\b,{\gaMma },0)\maPsTO
(0,\b,{\gaMMa }+ \B s ,0)
\vspAce{2pt}
\\
j\cOL (0,\b,{\GAMma },0)\MaPsTo (0,{\gAmMA }, \b,0) . \HspacE{185pt}
\End{arraY}\vspace{-1pt}$$ AN [*orDiNary*]{} point is a SInGular point In $\OMega_0^\perp$ of thE foRm $\< (0,\b, {\gamMa },0)\>$ SucH That eIthEr $\b=0$ anD $T({\gammA }^{1+Q})\ne 0$, or $T(\B^{1+q})\ne 0$ (recalL tHAt $ T(\b)=T({\gAMma })= T(\b{\gaMMA }) = 0 $). SInce Any $\b\in F^*$ has charactERiSTic polynomial $x^3+t(\b)x^2+T(\b^{1+q})+n(\b)$, THe ORDinAry RequiremenT cAn faiL For some $\B,{\GaMMA }$ If aND only if $q\equiv 1$ (Mod 3). Moreover, IF $\b\iN F-K$ theN ${\b ^Q \in \B k}\iff~{ \b^{q-1}\In K } \ifF \b^{(Q-1,Q^2+q+1)}\iN K \iff \b^{3}\in K
\ifF T(\b^{1+q})= 0$.
for $\pi$ in, siNce $T((a\pI+ \Pi^q)(a\pi + \pI^{Q }) ^q )=(a^2+a+1)T(\pi^{1+Q})$ the poIntS of The lINe $\<(0,A\pI+ \pi^Q,0,0), (0,0, a\PI+ \pi^Q,0) \>,$ $A\iN K,$ aRE orDinary if AnD oNly if $A^2+a+1\ne 0$, SO THAt alL poInts Are orDinary if $q\equiV 2$ (moD 3), but THerE are tWo linEs | ) =0\ne \b $ that partitio n the $(q+ 1)^2$ si ngu la r po ints in $\Omega_0^ \ perp $, sending $$\label{ge nerat or s }
\b e gi n{arr ay}{lll l }\ h spa ce {- 5.5 pt } u_ s\col (0 ,\b,{\g amma },0)\ map st o
(0,\b,{\g a mm a }+ \b s ,0 )
\vspace{2p t}\\
j\c ol (0 , \b,{\ gam ma }, 0)\map s to (0, {\gamma } ,\ b,0) . \hspac e { 18 5pt}
\end{array}\vspa c e{ - 1pt}$$ An [*or dinary *] { }p o int is a singula rpoint in $\Om e ga _ 0 ^ \pe r p$ of the for m $\< (0,\b , {\ gamma}, 0)\ > $ such that e i the r $\b=0$ an d $T ({\gamma}^{1+q } )\ne 0$ , or $T( \b^{1+ q}) \ne 0$( re ca llth a t $ T( \b) = T({ \gamma } )= T (\b{\ gamm a } ) = 0 $) . Si nce a ny $\b\in F^* $ h as c h ara cteri sticpoly no mial$x^3+T (\b)x ^2 +T(\b^{1+q})+N( \b)$ , the ord ina ry re qu ireme n t canfai l f or some $\b,{\ g amm a} $ if and only if $q\eq ui v 1$ (mod 3) . More o ve r, if $\b\i nF-K $ th e n ${\b ^q\ in \b K}\i ff~{ \ b ^{ q- 1}\in K } \iff \ b^{ (q- 1,q^2 + q+1) }\in K \iff \ b^{3} \ in K
\iff T(\b^{1+q})=0 $. Fo r $\p i$in, since $ T((a \ pi+\pi^ q )( a\p i + \p i^{q}) ^q )=(a^2+a+1)T(\pi^{1 +q })$ th e poi nts of the li ne $\<(0,a \ p i + \pi^q, 0,0) , (0,0, a\pi+ \ pi^q, 0) \>,$ $a \ in K,$ a re or dinary i f and onl y if $a^2+ a+1 \ne 0$ , s o th at all points a re o rd inary i f $ q\equiv 2$ (m od3), b ut thereare twoli ne s | ) =0\ne_\b $_that partition the $(q+1)^2$_singular points_in_$\Omega_0^\perp$, sending_$$\label{generators}
\begin{array}{llll}
\hspace{-5.5pt}u_s\col_(0,\b,{\gamma },0)\mapsto
(0,\b,{\gamma_}+ \b_s ,0)
\vspace{2pt}
\\
j\col (0,\b,{\gamma },0)\mapsto_(0,{\gamma }, \b,0)__. \hspace{185pt}
\end{array}\vspace{-1pt}$$ An [*ordinary*]{} point is a singular point in $\Omega_0^\perp$ of the form_$\<_(0,\b, {\gamma_},0)\>$_such_that either $\b=0$ and $T({\gamma_}^{1+q})\ne 0$, or $T(\b^{1+q})\ne 0$_(recall that_$ T(\b)=T({\gamma })= T(\b{\gamma }) = 0 $). Since_any_$\b\in F^*$ has_characteristic polynomial $x^3+T(\b)x^2+T(\b^{1+q})+N(\b)$, the ordinary requirement can fail for_some $\b,{\gamma }$ if and only_if $q\equiv 1$_(mod_3)._Moreover, if $\b\in F-K$_then ${\b ^q \in \b K}\iff~{_\b^{q-1}\in K } \iff \b^{(q-1,q^2+q+1)}\in_K \iff \b^{3}\in K
\iff _ T(\b^{1+q})= 0$.
For $\pi$_in, since $T((a\pi+ \pi^q)(a\pi +_\pi^{q })_^q )=(a^2+a+1)T(\pi^{1+q})$ the points of_the line $\<(0,a\pi+_\pi^q,0,0), _ (0,0, a\pi+_ \pi^q,0) \>,$ $a\in K,$ are_ordinary if and_only if $a^2+a+1\ne 0$, so that_all_points are ordinary_if_$q\equiv_2$ (mod_3), but there_are_two lines |
size scaling of the discontinuity and its effect on the scaling exponents, which have also previously been studied using a complementary approach with the addition of small, non-zero bending rigidity [@sharma_strain-controlled_2016]. Using these modified exponents, we test scaling relations recently predicted for fiber networks [@shivers_scaling_2019].
![ \[fig:1\] Rigidity phase diagram of central force networks. [Upon increasing the average]{} connectivity $z$ at $\gamma=0$, [a network]{} passes through three distinct regimes: (i) a disconnected structure for connectivity less than the percolation connectivity $z<z_p$ (ii) a percolated but floppy network for $z_p<z<z_c \simeq 2d $ and (iii) a rigid network for connectivity greater than $z_c$. Applying a sufficiently large finite strain to an otherwise floppy network with $z_p<z<z_c$ rigidifies the system. For a given $z$ in this range, a critical transition is observed with increasing strain, as indicated by the dashed arrow. The second-order line of transitions is characterized by a critical strain $\gamma_c(z)$ that varies linearly with $z$ near $z_c$ [@wyart_elasticity_2008] (see also Fig. \[fig:A3\] in the Appendix).](fig1){width="6cm" height="6cm"}
Simulation method
=================
To investigate the stiffness discontinuity in fiber networks, we use various network models including (i) triangular, (ii) *phantomized* triangular [@broedersz_criticality_2011; @licup_stress_2015], (iii) 2D and (iv) 3D jammed-packing-derived [@wyart_elasticity_2008; @tighe_force_2010; @baumgarten_normal_2018; @merkel_minimal-length_2019; @shivers_normal_2019], (v) Mikado [@wilhelm_elasticity_2003; @head_deformation_2003], and (vi) 2D Voronoi network [@heussinger_stiff_2006; @arzash_stress-stabilized_2019]. Triangular networks are built by depositing individual fibers of length $W$ on a periodic triangular lattice. The lattice spacing is $\ell_0=1$. A full triangular network has an average connectivity | size scaling of the discontinuity and its effect on the scaling exponents, which have also previously been study use a complementary approach with the addition of little, non - zero bending rigidity [ @sharma_strain - controlled_2016 ]. use these modify exponents, we test scaling relations recently predicted for character networks [ @shivers_scaling_2019 ].
! [ \[fig:1\ ] Rigidity phase diagram of cardinal force networks. [ Upon increasing the average ] { } connectivity $ z$ at $ \gamma=0 $, [ a network ] { } passes through three distinct government: (i) a disconnected structure for connectivity less than the percolation connectivity $ omega < z_p$ (ii) a percolated but floppy network for $ z_p < omega < z_c \simeq 2d $ and (iii) a rigid net for connectivity greater than $ z_c$. Applying a sufficiently large finite strain to an otherwise floppy network with $ z_p < omega < z_c$ rigidifies the system. For a given $ z$ in this range, a critical transition is observed with increase strain, as indicated by the dashed arrow. The second - order line of transitions is characterized by a critical strain $ \gamma_c(z)$ that varies linearly with $ z$ near $ z_c$ [ @wyart_elasticity_2008 ] (see besides Fig. \[fig: A3\ ] in the Appendix).](fig1){width="6 cm " height="6 cm " }
Simulation method acting
= = = = = = = = = = = = = = = = =
To investigate the stiffness discontinuity in fiber networks, we use diverse network models including (iodine) triangular, (ii) * phantomized * triangular [ @broedersz_criticality_2011; @licup_stress_2015 ], (iii) 2D and (iv) 3D jammed - packing - derived [ @wyart_elasticity_2008; @tighe_force_2010; @baumgarten_normal_2018; @merkel_minimal - length_2019; @shivers_normal_2019 ], (v) Mikado [ @wilhelm_elasticity_2003; @head_deformation_2003 ], and (vi) 2D Voronoi network [ @heussinger_stiff_2006; @arzash_stress - stabilized_2019 ]. Triangular network are built by depositing individual fibers of length $ W$ on a periodic triangular wicket. The lattice spacing is $ \ell_0=1$. A full triangular network has an average connectivity | sizf scaling of the discontlnuity and its eydect oi the sdaling ebponents, which have also pretiouwly btvn studied using a cooplementagy approaxh wmth the addition of small, non-zsvo beudmng rigidity [@shsrma_strain-wontrolled_2016]. Usitg tkese modified exponents, we test scalyng relstlons recently kredibtqd fkg niber networks [@shivers_scaling_2019].
![ \[fif:1\] Rigidpty phase diagram of central force networks. [Upoj increasing the agerage]{} conngdtiduty $z$ at $\gamoa=0$, [a netwogn]{} passes tgrough three distinct regimes: (i) a dixconnected srrufjure for coniectivpty less than the perwolatiom connectivity $z<z_'$ (ii) a percolated but flo'py network for $z_p<z<z_s \simeq 2d $ cnd (iii) a rigid netwoek for wonnactixuty grtatxr fhan $z_f$. A'plying a shfficiently large finite straim ei an otherwiss flop[y network with $z_p<z<z_c$ rigidifies the systtm. Fod a given $z$ in this rante, a critical transitlon is obferved with increasing strain, as indicated by the dashxd arxiw. Thd sfcond-order line of transitions is characterizqs ny a critical stvain $\gamma_c(z)$ that fagirf linearly wijh $z$ necd $a_c$ [@wyart_elasticity_2008] (see alfo Fit. \[fig:A3\] in uhe Alpendix).](fig1){width="6cm" height="6cm"}
Sumulation menhod
=================
Ro investigate the stiffness bisconjinuitu in fiber networks, we bse vadious netwogk models kncluding (i) triavgukas, (ii) *phantomized* triangulaw [@broederwz_crnticalith_2011; @livup_strqss_2015], (iii) 2D wnd (iy) 3D jammed-packing-degived [@wfart_elastifity_2008; @tighe_force_2010; @baumgarten_normal_2018; @merkel_minimak-langnh_2019; @shiverf_normsl_2019], (v) Mikado [@wylhelm_elasticijy_2003; @head_deyormatkon_2003], and (vi) 2D Voroioi network [@reussinger_stixv_2006; @arzash_strevs-stabilyzed_2019]. Triqngular vetworks are biilt by dvpjsitung individual fibcrs ow length $W$ on a 'triidic triangular lagtise. Tie laedice spacing is $\dll_0=1$. S fulu triangulav ndtwotk has an average cotnecfivity | size scaling of the discontinuity and its the exponents, which also previously been with addition of small, bending rigidity [@sharma_strain-controlled_2016]. these modified exponents, we test scaling recently predicted for fiber networks [@shivers_scaling_2019]. ![ \[fig:1\] Rigidity phase diagram of central networks. [Upon increasing the average]{} connectivity $z$ at $\gamma=0$, [a network]{} passes through distinct (i) disconnected for connectivity less than the percolation connectivity $z<z_p$ (ii) a percolated but floppy network for $z_p<z<z_c 2d $ and (iii) a rigid network for greater than $z_c$. Applying sufficiently large finite strain to otherwise network with rigidifies system. a given $z$ this range, a critical transition is observed with increasing strain, as indicated by the dashed arrow. The line of characterized by critical $\gamma_c(z)$ varies linearly with $z_c$ [@wyart_elasticity_2008] (see also Fig. \[fig:A3\] height="6cm"} Simulation method ================= To investigate the stiffness in fiber we use various network models including triangular, (ii) *phantomized* triangular [@broedersz_criticality_2011; @licup_stress_2015], (iii) 2D (iv) 3D jammed-packing-derived [@wyart_elasticity_2008; @tighe_force_2010; @baumgarten_normal_2018; @merkel_minimal-length_2019; @shivers_normal_2019], (v) Mikado [@wilhelm_elasticity_2003; @head_deformation_2003], and (vi) 2D [@heussinger_stiff_2006; @arzash_stress-stabilized_2019]. Triangular networks built by depositing fibers length on periodic triangular The lattice spacing is $\ell_0=1$. A full triangular network has an connectivity | size scaling of the discontinUity and its EffecT on The ScAlinG expOnents, which havE Also Previously been studied uSing a CoMPlemENtAry apProach wITh THE adDiTiOn oF sMAlL, non-zEro Bending Rigidity [@shArmA_sTrain-controlLEd_2016]. using these ModIfied exponenTs, wE test sCaLinG RelatIonS receNtly prEDicted For fiber nEtWOrks [@shIVers_scaLINg_2019].
![ \[Fig:1\] RIgidity phase diagrAM oF Central force neTworks. [upON iNCReaSinG the averagE]{} cOnnecTIvity $z$ aT $\GaMMA=0$, [A neTWork]{} passes thrOugh three diSTinCt regiMeS: (i) a DIsconnEcted StRUctUre for conneCtivIty less thAn the pERcolatiON connecTivity $Z<z_p$ (Ii) a PercOLaTeD buT fLOppY NeTwoRK foR $z_p<z<z_c \siMeQ 2d $ And (iiI) a riGID NEtwoRk fOr coNnectIvity greater tHan $Z_c$. ApPLyiNg a suFficiEntlY lArge fInite sTrain To An otherwise flopPy neTwork with $Z_p<z<Z_c$ RigIdIfies THe systEm. FOr a Given $z$ iN this raNGe, a CrITICaL transition is obserVeD WItH increasIng strAIn, As INdicated By The DashED Arrow. the sECoNd-order lIne of tRAnSiTions is ChAracteRiZed By a CritiCAl stRain $\gaMma_c(z)$ thaT variES linearly with $z$ NEar $z_c$ [@wyart_elaSTiCITy_2008] (SEe alSo FIg. \[fig:A3\] in the appeNDix).](fIg1){wiDTh="6Cm" hEIght="6cM"}
SimuLaTIoN Method
=================
To investigate ThE stiffNess dIscontinuity iN fiber netwORKS, we use vaRiouS NeTWork models inclUding (I) triangulaR, (Ii) *phantoMized* TriangulAr [@broederSZ_CriticalIty_2011; @LicUp_sTreSS_2015], (IiI) 2D and (iv) 3D jammeD-PAckiNg-Derived [@WyaRt_elastIciTy_2008; @tIghE_foRcE_2010; @baumgartEn_normal_2018; @MeRkEl_MiNimAl-lenGTh_2019; @shiverS_nOrmAl_2019], (V) MiKado [@wILhelm_eLastiCity_2003; @HeAd_DEfoRmation_2003], ANd (VI) 2d VorOnOi NetwOrk [@HeUssinGer_sTIff_2006; @Arzash_sTress-stabIliZEd_2019]. TrIaNgUlar netWorks are built By Depositing InDivIdual fIBErs of lenGth $W$ on a periodic triangulAR latticE. ThE lattIce sPacing is $\eLl_0=1$. A Full trIanGUlar neTwork hAs an aVeRagE COnnecTIViTy | size scaling of the discon tinuity an d its ef fec ton t he s caling exponen t s, w hich have also previou sly b ee n stu d ie d usi ng a co m pl e m ent ar yapp ro a ch with th e addit ion of sma ll, n on-zero bend i ng rigidity[@s harma_strain -co ntroll ed _20 1 6]. U sin g the se mod i fied e xponents, w e tests calingr e la tion s recently predic t ed for fiber netw orks [ @s h iv e r s_s cal ing_2019].
![ \[ f ig:1\]R ig i d i typ hase diagramof centralf orc e netw or ks. [Uponincre as i ngthe average ]{}connectiv ity $z $ at $\g a mma=0$, [a ne two rk] {} p a ss es th ro u ght hr eed ist inct reg im es : (i) a d i s c o nnec ted str uctur e for connect ivi ty l e ssthanthe p erco la tionconnec tivit y$z<z_p$ (ii) aperc olated bu t f lo ppy n etwor k for $ z_p <z< z_c \si meq 2d$ an d( i i i) a rigid network f or c on nectivit y grea t er t h an $z_c$ .App lyin g a suf fici e nt ly large finit e s tr ain toan other wi seflo ppy n e twor k with $z_p<z< z_c$r igidifies thes ystem. For ag iv e n $ z $ in th is range, a cri t ical tra n si tio n is o bserv ed wi t h increasing strain ,as ind icate d by the dash ed arrow.T h e second- orde r l i ne of transiti ons i s characte r ized bya cri tical st rain $\ga m m a_c(z)$tha t v ari esl i ne arly with $z$ n ear$z _c$ [@w yar t_elast ici ty_ 200 8](s ee also F ig. \[fi g: A3 \] i n t he Ap p endix).] (f ig1 ){ wid th="6 c m" hei ght=" 6cm" }
S i mul ation m e th o d
=== == == ==== === == =
To inv e sti gate th e stiffne ssd isco nt in uity in fiber networ ks , we use v ar iou s netw o r k models including (i) triangul a r, (ii) *p hanto mize d* triang ula r [@br oed e rsz_cr itical ity_2 01 1;@ l icup_ s t re ss_ 20 15], (iii) 2 D a nd (i v) 3Djammed- packing-derived [@ w yar t_elasticity_ 200 8; @ t i gh e_f o rc e _20 10 ; @b a u mgarten_normal_ 2018; @mer ke l _m inimal-len g th_ 20 19; @sh ivers_n ormal _ 2019],(v) Mikad o [@wilhe lm _ela s t ici ty_2003; @ head_def ormation_ 2 003], an d (vi ) 2 D Voro no i n etwor k [@he u ssi nger_ stiff_ 20 06; @a rzash _s tress-st abilized_2019]. Triangu lar ne twork s a re builtbyd epo siting in divi dual fiber s o f l ength $W $ on a per i od ict riang ular lattice.T he la t t ic e spacing i s $ \el l_0=1 $.A fulltria ngular network ha s an average co nnec t i vit y | size scaling_of the_discontinuity and its effect_on the_scaling_exponents, which_have_also previously been_studied using a_complementary approach with the_addition of small,_non-zero_bending rigidity [@sharma_strain-controlled_2016]. Using these modified exponents, we test scaling relations recently predicted for_fiber_networks [@shivers_scaling_2019].
![_\[fig:1\]_Rigidity_phase diagram of central force_networks. [Upon increasing the average]{}_connectivity $z$_at $\gamma=0$, [a network]{} passes through three distinct_regimes:_(i) a disconnected_structure for connectivity less than the percolation connectivity $z<z_p$_(ii) a percolated but floppy network_for $z_p<z<z_c \simeq_2d_$_and (iii) a rigid_network for connectivity greater than $z_c$._Applying a sufficiently large finite strain_to an otherwise floppy network with $z_p<z<z_c$_rigidifies the system. For a given_$z$ in this range, a_critical transition_is observed with increasing strain,_as indicated by_the dashed_arrow. The second-order_line of transitions is characterized by_a critical strain_$\gamma_c(z)$ that varies linearly with $z$_near_$z_c$ [@wyart_elasticity_2008] (see_also_Fig. \[fig:A3\]_in the_Appendix).](fig1){width="6cm" height="6cm"}
Simulation method
=================
To_investigate_the stiffness_discontinuity_in fiber networks, we use various_network_models including (i) triangular, (ii) *phantomized* triangular_[@broedersz_criticality_2011; @licup_stress_2015], (iii) 2D_and_(iv) 3D jammed-packing-derived [@wyart_elasticity_2008;_@tighe_force_2010; @baumgarten_normal_2018; @merkel_minimal-length_2019; @shivers_normal_2019], (v)_Mikado [@wilhelm_elasticity_2003; @head_deformation_2003], and (vi) 2D_Voronoi network_[@heussinger_stiff_2006; @arzash_stress-stabilized_2019]._Triangular networks are built by depositing individual fibers of length $W$_on a periodic triangular lattice. The_lattice spacing is $\ell_0=1$._A full_triangular_network has an_average_connectivity |
right) \gg Y^{\sigma -1/2}, \quad t\in \cA_2.\end{aligned}$$
We refer the reader to either [@Bou Section 1] or [@Ivic Chapter 11] for details of the following reduction from Lemma \[lem:zerodensity1\] which makes use of Heath-Brown’s twelfth power moment estimate [@HB0].
\[lem:zerodensity2\] Let $Y\le T^{A}$ be some parameter. There exists some $N$ satisfying $$\begin{aligned}
\label{eq:zerodensity1X}
Y^{4/3}<N<Y^{2+o(1)},\end{aligned}$$ and some sequence of complex numbers $a_n$ satisfying $|a_n|\le 1$ such that for some well spaced set $\cA\subseteq [0,T]$ with $$\begin{aligned}
N^{o(1)}\left|\sum_{N\le n \le 2N}a_n n^{it}\right|\ge N^{\sigma} \quad t\in \cA,\end{aligned}$$ we have $$\begin{aligned}
N(\sigma,T)\ll T^{o(1)}\left(|\cA|+T^2Y^{6(1-2\sigma)}\right).\end{aligned}$$
Proof of Theorem \[thm:main1\]
==============================
By Lemma \[lem:mainvlarge1\] we have $$\begin{aligned}
\label{eq:main1case22}
I(\delta T,\cA)\ll \frac{N^{2+o(1)}}{V^2}|\cA|+ \frac{N^{3/2+o(1)}}{V^2}W,\end{aligned}$$ where $$\begin{aligned}
W=\sum_{\substack{t_1,t_2\in \cA \\ |t_1-t_2|\le \delta T}}\left|\sum_{n\le \delta T/N}n^{-1/2+i(t_1-t_2)}\right|.\end{aligned}$$ For integer $\ell$, define $$\begin{aligned}
r(\ell)=|\{ (t_1,t_2)\in \cA\times \cA \ : \ \ell< t_1-t_2\le \ell+1 | right) \gg Y^{\sigma -1/2 }, \quad t\in \cA_2.\end{aligned}$$
We refer the reader to either [ @Bou Section 1 ] or [ @Ivic Chapter 11 ] for details of the following decrease from Lemma \[lem: zerodensity1\ ] which do use of Heath - Brown ’s twelfth power here and now appraisal [ @HB0 ].
\[lem: zerodensity2\ ] Let $ Y\le T^{A}$ be some parameter. There exists some $ N$ satisfy $ $ \begin{aligned }
\label{eq: zerodensity1X }
Y^{4/3}<N < Y^{2+o(1)},\end{aligned}$$ and some sequence of complex numbers $ a_n$ satisfy $ |a_n|\le 1 $ such that for some well spaced set $ \cA\subseteq [ 0,T]$ with $ $ \begin{aligned }
N^{o(1)}\left|\sum_{N\le n \le 2N}a_n n^{it}\right|\ge N^{\sigma } \quad t\in \cA,\end{aligned}$$ we get $ $ \begin{aligned }
N(\sigma, T)\ll T^{o(1)}\left(|\cA|+T^2Y^{6(1 - 2\sigma)}\right).\end{aligned}$$
Proof of Theorem \[thm: main1\ ]
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
By Lemma \[lem: mainvlarge1\ ] we have $ $ \begin{aligned }
\label{eq: main1case22 }
I(\delta T,\cA)\ll \frac{N^{2+o(1)}}{V^2}|\cA|+ \frac{N^{3/2+o(1)}}{V^2}W,\end{aligned}$$ where $ $ \begin{aligned }
W=\sum_{\substack{t_1,t_2\in \cA \\ |t_1 - t_2|\le \delta T}}\left|\sum_{n\le \delta T / N}n^{-1/2+i(t_1 - t_2)}\right|.\end{aligned}$$ For integer $ \ell$, define $ $ \begin{aligned }
r(\ell)=|\ { (t_1,t_2)\in \cA\times \cA \: \ \ell < t_1 - t_2\le \ell+1 | rigjt) \gg Y^{\sigma -1/2}, \quad t\in \cX_2.\end{aligned}$$
We rgfwr the readed to eitfer [@Bou Section 1] or [@Ivic Chaptec 11] foe detqils of the following feduction from Lenma \[ltm:zerodensity1\] whirg makes use or Heach-Urown’s twelfth kower moment estimate [@HB0].
\[lem:sefobensity2\] Let $Y\le T^{A}$ be some parameter. There rxlsts some $N$ sajisfypnd $$\befpn{cligned}
\label{eq:zerodensity1X}
Y^{4/3}<N<Y^{2+o(1)},\ehd{aligntd}$$ and some sequenve of complex numbers $a_n$ swtisvying $|a_n|\le 1$ such tjat for somg wejo spaced set $\cA\subseteq [0,T]$ with $$\begjn{aligned}
N^{o(1)}\left|\sum_{N\le n \le 2N}a_n v^{it}\riyht|\ge N^{\sigmq} \wuaf t\in \cA,\end{aoigneq}$$ we have $$\benpn{alignad}
N(\sigms,T)\ll T^{o(1)}\left(|\cA|+T^2N^{6(1-2\sigme)}\rigyt).\end{aligned}$$
Proof of Vheorem \[thm:main1\]
==============================
By Lemia \[lem:mainelcrge1\] we have $$\begin{alitnwd}
\labgl{eq:mdin1cxwe22}
I(\aelua V,\cA)\ml \fraf{N^{2+o(1)}}{T^2}|\cA|+ \frac{N^{3/2+o(1)}}{V^2}S,\end{aligned}$$ where $$\begin{aligned}
E=\stn_{\substack{t_1,t_2\in \cA \\ |t_1-e_2|\lq \delta T}}\left|\sum_{n\le \delta T/N}n^{-1/2+i(t_1-t_2)}\right|.\eng{aljgned}$$ For integer $\ell$, dwfine $$\begin{aligned}
r(\elp)=|\{ (t_1,t_2)\in \cA\eimes \cA \ : \ \ell< t_1-t_2\le \ell+1 | right) \gg Y^{\sigma -1/2}, \quad t\in \cA_2.\end{aligned}$$ the to either Section 1] or of following reduction from \[lem:zerodensity1\] which makes of Heath-Brown’s twelfth power moment estimate \[lem:zerodensity2\] Let $Y\le T^{A}$ be some parameter. There exists some $N$ satisfying $$\begin{aligned} Y^{4/3}<N<Y^{2+o(1)},\end{aligned}$$ and some sequence of complex numbers $a_n$ satisfying $|a_n|\le 1$ such that some spaced $\cA\subseteq with $$\begin{aligned} N^{o(1)}\left|\sum_{N\le n \le 2N}a_n n^{it}\right|\ge N^{\sigma} \quad t\in \cA,\end{aligned}$$ we have $$\begin{aligned} N(\sigma,T)\ll T^{o(1)}\left(|\cA|+T^2Y^{6(1-2\sigma)}\right).\end{aligned}$$ of Theorem \[thm:main1\] ============================== By Lemma \[lem:mainvlarge1\] we $$\begin{aligned} \label{eq:main1case22} I(\delta T,\cA)\ll \frac{N^{3/2+o(1)}}{V^2}W,\end{aligned}$$ where $$\begin{aligned} W=\sum_{\substack{t_1,t_2\in \cA |t_1-t_2|\le T}}\left|\sum_{n\le \delta For $\ell$, $$\begin{aligned} r(\ell)=|\{ (t_1,t_2)\in \cA \ : \ \ell< t_1-t_2\le \ell+1 | right) \gg Y^{\sigma -1/2}, \quad t\in \cA_2.\end{Aligned}$$
We rEfer tHe rEadEr To eiTher [@bou Section 1] or [@IvIC ChaPter 11] for details of the folLowinG rEDuctIOn From LEmma \[lem:ZErODEnsItY1\] wHicH mAKeS use oF HeAth-BrowN’s twelfth pOweR mOment estimatE [@hB0].
\[Lem:zerodenSitY2\] Let $Y\le T^{A}$ be sOme ParameTeR. ThERe exiSts Some $N$ SatisfYIng $$\begIn{aligned}
\LaBEl{eq:zeROdensitY1x}
y^{4/3}<N<y^{2+o(1)},\enD{aligned}$$ and some seQUeNCe of complex numBers $a_n$ SaTIsFYIng $|A_n|\lE 1$ such that fOr Some wELl spaceD SeT $\Ca\SubSEteq [0,T]$ with $$\begiN{aligned}
N^{o(1)}\lEFt|\sUm_{N\le n \Le 2n}a_n N^{It}\righT|\ge N^{\sIgMA} \quAd t\in \cA,\end{aLignEd}$$ we have $$\bEgin{alIGned}
N(\siGMa,T)\ll T^{o(1)}\Left(|\cA|+t^2Y^{6(1-2\sIgmA)}\rigHT).\eNd{AliGnED}$$
PrOOf Of THEorEm \[thm:maiN1\]
==============================
BY LEmma \[lEm:maINVLArge1\] We hAve $$\bEgin{aLigned}
\label{eq:MaiN1casE22}
i(\deLta T,\ca)\ll \frAc{N^{2+o(1)}}{v^2}|\ca|+ \frac{n^{3/2+o(1)}}{V^2}W,\enD{aligNeD}$$ where $$\begin{aligNed}
W=\Sum_{\substaCk{t_1,T_2\iN \cA \\ |T_1-t_2|\Le \delTA T}}\left|\Sum_{N\le \Delta T/N}N^{-1/2+i(t_1-t_2)}\rigHT|.\enD{aLIGNeD}$$ For integer $\ell$, defiNe $$\BEGiN{aligned}
R(\ell)=|\{ (t_1,t_2)\IN \ca\tIMes \cA \ : \ \ell< T_1-t_2\Le \eLl+1 | right) \gg Y^{\sigma -1/2} , \quad t\ in \c A_2 .\e nd {ali gned }$$
We refert he r eader to either [@BouSecti on 1] o r [ @Ivic Chapte r 1 1 ] fo rde tai ls of thefol lowingreductionfro mLemma \[lem: z er odensity1\ ] w hich makes u seof Hea th -Br o wn’stwe lfthpowerm omentestimate[@ H B0].
\ [lem:ze r o de nsit y2\] Let $Y\le T^ { A} $ be some param eter.Th e re e xis tssome $N$ s at isfyi n g $$\be g in { a l ign e d}
\label{eq: zerodensity 1 X}Y^{4/3 }< N<Y ^ {2+o(1 )},\e nd { ali gned}$$ and som e sequenc e of c o mplex n u mbers $ a_n$ s ati sfy ing$ |a _n |\l e1 $ s u ch th a t f or somewe ll spac ed s e t $ \cA\ sub sete q [0, T]$ with $$\b egi n{al i gne d}
N^ {o(1) }\le ft |\sum _{N\le n \l e2N}a_n n^{it}\r ight |\ge N^{\ sig ma } \ qu ad t\ i n \cA, \en d{a ligned} $$ we h a ve$$ \ b e gi n{aligned}
N(\sigm a, T ) \l l T^{o(1 )}\lef t (| \c A |+T^2Y^{ 6( 1-2 \sig m a )}\ri ght) . \e nd{align ed}$$Pr oo f of Th eo rem \[ th m:m ain 1\]
= = ==== ====== ======== ===== = ====
By Lemma \[lem:mainvla r ge 1 \ ]w e ha ve$$\begin{al igne d }
\l abel { eq :ma i n1cas e22}I( \ de l ta T,\cA)\ll \frac{ N^ {2+o(1 )}}{V ^2}|\cA|+ \fr ac{N^{3/2+ o ( 1 )}}{V^2} W,\e n d{ a ligned}$$ wher e $$\ begin{alig n ed}
W=\s um_{\ substack {t_1,t_2\ i n \cA \\|t_ 1-t _2| \le \ de lta T}}\left| \ s um_{ n\ le \del taT/N}n^{ -1/ 2+i (t_ 1-t _2 )}\right| .\end{al ig ne d} $$ Fo r int e ger $\el l$ , d ef ine $$\b e gin{al igned }
r( \e ll ) =|\ { (t_1, t _2 ) \ in \ cA \t imes \c A\ : \ \el l < t _1-t_2\ le \ell+1 | right) \gg_Y^{\sigma -1/2},_\quad t\in \cA_2.\end{aligned}$$
We refer_the reader_to_either [@Bou Section 1]_or [@Ivic_Chapter 11] for details_of the following_reduction from Lemma \[lem:zerodensity1\] which_makes use of_Heath-Brown’s_twelfth power moment estimate [@HB0].
\[lem:zerodensity2\] Let $Y\le T^{A}$ be some parameter. There exists some $N$_satisfying_$$\begin{aligned}
\label{eq:zerodensity1X}
Y^{4/3}<N<Y^{2+o(1)},\end{aligned}$$ and_some_sequence_of complex numbers $a_n$ satisfying_$|a_n|\le 1$ such that for_some well_spaced set $\cA\subseteq [0,T]$ with $$\begin{aligned}
N^{o(1)}\left|\sum_{N\le n \le_2N}a_n_n^{it}\right|\ge N^{\sigma} \quad_t\in \cA,\end{aligned}$$ we have $$\begin{aligned}
N(\sigma,T)\ll T^{o(1)}\left(|\cA|+T^2Y^{6(1-2\sigma)}\right).\end{aligned}$$
Proof of Theorem \[thm:main1\]
==============================
By Lemma \[lem:mainvlarge1\]_we have $$\begin{aligned}
\label{eq:main1case22}
I(\delta T,\cA)\ll \frac{N^{2+o(1)}}{V^2}|\cA|+ \frac{N^{3/2+o(1)}}{V^2}W,\end{aligned}$$_where $$\begin{aligned}
W=\sum_{\substack{t_1,t_2\in \cA_\\_|t_1-t_2|\le_\delta T}}\left|\sum_{n\le \delta T/N}n^{-1/2+i(t_1-t_2)}\right|.\end{aligned}$$_For integer $\ell$, define $$\begin{aligned}
r(\ell)=|\{ (t_1,t_2)\in_\cA\times \cA \ : \ \ell<_t_1-t_2\le \ell+1 |
84 protocol. The essence of this protocol is that the communicators can freely select the message mode and the control mode. In BB84 protocol, when Alice and Bob want to transform $n$ bit message, it need about $4n$ qubit. On the other hand, comparing with the ‘ping-pong’ protocol, we use a single qubit to realize the deterministic secure direct communication instead of using entanglement. Also, there maybe a denial-of-service(DoS) attack in the line $A\rightarrow B$ \[14\]. But any method of message authentification can protect the protocol against man-in-the-middle attacks with a reliable public channel.
In order to be practical and secure, a quantum key distribution scheme must be based on existing—or nearly existing—technology \[15\]. Experimental quantum key distribution was demonstrated for first time by Bennett, et al \[16\]. Since then, single photon source have been studied in recent years and a great variety of approaches has been proposed and implemented \[17-22\]. Today, several groups have shown that quantum key distribution is possible, even outside the laboratory. In principle, any two-level quantum system could be used to implement quantum cryptography (QC). In practice, all implementations have relied on photons. The reason is that their decoherence can be controlled and moderated. The technological challenges of the QC are the questions of how to produce single photons, how to transmit them, how to detect single photons, and how to exploit the intrinsic randomness of quantum processes to build random generators \[23\]. Considered the experimental feasibility, our protocol needs a single photon source and some linear optical elements and a single-photon detector. Recently, the full implementation of a quantum cryptography protocol using a stream of a single photon pulses generated by a stable and efficient source operating at room temperature was reported \[24\]. The single pulses are emitted on demand and the secure bit rate is 7700bits/s. And quantum logic operations using linear optical elements can be realized with today’s technology \[25\]. The implementation of the single-photon detection technology for quantum cryptography have been reported \[26\] and the values of $\sigma _{x},\sigma
_{y},$ and $\sigma _{z}$ of a polarization qubit on a single photon can be ascertained \[27\].Considered the experimental feasibility, this protocol can be realized with today’s technology. It is explained that when this paper was completed, we see the protocol \[28\] presented by Deng $et$ | 84 protocol. The essence of this protocol is that the communicators can freely select the message mode and the restraint mood. In BB84 protocol, when Alice and Bob want to transform $ n$ bit message, it want about $ 4n$ qubit. On the other hired hand, comparing with the ‘ ping - pong ’ protocol, we practice a individual qubit to realize the deterministic dependable direct communication instead of use web. Also, there maybe a abnegation - of - service(DoS) attack in the line $ A\rightarrow B$ \[14\ ]. But any method acting of message authentification can protect the protocol against man - in - the - middle attacks with a authentic public channel.
In order to be hardheaded and secure, a quantum cardinal distribution scheme must be base on existing — or nearly existing — engineering \[15\ ]. Experimental quantum key distribution was demonstrated for first time by Bennett, et al \[16\ ]. Since then, single photon source have been studied in recent years and a great variety of approaches has been proposed and implemented \[17 - 22\ ]. Today, several groups have shown that quantum cardinal distribution is potential, even outside the laboratory. In principle, any two - level quantum arrangement could be used to implement quantum cryptography (QC). In exercise, all implementations have trust on photons. The reason is that their decoherence can be controlled and moderated. The technical challenges of the QC are the questions of how to produce single photons, how to transmit them, how to detect unmarried photons, and how to exploit the intrinsic randomness of quantum processes to build random generator \[23\ ]. Considered the experimental feasibility, our protocol necessitate a single photon source and some linear optical elements and a individual - photon detector. Recently, the full implementation of a quantum cryptography protocol use a stream of a single photon pulses generated by a static and efficient source operating at room temperature was reported \[24\ ]. The single pulsation are emitted on requirement and the secure snatch rate is 7700bits / s. And quantum logic operation using linear optical elements can be realized with today ’s engineering \[25\ ]. The implementation of the single - photon detection engineering for quantum cryptography have been reported \[26\ ] and the values of $ \sigma _ { x},\sigma
_ { y},$ and $ \sigma _ { z}$ of a polarization qubit on a single photon can be ascertain \[27\].Considered the experimental feasibility, this protocol can be recognize with today ’s technology. It is explained that when this newspaper was completed, we see the protocol \[28\ ] present by Deng $ et$ | 84 prltocol. The essence of thls protocol is tkqt the commuhicators can freely select the messaje mide abd the control mode. In BB84 protobol, when Qlict and Bob want to transfovi $n$ glt mevwage, it need anout $4n$ qubid. On the other hxnb, comparing with the ‘ping-pong’ protocjl, we uxe a single qubij to gewlizs the deterministic secure direct dommunibation instead of using entanglement. Also, tjere maybe a denial-of-sfrvice(DoS) ajfach in the line $A\rightarrow B$ \[14\]. But ani method of message authentificagion ean protect tye ototocol agaiist mag-in-the-middle attacks with a reliable publlc chenneo.
In order to be practmcal and secure, a quwntum key dnstribution scheme muwt be bdsed on dzisginf—oc nsarly fxiating—technklogy \[15\]. Expeeimental quantum keu qpxtribution wzs demjnftrated for first time by Bennett, et al \[16\]. Sjnce then, single photon source have been stufied in rqcent years and a great variety of approaches has been orokowed ave lmplemented \[17-22\]. Today, several groups have shown egau qlantum key distrinution is possible, egem outside the uaborackrg. In principle, any two-levgl quabtum systtm coild be used to implement quqntum cryptoyraphy (QC). In practice, all implemzntatipns hsve relied on photons. Tke reaaon is that their deduherence can be zonnrolned and moderated. The techgological chaklenges of jhe QC wre the qufstiokv of how to producf sinyle pvotons, how to transmit them, how to detect single photonx, dnd how to zxploib the intrinsic randomness of quantui prozesses to guild rendom generaeors \[23\]. Considesgd the experikental fqasiviliry, our ofotocol needs s single ihjton source and some llnear kptical elementw abd a single-photpn aetqcnor. Rqwently, the fgll kmpurmentxtion of a wjantim cryptography protmcol using a stream of a single khoton pujses generatec by a stable and tfficixnt sonrce olerwting at room temperature was deported \[24\]. Thc single pulsqs ave eiitted on bemand and the secure bit rate is 7700bits/s. End quantum logic operarions using linear pktical elemenvs can be realised with today’s techbology \[25\]. The implekentation of the singme-photmn dehection technology for quantum cryptography have been reported \[26\] and the vqlues of $\sigma _{x},\sifma
_{y},$ and $\viyma _{z}$ of w poperization qubit om a single photon can be ascerteined \[27\].Conshdzred the experimental feasibikigy, this protozol can be realized with today’s yechnology. It is explained that when this paper was comileted, we see fhe protocul \[28\] presented by Drng $et$ | 84 protocol. The essence of this protocol the can freely the message mode BB84 when Alice and want to transform bit message, it need about $4n$ On the other hand, comparing with the ‘ping-pong’ protocol, we use a single to realize the deterministic secure direct communication instead of using entanglement. Also, there a attack the $A\rightarrow B$ \[14\]. But any method of message authentification can protect the protocol against man-in-the-middle attacks a reliable public channel. In order to be and secure, a quantum distribution scheme must be based existing—or existing—technology \[15\]. quantum distribution demonstrated for first by Bennett, et al \[16\]. Since then, single photon source have been studied in recent years and great variety has been and \[17-22\]. several groups have quantum key distribution is possible, even In principle, any two-level quantum system could be to implement cryptography (QC). In practice, all implementations relied on photons. The reason is that their can be controlled and moderated. The technological challenges of the QC are the questions of produce single photons, how transmit them, how detect photons, how exploit the randomness of quantum processes to build random generators \[23\]. Considered the feasibility, our protocol needs a single photon source and some elements a single-photon detector. the full implementation of quantum protocol using a stream single pulses stable efficient operating at room temperature reported \[24\]. The single pulses emitted on demand and 7700bits/s. And quantum logic operations using linear optical can be realized with today’s technology \[25\]. implementation of the single-photon detection technology for quantum cryptography have been reported and the $\sigma _{x},\sigma _{y},$ and $\sigma _{z}$ of a qubit on a single can be ascertained \[27\].Considered the experimental feasibility, this protocol be with today’s It is explained when this paper completed, we see \[28\] presented Deng | 84 protocol. The essence of this pRotocol is tHat thE coMmuNiCatoRs caN freely select tHE mesSage mode and the control mOde. In bB84 PRotoCOl, When ALice and bOb WANt tO tRaNsfOrM $N$ bIt mesSagE, it need About $4n$ qubiT. On ThE other hand, coMPaRing with thE ‘piNg-pong’ protocOl, wE use a sInGle QUbit tO reAlize The detERminisTic secure DiREct comMUnicatiON InSteaD of using entanglemENt. aLso, there maybe a Denial-Of-SErVICe(DOS) aTtack in the LiNe $A\riGHtarrow b$ \[14\]. buT ANY meTHod of message aUthentificaTIon Can proTeCt tHE protoCol agAiNSt mAn-in-the-middLe atTacks with A reliaBLe publiC Channel.
in ordeR to Be pRactICaL aNd sEcURe, a QUaNtuM Key DistribuTiOn SchemE musT BE BAsed On eXistIng—or Nearly existinG—teChnoLOgy \[15\]. experImentAl quAnTum keY distrIbutiOn Was demonstrated For fIrst time bY BeNnEtt, Et Al \[16\]. SinCE then, sIngLe pHoton soUrce havE BeeN sTUDIeD in recent years and a GrEAT vAriety of ApproaCHeS hAS been proPoSed And iMPLemenTed \[17-22\]. TODaY, several Groups HAvE sHown thaT qUantum KeY diStrIbutiON is pOssiblE, even outSide tHE laboratory. In pRInciple, any two-LEvEL QuANtum SysTem could be uSed tO ImplEmenT QuAntUM crypTograPhY (qC). iN practice, all implemeNtAtions Have rElied on photonS. The reason IS THat their DecoHErENce can be controLled aNd moderateD. the technOlogiCal challEnges of thE qc are the qUesTioNs oF hoW TO pRoduce single pHOTons, HoW to tranSmiT them, hoW to DetEct SinGlE photons, aNd how to eXpLoIt ThE inTrinsIC randomnEsS of QuAntUm proCEsses tO builD ranDoM gENerAtors \[23\]. CoNSiDERed tHe ExPeriMenTaL feasIbilITy, oUr protoCol needs a SinGLe phOtOn Source aNd some linear oPtIcal elemenTs And A singlE-PHoton detEctor. Recently, the full impLEmentatIon Of a quAntuM cryptogrAphY protoCol USing a sTream oF a sinGlE phOTOn pulSES gEneRaTed by a stabLE And EfficIeNt soUrce opeRating at room temperATurE was reported \[24\]. THe sInglE PUlSes ARe EMitTeD On dEMAnd and the secure Bit rate is 7700bItS/S. ANd quantum lOGic OpErationS using lInear OPtical eLements caN be realizEd With TODay’S technologY \[25\]. The implEmentatioN Of the SInGle-phOtoN detecTiOn tEchnoLogy foR QuaNtum cRyptogRaPhy havE been RePorted \[26\] anD the values of $\sigma _{x},\sigma
_{Y},$ and $\siGma _{z}$ oF a pOlarizatiOn qUBit On a single PhotOn can be ascErtAinEd \[27\].ConSidERed thE expERiMenTAl feaSibiLIty, this prOToCol CAN bE realized wiTH TOdaY’s tecHnoLOgy. It iS expLained that when thiS Paper was compleTed, wE SEe tHe pROtocOl \[28\] Presented by DenG $et$ | 84 protocol. The essence o f this pro tocol is th at the com municators can free ly select the messagemodean d the co ntrol mode.I nB B 84pr ot oco l, wh en Al ice and Bo b want totra ns form $n$ bit me ssage, itnee d about $4n$ qu bit. O nthe other ha nd, c ompari n g with the ‘pin g- p ong’ p r otocol, w eusea single qubit to re a lize the deter minist ic se c u redir ect commun ic ation instead of u s ing entanglement. Also, ther e ma ybe ade nia l -of-se rvice (D o S)attack in t he l ine $A\ri ghtarr o w B$ \[ 1 4\]. Bu t anymet hod ofm es sa geau t hen t if ica t ion can pro te ct theprot o c o l aga ins t ma n-in- the-middle at tac ks w i tha rel iable pub li c cha nnel.
In o rd er to be practi caland secur e,aqua nt um ke y distr ibu tio n schem e mustb e b as e d on existing—or nearl ye x is ting—tec hnolog y \ [1 5 \]. Expe ri men talq u antum key di stributi on was de mo nstrate dfor fi rs t t ime by B e nnet t, etal \[16\ ]. Si n ce then, singl e photon sourc e h a v eb eenstu died in rec enty ears and agre a t var ietyof ap p roaches has been pr op osed a nd im plemented \[1 7-22\]. To d a y , severa l gr o up s have shown th at qu antum keyd istribut ion i s possib le, eveno u tside th e l abo rat ory . In principle, a n y two -l evel qu ant um syst emcou ldbeus ed to imp lement q ua nt um c ryp togra p hy (QC). I n p ra cti ce, a l l impl ement atio ns h a vereliedo np h oton s. T he r eas on is t hatt hei r decoh erence ca n b e con tr ol led and moderated. T he technolog ic alchalle n g es of th e QC are the questionso f how t o p roduc e si ngle phot ons , howtot ransmi t them , how t o d e t ect s i n gl e p ho tons, andh o w t o exp lo it t he intr insic randomness o f qu antum process esto b u i ld ra n do m ge ne r ato r s \[23\]. Consid ered the e xp e ri mental fea s ibi li ty, our protoc ol ne e ds a si ngle phot on source a nd s o m e l inear opti cal elem ents anda sing l e- photo n d etecto r. Re centl y, the ful l imp lement at ion of a qu an tum cryp tography protocol using a str eam o f a single p hot o n p ulses gen erat ed by a st abl e a nd ef fic i ent s ourc e o per a tingat r o om temper a tu rew a sreported \[ 2 4 \ ].The s ing l e puls es a re emitted on dem a nd and the sec ureb i t r ate is 7 70 0bits/s. And q uan tu m logic op er ations usin g linear o p tical eleme nts ca n be re a l iz e d with tod ay’ s technol ogy \ [ 25\]. T he i m plemen tati on of th e sing l e-ph o t on detection tec hnolo g y forq uan tum c ry ptograp h y ha ve been re ported \[26 \] and the valu es of $ \s igma _ {x} ,\ sigma
_{y} , $ and $\s igma_{z}$ o fa po lar izatio n qu b i t ona si ng lephoton ca n be as ce r tai ned\[27\ ]. Cons idered th e experim ent a l feasi bi lit y , thisp ro t o col can be re alize d with today ’ s te c hn o logy. It is expla ined th a t w he n thispap e r was comp leted, we see t he p rotocol\[ 28\] p res en ted by Deng $et $ | 84 protocol._The essence_of this protocol is_that the_communicators_can freely_select_the message mode_and the control_mode. In BB84 protocol,_when Alice and_Bob_want to transform $n$ bit message, it need about $4n$ qubit. On the other_hand,_comparing with_the_‘ping-pong’_protocol, we use a single_qubit to realize the deterministic_secure direct_communication instead of using entanglement. Also, there maybe_a_denial-of-service(DoS) attack in_the line $A\rightarrow B$ \[14\]. But any method of_message authentification can protect the protocol_against man-in-the-middle attacks_with_a_reliable public channel.
In order_to be practical and secure, a_quantum key distribution scheme must be_based on existing—or nearly existing—technology \[15\]. Experimental_quantum key distribution was demonstrated for_first time by Bennett, et_al \[16\]._Since then, single photon source_have been studied_in recent_years and a_great variety of approaches has been_proposed and implemented_\[17-22\]. Today, several groups have shown_that_quantum key distribution_is_possible,_even outside_the laboratory. In_principle,_any two-level_quantum_system could be used to implement_quantum_cryptography (QC). In practice, all implementations have_relied on photons. The_reason_is that their decoherence_can be controlled and moderated._The technological challenges of the QC_are the_questions of_how to produce single photons, how to transmit them, how to_detect single photons, and how to_exploit the intrinsic randomness_of quantum_processes_to build random_generators_\[23\]. Considered_the experimental feasibility, our protocol needs a_single photon_source and some linear optical elements_and a single-photon detector._Recently,_the full implementation of a quantum_cryptography protocol using a stream of_a single photon pulses generated_by_a_stable and efficient source operating_at room temperature was reported \[24\]._The single pulses_are emitted on demand and the secure_bit_rate is 7700bits/s. And quantum logic_operations_using linear optical elements can be_realized_with_today’s technology \[25\]. The implementation_of the single-photon detection technology for_quantum cryptography have been reported \[26\] and the values_of $\sigma _{x},\sigma
_{y},$_and $\sigma _{z}$ of a_polarization_qubit_on a single photon can be ascertained \[27\].Considered the experimental_feasibility, this_protocol can be_realized with today’s technology. It is explained that when this_paper was completed, we see the protocol_\[28\] presented by Deng $et$ |
metry\_properties\]); note that if the noise $\xi(t)$ has time-reversal symmetry, the term $\propto \omega_1\omega_2(\omega_1+\omega_2)$ in the integrand in Eq. (\[eq:2nd\_moment\_weak\_noise\]) can be disregarded.
For weak noise $\xi(t)$, the leading contribution to the variance of $u$ comes from the second-order term $\propto \Xi_2$. To reveal a nonzero third-order noise correlator, in addition to the variance of $u$ one should measure the third cumulant of $u$, $$\begin{aligned}
\label{eq:3rd_u_cumulant}
&&\frac{\langle u^3\rangle - 3\langle u\rangle\langle u^2\rangle +2\langle u\rangle^3}{\langle u\rangle^3}
\approx -i\int d\omega_1d\omega_2\Xi_3^{(\Lambda)}(\omega_1,\omega_2)
\nonumber\\
&& \times\left[\Lambda^3 + \Lambda(\omega_1^2 + \omega_2^2 + \omega_1\omega_2) + i\omega_1\omega_2(\omega_1+\omega_2)\right].\end{aligned}$$ One can see that, in the case of weak $\delta$-correlated noise, Eqs. (\[eq:Lorentzian\_response\]) — (\[eq:moments\_delta\_corr\]) agree with Eqs. (\[eq:3rd\_order\_linear\_u\]) — (\[eq:3rd\_u\_cumulant\]). However, the results of this section are not limited to $\delta$-correlated noise.
Quantum formulation
===================
The above arguments can be immediately extended to the quantum regime, since the responses of quantum and classical harmonic oscillators to resonant modulation are the same. In the absence of coupling to a thermal reservoir, the Hamiltonian of the oscillator in a resonant field $F\cos\omega_Ft$ in the presence of weak classical frequency noise $\xi(t)$ is $$\label{eq:H_0}
H_0=\hbar [\omega_0+\xi(t)]a^{\dagger}a - qF\cos\omega_Ft,$$ where $a= (2\hbar\omega_0 | metry\_properties\ ]); note that if the noise $ \xi(t)$ has time - transposition isotropy, the term $ \propto \omega_1\omega_2(\omega_1+\omega_2)$ in the integrand in Eq. (\[eq:2nd\_moment\_weak\_noise\ ]) can be disregarded.
For fallible randomness $ \xi(t)$, the leading contribution to the variance of $ u$ comes from the second - holy order term $ \propto \Xi_2$. To reveal a nonzero third - ordering noise correlator, in addition to the variability of $ u$ one should measure the third cumulant of $ u$, $ $ \begin{aligned }
\label{eq:3rd_u_cumulant }
& & \frac{\langle u^3\rangle - 3\langle u\rangle\langle u^2\rangle +2\langle u\rangle^3}{\langle u\rangle^3 }
\approx -i\int d\omega_1d\omega_2\Xi_3^{(\Lambda)}(\omega_1,\omega_2)
\nonumber\\
& & \times\left[\Lambda^3 + \Lambda(\omega_1 ^ 2 + \omega_2 ^ 2 + \omega_1\omega_2) + i\omega_1\omega_2(\omega_1+\omega_2)\right].\end{aligned}$$ One can see that, in the sheath of decrepit $ \delta$-correlated noise, Eqs. (\[eq: Lorentzian\_response\ ]) — (\[eq: moments\_delta\_corr\ ]) match with Eqs. (\[eq:3rd\_order\_linear\_u\ ]) — (\[eq:3rd\_u\_cumulant\ ]). However, the results of this section are not restrict to $ \delta$-correlated noise.
Quantum formulation
= = = = = = = = = = = = = = = = = = =
The above arguments can be immediately extend to the quantum regime, since the responses of quantum and classical consonant oscillators to evocative modulation are the same. In the absence of coupling to a thermal reservoir, the Hamiltonian of the oscillator in a resonant discipline $ F\cos\omega_Ft$ in the presence of weak classical frequency noise $ \xi(t)$ is $ $ \label{eq: H_0 }
H_0=\hbar [ \omega_0+\xi(t)]a^{\dagger}a - qF\cos\omega_Ft,$$ where $ a= (2\hbar\omega_0 | metgy\_properties\]); note that in the noise $\xi(t)$ kqs timx-reverszl symmegry, the term $\propto \omega_1\omeja_2(\omwga_1+\omtya_2)$ in the integrand kn Eq. (\[eq:2nd\_loment\_weqk\_nomse\]) can be disrejzrded.
Fov weai noivx $\xi(t)$, the leadikg contribudion to the vasixnee of $u$ comes from the second-order tqrm $\proltl \Xi_2$. To reveal a npgzerk third-order noise correlator, in asdition to the variamce of $u$ one should measurf thf third cumulant ov $u$, $$\begin{alugneq}
\oabel{eq:3rd_u_cuoulant}
&&\frac{\langle u^3\rangme - 3\langle u\rangle\langle u^2\rangld +2\lanyle u\rangle^3}{\oabglf u\rangle^3}
\appcox -i\igt d\omega_1d\omcba_2\Xi_3^{(\Lakbda)}(\omeba_1,\omega_2)
\nonumbev\\
&& \timxs\ledt[\Lambda^3 + \Lambda(\omega_1^2 + \omega_2^2 + \omega_1\omegw_2) + i\omega_1\mmzga_2(\omega_1+\omega_2)\right].\end{qlugned}$$ One can wee thzt, ih the fasx of weak $\dslta$-correlared noise, Eqs. (\[eq:Loremtspsn\_response\]) — (\[sq:momegtf\_delta\_corr\]) agree with Eqs. (\[eq:3rd\_order\_lineag\_u\]) — (\[sq:3rd\_u\_cumulant\]). However, tye results of this seftion are not limited to $\delta$-correlated noise.
Quantum formglatikv
===================
Tht qbove qrhuments can be immediately extended to the quwhtim regime, since bhe responses of qiajtii and classicxl harmonjc oscillators to gesonanj moduoation art the same. In the absence of coupling to a tkernal reservoir, the Kamiltonian uf tne osvillator in a resonant yield $R\cos\omega_Ft$ in the pddsence of weak cuasxiwal frequency noise $\xi(t)$ is $$\label{eq:H_0}
I_0=\hbar [\omega_0+\xk(t)]a^{\dsgger}a - qF\cos\omeha_Ft,$$ where $a= (2\hbar\omega_0 | metry\_properties\]); note that if the noise $\xi(t)$ symmetry, term $\propto in the integrand disregarded. weak noise $\xi(t)$, leading contribution to variance of $u$ comes from the term $\propto \Xi_2$. To reveal a nonzero third-order noise correlator, in addition to variance of $u$ one should measure the third cumulant of $u$, $$\begin{aligned} \label{eq:3rd_u_cumulant} u^3\rangle 3\langle u^2\rangle u\rangle^3}{\langle u\rangle^3} \approx -i\int d\omega_1d\omega_2\Xi_3^{(\Lambda)}(\omega_1,\omega_2) \nonumber\\ && \times\left[\Lambda^3 + \Lambda(\omega_1^2 + \omega_2^2 + \omega_1\omega_2) + i\omega_1\omega_2(\omega_1+\omega_2)\right].\end{aligned}$$ can see that, in the case of weak noise, Eqs. (\[eq:Lorentzian\_response\]) — agree with Eqs. (\[eq:3rd\_order\_linear\_u\]) — However, results of section not to $\delta$-correlated noise. formulation =================== The above arguments can be immediately extended to the quantum regime, since the responses of and classical to resonant are same. the absence of a thermal reservoir, the Hamiltonian of a resonant field $F\cos\omega_Ft$ in the presence of classical frequency $\xi(t)$ is $$\label{eq:H_0} H_0=\hbar [\omega_0+\xi(t)]a^{\dagger}a - where $a= (2\hbar\omega_0 | metry\_properties\]); note that if The noise $\xi(T)$ has tIme-RevErSal sYmmeTry, the term $\propTO \omeGa_1\omega_2(\omega_1+\omega_2)$ in the IntegRaND in EQ. (\[Eq:2Nd\_momEnt\_weak\_NOiSE\]) Can Be DiSreGaRDeD.
For wEak Noise $\xi(T)$, the leadinG coNtRibution to thE VaRiance of $u$ cOmeS from the secoNd-oRder teRm $\ProPTo \Xi_2$. TO reVeal a NonzerO Third-oRder noise CoRRelatoR, In additION tO the Variance of $u$ one shoULd MEasure the third CumulaNt OF $u$, $$\BEGin{AliGned}
\label{eQ:3rD_u_cumULant}
&&\fraC{\LaNGLE u^3\rANgle - 3\langle u\raNgle\langle u^2\RAngLe +2\langLe U\raNGle^3}{\lanGle u\rAnGLe^3}
\aPprox -i\int d\oMega_1D\omega_2\Xi_3^{(\LAmbda)}(\oMEga_1,\omegA_2)
\NonumbeR\\
&& \times\LefT[\LaMbda^3 + \lAmBdA(\omEgA_1^2 + \OmeGA_2^2 + \oMegA_1\OmeGa_2) + i\omega_1\OmEgA_2(\omegA_1+\omeGA_2)\RIGht].\eNd{aLignEd}$$ One Can see that, in tHe cAse oF WeaK $\deltA$-corrElatEd Noise, eqs. (\[eq:LOrentZiAn\_response\]) — (\[eq:momEnts\_Delta\_corr\]) AgrEe WitH EQs. (\[eq:3rD\_Order\_lIneAr\_u\]) — (\[Eq:3rd\_u\_cuMulant\]). HOWevEr, THE ReSults of this section ArE NOt Limited tO $\delta$-COrReLAted noisE.
QUanTum fORMulatIon
===================
THE aBove arguMents cAN bE iMmediatElY extenDeD to The QuantUM regIme, sinCe the resPonseS Of quantum and clASsical harmoniC OsCILlATors To rEsonant moduLatiON are The sAMe. in tHE abseNce of CoUPlINg to a thermal reservoIr, The HamIltonIan of the oscilLator in a reSONAnt field $f\cos\OMeGA_Ft$ in the presenCe of wEak classicAL frequenCy noiSe $\xi(t)$ is $$\lAbel{eq:H_0}
H_0=\hBAR [\omega_0+\xi(T)]a^{\dAggEr}a - QF\cOS\OmEga_Ft,$$ where $a= (2\hbAR\OmegA_0 | metry\_properties\]); note that if t he no ise $\ xi (t)$ has time-reversal symm etry, the term $\propt o \om eg a _1\o m eg a_2(\ omega_1 + \o m e ga_ 2) $inth e i ntegr and in Eq. (\[eq:2nd \_m om ent\_weak\_n o is e\]) can b e d isregarded.
Fo r weak n ois e $\xi (t) $, th e lead i ng con tribution t o the v a rianceo f $ u$ c omes from the sec o nd - order term $\p ropto\X i _2 $ . To re veal a non ze ro th i rd-orde r n o i s e c o rrelator, inaddition to the varia nc e o f $u$ o ne sh ou l d m easure thethir d cumulan t of $ u $, $$\b e gin{ali gned}\la bel {eq: 3 rd _u _cu mu l ant }
& &\f r ac{ \langleu^ 3\ rangl e -3 \ l a ngle u\ rang le\la ngle u^2\rang le+2\l a ngl e u\r angle ^3}{ \l angle u\ran gle^3 }\approx -i\intd\om ega_1d\om ega _2 \Xi _3 ^{(\L a mbda)} (\o meg a_1,\om ega_2)\ non um b e r \\
&& \times\left[\L am b d a^ 3 + \Lam bda(\o m eg a_ 1 ^2 + \om eg a_2 ^2 + \ omega _1\o m eg a_2) + i \omega _ 1\ om ega_2(\ om ega_1+ \o meg a_2 )\rig h t].\ end{al igned}$$ Onec an see that, i n the case ofw ea k $\ d elta $-c orrelated n oise , Eqs . (\ [ eq :Lo r entzi an\_r es p on s e\]) — (\[eq:moment s\ _delta \_cor r\]) agree wi th Eqs. (\ [ e q :3rd\_or der\ _ li n ear\_u\]) — (\ [eq:3 rd\_u\_cum u lant\]). Howe ver, the resultso f this se cti onare no t li mited to $\de l t a$-c or related no ise.
Q uan tum fo rmu la tion
==== ======== == == == =
Th e abo v e argume nt s c an be imme d iately exte nded t ot hequantum re g i me,si nc e th e r es ponse s of qua ntum an d classic alh armo ni coscilla tors to reson an t modulati on ar e thes a me. In t he absence of couplingt o a the rma l res ervo ir, the H ami ltonia n o f the o scilla tor i na r e s onant f ie ld$F \cos\omega _ F t$in th epres ence of weak classical fr e que ncy noise $\x i(t )$ i s $$ \la b el { eq: H_ 0 }
H _ 0 =\hbar [\omega_ 0+\xi(t)]a ^{ \ da gger}a - q F \co s\ omega_F t,$$ wh ere $ a = (2\hb ar\omega_ 0 | metry\_properties\]); note_that if_the noise $\xi(t)$ has_time-reversal symmetry,_the_term $\propto_\omega_1\omega_2(\omega_1+\omega_2)$_in the integrand_in Eq. (\[eq:2nd\_moment\_weak\_noise\]) can_be disregarded.
For weak noise_$\xi(t)$, the leading_contribution_to the variance of $u$ comes from the second-order term $\propto \Xi_2$. To reveal_a_nonzero third-order_noise_correlator,_in addition to the variance_of $u$ one should measure_the third_cumulant of $u$, $$\begin{aligned}
\label{eq:3rd_u_cumulant}
&&\frac{\langle u^3\rangle - 3\langle u\rangle\langle_u^2\rangle_+2\langle u\rangle^3}{\langle u\rangle^3}
\approx_-i\int d\omega_1d\omega_2\Xi_3^{(\Lambda)}(\omega_1,\omega_2)
\nonumber\\
&& \times\left[\Lambda^3 + \Lambda(\omega_1^2 + \omega_2^2 + \omega_1\omega_2)_+ i\omega_1\omega_2(\omega_1+\omega_2)\right].\end{aligned}$$ One can see that,_in the case_of_weak_$\delta$-correlated noise, Eqs. (\[eq:Lorentzian\_response\]) —_(\[eq:moments\_delta\_corr\]) agree with Eqs. (\[eq:3rd\_order\_linear\_u\]) — (\[eq:3rd\_u\_cumulant\])._However, the results of this section_are not limited to $\delta$-correlated noise.
Quantum formulation
===================
The_above arguments can be immediately extended_to the quantum regime, since_the responses_of quantum and classical harmonic_oscillators to resonant_modulation are_the same. In_the absence of coupling to a_thermal reservoir, the_Hamiltonian of the oscillator in a_resonant_field $F\cos\omega_Ft$ in_the_presence_of weak_classical frequency noise_$\xi(t)$_is $$\label{eq:H_0}
H_0=\hbar_[\omega_0+\xi(t)]a^{\dagger}a_- qF\cos\omega_Ft,$$ where $a= (2\hbar\omega_0 |
7 $\pm$ 2 [MeV/]{}$c^2$, i.e. compatible with the detector resolution. Therefore, only an upper limit is derived: the full width of the signal is smaller than 15 at 95 % C.L. There is no natural explanation of such a small value, neither for the D$^{*'}$ nor for higher orbital excitations [@pene].
Various checks were performed. Varying the background shape and the kinematical cuts has no effect within statistics. No peculiar double counting was noticed, and the signal is stable when the $\pi^*$ is added to the $D^0 \pi^+\pi^-$ tracks in the vertex fit. As explained above, mass shifts are studied using $D^0_1$ and $D^{*0}_2$ narrow states, and a conservative systematic error of 6 is attached to the mass measurement.
The production rate of this signal can be compared with that of the $D^0_1$ and $D^{*0}_2$ narrow states:
$$\begin{array}{rcl}
\frac
{ < N_{D^{*'}} > \times Br ( D^{*'} \rightarrow D^* \pi^+ \pi^- ) }
{\sum_{J=1,2}
< N_{D^{(*)}_J} > \times Br ( D^{(*)}_J \rightarrow D^* \pi ) }
\\
= 0.49 \pm 0.18 (stat) \pm 0.10 (syst)
\end{array}
\label{eq:ratio}$$
Most of the systematic uncertainties cancel in this ratio. The quoted systematics is due to the Monte-Carlo statistics, and to the uncertainties on widths and on the kaon rejection. This result is compatible, within its large errors, with the value obtained using the thermodynamical models already mentioned for orbital states [@n483; @becatini].
Conclusion
==========
Using about 7000 exclusively reconstructed $D^*$ mesons, the $D^0_1$ and $D^{*0}_2$ multiplicities are measured in [ ]{}events, and found to be consistent with theoretical calculations. The measured multiplicities in [ ]{}events are consistent with the ones in [ ]{}events, both for the $D^0_1$ and $D^{*0}_2$.
The total D$^{**}$ production rate, involving | 7 $ \pm$ 2 [ MeV/]{}$c^2 $, i.e. compatible with the detector resolution. Therefore, only an upper terminus ad quem is derive: the full width of the signal is smaller than 15 at 95% C.L. There be no lifelike explanation of such a small value, neither for the D$^{*'}$ nor for higher orbital excitations [ @pene ].
assorted checks were performed. change the background shape and the kinematical cut has no impression within statistic. No peculiar double count was noticed, and the signal is static when the $ \pi^*$ is add to the $ D^0 \pi^+\pi^-$ tracks in the vertex fit. As explained above, aggregate shifts are studied using $ D^0_1 $ and $ D^{*0}_2 $ narrow states, and a cautious systematic error of 6 is attached to the mass measurement.
The production pace of this signal can be compared with that of the $ D^0_1 $ and $ D^{*0}_2 $ narrow states:
$ $ \begin{array}{rcl }
\frac
{ < N_{D^ { *' } } > \times Br (D^ { *' } \rightarrow D^ * \pi^+ \pi^-) }
{ \sum_{J=1,2 }
< N_{D^{(*)}_J } > \times Br (D^{(*)}_J \rightarrow D^ * \pi) }
\\
= 0.49 \pm 0.18 (stat) \pm 0.10 (syst)
\end{array }
\label{eq: ratio}$$
Most of the systematic uncertainties cancel in this proportion. The quote systematics is due to the Monte - Carlo statistics, and to the uncertainties on widths and on the kaon rejection. This result is compatible, within its large errors, with the value obtained using the thermodynamic model already mentioned for orbital department of state [ @n483; @becatini ].
Conclusion
= = = = = = = = = =
use about 7000 exclusively reconstructed $ D^*$ meson, the $ D^0_1 $ and $ D^{*0}_2 $ multiplicities are measure in [ ] { } events, and found to be reproducible with theoretical calculations. The measured multiplicities in [ ] { } events are consistent with the ones in [ ] { } event, both for the $ D^0_1 $ and $ D^{*0}_2$.
The total D$^{**}$ production rate, involving | 7 $\pl$ 2 [MeV/]{}$c^2$, i.e. compatible witm the detector rgsilutioi. Thererore, onlh an upper limit is derived: vhe dull qidth of the signal is smaller nhan 15 at 95 % C.L. Uhere is no naturem explakction lf sbci a small value, neither fmr the D$^{*'}$ nor fmr hngher orbital excitations [@pene].
Various checks wfre performed. Darypnd ths background shape and the kinematjcal cuus has no effect wothin statistics. No peculiwr dluble counting was noticed, ane thq signal is sgable when the $\pi^*$ is asded to the $D^0 \pi^+\pi^-$ tracks in thd vercex fit. As grplalted above, mess shpfts are studlvd usinc $D^0_1$ and $D^{*0}_2$ narrow statcs, anv a xonservative systematmc error of 6 is attacred to tha jass measurement.
Tye prodoctiot ragw ow tgix aignal cai be comparsd with thar of the $D^0_1$ and $D^{*0}_2$ natrjq states:
$$\begin{zrray}{rsl}
\srac
{ < N_{D^{*'}} > \times Br ( D^{*'} \rightarrow D^* \pi^+ \pp^- ) }
{\shm_{J=1,2}
< N_{D^{(*)}_J} > \times Br ( D^{(*)}_J \rughtarrow D^* \pi ) }
\\
= 0.49 \pm 0.18 (stat) \pm 0.10 (syst)
\end{array}
\label{eq:ratio}$$
Most of the systematic uncectxinuicf xajcel in this ratio. The quoted systematics is qhe tp the Monte-Carjo statistivs, amq to the uncettaintiza kn widths and on tje kaon rejextion. Thif rexult is compatible, within irs large errjes, with the value lbtained usnng thg therkodynamical models alrecdy mehtioned for orbital agates [@n483; @becatini].
Cunckuvion
==========
Using about 7000 exclusivejy reconsvructzd $D^*$ mesuns, jhe $D^0_1$ agd $D^{*0}_2$ multiolicibhes are measured ij [ ]{}evgnts, atd found tl be consistent with theoretical calculations. Tve keasured multlplicities in [ ]{}qvents are conxistent with the ones pn [ ]{}eventv, both for ehe $D^0_1$ and $D^{*0}_2$.
Tha total D$^{**}$ provuction rwte, unvooving | 7 $\pm$ 2 [MeV/]{}$c^2$, i.e. compatible with resolution. only an limit is derived: signal smaller than 15 95 % C.L. is no natural explanation of such small value, neither for the D$^{*'}$ nor for higher orbital excitations [@pene]. Various were performed. Varying the background shape and the kinematical cuts has no effect statistics. peculiar counting noticed, and the signal is stable when the $\pi^*$ is added to the $D^0 \pi^+\pi^-$ tracks the vertex fit. As explained above, mass shifts studied using $D^0_1$ and narrow states, and a conservative error 6 is to mass The production rate this signal can be compared with that of the $D^0_1$ and $D^{*0}_2$ narrow states: $$\begin{array}{rcl} \frac { N_{D^{*'}} > ( D^{*'} D^* \pi^- } {\sum_{J=1,2} < \times Br ( D^{(*)}_J \rightarrow D^* \\ = 0.49 \pm 0.18 (stat) \pm 0.10 \end{array} \label{eq:ratio}$$ of the systematic uncertainties cancel in ratio. The quoted systematics is due to the statistics, and to the uncertainties on widths and on the kaon rejection. This result is its large errors, with value obtained using thermodynamical already for states [@n483; Conclusion ========== Using about 7000 exclusively reconstructed $D^*$ mesons, the $D^0_1$ $D^{*0}_2$ multiplicities are measured in [ ]{}events, and found to with calculations. The measured in [ ]{}events are with ones in [ ]{}events, the and D$^{**}$ rate, | 7 $\pm$ 2 [MeV/]{}$c^2$, i.e. compatible with the Detector reSolutIon. theReFore, Only An upper limit is DErivEd: the full width of the sigNal is SmALler THaN 15 at 95 % C.L. there is NO nATUraL eXpLanAtIOn Of sucH a sMall valUe, neither fOr tHe d$^{*'}$ nor for higheR OrBital excitAtiOns [@pene].
VarioUs cHecks wErE peRFormeD. VaRying The bacKGround Shape and tHe KInematICal cuts HAS nO effEct within statistiCS. NO Peculiar double CountiNg WAs NOTicEd, aNd the signaL iS stabLE when thE $\Pi^*$ IS ADdeD To the $D^0 \pi^+\pi^-$ traCks in the verTEx fIt. As exPlAinED above, Mass sHiFTs aRe studied usIng $D^0_1$ And $D^{*0}_2$ narroW stateS, And a conSErvativE systeMatIc eRror OF 6 iS aTtaChED to THe MasS MeaSurement.
thE pRoducTion RATE Of thIs sIgnaL can bE compared with ThaT of tHE $D^0_1$ aNd $D^{*0}_2$ naRrow sTateS:
$$\bEgin{aRray}{rcL}
\frac
{ < n_{D^{*'}} > \Times Br ( D^{*'} \rightarRow D^* \Pi^+ \pi^- ) }
{\sum_{J=1,2}
< N_{d^{(*)}_J} > \tImEs BR ( D^{(*)}_j \righTArrow D^* \Pi ) }
\\
= 0.49 \pM 0.18 (stAt) \pm 0.10 (sysT)
\end{arrAY}
\laBeL{EQ:RaTio}$$
Most of the systemAtIC UnCertaintIes canCEl In THis ratio. thE quOted SYStemaTics IS dUe to the MOnte-CaRLo StAtisticS, aNd to thE uNceRtaIntieS On wiDths anD on the kaOn rejECtion. This resulT Is compatible, wIThIN ItS LargE erRors, with the ValuE ObtaIned USiNg tHE therModynAmICaL Models already mentioNeD for orBital States [@n483; @becatiNi].
ConclusiON
==========
uSing abouT 7000 excLUsIVely reconstrucTed $D^*$ mEsons, the $D^0_1$ aND $D^{*0}_2$ multipLicitIes are meAsured in [ ]{}eVENts, and foUnd To bE coNsiSTEnT with theoretiCAL calCuLations. the MeasureD muLtiPliCitIeS in [ ]{}events Are consiStEnT wItH thE ones IN [ ]{}events, bOtH foR tHe $D^0_1$ And $D^{*0}_2$.
THE total d$^{**}$ prodUctiOn RaTE, inVolving | 7 $\pm$ 2 [MeV/]{}$c^2$,i.e. compa tible wi thth e de tect or resolution. Ther efore, only an upper l imitis deri v ed : the full w i dt h ofth esig na l i s sma lle r than15 at 95 % C. L. There is no na tural expl ana tion of such asmallva lue , neit her forthe D$ ^ {*'}$nor for h ig h er orb i tal exc i t at ions [@pene].
Variou s c h ecks were perf ormed. V a ry i n g t hebackground s hapea nd thek in e m a tic a l cuts has no effect wit h instatis ti cs. No pec uliar d o ubl e countingwasnoticed,and th e signal is stab le whe n t he$\pi ^ *$ i s a dd e d t o t he$ D^0 \pi^+\p i^ -$ trac ks i n t h e ve rte x fi t. As explained ab ove , ma s s s hifts arestud ie d usi ng $D^ 0_1$an d $D^{*0}_2$ na rrow states,and a co ns ervat i ve sys tem ati c error of 6 i s at ta c h e dto the mass measur em e n t.
The pr oducti o nra t e of thi ssig nalc a n becomp a re d with t hat of th e$D^0_1$ a nd $D^ {* 0}_ 2$narro w sta tes:
$$\begin {arra y }{rcl}
\frac
{ < N_{D^{*'}}> \ t i me s Br( D ^{*'} \righ tarr o w D^ * \p i ^+ \p i ^- )}
{\s um _ {J = 1,2}
< N_{D^{(*)}_J }> \tim es Br ( D^{(*)}_J\rightarro w D ^* \pi ) } \ \= 0.49 \pm 0.1 8 (st at) \pm 0. 1 0 (syst)
\end {array}
\label{e q : ratio}$$
M ost of th e sy stematic unce r t aint ie s cance l i n thisrat io. Th e q uo ted syste matics i sdu eto th e Mon t e-Carlost ati st ics , and to the unce rtai nt ie s on widths an d on t he k aonrej ec tion. Thi s re sult is compatib le, with in i ts larg e errors, wit hthe valueob tai ned us i n g the th ermodynamical models al r eady me nti onedfororbital s tat es [@n 483 ; @beca tini].
Con cl usi o n
==== = = == ==
U sing about 7 000 excl us ivel y recon structed $D^*$ mes o ns, the $D^0_1$and $D^ { * 0} _2$ mu l tip li c iti e s are measured i n [ ]{}eve nt s ,and foundt o b econsist ent wit h the o retical calculat ions. The m easu r e d m ultiplicit ies in [ ]{}event s arec on siste ntwith t he on es in [ ]{} e ven ts, b oth fo rthe $D ^0_1$ a nd $D^{* 0}_2$.
The total D$^{* *}$ pr oduct ion rate, in vol v ing | 7_$\pm$ 2 [MeV/]{}$c^2$,_i.e. compatible with the_detector resolution._Therefore,_only an_upper_limit is derived:_the full width_of the signal is_smaller than 15 at_95_% C.L. There is no natural explanation of such a small value, neither for_the_D$^{*'}$ nor_for_higher_orbital excitations [@pene].
Various checks were performed._Varying the background shape and_the kinematical_cuts has no effect within statistics. No peculiar_double_counting was noticed,_and the signal is stable when the $\pi^*$ is_added to the $D^0 \pi^+\pi^-$ tracks_in the vertex_fit._As_explained above, mass shifts_are studied using $D^0_1$ and $D^{*0}_2$_narrow states, and a conservative systematic_error of 6 is attached to the mass_measurement.
The production rate of this signal_can be compared with that_of the_$D^0_1$ and $D^{*0}_2$ narrow states:
$$\begin{array}{rcl}
\frac
{_< N_{D^{*'}} >_\times Br_( D^{*'} \rightarrow_D^* \pi^+ \pi^- ) }
{\sum_{J=1,2}
< N_{D^{(*)}_J}_> \times Br_( D^{(*)}_J \rightarrow D^* \pi )_}_
\\
= 0.49 \pm_0.18_(stat)_\pm 0.10_(syst)
\end{array}
\label{eq:ratio}$$
Most of_the_systematic uncertainties_cancel_in this ratio. The quoted systematics_is_due to the Monte-Carlo statistics, and to_the uncertainties on widths_and_on the kaon rejection._This result is compatible, within_its large errors, with the value_obtained using_the thermodynamical_models already mentioned for orbital states [@n483; @becatini].
Conclusion
==========
Using about 7000 exclusively reconstructed_$D^*$ mesons, the $D^0_1$ and $D^{*0}_2$_multiplicities are measured in_[ ]{}events,_and_found to be_consistent_with theoretical_calculations. The measured multiplicities in [ ]{}events_are consistent_with the ones in [ ]{}events,_both for the $D^0_1$_and_$D^{*0}_2$.
The total D$^{**}$ production rate, involving |
in [@Sun; @LS], and they are closely related to orthogonal algebraic polynomials on the domain bounded by Steiner’s hyercycloid [@K; @LSX], much as Chebyshev polynomials arise from exponentials. In fact, the trigonometric functions arise from the exponentials by symmetry are called generalized cosine functions in [@LSX; @Sun], and there are also generalized sine functions that are anti-symmetric. Our results on the hexagonal domain can be easily translated to results in terms of generalized cosines. Our results on summability can also be translated to orthogonal expansions of algebraic polynomials on the domain bounded by hypercycloid, but the same cannot be said on our results on best approximation. In fact, just like the case of best approximation by polynomial on the interval, the approximation should be better at the boundary for polynomial approximation on the hypercycloid domain. For example, our Bernstein type inequality (Theorem 4.8) can be translated into a Markov type inequality for algebraic polynomials. A modulus of smoothness will need to be defined to take into account of the boundary effect of the hypercycloid domain, which is not trivial and will not be considered in this paper. Some of our results, especially those on the best approximation, can be extended to higher dimensions. We choose to stay on the hexagonal domain to keep an uniformity of the paper and to stay away from overwhelming notations.
The paper is organized as follows. Definitions and background materials will be given in Section 2. In Section 3 we study the Abel summability, aka Poisson integral, and Cesàro $(C,\delta)$ means of the Fourer series on the hexagon, where several compact formulas for the kernel functions will be deduced. One interesting result shows that the $(C,2)$ means are nonnegative, akin to the Fejèr means for the classical Fourier series. In Section 4 we study best approximation by trigonometric functions on the hexagonal domain and establish both direct and inverse theorems in terms of a modulus of smoothness.
Fourier series on the regular hexagon
=====================================
Below we briefly sum up what we need on Fourier analysis on hexagonal domain. We refer to [@LSX] for further details. The hexagonal lattice is given by $H \ZZ^2$, where the matrix $H$ and the spectral set $\Omega_H$ are given by $$H=\begin{pmatrix} \sqrt{3} & 0\\ -1 & 2\end{ | in [ @Sun; @LS ], and they are closely related to orthogonal algebraic polynomials on the knowledge domain restrict by Steiner ’s hyercycloid [ @K; @LSX ], much as Chebyshev polynomials arise from exponential. In fact, the trigonometric functions arise from the exponentials by isotropy are address generalized cosine functions in [ @LSX; @Sun ], and there embody also generalized sine function that are anti - symmetric. Our results on the hexangular domain can be easily translated to results in terms of generalized cosines. Our results on summability can besides be translated to orthogonal expansions of algebraic polynomial on the domain bounded by hypercycloid, but the same cannot be said on our results on good approximation. In fact, just like the case of best approximation by polynomial on the time interval, the approximation should be better at the boundary for polynomial approximation on the hypercycloid domain. For example, our Bernstein character inequality (Theorem 4.8) can be translate into a Markov type inequality for algebraic polynomials. A modulus of smoothness will need to be defined to take into account of the boundary effect of the hypercycloid domain, which is not trivial and will not be consider in this paper. Some of our results, specially those on the good approximation, can be extended to higher dimension. We choose to stay on the hexagonal domain to keep an uniformity of the newspaper and to stay away from overwhelming notations.
The paper is organized as succeed. Definitions and background materials will be given in incision 2. In Section 3 we analyze the Abel summability, aka Poisson integral, and Cesàro $ (C,\delta)$ means of the Fourer series on the hexagon, where several compact rule for the kernel functions will be deduced. One interesting result usher that the $ (C,2)$ means are nonnegative, akin to the Fejèr means for the classical Fourier series. In department 4 we study best estimate by trigonometric functions on the hexagonal domain and establish both lineal and inverse theorems in terms of a modulus of smoothness.
Fourier serial on the regular hexagon
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Below we briefly sum up what we need on Fourier analysis on hexangular domain. We refer to [ @LSX ] for further details. The hexangular lattice is given by $ H \ZZ^2 $, where the matrix $ H$ and the spectral set $ \Omega_H$ are pass by $ $ H=\begin{pmatrix } \sqrt{3 } & 0\\ -1 & 2\end { | in [@Sun; @LS], and they are clorely related to orthogmnal amgebraic polynomials on the domain blubded vy Steiner’s hyercycloia [@K; @LSX], mlch as Chwbysiev polynomials edise from expkkenticlw. In fact, the jrigonometriw functions arhsd yrom the exponentials by symmetry arq callec heneralized cofine suncfpoks in [@LSX; @Sun], and there are also generanized sine fumctions that are anti-symmehric. Our results on thf hexagonal domwun can be earily translated to resolts in terms of generalized cosknes. Pur resultw in dommability cen alsj be translabvd to osthogonsl expansions pf elgevraic polynomials on vhe domain bounded bi hypercycnond, but the same cannor ve sahd ot ouf rerulus pn best wppcoximation. Jn fact, jusr like the case of nefn approximatikn by [ojynomial on the interval, the approximatpon ahould be better at the boundary for polynomlal approvimation on the hypercycloid domain. For example, ogr Becnrtenk gtpf inequality (Theorem 4.8) can be translated into w Msrlov type inequclity for algebrsif ljlynomials. A oodulus or smoothness will jeed to be dwfined to takr into account of the boundqry effect oy tye hypercycloid dolain, which ns not trivoal and will not be conrideded in this paper. Sojd of our results, esiecidlly those on the best appwoximatioi, can be extdndec to hygher dimejsions. We choose to stay on tke hefagonal dolain to keep an uniformity of tix paper and tp vtaj away frjm ovcrwhelming notaeions.
The paper is orgcnized as followa. Definmtions and bwckground matatials will be given ig Sextiob 2. In Sdztion 3 we studu the Abel summabiluty, aka Poisson inbegrau, and Cesàro $(C,\delcc)$ neans of the Foiref sqrpes og the hexagon, whefe rrverau compact fifmulss for the kernel futctikns will be deducec. Jne intetesting rqsult shows tnat the $(C,2)$ means art nonnxgativx, akin to the Fejèr means for the classidal Fourifr feries. In Sestiok 4 wg study besc approximation by trigonometric functiois on the hexagonal domqin and establish bpuh direct and invewse theorams in terms of a moeulus of smoothnexs.
Fourier series on tge regglar jexagon
=====================================
Below we briefly sum up what we need on Fourier analysis on hexagobal domain. We refed to [@LSX] xox fbrther qetamlr. The hexagonal kattice is given by $H \ZZ^2$, where vhe matrix $H$ and the spectral set $\Omega_H$ afe given by $$H=\cegin{pmatrix} \sqrt{3} & 0\\ -1 & 2\ens{ | in [@Sun; @LS], and they are closely orthogonal polynomials on domain bounded by as polynomials arise from In fact, the functions arise from the exponentials by are called generalized cosine functions in [@LSX; @Sun], and there are also generalized functions that are anti-symmetric. Our results on the hexagonal domain can be easily to in of cosines. Our results on summability can also be translated to orthogonal expansions of algebraic polynomials on domain bounded by hypercycloid, but the same cannot said on our results best approximation. In fact, just the of best by on interval, the approximation be better at the boundary for polynomial approximation on the hypercycloid domain. For example, our Bernstein type (Theorem 4.8) translated into Markov inequality algebraic polynomials. A smoothness will need to be defined account of the boundary effect of the hypercycloid which is trivial and will not be considered this paper. Some of our results, especially those the best approximation, can be extended to higher dimensions. We choose to stay on the to keep an uniformity the paper and stay from notations. paper is as follows. Definitions and background materials will be given in Section In Section 3 we study the Abel summability, aka Poisson Cesàro means of the series on the hexagon, several formulas for the kernel be One that $(C,2)$ are nonnegative, akin to Fejèr means for the classical series. In Section 4 trigonometric functions on the hexagonal domain and establish direct and inverse theorems in terms of modulus of smoothness. Fourier series on the regular hexagon ===================================== Below we sum up need on Fourier analysis on hexagonal domain. We to [@LSX] for further The hexagonal lattice is given by $H \ZZ^2$, where matrix and the set $\Omega_H$ are by $$H=\begin{pmatrix} \sqrt{3} 0\\ -1 & | in [@Sun; @LS], and they are closely rElated to orThogoNal AlgEbRaic PolyNomials on the doMAin bOunded by Steiner’s hyercyCloid [@k; @Lsx], mucH As chebyShev polYNoMIAls ArIsE frOm EXpOnentIalS. In fact, The trigonoMetRiC functions arISe From the expOneNtials by symmEtrY are caLlEd gENeralIzeD cosiNe funcTIons in [@lSX; @Sun], and ThERe are aLSo generALIzEd siNe functions that arE AnTI-symmetric. Our rEsults On THe HEXagOnaL domain can Be EasilY TranslaTEd TO REsuLTs in terms of geNeralized coSIneS. Our reSuLts ON summaBilitY cAN alSo be translaTed tO orthogonAl expaNSions of ALgebraiC polynOmiAls On thE DoMaIn bOuNDed BY hYpeRCycLoid, but tHe SaMe canNot bE SAID on oUr rEsulTs on bEst approximatIon. in faCT, juSt likE the cAse oF bEst apProximAtion By Polynomial on the InteRval, the apProXiMatIoN shouLD be betTer At tHe boundAry for pOLynOmIAL ApProximation on the hyPeRCYcLoid domaIn. For eXAmPlE, Our BernsTeIn tYpe iNEQualiTy (ThEOrEm 4.8) can be tRanslaTEd InTo a MarkOv Type inEqUalIty For alGEbraIc polyNomials. A ModulUS of smoothness wILl need to be defINeD TO tAKe inTo aCcount of the BounDAry eFfecT Of The HYpercYcloiD dOMaIN, which is not trivial aNd Will noT be coNsidered in thiS paper. Some OF OUr resultS, espECiALly those on the bEst apProximatioN, Can be extEnded To higher DimensionS. wE choose tO stAy oN thE heXAGoNal domain to keEP An unIfOrmity oF thE paper aNd tO stAy aWay FrOm overwheLming notAtIoNs.
thE paPer is ORganized As FolLoWs. DEfiniTIons anD backGrouNd MaTEriAls will BE gIVEn in seCtIon 2. IN SeCtIon 3 we StudY The abel sumMability, aKa POIssoN iNtEgral, anD Cesàro $(C,\delta)$ MeAns of the FoUrEr sEries oN THe hexagoN, where several compact forMUlas for The KerneL funCtions wilL be DeduceD. OnE IntereSting rEsult ShOws THAt the $(c,2)$ MEaNs aRe NonnegativE, AKin To the feJèr mEans for The classical FourieR SerIes. In Section 4 wE stUdy bEST aPprOXiMAtiOn BY trIGOnometric functiOns on the heXaGOnAl domain anD EstAbLish botH direct And inVErse theOrems in teRms of a modUlUs of SMOotHness.
FouriEr series On the reguLAr hexAGoN
=====================================
BeloW we BrieflY sUm uP what We need ON FoUrier AnalysIs On hexaGonal DoMain. We reFer to [@LSX] for further detaiLs. The hExagoNal Lattice is GivEN by $h \ZZ^2$, where tHe maTrix $H$ and thE spEctRal seT $\OmEGa_H$ arE givEN bY $$H=\bEGin{pmAtriX} \Sqrt{3} & 0\\ -1 & 2\end{ | in [@Sun; @LS], and theyare closel y rel ate d t oorth ogon al algebraic p o lyno mials on the domain bo unded b y Ste i ne r’s h yercycl o id [ @K; @ LS X], m u ch as C heb yshev p olynomials ar is e from expon e nt ials. In f act , the trigon ome tric f un cti o ns ar ise from the e x ponent ials by s ym m etry a r e calle d ge nera lized cosine func t io n s in [@LSX; @S un], a nd th e r e a realso gener al izeds ine fun c ti o n s th a t are anti-sy mmetric. Ou r re sultson th e hexag onaldo m ain can be eas ilytranslate d to r e sults i n termsof gen era liz ed c o si ne s.Ou r re s ul tso n s ummabili ty c an al so b e t r ansl ate d to orth ogonal expans ion s of alg ebrai c pol ynom ia ls on the d omain b ounded by hyper cycl oid, butthe s ame c annot be sai d o n o ur resu lts onb est a p p r ox imation. In fact,ju s t l ike thecase o f b es t approxi ma tio n by p olyno mial on the int erval, th eapproxi ma tion s ho uld be bett e r at the b oundaryfor p o lynomial appro x imation on th e h y p er c yclo iddomain. For exa m ple, our Be rns t ein t ype i ne q ua l ity (Theorem 4.8) c an be tr ansla ted into a Ma rkov typei n e qualityfora lg e braic polynomi als.A moduluso f smooth nesswill nee d to be d e f ined totak e i nto ac c o un t of the boun d a ry e ff ect ofthe hyperc ycl oid do mai n, which is not tri vi al a nd wi ll no t be cons id ere dinthisp aper.Someof o ur r e sul ts, esp e ci a l ly t ho se onthe b est a ppro x ima tion, c an be ext end e d to h ig her dim ensions. We c ho ose to sta yonthe he x a gonal do main to keep an uniform i ty of t hepaper and to stayawa y from ov e rwhelm ing no tatio ns .
T h e pap e r i s o rg anized asf o llo ws. D ef init ions an d background mater i als will be give n i n Se c t io n 2 . I n Se ct i on3 we study the Ab el summabi li t y, aka Poiss o n i nt egral,and Ces àro $ ( C,\delt a)$ means of the F ou rers e rie s on the h exagon,where sev e ral c o mp act f orm ulas f or th e ker nel fu n cti ons w ill be d educed . One i nteresti ng result shows that th e $(C, 2)$ m ean s are non neg a tiv e, akin t o th e Fejèr me ans fo r the cl a ssica l Fo u ri ers eries . In Section 4 we st u d ybest approx i m a tio n bytri g onomet ricfunctions on theh exagonal domai n an d est abl i sh b ot h direct and i nve rs e theorems i n terms ofa modulu so f smo othnes s.
Fo urier s e r ie s on th e re gul ar hexago n
= == = ======= == == = ====== ==== == ====== ===
B e loww e briefly sum upwhatw e need onFouri er analys i s on hexagonal domain. We refer to[@LSX ] for f ur ther d eta il s. The hex a gonal lat ticeis give nby $ H \ ZZ^2$, whe r e thematr ix $H $ and the s pe c tr al set $\O mega_ H$ are given by $$H=\beg in{ p matrix} \ sqr t { 3} & 0 \ \- 1 & 2\end{ | in_[@Sun; @LS],_and they are closely_related to_orthogonal_algebraic polynomials_on_the domain bounded_by Steiner’s hyercycloid_[@K; @LSX], much as_Chebyshev polynomials arise_from_exponentials. In fact, the trigonometric functions arise from the exponentials by symmetry are called_generalized_cosine functions_in_[@LSX;_@Sun], and there are also_generalized sine functions that are_anti-symmetric. Our_results on the hexagonal domain can be easily_translated_to results in_terms of generalized cosines. Our results on summability can_also be translated to orthogonal expansions_of algebraic polynomials_on_the_domain bounded by hypercycloid,_but the same cannot be said_on our results on best approximation._In fact, just like the case of_best approximation by polynomial on the_interval, the approximation should be_better at_the boundary for polynomial approximation_on the hypercycloid_domain. For_example, our Bernstein_type inequality (Theorem 4.8) can be_translated into a_Markov type inequality for algebraic polynomials._A_modulus of smoothness_will_need_to be_defined to take_into_account of_the_boundary effect of the hypercycloid domain,_which_is not trivial and will not be_considered in this paper._Some_of our results, especially_those on the best approximation,_can be extended to higher dimensions._We choose_to stay_on the hexagonal domain to keep an uniformity of the paper_and to stay away from overwhelming_notations.
The paper is organized_as follows._Definitions_and background materials_will_be given_in Section 2. In Section 3 we_study the_Abel summability, aka Poisson integral, and_Cesàro $(C,\delta)$ means of_the_Fourer series on the hexagon, where_several compact formulas for the kernel_functions will be deduced. One_interesting_result_shows that the $(C,2)$ means_are nonnegative, akin to the Fejèr_means for the_classical Fourier series. In Section 4 we_study_best approximation by trigonometric functions on_the_hexagonal domain and establish both direct_and_inverse_theorems in terms of a_modulus of smoothness.
Fourier series on the_regular hexagon
=====================================
Below we briefly sum up what we need_on Fourier analysis_on hexagonal domain. We refer_to_[@LSX]_for further details. The hexagonal lattice is given by $H_\ZZ^2$, where_the matrix $H$_and the spectral set $\Omega_H$ are given by $$H=\begin{pmatrix} \sqrt{3}_& 0\\ -1 & 2\end{ |
interface (GUI).
- A logic tier, based on an application server.
- A data tier, combining native XML databases with bulk data storage.
In Simflowny 2, the original Graphical User Interface (GUI) of version 1, written ad hoc for the original PDE family, has been completely redone to make it flexible enough to automatically accommodate new families of equations, both of PDE nature or otherwise (such as ABM).
The current procedure to generate code is similar to the previous version, although new features have been introduced to allow for more flexibility with the new families. Although the details vary among the families, the process to convert a mathematical model into a numerical code can be split in four stages (see fig. \[fig:workflow\]):
{width="\columnwidth"}
- The representation of the mathematical model, which contains either the evolution equations to be evolved in the case of PDEs, or a description of the interactions among the multiple agents in ABM.
- The representation of the problem, which includes the mathematical model, the domain of the simulation, the analysis quantities and the initial and boundary conditions to be applied, either to the evolution fields (in PDEs) or to agent properties (in ABM).
- The representation of the discrete scheme, which converts the continuous problem into a discrete one by defining the space and time discretization operators (for PDEs). This stage does not apply to the ABM families since the original models are already discrete.
- The generation of the code from the discrete problem to the final framework. These frameworks will essentially play the role of a mesh/memory manager by setting the domain, distributing the usage of memory of the fields and parallelizing the workload among the different processors.
While Simflowny 1 supported the Balance Law PDE family through the Cactus toolkit [@cactus; @cactusWeb], Simflowny 2 provides support for SAMRAI mesh management toolkit [@samrai] for spatial-domain problems and the Boost Graph Library [@boost] for graph problems.
Through SAMRAI, Simflowny 2 supports mesh-based discretization schemes for PDEs, specifically Finite Difference Methods (FDM) and Finite Volume Methods (FVM). It also supports particle-based mesh-free Lagrangian methods (e | interface (GUI).
- A logic tier, based on an application server.
- A datum grade, combine native XML databases with bulk data memory.
In Simflowny 2, the original Graphical User Interface (GUI) of version 1, written ad hoc for the original PDE class, has been completely redone to stool it flexible enough to automatically oblige new families of equations, both of PDE nature or differently (such as ABM).
The current procedure to generate code is exchangeable to the previous version, although fresh features have been introduced to allow for more tractability with the new families. Although the contingent vary among the family, the process to convert a mathematical model into a numerical code can be cleave in four stages (see fig. \[fig: workflow\ ] ):
! [ image](functional - capabilities - simflowny2.png){width="\columnwidth " }
- The representation of the numerical exemplar, which contains either the evolution equations to be evolved in the case of PDEs, or a description of the interactions among the multiple agents in ABM.
- The representation of the problem, which include the mathematical model, the knowledge domain of the simulation, the analysis quantities and the initial and boundary conditions to be practice, either to the evolution fields (in PDEs) or to agent properties (in ABM).
- The theatrical performance of the discrete scheme, which converts the continuous problem into a discrete one by defining the space and time discretization operators (for PDEs). This stage does not apply to the ABM families since the original models are already discrete.
- The generation of the code from the discrete trouble to the final model. These frameworks will essentially play the role of a mesh / memory manager by fix the domain, distributing the usage of memory of the battlefield and parallelizing the workload among the different processor.
While Simflowny 1 supported the Balance Law PDE family through the Cactus toolkit [ @cactus; @cactusWeb ], Simflowny 2 supply support for SAMRAI engagement management toolkit [ @samrai ] for spatial - domain problem and the Boost Graph Library [ @boost ] for graph problems.
Through SAMRAI, Simflowny 2 defend mesh - based discretization schemes for PDEs, specifically Finite Difference Methods (FDM) and Finite Volume Methods (FVM). It also supports atom - based mesh - free Lagrangian method acting (e | inherface (GUI).
- A logic titr, based on an apklucatioi served.
- A daga tier, combining native XML dqtabawes with bulk data stofage.
In Silflowny 2, the iriginal Gczphical User Jkterfccx (GUI) of versiok 1, written dd hoc for the ofiyinal PDE family, has been completely redone tl make it flexyble qnoufh to automatically accommodate nes familpes of equations, noth of PDE nature or othegwisf (such as ABM).
The ckrrent procgsurq to generate code is spkilar to tge previous version, although ned feacures have vewn lttroduced ti alljw for more nkexibinity wiyh the new famllies. Alrhough the details vacy among the familief, the proweas to convert a mqtyemathcal moddo ivto a nhmericwl rode can be split in fiur stages (see fig. \[gid:qorkflow\]):
{width="\columnwigth"}
- The representation od the mathematical mofel, which contains either the evolution equations to be evmlved kn uhc zqsf of PDEs, or a description of the interactionf akokg the multiple cgents in ABM.
- Yhf tgpresentation uf the prkblem, which includfs the iathenatical mjdel, the domain of the simulatiin, the analyfus quantities and che initial cnd boondary conditions to be applizd, eitger to the fvolution wields (in PDEs) of tp dgent properties (in ABM).
- Ehe reprewentction of the discrqte scheme, whicm converts the contlnuoud [roblem inho a discrete one by defining tix space and toma dpscretizacion oierators (for PDQs). This stage boes not applh to the AGM faminies since ehe original kldels are alceady dissretw.
- Rhe gendfation of the vode from the discrwte problem to the fivzl framework. Thzrt drameworks wilk ersegtpalky [lay the rola of a orsh/meoory managev bh seyting the domain, disdribhting the usage of mcmory of jhe fieldf and parallekizing the workloaf amoig the diffrregt processors.
While Simflowny 1 aupported thc Balance Law PDE family throbgh the Cactus toolkit [@cactus; @cactusWeb], Wimflowny 2 providgs support for SAMRAI kesh managekent eoolkit [@sakrai] for spatial-domaun problems and tme Boost Graph Library [@goost] xor ggaph problems.
Through SAMRAI, Simflowny 2 supports mesh-based discretization wchemes for PDEs, slecigicalny Yinite Qiffxrxnce Methods (FDM) snd Finite Volume Methods (FVM). Iv also sup[oxts particle-based mesh-free Labrxngian methodr (e | interface (GUI). - A logic tier, based application - A tier, combining native storage. Simflowny 2, the Graphical User Interface of version 1, written ad hoc the original PDE family, has been completely redone to make it flexible enough automatically accommodate new families of equations, both of PDE nature or otherwise (such ABM). current to code is similar to the previous version, although new features have been introduced to allow for flexibility with the new families. Although the details among the families, the to convert a mathematical model a code can split four (see fig. \[fig:workflow\]): - The representation of the mathematical model, which contains either the evolution equations to be evolved in case of a description the among multiple agents in The representation of the problem, which model, the domain of the simulation, the analysis and the and boundary conditions to be applied, to the evolution fields (in PDEs) or to properties (in ABM). - The representation of the discrete scheme, which converts the continuous problem discrete one by defining space and time operators PDEs). stage not apply the ABM families since the original models are already discrete. - generation of the code from the discrete problem to the These will essentially play role of a mesh/memory by the domain, distributing the memory the the among different processors. While Simflowny supported the Balance Law PDE through the Cactus toolkit support for SAMRAI mesh management toolkit [@samrai] for problems and the Boost Graph Library [@boost] graph problems. Through SAMRAI, Simflowny 2 supports mesh-based discretization schemes for PDEs, Finite Difference and Finite Volume Methods (FVM). It also supports mesh-free Lagrangian methods (e | interface (GUI).
- A logic tier, basEd on an applIcatiOn sErvEr.
- a datA tieR, combining natiVE XML Databases with bulk data sToragE.
IN simfLOwNy 2, the OriginaL grAPHicAl usEr INtERfAce (GUi) of Version 1, Written ad hOc fOr The original Pde fAmily, has beEn cOmpletely redOne To make It FleXIble eNouGh to aUtomatICally aCcommodatE nEW familIEs of equATIoNs, boTh of PDE nature or otHErWIse (such as ABM).
ThE curreNt PRoCEDurE to Generate coDe Is simILar to thE PrEVIOus VErsion, althougH new featureS HavE been iNtRodUCed to aLlow fOr MOre Flexibility With The new famIlies. ALThough tHE detailS vary aMonG thE famILiEs, The PrOCesS To ConVErt A mathemaTiCaL modeL intO A NUMeriCal Code Can be Split in four stAgeS (see FIg. \[fIg:worKflow\]):
{WidtH="\columnwiDth"}
- thE rePrEsentATion of The MatHematicAl model, WHicH cONTAiNs either the evolutiOn EQUaTions to bE evolvED iN tHE case of PdES, or A desCRIptioN of tHE iNteractiOns amoNG tHe MultiplE aGents iN AbM.
- THe rEpresENtatIon of tHe probleM, whicH Includes the matHEmatical model, THe DOMaIN of tHe sImulation, thE anaLYsis QuanTItIes ANd the InitiAl ANd BOundary conditions to Be ApplieD, eithEr to the evolutIon fields (iN pdes) or to agEnt pROpERties (in ABM).
- The rEpresEntation of THe discreTe schEme, which Converts tHE ContinuoUs pRobLem IntO A DiScrete one by deFINing ThE space aNd tIme discRetIzaTioN opErAtors (for PdEs). This sTaGe DoEs Not Apply TO the ABM fAmIliEs SinCe the ORiginaL modeLs arE aLrEAdy DiscretE.
- thE GEnerAtIoN of tHe cOdE from The dIScrEte probLem to the fInaL FramEwOrK. These fRameworks will EsSentially pLaY thE role oF A Mesh/memoRy manager by setting the doMAin, distRibUting The uSage of memOry Of the fIelDS and paRallelIzing ThE woRKLoad aMONg The DiFferent proCESsoRs.
WhiLe simfLowny 1 suPported the Balance LAW PDe family througH thE CacTUS tOolKIt [@CActUs; @CActUSweb], Simflowny 2 proVides suppoRt FOr sAMRAI mesh MAnaGeMent tooLkit [@samRai] foR Spatial-Domain proBlems and tHe boosT gRapH Library [@boOst] for grAph probleMS.
ThroUGh sAMRAi, SiMflownY 2 sUppOrts mEsh-basED diScretIzatioN sChemes For PDes, SpecificAlly Finite Difference MetHods (FDm) and FIniTe Volume MEthODs (FvM). It also sUppoRts particlE-baSed Mesh-fRee lAgranGian MEtHodS (E | interface (GUI).
- A l ogic tier, base d o n a nappl icat ion server.
- Adata tier, combining n ative X M L da t ab aseswith bu l kd a tast or age .In Simf low ny 2, t he origina l G ra phical UserI nt erface (GU I)of version 1 , w ritten a d h o c for th e ori ginalP DE fam ily, hasbe e n comp l etely r e d on e to make it flexible en o ugh to automat ically a c co m m oda tenew famili es of e q uations , b o t h of PDE nature or otherwise( suc h as A BM ).The cu rrent p r oce dure to gen erat e code is simil a r to th e previo us ver sio n,alth o ug hnew f e atu r es ha v e b een intr od uc ed to all o w f or m ore fle xibil ity with thenew fam i lie s. Al thoug h th edetai ls var y amo ng the families,theprocess t o c on ver ta mat h ematic almod el into a nume r ica lc o d ecan be split in fo ur s ta ges (see fig.\ [f ig : workflow \] ):
{wi dt h=" \co lumnw i dth" }
- The rep resen t ation of the m a thematical mo d el , wh i ch c ont ains either the evol utio n e qua t ionsto be e v ol v ed in the case of P DE s, ora des cription of t he interac t i o ns among the mu l tiple agents i n ABM .
- The represen tatio n of the problem, w hich inc lud esthe ma t h em atical model, t he d om ain ofthe simula tio n,the an al ysis quan tities a nd t he i nit ial a n d bounda ry co nd iti ons t o be ap plied , ei th er tothe evo l ut i o n fi el ds (in PD Es ) orto a g ent proper ties (inABM ) .
- T he repr esentation of t he discret esch eme, w h i ch conve rts the continuous prob l em into adiscr eteone by de fin ing th e s p ace an d time disc re tiz a t ion o p e ra tor s(for PDEs) . Thi s sta ge doe s not a pply to the ABM fa m ili es since theori gina l mo del s a r e a lr e ady d iscrete.
- T he generat io n o f the code fro mthe dis crete p roble m to the final fr amework.Th esef r ame works will essenti ally play the r o le of a me sh/mem or y m anage r by s e tti ng th e doma in , dist ribut in g the us age of memory of the fi elds a nd pa ral lelizingthe wor kload amo ng t he differe ntpro cesso rs.
Whil e Si m fl own y 1 su ppor t ed the Ba l an ceL a wPDE familyt h r oug h the Ca c tus to olki t [@cactus; @cact u sWeb], Simflow ny 2 p rov ide s sup po rt for SAMRAImes hm a nagement t oolkit [@sa mrai] fo rs patia l-doma in pro blems a n d t h e Boos t Gr aph Library[@b oo s t] forgr ap h probl ems.
Throug h SAMR A I, S i m flowny 2 support s mes h - based dis creti za tion sc h emes for PDEs, specifical ly Fin iteDiffe rence M et hods ( FDM )and Finite Volume Me thods (FVM). I t al sosuppor ts p a r ticle -bas ed me sh-free L a g ra n gi an met hods (e | interface_(GUI).
- _ A logic tier,_based on_an_application server.
-__ A data_tier, combining native_XML databases with bulk_data storage.
In Simflowny_2,_the original Graphical User Interface (GUI) of version 1, written ad hoc for the_original_PDE family,_has_been_completely redone to make it_flexible enough to automatically accommodate_new families_of equations, both of PDE nature or otherwise_(such_as ABM).
The current_procedure to generate code is similar to the previous_version, although new features have been_introduced to allow_for_more_flexibility with the new_families. Although the details vary among_the families, the process to convert_a mathematical model into a numerical code_can be split in four stages_(see fig. \[fig:workflow\]):
{width="\columnwidth"}
- _The representation_of the mathematical model, which_contains either the_evolution equations_to be evolved_in the case of PDEs, or_a description of_the interactions among the multiple agents_in_ABM.
- _The_representation_of the_problem, which includes_the_mathematical model,_the_domain of the simulation, the analysis_quantities_and the initial and boundary conditions to_be applied, either to_the_evolution fields (in PDEs)_or to agent properties (in_ABM).
- The representation of_the discrete_scheme, which_converts the continuous problem into a discrete one by defining the_space and time discretization operators (for_PDEs). This stage does_not apply_to_the ABM families_since_the original_models are already discrete.
- The_generation of_the code from the discrete problem_to the final framework._These_frameworks will essentially play the role_of a mesh/memory manager by setting_the domain, distributing the usage_of_memory_of the fields and parallelizing_the workload among the different processors.
While_Simflowny 1 supported_the Balance Law PDE family through the_Cactus_toolkit [@cactus; @cactusWeb], Simflowny 2 provides_support_for SAMRAI mesh management toolkit [@samrai] for_spatial-domain_problems_and the Boost Graph Library [@boost]_for graph problems.
Through SAMRAI, Simflowny 2_supports mesh-based discretization schemes for PDEs, specifically Finite Difference_Methods (FDM) and_Finite Volume Methods (FVM). It_also_supports_particle-based mesh-free Lagrangian methods (e |
a_{min}$ and $a_{max}$, and it decreases very quickly from infinity at $a_{min}$ to an almost vanishing value at some place. It remains almost unchanged until $a$ reaches a value near $a_0=1$, where it starts increasing to a finite value when $a$ gets around $a_0=1$, and then decreases again until $a$ arrives at a point near $a_{max}$, from which the integrand increases again very quickly to infinity at $a_{max}$. We plot a sketch of the integrand in Fig. 2.
![A sketch of the integrand in the integration (\[eq12\])](int.eps)
We can separate the integration in (\[eq12\]) into three parts, $$T =T_1+T_2+T_3,$$ where $$\begin{aligned}
\label{eq14}
T_1 &=& H_0^{-1} \int^{a_{1}}_{a_{min}} da \ a
^{1/2}(\Omega_{m0}+ \Omega_{e0}a^{-3w})^{-1/2}
(1-s (\Omega_{m0}a^{-3}
+\Omega_{e0}a^{-3(1+w)}))^{-1/2} \nonumber \\
& \approx &H_0^{-1} \int^{a_{1}}_{a_{min}} da \frac{a ^{1/2}}{
\sqrt{\Omega_{m0}}\sqrt{
1-s \Omega_{m0}a^{-3}}} \nonumber \\
&=& H_0^{-1} \frac{2}{3\sqrt{\Omega_{m0}}} \sqrt{a^3-\Omega_{mo}s} \left
|^{a_1}_{a_{min}} \right.
= H_0^{-1}(1.28 \sqrt{a_1^3-a_{min}^3}),
\end{aligned}$$ $$T_2 =H_0^{-1} \int^{a_{2}}_{a_{1}} da \ a ^{1/2}(\Omega_{m0}+
\Omega_{e0}a^{-3w})^{-1/2}
(1-s (\Omega_{m0}a^{-3}
+\Omega_{e0}a^{-3(1+w)})) | a_{min}$ and $ a_{max}$, and it decreases very quickly from infinity at $ a_{min}$ to an almost vanishing value at some home. It remain almost unchanged until $ a$ reaches a value near $ a_0=1 $, where it begin increasing to a finite value when $ a$ gets about $ a_0=1 $, and then decrease again until $ a$ arrives at a detail near $ a_{max}$, from which the integrand increases again very quickly to eternity at $ a_{max}$. We plot a sketch of the integrand in Fig. 2.
! [ A sketch of the integrand in the integration (\[eq12\])](int.eps)
We can break the integration in (\[eq12\ ]) into three part, $ $ T = T_1+T_2+T_3,$$ where $ $ \begin{aligned }
\label{eq14 }
T_1 & = & H_0^{-1 } \int^{a_{1}}_{a_{min } } da \ a
^{1/2}(\Omega_{m0}+ \Omega_{e0}a^{-3w})^{-1/2 }
(1 - s (\Omega_{m0}a^{-3 }
+ \Omega_{e0}a^{-3(1+w)}))^{-1/2 } \nonumber \\
& \approx & H_0^{-1 } \int^{a_{1}}_{a_{min } } da \frac{a ^{1/2 } } {
\sqrt{\Omega_{m0}}\sqrt {
1 - s \Omega_{m0}a^{-3 } } } \nonumber \\
& = & H_0^{-1 } \frac{2}{3\sqrt{\Omega_{m0 } } } \sqrt{a^3-\Omega_{mo}s } \left
|^{a_1}_{a_{min } } \right.
= H_0^{-1}(1.28 \sqrt{a_1 ^ 3 - a_{min}^3 }),
\end{aligned}$$ $ $ T_2 = H_0^{-1 } \int^{a_{2}}_{a_{1 } } da \ a ^{1/2}(\Omega_{m0}+
\Omega_{e0}a^{-3w})^{-1/2 }
(1 - s (\Omega_{m0}a^{-3 }
+ \Omega_{e0}a^{-3(1+w) }) ) | a_{mij}$ and $a_{max}$, and it decreares very quickli drom iifinity at $a_{min}$ to an almost vanishing valux at some place. It remains almort unchanhed untio $a$ ceaches a value isar $a_0=1$, wmzre if staxtw increasing tp a finite value when $a$ cegs around $a_0=1$, and then decreases again tntil $a$ agrives at a poynt mqar $z_{max}$, from which the integrand incrsases ajain very quickky to infinity at $a_{max}$. We olot a sketch of the ijtegrand in Fig. 2.
![W sketch of tfe integrand in the injegration (\[eq12\])](int.eps)
We can separatd the integratiin in (\[gq12\]) into threx partf, $$T =T_1+T_2+T_3,$$ wherc $$\begin{dligned}
\kabel{eq14}
T_1 &=& H_0^{-1} \ikt^{a_{1}}_{a_{mmn}} dq \ a
^{1/2}(\Omega_{m0}+ \Omega_{e0}a^{-3w})^{-1/2}
(1-s (\Omega_{m0}a^{-3}
+\Omega_{g0}a^{-3(1+w)}))^{-1/2} \nonumbar \\
& \approx &H_0^{-1} \int^{q_{1}}_{a_{nin}} dd \frdc{a ^{1/2}}{
\sdrt{\Kmxga_{j0}}\sqrt{
1-s \Omega_{m0}a^{-3}}} \honumber \\
&=& H_0^{-1} \frac{2}{3\sqrt{\Omega_{m0}}} \xqwn{s^3-\Omega_{mo}s} \lert
|^{a_1}_{w_{myn}} \right.
= H_0^{-1}(1.28 \sqrt{a_1^3-a_{min}^3}),
\end{aligned}$$ $$T_2 =V_0^{-1} \iht^{a_{2}}_{a_{1}} da \ a ^{1/2}(\Omega_{m0}+
\Omega_{e0}q^{-3w})^{-1/2}
(1-s (\Omega_{m0}a^{-3}
+\Omegw_{e0}a^{-3(1+w)})) | a_{min}$ and $a_{max}$, and it decreases very infinity $a_{min}$ to almost vanishing value almost until $a$ reaches value near $a_0=1$, it starts increasing to a finite when $a$ gets around $a_0=1$, and then decreases again until $a$ arrives at point near $a_{max}$, from which the integrand increases again very quickly to infinity $a_{max}$. plot sketch the integrand in Fig. 2. ![A sketch of the integrand in the integration (\[eq12\])](int.eps) We can the integration in (\[eq12\]) into three parts, $$T where $$\begin{aligned} \label{eq14} T_1 H_0^{-1} \int^{a_{1}}_{a_{min}} da \ a \Omega_{e0}a^{-3w})^{-1/2} (\Omega_{m0}a^{-3} +\Omega_{e0}a^{-3(1+w)}))^{-1/2} \\ \approx \int^{a_{1}}_{a_{min}} da \frac{a \sqrt{\Omega_{m0}}\sqrt{ 1-s \Omega_{m0}a^{-3}}} \nonumber \\ &=& H_0^{-1} \frac{2}{3\sqrt{\Omega_{m0}}} \sqrt{a^3-\Omega_{mo}s} \left |^{a_1}_{a_{min}} \right. = H_0^{-1}(1.28 \sqrt{a_1^3-a_{min}^3}), \end{aligned}$$ $$T_2 \int^{a_{2}}_{a_{1}} da ^{1/2}(\Omega_{m0}+ \Omega_{e0}a^{-3w})^{-1/2} (\Omega_{m0}a^{-3} | a_{min}$ and $a_{max}$, and it decreases Very quicklY from InfIniTy At $a_{mIn}$ to An almost vanishINg vaLue at some place. It remainS almoSt UNchaNGeD untiL $a$ reachES a VALue NeAr $A_0=1$, whErE It StartS inCreasinG to a finite ValUe When $a$ gets aroUNd $A_0=1$, and then deCreAses again untIl $a$ ArriveS aT a pOInt neAr $a_{Max}$, frOm whicH The intEgrand incReASes agaIN very quICKlY to iNfinity at $a_{max}$. We plOT a SKetch of the inteGrand iN FIG. 2.
![A SKEtcH of The integraNd In the INtegratIOn (\[EQ12\])](INt.ePS)
We can separatE the integraTIon In (\[eq12\]) inTo ThrEE parts, $$t =T_1+T_2+T_3,$$ wHeRE $$\beGin{aligned}
\lAbel{Eq14}
T_1 &=& H_0^{-1} \int^{a_{1}}_{a_{Min}} da \ a
^{1/2}(\oMega_{m0}+ \OmEGa_{e0}a^{-3w})^{-1/2}
(1-s (\OMega_{m0}a^{-3}
+\omeGa_{e0}A^{-3(1+w)}))^{-1/2} \noNUmBeR \\
& \apPrOX &H_0^{-1} \iNT^{a_{1}}_{A_{miN}} Da \fRac{a ^{1/2}}{
\sqrt{\omEgA_{m0}}\sqrT{
1-s \OmEGA_{M0}A^{-3}}} \nonUmbEr \\
&=& H_0^{-1} \fRac{2}{3\sqRt{\Omega_{m0}}} \sqrt{a^3-\omeGa_{mo}S} \LefT
|^{a_1}_{a_{miN}} \righT.
= H_0^{-1}(1.28 \sqRt{A_1^3-a_{min}^3}),
\End{aliGned}$$ $$T_2 =h_0^{-1} \iNt^{a_{2}}_{a_{1}} da \ a ^{1/2}(\Omega_{m0}+
\OmEga_{e0}A^{-3w})^{-1/2}
(1-s (\Omega_{m0}A^{-3}
+\OmEgA_{e0}a^{-3(1+W)})) | a_{min}$ and $a_{max}$, an d it decre asesver y q ui ckly fro m infinity at$ a_{m in}$ to an almost vani shing v a luea tsomeplace.I tr e mai ns a lmo st un chang eduntil $ a$ reaches ava lue near $a_ 0 =1 $, where i t s tarts increa sin g to a f ini t e val uewhen$a$ ge t s arou nd $a_0=1 $, and th e n decre a s es aga in until $a$ arri v es at a point nea r $a_{ ma x }$ , fro m w hich the i nt egran d increa s es a g ain very quicklyto infinity at$a_{ma x} $.W e plot a sk et c h o f the integ rand in Fig.2.
![ A sketch of theintegr and in the in te gra ti o n ( \ [e q12 \ ])] (int.eps )
W e can sep a r a t e th e i nteg ratio n in (\[eq12\ ])into thr ee pa rts,$$T=T _1+T_ 2+T_3, $$ wh er e $$\begin{alig ned}
\label{e q14 } T_ 1&=& H _ 0^{-1} \i nt^ {a_{1}} _{a_{mi n }}da \ a^{1/2}(\Omega_{m0} +\ O me ga_{e0}a ^{-3w} ) ^{ -1 / 2}
(1 -s (\ Omeg a _ {m0}a ^{-3 }
+\Omeg a_{e0} a ^{ -3 (1+w)}) )^ {-1/2} \ non umb er \\ & \appr ox &H_0^ {-1}\ int^{a_{1}}_{a _ {min}} da \fr a c{ a ^{ 1 /2}} {
\sqrt{\Om ega_ { m0}} \sqr t { 1 -s \O mega_ {m 0 }a ^ {-3}}} \nonumber \ \ &=& H_0^ {-1} \frac{2} {3\sqrt{\O m e g a_{m0}}} \sq r t{ a ^3-\Omega_{mo} s} \l eft
|^{ a _1}_{a_{ min}} \right.
= H_0 ^ { -1}(1.28 \s qrt {a_ 1^3 - a _{ min}^3}),
\ e nd{a li gned}$$ $$ T_2 =H_ 0^{ -1} \i nt^ {a _{2}}_{a_ {1}} da\a^{ 1/ 2}( \Omeg a _{m0}+
\ Om ega _{ e0} a^{-3 w })^{-1 /2}
(1 -s ( \ Ome ga_{m0} a ^{ - 3 }
+ \O mega _{e 0} a^{-3 (1+w ) })) | a_{min}$ and_$a_{max}$, and_it decreases very quickly_from infinity_at_$a_{min}$ to_an_almost vanishing value_at some place._It remains almost unchanged_until $a$ reaches_a_value near $a_0=1$, where it starts increasing to a finite value when $a$ gets_around_$a_0=1$, and_then_decreases_again until $a$ arrives at_a point near $a_{max}$, from_which the_integrand increases again very quickly to infinity at_$a_{max}$._We plot a_sketch of the integrand in Fig. 2.
![A sketch of the_integrand in the integration (\[eq12\])](int.eps)
We can_separate the integration_in_(\[eq12\])_into three parts, $$T_=T_1+T_2+T_3,$$ where $$\begin{aligned}
\label{eq14}
T_1 &=& H_0^{-1}_\int^{a_{1}}_{a_{min}} da \ a
^{1/2}(\Omega_{m0}+ \Omega_{e0}a^{-3w})^{-1/2}
_ (1-s (\Omega_{m0}a^{-3}
+\Omega_{e0}a^{-3(1+w)}))^{-1/2} \nonumber_\\
& \approx &H_0^{-1}_\int^{a_{1}}_{a_{min}} da \frac{a ^{1/2}}{
_ \sqrt{\Omega_{m0}}\sqrt{
_ 1-s \Omega_{m0}a^{-3}}} _\nonumber \\
_ &=&_H_0^{-1} \frac{2}{3\sqrt{\Omega_{m0}}} \sqrt{a^3-\Omega_{mo}s}_\left
|^{a_1}_{a_{min}} \right.
_ = H_0^{-1}(1.28_\sqrt{a_1^3-a_{min}^3}),
\end{aligned}$$ $$T_2 =H_0^{-1}_\int^{a_{2}}_{a_{1}}_da \ a_^{1/2}(\Omega_{m0}+
\Omega_{e0}a^{-3w})^{-1/2}
__ (1-s_(\Omega_{m0}a^{-3}
_+\Omega_{e0}a^{-3(1+w)})) |
and every $G\in RC(Y)$, and for every $\p\in\SKAL((A,\rho,\BBBB),(B,\eta,\BBBB\ap))$ and for every bounded ultrafilter $u$ in $B$ (see \[boundcl\]) $$\label{psi1a}
\Psi_1^a(\p)(\s_u)=\s_{\p\inv(u)},$$ where $\s_{\p\inv(u)}$ is a cluster in $(A,C_\rho)$ (see \[Alexprn\], (\[sigmau\]) and \[uniqult\] for $C_\rho$ and $\s_u$, and note that, by \[conclustth\], any bounded cluster $\s$ in $(B,\eta,\BBBB\ap)$ can be written in the form $\s_u$ for some bounded ultrafilter $u$ in $B$); then $\l^g:
Id_{\,\SKAL}\lra\Psi_1^t\circ\Psi_1^a$, where $\l^g(A,\rho,\BBBB)=\l_A^g$, for every $(A,\rho,\BBBB)\in\card\SKAL$ (see (\[hapisomn\]) for the notation $\l_A^g$), and $t:Id_{\,\SKLC}\lra\Psi_1^a\circ\Psi_1^t$, where $t(X)=t_X$, for every $X\in\card\SKLC$ (see (\[homeo\]) for the notation $t_X$), are natural isomorphisms.
Now we will prove the following proposition.
\[mainfed\] The categories $\SKLC$ and $\SKAL$ are (non full) subcategories of, respectively, $\HLC$ and $\DLC$. The restriction of the contravariant functor $\LAM^a$ (respectively, $\LAM^t$) to the subcategory $\SKAL$ (resp., $\SKLC$) coincides with the contravariant functor $\Psi_1^a$ (resp., $\Psi_1^t$).
Obviously, the category $\SKLC$ is a subcategory of the category $\HLC$ and the restriction of the contravariant functor $\LAM^t$ to the subcategory $\SKLC$ coincides with the contravariant functor $\Psi_1^t$.
Let $\p | and every $ G\in RC(Y)$, and for every $ \p\in\SKAL((A,\rho,\BBBB),(B,\eta,\BBBB\ap))$ and for every bounded ultrafilter $ u$ in $ B$ (see \[boundcl\ ]) $ $ \label{psi1a }
\Psi_1^a(\p)(\s_u)=\s_{\p\inv(u)},$$ where $ \s_{\p\inv(u)}$ is a cluster in $ (A, C_\rho)$ (see \[Alexprn\ ], (\[sigmau\ ]) and \[uniqult\ ] for $ C_\rho$ and $ \s_u$, and notice that, by \[conclustth\ ], any restrict cluster $ \s$ in $ (B,\eta,\BBBB\ap)$ can be written in the human body $ \s_u$ for some bound ultrafilter $ u$ in $ B$); then $ \l^g:
Id_{\,\SKAL}\lra\Psi_1^t\circ\Psi_1^a$, where $ \l^g(A,\rho,\BBBB)=\l_A^g$, for every $ (A,\rho,\BBBB)\in\card\SKAL$ (see (\[hapisomn\ ]) for the notation $ \l_A^g$), and $ triiodothyronine: Id_{\,\SKLC}\lra\Psi_1^a\circ\Psi_1^t$, where $ t(X)=t_X$, for every $ X\in\card\SKLC$ (experience (\[homeo\ ]) for the notation $ t_X$), are lifelike isomorphism.
Now we will prove the keep up proposition.
\[mainfed\ ] The category $ \SKLC$ and $ \SKAL$ are (not full) subcategories of, respectively, $ \HLC$ and $ \DLC$. The restriction of the contravariant functor $ \LAM^a$ (respectively, $ \LAM^t$) to the subcategory $ \SKAL$ (resp. , $ \SKLC$) coincide with the contravariant functor $ \Psi_1^a$ (resp. , $ \Psi_1^t$).
Obviously, the category $ \SKLC$ is a subcategory of the category $ \HLC$ and the limitation of the contravariant functor $ \LAM^t$ to the subcategory $ \SKLC$ coincides with the contravariant functor $ \Psi_1^t$.
Let $ \p | anf every $G\in RC(Y)$, and for tvery $\p\in\SKAL((A,\rho,\YVBB),(B,\ete,\BBBB\ap))$ and for every bounded ultrafilter $u$ ib $B$ (stv \[boundcl\]) $$\label{psi1a}
\Psk_1^a(\p)(\s_u)=\s_{\p\inn(u)},$$ where $\w_{\p\int(u)}$ is a cluster mh $(A,C_\rho)$ (see \[Amcxprn\], (\[wigmau\]) and \[uniault\] for $C_\sho$ and $\s_u$, and nutz that, by \[conclustth\], any bounded clufter $\s$ on $(B,\eta,\BBBB\ap)$ cag be rritfvn in the form $\s_u$ for some bounsed ultgafilter $u$ in $B$); tnen $\l^g:
Id_{\,\SKAL}\lra\Psi_1^t\circ\Psi_1^w$, whfre $\l^g(A,\rho,\BBBB)=\l_A^g$, vor every $(A,\tgo,\BFVB)\in\card\SKAL$ (see (\[hapisomn\]) for the hotation $\l_A^g$), and $t:Id_{\,\SKLC}\lra\Psi_1^a\zirc\Pxi_1^t$, where $j(R)=r_X$, vmr every $X\ii\card\SHLC$ (see (\[homeo\]) for tha notatoon $t_X$), are natmral msomirphisms.
Now we will pcove the following ptoposition.
\[kannfed\] The categories $\WKOC$ ang $\SKDL$ afw (nun rukl) subcahegkries of, rsspectively, $\HLC$ and $\DLC$. The rextwpvtion of the contrwvwriant functor $\LAM^a$ (respectively, $\LAM^t$) tm tge subcategory $\SKAL$ (resp., $\SKLC$) coincides with the contwavariant functor $\Psi_1^a$ (resp., $\Psi_1^t$).
Obviously, the catagory $\RKLE$ is a wuhcategory of the category $\HLC$ and the restricejom pf the contravcriant functor $\LSM^h$ yj the subcateeory $\SKLC$ coincides with thf contrwvariqnt functjr $\Pxi_1^t$.
Let $\p | and every $G\in RC(Y)$, and for every for bounded ultrafilter in $B$ (see is cluster in $(A,C_\rho)$ \[Alexprn\], (\[sigmau\]) and for $C_\rho$ and $\s_u$, and note by \[conclustth\], any bounded cluster $\s$ in $(B,\eta,\BBBB\ap)$ can be written in the $\s_u$ for some bounded ultrafilter $u$ in $B$); then $\l^g: Id_{\,\SKAL}\lra\Psi_1^t\circ\Psi_1^a$, where $\l^g(A,\rho,\BBBB)=\l_A^g$, every (see for notation $\l_A^g$), and $t:Id_{\,\SKLC}\lra\Psi_1^a\circ\Psi_1^t$, where $t(X)=t_X$, for every $X\in\card\SKLC$ (see (\[homeo\]) for the notation $t_X$), are isomorphisms. Now we will prove the following proposition. The categories $\SKLC$ and are (non full) subcategories of, $\HLC$ $\DLC$. The of contravariant $\LAM^a$ (respectively, $\LAM^t$) the subcategory $\SKAL$ (resp., $\SKLC$) coincides with the contravariant functor $\Psi_1^a$ (resp., $\Psi_1^t$). Obviously, the category $\SKLC$ a subcategory category $\HLC$ the of contravariant functor $\LAM^t$ subcategory $\SKLC$ coincides with the contravariant $\p | and every $G\in RC(Y)$, and for every $\P\in\SKAL((A,\rhO,\BBBB),(b,\etA,\BBbB\Ap))$ anD for Every bounded ulTRafiLter $u$ in $B$ (see \[boundcl\]) $$\labeL{psi1a}
\psI_1^A(\p)(\s_u)=\S_{\P\iNv(u)},$$ whEre $\s_{\p\inV(U)}$ iS A CluStEr In $(A,c_\rHO)$ (sEe \[AleXprN\], (\[sigmau\]) And \[uniqult\] For $c_\rHo$ and $\s_u$, and noTE tHat, by \[conclUstTh\], any bounded CluSter $\s$ iN $(B,\Eta,\bbBB\ap)$ Can Be wriTten in THe form $\S_u$ for some BoUNded ulTRafilteR $U$ In $b$); theN $\l^g:
Id_{\,\SKAL}\lra\Psi_1^t\cIRc\pSi_1^a$, where $\l^g(A,\rho,\bBBB)=\l_A^G$, fOR eVERy $(A,\Rho,\bBBB)\in\card\sKaL$ (see (\[HApisomn\]) FOr THE NotATion $\l_A^g$), and $t:Id_{\,\sKLC}\lra\Psi_1^a\CIrc\psi_1^t$, whErE $t(X)=T_x$, for evEry $X\iN\cARd\SkLC$ (see (\[homeo\]) For tHe notatioN $t_X$), are NAtural iSOmorphiSms.
Now We wIll ProvE ThE fOllOwINg pROpOsiTIon.
\[Mainfed\] THe CaTegorIes $\Sklc$ ANd $\SKaL$ aRe (noN full) Subcategories Of, rEspeCTivEly, $\HLc$ and $\DlC$. ThE rEstriCtion oF the cOnTravariant functOr $\LAm^a$ (respectIveLy, $\lAM^T$) tO the sUBcategOry $\sKAl$ (resp., $\SKlC$) coincIDes WiTH THe Contravariant functOr $\pSI_1^a$ (Resp., $\Psi_1^t$).
obviouSLy, ThE Category $\sKlC$ iS a suBCAtegoRy of THe Category $\hLC$ and THe ReStrictiOn Of the cOnTraVarIant fUNctoR $\LAM^t$ tO the subcAtegoRY $\SKLC$ coincides WIth the contravARiANT fUNctoR $\PsI_1^t$.
Let $\p | and every $G\in RC(Y)$, a nd for eve ry $\ p\i n\S KA L((A ,\rh o,\BBBB),(B,\e t a,\B BBB\ap))$ and for ever y bou nd e d ul t ra filte r $u$ i n $ B $ (s ee \ [bo un d cl \]) $ $\l abel{ps i1a}
\Psi_ 1^a (\ p)(\s_u)=\s_ { \p \inv(u)},$ $ w here $\s_{\p \in v(u)}$ i s a clust erin $( A,C_\r h o)$ (s ee \[Alex pr n \], (\ [ sigmau\ ] ) a nd \ [uniqult\] for $C _ \r h o$ and $\s_u$, and n ot e t h a t,by\[conclust th \], a n y bound e dc l u ste r $\s$ in $(B, \eta,\BBBB\ a p)$ can b ewri t ten in thefo r m $ \s_u$ for s omebounded u ltrafi l ter $u$ in $B$) ; then $\ l^g :
Id _ {\ ,\ SKA L} \ lra \ Ps i_1 ^ t\c irc\Psi_ 1^ a$ , whe re $ \ l ^ g (A,\ rho ,\BB BB)=\ l_A^g$, for e ver y $( A ,\r ho,\B BBB)\ in\c ar d\SKA L$ (se e (\[ ha pisomn\]) for t he n otation $ \l_ A^ g$) ,and $ t :Id_{\ ,\S KLC }\lra\P si_1^a\ c irc \P s i _ 1^ t$, where $t(X)=t_ X$ , fo r every$X\in\ c ar d\ S KLC$ (se e(\[ home o \ ]) fo r th e n otation$t_X$) , a re natura lisomor ph ism s.
Noww e wi ll pro ve the f ollow i ng proposition .
\[mainfed\] Th e ca t egor ies $\SKLC$ an d $\ S KAL$ are (n onf ull)subca te g or i es of, respectively ,$\HLC$ and$\DLC$. The r estriction o f the con trav a ri a nt functor $\L AM^a$ (respecti v ely, $\L AM^t$ ) to the subcateg o r y $\SKAL $ ( res p., $\ S K LC $) coincidesw i th t he contra var iant fu nct or$\P si_ 1^ a$ (resp. , $\Psi_ 1^ t$ ).
Obv iousl y , the ca te gor y$\S KLC$i s a su bcate gory o ft hecategor y $ \ H LC$an dtheres tr ictio n of the contra variant f unc t or $ \L AM ^t$ tothe subcatego ry $\SKLC$ c oi nci des wi t h the con travariant functor $\Ps i _1^t$.
Le t $\p | and_every $G\in_RC(Y)$, and for every_$\p\in\SKAL((A,\rho,\BBBB),(B,\eta,\BBBB\ap))$ and_for_every bounded_ultrafilter_$u$ in $B$_(see \[boundcl\]) $$\label{psi1a}
\Psi_1^a(\p)(\s_u)=\s_{\p\inv(u)},$$_where $\s_{\p\inv(u)}$ is a_cluster in $(A,C_\rho)$_(see_\[Alexprn\], (\[sigmau\]) and \[uniqult\] for $C_\rho$ and $\s_u$, and note that, by \[conclustth\], any_bounded_cluster $\s$_in_$(B,\eta,\BBBB\ap)$_can be written in the_form $\s_u$ for some bounded_ultrafilter $u$_in $B$); then $\l^g:
Id_{\,\SKAL}\lra\Psi_1^t\circ\Psi_1^a$, where $\l^g(A,\rho,\BBBB)=\l_A^g$, for every_$(A,\rho,\BBBB)\in\card\SKAL$_(see (\[hapisomn\]) for_the notation $\l_A^g$), and $t:Id_{\,\SKLC}\lra\Psi_1^a\circ\Psi_1^t$, where $t(X)=t_X$, for every_$X\in\card\SKLC$ (see (\[homeo\]) for the notation_$t_X$), are natural_isomorphisms.
Now_we_will prove the following_proposition.
\[mainfed\] The categories $\SKLC$ and $\SKAL$_are (non full) subcategories of, respectively,_$\HLC$ and $\DLC$. The restriction of the_contravariant functor $\LAM^a$ (respectively, $\LAM^t$) to_the subcategory $\SKAL$ (resp., $\SKLC$)_coincides with_the contravariant functor $\Psi_1^a$ (resp.,_$\Psi_1^t$).
Obviously, the category_$\SKLC$ is_a subcategory of_the category $\HLC$ and the restriction_of the contravariant_functor $\LAM^t$ to the subcategory $\SKLC$_coincides_with the contravariant_functor_$\Psi_1^t$.
Let_$\p |
^-=\min\{0,r\}$, and the paths $\gamma_+,\gamma_-$ satisfying $-\gamma_+\subseteq\gamma_-$ with, for some $\phi_+\in (\pi/3,\pi/2)$, $\phi_-\in (\pi/2,\pi-\phi_+)$. The Airy$_{2\to 1}$ process crosses over between the Airy$_2$ and the Airy$_1$ processes in the sense that ${\mathrm{Airy}_{2\to 1}}(t+\tau)$ converges to $2^{1/3}{\mathrm{Airy}_1}(2^{-2/3}t)$ as $\tau\to\infty$ and to ${\mathrm{Airy}_2}(1;t)$ (the Airy$_2$ process, i.e. the top line of the multi-layer Airy$_2$ process) when $\tau\to-\infty$ (in the sense of finite dimensional distributions). It is expected to govern the asymptotic spatial fluctuations in random growth models when the initial conditions are deterministic near the point where the hydrodynamic profile changes from flat to curved. In particular, it is shown in [@bfs] that it governs the asymptotic fluctuations near the profile switch point for the totally asymmetric simple exclusion process starting with particles only at the even negative integers.
We take again $X={\mathbb{R}}$ and $\mu$ the Lebesgue measure, and for $i<j$ we define $$\begin{gathered}
{\mathcal{W}}_{t_i,t_j}(x,y)=e^{(t_j-t_i)\Delta}(x-(t_i^-)^2,y-(t_j^-)^2),\quad\,\,
K_{t_i}(x,y)={K_{2\to 1}}^{t_i}(x,y):={K^{{\rm ext}}_{2\to 1}}(t_i,x;t_i,y)\\
{\mathcal{W}}_{t_j,t_i}K_{t_i}(x,y)={K^{{\rm ext}}_{2\to 1}}(t_j,x;t_i,y).\\end{gathered}$$ Proceeding as in Section \[sec:pearcey\] one checks that these choices satisfy Assumption \[assum:1\], and hence (under the additional assumptions) we may apply Theorem \[thm:extendedToBVP\]. Using the translation invariance of | ^-=\min\{0,r\}$, and the paths $ \gamma_+,\gamma_-$ satisfying $ -\gamma_+\subseteq\gamma_-$ with, for some $ \phi_+\in (\pi/3,\pi/2)$, $ \phi_-\in (\pi/2,\pi-\phi_+)$. The Airy$_{2\to 1}$ process crosses over between the Airy$_2 $ and the Airy$_1 $ summons in the common sense that $ { \mathrm{Airy}_{2\to 1}}(t+\tau)$ converges to $ 2^{1/3}{\mathrm{Airy}_1}(2^{-2/3}t)$ as $ \tau\to\infty$ and to $ { \mathrm{Airy}_2}(1;t)$ (the Airy$_2 $ process, i.e. the top line of the multi - layer Airy$_2 $ procedure) when $ \tau\to-\infty$ (in the common sense of finite dimensional distributions). It is expected to govern the asymptotic spatial fluctuation in random growth models when the initial condition are deterministic near the point where the hydrodynamic profile change from flat to curved. In particular, it is shown in [ @bfs ] that it governs the asymptotic fluctuations near the visibility switch point for the totally asymmetrical simple exclusion process starting with particles merely at the even negative integers.
We take again $ X={\mathbb{R}}$ and $ \mu$ the Lebesgue measure, and for $ i < j$ we specify $ $ \begin{gathered }
{ \mathcal{W}}_{t_i, t_j}(x, y)=e^{(t_j - t_i)\Delta}(x-(t_i^-)^2,y-(t_j^-)^2),\quad\,\,
K_{t_i}(x, y)={K_{2\to 1}}^{t_i}(x, y):={K^{{\rm ext}}_{2\to 1}}(t_i, x;t_i, y)\\
{ \mathcal{W}}_{t_j, t_i}K_{t_i}(x, y)={K^{{\rm ext}}_{2\to 1}}(t_j, x;t_i, y).\\end{gathered}$$ Proceeding as in Section \[sec: pearcey\ ] one checks that these choices satisfy Assumption \[assum:1\ ], and hence (under the additional assumptions) we may apply Theorem \[thm: extendedToBVP\ ]. Using the translation invariance of | ^-=\min\{0,g\}$, and the paths $\gamma_+,\gamoa_-$ satisfying $-\gcnma_+\subveteq\gzmma_-$ witf, for some $\phi_+\in (\pi/3,\pi/2)$, $\phi_-\in (\'i/2,\pi-\phi_+)$. Tye Airy$_{2\to 1}$ process crorses over between the Qiry$_2$ and tis Airy$_1$ ixocesacs in vhe sense that ${\kathrm{Airy}_{2\do 1}}(t+\tau)$ convercer co $2^{1/3}{\mathrm{Airy}_1}(2^{-2/3}t)$ as $\tau\to\infty$ and to ${\iathrm{Aorj}_2}(1;t)$ (the Airy$_2$ prjcesx, i.e. fhe top line of the multi-layer Airg$_2$ procevs) when $\tau\to-\onfty$ (in the sense of finihe dlmensional distribktions). It iw ex[wcted to govdrn the asjkptotic spztial fluctuations in random gruwth kodels wheb rhe hnitial convitionf are determlmistic near tne point where thx hyerodynamic profile chenges from flat to corved. In pdrcicular, it is shown ib [@vfs] tvat ht gucervs uhe aaymptohic fluctuatikns near thw profile switch poone for the totamly asrmietric simple exclusion process startinc wjth particles only at tye even negative intehers.
We tahe again $X={\mathbb{R}}$ and $\mu$ the Lebesgue measure, and for $m<j$ we dcfind $$\bfgin{gathered}
{\mathcal{W}}_{t_i,t_j}(x,y)=e^{(t_j-t_i)\Delta}(x-(t_i^-)^2,y-(t_j^-)^2),\qtzd\,\,
L_{t_i}(x,y)={K_{2\to 1}}^{t_i}(x,y):={K^{{\vm ext}}_{2\to 1}}(t_i,x;t_i,y)\\
{\mstjcsj{W}}_{t_j,t_i}K_{t_i}(x,y)={K^{{\ro ext}}_{2\to 1}}(t_n,x;t_i,y).\\end{gathered}$$ Pgoceedigg as in Sectijn \[src:pearcey\] one checks that tyese choices watisfy Assumption \[assum:1\], and kence (onder yhe additional assumptiuns) se may applj Theorem \[ghm:extendedToBVP\]. Uspng dhe translation invariance of | ^-=\min\{0,r\}$, and the paths $\gamma_+,\gamma_-$ satisfying $-\gamma_+\subseteq\gamma_-$ some (\pi/3,\pi/2)$, $\phi_-\in The Airy$_{2\to 1}$ Airy$_2$ the Airy$_1$ processes the sense that 1}}(t+\tau)$ converges to $2^{1/3}{\mathrm{Airy}_1}(2^{-2/3}t)$ as $\tau\to\infty$ to ${\mathrm{Airy}_2}(1;t)$ (the Airy$_2$ process, i.e. the top line of the multi-layer Airy$_2$ when $\tau\to-\infty$ (in the sense of finite dimensional distributions). It is expected to the spatial in growth models when the initial conditions are deterministic near the point where the hydrodynamic profile changes flat to curved. In particular, it is shown [@bfs] that it governs asymptotic fluctuations near the profile point the totally simple process with particles only the even negative integers. We take again $X={\mathbb{R}}$ and $\mu$ the Lebesgue measure, and for $i<j$ we $$\begin{gathered} {\mathcal{W}}_{t_i,t_j}(x,y)=e^{(t_j-t_i)\Delta}(x-(t_i^-)^2,y-(t_j^-)^2),\quad\,\, ext}}_{2\to 1}}(t_i,x;t_i,y)\\ ext}}_{2\to Proceeding in Section \[sec:pearcey\] that these choices satisfy Assumption \[assum:1\], the additional assumptions) we may apply Theorem \[thm:extendedToBVP\]. the translation of | ^-=\min\{0,r\}$, and the paths $\gamma_+,\gamma_-$ Satisfying $-\Gamma_+\SubSetEq\GammA_-$ witH, for some $\phi_+\in (\pI/3,\Pi/2)$, $\phI_-\in (\pi/2,\pi-\phi_+)$. The Airy$_{2\to 1}$ proCess cRoSSes oVEr BetweEn the AiRY$_2$ aND The aiRy$_1$ ProCeSSeS in thE seNse that ${\Mathrm{Airy}_{2\To 1}}(t+\TaU)$ converges to $2^{1/3}{\MAtHrm{Airy}_1}(2^{-2/3}t)$ as $\Tau\To\infty$ and to ${\MatHrm{AirY}_2}(1;t)$ (The aIry$_2$ prOceSs, i.e. tHe top lINe of thE multi-layEr aIry$_2$ proCEss) when $\TAU\tO-\infTy$ (in the sense of finITe DImensional distRibutiOnS). it IS ExpEctEd to govern ThE asymPTotic spATiAL FLucTUations in randOm growth modELs wHen the InItiAL condiTions ArE DetErministic nEar tHe point whEre the HYdrodynAMic profIle chaNgeS frOm flAT tO cUrvEd. iN paRTiCulAR, it Is shown iN [@bFs] That iT govERNS The aSymPtotIc fluCtuations near The ProfILe sWitch Point For tHe TotalLy asymMetriC sImple exclusion pRoceSs startinG wiTh ParTiCles oNLy at thE evEn nEgative IntegerS.
we tAkE AGAiN $X={\mathbb{R}}$ and $\mu$ the LEbESGuE measure, And for $I<J$ wE dEFine $$\begiN{gAthEred}
{\MAThcal{w}}_{t_i,t_J}(X,y)=E^{(t_j-t_i)\DelTa}(x-(t_i^-)^2,y-(T_J^-)^2),\qUaD\,\,
K_{t_i}(x,y)={K_{2\To 1}}^{T_i}(x,y):={K^{{\rM eXt}}_{2\tO 1}}(t_i,X;t_i,y)\\
{\mAThcaL{W}}_{t_j,t_i}k_{t_i}(x,y)={K^{{\rm Ext}}_{2\to 1}}(T_J,x;t_i,y).\\end{gatherED}$$ Proceeding as IN SECTiON \[sec:PeaRcey\] one checKs thAT theSe chOIcEs sATisfy assumPtIOn \[ASsum:1\], and hence (under thE aDditioNal asSumptions) we maY apply TheoREM \[Thm:extenDedTObVp\]. using the translAtion Invariance OF | ^-=\min\{0,r\}$, and the p aths $\gam ma_+, \ga mma _- $ sa tisf ying $-\gamma_ + \sub seteq\gamma_-$ with, f or so me $\ph i _+ \in ( \pi/3,\ p i/ 2 ) $,$\ ph i_- \i n ( \pi/2 ,\p i-\phi_ +)$. The A iry $_ {2\to 1}$ pr o ce ss crosses ov er between t heAiry$_ 2$ an d theAir y$_1$ proce s ses in the sens et hat ${ \ mathrm{ A i ry }_{2 \to 1}}(t+\tau)$c on v erges to $2^{1 /3}{\m at h rm { A iry }_1 }(2^{-2/3} t) $ as$ \tau\to \ in f t y $ a n d to ${\mathr m{Airy}_2}( 1 ;t) $ (the A iry $ _2$ pr ocess ,i .e. the top li ne o f the mul ti-lay e r Airy$ _ 2$ proc ess) w hen $\ tau\ t o- \i nft y$ (in th e s e nse of fini te d imens iona l d i stri but ions ). It is expectedtogove r n t he as ympto ticsp atial fluct uatio ns in random grow th m odels whe n t he in it ial c o nditio nsare determ inistic nea rt h e p oint where the hyd ro d y na mic prof ile ch a ng es from fla ttocurv e d . Inpart i cu lar, itis sho w nin [@bfs] t hat it g ove rns thea symp toticfluctuat ionsn ear the profil e switch point fo r th e tot all y asymmetri c si m pleexcl u si onp roces s sta rt i ng with particles only a t theevennegative inte gers.
Wet a k e again$X={ \ ma t hbb{R}}$ and $ \mu$the Lebesg u e measur e, an d for $i <j$ we de f i ne $$\be gin {ga the red } {\mathcal{W}} _ { t_i, t_ j}(x,y) =e^ {(t_j-t _i) \De lta }(x -( t_i^-)^2, y-(t_j^- )^ 2) ,\ qu ad\ ,\,
K_{t_i}( x, y)= {K _{2 \to 1 } }^{t_i }(x,y ):={ K^ {{ \ rmext}}_{ 2 \t o 1}}( t_ i, x;t_ i,y )\ \
{ \mat h cal {W}}_{t _j,t_i}K_ {t_ i }(x, y) ={ K^{{\rm ext}}_{2\to1} }(t_j,x;t_ i, y). \\end{ g a thered}$ $ Proceeding as in Sect i on \[se c:p earce y\]one check s t hat th ese choice s sati sfy A ss ump t i on \[ a s su m:1 \] , and henc e (un der t he add itional assumptions) we m a y a pply Theorem\[t hm:e x t en ded T oB V P\] .U sin g the translation invarianc eo f | ^-=\min\{0,r\}$, and_the paths_$\gamma_+,\gamma_-$ satisfying $-\gamma_+\subseteq\gamma_-$ with,_for some_$\phi_+\in_(\pi/3,\pi/2)$, $\phi_-\in_(\pi/2,\pi-\phi_+)$._The Airy$_{2\to 1}$_process crosses over_between the Airy$_2$ and_the Airy$_1$ processes_in_the sense that ${\mathrm{Airy}_{2\to 1}}(t+\tau)$ converges to $2^{1/3}{\mathrm{Airy}_1}(2^{-2/3}t)$ as $\tau\to\infty$ and to ${\mathrm{Airy}_2}(1;t)$ (the_Airy$_2$_process, i.e._the_top_line of the multi-layer Airy$_2$_process) when $\tau\to-\infty$ (in the_sense of_finite dimensional distributions). It is expected to govern_the_asymptotic spatial fluctuations_in random growth models when the initial conditions are_deterministic near the point where the_hydrodynamic profile changes_from_flat_to curved. In particular,_it is shown in [@bfs] that_it governs the asymptotic fluctuations near_the profile switch point for the totally_asymmetric simple exclusion process starting with_particles only at the even_negative integers.
We_take again $X={\mathbb{R}}$ and $\mu$_the Lebesgue measure,_and for_$i<j$ we define_$$\begin{gathered}
{\mathcal{W}}_{t_i,t_j}(x,y)=e^{(t_j-t_i)\Delta}(x-(t_i^-)^2,y-(t_j^-)^2),\quad\,\,
K_{t_i}(x,y)={K_{2\to 1}}^{t_i}(x,y):={K^{{\rm_ext}}_{2\to 1}}(t_i,x;t_i,y)\\
_{\mathcal{W}}_{t_j,t_i}K_{t_i}(x,y)={K^{{\rm ext}}_{2\to 1}}(t_j,x;t_i,y).\\end{gathered}$$ Proceeding as in_Section_\[sec:pearcey\] one checks_that_these_choices satisfy_Assumption \[assum:1\], and_hence_(under the_additional_assumptions) we may apply Theorem \[thm:extendedToBVP\]._Using_the translation invariance of |
{j}\bar{l}}E_{ik}} \over {(i,k)}} \right ).
\label{6}$$ For $u=\iota$ or $v=\iota, \iota^2=0$ superalgebra $osp(m|2n)$ is contracted to inhomogeneous superalgebra, which is semidirect sum $ \{E_{i\bar{j}}\} \S (so(m) \bigoplus sp(2n)),$ with all anticommutators of the odd generators equal to zero $\{E_{i\bar{j}},E_{k\bar{p}} \} = 0.$
Example: CK contractions of $osp(3|2)$
--------------------------------------
This superalgebra has $so(3)$ as even subalgebra therefore their contractions to the kinematical $(1+1)$ Poincare, Newton and Galilei superalgebras may be fulfilled according to general CK scheme of the first section. But unlike of two odd generators of $osp(1|2)$ the superalgebra $osp(3|2)$ has six odd generators. In the basis $X_{ik}=E_{ki}, \, k,i=1,2,3, \, F=\displaystyle{\frac{1}{2}}E_{44}, \,
E=-\displaystyle{\frac{1}{2}}E_{55}, \,
H=-E_{45}, \, Q_k=E_{k4}, \, Q_{-k}=E_{k5}$ the generators are affected by the contraction coefficients $j_1,j_2$ in the following way $$X_{ik}\to (i,k)X_{ik}, \quad Q_{\pm k}\to (1,k)Q_{\pm k}
\label{12}$$ and $H,F,E $ are remained unchanged. Then superalgebra $osp(3;j|2)$ is given by $$[X_{12},X_{13}]=j_1^2X_{23}, \quad [X_{13},X_{23}]=j_2^2X_{12}, \quad
[X_{23},X_{12}]=X_{13},$$ $$[H,E]=2E, \quad [H,F]=-2F, \quad [E,F]=H,$$ $$[X_{ik},Q_{\pm i}]=Q_{\pm k}, \quad
[X_{ik},Q_{\pm k}]=-(i | { j}\bar{l}}E_{ik } } \over { (i, k) } } \right).
\label{6}$$ For $ u=\iota$ or $ v=\iota, \iota^2=0 $ superalgebra $ osp(m|2n)$ is contracted to inhomogeneous superalgebra, which is semidirect sum $ \{E_{i\bar{j}}\ } \S (so(m) \bigoplus sp(2n)),$ with all anticommutators of the odd generators adequate to zero $ \{E_{i\bar{j}},E_{k\bar{p } } \ } = 0.$
case: CK contractions of $ osp(3|2)$
--------------------------------------
This superalgebra has $ so(3)$ equally even subalgebra therefore their contraction to the kinematical $ (1 + 1)$ Poincare, Newton and Galilei superalgebras may be fulfilled according to cosmopolitan CK outline of the first part. But unlike of two odd generators of $ osp(1|2)$ the superalgebra $ osp(3|2)$ suffer six odd generator. In the footing $ X_{ik}=E_{ki }, \, k, i=1,2,3, \, F=\displaystyle{\frac{1}{2}}E_{44 }, \,
E=-\displaystyle{\frac{1}{2}}E_{55 }, \,
H=-E_{45 }, \, Q_k = E_{k4 }, \, Q_{-k}=E_{k5}$ the generators are affected by the compression coefficients $ j_1,j_2 $ in the following way $ $ X_{ik}\to (iodine, k)X_{ik }, \quad Q_{\pm k}\to (1,k)Q_{\pm kelvin }
\label{12}$$ and $ H, F, E $ are remained unchanged. Then superalgebra $ osp(3;j|2)$ is given by $ $ [ X_{12},X_{13}]=j_1 ^ 2X_{23 }, \quad [ X_{13},X_{23}]=j_2 ^ 2X_{12 }, \quad
[ X_{23},X_{12}]=X_{13},$$ $ $ [ heat content, E]=2E, \quad [ H, F]=-2F, \quad [ E, F]=H,$$ $ $ [ X_{ik},Q_{\pm i}]=Q_{\pm k }, \quad
[ X_{ik},Q_{\pm k}]=-(i | {j}\bag{l}}E_{ik}} \over {(i,k)}} \right ).
\labeu{6}$$ For $u=\iota$ or $r=\uota, \imta^2=0$ suleralgebfa $osp(m|2n)$ is contracted to iniomoteneoys superalgebra, which ks semidigect sum $ \{E_{i\ber{j}}\} \S (so(m) \bigoplna sp(2n)),$ wlch alm antncimmutators of jhe odd genesators equal tm xexo $\{E_{i\bar{j}},E_{k\bar{p}} \} = 0.$
Example: CK contraceions og $lsp(3|2)$
--------------------------------------
This superajgebga has $so(3)$ as even subalgebra therefore tgeir coitractions to tne kinematical $(1+1)$ Poincare, Nfwtoj and Galilei supegalgebras mqy bq fulfilled azcording to general CK scheme of the first section. Bug unlnke of two ide gfterators of $osp(1|2)$ nhe superalgenga $osp(3|2)$ vas six odd generatorx. Ii thw basis $X_{ik}=E_{ki}, \, k,i=1,2,3, \, F=\visplaystyle{\frac{1}{2}}E_{44}, \,
E=-\dysplaystyne{\yrac{1}{2}}E_{55}, \,
H=-E_{45}, \, Q_k=E_{k4}, \, Q_{-k}=E_{k5}$ rhw gengratoss afw awfedtxd gy the coitraction ckefficients $j_1,j_2$ in the followinb rqy $$X_{ik}\to (i,k)X_{ii}, \quad Q_{\[m k}\to (1,k)Q_{\pm k}
\label{12}$$ and $H,F,E $ are remaineg uhchanged. Then superalgevra $osp(3;j|2)$ is given by $$[D_{12},X_{13}]=j_1^2X_{23}, \quad [X_{13},X_{23}]=j_2^2X_{12}, \quad
[X_{23},X_{12}]=X_{13},$$ $$[H,E]=2E, \quad [H,F]=-2F, \quad [E,F]=H,$$ $$[X_{ik},Q_{\pm i}]=Q_{\pm k}, \quad
[E_{iy},Q_{\pn h}]=-(k | {j}\bar{l}}E_{ik}} \over {(i,k)}} \right ). \label{6}$$ For $v=\iota, superalgebra $osp(m|2n)$ contracted to inhomogeneous $ \S (so(m) \bigoplus with all anticommutators the odd generators equal to zero \} = 0.$ Example: CK contractions of $osp(3|2)$ -------------------------------------- This superalgebra has $so(3)$ even subalgebra therefore their contractions to the kinematical $(1+1)$ Poincare, Newton and Galilei may fulfilled to CK scheme of the first section. But unlike of two odd generators of $osp(1|2)$ the superalgebra has six odd generators. In the basis $X_{ik}=E_{ki}, k,i=1,2,3, \, F=\displaystyle{\frac{1}{2}}E_{44}, \, \, H=-E_{45}, \, Q_k=E_{k4}, \, the are affected the coefficients in the following $$X_{ik}\to (i,k)X_{ik}, \quad Q_{\pm k}\to (1,k)Q_{\pm k} \label{12}$$ and $H,F,E $ are remained unchanged. Then superalgebra $osp(3;j|2)$ given by [X_{13},X_{23}]=j_2^2X_{12}, \quad $$[H,E]=2E, [H,F]=-2F, [E,F]=H,$$ $$[X_{ik},Q_{\pm i}]=Q_{\pm [X_{ik},Q_{\pm k}]=-(i | {j}\bar{l}}E_{ik}} \over {(i,k)}} \right ).
\label{6}$$ FOr $u=\iota$ or $v=\Iota, \iOta^2=0$ SupErAlgeBra $oSp(m|2n)$ is contractED to iNhomogeneous superalgebRa, whiCh IS semIDiRect sUm $ \{E_{i\bar{J}}\} \s (sO(M) \BigOpLuS sp(2N)),$ wITh All anTicOmmutatOrs of the odD geNeRators equal tO ZeRo $\{E_{i\bar{j}},E_{k\Bar{P}} \} = 0.$
Example: CK coNtrActionS oF $osP(3|2)$
--------------------------------------
this sUpeRalgeBra has $SO(3)$ as eveN subalgebRa THerefoRE their cONTrActiOns to the kinematicAL $(1+1)$ POIncare, Newton anD GalilEi SUpERAlgEbrAs may be fulFiLled aCCording TO gENERal ck scheme of the fIrst section. bUt uNlike oF tWo oDD generAtors Of $OSp(1|2)$ tHe superalgeBra $oSp(3|2)$ has six oDd geneRAtors. In THe basis $x_{ik}=E_{ki}, \, K,i=1,2,3, \, F=\DisPlaySTyLe{\FraC{1}{2}}E_{44}, \,
e=-\DisPLaYstYLe{\fRac{1}{2}}E_{55}, \,
H=-E_{45}, \, Q_k=e_{k4}, \, q_{-k}=e_{k5}$ the GeneRATORs arE afFectEd by tHe contraction CoeFficIEntS $j_1,j_2$ in The foLlowInG way $$X_{Ik}\to (i,k)x_{ik}, \quAd q_{\pm k}\to (1,k)Q_{\pm k}
\labeL{12}$$ and $h,F,E $ are remAinEd UncHaNged. THEn supeRalGebRa $osp(3;j|2)$ iS given bY $$[x_{12},X_{13}]=j_1^2x_{23}, \qUAD [x_{13},X_{23}]=J_2^2X_{12}, \quad
[X_{23},X_{12}]=X_{13},$$ $$[H,E]=2E, \quad [H,F]=-2f, \qUAD [E,f]=H,$$ $$[X_{ik},Q_{\pm I}]=Q_{\pm k}, \qUAd
[x_{iK},q_{\pm k}]=-(i | {j}\bar{l}}E_{ik}} \over { (i,k)}} \r ight).\la be l{6} $$ F or $u=\iota$ o r $v= \iota, \iota^2=0$ supe ralge br a $os p (m |2n)$ is con t ra c t edto i nho mo g en eoussup eralgeb ra, whichisse midirect sum $\{E_{i\bar {j} }\} \S (so(m ) \ bigopl us sp ( 2n)), $ w ith a ll ant i commut ators ofth e odd g e nerator s eq ualto zero $\{E_{i\b a r{ j }},E_{k\bar{p} } \} = 0 . $E xam ple : CK contr ac tions of $osp ( 3| 2 ) $
-- - ------------- ----------- - --- ------ -
Th i s supe ralge br a ha s $so(3)$ a s ev en subalg ebra t h erefore their c ontrac tio nsto t h eki nem at i cal $( 1+1 ) $ P oincare, N ew ton a nd G a l i l ei s upe ralg ebras may be fulfi lle d ac c ord ing t o gen eral C K sch eme of thefi rst section. Bu t un like of t wood d g en erato r s of $ osp (1| 2)$ the supera l geb ra $ o sp (3|2)$ has six odd g e n er ators. I n theb as is $X_{ik}= E_ {ki }, \ , k,i=1 ,2,3 , \ , F=\dis playst y le {\ frac{1} {2 }}E_{4 4} , \ ,
E =-\di s play style{ \frac{1} {2}}E _ {55}, \,
H=-E_ { 45}, \, Q_k=E _ {k 4 } ,\ , Q_ {-k }=E_{k5}$ t he g e nera tors ar e a f fecte d byth e c o ntraction coefficie nt s $j_1 ,j_2$ in the follo wing way $ $ X _ {ik}\to(i,k ) X_ { ik}, \quad Q_{ \pm k }\to (1,k) Q _{\pm k}
\lab el{12}$$ and $H,F , E $ are r ema ine d u nch a n ge d. Then super a l gebr a$osp(3; j|2 )$ is g ive n b y $ $[X _{ 12},X_{13 }]=j_1^2 X_ {2 3} ,\qu ad [X _ {13},X_{ 23 }]= j_ 2^2 X_{12 } , \qua d
[X_ {23} ,X _{ 1 2}] =X_{13} , $$ $ $[H, E] =2 E, \ qua d[H,F] =-2F , \q uad [E, F]=H,$$ $ $[X _ {ik} ,Q _{ \pm i}] =Q_{\pm k}, \ qu ad
[X_{ik} ,Q _{\ pm k}] = - (i | {j}\bar{l}}E_{ik}} \over_{(i,k)}} \right_).
\label{6}$$ For $u=\iota$ or_$v=\iota, \iota^2=0$_superalgebra_$osp(m|2n)$ is_contracted_to inhomogeneous superalgebra,_which is semidirect_sum $ \{E_{i\bar{j}}\} \S_(so(m) \bigoplus sp(2n)),$_with_all anticommutators of the odd generators equal to zero $\{E_{i\bar{j}},E_{k\bar{p}} \} = 0.$
Example: CK_contractions_of $osp(3|2)$
--------------------------------------
This_superalgebra_has_$so(3)$ as even subalgebra therefore_their contractions to the kinematical_$(1+1)$ Poincare,_Newton and Galilei superalgebras may be fulfilled according_to_general CK scheme_of the first section. But unlike of two odd_generators of $osp(1|2)$ the superalgebra $osp(3|2)$_has six odd_generators._In_the basis $X_{ik}=E_{ki}, \,_k,i=1,2,3, \, F=\displaystyle{\frac{1}{2}}E_{44}, \,
E=-\displaystyle{\frac{1}{2}}E_{55}, \,
H=-E_{45}, \,_Q_k=E_{k4}, \, Q_{-k}=E_{k5}$ the generators are_affected by the contraction coefficients $j_1,j_2$ in_the following way $$X_{ik}\to (i,k)X_{ik}, \quad_Q_{\pm k}\to (1,k)Q_{\pm k}
\label{12}$$ and_$H,F,E $_are remained unchanged. Then superalgebra_$osp(3;j|2)$ is given_by $$[X_{12},X_{13}]=j_1^2X_{23},_\quad [X_{13},X_{23}]=j_2^2X_{12}, \quad
[X_{23},X_{12}]=X_{13},$$_$$[H,E]=2E, \quad [H,F]=-2F, \quad [E,F]=H,$$ $$[X_{ik},Q_{\pm_i}]=Q_{\pm k}, \quad
[X_{ik},Q_{\pm_k}]=-(i |
^{[3]}$}
\put(197, 42){$S$}
\put(307, 94){$R^{[L]}$}
\put(309, 42){$S$}
\put(335, 79){\footnotesize $c_L$}
\put(360, 67){$Q$}
\put(363, 37){$Y$}
\put(29, 6){\footnotesize $s_1$}
\put(118, 6){\footnotesize $s_2$}
\put(207, 6){\footnotesize $s_3$}
\put(319, 6){\footnotesize $s_L$}
\end{overpic}$$ where structure fragments $S$ are Clebsh-Gordan coefficients. The tensor $Y$ can be any random tensor; no algorithm based on a SU(2) invariant benchmark can determine or variate $Y$ because it is the only non SU(2) invariant component of the MPS network.
[1]{}
Michael A. Nielsen, Isaac L. Chuang, *Quantum Computation and Quantum Informsation*, Cambridge University Press (2000).
Steven R. White, *Density matrix formulation for quantum renormalization groups*, Phys. Rev. Lett. 69, 2863-2866 (1992).
Kenneth G. Wilson, *The renormalization group: Critical phenomena and the Kondo problem*, Rev. Mod. Phys. 47, 773-840 (1975).
Steven R. White, *Density Matrix Renormalization Group algorithms with a single center site*, Phys. Rev. B 72, 180403 (2005).
Ulrich Schollwöck, *The density-matrix renormalization group*, Rev. Mod. Phys. 77, 259–315 (2005).
F. Verstraete, D. Porras and J. I. Cirac, *Density Matrix Renormalization Group and Periodic Boundary Conditions: a Quantum Information Perspective*, Phys. Rev. Lett. 93, 227205 (2004).
G. Vidal, J. I. Latorre, E. Rico, A. Kitaev, *Entanglement in quantum critical phenomena*, Phys.Rev.Lett.90:227902,2003.
J. I. Latorre, E. Rico, G. Vidal, *Ground state entanglement in quantum spin chains*, Quant | ^{[3]}$ }
\put(197, 42){$S$ }
\put(307, 94){$R^{[L]}$ }
\put(309, 42){$S$ }
\put(335, 79){\footnotesize $ c_L$ }
\put(360, 67){$Q$ }
\put(363, 37){$Y$ }
\put(29, 6){\footnotesize $ s_1 $ }
\put(118, 6){\footnotesize $ s_2 $ }
\put(207, 6){\footnotesize $ s_3 $ }
\put(319, 6){\footnotesize $ s_L$ }
\end{overpic}$$ where structure fragments $ S$ are Clebsh - Gordan coefficients. The tensor $ Y$ can be any random tensor; no algorithm based on a SU(2) changeless benchmark can settle or variate $ Y$ because it is the only non SU(2) invariant part of the MPS network.
[ 1 ] { }
Michael A. Nielsen, Isaac L. Chuang, * Quantum Computation and Quantum Informsation *, Cambridge University Press (2000).
Steven R. White, * concentration matrix formulation for quantum renormalization groups *, Phys. Rev. Lett. 69, 2863 - 2866 (1992).
Kenneth G. Wilson, * The renormalization group: Critical phenomenon and the Kondo trouble *, Rev. Mod. Phys. 47, 773 - 840 (1975).
Steven R. White, * Density Matrix Renormalization Group algorithms with a single center site *, Phys. Rev. B 72, 180403 (2005).
Ulrich Schollwöck, * The density - matrix renormalization group *, Rev. Mod. Phys. 77, 259–315 (2005).
F. Verstraete, D. Porras and J. I. Cirac, * Density Matrix Renormalization Group and Periodic Boundary Conditions: a Quantum Information Perspective *, Phys. Rev. Lett. 93, 227205 (2004).
G. Vidal, J. I. Latorre, E. Rico, A. Kitaev, * Entanglement in quantum critical phenomena *, Phys. Rev. Lett.90:227902,2003.
J. I. Latorre, E. Rico, G. Vidal, * footing state entanglement in quantum tailspin chains *, Quant | ^{[3]}$}
\puh(197, 42){$S$}
\put(307, 94){$R^{[L]}$}
\put(309, 42){$S$}
\put(335, 79){\foounotesize $c_L$}
\put(360, 67){$Q$}
\put(363, 37){$B$}
\put(29, 6){\fkotnoteskze $s_1$}
\put(118, 6){\footnotesize $s_2$}
\put(207, 6){\fiotnouvsize $s_3$}
\put(319, 6){\footnoteskze $s_L$}
\end{lverpic}$$ qhert structure fragmxhts $S$ avz Clegdh-Goxden coefficients. The tensos $Y$ can be any rxnbom tensor; no algorithm based on a ST(2) invaroajt benchmark cwn dttewmins or variate $Y$ because it is the ohly non SU(2) invariant component of the MPS netwlrk.
[1]{}
Mlchael A. Nielsen, Idaac L. Chuabg, *Qtqntum Computxtion and Quantum Infotmsation*, Cambridge University Prdss (2000).
Sceven R. Whijz, *Dejvity matrix formllation for qmsntum senormakization groupx*, Piys. Eev. Lett. 69, 2863-2866 (1992).
Kenneth G. Xilson, *The renormalieation grogp: Critical phenomeba and jhe Kmndo proclej*, Cev. Mod. Pjys. 47, 773-840 (1975).
Steven D. White, *Denwity Matrix Renormakisqtion Group amgoritrmf with a single center site*, Phys. Rev. B 72, 180403 (2005).
Umrich Schollwöck, *The denwity-matrix renormalizwtion grotp*, Rev. Mod. Phys. 77, 259–315 (2005).
F. Verstraete, D. Porras and J. I. Chrac, *Vevsiuy Matrkz Genormalization Group and Periodic Boundary Cjhdotpons: a Quantum Innormation Perspectovf*, Lrys. Rev. Lett. 93, 227205 (2004).
G. Vibzl, J. I. Latorre, E. Ricl, A. Kitwev, *Ebtanglemegt im quantum critical phenomenq*, Phys.Rev.Letn.90:227902,2003.
J. I. Latorre, E. Rico, G. Ridal, *Ground staje entsnglement in quantum spnn chajns*, Quant | ^{[3]}$} \put(197, 42){$S$} \put(307, 94){$R^{[L]}$} \put(309, 42){$S$} $c_L$} 67){$Q$} \put(363, \put(29, 6){\footnotesize $s_1$} $s_3$} 6){\footnotesize $s_L$} \end{overpic}$$ structure fragments $S$ Clebsh-Gordan coefficients. The tensor $Y$ can any random tensor; no algorithm based on a SU(2) invariant benchmark can determine variate $Y$ because it is the only non SU(2) invariant component of the network. Michael Nielsen, L. Chuang, *Quantum Computation and Quantum Informsation*, Cambridge University Press (2000). Steven R. White, *Density matrix for quantum renormalization groups*, Phys. Rev. Lett. 69, (1992). Kenneth G. Wilson, renormalization group: Critical phenomena and Kondo Rev. Mod. 47, (1975). R. White, *Density Renormalization Group algorithms with a single center site*, Phys. Rev. B 72, 180403 (2005). Ulrich Schollwöck, *The renormalization group*, Phys. 77, (2005). Verstraete, Porras and J. *Density Matrix Renormalization Group and Periodic Quantum Information Perspective*, Phys. Rev. Lett. 93, 227205 G. Vidal, I. Latorre, E. Rico, A. Kitaev, in quantum critical phenomena*, Phys.Rev.Lett.90:227902,2003. J. I. Latorre, Rico, G. Vidal, *Ground state entanglement in quantum spin chains*, Quant | ^{[3]}$}
\put(197, 42){$S$}
\put(307, 94){$R^{[L]}$}
\put(309, 42){$S$}
\put(335, 79){\footnotesIze $c_L$}
\put(360, 67){$Q$}
\pUt(363, 37){$Y$}
\puT(29, 6){\foOtnOtEsizE $s_1$}
\puT(118, 6){\footnotesize $s_2$}
\PUt(207, 6){\foOtnotesize $s_3$}
\put(319, 6){\footnoteSize $s_l$}
\eND{oveRPiC}$$ wherE structURe FRAgmEnTs $s$ arE CLEbSh-GorDan CoefficIents. The teNsoR $Y$ Can be any randOM tEnsor; no algOriThm based on a Su(2) inVarianT bEncHMark cAn dEtermIne or vARiate $Y$ Because it Is THe only NOn SU(2) invARIaNt coMponent of the MPS neTWoRK.
[1]{}
Michael A. NielsEn, IsaaC L. cHuANG, *QuAntUm ComputatIoN and QUAntum InFOrMSATioN*, cambridge UnivErsity Press (2000).
sTevEn R. WhiTe, *denSIty matRix foRmULatIon for quantUm reNormalizaTion grOUps*, Phys. rEv. Lett. 69, 2863-2866 (1992).
KEnneth g. WiLsoN, *The REnOrMalIzATioN GrOup: cRitIcal phenOmEnA and tHe KoNDO PRoblEm*, REv. MoD. Phys. 47, 773-840 (1975).
steven R. White, *DEnsIty MATriX RenoRmaliZatiOn group AlgoriThms wItH a single center sIte*, PHys. Rev. B 72, 180403 (2005).
UlRicH SChoLlWöck, *THE densiTy-mAtrIx renorMalizatIOn gRoUP*, rEv. mod. Phys. 77, 259–315 (2005).
F. Verstraete, d. PORRaS and J. I. CiRac, *DenSItY MATrix RenoRmAliZatiON group And PERiOdic BounDary CoNDiTiOns: a QuaNtUm InfoRmAtiOn PErspeCTive*, phys. ReV. Lett. 93, 227205 (2004).
G. ViDal, J. I. lAtorre, E. Rico, A. KiTAev, *EntanglemeNT iN QUaNTum cRitIcal phenomeNa*, PhYS.Rev.lett.90:227902,2003.
j. i. LAtoRRe, E. RiCo, G. ViDaL, *grOUnd state entanglemenT iN quantUm spiN chains*, Quant | ^{[3]}$}
\put(197, 42){$S $}
\put(3 07, 9 4){ $R^ {[ L]}$ }
\ put(309, 42){$ S $}
\put(335, 79){\footnot esize $ c _L$}
\put( 360, 67 ) {$ Q $ }
\p ut (36 3, 37 ){$Y$ }
\put(29 , 6){\foot not es ize $s_1$}
\ pu t(118, 6){ \fo otnotesize $ s_2 $}
\p ut (20 7 , 6){ \fo otnot esize$ s_3$} \put(319 ,6 ){\foo t notesiz e $s _L$}
\end{overpic}$$w he r e structure fr agment s$ S$ a reCle bsh-Gordan c oeffi c ients.T he t e nso r $Y$ can be a ny random t e nso r; noal gor i thm ba sed o na SU (2) invaria nt b enchmarkcan de t ermineo r varia te $Y$ be cau se i t i sthe o n lyn on SU ( 2)invarian tco mpone nt o f t h e MP S n etwo rk.
[1]{}
Michae l A . Ni e lse n, Is aac L . Ch ua ng, * Quantu m Com pu tation and Quan tumInformsat ion *, Ca mb ridge Univer sit y P ress (2 000).
S tev en R . W hite, *Density mat ri x fo rmulatio n forq ua nt u m renorm al iza tion g roups *, P h ys . Rev. L ett. 6 9 ,28 63-2866 ( 1992).
Ken net h G.W ilso n, *Th e renorm aliza t ion group: Cri t ical phenomen a a n d t h e Ko ndo problem*,Rev. Mod. Phy s .47, 773-8 40 (1 97 5 ).
Steven R. White, * De nsityMatri x Renormaliza tion Group a l gorithms wit h a single centersite* , Phys. Re v . B 72,18040 3 (2005) .
Ulrich S chollwöc k,*Th e d ens i t y- matrix renorm a l izat io n group *,Rev. Mo d.Phy s.77, 2 59–315 (2 005).
F .Ve rs tr aet e, D. Porras a nd J. I . C irac, *Densi ty Ma trix R en o rma lizatio n G r o up a nd P erio dic B ounda ry C o ndi tions:a Quantum In f orma ti on Perspe ctive*, Phys. R ev. Lett.93 , 2 27205( 2 004).
G . Vidal, J. I. Latorre, E. Rico , A . Kit aev, *Entangl eme nt inqua n tum cr itical phen om ena * , Phys . R ev .Le tt .90:227902 , 2 003 .
J. I . La torre,E. Rico, G. Vidal, *Gr ound state en tan glem e n tinq ua n tum s p inc h ains*, Quant | ^{[3]}$}
\put(197,_42){$S$}
\put(307,_94){$R^{[L]}$}
\put(309, 42){$S$}
\put(335,_79){\footnotesize $c_L$}_
_\put(360, 67){$Q$}
_\put(363,_37){$Y$}
\put(29, 6){\footnotesize_$s_1$}
\put(118, 6){\footnotesize_$s_2$}
\put(207, 6){\footnotesize $s_3$}
_\put(319, 6){\footnotesize $s_L$}
\end{overpic}$$_where_structure fragments $S$ are Clebsh-Gordan coefficients. The tensor $Y$ can be any random tensor;_no_algorithm based_on_a_SU(2) invariant benchmark can determine_or variate $Y$ because it_is the_only non SU(2) invariant component of the MPS_network.
[1]{}
Michael_A. Nielsen, Isaac_L. Chuang, *Quantum Computation and Quantum Informsation*, Cambridge University_Press (2000).
Steven R. White, *Density matrix_formulation for quantum_renormalization_groups*,_Phys. Rev. Lett. 69,_2863-2866 (1992).
Kenneth G. Wilson, *The renormalization_group: Critical phenomena and the Kondo_problem*, Rev. Mod. Phys. 47, 773-840 (1975).
Steven_R. White, *Density Matrix Renormalization Group_algorithms with a single center_site*, Phys._Rev. B 72, 180403 (2005).
Ulrich_Schollwöck, *The density-matrix_renormalization group*,_Rev. Mod. Phys._77, 259–315 (2005).
F. Verstraete, D. Porras_and J. I._Cirac, *Density Matrix Renormalization Group and_Periodic_Boundary Conditions: a_Quantum_Information_Perspective*, Phys._Rev. Lett. 93,_227205_(2004).
G. Vidal,_J._I. Latorre, E. Rico, A. Kitaev,_*Entanglement_in quantum critical phenomena*, Phys.Rev.Lett.90:227902,2003.
J. I. Latorre,_E. Rico, G. Vidal,_*Ground_state entanglement in quantum_spin chains*, Quant |
2000 *Phys. Rev. Lett.* [**85**]{} 3313
Cerf N J, Bourennane M, Karlsson A and Gisin N 2002 *Phys. Rev. Lett.* [**88**]{} 127902
Vaidman L, Aharonov Y and Albert D Z 1987 *Phys. Rev. Lett.* [**58**]{} 1385
Aharonov Y and Englert B-G 2001 *Z. Naturforsch. A* [**56**]{} 16
Aravind P K 2003 *Z. Naturforsch. A* [**58**]{} 2212
Hayashi A, Horibe M and Hashimoto T 2005 *Phys. Rev. A* [**71**]{} 052331
Lee J, Kim M S and Brukner [Č]{} 2003 *Phys. Rev. Lett.* [**91**]{} 087902
Barnum H 2002 quant-ph/0205155
Klappenecker A and Rötteler M 2005 *Proc. ITIS* 1740-1744
Lubkin E 1978 *J. Math Phys.* [**19**]{} 1028
Bandyopadhyay S, Boykin P O, Roychowdhury V and Vatan F 2002 *Algorithmica* [**34**]{} 512
Wocjan P and Beth T 2004 quant-ph/0407081
Weiner M 2009 arXiv:0902.0639.
Paterek T 2007 *Phys. Lett. A* [**367**]{} 57
Grassl M 2004 quant-ph/0406175v2
Paterek T, Dakić B and Brukner [Č]{} 2009 *Phys. Rev. A* [**79**]{} 012109
Brierley S, Weigert S and Bengtsson I 2010 *Quant. Inf. Comp.* [**10**]{} 803
Reck M, Zeilinger A, Bernstein H J and Bertani P 1994 *Phys. Rev. Lett.* [**73**]{} 58
Weihs G, Reck M, Weinfurter H, and Zeilinger A 1996 [*Phys. Rev. A*]{} 893
Żukowski M, Zeilinger A and Horne M 1997 *Phys. Rev. A* [**55**]{} 2564
Politi A, Cryan M J, Rarity J G, Yu S and O | 2000 * Phys. Rev. Lett. * [ * * 85 * * ] { } 3313
Cerf N J, Bourennane M, Karlsson A and Gisin N 2002 * Phys. Rev. Lett. * [ * * 88 * * ] { } 127902
Vaidman L, Aharonov Y and Albert D Z 1987 * Phys. Rev. Lett. * [ * * 58 * * ] { } 1385
Aharonov Y and Englert B - G 2001 * Z. Naturforsch. A * [ * * 56 * * ] { } 16
Aravind P K 2003 * Z. Naturforsch. A * [ * * 58 * * ] { } 2212
Hayashi A, Horibe M and Hashimoto T 2005 * Phys. Rev. A * [ * * 71 * * ] { } 052331
Lee J, Kim M S and Brukner [ Č ] { } 2003 * Phys. Rev. Lett. * [ * * 91 * * ] { } 087902
Barnum H 2002 quant - ph/0205155
Klappenecker A and Rötteler M 2005 * Proc. ITIS * 1740 - 1744
Lubkin E 1978 * J. Math Phys. * [ * * 19 * * ] { } 1028
Bandyopadhyay S, Boykin phosphorus oxygen, Roychowdhury V and Vatan F 2002 * Algorithmica * [ * * 34 * * ] { } 512
Wocjan P and Beth T 2004 quant - ph/0407081
Weiner M 2009 arXiv:0902.0639.
Paterek T 2007 * Phys. Lett. A * [ * * 367 * * ] { } 57
Grassl M 2004 quant - ph/0406175v2
Paterek T, Dakić B and Brukner [ Č ] { } 2009 * Phys. Rev. A * [ * * 79 * * ] { } 012109
Brierley S, Weigert S and Bengtsson I 2010 * Quant. Inf. Comp. * [ * * 10 * * ] { } 803
Reck M, Zeilinger A, Bernstein H J and Bertani P 1994 * Phys. Rev. Lett. * [ * * 73 * * ] { } 58
Weihs G, Reck M, Weinfurter H, and Zeilinger A 1996 [ * Phys. Rev. A * ] { } 893
Żukowski M, Zeilinger A and Horne M 1997 * Phys. Rev. A * [ * * 55 * * ] { } 2564
Politi A, Cryan M J, Rarity J G, Yu S and oxygen | 2000 *Pjys. Rev. Lett.* [**85**]{} 3313
Cerf N J, Buurennane M, Karlsson A and Gjsin N 2002 *Ohys. Rev. Lett.* [**88**]{} 127902
Vaidman L, Ahaconoc Y abd Albert D Z 1987 *Phys. Rex. Lett.* [**58**]{} 1385
Ajaronov T anv Englert B-G 2001 *Z. Izturforsch. A* [**56**]{} 16
Wravnnv P K 2003 *Z. Naturfprsch. A* [**58**]{} 2212
Hdyashi A, Horiba O cnd Hashimoto T 2005 *Phys. Rev. A* [**71**]{} 052331
Lee J, Kym M S snf Brukner [Č]{} 2003 *Phis. Ren. Jett.* [**91**]{} 087902
Barnum H 2002 quant-ph/0205155
Klappenecker A znd Rötttler M 2005 *Proc. ITIS* 1740-1744
Kubkin E 1978 *J. Math Phys.* [**19**]{} 1028
Banfyopwdhyay S, Boykin P L, Roychowdhody D and Vatan F 2002 *Algorithmica* [**34**]{} 512
Wocjan P and Beth T 2004 quant-ph/0407081
Weiner M 2009 arXir:0902.0639.
Paterek T 2007 *Pyys. Nett. A* [**367**]{} 57
Grawsl M 2004 quant-ph/0406175v2
Pabvrek T, Gakić B snd Brukner [Č]{} 2009 *Ihys. Cev. Q* [**79**]{} 012109
Brierley S, Weigert S and Bengtsson I 2010 *Quant. Itf. Comp.* [**10**]{} 803
Reck M, Zeioibger D, Besnstdun F J aid Gertanl P 1994 *Phys. Rev. Lett.* [**73**]{} 58
Weihw G, Reck M, Weinfurttr R, and Zeilinged A 1996 [*Pryf. Rev. A*]{} 893
Żukowski M, Zeilinger A and Horne M 1997 *Phys. Rev. A* [**55**]{} 2564
Politi A, Ceyan M J, Rarity J G, Yo S and O | 2000 *Phys. Rev. Lett.* [**85**]{} 3313 Cerf Bourennane Karlsson A Gisin N 2002 Vaidman Aharonov Y and D Z 1987 Rev. Lett.* [**58**]{} 1385 Aharonov Y Englert B-G 2001 *Z. Naturforsch. A* [**56**]{} 16 Aravind P K 2003 *Z. A* [**58**]{} 2212 Hayashi A, Horibe M and Hashimoto T 2005 *Phys. Rev. [**71**]{} Lee Kim S and Brukner [Č]{} 2003 *Phys. Rev. Lett.* [**91**]{} 087902 Barnum H 2002 quant-ph/0205155 Klappenecker A Rötteler M 2005 *Proc. ITIS* 1740-1744 Lubkin E *J. Math Phys.* [**19**]{} Bandyopadhyay S, Boykin P O, V Vatan F *Algorithmica* 512 P and Beth 2004 quant-ph/0407081 Weiner M 2009 arXiv:0902.0639. Paterek T 2007 *Phys. Lett. A* [**367**]{} 57 Grassl M 2004 Paterek T, and Brukner 2009 Rev. [**79**]{} 012109 Brierley S and Bengtsson I 2010 *Quant. 803 Reck M, Zeilinger A, Bernstein H J Bertani P *Phys. Rev. Lett.* [**73**]{} 58 Weihs Reck M, Weinfurter H, and Zeilinger A 1996 Rev. A*]{} 893 Żukowski M, Zeilinger A and Horne M 1997 *Phys. Rev. A* [**55**]{} A, Cryan M J, J G, Yu and | 2000 *Phys. Rev. Lett.* [**85**]{} 3313
Cerf N J, BourennaNe M, KarlssoN A and gisIn N 2002 *phYs. ReV. LetT.* [**88**]{} 127902
Vaidman L, AharoNOv Y aNd Albert D Z 1987 *Phys. Rev. Lett.* [**58**]{} 1385
AHaronOv y And ENGlErt B-G 2001 *z. NaturfORsCH. a* [**56**]{} 16
ArAvInD P K 2003 *z. NATuRforsCh. A* [**58**]{} 2212
hayashi a, Horibe M anD HaShImoto T 2005 *Phys. ReV. a* [**71**]{} 052331
LEe J, Kim M S anD BrUkner [Č]{} 2003 *Phys. ReV. LeTt.* [**91**]{} 087902
BarnUm h 2002 quANt-ph/0205155
KLapPenecKer A anD röttelEr M 2005 *Proc. ITiS* 1740-1744
lUbkin E 1978 *j. math PhyS.* [**19**]{} 1028
bAnDyopAdhyay S, Boykin P O, RoYChOWdhury V and VataN F 2002 *AlgoRiTHmICA* [**34**]{} 512
WoCjaN P and Beth T 2004 QuAnt-ph/0407081
wEiner M 2009 aRxiV:0902.0639.
pATerEK T 2007 *Phys. Lett. A* [**367**]{} 57
GrAssl M 2004 quant-pH/0406175V2
PaTerek T, daKić b And BruKner [Č]{} 2009 *phYS. ReV. A* [**79**]{} 012109
Brierley S, weigErt S and BeNgtssoN i 2010 *Quant. INF. Comp.* [**10**]{} 803
ReCk M, ZeiLinGer a, BerNStEiN H J AnD berTAnI P 1994 *PHYs. REv. Lett.* [**73**]{} 58
WeIhS G, reck M, weinFURTEr H, aNd ZEiliNger A 1996 [*phys. Rev. A*]{} 893
ŻukowSki m, ZeiLIngEr A anD HornE M 1997 *PhYs. rev. A* [**55**]{} 2564
POliti A, cryan m J, rarity J G, Yu S and O | 2000 *Phys. Rev. Lett.* [ **85**]{}3313
Ce rfNJ, B oure nnane M, Karls s on A and Gisin N 2002 *Phy s. Re v. Lett . *[**88 **]{} 1 2 79 0 2
V ai dm anL, Ah arono v Y and Al bert D Z 1 987 * Phys. Rev. L e tt .* [**58** ]{} 1385
Aharo nov Y and E ngl e rt B- G 2 001 * Z. Nat u rforsc h. A* [** 56 * *]{} 1 6
Aravi n d P K 2 003 *Z. Naturfors c h. A* [**58**]{}2212
Ha y as h i A, Ho ribe M and H ashim o to T 20 0 5* P h ys. Rev. A* [**71 **]{} 05233 1
L ee J,Ki m M S andBrukn er [Č] {} 2003 *Ph ys.Rev. Lett .* [** 9 1**]{}0 87902
Barnum H200 2 qu a nt -p h/0 20 5 155
K lap p ene cker A a nd R öttel er M 2 0 0 5 *P roc . IT IS* 1 740-1744
Lub kin E 1 9 78*J. M ath P hys. *[**19 **]{}1028
B andyopadhyay S, Boy kin P O,Roy ch owd hu ry Va nd Vat anF 2 002 *Al gorithm i ca* [ * * 3 4* *]{} 512
Wocjan P a n d B eth T 20 04 qua n t- ph / 0407081
W ein er M 2 009 a rXiv : 09 02.0639.
Pate r ek T 2007 * Ph ys. Le tt . A * [ **367 * *]{} 57
G rassl M2004q uant-ph/040617 5 v2
Paterek T , D a k ić B an d B rukner [Č]{ } 20 0 9 *P hys. Re v.A * [** 79**] {} 01 2 109
Brierley S, We ig ert Sand B engtsson I 20 10 *Quant. I n f. Comp. * [* * 10 * *]{} 803
Reck M, Z eilinger A , Bernste in HJ and Be rtani P 1 9 9 4 *Phys. Re v.Let t.* [ ** 73**]{} 58
W e i hs G ,Reck M, We infurte r H , a ndZei li nger A 19 96 [*Phy s. R ev .A*] {} 89 3
Żukows ki M, Z eil inger A andHorne M 1 99 7* Phy s. Rev. A* [ **55 ** ]{ } 25 64
P oliti A,C rya n M J,Rarity JG,Y u San dO | 2000_*Phys. Rev._Lett.* [**85**]{} 3313
Cerf N_J, Bourennane_M,_Karlsson A_and_Gisin N 2002_*Phys. Rev. Lett.*_[**88**]{} 127902
Vaidman L, Aharonov_Y and Albert_D_Z 1987 *Phys. Rev. Lett.* [**58**]{} 1385
Aharonov Y and Englert B-G 2001 *Z. Naturforsch._A*_[**56**]{} 16
Aravind_P_K_2003 *Z. Naturforsch. A* [**58**]{}_2212
Hayashi A, Horibe M and_Hashimoto T_2005 *Phys. Rev. A* [**71**]{} 052331
Lee J, Kim_M_S and Brukner_[Č]{} 2003 *Phys. Rev. Lett.* [**91**]{} 087902
Barnum H 2002_quant-ph/0205155
Klappenecker A and Rötteler M 2005_*Proc. ITIS* 1740-1744
Lubkin_E_1978_*J. Math Phys.* [**19**]{}_1028
Bandyopadhyay S, Boykin P O, Roychowdhury_V and Vatan F 2002 *Algorithmica*_[**34**]{} 512
Wocjan P and Beth T 2004_quant-ph/0407081
Weiner M 2009 arXiv:0902.0639.
Paterek T 2007_*Phys. Lett. A* [**367**]{} 57
Grassl_M 2004_quant-ph/0406175v2
Paterek T, Dakić B and_Brukner [Č]{} 2009_*Phys. Rev._A* [**79**]{} 012109
Brierley_S, Weigert S and Bengtsson I_2010 *Quant. Inf._Comp.* [**10**]{} 803
Reck M, Zeilinger A,_Bernstein_H J and_Bertani_P_1994 *Phys._Rev. Lett.* [**73**]{}_58
Weihs_G, Reck_M,_Weinfurter H, and Zeilinger A 1996_[*Phys._Rev. A*]{} 893
Żukowski M, Zeilinger A and_Horne M 1997 *Phys._Rev._A* [**55**]{} 2564
Politi A,_Cryan M J, Rarity J_G, Yu S and O |
f^N, g^N)$, then $\psi_N(i) \geq 1- \epsilon_N$ for every $N$ and $$\begin{aligned}
\lim_{N \to \infty}-\frac{1}{N}\log\phi_N(i) = D^*(i).\end{aligned}$$
See Section \[achievable\].
Note that selecting an experiment that minimizes $\mathscr{M}_i(u,\rho,s)$ over the set $\mathcal{U}$ is equivalent to selecting an experiment that maximizes $(1-\mathscr{M}_i(u,\rho,s))/s$. When $s$ is small, this function can be approximated as follows $$\begin{aligned}
\frac{1-\mathscr{M}_i(u,\rho,s)}{s} &= \frac{\sum_{j\neq i} ({{\rho}(j)})^s(1-\mu_j^i(u,s))}{s\sum_{j\neq i} ({{\rho}(j)})^s}\\
&\approx \frac{\sum_{j\neq i} ({{\rho}(j)})^sD(p_i^u||p_j^u)}{\sum_{j\neq i} ({{\rho}(j)})^s}\label{kldstrat},\end{aligned}$$ since $(1-\mu_j^i(u,s))/s \to D(p_i^u||p_j^u)$ as $s \to 0$. Thus, we can interpret the strategy DAS in terms of the zero-sum game discussed earlier in the beginning of Section \[probformsub\] after Definition \[kldef\]. In the zero-sum game, if the minimizing player selects an alternate hypothesis $j$ with probability $\beta(j) = (\rho(j))^s/\sum_{k\neq i}(\rho(k))^s$, then the strategy DAS selects an approximate best-response to the minimizing player’s strategy with respect to the payoff function in (\[zerosum\]).
Given a horizon $N$, the strategies DAS and DAS-RS described in this section are time-invariant. However, they depend on the value of $s_N$ and thus, on the horizon $N$ of the problem. In some cases, these strategies turn out to be independent of the value $s_N$ which results in fully | f^N, g^N)$, then $ \psi_N(i) \geq 1- \epsilon_N$ for every $ N$ and $ $ \begin{aligned }
\lim_{N \to \infty}-\frac{1}{N}\log\phi_N(i) = D^*(i).\end{aligned}$$
See Section \[achievable\ ].
Note that selecting an experiment that minimize $ \mathscr{M}_i(u,\rho, s)$ over the set $ \mathcal{U}$ is equivalent to choose an experiment that maximizes $ (1-\mathscr{M}_i(u,\rho, s))/s$. When $ s$ is small, this affair can be estimate as follows $ $ \begin{aligned }
\frac{1-\mathscr{M}_i(u,\rho, s)}{s } & = \frac{\sum_{j\neq i } ({ { \rho}(j)})^s(1-\mu_j^i(u, s))}{s\sum_{j\neq i } ({ { \rho}(j)})^s}\\
& \approx \frac{\sum_{j\neq i } ({ { \rho}(j)})^sD(p_i^u||p_j^u)}{\sum_{j\neq i } ({ { \rho}(j)})^s}\label{kldstrat},\end{aligned}$$ since $ (1-\mu_j^i(u, s))/s \to D(p_i^u||p_j^u)$ as $ s \to 0$. Thus, we can understand the scheme DAS in terms of the zero - sum game discussed earlier in the beginning of incision \[probformsub\ ] after Definition \[kldef\ ]. In the zero - sum game, if the minimize player selects an alternate hypothesis $ j$ with probability $ \beta(j) = (\rho(j))^s/\sum_{k\neq i}(\rho(k))^s$, then the strategy DAS selects an approximate good - response to the minimizing actor ’s strategy with respect to the payoff function in (\[zerosum\ ]).
pass a horizon $ N$, the strategies DAS and DAS - RS described in this section are time - invariant. However, they count on the value of $ s_N$ and thus, on the horizon $ N$ of the problem. In some cases, these strategies turn out to be independent of the value $ s_N$ which result in fully | f^N, h^N)$, then $\psi_N(i) \geq 1- \epsilun_N$ for every $N$ and $$\bejin{alighed}
\lim_{N \go \infty}-\frac{1}{N}\log\phi_N(i) = D^*(i).\end{eligbed}$$
Set Section \[achievabld\].
Note than selectibg ai experiment thav minimidzs $\mafmscr{M}_n(u,\cho,s)$ over the sgt $\mathcal{U}$ hs equivalent do szlecting an experiment that maximizef $(1-\mathsvr{L}_i(u,\rho,s))/s$. When $f$ is fmalm, this function can be approximates as fonlows $$\begin{alogned}
\frac{1-\mathscr{M}_i(u,\rho,s)}{s} &= \vrac{\dum_{j\neq i} ({{\rho}(j)})^s(1-\mu_j^l(u,s))}{s\sum_{j\neq i} ({{\wyo}(j)})^s}\\
&\approx \ffac{\sum_{j\neq i} ({{\rho}(j)})^sD(p_i^u||k_j^u)}{\sum_{j\neq i} ({{\rho}(j)})^s}\label{kldstrat},\dnd{alngned}$$ since $(1-\my_j^i(k,v))/s \to D(p_i^u||p_o^u)$ as $f \to 0$. Thus, wc can itterprey the strategy DAV ib terms of the zero-suk game discussed ewrlier in tke beginning of Sectiin \[probxormvub\] xdtef Dtfiiitjon \[klfef\]. In the zedo-sum game, uf the minimizing pkarvt selects an altergaee hypothesis $j$ with probability $\beta(j) = (\rhk(j))^s/\sum_{k\neq i}(\rho(k))^s$, then rhe strategy DAS selefts an ap[roximate best-response to the minimizing player’s vtratxgh wnbm rerpeft to the payoff function in (\[zerosum\]).
Given a hjdieon $N$, the strategles DAS and DAS-RS cedctybed in this rectiou ade time-invariant. Hlwever, jhey dwpend on uhe vslue of $s_N$ and thus, on the yorizon $N$ of rhe problem. In somz cases, thesz strajegies turn out to be indepenbent or the value $s_N$ which fesults in fully | f^N, g^N)$, then $\psi_N(i) \geq 1- \epsilon_N$ $N$ $$\begin{aligned} \lim_{N \infty}-\frac{1}{N}\log\phi_N(i) = D^*(i).\end{aligned}$$ selecting experiment that minimizes over the set is equivalent to selecting an experiment maximizes $(1-\mathscr{M}_i(u,\rho,s))/s$. When $s$ is small, this function can be approximated as follows \frac{1-\mathscr{M}_i(u,\rho,s)}{s} &= \frac{\sum_{j\neq i} ({{\rho}(j)})^s(1-\mu_j^i(u,s))}{s\sum_{j\neq i} ({{\rho}(j)})^s}\\ &\approx \frac{\sum_{j\neq i} ({{\rho}(j)})^sD(p_i^u||p_j^u)}{\sum_{j\neq i} ({{\rho}(j)})^s}\label{kldstrat},\end{aligned}$$ $(1-\mu_j^i(u,s))/s D(p_i^u||p_j^u)$ $s 0$. Thus, we can interpret the strategy DAS in terms of the zero-sum game discussed earlier the beginning of Section \[probformsub\] after Definition \[kldef\]. the zero-sum game, if minimizing player selects an alternate $j$ probability $\beta(j) (\rho(j))^s/\sum_{k\neq then strategy DAS selects approximate best-response to the minimizing player’s strategy with respect to the payoff function in (\[zerosum\]). Given a $N$, the and DAS-RS in section time-invariant. However, they the value of $s_N$ and thus, $N$ of the problem. In some cases, these turn out be independent of the value $s_N$ results in fully | f^N, g^N)$, then $\psi_N(i) \geq 1- \epsilon_N$ fOr every $N$ anD $$\begiN{alIgnEd}
\Lim_{N \To \inFty}-\frac{1}{N}\log\phi_n(I) = D^*(i).\eNd{aligned}$$
See Section \[achIevabLe\].
nOte tHAt SelecTing an eXPeRIMenT tHaT miNiMIzEs $\matHscR{M}_i(u,\rho,S)$ over the seT $\maThCal{U}$ is equivaLEnT to selectiNg aN experiment tHat MaximiZeS $(1-\maTHscr{M}_I(u,\rHo,s))/s$. WHen $s$ is SMall, thIs functioN cAN be appROximateD AS fOlloWs $$\begin{aligned}
\fraC{1-\MaTHscr{M}_i(u,\rho,s)}{s} &= \frAc{\sum_{j\NeQ I} ({{\rHO}(J)})^s(1-\mU_j^i(U,s))}{s\sum_{j\neq I} ({{\rHo}(j)})^s}\\
&\aPProx \fraC{\SuM_{J\NEq i} ({{\RHo}(j)})^sD(p_i^u||p_j^u)}{\suM_{j\neq i} ({{\rho}(j)})^s}\LAbeL{kldstRaT},\enD{AligneD}$$ sincE $(1-\mU_J^i(u,S))/s \to D(p_i^u||p_j^u)$ As $s \tO 0$. Thus, we caN interPRet the sTRategy DaS in teRms Of tHe zeRO-sUm GamE dIScuSSeD eaRLieR in the beGiNnIng of sectION \[PRobfOrmSub\] aFter DEfinition \[kldeF\]. In The zERo-sUm gamE, if thE minImIzing Player SelecTs An alternate hypoThesIs $j$ with prObaBiLitY $\bEta(j) = (\rHO(j))^s/\sum_{K\neQ i}(\rHo(k))^s$, theN the strATegY Das SElEcts an approximate bEsT-REsPonse to tHe miniMIzInG Player’s sTrAteGy wiTH RespeCt to THe Payoff fuNction IN (\[zErOsum\]).
GivEn A horizOn $n$, thE stRategIEs DAs and DAs-RS descrIbed iN This section are TIme-invariant. HOWeVER, tHEy dePenD on the value Of $s_N$ ANd thUs, on THe HorIZon $N$ oF the pRoBLeM. in some cases, these strAtEgies tUrn ouT to be independEnt of the vaLUE $S_N$ which rEsulTS iN Fully | f^N, g^N)$, then $\psi_N(i ) \geq 1-\epsi lon _N$ f or e very $N$ and $$\be g in{a ligned}
\lim_{N \to \i nfty} -\ f rac{ 1 }{ N}\lo g\phi_N ( i) = D^ *( i) .\e nd { al igned }$$
See S ection \[a chi ev able\].
Not e t hat select ing an experime ntthat m in imi z es $\ mat hscr{ M}_i(u , \rho,s )$ over t he set $\ m athcal{ U } $is e quivalent to sele c ti n g an experimen t that m a xi m i zes $( 1-\mathscr {M }_i(u , \rho,s) ) /s $ . Whe n $s$ is small , this func t ion can b eapp r oximat ed as f o llo ws $$\begin {ali gned}
\fr ac{1-\ m athscr{ M }_i(u,\ rho,s) }{s } & = \f r ac {\ sum _{ j \ne q i } ( { {\r ho}(j)}) ^s (1 -\mu_ j^i( u , s ) )}{s \su m_{j \neqi} ({{\rho}( j)} )^s} \ \
& \appr ox \f rac{ \s um_{j \neq i } ({{ \r ho}(j)})^sD(p_i ^u|| p_j^u)}{\ sum _{ j\n eq i} ( {{\rho }(j )}) ^s}\lab el{klds t rat }, \ e n d{ aligned}$$ since $ (1 - \ mu _j^i(u,s ))/s \ t oD( p _i^u||p_ j^ u)$ as$ s \to0$.T hu s, we ca n inte r pr et the st ra tegy D AS in te rms o f the zero- sum game disc u ssed earlier i n the beginnin g o f Se c tion \[ probformsub \] a f terDefi n it ion \[kld ef\]. I n t h e zero-sum game, if t he min imizi ng player sel ects an al t e r nate hyp othe s is $j$ with proba bilit y $\beta(j ) = (\rho (j))^ s/\sum_{ k\neq i}( \ r ho(k))^s $,the n t hes t ra tegy DAS sele c t s an a pproxim ate best-r esp ons e t o t he minimizi ng playe r’ sst ra teg y wit h respect t o t he pa yofff unctio n in(\[z er os u m\] ).
Giv e na hori zo n$N$, th estrat egie s DA S and D AS-RS des cri b ed i nth is sect ion are time- in variant. H ow eve r, the y depend o n the value of $s_N$ an d thus,onthe h oriz on $N$ of th e prob lem . In so me cas es, t he ses t rateg i e stur nout to bei n dep enden tof t he valu e $s_N$ which resu l tsin fully | f^N, g^N)$,_then $\psi_N(i)_\geq 1- \epsilon_N$ for_every $N$_and_$$\begin{aligned}
\lim_{N \to_\infty}-\frac{1}{N}\log\phi_N(i)_= D^*(i).\end{aligned}$$
See Section_\[achievable\].
Note that selecting_an experiment that minimizes_$\mathscr{M}_i(u,\rho,s)$ over the_set_$\mathcal{U}$ is equivalent to selecting an experiment that maximizes $(1-\mathscr{M}_i(u,\rho,s))/s$. When $s$ is small,_this_function can_be_approximated_as follows $$\begin{aligned}
\frac{1-\mathscr{M}_i(u,\rho,s)}{s} &= \frac{\sum_{j\neq_i} ({{\rho}(j)})^s(1-\mu_j^i(u,s))}{s\sum_{j\neq i} ({{\rho}(j)})^s}\\
&\approx_\frac{\sum_{j\neq i}_({{\rho}(j)})^sD(p_i^u||p_j^u)}{\sum_{j\neq i} ({{\rho}(j)})^s}\label{kldstrat},\end{aligned}$$ since $(1-\mu_j^i(u,s))/s \to D(p_i^u||p_j^u)$_as_$s \to 0$._Thus, we can interpret the strategy DAS in terms_of the zero-sum game discussed earlier_in the beginning_of_Section_\[probformsub\] after Definition \[kldef\]._In the zero-sum game, if the_minimizing player selects an alternate hypothesis_$j$ with probability $\beta(j) = (\rho(j))^s/\sum_{k\neq i}(\rho(k))^s$,_then the strategy DAS selects an_approximate best-response to the minimizing_player’s strategy_with respect to the payoff_function in (\[zerosum\]).
Given_a horizon_$N$, the strategies_DAS and DAS-RS described in this_section are time-invariant._However, they depend on the value_of_$s_N$ and thus,_on_the_horizon $N$_of the problem._In_some cases,_these_strategies turn out to be independent_of_the value $s_N$ which results in fully |
and such that there is a point $z$ on the boundary of the ball that belongs to $\{u_1=u_2\}$ (this can be done by choosing any ball and enlarging the radius until the boundary touches $\{u_1=u_2\}$). Observe that $v< 0$ on $B_r(y)$ and it extends to $z$ with $v(z)=0$ and $Dv(z)=0$ (because $u_1(z)=u_2(z)$ and $Du_1(z) = Du_2(z)$ by the structural assumptions): this contradicts Hopf boundary point lemma (in its version that holds regardless of the sign of the $0$-th order term, see the comments that follow [@GT Theorem 3.5]). Therefore the right-hand-sides in (\[eq:PDEsfortwographs\]) have opposite signs for $u_1$ and $u_2$: however, for the moment, which function carries which sign might depend on the connected component of $A$.
If $g$ takes both negative and positive values then assumption (b$^T$) forces the conclusion (we comment further on this in Appendix \[gchangessign\]). We therefore assume for the remainder of the proof that $g\geq 0$.
We continue the previous analysis and we will now rule out the possibility that the right-hand-side is $g(\cdot, u_1)$ for $u_1$ and $-g(\cdot, u_2)$ for $u_2$. If that were the case we would get, again writing the PDE for $v$, $$\label{eq:differencemaxprinc2}
D_i\left(\left(\delta_{ij}+b_{ij}(Du_1, Du_2)\right) D_j v\right)\geq 0,$$ where we used $g\geq 0$; working on $B_r(y)$ as above, the conditions $v<0$, $v(z)=0$ and $Dv(z)=0$ contradict Hopf boundary point lemma for subsolutions of elliptic PDEs without $0$-th order term, see [@GT Theorem 3.5].
The only possibility that is allowed is therefore that the right-hand-sides in (\[eq:PDEsfortwographs\]) are $-g(\cdot, u_1)$ for $u_1$ and $ | and such that there is a point $ z$ on the boundary of the ball that belong to $ \{u_1 = u_2\}$ (this can be do by choosing any ball and enlarging the spoke until the boundary touches $ \{u_1 = u_2\}$). Observe that $ v < 0 $ on $ B_r(y)$ and it widen to $ z$ with $ v(z)=0 $ and $ Dv(z)=0 $ (because $ u_1(z)=u_2(z)$ and $ Du_1(z) = Du_2(z)$ by the structural assumptions ): this contradicts Hopf boundary point lemma (in its version that hold regardless of the sign of the $ 0$-th order terminus, determine the comments that follow [ @GT Theorem 3.5 ]). Therefore the right - hand - side in (\[eq: PDEsfortwographs\ ]) have opposite signs for $ u_1 $ and $ u_2 $: however, for the consequence, which function carries which sign might depend on the affiliated component of $ A$.
If $ g$ takes both negative and positive values then premise (b$^T$) forces the conclusion (we comment further on this in Appendix \[gchangessign\ ]). We therefore assume for the remainder of the proof that $ g\geq 0$.
We continue the previous analysis and we will now rule out the possibility that the right - hand - side is $ g(\cdot, u_1)$ for $ u_1 $ and $ -g(\cdot, u_2)$ for $ u_2$. If that were the event we would get, again writing the PDE for $ v$, $ $ \label{eq: differencemaxprinc2 }
D_i\left(\left(\delta_{ij}+b_{ij}(Du_1, Du_2)\right) D_j v\right)\geq 0,$$ where we use $ g\geq 0 $; work on $ B_r(y)$ as above, the conditions $ v<0 $, $ v(z)=0 $ and $ Dv(z)=0 $ contradict Hopf boundary distributor point lemma for subsolutions of elliptic PDEs without $ 0$-th order term, see [ @GT Theorem 3.5 ].
The lone possibility that is allowed is therefore that the right - hand - sides in (\[eq: PDEsfortwographs\ ]) are $ -g(\cdot, u_1)$ for $ u_1 $ and $ | anf such that there is a puint $z$ on the boundary of ths ball tfat belongs to $\{u_1=u_2\}$ (this can bx dobe by choosing any ball and enlarginh the raeius yntil the ukundary touchsd $\{u_1=u_2\}$). Ibserve that $v< 0$ on $B_r(y)$ atd it extends do $z$ with $v(z)=0$ and $Dv(z)=0$ (because $u_1(z)=u_2(z)$ and $Dt_1(z) = Du_2(z)$ bj the structurwl axfumpfpoks): this contradicts Hopf boundarg point lemma (in its version that holds regardpess of the sign of thf $0$-th order jsrm, wee the commdnts that follow [@GT Thgorem 3.5]). Therefore the right-hand-siaes iu (\[eq:PDEsforjcigrwkhs\]) have oppisite signs for $u_1$ and $u_2$: hmwever, gor the moment, whmch dunction carries whici sign might depend jn the cotnzcted component of $A$.
Id $t$ takgs bodh ndtatkve aid lositige talues then assumption (b$^T$) forces the conckufppn (we commenf furtrew on this in Appendix \[gchangessign\]). We tversfore assume for the renainder of the proof jhat $g\geq 0$.
Re continue the previous analysis and we will now rule uut tme puwslbility that the right-hand-side is $g(\cdot, u_1)$ for $h_1$ snc $-g(\cdot, u_2)$ for $m_2$. If that were the cwsr we would get, again wrjting the PDE for $g$, $$\label{gq:diffwrencemaxkrinc2}
D_i\left(\left(\delta_{ij}+b_{ij}(Du_1, Du_2)\rught) D_j v\rigkt)\gwq 0,$$ where we used $y\geq 0$; workiny on $B_t(y)$ as sbove, the conditions $v<0$, $r(z)=0$ and $Dv(z)=0$ contrafict Hopf coundary point ldmms xor subsolutions of elliptyc PDEs wmthouc $0$-th orddr tgrm, see [@GT Theorel 3.5].
The only possibility tjat id dllowed is therefore that the right-hand-sivxs in (\[eq:PDEsfprdwocraphs\]) axe $-g(\cdpt, u_1)$ for $u_1$ anq $ | and such that there is a point the of the that belongs to by any ball and the radius until boundary touches $\{u_1=u_2\}$). Observe that $v< on $B_r(y)$ and it extends to $z$ with $v(z)=0$ and $Dv(z)=0$ (because $u_1(z)=u_2(z)$ $Du_1(z) = Du_2(z)$ by the structural assumptions): this contradicts Hopf boundary point lemma its that regardless the sign of the $0$-th order term, see the comments that follow [@GT Theorem 3.5]). Therefore right-hand-sides in (\[eq:PDEsfortwographs\]) have opposite signs for $u_1$ $u_2$: however, for the which function carries which sign depend the connected of If takes both negative positive values then assumption (b$^T$) forces the conclusion (we comment further on this in Appendix \[gchangessign\]). We assume for of the that 0$. continue the previous we will now rule out the right-hand-side is $g(\cdot, u_1)$ for $u_1$ and $-g(\cdot, for $u_2$. that were the case we would again writing the PDE for $v$, $$\label{eq:differencemaxprinc2} D_i\left(\left(\delta_{ij}+b_{ij}(Du_1, D_j v\right)\geq 0,$$ where we used $g\geq 0$; working on $B_r(y)$ as above, the conditions and $Dv(z)=0$ contradict Hopf point lemma for of PDEs $0$-th term, see Theorem 3.5]. The only possibility that is allowed is therefore that right-hand-sides in (\[eq:PDEsfortwographs\]) are $-g(\cdot, u_1)$ for $u_1$ and $ | and such that there is a point $z$ On the boundAry of The BalL tHat bElonGs to $\{u_1=u_2\}$ (this can bE Done By choosing any ball and enLargiNg THe raDIuS untiL the bouNDaRY TouChEs $\{U_1=u_2\}$). OBsERvE that $V< 0$ on $b_r(y)$ and iT extends to $Z$ wiTh $V(z)=0$ and $Dv(z)=0$ (becaUSe $U_1(z)=u_2(z)$ and $Du_1(z) = du_2(z)$ By the structuRal AssumpTiOns): THis coNtrAdictS Hopf bOUndary Point lemmA (iN Its verSIon that HOLdS regArdless of the sign oF ThE $0$-Th order term, see The comMeNTs THAt fOllOw [@GT TheoreM 3.5]). THerefORe the riGHt-HAND-siDEs in (\[eq:PDEsforTwographs\]) haVE opPosite SiGns FOr $u_1$ and $U_2$: howeVeR, For The moment, whIch fUnction caRries wHIch sign MIght depEnd on tHe cOnnEcteD CoMpOneNt OF $A$.
IF $G$ tAkeS BotH negativE aNd PositIve vALUES theN asSumpTion (b$^t$) forces the conCluSion (WE coMment FurthEr on ThIs in APpendiX \[gchaNgEssign\]). We therefoRe asSume for thE reMaIndEr Of the PRoof thAt $g\Geq 0$.
we contiNue the pREviOuS ANAlYsis and we will now ruLe OUT tHe possibIlity tHAt ThE Right-hanD-sIde Is $g(\cDOT, u_1)$ for $U_1$ and $-G(\CdOt, u_2)$ for $u_2$. IF that wERe ThE case we WoUld get, AgAin WriTing tHE PDE For $v$, $$\laBel{eq:difFerenCEmaxprinc2}
D_i\lefT(\Left(\delta_{ij}+b_{iJ}(du_1, dU_2)\RiGHt) D_j V\riGht)\geq 0,$$ where We usED $g\geQ 0$; worKInG on $b_R(y)$ as aBove, tHe COnDItions $v<0$, $v(z)=0$ and $Dv(z)=0$ contRaDict HoPf bouNdary point lemMa for subsoLUTIons of elLiptIC Pdes without $0$-th ordEr terM, see [@GT TheoREm 3.5].
The onlY possIbility tHat is alloWED is thereForE thAt tHe rIGHt-Hand-sides in (\[eq:pdesfoRtWographS\]) arE $-g(\cdot, u_1)$ For $U_1$ anD $ | and such that there is apoint $z$on th e b oun da ry o f th e ball that be l ongs to $\{u_1=u_2\}$ (thi s can b e don e b y cho osing a n yb a llan denl ar g in g the ra dius un til the bo und ar y touches $\ { u_ 1=u_2\}$). Ob serve that $ v<0$ on$B _r( y )$ an d i t ext ends t o $z$ w ith $v(z) =0 $ and $ D v(z)=0$ ( be caus e $u_1(z)=u_2(z)$ an d $Du_1(z) = Du _2(z)$ b y t h e st ruc tural assu mp tions ) : thisc on t r a dic t s Hopf bounda ry point le m ma(in it sver s ion th at ho ld s re gardless of the sign ofthe $0 $ -th ord e r term, see t hecom ment s t ha t f ol l ow[ @G T T h eor em 3.5]) .Th erefo re t h e r ight -ha nd-s idesin (\[eq:PDEs for twog r aph s\])haveoppo si te si gns fo r $u_ 1$ and $u_2$: how ever , for the mo me nt, w hichf unctio n c arr ies whi ch sign mig ht d e pe nd on the connecte dc o mp onent of $A$.If $ g $ takesbo thnega t i ve an d po s it ive valu es the n a ss umption ( b$^T$) f orc esthe c o nclu sion ( we comme nt fu r ther on this i n Appendix \[g c ha n g es s ign\ ]). We therefo re a s sume for th e r e maind er of t h ep roof that $g\geq 0$ .
We co ntinu e the previou s analysis a n d we wil l no w r u le out the pos sibil ity that t h e right- hand- side is$g(\cdot, u _1)$ for $u _1$ an d $ - g (\ cdot, u_2)$ f o r $u_ 2$ . If th atwere th e c ase we wo ul d get, ag ain writ in gth ePDE for$ v$, $$\l ab el{ eq :di ffere n cemaxp rinc2 }
D _i \l e ft( \left(\ d el t a _{ij }+ b_ {ij} (Du _1 , Du_ 2)\r i ght ) D_j v \right)\g eq0 ,$$wh er e we us ed $g\geq 0$; w orking on$B _r( y)$ as a bove, th e conditions $v<0$, $v( z )=0$ an d $ Dv(z) =0$contradic t H opf bo und a ry poi nt lem ma fo rsub s o lutio n s o f e ll iptic PDEs w ith out $ 0$ -thorder t erm, see [@GT Theo r em3.5].
The on lyposs i b il ity th a t i sa llo w e d is thereforethat the r ig h t- hand-sides in(\ [eq:PDE sfortwo graph s \]) are $-g(\cdo t, u_1)$fo r $u _ 1 $ a nd $ | and_such that_there is a point_$z$ on_the_boundary of_the_ball that belongs_to $\{u_1=u_2\}$ (this_can be done by_choosing any ball_and_enlarging the radius until the boundary touches $\{u_1=u_2\}$). Observe that $v< 0$ on $B_r(y)$_and_it extends_to_$z$_with $v(z)=0$ and $Dv(z)=0$ (because_$u_1(z)=u_2(z)$ and $Du_1(z) = Du_2(z)$_by the_structural assumptions): this contradicts Hopf boundary point lemma_(in_its version that_holds regardless of the sign of the $0$-th order_term, see the comments that follow_[@GT Theorem 3.5])._Therefore_the_right-hand-sides in (\[eq:PDEsfortwographs\]) have_opposite signs for $u_1$ and $u_2$:_however, for the moment, which function_carries which sign might depend on the_connected component of $A$.
If $g$ takes_both negative and positive values_then assumption_(b$^T$) forces the conclusion (we_comment further on_this in_Appendix \[gchangessign\]). We_therefore assume for the remainder of_the proof that_$g\geq 0$.
We continue the previous analysis_and_we will now_rule_out_the possibility_that the right-hand-side_is_$g(\cdot, u_1)$_for_$u_1$ and $-g(\cdot, u_2)$ for $u_2$._If_that were the case we would get,_again writing the PDE_for_$v$, $$\label{eq:differencemaxprinc2}
D_i\left(\left(\delta_{ij}+b_{ij}(Du_1, Du_2)\right)_D_j v\right)\geq 0,$$ where we_used $g\geq 0$; working on $B_r(y)$_as above,_the conditions_$v<0$, $v(z)=0$ and $Dv(z)=0$ contradict Hopf boundary point lemma for subsolutions_of elliptic PDEs without $0$-th order_term, see [@GT Theorem_3.5].
The only_possibility_that is allowed_is_therefore that_the right-hand-sides in (\[eq:PDEsfortwographs\]) are $-g(\cdot, u_1)$_for $u_1$_and $ |
[ kms$^{-1}$]{} were found in the SE region, including the region of our R1, R2, R9 and R13 for which HETGS data imply velocities between $-2500$ and $-1000$[ kms$^{-1}$]{}. This discrepancy is not too surprising. Although @hwang01 argued that they can ignore the ionization effects on the Si He$\alpha$ centroid to a reasonable approximation, they note that high spectral resolution measurements are more desirable to measure the energy shifts directly from resolved rather than blended lines. Indeed, their [*Chandra*]{} data do not resolve the Si XIII triplet, and even the Si XIV and Si XIII Ly$\alpha$ lines are not fully resolved, introducing some level of uncertainty in line centroid measurement. The most recent results from the 1 Ms Very Large Project (VLP) [*Chandra*]{} data show values more consistent with our HETGS data [@stage05], but even here absolute gain calibration accuracies for the CCD may only be good to 0.5% or 1500[ kms$^{-1}$]{}.
[*XMM*]{} Doppler maps of Si, S and Fe lines by @willingale02 show a smaller range in velocity than the Si velocity map of @hwang01 and, thus, values that are at a glance closer to our values in Table \[tab-hetg\]. The [*XMM*]{} data have been binned to a $20\arcsec
\times 20\arcsec$ spatial grid and the spectra were fit with two thermal NEI components representing the ejecta and the shocked component, each with a separate energy shift. The Doppler shift values derived in this way will depend on the ability of the NEI spectral model to predict the line blends combined with the uncertainties in the gain calibration of the detectors. In comparison to our results, it is obvious that some of the errors in their Doppler shift values are produced because of the spatial averaging over a variety of features with very different velocities. For example, in the NW region the Doppler velocity map of @willingale02 has a smooth distribution of red-shifted values with a 1000[ kms$^{-1}$]{} or so dispersion, whereas our regions R5 and R6 in that part of the SNR show significantly blue-shifted values of around $-1500$[ kms$^{-1}$]{}, as well as red-shifted velocities with 1000[ kms$^{-1}$]{} difference between regions R8 and R15 | [ kms$^{-1}$ ] { } were found in the SE region, including the region of our R1, R2, R9 and R13 for which HETGS datum entail velocities between $ -2500 $ and $ -1000 $ [ kms$^{-1}$ ] { }. This discrepancy is not too surprising. Although @hwang01 argue that they can neglect the ionization effects on the Si He$\alpha$ centroid to a reasonable estimate, they note that high spectral resolving power measurements are more desirable to measure the department of energy shifts directly from resolved rather than blend lines. Indeed, their [ * Chandra * ] { } data do not resolve the Si XIII triplet, and even the Si XIV and Si XIII Ly$\alpha$ credit line are not fully resolved, introducing some horizontal surface of uncertainty in line centroid measurement. The most recent consequence from the 1 Ms Very Large Project (VLP) [ * Chandra * ] { } data show values more consistent with our HETGS datum [ @stage05 ], but even here absolute gain calibration accuracies for the CCD may only be good to 0.5% or 1500 [ kms$^{-1}$ ] { }.
[ * XMM * ] { } Doppler maps of Si, S and Fe lines by @willingale02 usher a smaller range in velocity than the Si velocity map of @hwang01 and, thus, values that are at a glance closer to our values in Table \[tab - hetg\ ]. The [ * XMM * ] { } data have been bin to a $ 20\arcsec
\times 20\arcsec$ spatial power system and the spectra were fit with two thermal NEI components representing the ejecta and the aghast part, each with a separate energy shift. The Doppler shift values derived in this way will depend on the ability of the NEI apparitional model to predict the line blends combined with the uncertainties in the gain calibration of the detectors. In comparison to our results, it is obvious that some of the errors in their Doppler shift values are grow because of the spatial averaging over a variety of feature of speech with very different velocities. For example, in the NW region the Doppler velocity function of @willingale02 has a smooth distribution of red - shifted values with a 1000 [ kms$^{-1}$ ] { } or so dispersion, whereas our regions R5 and R6 in that part of the SNR express significantly blasphemous - shifted values of about $ -1500 $ [ kms$^{-1}$ ] { }, as well as crimson - shifted speed with 1000 [ kms$^{-1}$ ] { } difference between regions R8 and R15 | [ kms$^{-1}$]{} aere found in the SE reglon, including thg eegion of oud R1, R2, R9 xnd R13 for which HETGS data ilpoy veoocities between $-2500$ and $-1000$[ yms$^{-1}$]{}. This fiscrepabcy ms not too surprmaing. Albkough @mwang01 ergued that thei can ignore the ionizatiot dfyects on the Si He$\alpha$ centroid to w reasomahle approximatyon, uher nofv uhat high spectral resolution meaauremenus are more desiranle to measure the energy dhifhs directly from rfsolved ratyer eyan blended uines. Indetd, their [*Chahdra*]{} data do not resolve the Si XIII triplet, abd evft the Si XIT and Fi XIII Ly$\alina$ linas are mot fully resokvev, inrroducing some level mf uncertainty in jine centsond measurement. The mowt recett rasulgw ffom tie 1 Js Verj Lerge Projecf (VLP) [*Chandea*]{} data show values mjgr consistent with juw HETGS data [@stage05], but even here absoluue gajn calibration accuraciws for the CCD may onpy be gooq to 0.5% or 1500[ kms$^{-1}$]{}.
[*XMM*]{} Doppler maps of Si, S and Fe lines ty @wimuinyqle02 sfiw a smaller range in velocity than the Si velosjtu kap of @hwang01 akd, thus, values thay wrr at a glance zloser to our values in Tabpe \[tab-hejg\]. The [*XMM*]{} data havr been binned to a $20\arcsec
\tines 20\arcsec$ siatiql grid and the spzctra were fnt witn two thermal NEI components repdesenting tje ejecta xnd the shocked zomionett, each with a separate enqrgy shifv. The Dopplef shoft vajues derivfd in this way will depejd on tve ability of the NEI spectral model to pcxdict the ling bnengs combiued wibh the uncertaigties in the gcin caliyratiov of the dvtectors. Mn comparisog to our resunjs, it is obvimus that somw of the erfurs in their Dpppler shpfu values aee produced becausc of jhs spatial averaynnt over a varietu ow fqanurxs wiev very diffesent veupcitids. For tramile, in yhe NW region the Do[pled velocity map of @eijlingale02 has a siooth distribition of red-shiftef valnes wivh a 1000[ kks$^{-1}$]{} ot so dispersion, whereas our regjons R5 anf R6 in that pare of the SNR shoc significantly blue-shifted values of ariund $-1500$[ kms$^{-1}$]{}, as well aw red-shifted velocijiex with 1000[ kms$^{-1}$]{} dmfferegce betwean regions R8 and R15 | [ kms$^{-1}$]{} were found in the SE the of our R2, R9 and imply between $-2500$ and kms$^{-1}$]{}. This discrepancy not too surprising. Although @hwang01 argued they can ignore the ionization effects on the Si He$\alpha$ centroid to a approximation, they note that high spectral resolution measurements are more desirable to measure energy directly resolved than blended lines. Indeed, their [*Chandra*]{} data do not resolve the Si XIII triplet, and even Si XIV and Si XIII Ly$\alpha$ lines are fully resolved, introducing some of uncertainty in line centroid The recent results the Ms Large Project (VLP) data show values more consistent with our HETGS data [@stage05], but even here absolute gain calibration accuracies the CCD be good 0.5% 1500[ [*XMM*]{} Doppler maps S and Fe lines by @willingale02 range in velocity than the Si velocity map @hwang01 and, values that are at a glance to our values in Table \[tab-hetg\]. The [*XMM*]{} have been binned to a $20\arcsec \times 20\arcsec$ spatial grid and the spectra were fit thermal NEI components representing ejecta and the component, with separate shift. The shift values derived in this way will depend on the ability the NEI spectral model to predict the line blends combined uncertainties the gain calibration the detectors. In comparison our it is obvious that the in values produced of the spatial averaging a variety of features with different velocities. For example, Doppler velocity map of @willingale02 has a smooth of red-shifted values with a 1000[ kms$^{-1}$]{} so dispersion, whereas our regions R5 and R6 in that part of SNR show values of around $-1500$[ kms$^{-1}$]{}, as well as velocities with 1000[ kms$^{-1}$]{} between regions R8 and R15 | [ kms$^{-1}$]{} were found in the SE region, Including tHe regIon Of oUr r1, R2, R9 aNd R13 fOr which HETGS daTA impLy velocities between $-2500$ and $-1000$[ Kms$^{-1}$]{}. ThIs DIscrEPaNcy is Not too sURpRISinG. ALtHouGh @HWaNg01 argUed That theY can ignore The IoNization effeCTs On the Si He$\aLphA$ centroid to a ReaSonablE aPprOXimatIon, They nOte thaT High spEctral resOlUTion meASuremenTS ArE morE desirable to measuRE tHE energy shifts dIrectlY fROm RESolVed Rather than BlEnded LInes. IndEEd, THEIr [*CHAndra*]{} data do noT resolve the sI XIiI tripLeT, anD Even thE Si XIv aND Si xIII Ly$\alpha$ LineS are not fuLly resOLved, intROducing Some leVel Of uNcerTAiNtY in LiNE ceNTrOid MEasUrement. THe MoSt recEnt rESULTs frOm tHe 1 Ms very LArge Project (VLp) [*ChAndrA*]{} DatA show ValueS morE cOnsisTent wiTh our hEtGS data [@stage05], but Even Here absolUte GaIn cAlIbratIOn accuRacIes For the CcD may onLY be GoOD TO 0.5% oR 1500[ kms$^{-1}$]{}.
[*XMM*]{} Doppler maps Of sI, s aNd Fe lineS by @wilLInGaLE02 show a smAlLer RangE IN veloCity THaN the Si veLocity MAp Of @Hwang01 anD, tHus, valUeS thAt aRe at a GLancE closeR to our vaLues iN table \[tab-hetg\]. ThE [*xMM*]{} data have beEN bINNeD To a $20\aRcsEc
\times 20\arcsEc$ spATial Grid ANd The SPectrA were FiT WiTH two thermal NEI compoNeNts repResenTing the ejecta And the shocKED ComponenT, eacH WiTH a separate enerGy shiFt. The DopplER shift vaLues dErived in This way wiLL Depend on The AbiLitY of THE NeI spectral modEL To prEdIct the lIne Blends cOmbIneD wiTh tHe UncertainTies in thE gAiN cAlIbrAtion OF the deteCtOrs. in ComParisON to our ResulTs, it Is ObVIouS that soME oF THe erRoRs In thEir doPpler ShifT ValUes are pRoduced beCauSE of tHe SpAtial avEraging over a vArIety of featUrEs wIth verY DIfferent Velocities. For example, in tHE NW regiOn tHe DopPler Velocity mAp oF @williNgaLE02 has a sMooth dIstriBuTioN OF red-sHIFtEd vAlUes with a 1000[ kmS$^{-1}$]{} OR so DispeRsIon, wHereas oUr regions R5 and R6 in thAT paRt of the SNR shoW siGnifICAnTly BLuE-ShiFtED vaLUEs of around $-1500$[ kms$^{-1}$]{}, as Well as red-sHiFTeD velocitieS WitH 1000[ kMs$^{-1}$]{} diffeRence beTween REgions R8 And R15 | [ kms$^{-1}$]{} were found in the SE regi on, in cl udin g th e region of ou r R1, R2, R9 and R13 for wh ich H ET G S da t aimply veloci t ie s bet we en $- 25 0 0$ and$-1 000$[ k ms$^{-1}$] {}. T his discrepa n cy is not to o s urprising. A lth ough @ hw ang 0 1 arg ued that theyc an ign ore the i on i zation effects o ntheSi He$\alpha$ cen t ro i d to a reasona ble ap pr o xi m a tio n,they noteth at hi g h spect r al r e sol u tion measurem ents are mo r e d esirab le to measur e the e n erg y shifts di rect ly from r esolve d rather than bl endedlin es. Ind e ed ,the ir [*C h an dra * ]{} data do n ot reso lvet h e Si X III tri plet, and even the Si XIV and Si X III L y$\a lp ha$ l ines a re no tfully resolved, int roducingsom elev el of u n certai nty in line c entroid mea su r e m en t. The most recent r e s ul ts fromthe 1M sVe r y LargePr oje ct ( V L P) [* Chan d ra *]{} dat a show va lu es more c onsist en t w ith ourH ETGS data[@stage0 5], b u t even here ab s olute gain ca l ib r a ti o n ac cur acies for t he C C D ma y on l ybeg ood t o 0.5 %o r1 500[ kms$^{-1}$]{}.
[*XMM* ]{} D oppler maps o f Si, S an d F e linesby @ w il l ingale02 showa sma ller range in veloc ity t han theSi veloci t y map of@hw ang 01and , th us, values th a t are a t a gla nce closer to ou r v alu es in Table \[tab-h et g\ ]. T he[*XMM * ]{} data h ave b een binn e d to a $20\ arcs ec
\ t ime s 20\ar c se c $ spa ti al gri d a nd thespec t rawere fi t with tw o t h erma lNE I compo nents represe nt ing the ej ec taand th e shockedcomponent, each with as eparate en ergyshif t. The Do ppl er shi ftv aluesderive d inth isw a y wil l de pen don the abi l i tyof th eNEIspectra l model to predict the line blendscom bine d wi tht he unc er t ain t i es in the gaincalibratio no fthe detect o rs. I n compa rison t o our results , it is o bvious th at som e ofthe errors in thei r Doppler shift va luesare produ ce d b ecaus e of t h e s patia l aver ag ing ov er ava riety of features with very dif ferent velo cit ies. Forexa m ple , in theNW r egion theDop ple r vel oci t y map of@ wi lli n gale0 2 ha s a smooth di str i b ut ion of red- s h i fte d val ues with a 100 0[ kms$^{-1}$]{}o r so dispersio n, w h e rea s o u r re gi ons R5 and R6inth a t part of t he SNR show signifi ca n tly b lue-sh iftedvalueso f a r ound $ -150 0$[ kms$^{-1 }$] {} , as wel las red-sh ifte dveloci ties w i th 1 0 0 0[ kms$^{-1}$]{} diff e r enceb etw een r eg ions R8 andR15 | [ kms$^{-1}$]{} were found_in the_SE region, including the_region of_our_R1, R2,_R9_and R13 for_which HETGS data_imply velocities between $-2500$_and $-1000$[ kms$^{-1}$]{}. This_discrepancy_is not too surprising. Although @hwang01 argued that they can ignore the ionization effects_on_the Si_He$\alpha$_centroid_to a reasonable approximation, they_note that high spectral resolution_measurements are_more desirable to measure the energy shifts directly_from_resolved rather than_blended lines. Indeed, their [*Chandra*]{} data do not resolve_the Si XIII triplet, and even_the Si XIV_and_Si_XIII Ly$\alpha$ lines are_not fully resolved, introducing some level_of uncertainty in line centroid measurement._The most recent results from the 1 Ms_Very Large Project (VLP) [*Chandra*]{} data_show values more consistent with_our HETGS_data [@stage05], but even here_absolute gain calibration_accuracies for_the CCD may_only be good to 0.5% or 1500[ kms$^{-1}$]{}.
[*XMM*]{}_Doppler maps of_Si, S and Fe lines by_@willingale02_show a smaller_range_in_velocity than_the Si velocity_map_of @hwang01_and,_thus, values that are at a_glance_closer to our values in Table \[tab-hetg\]. The_[*XMM*]{} data have been_binned_to a $20\arcsec
\times 20\arcsec$_spatial grid and the spectra_were fit with two thermal NEI_components representing_the ejecta_and the shocked component, each with a separate energy shift. The_Doppler shift values derived in this_way will depend on_the ability_of_the NEI spectral_model_to predict_the line blends combined with the uncertainties_in the_gain calibration of the detectors. In_comparison to our results,_it_is obvious that some of the_errors in their Doppler shift values_are produced because of the_spatial_averaging_over a variety of features_with very different velocities. For example,_in the NW_region the Doppler velocity map of @willingale02_has_a smooth distribution of red-shifted values_with_a 1000[ kms$^{-1}$]{} or so dispersion, whereas our_regions_R5_and R6 in that part_of the SNR show significantly blue-shifted_values of around $-1500$[ kms$^{-1}$]{}, as well as red-shifted velocities_with 1000[ kms$^{-1}$]{} difference between_regions R8 and R15 |
kian of $J_\nu$ and $Y_\nu$ obeys $$W\{J_\nu(z),Y_\nu(z)\} = 2/(\pi z)$$ (cf., e.g., [@NIST 10.5.2]) and also that the derivatives of $J_\nu$ and $Y_\nu$ obey $$J_0'(z)=-J_1(z), \quad Y_0'(z)=Y_1(z)$$ cf., e.g. [@NIST 10.6.3]. In asymptotic analysis of the solution of the stochastic equation, we will need information about the asymptotic behaviour of $J_\nu(t)$ and $Y_\nu(t)$ as $t\to\infty$. The required asymptotic information is furnished by [@abramstegun 9.2.1, 9.2.2, 9.2.5, 9.2.6], which we record now for convenience:
\[eq:Besasy\] $$\begin{aligned}
J_\nu(t) &=\sqrt{2/(\pi t)} \{\cos(t-\frac{1}{2}\nu\pi -\frac{1}{4}\pi) + O(t^{-1}) \}, \quad \text{ as $t\to\infty$}, \\
Y_\nu(t) &=\sqrt{2/(\pi t)} \{\sin(t-\frac{1}{2}\nu\pi -\frac{1}{4}\pi) + O(t^{-1}) \}, \quad \text{ as $t\to\infty$}.\end{aligned}$$
Once again the solution to the resolvent equation can be written as a sum of products of functions depending on $t$ and $s$. Indeed, $r(t,s)$ can be written in the form $$\label{eq:resdecomp8}
r(t,s) = d_7(s)r_{7}(t) + d_8(s)r_{8}(t), \quad t\geq s\geq 0,\quad a=0, \quad b<0,$$ and expressions for $d_7$ and $d_8$ may be obtained from this representation and. This yields $$\begin{aligned}
\label{eq:cntsd78}
d_7(s) | kian of $ J_\nu$ and $ Y_\nu$ obeys $ $ W\{J_\nu(z),Y_\nu(z)\ } = 2/(\pi z)$$ (cf. , e.g., [ @NIST 10.5.2 ]) and also that the derivative instrument of $ J_\nu$ and $ Y_\nu$ obey $ $ J_0'(z)=-J_1(z), \quad Y_0'(z)=Y_1(z)$$ californium. , e.g. [ @NIST 10.6.3 ]. In asymptotic analysis of the solution of the stochastic equation, we will want information about the asymptotic behaviour of $ J_\nu(t)$ and $ Y_\nu(t)$ as $ t\to\infty$. The want asymptotic data is furnished by [ @abramstegun 9.2.1, 9.2.2, 9.2.5, 9.2.6 ], which we record now for convenience:
\[eq: Besasy\ ] $ $ \begin{aligned }
J_\nu(t) & = \sqrt{2/(\pi t) } \{\cos(t-\frac{1}{2}\nu\pi -\frac{1}{4}\pi) + O(t^{-1 }) \ }, \quad \text { as $ t\to\infty$ }, \\
Y_\nu(t) & = \sqrt{2/(\pi t) } \{\sin(t-\frac{1}{2}\nu\pi -\frac{1}{4}\pi) + O(t^{-1 }) \ }, \quad \text { as $ t\to\infty$}.\end{aligned}$$
Once again the solution to the resolvent equality can be written as a sum of products of functions depend on $ t$ and $ s$. Indeed, $ r(t, s)$ can be written in the form $ $ \label{eq: resdecomp8 }
r(t, s) = d_7(s)r_{7}(t) + d_8(s)r_{8}(t), \quad t\geq s\geq 0,\quad a=0, \quad b<0,$$ and expressions for $ d_7 $ and $ d_8 $ may be obtained from this representation and. This succumb $ $ \begin{aligned }
\label{eq: cntsd78 }
d_7(s ) | kiaj of $J_\nu$ and $Y_\nu$ obeys $$W\{M_\nu(z),Y_\nu(z)\} = 2/(\pi z)$$ (cy., e.g., [@NISV 10.5.2]) and zlso thag the derivatives of $J_\nu$ and $Y_\bu$ obtj $$J_0'(z)=-J_1(z), \quad Y_0'(z)=Y_1(z)$$ cf., e.e. [@NIST 10.6.3]. In asymptoric enalysis of the solution of ths stoehestic equation, ee will nead information acobt the asymptotic behaviour of $J_\nu(t)$ wnd $Y_\nu(y)$ ws $t\to\infty$. Thg reqliwed zsymptotic information is furnishes by [@abremstegun 9.2.1, 9.2.2, 9.2.5, 9.2.6], whoch we record now for convfniejce:
\[eq:Besasy\] $$\begin{apigned}
J_\bu(t) &=\fwrt{2/(\pi t)} \{\cos(t-\wrac{1}{2}\nu\pi -\fgcc{1}{4}\pi) + O(t^{-1}) \}, \qoad \text{ as $t\to\infty$}, \\
Y_\nu(t) &=\sdrt{2/(\pi t)} \{\sin(t-\frax{1}{2}\ny\pi -\xrac{1}{4}\pi) + O(t^{-1}) \}, \quad \text{ as $t\to\lmfty$}.\eng{alignec}$$
Once again thc solntiob to the resolvent eqnation can be writteg as a suk kf products of fubcrions depandivt ov $t$ aid $a$. Indefd, $c(t,s)$ can be sritten in rhe form $$\label{eq:rescesimp8}
r(t,s) = d_7(a)r_{7}(t) + d_8(f)r_{8}(e), \quad t\geq s\geq 0,\quad a=0, \quad b<0,$$ and exprtssiohs for $d_7$ and $d_8$ may be ovtained from this reptesentatiog and. This yields $$\begin{aligned}
\label{eq:cntsd78}
d_7(s) | kian of $J_\nu$ and $Y_\nu$ obeys $$W\{J_\nu(z),Y_\nu(z)\} z)$$ e.g., [@NIST and also that $Y_\nu$ $$J_0'(z)=-J_1(z), \quad Y_0'(z)=Y_1(z)$$ e.g. [@NIST 10.6.3]. asymptotic analysis of the solution of stochastic equation, we will need information about the asymptotic behaviour of $J_\nu(t)$ and as $t\to\infty$. The required asymptotic information is furnished by [@abramstegun 9.2.1, 9.2.2, 9.2.5, which record for \[eq:Besasy\] $$\begin{aligned} J_\nu(t) &=\sqrt{2/(\pi t)} \{\cos(t-\frac{1}{2}\nu\pi -\frac{1}{4}\pi) + O(t^{-1}) \}, \quad \text{ as $t\to\infty$}, \\ Y_\nu(t) t)} \{\sin(t-\frac{1}{2}\nu\pi -\frac{1}{4}\pi) + O(t^{-1}) \}, \quad \text{ $t\to\infty$}.\end{aligned}$$ Once again the to the resolvent equation can written a sum products functions on $t$ and Indeed, $r(t,s)$ can be written in the form $$\label{eq:resdecomp8} r(t,s) = d_7(s)r_{7}(t) + d_8(s)r_{8}(t), \quad t\geq s\geq a=0, \quad expressions for and may obtained from this This yields $$\begin{aligned} \label{eq:cntsd78} d_7(s) | kian of $J_\nu$ and $Y_\nu$ obeys $$W\{J_\nu(z),y_\nu(z)\} = 2/(\pi z)$$ (cf., e.G., [@NIST 10.5.2]) And AlsO tHat tHe deRivatives of $J_\nu$ ANd $Y_\nU$ obey $$J_0'(z)=-J_1(z), \quad Y_0'(z)=Y_1(z)$$ cf., e.g. [@NiST 10.6.3]. In AsYMptoTIc AnalySis of thE SoLUTioN oF tHe sToCHaStic eQuaTion, we wIll need infOrmAtIon about the aSYmPtotic behaVioUr of $J_\nu(t)$ and $Y_\Nu(t)$ As $t\to\iNfTy$. THE requIreD asymPtotic INformaTion is furNiSHed by [@aBRamstegUN 9.2.1, 9.2.2, 9.2.5, 9.2.6], WhIch wE record now for convENiENce:
\[eq:Besasy\] $$\begIn{aligNeD}
j_\nU(T) &=\SqrT{2/(\pi T)} \{\cos(t-\frac{1}{2}\nU\pI -\frac{1}{4}\PI) + O(t^{-1}) \}, \quad \TExT{ AS $T\to\INfty$}, \\
Y_\nu(t) &=\sqrt{2/(\pI t)} \{\sin(t-\frac{1}{2}\nU\Pi -\fRac{1}{4}\pi) + O(T^{-1}) \}, \qUad \TExt{ as $t\To\infTy$}.\ENd{aLigned}$$
Once aGain The solutiOn to thE ResolveNT equatiOn can bE wrIttEn as A SuM oF prOdUCts OF fUncTIonS dependiNg On $T$ and $s$. indeED, $R(T,S)$ can Be wRittEn in tHe form $$\label{eq:ResDecoMP8}
r(t,S) = d_7(s)r_{7}(t) + D_8(s)r_{8}(t), \qUad t\GeQ s\geq 0,\Quad a=0, \qUad b<0,$$ aNd Expressions for $d_7$ And $d_8$ May be obtaIneD fRom ThIs repREsentaTioN anD. This yiElds $$\begIN{alIgNED}
\LaBel{eq:cntsd78}
d_7(s) | kian of $J_\nu$ and $Y_\nu $ obeys $$ W\{J_ \nu (z) ,Y _\nu (z)\ } = 2/(\pi z)$ $ (cf ., e.g., [@NIST 10.5.2 ]) an da lsot ha t the deriva t iv e s of $ J_ \nu $a nd $Y_\ nu$ obey $ $J_0'(z)=- J_1 (z ), \quad Y_0 ' (z )=Y_1(z)$$ cf ., e.g. [@NI ST10.6.3 ]. In asymp tot ic an alysis of the solution o f the s t ochasti c eq uati on, we will needi nf o rmation aboutthe as ym p to t i c b eha viour of $ J_ \nu(t ) $ and $ Y _\ n u ( t)$ as $t\to\inft y$. The req u ire d asym pt oti c infor matio ni s f urnished by [@a bramstegu n 9.2. 1 , 9.2.2 , 9.2.5, 9.2.6 ],whi ch w e r ec ord n o w f o rcon v eni ence:
\ [e q: Besas y\]$ $ \ b egin {al igne d}
J_\nu(t) &= \sq rt{2 / (\p i t)} \{\c os(t -\ frac{ 1}{2}\ nu\pi - \frac{1}{4}\pi) + O (t^{-1})\}, \ qua d\text { as $t \to \in fty$}, \\
Y_\ nu ( t ) & =\sqrt{2/(\pi t)}\{ \ s in (t-\frac {1}{2} \ nu \p i -\frac{ 1} {4} \pi) + O(t^ {-1} ) \ }, \quad \text { a s$t\to\i nf ty$}.\ en d{a lig ned}$ $
On ce aga in the s oluti o n to the resol v ent equationc an b ew ritt enas a sum of pro d ucts off un cti o ns de pendi ng on $t$ and $s$. Indeed ,$r(t,s )$ ca n be writtenin the for m $ $\label{ eq:r e sd e comp8}
r(t ,s) = d_7(s)r_{ 7 }(t) + d _8(s) r_{8}(t) , \quad t \ g eq s\geq 0, \qu ada=0 , \q uad b<0,$$ an d expr es sions f or$d_7$ a nd$d_ 8$may b e obtaine d from t hi sre pr ese ntati o n and. T hi s y ie lds $$\b e gin{al igned }
\l ab el { eq: cntsd78 }
d_7 (s ) | kian of_$J_\nu$ and_$Y_\nu$ obeys $$W\{J_\nu(z),Y_\nu(z)\} =_2/(\pi z)$$_(cf.,_e.g., [@NIST 10.5.2])_and_also that the_derivatives of $J_\nu$_and $Y_\nu$ obey $$J_0'(z)=-J_1(z),_\quad Y_0'(z)=Y_1(z)$$ cf.,_e.g. [@NIST_10.6.3]. In asymptotic analysis of the solution of the stochastic equation, we will need_information_about the_asymptotic_behaviour_of $J_\nu(t)$ and $Y_\nu(t)$ as_$t\to\infty$. The required asymptotic information_is furnished_by [@abramstegun 9.2.1, 9.2.2, 9.2.5, 9.2.6], which we record_now_for convenience:
\[eq:Besasy\] $$\begin{aligned}
_ J_\nu(t) &=\sqrt{2/(\pi t)} \{\cos(t-\frac{1}{2}\nu\pi -\frac{1}{4}\pi) +_O(t^{-1}) \}, \quad \text{ as $t\to\infty$},_ \\
___Y_\nu(t) &=\sqrt{2/(\pi t)} \{\sin(t-\frac{1}{2}\nu\pi_-\frac{1}{4}\pi) + O(t^{-1}) \}, \quad \text{_as $t\to\infty$}.\end{aligned}$$
Once again the solution to_the resolvent equation can be written as_a sum of products of functions_depending on $t$ and $s$._Indeed, $r(t,s)$_can be written in the_form $$\label{eq:resdecomp8}
_ _r(t,s) = d_7(s)r_{7}(t)_+ d_8(s)r_{8}(t), \quad t\geq s\geq 0,\quad_a=0, \quad b<0,$$_and expressions for $d_7$ and $d_8$_may_be obtained from_this_representation_and. This_yields $$\begin{aligned}
\label{eq:cntsd78}
__ d_7(s) |
Au-coated sphere. Our measurements are performed by means of dynamic atomic force microscope (AFM) operated in the frequency-shift technique described in detail in Refs. [@27; @28]. We demonstrate significant increase in the gradient of the Casimir force in comparison with that between a Si plate covered with a SiO${}_2$ film and an Au-coated sphere, i.e., in the absence of graphene sheet. At short separations this increase is up to a factor 4-5 larger than the total experimental error in the measurement of the force gradient determined at a 67% confidence level. We also compare the experimental results with an approximate theory where the gradients of the Casimir force between a Si-SiO${}_2$ system and Au-coated sphere and between a graphene described by the Dirac model and the same sphere are computed independently of one another using the Lifshitz theory and then are added. Some excess of the theoretical force gradient over the experimental one is attributed to the screening of the Si-SiO${}_2$ surface by a graphene sheet.
The paper is organized as follows. In Sec. II we briefly describe the detection system, the measurement scheem and the sample preparation. Section III contains the measurement results and their comparison with theory. Section IV contains our conclusions.
Experimental setup
==================
The detection system used in our measurements consists of an AFM cantilever with attached hollow glass microsphere coated with Au, piezoelectric actuators, fiber interferometers, light source, and phase locked loop (PLL). The thickness of the Au coating and the radius of the coated sphere were measured to be 280nm and $54.10\pm 0.09\,\mu$m using an AFM and a scanning electron microscope, respectively. A turbo-pump, oil-free dry scroll mechanical-pump and ion-pump were used to achieve high vacuum down to $10^{-9}\,$Torr (see Refs. [@27; @28] for detail of the setup).
In the dynamic measurement scheme the total force $F_{\rm tot}(a)=F_{\rm el}(a)+F(a)$ acting on the sphere \[where $F_{\rm el}(a)$ and $F(a)$ are the electric and Casimir force, respectively, and $a$ is the separation distance between the sphere and graphene\] modifies the resonant natural frequency of the oscillator. The change in the frequency $\Delta\omega=\omega_r-\omega_ | Au - coated sphere. Our measurements are performed by mean of active nuclear force microscope (AFM) operated in the frequency - shift proficiency described in detail in Refs. [ @27; @28 ]. We demonstrate meaning addition in the gradient of the Casimir force in comparison with that between a Si plate covered with a SiO${}_2 $ film and an Au - coated celestial sphere, i.e., in the absence of graphene sheet. At short interval this increase is up to a factor 4 - 5 larger than the total experimental error in the measurement of the force gradient determine at a 67% confidence level. We also compare the experimental resultant role with an approximate theory where the gradients of the Casimir force between a Si - SiO${}_2 $ system and gold - coated sphere and between a graphene described by the Dirac model and the same sector are computed independently of one another using the Lifshitz theory and then are added. Some excess of the theoretical force gradient over the experimental one is attributed to the screening of the Si - SiO${}_2 $ surface by a graphene sheet.
The paper is organized as follows. In Sec. II we briefly describe the detection system, the measurement scheem and the sample distribution formulation. Section III contains the measurement resultant role and their comparison with hypothesis. incision IV contains our conclusions.
Experimental setup
= = = = = = = = = = = = = = = = = =
The detection system used in our measurements consist of an AFM cantilever with attached hollow glass microsphere coated with Au, piezoelectric actuators, fiber interferometer, light source, and phase locked loop (PLL). The thickness of the Au coating and the radius of the coated sphere were measure to be 280 nm and $ 54.10\pm 0.09\,\mu$m using an AFM and a scanning electron microscope, respectively. A turbo - pump, oil - free dry coil mechanical - pump and ion - pump were used to achieve high vacuum down to $ 10^{-9}\,$Torr (see Refs. [ @27; @28 ] for detail of the setup).
In the dynamic measurement outline the total power $ F_{\rm tot}(a)=F_{\rm el}(a)+F(a)$ acting on the sphere \[where $ F_{\rm el}(a)$ and $ F(a)$ are the electric and Casimir military unit, respectively, and $ a$ is the separation distance between the sector and graphene\ ] modifies the evocative natural frequency of the oscillator. The change in the frequency $ \Delta\omega=\omega_r-\omega _ | Au-foated sphere. Our measurtments are performed by keans kf dynamkc atomic force microscope (AHM) operattb in the frequency-shkft technpque desceibev in detail in Rxrs. [@27; @28]. We demonsfvate vmgnificant incrgase in the cradient of tha Zadimir force in comparison with that betweem w Si plate covgred eyth z SiO${}_2$ film and an Au-coated sphere, j.e., in tie absence of gtaphene sheet. At short sepwratlons this increase is up to a faceir 4-5 larger tfan the touan experimehtal error in the measurement ow the force graeiwnt getermined et a 67% bonfidence leyvl. We anso comlare the experlmentel rwsults with an approxmmate theory where tre gradietta of the Casimir doece bgtweet a Ru-SiU${}_2$ sgsvem and Ak-coeted sphere and betweeb a graphene descrineq by the Dirac model agd the same sphere are computed indepengenfly of one another usint the Lifshitz theory and then are added. Some excess of the theoretical force gsadieit ovtr the dzpfrimental one is attributed to the screening jr uhe Si-SiO${}_2$ surface by a graphrnf xreet.
The paper is oryznjzed as follows. In Sec. II re bruefly dessribr the detection system, the neasurement fxheem and the samppe preparatnon. Sevtion III contains the measuxement results anf their ckoparison with thdorj. Sewtion IV contains our concjusions.
Ex'erimzntal segup
==================
Tne detqction syshem used in our measuremfnts eonsivts of an WFM cantilever with attached hollow glass mictos[hege coated with Au, piezoeleceric actuators, fiber nnterfdrometers, might smurce, and prase locked lmlp (PLL). The tiickness jf tye Ay coative and the radiis of the coated spyere were measured to ge 280nm and $54.10\pm 0.09\,\mu$n uwing an AFM and a rcagnpng ejactron microvcopd, rdxpectkvely. A turno-pjmp, pil-free dry scroll machahical-pump and ion-pimi were usgd to achyeve high vacium down to $10^{-9}\,$Torr (ste Refv. [@27; @28] hor deyaij of the setup).
In the dynamic msasuremenh sgheme the totwl firce $F_{\rm tot}(c)=F_{\rm el}(a)+F(a)$ acting on the sphere \[where $F_{\rn el}(a)$ and $F(a)$ are jhe electric and Casiklr force, res'ectivqly, and $a$ is the separation dustance between tme sphere and graphene\] modifhes tje resonant natural frequency of the oscillator. The change in the frequenxy $\Delva\jmega=\omega_r-\kmegs_ | Au-coated sphere. Our measurements are performed by dynamic force microscope operated in the in [@27; @28]. We significant increase in gradient of the Casimir force in with that between a Si plate covered with a SiO${}_2$ film and an sphere, i.e., in the absence of graphene sheet. At short separations this increase up a 4-5 than the total experimental error in the measurement of the force gradient determined at a 67% level. We also compare the experimental results with approximate theory where the of the Casimir force between Si-SiO${}_2$ and Au-coated and a described by the model and the same sphere are computed independently of one another using the Lifshitz theory and then added. Some the theoretical gradient the one is attributed screening of the Si-SiO${}_2$ surface by The paper is organized as follows. In Sec. we briefly the detection system, the measurement scheem the sample preparation. Section III contains the measurement and their comparison with theory. Section IV contains our conclusions. Experimental setup ================== The detection in our measurements consists an AFM cantilever attached glass coated Au, piezoelectric fiber interferometers, light source, and phase locked loop (PLL). The thickness the Au coating and the radius of the coated sphere to 280nm and $54.10\pm using an AFM and scanning microscope, respectively. A turbo-pump, scroll and to high down to $10^{-9}\,$Torr (see [@27; @28] for detail of setup). In the dynamic $F_{\rm tot}(a)=F_{\rm el}(a)+F(a)$ acting on the sphere \[where el}(a)$ and $F(a)$ are the electric and force, respectively, and $a$ is the separation distance between the sphere and modifies the frequency of the oscillator. The change in the $\Delta\omega=\omega_r-\omega_ | Au-coated sphere. Our measuremEnts are perFormeD by MeaNs Of dyNamiC atomic force miCRoscOpe (AFM) operated in the freQuencY-sHIft tEChNique DescribED iN DEtaIl In refS. [@27; @28]. WE DeMonstRatE signifIcant increAse In The gradient oF ThE Casimir foRce In comparison WitH that bEtWeeN A Si plAte CoverEd with A siO${}_2$ filM and an Au-cOaTEd spheRE, i.e., in thE ABsEnce Of graphene sheet. At SHoRT separations thIs incrEaSE iS UP to A faCtor 4-5 larger ThAn the TOtal expERiMENTal ERror in the measUrement of thE ForCe gradIeNt dETerminEd at a 67% CoNFidEnce level. We Also Compare thE experIMental rESults wiTh an apProXimAte tHEoRy WheRe THe gRAdIenTS of The CasimIr FoRce beTweeN A sI-siO${}_2$ sYstEm anD Au-coAted sphere and BetWeen A GraPhene DescrIbed By The DiRac modEl and ThE same sphere are cOmpuTed indepeNdeNtLy oF oNe anoTHer usiNg tHe LIfshitz Theory aND thEn ARE AdDed. Some excess of the ThEOReTical forCe gradIEnT oVEr the expErImeNtal ONE is atTribUTeD to the scReeninG Of ThE Si-SiO${}_2$ sUrFace by A gRapHenE sheeT.
the pAper is OrganizeD as foLLows. In Sec. II we bRIefly describe THe DETeCTion SysTem, the measuRemeNT schEem aND tHe sAMple pReparAtIOn. sEction III contains thE mEasureMent rEsults and theiR comparisoN WITh theory. sectIOn iv contains our coNclusIons.
ExperiMEntal setUp
==================
The DetectioN system usED In our meaSurEmeNts ConSIStS of an AFM cantiLEVer wItH attachEd hOllow glAss MicRosPheRe Coated witH Au, piezoElEcTrIc ActUatorS, Fiber intErFerOmEteRs, ligHT sourcE, and pHase LoCkED loOp (PLL). ThE ThICKnesS oF tHe Au CoaTiNg and The rADiuS of the cOated spheRe wERe meAsUrEd to be 280nM and $54.10\pm 0.09\,\mu$m usinG aN AFM and a scAnNinG electRON microscOpe, respectively. A turbo-puMP, oil-freE drY scroLl meChanical-pUmp And ion-PumP Were usEd to acHieve HiGh vACUum doWN To $10^{-9}\,$torR (sEe Refs. [@27; @28] for dETAil Of the SeTup).
IN the dynAmic measurement schEMe tHe total force $F_{\Rm tOt}(a)=F_{\RM El}(A)+F(a)$ ACtINg oN tHE spHERe \[where $F_{\rm el}(a)$ anD $F(a)$ are the eLeCTrIc and CasimIR foRcE, respecTively, aNd $a$ is THe separAtion distAnce betweEn The sPHEre And graphenE\] modifieS the resonANt natURaL freqUenCy of thE oSciLlatoR. The chANge In the FrequeNcY $\Delta\Omega=\OmEga_r-\omegA_ | Au-coated sphere. Our mea surementsare p erf orm ed bymean s of dynamic a t omic force microscope (AFM ) ope ra t ed i n t he fr equency - sh i f t t ec hn iqu ed es cribe d i n detai l in Refs. [@ 27 ; @28]. We d e mo nstrate si gni ficant incre ase in th egra d ientofthe C asimir forcein compar is o n with that be t w ee n aSi plate coveredw it h a SiO${}_2$ f ilm an da nA u -co ate d sphere,i. e., i n the ab s en c e ofg raphene sheet . At shorts epa ration sthi s incre ase i su p t o a factor4-5larger th an the total e x perimen tal er ror in the me as ure me n t o f t hef orc e gradie nt d eterm ined a t a 67 % c onfi dence level. We al socomp a rethe e xperi ment al resu lts wi th an a pproximate theo ry w here thegra di ent sof th e Casim irfor ce betw een a S i -Si O$ { } _ 2$ system and Au-coa te d sp here and betwe e nag raphenede scr ibed b y the Dir a cmodel an d thes am espherear e comp ut edind epend e ntly of on e anothe r usi n g the Lifshitz theory and th e na r ea dded . S ome excessof t h e th eore t ic alf orcegradi en t o v er the experimental o ne isattri buted to thescreeningo f the Si-S iO${ } _2 $ surface by agraph ene sheet.
The pap er is organiz ed as fol l o ws. In S ec. II we br i e fl y describe th e dete ct ion sys tem , the m eas ure men t s ch eem and t he sampl epr ep ar ati on. S e ction II Icon ta ins them easure mentresu lt sa ndtheir c o mp a r ison w it h th eor y. Sect ionI V c ontains our conc lus i ons.
Ex perimen tal setup
=== == ========== == =
The de t e ction sy stem used in our measur e ments c ons istsof a n AFM can til ever w ith attach ed hol low g la ssm i crosp h e re co at ed with Au , pie zoele ct ricactuato rs, fiber interfer o met ers, light so urc e, a n d p has e l o cke dl oop ( PLL). The thick ness of th eA ucoating an d th eradiusof thecoate d sphere were mea sured tobe 280 n m an d $54.10\p m 0.09\, \mu$m usi n g anA FM anda s cannin gele ctron micro s cop e, re specti ve ly. Aturbo -p ump, oil -free dry scroll mechan ical-p ump a ndion-pumpwer e us ed to ach ieve high vacu umdow n to$10 ^ {-9}\ ,$To r r(se e Refs . [@ 2 7; @28] f o rdet a i lof the setu p ) .
I n the dy n amic m easu rement scheme the total force $F _{\r m tot }(a ) =F_{ \r m el}(a)+F(a)$ ac ti n g on thesp here \[wher e $F_{\r me l}(a) $ and$F(a)$ are th e el e ctricandCas imir forc e,re s pective ly ,a nd $a$ isth e sepa ration dist a n ce between the s phere a nd gr a phe ne\]mo difiest he r esonant na tural frequ ency o f th e osc illator .The ch ang ein the fre q uency $\D elta\ omega=\ om ega_ r-\ omega_ | Au-coated_sphere. Our_measurements are performed by_means of_dynamic_atomic force_microscope_(AFM) operated in_the frequency-shift technique_described in detail in_Refs. [@27; @28]. We_demonstrate_significant increase in the gradient of the Casimir force in comparison with that between_a_Si plate_covered_with_a SiO${}_2$ film and an_Au-coated sphere, i.e., in the_absence of_graphene sheet. At short separations this increase is_up_to a factor_4-5 larger than the total experimental error in the_measurement of the force gradient determined_at a 67%_confidence_level._We also compare the_experimental results with an approximate theory_where the gradients of the Casimir_force between a Si-SiO${}_2$ system and Au-coated_sphere and between a graphene described_by the Dirac model and_the same_sphere are computed independently of_one another using_the Lifshitz_theory and then_are added. Some excess of the_theoretical force gradient_over the experimental one is attributed_to_the screening of_the_Si-SiO${}_2$_surface by_a graphene sheet.
The_paper_is organized_as_follows. In Sec. II we briefly describe_the_detection system, the measurement scheem and the_sample preparation. Section III_contains_the measurement results and_their comparison with theory. Section_IV contains our conclusions.
Experimental setup
==================
The detection_system used_in our_measurements consists of an AFM cantilever with attached hollow glass microsphere_coated with Au, piezoelectric actuators, fiber_interferometers, light source, and_phase locked_loop_(PLL). The thickness_of_the Au_coating and the radius of the coated_sphere were_measured to be 280nm and $54.10\pm_0.09\,\mu$m using an AFM_and_a scanning electron microscope, respectively. A_turbo-pump, oil-free dry scroll mechanical-pump and_ion-pump were used to achieve_high_vacuum_down to $10^{-9}\,$Torr (see Refs. [@27;_@28] for detail of the setup).
In_the dynamic measurement_scheme the total force $F_{\rm tot}(a)=F_{\rm el}(a)+F(a)$_acting_on the sphere \[where $F_{\rm el}(a)$_and_$F(a)$ are the electric and Casimir_force,_respectively,_and $a$ is the separation_distance between the sphere and graphene\]_modifies the resonant natural frequency of the oscillator. The_change in the_frequency $\Delta\omega=\omega_r-\omega_ |
beta^{(h)} \cdot {\tilde {\bar \beta}}^{(h)} \Big],\nonumber\\
&&\nonumber\\
{\cal L}_{b} \longrightarrow \tilde {\cal L}_{b} & = & \tilde {\cal L}_0 + \frac {\partial}
{\partial \bar \theta}\, \frac {\partial} {\partial \theta}\, \Big [ \frac {i} {2} \, \tilde {\cal A}_\mu^{(h)}
\cdot \tilde {\cal A}^{\mu (h)} + \tilde {\cal F}^{(h)} \cdot {\tilde {\bar {\cal F}}}^{(h)}
+ \frac {i}{2} \, \tilde \Phi_\mu^{(h)} \cdot \tilde \Phi^{\mu (h)}
+ \frac{1}{2}\, \tilde \beta^{(h)} \cdot {\tilde {\bar \beta}}^{(h)} \Big].\nonumber\\\end{aligned}$$ Now, the (anti-)BRST invariance of the above coupled Lagrangian densities is straightforward because of the fact $(\partial_\theta^2 = \partial_{\bar \theta}^2 = 0)$. Thus, we have $$\begin{aligned}
\frac {\partial}{\partial \theta}\; \tilde{\cal L}_{\bar b}\bigg|_{\bar \theta = 0} = 0,\qquad
\frac {\partial}{\partial \bar \theta}\; \tilde{\cal L}_b \bigg|_{\theta = 0} = 0,\end{aligned}$$ which imply the (anti-)BRST invariance of the coupled Lagrangian densities within the framework of superfield formalism.
[99]{} Schwinger, J. S.: Phys. Rev. [**125**]{}, 397 (1962)
Schwinger, J. S.: Phys. Rev. [**128**]{}, 2425 (1962)
Deser, S., Jackiw, R., Templeton, S.: Ann. Phys. [**140**]{}, 372 (1982) \[Erratum-ibid. [**185**]{}, 406 (1988)\]
Deser, S., Jackiw, R., Templeton, S.: Phys. Rev. Lett. [**48**]{}, 975 (1982)
Freedman, D. Z., Townsend, P. K.: Nucl. Phys. B [**177**]{}, 282 (1981)
Allen, T. J., Bowick, M. J., Lahiri, A.: Mod. | beta^{(h) } \cdot { \tilde { \bar \beta}}^{(h) } \Big],\nonumber\\
& & \nonumber\\
{ \cal L}_{b } \longrightarrow \tilde { \cal L}_{b } & = & \tilde { \cal L}_0 + \frac { \partial }
{ \partial \bar \theta}\, \frac { \partial } { \partial \theta}\, \Big [ \frac { i } { 2 } \, \tilde { \cal A}_\mu^{(h) }
\cdot \tilde { \cal A}^{\mu (h) } + \tilde { \cal F}^{(h) } \cdot { \tilde { \bar { \cal F}}}^{(h) }
+ \frac { i}{2 } \, \tilde \Phi_\mu^{(h) } \cdot \tilde \Phi^{\mu (h) }
+ \frac{1}{2}\, \tilde \beta^{(h) } \cdot { \tilde { \bar \beta}}^{(h) } \Big].\nonumber\\\end{aligned}$$ Now, the (anti-)BRST invariability of the above pair Lagrangian densities is straightforward because of the fact $ (\partial_\theta^2 = \partial_{\bar \theta}^2 = 0)$. therefore, we have $ $ \begin{aligned }
\frac { \partial}{\partial \theta}\; \tilde{\cal L}_{\bar b}\bigg|_{\bar \theta = 0 } = 0,\qquad
\frac { \partial}{\partial \bar \theta}\; \tilde{\cal L}_b \bigg|_{\theta = 0 } = 0,\end{aligned}$$ which incriminate the (anti-)BRST invariance of the coupled Lagrangian density within the framework of superfield formalism.
[ 99 ] { } Schwinger, J. S.: Phys. Rev. [ * * 125 * * ] { }, 397 (1962)
Schwinger, J. S.: Phys. Rev. [ * * 128 * * ] { }, 2425 (1962)
Deser, S., Jackiw, R., Templeton, S.: Ann. Phys. [ * * 140 * * ] { }, 372 (1982) \[Erratum - ibid. [ * * 185 * * ] { }, 406 (1988)\ ]
Deser, S., Jackiw, R., Templeton, S.: Phys. Rev. Lett. [ * * 48 * * ] { }, 975 (1982)
Freedman, D. Z., Townsend, P. K.: Nucl. Phys. B [ * * 177 * * ] { }, 282 (1981)
Allen, T. J., Bowick, M. J., Lahiri, A.: Mod. | betw^{(h)} \cdot {\tilde {\bar \beta}}^{(h)} \Nig],\nonumber\\
&&\nonumywr\\
{\cal N}_{b} \lonfrightarfow \tilde {\cal L}_{b} & = & \tilde {\cap O}_0 + \frqc {\partial}
{\partial \bar \theta}\, \frwc {\partiql} {\pertial \theta}\, \Bmf [ \frac {i} {2} \, \tjpde {\eao A}_\mu^{(h)}
\cdot \tikde {\cal A}^{\mg (h)} + \tilde {\cal F}^{(f)} \edot {\tilde {\bar {\cal F}}}^{(h)}
+ \frac {i}{2} \, \tilde \Phi_\mu^{(h)} \cfot \tilde \Phi^{\mo (h)}
+ \gwac{1}{2}\, \fplbe \beta^{(h)} \cdot {\tilde {\bar \beta}}^{(h)} \Big].\nonlmber\\\end{aligned}$$ Npw, the (anti-)BRST invariance of hhe above coupled Pagrangian eensyries is strakghtforward because of the fact $(\partial_\theta^2 = \partial_{\car \tketa}^2 = 0)$. Thus, ww hwee $$\begin{alijned}
\frwc {\partial}{\pavnial \thata}\; \tikde{\cal L}_{\bar b}\blgg|_{\bac \thwta = 0} = 0,\qquad
\frac {\parvial}{\partial \bar \thetw}\; \tilde{\cdl L}_b \bigg|_{\theta = 0} = 0,\wne{aligted}$$ fhicf imoly tie (znti-)BRDT mnvariance kf the coupoed Lagrangian densotyvx within the framerowk of superfield formalism.
[99]{} Schwinger, J. V.: Pgys. Rev. [**125**]{}, 397 (1962)
Schwinger, J. S.: Phys. Rev. [**128**]{}, 2425 (1962)
Deser, S., Jwckiw, R., Tqmpleton, S.: Ann. Phys. [**140**]{}, 372 (1982) \[Erratum-ibid. [**185**]{}, 406 (1988)\]
Deser, S., Jackhw, R., Veoplttin, S.: Oyyd. Rev. Lett. [**48**]{}, 975 (1982)
Freedman, D. Z., Townsend, P. K.: Nucl. Prgs. B [**177**]{}, 282 (1981)
Allen, T. J., Bjwick, M. J., Lshlro, A.: Mod. | beta^{(h)} \cdot {\tilde {\bar \beta}}^{(h)} \Big],\nonumber\\ &&\nonumber\\ \longrightarrow {\cal L}_{b} = & \tilde {\partial \theta}\, \frac {\partial} \theta}\, \Big [ {i} {2} \, \tilde {\cal A}_\mu^{(h)} \tilde {\cal A}^{\mu (h)} + \tilde {\cal F}^{(h)} \cdot {\tilde {\bar {\cal F}}}^{(h)} \frac {i}{2} \, \tilde \Phi_\mu^{(h)} \cdot \tilde \Phi^{\mu (h)} + \frac{1}{2}\, \tilde \beta^{(h)} {\tilde \beta}}^{(h)} Now, (anti-)BRST invariance of the above coupled Lagrangian densities is straightforward because of the fact $(\partial_\theta^2 = \theta}^2 = 0)$. Thus, we have $$\begin{aligned} \frac \theta}\; \tilde{\cal L}_{\bar b}\bigg|_{\bar = 0} = 0,\qquad \frac \bar \tilde{\cal L}_b = = which imply the invariance of the coupled Lagrangian densities within the framework of superfield formalism. [99]{} Schwinger, J. S.: Phys. [**125**]{}, 397 J. S.: Rev. 2425 Deser, S., Jackiw, S.: Ann. Phys. [**140**]{}, 372 (1982) (1988)\] Deser, S., Jackiw, R., Templeton, S.: Phys. Lett. [**48**]{}, (1982) Freedman, D. Z., Townsend, P. Nucl. Phys. B [**177**]{}, 282 (1981) Allen, T. Bowick, M. J., Lahiri, A.: Mod. | beta^{(h)} \cdot {\tilde {\bar \beta}}^{(h)} \Big],\Nonumber\\
&&\noNumbeR\\
{\caL L}_{b} \LoNgriGhtaRrow \tilde {\cal L}_{b} & = & \TIlde {\Cal L}_0 + \frac {\partial}
{\partial \Bar \thEtA}\, \Frac {\PArTial} {\pArtial \tHEtA}\, \bIg [ \fRaC {i} {2} \, \TilDe {\CAl a}_\mu^{(h)}
\cDot \Tilde {\caL A}^{\mu (h)} + \tilde {\Cal f}^{(h)} \Cdot {\tilde {\bar {\CAl f}}}^{(h)}
+ \frac {i}{2} \, \tilDe \PHi_\mu^{(h)} \cdot \tilDe \PHi^{\mu (h)}
+ \fRaC{1}{2}\, \tiLDe \betA^{(h)} \cDot {\tiLde {\bar \BEta}}^{(h)} \BiG].\nonumber\\\EnD{AligneD}$$ now, the (aNTI-)BrST iNvariance of the aboVE cOUpled LagrangiaN densiTiES iS STraIghTforward beCaUse of THe fact $(\pARtIAL_\TheTA^2 = \partial_{\bar \thEta}^2 = 0)$. Thus, we haVE $$\beGin{aliGnEd}
\fRAc {\partIal}{\paRtIAl \tHeta}\; \tilde{\caL L}_{\baR b}\bigg|_{\bar \Theta = 0} = 0,\qQUad
\frac {\PArtial}{\pArtial \Bar \TheTa}\; \tiLDe{\CaL L}_b \BiGG|_{\thETa = 0} = 0,\End{ALigNed}$$ which ImPlY the (aNti-)Brst INvarIanCe of The coUpled LagrangiAn dEnsiTIes WithiN the fRameWoRk of sUperfiEld foRmAlism.
[99]{} Schwinger, J. s.: PhyS. Rev. [**125**]{}, 397 (1962)
SchwiNgeR, J. s.: PhYs. rev. [**128**]{}, 2425 (1962)
DeSEr, S., JacKiw, r., TeMpleton, s.: Ann. PhyS. [**140**]{}, 372 (1982) \[errAtUM-IBiD. [**185**]{}, 406 (1988)\]
Deser, S., Jackiw, R., TempLeTON, S.: phys. Rev. LEtt. [**48**]{}, 975 (1982)
FreEDmAn, d. z., TownsenD, P. k.: NuCl. PhYS. b [**177**]{}, 282 (1981)
AlleN, T. J., BOWiCk, M. J., LahiRi, A.: Mod. | beta^{(h)} \cdot {\tilde { \bar \beta }}^{( h)} \B ig ],\n onum ber\\
&&\nonum b er\\
{\cal L}_{b} \longrig htarr ow \til d e{\cal L}_{b} &= & \ ti ld e { \c a lL}_0+ \ frac {\ partial} {\p ar tial \bar \t h et a}\, \frac {\ partial} {\p art ial \ th eta } \, \ Big [ \f rac {i } {2} \ , \tilde{\ c al A}_ \ mu^{(h) }
\ cdot \tilde {\cal A}^ { \m u (h)} + \tilde {\cal F } ^{ ( h )}\cd ot {\tilde { \bar{ \cal F} } }^ { ( h )}+ \frac {i}{2 } \, \tilde \Ph i_\mu^ {( h)} \cdot\tild e\ Phi ^{\mu (h)}
+ \ frac{1}{2 }\, \t i lde \b e ta^{(h) } \cdo t { \ti lde{ \b ar \b et a }}^ { (h )}\Bi g].\nonu mb er \\\en d{al i g n e d}$$ No w, t he (a nti-)BRST inv ari ance ofthe a bovecoup le d Lag rangia n den si ties is straigh tfor ward beca use o f t he fact $(\par tia l_\ theta^2 = \par t ial _{ \ b a r\theta}^2 = 0)$. T hu s , w e have $ $\begi n {a li g ned}
\fr ac {\ part i a l}{\p arti a l\theta}\ ; \ti l de {\ cal L}_ {\ bar b} \b igg |_{ \bar\ thet a = 0} = 0,\q quad\ frac {\partial } {\partial \ba r \ t h et a }\; \t ilde{\cal L }_b\ bigg |_{\ t he ta= 0} = 0,\e nd { al i gned}$$ which imply t he (an ti-)B RST invarianc e of the c o u p led Lagr angi a nd ensities withi n the framework of super field formali sm.
[99] { } Schwing er, J. S. : P h y s. Rev. [**125* * ] {},39 7 (1962 )
Schwing er, J. S. : P hy s. Rev. [ **128**] {} ,24 25 (1 962)Deser, S ., Ja ck iw, R.,T emplet on, S .: A nn .P hys . [**14 0 ** ] { }, 3 72 ( 1982 ) \ [E rratu m-ib i d.[**185* *]{}, 406 (1 9 88)\ ]
D eser, S ., Jackiw, R. ,Templeton, S .:Phys.R e v. Lett. [**48**]{}, 975 (1982)
Freedm an, D. Z ., T ownsend,P.K.: Nu cl. Phys.B [**1 77**] {} , 2 8 2 (198 1 )
All en , T. J., B o w ick , M.J. , La hiri, A .: Mod. | beta^{(h)} \cdot_{\tilde {\bar_\beta}}^{(h)} \Big],\nonumber\\
&&\nonumber\\
{\cal L}_{b} \longrightarrow_\tilde {\cal_L}_{b}_& =_&_\tilde {\cal L}_0_+ \frac {\partial}_
{\partial \bar \theta}\, \frac_{\partial} {\partial _\theta}\,_ \Big [ \frac {i} {2} \, \tilde {\cal A}_\mu^{(h)}
\cdot \tilde {\cal A}^{\mu_(h)}_+ \tilde_{\cal_F}^{(h)}_\cdot {\tilde {\bar {\cal F}}}^{(h)}_
+ \frac {i}{2} \, \tilde_\Phi_\mu^{(h)} \cdot_\tilde \Phi^{\mu (h)}
+ \frac{1}{2}\, \tilde \beta^{(h)}_\cdot_{\tilde {\bar \beta}}^{(h)}_ \Big].\nonumber\\\end{aligned}$$ Now, the (anti-)BRST invariance of the above_coupled Lagrangian densities is straightforward because_of the fact_$(\partial_\theta^2_=_\partial_{\bar \theta}^2 = 0)$._Thus, we have $$\begin{aligned}
\frac {\partial}{\partial \theta}\;_ \tilde{\cal L}_{\bar b}\bigg|_{\bar \theta =_0} = 0,\qquad
\frac {\partial}{\partial \bar \theta}\;_ \tilde{\cal L}_b \bigg|_{\theta = 0}_= 0,\end{aligned}$$ which imply the_(anti-)BRST invariance_of the coupled Lagrangian densities_within the framework_of superfield_formalism.
[99]{} Schwinger, J._S.: Phys. Rev. [**125**]{}, 397 (1962)
Schwinger,_J. S.: Phys._Rev. [**128**]{}, 2425 (1962)
Deser, S., Jackiw,_R.,_Templeton, S.: Ann._Phys._[**140**]{},_372 (1982)_\[Erratum-ibid. [**185**]{}, 406_(1988)\]
Deser,_S., Jackiw,_R.,_Templeton, S.: Phys. Rev. Lett. [**48**]{},_975_(1982)
Freedman, D. Z., Townsend, P. K.: Nucl._Phys. B [**177**]{}, 282_(1981)
Allen,_T. J., Bowick, M._J., Lahiri, A.: Mod. |
.eps){width="0.85\linewidth"}
Once the link is removed in a rewiring event, nodes can be left without any link, so that the phase of previously connected and synchronized oscillators will start to drift away one from each other due to their natural frequencies difference $\delta \omega$, loosing any information regarding their previous mutual synchronization when their phase difference approaches $\pi /2$. This allows one to define the typical local desynchronization timescale $\tau_{LD}$ such that $$\tau_D \langle \delta \omega \rangle \approx \frac{\pi}{2}$$ with being the average natural frequency difference. For Gaussian distributed natural frequency one of course has $$\langle \delta \omega \rangle =\sqrt{\int_{-\infty}^{\infty} d\omega_1d\omega_2\, P_\sigma(\omega_1)P_\sigma(\omega_2)(\omega_1-\omega_2)^2}=\sqrt{2}\sigma \;$$ which finally yields $$\tau_{LD} \approx \frac{\pi}{2\sqrt{2}\,\sigma}$$
Before proceeding further, one comment is in order about our estimate of the typical local desynchronization timescale. We have computed it as the time required by a typical pair of oscillators to desynchronize. This is of course different from the average of individual couples desynchronization times $ \langle \pi /(2\delta\omega ) \rangle$, which is dominated by oscillators couples with almost degenerate natural frequencies, $\delta \omega \approx 0$. These latters, however, characterized by a very large local desinchronization time, are far from being representative of the typical behavior of random oscillators couples.
![Critical rewiring time $T_c$ as function of $q$ for $q<1$ for $\sigma=1$ and different values of the coupling constant (increasing along the cyan arrow). Respectively, from left to right: $\varepsilon=32$ (blue circles ), $\varepsilon=16$ (red circles) and $\varepsilon=8$ (black circles). Error bars give the estimated upper and lower boundaries for $T_c(q)$ as discussed in the main text. The dashed straight lines (same color coding) mark the linear prediction of Eq. \[eq\_final\] (see Section \[ave\_net\]). (Inset): The slope $s$ of each ${\epsilon}$ curve, evaluated by linear regression of the main panel data | .eps){width="0.85\linewidth " }
Once the link is removed in a rewiring event, node can be leave without any link, so that the phase of previously connected and synchronized oscillator will start to drift away one from each early ascribable to their natural frequencies difference $ \delta \omega$, loosing any information involve their previous mutual synchronism when their phase difference approaches $ \pi /2$. This allow one to define the typical local asynchronism timescale $ \tau_{LD}$ such that $ $ \tau_D \langle \delta \omega \rangle \approx \frac{\pi}{2}$$ with being the average natural frequency difference. For Gaussian spread natural frequency one of course hold $ $ \langle \delta \omega \rangle = \sqrt{\int_{-\infty}^{\infty } d\omega_1d\omega_2\, P_\sigma(\omega_1)P_\sigma(\omega_2)(\omega_1-\omega_2)^2}=\sqrt{2}\sigma \;$$ which last yields $ $ \tau_{LD } \approx \frac{\pi}{2\sqrt{2}\,\sigma}$$
Before proceed further, one comment is in order about our appraisal of the typical local desynchronization timescale. We have computed it as the time required by a typical couple of oscillator to desynchronize. This is of class different from the average of individual couples desynchronization times $ \langle \pi /(2\delta\omega) \rangle$, which is dominated by oscillators couples with almost degenerate lifelike frequency, $ \delta \omega \approx 0$. These latters, however, characterize by a very big local desinchronization time, are far from being representative of the typical behavior of random oscillator couples.
! [ Critical rewire time $ T_c$ as function of $ q$ for $ q<1 $ for $ \sigma=1 $ and different value of the coupling constant (increasing along the cyan arrow). Respectively, from left to right: $ \varepsilon=32 $ (blasphemous circles), $ \varepsilon=16 $ (red circles) and $ \varepsilon=8 $ (black circle). Error browning automatic rifle give the estimated upper and lower boundaries for $ T_c(q)$ as discussed in the chief text. The dashed straight lines (like color cryptography) mark the linear prediction of Eq. \[eq\_final\ ] (see Section \[ave\_net\ ]). (Inset ): The slope $ s$ of each $ { \epsilon}$ bend, evaluated by linear arrested development of the main panel data | .eps){aidth="0.85\linewidth"}
Once the llnk is removed iu a rewmring ebent, nodds can be left without any lmnk, wo thqt the phase of previojsly connvcted and synrhronized oscillefors will stadb to brmft away one frpm each otver due to thehr nctural frequencies difference $\delta \jmega$, lpoding any inforiatipg refarding their previous mutual syncgronizauion when their phsse difference approaches $\oi /2$. Hhis allows one to define the typyxal local derynchronization timesczle $\tau_{LD}$ such that $$\tau_D \langle \deltc \omega \rantlw \aokrox \frac{\pi}{2}$$ xith bving the average natusal freauency differekce. Fmr Taussian distributed iatural frequency ong of coursa kas $$\langle \delta \omegq \eanglg =\sqrd{\int_{-\kbfth}^{\inrtb} d\kmega_1d\lmeja_2\, P_\sigma(\omsga_1)P_\sigma(\omwga_2)(\omega_1-\omega_2)^2}=\sqrt{2}\sibmw \;$$ which finalmy yiejdf $$\tau_{LD} \approx \frac{\pi}{2\sqrt{2}\,\sigma}$$
Before prmcesding further, one commebt is in order about lur estimwte of the typical local desynchronization timescdle. Wx favt gjopuhed it as the time required by a typical pair kf oxcillators to besynchronize. Thos ix of course diwferenc fdom the average of indiviqual xouples dtsyncnronization times $ \langle \pu /(2\delta\omega ) \rangle$, which is dlminated by oscollatprs couples with almost degsnerate natkral freqhdncies, $\delta \omeea \sp[rox 0$. Thtre latters, howevew, charactxrizeb by a vdry karge jocal desijchrokhzation time, are fwr frlm being repgesentative of the typical behatmor of random ovcinlators eouplex.
![Critical rewyring time $T_c$ cs functnon of $q$ for $q<1$ fkr $\sigme=1$ and differqnt values of jhe coupling ronstant (yncrwasibg alone the cyan arroe). Respectpvtly, from lwft to right: $\varepxiluh=32$ (blue circles ), $\vatwpsilon=16$ (red citclds) wnf $\tarepfhlon=8$ (black chrclds). Dtror cars give tme dstikated upper and lowes bohndaries for $T_c(q)$ ax qiscussee in the main text. Thr dashed straight pines (saoe cokor coding) mark the linear predicfion of Ee. \[ez\_final\] (see Sqctiin \[ave\_net\]). (Inxet): The slope $s$ of each ${\epsilon}$ curve, etaluated by linear regtession of the main pcntl data | .eps){width="0.85\linewidth"} Once the link is removed in event, can be without any link, previously and synchronized oscillators start to drift one from each other due to natural frequencies difference $\delta \omega$, loosing any information regarding their previous mutual synchronization their phase difference approaches $\pi /2$. This allows one to define the typical desynchronization $\tau_{LD}$ that \langle \delta \omega \rangle \approx \frac{\pi}{2}$$ with being the average natural frequency difference. For Gaussian distributed frequency one of course has $$\langle \delta \omega =\sqrt{\int_{-\infty}^{\infty} d\omega_1d\omega_2\, P_\sigma(\omega_1)P_\sigma(\omega_2)(\omega_1-\omega_2)^2}=\sqrt{2}\sigma \;$$ finally yields $$\tau_{LD} \approx \frac{\pi}{2\sqrt{2}\,\sigma}$$ proceeding one comment in about estimate of the local desynchronization timescale. We have computed it as the time required by a typical pair of oscillators desynchronize. This course different the of couples desynchronization times \pi /(2\delta\omega ) \rangle$, which is couples with almost degenerate natural frequencies, $\delta \omega 0$. These however, characterized by a very large desinchronization time, are far from being representative of typical behavior of random oscillators couples. ![Critical rewiring time $T_c$ as function of $q$ for $\sigma=1$ and different values the coupling constant along cyan Respectively, left to $\varepsilon=32$ (blue circles ), $\varepsilon=16$ (red circles) and $\varepsilon=8$ (black circles). bars give the estimated upper and lower boundaries for $T_c(q)$ in main text. The straight lines (same color mark linear prediction of Eq. Section (Inset): of ${\epsilon}$ evaluated by linear regression the main panel data | .eps){width="0.85\linewidth"}
Once the lInk is removEd in a RewIriNg EvenT, nodEs can be left witHOut aNy link, so that the phase of PreviOuSLy coNNeCted aNd synchROnIZEd oScIlLatOrS WiLl staRt tO drift aWay one from EacH oTher due to theIR nAtural freqUenCies differenCe $\dElta \omEgA$, loOSing aNy iNformAtion rEGardinG their preViOUs mutuAL synchrONIzAtioN when their phase diFFeREnce approaches $\Pi /2$. This AlLOwS ONe tO deFine the typIcAl locAL desyncHRoNIZAtiON timescale $\tau_{lD}$ such that $$\tAU_D \lAngle \dElTa \oMEga \ranGle \apPrOX \frAc{\pi}{2}$$ with beiNg thE average nAtural FRequencY DiffereNce. For gauSsiAn diSTrIbUteD nATurAL fReqUEncY one of coUrSe Has $$\laNgle \DELTA \omeGa \rAnglE =\sqrt{\Int_{-\infty}^{\infty} D\omEga_1d\OMegA_2\, P_\sigMa(\omeGa_1)P_\sIgMa(\omeGa_2)(\omegA_1-\omegA_2)^2}=\sQrt{2}\sigma \;$$ which fiNallY yields $$\taU_{LD} \ApProX \fRac{\pi}{2\SQrt{2}\,\sigMa}$$
BEfoRe proceEding fuRTheR, oNE COmMent is in order about OuR EStImate of tHe typiCAl LoCAl desyncHrOniZatiON TimesCale. wE hAve compuTed it aS ThE tIme requIrEd by a tYpIcaL paIr of oSCillAtors tO desynchRonizE. this is of course DIfferent from tHE aVERaGE of iNdiVidual couplEs deSYnchRoniZAtIon TImes $ \lAngle \Pi /(2\DElTA\omega ) \rangle$, which is DoMinateD by osCillators coupLes with almOST DegeneraTe naTUrAL frequencies, $\deLta \omEga \approx 0$. THEse latteRs, howEver, charActerized BY A very larGe lOcaL deSinCHRoNization time, aRE Far fRoM being rEprEsentatIve Of tHe tYpiCaL behavior Of random OsCiLlAtOrs CouplES.
![CriticaL rEwiRiNg tIme $T_c$ AS functIon of $Q$ for $Q<1$ fOr $\SIgmA=1$ and difFErENT valUeS oF the CouPlIng coNstaNT (inCreasinG along the CyaN ArroW). REsPectiveLy, from left to rIgHt: $\varepsilOn=32$ (BluE circlES ), $\VarepsilOn=16$ (red circles) and $\varepsilON=8$ (black cIrcLes). ErRor bArs give thE esTimateD upPEr and lOwer boUndarIeS foR $t_C(q)$ as dISCuSseD iN the main teXT. the DasheD sTraiGht lineS (same color coding) maRK thE linear predicTioN of EQ. \[EQ\_fInaL\] (SeE secTiON \[avE\_NEt\]). (Inset): The slope $S$ of each ${\epsIlON}$ cUrve, evaluaTEd bY lInear reGressioN of thE Main panEl data | .eps){width="0.85\linewidt h"}
Oncethe l ink is r emov ed i n a rewiring e v ent, nodes can be left wit houtan y lin k ,so th at thep ha s e of p re vio us l yconne cte d and s ynchronize d o sc illators wil l s tart to dr ift away one fr omeach o th erd ue to th eir n atural freque ncies dif fe r ence $ \ delta \ o m eg a$,loosing any infor m at i on regarding t heir p re v io u s mu tua l synchron iz ation when th e ir p h ase difference ap proaches $\ p i / 2$. Th is al l ows on e tode f ine the typica l lo cal desyn chroni z ation t i mescale $\tau _{L D}$ suc h t ha t $ $\ t au_ D \ lan g le\delta \ om eg a \ra ngle \ a p prox \f rac{ \pi}{ 2}$$ with bei ngthea ver age n atura l fr eq uency diffe rence .For Gaussian di stri buted nat ura lfre qu encyo ne ofcou rse has $$ \langle \de lt a \ om ega \rangle =\sqrt {\ i n t_ {-\infty }^{\in f ty }d \omega_1 d\ ome ga_2 \ , P_\s igma ( \o mega_1)P _\sigm a (\ om ega_2)( \o mega_1 -\ ome ga_ 2)^2} = \sqr t{2}\s igma \;$ $ whi c h finally yiel d s $$\tau_{LD} \a p p ro x \fr ac{ \pi}{2\sqrt {2}\ , \sig ma}$ $
Bef o re pr oceed in g f u rther, one commentis in or der a bout our esti mate of th e t ypical l ocal de s ynchronization time scale. Weh ave comp utedit as th e time re q u ired bya t ypi cal pa i r o f oscillators t o de sy nchroni ze. This i s o f c our sedi fferent f rom theav er ag eofindiv i dual cou pl esde syn chron i zation time s $\l an g le\pi /(2 \ de l t a\om eg a) \r ang le $, wh ichi s d ominate d by osci lla t orsco up les wit h almost dege ne rate natur al fr equenc i e s, $\del ta \omega \approx 0$. T h ese lat ter s, ho weve r, charac ter ized b y a very l arge l ocalde sin c h roniz a t io n t im e, are far f rom bein grepr esentat ive of the typical beh avior of rand omosci l l at ors co u ple s.
![ C r itical rewiring time $T_c $a sfunction o f $q $for $q< 1$ for$\sig m a=1$ an d differe nt values o f th e cou pling cons tant (in creasinga longt he cyan ar row).Re spe ctive ly, fr o m l eft t o righ t: $\var epsil on =32$ (bl ue circles ), $\varepsi lon=16 $ (re d c ircles) a nd$ \va repsilon= 8$ ( black circ les ).Error ba r s giv e th e e sti m ateduppe r and lowe r b oun d a ri es for $T_c ( q ) $ a s dis cus s ed inthemain text. The da s hed straight l ines ( sam e c o lorco ding) mark the li ne a r predict io n of Eq. \[ eq\_fina l\ ] (see Secti on \[a ve\_net \ ] ). (Inset ): T heslope $s$ of e a ch ${\e ps il o n}$ cu rve, e valuat ed byl inea r regression of th e mai n panel dat a | .eps){width="0.85\linewidth"}
Once the_link is_removed in a rewiring_event, nodes_can_be left_without_any link, so_that the phase_of previously connected and_synchronized oscillators will_start_to drift away one from each other due to their natural frequencies difference $\delta_\omega$,_loosing any_information_regarding_their previous mutual synchronization when_their phase difference approaches $\pi_/2$. This_allows one to define the typical local desynchronization_timescale_$\tau_{LD}$ such that_$$\tau_D \langle \delta \omega \rangle \approx \frac{\pi}{2}$$ with being_the average natural frequency difference. For_Gaussian distributed natural_frequency_one_of course has $$\langle_\delta \omega \rangle =\sqrt{\int_{-\infty}^{\infty} d\omega_1d\omega_2\, P_\sigma(\omega_1)P_\sigma(\omega_2)(\omega_1-\omega_2)^2}=\sqrt{2}\sigma_\;$$ which finally yields $$\tau_{LD} \approx_\frac{\pi}{2\sqrt{2}\,\sigma}$$
Before proceeding further, one comment is in_order about our estimate of the_typical local desynchronization timescale. We_have computed_it as the time required_by a typical_pair of_oscillators to desynchronize._This is of course different from_the average of_individual couples desynchronization times $ \langle_\pi_/(2\delta\omega ) \rangle$,_which_is_dominated by_oscillators couples with_almost_degenerate natural_frequencies,_$\delta \omega \approx 0$. These latters,_however,_characterized by a very large local desinchronization_time, are far from_being_representative of the typical_behavior of random oscillators couples.
![Critical_rewiring time $T_c$ as function of_$q$ for_$q<1$ for_$\sigma=1$ and different values of the coupling constant (increasing along the_cyan arrow). Respectively, from left to_right: $\varepsilon=32$ (blue circles_), $\varepsilon=16$_(red_circles) and $\varepsilon=8$_(black_circles). Error_bars give the estimated upper and lower_boundaries for_$T_c(q)$ as discussed in the main_text. The dashed straight_lines_(same color coding) mark the linear_prediction of Eq. \[eq\_final\] (see Section_\[ave\_net\]). (Inset): The slope $s$_of_each_${\epsilon}$ curve, evaluated by linear_regression of the main panel data |
to the dimension of the control distribution, the system will be called a *fully actuated nonholonomic system*.
A **solution of a fully actuated nonholonomic problem** is an admissible curve $\gamma:I\rightarrow\mathcal{D}$ such that $$\nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))\in \Gamma (\tau_D),$$ or, equivalently, $$\nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))=u^{A}(t)e_{A}(\tau_{\mathcal{D}}(\gamma(t)),$$ where $u^{A}$ are the control inputs.
Locally, the equations may be written as $$\begin{aligned}
\dot{q}^{i}&=&\rho_{A}^{i}y^{A}\\
\dot{y}^{C}&=&-\Gamma_{AB}^{C}y^{A}y^{B}-(\mathcal{G}^{\mathcal{D}})^{CB}\rho_{B}^{i}\frac{\partial V}{\partial q^{i}}+u^{C}.\end{aligned}$$
Given a cost function $$\begin{aligned}
C&:&\mathcal{D}\times U\rightarrow\mathbb{R}\\
&&(q^{i},y^{A}, u^A)\mapsto C(q^{i},y^{A},u^{A})\end{aligned}$$ the *optimal control problem* consists of finding an admissible curve $\gamma:I\rightarrow\mathcal{D}$ solution of the fully actuated nonholonomic problem given initial and final boundary conditions on $\mathcal{D}$ and minimizing the functional $$\mathcal{J}(\gamma(t),u(t)):=\int_{0}^{T}C(\gamma(t),u(t))dt,$$ where $\gamma$ is an admissible curve.
We define the submanifold $\mathcal{D}^{(2)}$ of $T\mathcal{D}$ by $$\label{D2}
\mathcal{D}^{(2)}:=\{v\in T\mathcal{D}\mid v=\dot{\gamma}(0)\hbox{
where } \gamma:I\rightarrow\mathcal{D} \hbox{ is admissible}\ | to the dimension of the control distribution, the system will be called a * amply trip nonholonomic system *.
A * * solution of a fully actuated nonholonomic trouble * * is an admissible curve $ \gamma: I\rightarrow\mathcal{D}$ such that $ $ \nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))\in \Gamma (\tau_D),$$ or, equivalently, $ $ \nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))=u^{A}(t)e_{A}(\tau_{\mathcal{D}}(\gamma(t)),$$ where $ u^{A}$ are the control inputs.
Locally, the equation may be written as $ $ \begin{aligned }
\dot{q}^{i}&=&\rho_{A}^{i}y^{A}\\
\dot{y}^{C}&=&-\Gamma_{AB}^{C}y^{A}y^{B}-(\mathcal{G}^{\mathcal{D}})^{CB}\rho_{B}^{i}\frac{\partial V}{\partial q^{i}}+u^{C}.\end{aligned}$$
Given a monetary value function $ $ \begin{aligned }
C&:&\mathcal{D}\times U\rightarrow\mathbb{R}\\
& & (q^{i},y^{A }, u^A)\mapsto C(q^{i},y^{A},u^{A})\end{aligned}$$ the * optimal command problem * consists of recover an admissible curvature $ \gamma: I\rightarrow\mathcal{D}$ solution of the fully actuated nonholonomic problem given initial and final boundary condition on $ \mathcal{D}$ and minimizing the functional $ $ \mathcal{J}(\gamma(t),u(t)):=\int_{0}^{T}C(\gamma(t),u(t))dt,$$ where $ \gamma$ is an admissible curve.
We specify the submanifold $ \mathcal{D}^{(2)}$ of $ T\mathcal{D}$ by $ $ \label{D2 }
\mathcal{D}^{(2)}:=\{v\in T\mathcal{D}\mid v=\dot{\gamma}(0)\hbox {
where } \gamma: I\rightarrow\mathcal{D } \hbox { is admissible}\ | to the dimension of the coktrol distribution, the vystem will be called a *fully actuated noniolobomic system*.
A **solution of a fully acnuated nobholibomic problem** is ak admjdsibnx curve $\gamma:I\rlghtarrow\madhcal{D}$ such thdt $$\ncbla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\iathcal{B}^{\mwthcal{D}}}V(\tau_{\matrcal{C}}(\dammz(n)))\ik \Gamma (\tau_D),$$ or, equivalently, $$\nabma_{\gamma(u)}^{\mathcal{G}^{\mathcal{D}}}\gsmma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\hau_{\mwthcal{D}}(\gamma(t)))=u^{A}(t)e_{A}(\hau_{\mathcal{D}}(\tammw(r)),$$ where $u^{A}$ afe the conurml inputs.
Lkcally, the equations may be wrigten cs $$\begin{alitnwd}
\dlj{q}^{i}&=&\rho_{A}^{i}y^{A}\\
\dov{y}^{C}&=&-\Gamia_{AB}^{C}y^{A}y^{B}-(\matmbal{G}^{\matvcal{D}})^{CB}\tho_{B}^{i}\frac{\partisl T}{\parrial q^{i}}+u^{C}.\end{aligned}$$
Giten a cost function $$\fegin{aligteb}
C&:&\mathcal{D}\times U\righraerow\mdthbt{R}\\
&&(q^{i},h^{Q}, u^X)\makstp D(q^{i},y^{A},u^{W})\env{aligned}$$ ths *optimal cintrol problem* consosew of finding zn admysfible curve $\gamma:I\rightarrow\mathcal{D}$ sonutjon of the fully actuatwd nonholonomic problgm given igitial and final boundary conditions on $\mathcal{D}$ dnd mmnkmieikn thd fknctional $$\mathcal{J}(\gamma(t),u(t)):=\int_{0}^{T}C(\gamma(t),u(t))dt,$$ whews $\bakma$ is an admifsible curvr.
Wf cgfine the submxnifolb $\mzthcal{D}^{(2)}$ of $T\mathcap{D}$ by $$\lwbel{D2}
\nathcal{D}^{(2)}:=\{v\yn T\kathcal{D}\mid v=\dot{\gamma}(0)\hbox{
whwre } \gamma:I\rpghtqrrow\mathcal{D} \hbox{ is admissiyle}\ | to the dimension of the control distribution, will called a actuated nonholonomic system*. actuated problem** is an curve $\gamma:I\rightarrow\mathcal{D}$ such $$\nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))\in \Gamma (\tau_D),$$ or, equivalently, $$\nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))=u^{A}(t)e_{A}(\tau_{\mathcal{D}}(\gamma(t)),$$ $u^{A}$ are the control inputs. Locally, the equations may be written as $$\begin{aligned} \dot{y}^{C}&=&-\Gamma_{AB}^{C}y^{A}y^{B}-(\mathcal{G}^{\mathcal{D}})^{CB}\rho_{B}^{i}\frac{\partial V}{\partial q^{i}}+u^{C}.\end{aligned}$$ Given a cost function $$\begin{aligned} C&:&\mathcal{D}\times U\rightarrow\mathbb{R}\\ &&(q^{i},y^{A}, u^A)\mapsto C(q^{i},y^{A},u^{A})\end{aligned}$$ *optimal problem* of an admissible curve $\gamma:I\rightarrow\mathcal{D}$ solution of the fully actuated nonholonomic problem given initial and final boundary on $\mathcal{D}$ and minimizing the functional $$\mathcal{J}(\gamma(t),u(t)):=\int_{0}^{T}C(\gamma(t),u(t))dt,$$ where is an admissible curve. define the submanifold $\mathcal{D}^{(2)}$ of by \mathcal{D}^{(2)}:=\{v\in T\mathcal{D}\mid where \gamma:I\rightarrow\mathcal{D} is admissible}\ | to the dimension of the controL distributIon, thE sySteM wIll bE calLed a *fully actuaTEd noNholonomic system*.
A **solutIon of A fULly aCTuAted nOnholonOMiC PRobLeM** iS an AdMIsSible CurVe $\gamma:i\rightarroW\maThCal{D}$ such that $$\NAbLa_{\gamma(t)}^{\maThcAl{G}^{\mathcal{D}}}\gAmmA(t)+grad_{\MaThcAL{G}^{\matHcaL{D}}}V(\taU_{\mathcAL{D}}(\gammA(t)))\in \Gamma (\TaU_d),$$ or, equIValentlY, $$\NAbLa_{\gaMma(t)}^{\mathcal{G}^{\mathcAL{D}}}\GAmma(t)+grad_{\mathcAl{G}^{\matHcAL{D}}}v(\TAu_{\mAthCal{D}}(\gamma(t)))=U^{A}(T)e_{A}(\taU_{\Mathcal{d}}(\GaMMA(T)),$$ whERe $u^{A}$ are the conTrol inputs.
LOCalLy, the eQuAtiONs may bE writTeN As $$\bEgin{aligned}
\Dot{q}^{I}&=&\rho_{A}^{i}y^{A}\\
\dOt{y}^{C}&=&-\GaMMa_{AB}^{C}y^{A}Y^{b}-(\mathcaL{G}^{\mathCal{d}})^{CB}\Rho_{B}^{I}\FrAc{\ParTiAL V}{\pARtIal Q^{I}}+u^{C}.\End{alignEd}$$
giVen a cOst fUNCTIon $$\bEgiN{aliGned}
C&:&\Mathcal{D}\times u\riGhtaRRow\MathbB{R}\\
&&(q^{i},y^{a}, u^A)\mApSto C(q^{I},y^{A},u^{A})\eNd{aliGnEd}$$ the *optimal conTrol Problem* coNsiStS of FiNding AN admisSibLe cUrve $\gamMa:I\righTArrOw\MATHcAl{D}$ solution of the fuLlY ACtUated nonHolonoMIc PrOBlem giveN iNitIal aND Final BounDArY conditiOns on $\mAThCaL{D}$ and miNiMizing ThE fuNctIonal $$\MAthcAl{J}(\gamMa(t),u(t)):=\int_{0}^{t}C(\gamMA(t),u(t))dt,$$ where $\gamMA$ is an admissibLE cURVe.
wE defIne The submanifOld $\mAThcaL{D}^{(2)}$ of $t\MaThcAL{D}$ by $$\lAbel{D2}
\MaTHcAL{D}^{(2)}:=\{v\in T\mathcal{D}\mid v=\dOt{\Gamma}(0)\hBox{
whEre } \gamma:I\righTarrow\mathCAL{d} \hbox{ is aDmisSIbLE}\ | to the dimension of the c ontrol dis tribu tio n,th e sy stem will be calle d a * fully actuated nonholo nomic s y stem * .
A ** solutio n o f a f ul ly ac tu a te d non hol onomicproblem**isan admissiblec ur ve $\gamma :I\ rightarrow\m ath cal{D} $suc h that $$ \nabl a_{\ga m ma(t)} ^{\mathca l{ G }^{\ma t hcal{D} } } \g amma (t)+grad_{\mathca l {G } ^{\mathcal{D}} }V(\ta u_ { \m a t hca l{D }}(\gamma( t) ))\in \Gamma( \t a u _ D), $ $ or, equival ently, $$\n a bla _{\gam ma (t) } ^{\mat hcal{ G} ^ {\m athcal{D}}} \gam ma(t)+gra d_{\ma t hcal{G} ^ {\mathc al{D}} }V( \ta u_{\ m at hc al{ D} } (\g a mm a(t ) ))= u^{A}(t) e_ {A }(\ta u_{\ m a t h cal{ D}} (\ga mma(t )),$$ where $ u^{ A}$a rethe c ontro l in pu ts.
Locall y, th eequations may b e wr itten as$$\ be gin {a ligne d }
\dot {q} ^{i }&=&\rh o_{A}^{ i }y^ {A } \ \
\ dot{y}^{C}&=&-\Gam ma _ { AB }^{C}y^{ A}y^{B } -( \m a thcal{G} ^{ \ma thca l { D}})^ {CB} \ rh o_{B}^{i }\frac { \p ar tial V} {\ partia lq^{ i}} +u^{C } .\en d{alig ned}$$
Given a cost functio n $$\begin{ali g ne d }
C & :&\m ath cal{D}\time s U\ r ight arro w \m ath b b{R}\ \
&&( q^ { i} , y^{A}, u^A)\mapstoC( q^{i}, y^{A} ,u^{A})\end{a ligned}$$t h e *optima l co n tr o l problem* con sists of findin g an admi ssibl e curve$\gamma:I \ r ightarro w\m ath cal {D} $ so lution of the f ully a ctuated no nholono mic pr obl emgi ven initi al and f in al b ou nda ry co n ditionson $\ ma thc al{D} $ and m inimi zing t he fun ctional $$ \ m athc al {J }(\g amm a( t),u( t)): = \in t_{0}^{ T}C(\gamm a(t ) ,u(t )) dt ,$$ whe re $\gamma$ i san admissi bl e c urve.W e define the submanifold $\math c al{D}^{ (2) }$ of $T\ mathcal{D }$by $$\ lab e l{D2}\mathc al{D} ^{ (2) } : =\{v\ i n T \ma th cal{D}\mid v =\d ot{\g am ma}( 0)\hbox {
where } \gamma:I \ rig htarrow\mathc al{ D} \ h b ox { i s a d mis si b le} \ | to_the dimension_of the control distribution,_the system_will_be called_a_*fully actuated nonholonomic_system*.
A **solution of_a fully actuated nonholonomic_problem** is an_admissible_curve $\gamma:I\rightarrow\mathcal{D}$ such that $$\nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))\in \Gamma (\tau_D),$$ or, equivalently, $$\nabla_{\gamma(t)}^{\mathcal{G}^{\mathcal{D}}}\gamma(t)+grad_{\mathcal{G}^{\mathcal{D}}}V(\tau_{\mathcal{D}}(\gamma(t)))=u^{A}(t)e_{A}(\tau_{\mathcal{D}}(\gamma(t)),$$ where $u^{A}$ are the_control_inputs.
Locally, the_equations_may_be written as $$\begin{aligned}
\dot{q}^{i}&=&\rho_{A}^{i}y^{A}\\
\dot{y}^{C}&=&-\Gamma_{AB}^{C}y^{A}y^{B}-(\mathcal{G}^{\mathcal{D}})^{CB}\rho_{B}^{i}\frac{\partial V}{\partial_q^{i}}+u^{C}.\end{aligned}$$
Given a cost function $$\begin{aligned}
C&:&\mathcal{D}\times_U\rightarrow\mathbb{R}\\
&&(q^{i},y^{A}, u^A)\mapsto_C(q^{i},y^{A},u^{A})\end{aligned}$$ the *optimal control problem* consists of finding_an_admissible curve $\gamma:I\rightarrow\mathcal{D}$_solution of the fully actuated nonholonomic problem given initial_and final boundary conditions on $\mathcal{D}$_and minimizing the_functional_$$\mathcal{J}(\gamma(t),u(t)):=\int_{0}^{T}C(\gamma(t),u(t))dt,$$_where $\gamma$ is an_admissible curve.
We define the submanifold $\mathcal{D}^{(2)}$_of $T\mathcal{D}$ by $$\label{D2}
\mathcal{D}^{(2)}:=\{v\in T\mathcal{D}\mid v=\dot{\gamma}(0)\hbox{
where_} \gamma:I\rightarrow\mathcal{D} \hbox{ is admissible}\ |
usual one.
By Proposition \[prop:reflexible\], the characteristic ideals ${\rm char}_{R}(M)$ and ${\rm char}_{R}(N)$ are reflexive. Hence, by Lemma \[lem:inequality\] and the condition (G$_{1}$), we may assume that $R$ is a Gorenstein local ring with $\dim(R) \leq 1$.
By the horseshoe lemma, a projective resolution of $M$ can be built up inductively with the $n$-th item in the resolution of $M$ equal to the direct sum of the $n$-th items in projective resolutions of $N$ and $M/N$. Hence one can take the following commutative diagram with exact lows: $$\xymatrix{
0 \ar[r] & Y \ar[r]^-{\beta} \ar@{^{(}->}[d] & R^{r} \ar[r] \ar@{^{(}->}[d] & N \ar[r] \ar@{^{(}->}[d] & 0
\\
0 \ar[r] & X \ar[r]^-{\alpha} \ar[d] & R^{r+s} \ar[r] \ar[d] & M \ar[r] \ar[d] & 0
\\
0 \ar[r] & X/Y \ar[r] & R^{s} \ar[r] & M/N \ar[r] & 0.
}$$ Here $r$ is a positive integer and the homomorphism $R^{r} {\lhook\joinrel\longrightarrow}R^{r+s}$ is a split injection.
Since $R$ is Gorenstein and $\dim(R) \leq 1$, the module $\operatorname{Ext}_{R}^{1}(X/Y,R) \stackrel{\sim}{\longrightarrow} \operatorname{Ext}_{R}^{2}(M/N,R)$ vanishes. Hence by taking $R$-duals to the above commutative diagram, we get exact sequences of $R$-modules $$\xymatrix{
0 \ar[r] & (R^{*})^{s} \ar[r] \ar[d] & (R^{*})^{r+s} \ar[r] \ar[d]^-{\alpha^{*}} & (R^{*})^{r} \ar[r] \ar[d]^-{\beta^{*}} & | usual one.
By Proposition \[prop: reflexible\ ], the characteristic ideals $ { \rm char}_{R}(M)$ and $ { \rm char}_{R}(N)$ are reflexive. Hence, by Lemma \[lem: inequality\ ] and the circumstance (G$_{1}$), we may wear that $ R$ is a Gorenstein local ring with $ \dim(R) \leq 1$.
By the horseshoe lemma, a projective resolution of $ M$ can be build up inductively with the $ n$-th detail in the resolution of $ M$ equal to the lineal sum of the $ n$-th items in projective resolutions of $ N$ and $ M / N$. therefore one can remove the keep up commutative diagram with exact low: $ $ \xymatrix {
0 \ar[r ] & Y \ar[r]^-{\beta } \ar@{^{(}->}[d ] & R^{r } \ar[r ] \ar@{^{(}->}[d ] & N \ar[r ] \ar@{^{(}->}[d ] & 0
\\
0 \ar[r ] & X \ar[r]^-{\alpha } \ar[d ] & R^{r+s } \ar[r ] \ar[d ] & M \ar[r ] \ar[d ] & 0
\\
0 \ar[r ] & X / Y \ar[r ] & R^{s } \ar[r ] & M / N \ar[r ] & 0.
} $ $ Here $ r$ is a positivist integer and the homomorphism $ R^{r } { \lhook\joinrel\longrightarrow}R^{r+s}$ is a split injection.
Since $ R$ is Gorenstein and $ \dim(R) \leq 1 $, the faculty $ \operatorname{Ext}_{R}^{1}(X / Y, R) \stackrel{\sim}{\longrightarrow } \operatorname{Ext}_{R}^{2}(M / N, R)$ vanishes. Hence by take $ R$-duals to the above commutative diagram, we catch exact sequences of $ R$-modules $ $ \xymatrix {
0 \ar[r ] & (R^{*})^{s } \ar[r ] \ar[d ] & (R^{*})^{r+s } \ar[r ] \ar[d]^-{\alpha^ { * } } & (R^{*})^{r } \ar[r ] \ar[d]^-{\beta^ { * } } & | uskal one.
By Proposition \[prok:reflexible\], the ckqractecistic jdeals ${\ro char}_{R}(M)$ and ${\rm char}_{R}(N)$ are rxflezive. Yence, by Lemma \[lem:inequxlity\] and the coneitiib (G$_{1}$), we may assume bkat $R$ ls a Yocenstein local ting with $\dik(R) \leq 1$.
By the vofszshoe lemma, a projective resolution jf $M$ cam he built up inquctpvqly sptm the $n$-th item in the resolution of $M$ exual to the dorect sum of the $n$-th items in orojective resolutlons of $N$ abd $M/G$. Hence one cxn take tht yollowing ckmmutative diagram with exact luws: $$\xvmatrix{
0 \ar[r] & T \ag[t]^-{\beta} \ar@{^{(}->}[d] & R^{c} \ar[r] \wr@{^{(}->}[d] & N \ar[r] \av@{^{(}->}[c] & 0
\\
0 \as[r] & X \at[r]^-{\alpha} \ar[d] & R^{v+s} \ar[c] \ar[e] & M \ar[r] \ar[d] & 0
\\
0 \ar[r] & X/B \ar[r] & R^{s} \ar[r] & M/N \ar[t] & 0.
}$$ Here $s$ ns a positive integer abd thg hommmoroyiso $R^{d} {\khkok\joijrem\longrightzrrow}R^{r+s}$ is a split injection.
Sonsv $R$ is Gorensfein agd $\dim(R) \leq 1$, the module $\operatorname{Ext}_{R}^{1}(X/J,R) \sfackrel{\sim}{\longrightarroq} \operatorname{Ext}_{R}^{2}(M/N,R)$ vanishes. Hence by taking $R$-duals to the above commutative giagrem, we gct ebqch sequences of $R$-modules $$\xymatrix{
0 \ar[r] & (R^{*})^{s} \ar[r] \wd[d] & (G^{*})^{r+s} \ar[r] \ar[d]^-{\alpha^{*}} & (R^{*})^{r} \ar[r] \ar[d]^-{\neha^{*}} & | usual one. By Proposition \[prop:reflexible\], the characteristic char}_{R}(M)$ ${\rm char}_{R}(N)$ reflexive. Hence, by (G$_{1}$), may assume that is a Gorenstein ring with $\dim(R) \leq 1$. By horseshoe lemma, a projective resolution of $M$ can be built up inductively with $n$-th item in the resolution of $M$ equal to the direct sum of $n$-th in resolutions $N$ and $M/N$. Hence one can take the following commutative diagram with exact lows: $$\xymatrix{ 0 & Y \ar[r]^-{\beta} \ar@{^{(}->}[d] & R^{r} \ar[r] \ar@{^{(}->}[d] N \ar[r] \ar@{^{(}->}[d] & \\ 0 \ar[r] & X \ar[d] R^{r+s} \ar[r] & \ar[r] & 0 \\ \ar[r] & X/Y \ar[r] & R^{s} \ar[r] & M/N \ar[r] & 0. }$$ Here $r$ is a integer and $R^{r} {\lhook\joinrel\longrightarrow}R^{r+s}$ a injection. $R$ is Gorenstein \leq 1$, the module $\operatorname{Ext}_{R}^{1}(X/Y,R) \stackrel{\sim}{\longrightarrow} by taking $R$-duals to the above commutative diagram, get exact of $R$-modules $$\xymatrix{ 0 \ar[r] & \ar[r] \ar[d] & (R^{*})^{r+s} \ar[r] \ar[d]^-{\alpha^{*}} & (R^{*})^{r} \ar[d]^-{\beta^{*}} & | usual one.
By Proposition \[prop:Reflexible\], The chAraCteRiStic IdeaLs ${\rm char}_{R}(M)$ and ${\rM Char}_{r}(N)$ are reflexive. Hence, by LEmma \[lEm:INequALiTy\] and The condITiON (g$_{1}$), we MaY aSsuMe THaT $R$ is a gorEnstein Local ring wIth $\DiM(R) \leq 1$.
By the hoRSeShoe lemma, a ProJective resolUtiOn of $M$ cAn Be bUIlt up IndUctivEly witH The $n$-th Item in the ReSOlutioN Of $M$ equaL TO tHe diRect sum of the $n$-th itEMs IN projective resOlutioNs OF $N$ AND $M/N$. henCe one can taKe The foLLowing cOMmUTATivE Diagram with exAct lows: $$\xymaTRix{
0 \Ar[r] & Y \ar[R]^-{\bEta} \AR@{^{(}->}[d] & R^{r} \ar[R] \ar@{^{(}->}[d] & N \Ar[R] \Ar@{^{(}->}[d] & 0
\\
0 \Ar[r] & X \ar[r]^-{\alphA} \ar[d] & r^{r+s} \ar[r] \ar[d] & m \ar[r] \ar[D] & 0
\\
0 \Ar[r] & X/Y \ar[R] & r^{s} \ar[r] & M/N \Ar[r] & 0.
}$$ HerE $r$ iS a pOsitIVe InTegEr ANd tHE hOmoMOrpHism $R^{r} {\lhOoK\jOinreL\lonGRIGHtarRow}r^{r+s}$ iS a splIt injection.
SiNce $r$ is GORenStein And $\diM(R) \leQ 1$, tHe modUle $\opeRatorNaMe{Ext}_{R}^{1}(X/Y,R) \stackrEl{\siM}{\longrighTarRoW} \opErAtornAMe{Ext}_{R}^{2}(m/N,R)$ VanIshes. HeNce by taKIng $r$-dUALS tO the above commutatiVe DIAgRam, we get Exact sEQuEnCEs of $R$-modUlEs $$\xYmatRIX{
0 \ar[r] & (R^{*})^{S} \ar[r] \AR[d] & (r^{*})^{r+s} \ar[r] \ar[D]^-{\alpha^{*}} & (r^{*})^{R} \aR[r] \Ar[d]^-{\beta^{*}} & | usual one.
By Propositio n \[prop:r eflex ibl e\] ,thechar acteristic ide a ls $ {\rm char}_{R}(M)$ and ${\r mc har} _ {R }(N)$ are re f le x i ve. H en ce, b y L emma\[l em:ineq uality\] a ndth e condition( G$ _{1}$), we ma y assume tha t $ R$ isaGor e nstei n l ocalring w i th $\d im(R) \le q1 $.
By the hor s e sh oe l emma, a projectiv e r e solution of $M $ canbe bu i l t u p i nductively w ith t h e $n$-t h i t e m in the resolutio n of $M$ eq u alto the d ire c t sumof th e$ n$- th items in pro jective r esolut i ons of$ N$ and$M/N$. He nce one ca ntak et hef ol low i ngcommutat iv ediagr am w i t h exac t l ows: $$\x ymatrix{
0 \a r[r ] &Y \a r[r]^ -{\be ta}\a r@{^{ (}->}[ d] &R^ {r} \ar[r] \ar@ {^{( }->}[d] & N\a r[r ]\ar@{ ^ {(}->} [d] &0
\\
0 \ar[r] & X \ a r [ r] ^-{\alpha} \ar[d]&R ^ {r +s} \ar[ r] \ar [ d] & M \ar[r] \ ar[ d] & 0
\\
0 \ar [ r] & X/Y \ ar[r]& R ^{ s} \ar[ r] & M/N \ ar[ r]& 0.}$$Here $ r$ is aposit i ve integer and the homomorph i sm $ R^ { r} { \lh ook\joinrel \lon g righ tarr o w} R^{ r +s}$is asp l it injection.
Since $ R$ is Go renst ein and $\dim (R) \leq 1 $ , the modu le $ \ op e ratorname{Ext} _{R}^ {1}(X/Y,R) \stackre l{\si m}{\long rightarro w } \operat orn ame {Ex t}_ { R }^ {2}(M/N,R)$ v a n ishe s. Hencebytaking$R$ -du als to t he abovecommutat iv edi ag ram , weg et exact s equ en ces of $ R $-modu les $ $\xy ma tr i x{0 \ar[r ] & ( R^{* }) ^{ s} \ ar[ r] \ar[ d] & (R^ {*})^{r +s} \ar[r ] \ a r[d] ^- {\ alpha^{ *}} & (R^{*}) ^{ r} \ar[r]\a r[d ]^-{\b e t a^{*}} & | usual_one.
By Proposition \[prop:reflexible\],_the characteristic ideals ${\rm_char}_{R}(M)$ and_${\rm_char}_{R}(N)$ are_reflexive._Hence, by Lemma \[lem:inequality\]_and the condition_(G$_{1}$), we may assume_that $R$ is_a_Gorenstein local ring with $\dim(R) \leq 1$.
By the horseshoe lemma, a projective resolution of_$M$_can be_built_up_inductively with the $n$-th item_in the resolution of $M$_equal to_the direct sum of the $n$-th items in_projective_resolutions of $N$_and $M/N$. Hence one can take the following commutative_diagram with exact lows: $$\xymatrix{
0 \ar[r]_& Y \ar[r]^-{\beta}_\ar@{^{(}->}[d]_&_R^{r} \ar[r] \ar@{^{(}->}[d] &_N \ar[r] \ar@{^{(}->}[d] & 0
\\
0_\ar[r] & X \ar[r]^-{\alpha} \ar[d] &_R^{r+s} \ar[r] \ar[d] & M \ar[r] \ar[d]_& 0
\\
0 \ar[r] & X/Y \ar[r]_& R^{s} \ar[r] & M/N_\ar[r] &_0.
}$$ Here $r$ is_a positive integer_and the_homomorphism $R^{r} {\lhook\joinrel\longrightarrow}R^{r+s}$_is a split injection.
Since $R$ is_Gorenstein and $\dim(R)_\leq 1$, the module $\operatorname{Ext}_{R}^{1}(X/Y,R) \stackrel{\sim}{\longrightarrow}_\operatorname{Ext}_{R}^{2}(M/N,R)$_vanishes. Hence by_taking_$R$-duals_to the_above commutative diagram,_we_get exact_sequences_of $R$-modules $$\xymatrix{
0 \ar[r] & (R^{*})^{s}_\ar[r]_\ar[d] & (R^{*})^{r+s} \ar[r] \ar[d]^-{\alpha^{*}} & (R^{*})^{r}_\ar[r] \ar[d]^-{\beta^{*}} & |
the solutions to separations of multiples of the KI resolution only. Thus, at the 99.7 % (67 %) confidence level of rejecting binary models, it is very unlikely that more than one out of our five targets have a binary companion of 0.08 (0.05) brightness of the primary.]{} The masses that such flux limits represent depend on both the luminosity of the primary and the assumed age of the system. Based on the PMS evolutionary tracks of @1999ApJ...525..772P and an age of 1 Myr, flux ratios of 0.1 would refer to companion masses of about 1/4 of the primary mass.
The fact that we can rule out binaries for the major part of the tested FoV, and for a brightness ratio down to 0.05 is mostly related to the measured small [*differential*]{} visibility variations. In contrast, the [*absolute*]{} level constrains binary parameters to a lesser extent only, but is rather linked to the amount of (over-)resolved emission. An equal brightness binary would result in large ($\gtrsim\,0.2$) squared visibility variations over the probed range of spatial frequencies of the KI baseline. This amplitude of the visibility variation decreases with decreasing companion brightness (see the left panels in Fig. \[fig:80\]). Our binary search is limited to $FR\,\gtrsim\,0.05$ because even lower $FR$ would result in differential visibility variations compatible with models of a single resolved, extended emission structure, as discussed in the next section. Fitting binary models to such data of low differential visibility variation is very ambiguous. The visibilities of GM Aur show a statistically significant differential visibility signal of $\sim\,0.1$, different from the other four targets, but our analysis shows that this still small change in visibility does not appreciably alter the range of allowed binary solutions. We discuss this visibility trend further below in Sect. \[sec:111\].
The set of binary parameters, excluded by our data, does not significantly depend on the fact that so far we probed the data against simple, dust-free binary models only. It is possible that some extended, circumbinary dust emission has been resolved out by the interferometer, resulting in the visibilities below unity. This raises the concern that the putative binary signature in the visibility variation is not recovered by the analysis because the absolute visibility level has changed with respect to the calculated one, due to the missing over-resolved dust emission in the correlated flux. To probe if | the solutions to separations of multiples of the KI resolution merely. therefore, at the 99.7 % (67 %) confidence level of disapprove binary model, it is very unlikely that more than one out of our five targets hold a binary companion of 0.08 (0.05) brightness of the primary. ] { } The mass that such flux limits defend depend on both the luminosity of the primary and the assumed long time of the arrangement. Based on the PMS evolutionary tracks of @1999ApJ... 525.. 772P and an age of 1 Myr, flux ratios of 0.1 would mention to companion masses of about 1/4 of the primary bulk.
The fact that we can rule out binaries for the major part of the test FoV, and for a brightness ratio down to 0.05 is mostly related to the measured belittled [ * differential * ] { } visibility variations. In contrast, the [ * absolute * ] { } level constrains binary parameters to a lesser extent only, but is rather linked to the amount of (over-)resolved emission. An equal brightness binary would result in large ($ \gtrsim\,0.2 $) squared visibility mutant over the probed range of spatial frequency of the KI baseline. This amplitude of the visibility variation decreases with decreasing companion luminosity (see the leftover panels in Fig. \[fig:80\ ]). Our binary search is limited to $ FR\,\gtrsim\,0.05 $ because even lower $ FR$ would leave in differential visibility variations compatible with models of a single resolved, extended emission social organization, as discussed in the next section. Fitting binary models to such data of low differential visibility variation is very equivocal. The visibility of GM Aur show a statistically significant differential visibility signal of $ \sim\,0.1 $, different from the other four aim, but our analysis shows that this still small change in visibility does not appreciably alter the range of allowed binary solution. We discuss this visibility course further below in Sect. \[sec:111\ ].
The stage set of binary parameters, exclude by our data, does not importantly depend on the fact that so far we probe the data against simple, debris - free binary models only. It is possible that some extended, circumbinary dust emission has been resolved out by the interferometer, resulting in the visibilities below unity. This raises the business that the putative binary signature in the visibility variation is not reclaim by the analysis because the absolute visibility level has changed with respect to the calculated one, due to the missing over - purpose dust emission in the correlated flux. To probe if | thf solutions to separatioks of multiples of the NI resklution unly. Thus, at the 99.7 % (67 %) confidencx lecel od rejecting binary moddls, it is very unoikeot that morx than okz out lf obr five targets mave a binasy companion ox 0.08 (0.05) yrightness of the primary.]{} The masses that sicj flux limits teprexqnt svptnd on both the luminosity of the primarj and the assumed age of the system. Based oj thf PMS evolutionary tracks of @1999QpJ...525..772P qnd an age ow 1 Myr, flux ratios of 0.1 sould refer to companion masses of ayout 1/4 of thg lrlkary mass.
Thx fact that we can rule oud binaroes for the mamor pert if the tested FoV, and for a brightness tatio down tk 0.05 is mostly relaree to jhe maasufwd rmaml [*djfferejtiel*]{} visibilify variatiobs. In contrast, the [*sbfilute*]{} level cknstraynf binary parameters to a lesser extent mnlg, but is rather linked ro the amount of (over-)tesolved eiission. An equal brightness binary would result it larje ($\gtxwim\,0.2$) sdyaged visibility variations over the probed rands pf spatial frequcncies of the KI bssfloge. This amplijude of ths visibility variahion desreasws with dtcreaxing companion brightness (swe the left ianeos in Fig. \[fig:80\]). Our bnnary search is kimitrd to $FR\,\gtrsim\,0.05$ because zven lkwer $FR$ woupd result kn differential xispbilhty variations compatible rith modeos oy a singue rgsolved, extended fmisslmn structure, as didcussgd in dhe next sfction. Fitting binary models to such data of kof dpfferenticl vislbility variatijn is very ambnguous. Tke viskbilities kf GM Anr show a stwtistically shhnificant dihferentiaj viwibioity sieval of $\sim\,0.1$, difgerent from the othwr four targets, bub our znalysis shows ckau rhis still smakl zhaggv ii visytility does tot xppfrciabuy alter thc rxnge of allowed binary smlutjons. We discuss thos visibiluty trenq further belpw in Sect. \[sec:111\].
The stt of uinary paraketgrs, excluded by our data, does nkt signiflcaktly depend og thc fast that so far we probed the data against simple, vust-free binary models inly. It is possible uhat some extxnded, sircumbindry dust emission haw been resolved omt by the interferometsr, resgltinh in the visibilities below unity. This raises the concern that the putatice binerr signature in yhe vpsiyilnty varyatili is not recoverec by the analysis because the ausolute viviyility level has changed with rdspect to the calculated one, due to tge missimg over-resolved dust emission im the correlated flux. To probe id | the solutions to separations of multiples of resolution Thus, at 99.7 % (67 binary it is very that more than out of our five targets have binary companion of 0.08 (0.05) brightness of the primary.]{} The masses that such limits represent depend on both the luminosity of the primary and the assumed of system. on PMS evolutionary tracks of @1999ApJ...525..772P and an age of 1 Myr, flux ratios of 0.1 would to companion masses of about 1/4 of the mass. The fact that can rule out binaries for major of the FoV, for brightness ratio down 0.05 is mostly related to the measured small [*differential*]{} visibility variations. In contrast, the [*absolute*]{} level constrains parameters to extent only, is linked the amount of An equal brightness binary would result squared visibility variations over the probed range of frequencies of KI baseline. This amplitude of the variation decreases with decreasing companion brightness (see the panels in Fig. \[fig:80\]). Our binary search is limited to $FR\,\gtrsim\,0.05$ because even lower $FR$ in differential visibility variations with models of single extended structure, discussed in next section. Fitting binary models to such data of low differential variation is very ambiguous. The visibilities of GM Aur show significant visibility signal of different from the other targets, our analysis shows that small in appreciably the of allowed binary solutions. discuss this visibility trend further in Sect. \[sec:111\]. The by our data, does not significantly depend on fact that so far we probed the against simple, dust-free binary models only. It is possible that some extended, dust emission resolved out by the interferometer, resulting in the below unity. This raises concern that the putative binary signature in the visibility is recovered by analysis because the visibility level has with respect to one, due the over-resolved the correlated flux. To probe if | the solutions to separations Of multipleS of thE KI ResOlUtioN onlY. Thus, at the 99.7 % (67 %) confIDencE level of rejecting binarY modeLs, IT is vERy UnlikEly that MOrE THan OnE oUt oF oUR fIve taRgeTs have a Binary compAniOn Of 0.08 (0.05) brightness OF tHe primary.]{} THe mAsses that sucH flUx limiTs RepREsent DepEnd on Both thE LuminoSity of the PrIMary anD The assuMED aGe of The system. Based on tHE Pms evolutionary tRacks oF @1999APj...525..772P AND an Age Of 1 Myr, flux rAtIos of 0.1 WOuld refER tO COMpaNIon masses of abOut 1/4 of the priMAry Mass.
ThE fAct THat we cAn rulE oUT biNaries for thE majOr part of tHe testED FoV, and FOr a brigHtness RatIo dOwn tO 0.05 Is MoStlY rELatED tO thE MeaSured smaLl [*DiFfereNtiaL*]{} VISIbilIty VariAtionS. In contrast, thE [*abSoluTE*]{} leVel coNstraIns bInAry paRameteRs to a LeSser extent only, bUt is Rather linKed To The AmOunt oF (Over-)reSolVed EmissioN. An equaL BriGhTNESs Binary would result iN lARGe ($\Gtrsim\,0.2$) sqUared vISiBiLIty variaTiOns Over THE probEd raNGe Of spatiaL frequENcIeS of the Ki bAselinE. THis AmpLitudE Of thE visibIlity varIatioN Decreases with dECreasing compaNIoN BRiGHtneSs (sEe the left paNels IN Fig. \[Fig:80\]). OUR bInaRY searCh is lImITeD To $FR\,\gtrsim\,0.05$ because evEn Lower $Fr$ woulD result in diffErential viSIBIlity varIatiONs COmpatible with mOdels Of a single rESolved, exTendeD emissioN structurE, AS discussEd iN thE neXt sECTiOn. Fitting binaRY ModeLs To such dAta Of low diFfeRenTiaL viSiBility varIation is VeRy AmBiGuoUs. The VIsibilitIeS of gM aur Show a STatistIcallY sigNiFiCAnt DiffereNTiAL VisiBiLiTy siGnaL oF $\sim\,0.1$, dIffeREnt From the Other four TarGEts, bUt OuR analysIs shows that thIs Still small ChAngE in visIBIlity doeS not appreciably alter the RAnge of aLloWed biNary Solutions. we dIscuss ThiS VisibiLity trEnd fuRtHer BELow in sECt. \[Sec:111\].
thE set of binaRY ParAmeteRs, ExclUded by oUr data, does not signiFIcaNtly depend on tHe fAct tHAT sO faR We PRobEd THe dATA against simple, dUst-free binArY MoDels only. It IS poSsIble thaT some exTendeD, CircumbInary dust Emission hAs Been RESolVed out by thE interfeRometer, reSUltinG In The viSibIlitieS bEloW unitY. This rAIseS the cOncern ThAt the pUtatiVe Binary siGnature in the visibility vAriatiOn is nOt rEcovered bY thE AnaLysis becaUse tHe absolute VisIbiLity lEveL Has chAngeD WiTh rESpect To thE CalculateD OnE, duE TO tHe missing ovER-REsoLved dUst EMissioN in tHe correlated flux. TO Probe if | the solutions to separati ons of mul tiple s o f t he KIreso lution only. T h us,at the 99.7 % (67 %) c onfid en c e le v el of r ejectin g b i n ary m od els ,i tis ve ryunlikel y that mor e t ha n one out of ou r five tar get s have a bin ary compa ni ono f 0.0 8 ( 0.05) brigh t ness o f the pri ma r y.]{}T he mass e s t hatsuch flux limitsr ep r esent depend o n both t h el u min osi ty of thepr imary and the as s u m eda ge of the sys tem. Basedo n t he PMS e vol u tionar y tra ck s of @1999ApJ.. .525 ..772P an d an a g e of 1M yr, flu x rati osof0.1w ou ld re fe r to co mpa n ion massesof a bout1/4o f t he p rim arymass.
The fact th atwe c a n r ule o ut bi nari es forthe ma jor p ar t of the tested FoV , and for abr igh tn ess r a tio do wnto0.05 is mostly rel at e d to the measured smal l[ * di fferenti al*]{} vi si b ility va ri ati ons. I n con tras t ,the [*ab solute * ]{ }level c on strain sbin ary para m eter s to a lesserexten t only, but isr ather linkedt ot h ea moun t o f (over-)re solv e d em issi o n. An equal brig ht n es s binary would resul tin lar ge ($ \gtrsim\,0.2$ ) squaredv i s ibilityvari a ti o ns over the pr obedrange of s p atial fr equen cies ofthe KI ba s e line. Th isamp lit ude o fthe visibilit y vari at ion dec rea ses wit h d ecr eas ing c ompanionbrightne ss ( se ethe left panels i nFig .\[f ig:80 \ ]). Ou r bin aryse ar c h i s limit e dt o $FR \, \g trsi m\, 0. 05$ b ecau s e e ven low er $FR$ w oul d res ul tin diff erential visi bi lity varia ti ons compa t i ble with models of a single res o lved, e xte ndedemis sion stru ctu re, as di s cussed in th e nex tsec t i on. F i t ti ngbi nary model s tosuchda ta o f low d ifferential visibi l ity variation is ve ry a m b ig uou s .T hevi s ibi l i ties of GM Aurshow a sta ti s ti cally sign i fic an t diffe rential visi b ility s ignal of$\sim\,0. 1$ , di f f ere nt from th e otherfour targ e ts, b u tour a nal ysis s ho wsthatthis s t ill smal l chan ge in vi sibil it y does n ot appreciably alter th e rang e ofall owed bina rys olu tions. We dis cuss thisvis ibi litytre n d fur ther be low in Se ct.\ [sec:111\ ] .
Th e se t of binary p a ram eters , e x cluded byour data, does no t significantly dep e n d o n t h e fa ct that so far w e p ro b e d the da ta against si mple, du st - freebinary model s only. I ti s poss ible th at some e xte nd e d, circ um bi n ary du st e mi ssionhas be e n re s o lved out by theinter f e romet e r,resul ti ng in t h e vi sibilities below unit y. Thi s ra isesthe con ce rn tha t t he putativeb inary sig natur e in th evisi bil ity va riat i o n isnotre cov ered by t h e a n al ys i s b ecau se th eabso lute visi b ility le vel has cha ng edw i th res p ec t to the cal cul atedo n e, due tot he m i ss i ng ov er-res olveddust em i ssi on in the co r r elated fl ux. To pr o beif | the_solutions to_separations of multiples of_the KI_resolution_only. Thus,_at_the 99.7 % (67 %)_confidence level of_rejecting binary models, it_is very unlikely_that_more than one out of our five targets have a binary companion of 0.08_(0.05)_brightness of_the_primary.]{}_The masses that such flux_limits represent depend on both_the luminosity_of the primary and the assumed age of_the_system. Based on_the PMS evolutionary tracks of @1999ApJ...525..772P and an age_of 1 Myr, flux ratios of 0.1_would refer to_companion_masses_of about 1/4 of_the primary mass.
The fact that we_can rule out binaries for the_major part of the tested FoV, and_for a brightness ratio down to_0.05 is mostly related to_the measured_small [*differential*]{} visibility variations. In_contrast, the [*absolute*]{}_level constrains_binary parameters to_a lesser extent only, but is_rather linked to_the amount of (over-)resolved emission. An_equal_brightness binary would_result_in_large ($\gtrsim\,0.2$)_squared visibility variations_over_the probed_range_of spatial frequencies of the KI_baseline._This amplitude of the visibility variation decreases_with decreasing companion brightness_(see_the left panels in_Fig. \[fig:80\]). Our binary search is_limited to $FR\,\gtrsim\,0.05$ because even lower_$FR$ would_result in_differential visibility variations compatible with models of a single resolved, extended_emission structure, as discussed in the_next section. Fitting binary_models to_such_data of low_differential_visibility variation_is very ambiguous. The visibilities of GM_Aur show_a statistically significant differential visibility signal_of $\sim\,0.1$, different from_the_other four targets, but our analysis_shows that this still small change_in visibility does not appreciably_alter_the_range of allowed binary solutions._We discuss this visibility trend further_below in Sect. \[sec:111\].
The_set of binary parameters, excluded by our_data,_does not significantly depend on the_fact_that so far we probed the_data_against_simple, dust-free binary models only._It is possible that some extended,_circumbinary dust emission has been resolved out by the_interferometer, resulting in_the visibilities below unity. This_raises_the_concern that the putative binary signature in the visibility variation_is not_recovered by the_analysis because the absolute visibility level has changed with respect_to the calculated one, due to the_missing over-resolved dust emission in the correlated flux. To probe if |
\omega\right\vert \ll\left\vert \delta_{i}\right\vert $ such that $\delta
_{1}\approx\delta_{2}$. In the tight-binding limit, the external potential corresponds to the Hamiltonian $$H_{\text{e}}\left( t\right) =\hbar\Omega\sum_{l=0,1\text{; }q=\text{L,R}}\cos\left( \varphi_{l,q}\mathbf{+}\omega t\right) b_{l,q}^{\dag}b_{l,q},
\label{eq:He}$$ where $\varphi_{l,\text{L/R}}=\mp\frac{k_{x}\lambda_{\text{s}}}{4}+\frac{lk_{y}\lambda_{\text{s}}}{2}$. The wavevectors $k_{x}$ and $k_{y}$ can be represented as $k_{x}=k_{1}\cos\phi_{1}-k_{2}\cos\phi_{2}$ and $k_{y}=k_{1}\sin\phi_{1}-k_{2}\sin\phi_{2}$. To eliminate the time-dependent Hamiltonian $H_{\text{e}}\left( t\right) $, we perform a unitary transformation $U=\!\exp[-i\frac{\Omega}{\omega}\sum_{l=0,1\text{; }q=\text{L,R}}\sin\left(
\varphi_{l,q}\mathbf{+}\omega t\right) b_{l,q}^{\dag}b_{l,q}]$, which yields the following effective Hamiltonian$$\begin{aligned}
H_{\text{eff}}= & \sum_{l}\frac{\Delta}{2}\left( b_{l,\text{R}}^{\dag
}b_{l,\text{R}}-b_{l,\text{L}}^{\dag}b_{l,\text{L}}\right) \nonumber\\
& -\sum_{\substack{l\neq-1,0,1\\q=\text{L,R}}}J_{y}b_{l+1,q}^{\dag}b_{l,q}-\sum_{l\neq0,1}J_{x}b_{l,\text{L}}^{\dag}b_{l,\text{R}}+\text{H.c.}\nonumber\\
& -\sum | \omega\right\vert \ll\left\vert \delta_{i}\right\vert $ such that $ \delta
_ { 1}\approx\delta_{2}$. In the tight - binding limit, the external potential corresponds to the Hamiltonian $ $ H_{\text{e}}\left ( t\right) = \hbar\Omega\sum_{l=0,1\text {; } q=\text{L, R}}\cos\left ( \varphi_{l, q}\mathbf{+}\omega t\right) b_{l, q}^{\dag}b_{l, q },
\label{eq: He}$$ where $ \varphi_{l,\text{L / R}}=\mp\frac{k_{x}\lambda_{\text{s}}}{4}+\frac{lk_{y}\lambda_{\text{s}}}{2}$. The wavevectors $ k_{x}$ and $ k_{y}$ can be represent as $ k_{x}=k_{1}\cos\phi_{1}-k_{2}\cos\phi_{2}$ and $ k_{y}=k_{1}\sin\phi_{1}-k_{2}\sin\phi_{2}$. To rule out the time - dependent Hamiltonian $ H_{\text{e}}\left ( t\right) $, we perform a one transformation $ U=\!\exp[-i\frac{\Omega}{\omega}\sum_{l=0,1\text {; } q=\text{L, R}}\sin\left (
\varphi_{l, q}\mathbf{+}\omega t\right) b_{l, q}^{\dag}b_{l, q}]$, which yields the following effective Hamiltonian$$\begin{aligned }
H_{\text{eff}}= & \sum_{l}\frac{\Delta}{2}\left ( b_{l,\text{R}}^{\dag
} b_{l,\text{R}}-b_{l,\text{L}}^{\dag}b_{l,\text{L}}\right) \nonumber\\
& -\sum_{\substack{l\neq-1,0,1\\q=\text{L, R}}}J_{y}b_{l+1,q}^{\dag}b_{l, q}-\sum_{l\neq0,1}J_{x}b_{l,\text{L}}^{\dag}b_{l,\text{R}}+\text{H.c.}\nonumber\\
& -\sum |
\omeha\right\vert \ll\left\vert \dtlta_{i}\right\vert $ socy that $\delta
_{1}\zpprox\deuta_{2}$. In the tight-binding limiv, thw exttgnal potential corresoonds to nhe Hamilronien $$H_{\text{e}}\left( t\cjght) =\hncr\Omefw\sum_{n=0,1\vext{; }q=\text{L,R}}\cos\keft( \varpvi_{l,q}\mathbf{+}\omegd g\rnght) b_{l,q}^{\dag}b_{l,q},
\label{eq:He}$$ where $\varphy_{l,\text{L/T}}=\mo\frac{k_{x}\lambda_{\tgxt{s}}}{4}+\fgas{lk_{y}\mambda_{\text{s}}}{2}$. The wavevectors $k_{x}$ and $k_{y}$ can be representrd as $k_{x}=k_{1}\cos\phi_{1}-k_{2}\cos\phi_{2}$ and $k_{y}=k_{1}\din\phi_{1}-k_{2}\sin\phi_{2}$. To epiminate thg tiiw-dependent Hxmiltonian $H_{\text{e}}\left( t\right) $, we perform a unitary gransyormation $U=\!\grp[-i\fgdc{\Omega}{\omege}\sum_{l=0,1\tvxt{; }q=\text{L,R}}\sik\keft(
\vasphi_{l,q}\msthbf{+}\omega t\rinht) u_{l,q}^{\dqg}b_{l,q}]$, which yields thx following effectivg Hamiltonhau$$\begin{aligned}
H_{\text{eff}}= & \sum_{n}\fraw{\Delgq}{2}\lewt( b_{k,\tsxt{R}}^{\dah
}b_{l,\vext{R}}-b_{l,\text{M}}^{\dag}b_{l,\text{L}}\eight) \nonumber\\
& -\sum_{\xufwtack{l\neq-1,0,1\\q=\texf{L,R}}}J_{y}b_{j+1,q}^{\qag}b_{l,q}-\sum_{l\neq0,1}J_{x}b_{l,\text{L}}^{\dag}b_{l,\text{R}}+\text{H.c.}\nmnujber\\
& -\sum | \omega\right\vert \ll\left\vert \delta_{i}\right\vert $ such that $\delta the limit, the potential corresponds to }q=\text{L,R}}\cos\left( t\right) b_{l,q}^{\dag}b_{l,q}, \label{eq:He}$$ $\varphi_{l,\text{L/R}}=\mp\frac{k_{x}\lambda_{\text{s}}}{4}+\frac{lk_{y}\lambda_{\text{s}}}{2}$. The wavevectors and $k_{y}$ can be represented as and $k_{y}=k_{1}\sin\phi_{1}-k_{2}\sin\phi_{2}$. To eliminate the time-dependent Hamiltonian $H_{\text{e}}\left( t\right) $, we perform a transformation $U=\!\exp[-i\frac{\Omega}{\omega}\sum_{l=0,1\text{; }q=\text{L,R}}\sin\left( \varphi_{l,q}\mathbf{+}\omega t\right) b_{l,q}^{\dag}b_{l,q}]$, which yields the following effective Hamiltonian$$\begin{aligned} H_{\text{eff}}= \sum_{l}\frac{\Delta}{2}\left( }b_{l,\text{R}}-b_{l,\text{L}}^{\dag}b_{l,\text{L}}\right) & & -\sum |
\omega\right\vert \ll\left\vert \dElta_{i}\right\Vert $ sUch ThaT $\dElta
_{1}\ApprOx\delta_{2}$. In the tiGHt-biNding limit, the external pOtentIaL CorrESpOnds tO the HamILtONIan $$h_{\tExT{e}}\lEfT( T\rIght) =\hBar\omega\suM_{l=0,1\text{; }q=\texT{L,R}}\CoS\left( \varphi_{l,Q}\MaThbf{+}\omega t\RigHt) b_{l,q}^{\dag}b_{l,q},
\lAbeL{eq:He}$$ wHeRe $\vARphi_{l,\TexT{L/R}}=\mp\Frac{k_{x}\LAmbda_{\tExt{s}}}{4}+\frac{lK_{y}\LAmbda_{\tEXt{s}}}{2}$. The wAVEvEctoRs $k_{x}$ and $k_{y}$ can be repREsENted as $k_{x}=k_{1}\cos\phI_{1}-k_{2}\cos\pHi_{2}$ ANd $K_{Y}=K_{1}\siN\phI_{1}-k_{2}\sin\phi_{2}$. To ElIminaTE the timE-DePENDenT hamiltonian $H_{\tExt{e}}\left( t\riGHt) $, wE perfoRm A unITary trAnsfoRmATioN $U=\!\exp[-i\frac{\OMega}{\Omega}\sum_{l=0,1\Text{; }q=\tEXt{L,R}}\sin\LEft(
\varpHi_{l,q}\maThbF{+}\omEga t\RIgHt) B_{l,q}^{\DaG}B_{l,q}]$, WHiCh yIEldS the follOwInG effeCtivE hAMIltoNiaN$$\begIn{aliGned}
H_{\text{eff}}= & \sUm_{l}\Frac{\dEltA}{2}\left( B_{l,\texT{R}}^{\daG
}b_{L,\text{r}}-b_{l,\texT{L}}^{\dag}B_{l,\Text{L}}\right) \nonumBer\\
& -\sUm_{\substacK{l\nEq-1,0,1\\Q=\teXt{l,R}}}J_{y}b_{L+1,Q}^{\dag}b_{l,Q}-\suM_{l\nEq0,1}J_{x}b_{l,\tExt{L}}^{\dag}B_{L,\teXt{r}}+\TEXt{h.c.}\nonumber\\
& -\sum |
\omega\right\vert \ll\lef t\vert \de lta_{ i}\ rig ht \ver t $such that $\de l ta
_ {1}\approx\delta_{2}$. In t he tigh t -b indin g limit , t h e ex te rn alpo t en tialcor respond s to the H ami lt onian $$H_{\ t ex t{e}}\left ( t\right) =\ hba r\Omeg a\ sum _ {l=0, 1\t ext{; }q=\t e xt{L,R }}\cos\le ft ( \var p hi_{l,q } \ ma thbf {+}\omega t\right ) b _{l,q}^{\dag}b _{l,q} ,\ la b e l{e q:H e}$$ where $ \varp h i_{l,\t e xt { L / R}} = \mp\frac{k_{x }\lambda_{\ t ext {s}}}{ 4} +\f r ac{lk_ {y}\l am b da_ {\text{s}}} {2}$ . The wav evecto r s $k_{x } $ and $ k_{y}$ ca n b e re p re se nte da s $ k _{ x}= k _{1 }\cos\ph i_ {1 }-k_{ 2}\c o s \ p hi_{ 2}$ and $k_{ y}=k_{1}\sin\ phi _{1} - k_{ 2}\si n\phi _{2} $. To e limina te th etime-dependentHami ltonian $ H_{ \t ext {e }}\le f t( t\ rig ht) $, we perfor m aun i t a ry transformation $U =\ ! \ ex p[-i\fra c{\Ome g a} {\ o mega}\su m_ {l= 0,1\ t e xt{;}q=\ t ex t{L,R}}\ sin\le f t(
\ varphi_ {l ,q}\ma th bf{ +}\ omega t\ri ght) b_{l,q}^ {\dag } b_{l,q}]$, whi c h yields thef ol l o wi n g ef fec tive Hamilt onia n $$\b egin { al ign e d}
H_ {\tex t{ e ff } }= & \sum_{l}\fra c{ \Delta }{2}\ left( b_{l,\ text{R}}^{ \ d a g
}b_{l, \tex t {R } }-b_{l,\text{L }}^{\ dag}b_{l,\ t ext{L}}\ right ) \nonum ber\\
& - \ sum_{\su bst ack {l\ neq - 1 ,0 ,1\\q=\text{L , R }}}J _{ y}b_{l+ 1,q }^{\dag }b_ {l, q}- \su m_ {l\neq0,1 }J_{x}b_ {l ,\ te xt {L} }^{\d a g}b_{l,\ te xt{ R} }+\ text{ H .c.}\n onumb er\\
& - \su m |
\omega\right\vert \ll\left\vert_\delta_{i}\right\vert $_such that $\delta
_{1}\approx\delta_{2}$. In_the tight-binding_limit,_the external_potential_corresponds to the_Hamiltonian $$H_{\text{e}}\left( _t\right) =\hbar\Omega\sum_{l=0,1\text{; }q=\text{L,R}}\cos\left(_ \varphi_{l,q}\mathbf{+}\omega t\right)__b_{l,q}^{\dag}b_{l,q},
\label{eq:He}$$ where $\varphi_{l,\text{L/R}}=\mp\frac{k_{x}\lambda_{\text{s}}}{4}+\frac{lk_{y}\lambda_{\text{s}}}{2}$. The wavevectors $k_{x}$ and $k_{y}$ can be represented as $k_{x}=k_{1}\cos\phi_{1}-k_{2}\cos\phi_{2}$ and_$k_{y}=k_{1}\sin\phi_{1}-k_{2}\sin\phi_{2}$._To eliminate_the_time-dependent_Hamiltonian $H_{\text{e}}\left( t\right) _$, we perform a unitary_transformation $U=\!\exp[-i\frac{\Omega}{\omega}\sum_{l=0,1\text{;_}q=\text{L,R}}\sin\left(
\varphi_{l,q}\mathbf{+}\omega t\right) b_{l,q}^{\dag}b_{l,q}]$, which yields the following_effective_Hamiltonian$$\begin{aligned}
H_{\text{eff}}= &_ \sum_{l}\frac{\Delta}{2}\left( b_{l,\text{R}}^{\dag
}b_{l,\text{R}}-b_{l,\text{L}}^{\dag}b_{l,\text{L}}\right) \nonumber\\
& -\sum_{\substack{l\neq-1,0,1\\q=\text{L,R}}}J_{y}b_{l+1,q}^{\dag}b_{l,q}-\sum_{l\neq0,1}J_{x}b_{l,\text{L}}^{\dag}b_{l,\text{R}}+\text{H.c.}\nonumber\\
& -\sum |
Phys. Rep. 325, 83 (1999). X-G. Huang, Rep. Prog. Phys. 79, 076302 (2016).
J. Schwinger, Phys. Rev. 82, 664(1951). See for example, G. Dunne and T.M. Hall, Phys. Lett. B419, 322(1998); Phys. Rev. D60, 065002(1999). W. Pauli, Rev. Mod. Phys. 13, 203(1941). Q.-G. Lin, J. Phys. G. 25, 1793(1999). H.K. Lee and Y.S. Yoon, JHEP 0603, 078(2006). S.P. Gavrilov and D.M. Gitman Phys. Rev. D 87, 125025 (2013).
S.P. Kim, H.K. Lee and Y.S. Yoon, Phys.Rev.D82,025015(2010) and references therein. H.K. Lee and Y.S. Yoon, JHEP 0703,086 (2007). Y.M. Koh, H.K. Lee, W-G. Paeng and Y. Yoon, J. Korean Phys. Soc. 56, 1884 (2010). C. Giunti and A. Studenkin, Rev. Mod. Phys. 87, 531(2015). H.K.Lee, Phys. Rev. D84, 077302 (2011).
J. Ambjorn, R.J. Hughes and N.K. Nielson, Ann. Phys.(NY), 150, 92(1983):J. Ambjorn and P. Olesen, Nucl.Phys.B 330, 193(1990) and references therein. N.K. Nielson and P. Olesen, Nucl. Phys. B 144, 376(1978). N. Graham and R.L. Jaffe, Phys.Lett. B435, 145 (1998). H.K. Lee and Y. Yoon, Mod. Phys. Lett. A 22, 2081(2007). K. Fujikawa and R. Shrock, Phys. Rev. Lett. 45, 963 (1980). See for example, D. Forero, M. Tortola, J. Valle, Phys. Rev. D 90, 093006(2014). A.G. Beda, V.B. Brud | Phys. Rep. 325, 83 (1999). X - G. Huang, Rep. Prog. Phys. 79, 076302 (2016).
J. Schwinger, Phys. Rev. 82, 664(1951). See for example, G. Dunne and T.M. Hall, Phys. Lett. B419, 322(1998); Phys. Rev. D60, 065002(1999). W. Pauli, Rev. Mod. Phys. 13, 203(1941). Q.-G. Lin, J. Phys. G. 25, 1793(1999). H.K. Lee and Y.S. Yoon, JHEP 0603, 078(2006). S.P. Gavrilov and D.M. Gitman Phys. Rev. D 87, 125025 (2013).
S.P. Kim, H.K. Lee and Y.S. Yoon, Phys. Rev. D82,025015(2010) and references therein. H.K. Lee and Y.S. Yoon, JHEP 0703,086 (2007). Y.M. Koh, H.K. Lee, W - G. Paeng and Y. Yoon, J. Korean Phys. Soc. 56, 1884 (2010). C. Giunti and A. Studenkin, Rev. Mod. Phys. 87, 531(2015). H.K.Lee, Phys. Rev. D84, 077302 (2011).
J. Ambjorn, R.J. Hughes and N.K. Nielson, Ann. Phys.(NY), 150, 92(1983):J. Ambjorn and P. Olesen, Nucl. Phys. B 330, 193(1990) and reference point therein. N.K. Nielson and P. Olesen, Nucl. Phys. B 144, 376(1978). N. Graham and R.L. Jaffe, Phys. Lett. B435, 145 (1998). H.K. Lee and Y. Yoon, Mod. Phys. Lett. A 22, 2081(2007). K. Fujikawa and R. Shrock, Phys. Rev. Lett. 45, 963 (1980). examine for example, D. Forero, M. Tortola, J. Valle, Phys. Rev. D 90, 093006(2014). A.G. Beda, V.B. Brud | Phjs. Rep. 325, 83 (1999). X-G. Huang, Rep. Pvog. Phys. 79, 076302 (2016).
J. Schcunger, 'hys. Reb. 82, 664(1951). See wor example, G. Dunne and T.M. Hell, Phys. Oett. B419, 322(1998); Phys. Rev. D60, 065002(1999). W. Oauli, Rev. Mod. Phyw. 13, 203(1941). W.-T. Lin, J. Phba. G. 25, 1793(1999). H.K. Lee zkd Y.S. Boon, JHEP 0603, 078(2006). S.P. Navrilov ang D.M. Gitman Phfs. Rzv. D 87, 125025 (2013).
S.P. Kim, H.K. Lee and Y.S. Yoon, Phyf.Rev.D82,025015(2010) amd references thgrein. R.K. Lsv cnd Y.S. Yoon, JHEP 0703,086 (2007). Y.M. Koh, H.K. Les, W-G. Patng and Y. Yoon, J. Kprean Phys. Soc. 56, 1884 (2010). C. Giunti and A. Studenkin, Rev. Mld. Phys. 87, 531(2015). H.J.Lee, Phys. Rev. D84, 077302 (2011).
G. Ambjorn, G.L. Hughes ans N.K. Nielson, Ann. Phys.(NY), 150, 92(1983):J. Ambjurn aud P. Olesen, Nycl.Ovys.B 330, 193(1990) and ceferegces therein. N.K. Nielvon and P. Olesen, Nucl. Phbs. B 144, 376(1978). N. Graham and R.L. Jahfe, Phys.Lett. B435, 145 (1998). H.K. Jee and Y. Ykon, Mod. Phys. Lett. A 22, 2081(2007). K. Xujinawa qnd R. Ahcoci, Phys. Ret. Lett. 45, 963 (1980). Sse for example, D. Forero, M. Tortplw, J. Valle, Phys. Rev. D 90, 093006(2014). A.G. Beda, V.B. Brud | Phys. Rep. 325, 83 (1999). X-G. Huang, Phys. 076302 (2016). Schwinger, Phys. Rev. G. and T.M. Hall, Lett. B419, 322(1998); Rev. D60, 065002(1999). W. Pauli, Rev. Phys. 13, 203(1941). Q.-G. Lin, J. Phys. G. 25, 1793(1999). H.K. Lee and Yoon, JHEP 0603, 078(2006). S.P. Gavrilov and D.M. Gitman Phys. Rev. D 87, (2013). Kim, Lee Y.S. Yoon, Phys.Rev.D82,025015(2010) and references therein. H.K. Lee and Y.S. Yoon, JHEP 0703,086 (2007). Y.M. Koh, Lee, W-G. Paeng and Y. Yoon, J. Korean Soc. 56, 1884 (2010). Giunti and A. Studenkin, Rev. Phys. 531(2015). H.K.Lee, Rev. 077302 J. Ambjorn, R.J. and N.K. Nielson, Ann. Phys.(NY), 150, 92(1983):J. Ambjorn and P. Olesen, Nucl.Phys.B 330, 193(1990) and references therein. Nielson and Nucl. Phys. 144, N. and R.L. Jaffe, 145 (1998). H.K. Lee and Y. Lett. A 22, 2081(2007). K. Fujikawa and R. Phys. Rev. 45, 963 (1980). See for example, Forero, M. Tortola, J. Valle, Phys. Rev. D 093006(2014). A.G. Beda, V.B. Brud | Phys. Rep. 325, 83 (1999). X-G. Huang, Rep. Prog. Phys. 79, 076302 (2016).
j. Schwinger, phys. REv. 82, 664(1951). SEe fOr ExamPle, G. dunne and T.M. Hall, pHys. LEtt. B419, 322(1998); Phys. Rev. D60, 065002(1999). W. Pauli, Rev. MOd. PhyS. 13, 203(1941). Q.-g. lin, J. pHyS. G. 25, 1793(1999). H.K. LEe and Y.S. yOoN, jhEP 0603, 078(2006). s.P. gaVriLoV AnD D.M. GiTmaN Phys. ReV. D 87, 125025 (2013).
S.P. Kim, H.K. LEe aNd y.S. Yoon, Phys.ReV.d82,025015(2010) aNd referencEs tHerein. H.K. Lee aNd Y.s. Yoon, JhEp 0703,086 (2007). Y.M. kOh, H.K. LEe, W-g. PaenG and Y. YOOn, J. KorEan Phys. SoC. 56, 1884 (2010). C. gIunti aND A. StudeNKIn, rev. MOd. Phys. 87, 531(2015). H.K.Lee, Phys. ReV. d84, 077302 (2011).
J. aMbjorn, R.J. Hughes And N.K. NIeLSoN, aNn. PHys.(nY), 150, 92(1983):J. Ambjorn AnD P. OleSEn, Nucl.PHYs.b 330, 193(1990) AND reFErences thereiN. N.K. Nielson aND P. OLesen, NUcL. PhYS. B 144, 376(1978). N. GraHam anD R.l. jafFe, Phys.Lett. B435, 145 (1998). h.K. LeE and Y. Yoon, mod. PhyS. lett. A 22, 2081(2007). K. FUJikawa aNd R. ShrOck, phyS. Rev. lEtT. 45, 963 (1980). SEe fOr EXamPLe, d. FoREro, m. Tortola, j. VAlLe, PhyS. Rev. d 90, 093006(2014). a.g. bEda, V.b. BrUd | Phys. Rep. 325, 83 (1999) . X-G. Hua ng, R ep. Pr og . Ph ys.79, 076302 (20 1 6).
J. Schwinger, Phys. R ev. 8 2, 664( 1 95 1). S ee fore xa m p le, G .Dun ne an d T.M . H all, Ph ys. Lett.B41 9, 322(1998);P hy s. Rev. D6 0,065002(1999) . W . Paul i, Re v . Mod . P hys.13, 20 3 (1941) . Q.-G. L in , J. Ph y s. G. 2 5 , 1 793( 1999). H.K. Lee a n dY .S. Yoon, JHEP 0603, 0 7 8( 2 0 06) . S .P. Gavril ov andD .M. Git m an P h ys. Rev. D 87, 12 5025 (2013) .
S .P. Ki m, H. K . Leeand Y .S . Yo on, Phys.Re v.D8 2,025015( 2010)a nd refe r ences t herein . H .K. Lee an dY.S .Y oon , J HEP 070 3,086 (2 00 7) . Y.M . Ko h , H .K.Lee , W- G. Pa eng and Y. Yo on, J.K ore an Ph ys. S oc.56 , 188 4 (201 0). C .Giunti and A. S tude nkin, Rev . M od . P hy s. 87 , 531(2 015 ).H.K.Lee , Phys. Rev .D 8 4 ,077302 (2011).
J. A m b jo rn, R.J. Hughe s a nd N.K. Nie ls on, Ann . Phys. (NY) , 1 50, 92(1 983):J . A mb jorn an dP. Ole se n,Nuc l.Phy s .B 3 30, 19 3(1990)and r e ferences there i n. N.K. Niels o na n dP . Ol ese n, Nucl. Ph ys.B 144 , 37 6 (1 978 ) . N.Graha ma nd R.L. Jaffe, Phys.Le tt . B435 , 145 (1998). H.K. Lee and Y . Y oon, Mod . Ph y s. Lett. A 22, 20 81(20 07). K. Fu j ikawa an d R.Shrock,Phys. Rev . Lett. 45 , 9 63(19 80) . Se e for example , D. F or ero, M. To rtola,J.Val le, Ph ys . Rev. D90, 0930 06 (2 01 4) . A .G. B e da, V.B. B rud | Phys._Rep. 325,_83 (1999). X-G. Huang,_Rep. Prog._Phys._79, 076302_(2016).
J._Schwinger, Phys. Rev._82, 664(1951). See_for example, G. Dunne_and T.M. Hall,_Phys._Lett. B419, 322(1998); Phys. Rev. D60, 065002(1999). W. Pauli, Rev. Mod. Phys. 13, 203(1941)._Q.-G._Lin, J._Phys._G._25, 1793(1999). H.K. Lee and_Y.S. Yoon, JHEP 0603, 078(2006)._S.P. Gavrilov_and D.M. Gitman Phys. Rev. D 87, 125025_(2013).
S.P._Kim, H.K. Lee_and Y.S. Yoon, Phys.Rev.D82,025015(2010) and references therein. H.K. Lee_and Y.S. Yoon, JHEP 0703,086 (2007)._Y.M. Koh, H.K._Lee,_W-G._Paeng and Y. Yoon,_J. Korean Phys. Soc. 56, 1884_(2010). C. Giunti and A. Studenkin,_Rev. Mod. Phys. 87, 531(2015). H.K.Lee, Phys._Rev. D84, 077302 (2011).
J. Ambjorn, R.J._Hughes and N.K. Nielson, Ann._Phys.(NY), 150,_92(1983):J. Ambjorn and P. Olesen,_Nucl.Phys.B 330, 193(1990)_and references_therein. N.K. Nielson_and P. Olesen, Nucl. Phys. B_144, 376(1978). N._Graham and R.L. Jaffe, Phys.Lett. B435,_145_(1998). H.K. Lee_and_Y._Yoon, Mod._Phys. Lett. A_22,_2081(2007). K._Fujikawa_and R. Shrock, Phys. Rev. Lett._45,_963 (1980). See for example, D. Forero,_M. Tortola, J. Valle,_Phys._Rev. D 90, 093006(2014)._A.G. Beda, V.B. Brud |
beta$ to ${\mathcal{C}}$ does preserve finite products; this is part of proof of [@Glicksberg Thm. 3]. In conclusion, [Theorem [\[thm:main\]]{}]{}(a) applies to the situation $$\xymatrix{
{\mathcal{C}}=\mathsf{PsLocComp} \ar@<0.5ex>[r]^-{F=\beta} & \mathsf{CompHaus}={\mathcal{D}}\,, \ar@<0.5ex>[l]^-{G}
}$$ and shows that every algebraic structure on a pseudocompact and locally compact topological space $X$ ascends uniquely along the canonical map $\eta_X \colon X \to \beta X$ (which is the unit of the adjunction), as asserted in Theorem A(b).
It is proved in Munkres [@Munkres Cor. 82.2] (see also May [@May Chap. 3§8]) that a topological space has a universal covering space if and only if it is path connected, locally path connected, and semi-locally simply connected[^8]. Write ${\mathcal{D}}$ for the category of all such spaces, and ${\mathcal{C}}=\mathsf{SConn}$ for the category of simply connected topological spaces. Note that ${\mathcal{C}}$ and ${\mathcal{D}}$ are closed under finite products in ${\mathsf{Top}}$; see e.g. [@Chevalley II§7 Prop. 1 and II§8 Prop. 4].
Let ${\mathsf{Top}}_*$ be the category of pointed topological spaces and denote by ${\mathcal{C}}_*$ and ${\mathcal{D}}_*$ the full subcategories of ${\mathsf{Top}}_*$ whose objects are the ones in ${\mathcal{C}}$ and ${\mathcal{D}}$, respectively. As noted above, the inclusion functor $F \colon {\mathcal{C}}_* \to {\mathcal{D}}_*$ preserves finite products. Since we work with pointed spaces, the universal covering space ${\mathcal{C}}_* \ni \tilde{X} \to X$ of a space $X \in {\mathcal{D}}_*$ has the unique mapping property, see e.g. [@Chevalley II§8 Prop. 1], in other words, $F(\tilde{X})=\tilde{X} \to X$ is a universal arrow from | beta$ to $ { \mathcal{C}}$ does preserve finite products; this is part of proof of [ @Glicksberg Thm. 3 ]. In ending, [ Theorem [ \[thm: main\]]{}]{}(a) enforce to the situation $ $ \xymatrix {
{ \mathcal{C}}=\mathsf{PsLocComp } \ar@<0.5ex>[r]^-{F=\beta } & \mathsf{CompHaus}={\mathcal{D}}\, , \ar@<0.5ex>[l]^-{G }
} $ $ and testify that every algebraic social organization on a pseudocompact and locally compact topological space $ X$ ascends uniquely along the canonical function $ \eta_X \colon X \to \beta X$ (which is the whole of the adjunction), as assert in Theorem A(b).
It is proved in Munkres [ @Munkres Cor. 82.2 ] (see besides May [ @May Chap. 3§8 ]) that a topological space has a universal covering quad if and only if it is path connected, locally path connected, and semi - locally merely connected[^8 ]. Write $ { \mathcal{D}}$ for the category of all such spaces, and $ { \mathcal{C}}=\mathsf{SConn}$ for the class of simply connected topological space. notice that $ { \mathcal{C}}$ and $ { \mathcal{D}}$ are closed under finite products in $ { \mathsf{Top}}$; see e.g. [ @Chevalley II§7 Prop. 1 and II§8 Prop. 4 ].
Let $ { \mathsf{Top}}_*$ be the class of pointed topological spaces and denote by $ { \mathcal{C}}_*$ and $ { \mathcal{D}}_*$ the full subcategories of $ { \mathsf{Top}}_*$ whose objects are the ones in $ { \mathcal{C}}$ and $ { \mathcal{D}}$, respectively. As noted above, the inclusion body functor $ F \colon { \mathcal{C } } _ * \to { \mathcal{D}}_*$ continue finite products. Since we work with pointed spaces, the universal covering space $ { \mathcal{C } } _ * \ni \tilde{X } \to X$ of a space $ X \in { \mathcal{D}}_*$ has the unique function place, see e.g. [ @Chevalley II§8 Prop. 1 ], in other actor's line, $ F(\tilde{X})=\tilde{X } \to X$ is a universal arrow from | betw$ to ${\mathcal{C}}$ does presevve finite produers; thiv is pzrt of pfoof of [@Glicksberg Thm. 3]. In coicluwion, [Ukeorem [\[thm:main\]]{}]{}(a) applids to the situatiin $$\xbmatrix{
{\mathcem{C}}=\mathsn{'sLocDlmp} \cr@<0.5xx>[r]^-{F=\beta} & \mathsn{CompHaus}={\madhcal{D}}\,, \ar@<0.5ex>[l]^-{G}
}$$ xnb shows that every algebraic structuwe on a pdeudocompact agd lpsallg compact topological space $X$ ascehds unixuely along tne canonical map $\eta_X \coloj X \ho \beta X$ (which is the unit od thq adjunction), xs asserted in Theorem Z(b).
It is proved in Munkres [@Munkrer Cor. 82.2] (see also Nat [@Maj Chap. 3§8]) that e topojogical spacc has a univerxal covering siace mf abd only if it is path connected, locally path conteeted, and semi-locally winply wonnactea[^8]. Wrkte ${\methdal{D}}$ flr vhe categorg of all suxh spaces, and ${\mathcsl{S}}=\nathsf{SConn}$ fkr the cwtegory of simply connected topological spzces. Note that ${\mathcal{C}}$ and ${\mathcal{D}}$ are cloded under finite products in ${\mathsf{Top}}$; see e.g. [@Chevalley II§7 Psop. 1 aid II§8 Krip. 4].
Let ${\nahhsf{Top}}_*$ be the category of pointed topologicaj skacvs and denote by ${\iathcal{C}}_*$ anc ${\layrcal{D}}_*$ the fulu subccfefories of ${\mathsf{Too}}_*$ whose objexts are tre omes in ${\mathcal{C}}$ and ${\mathcal{E}}$, respectivejt. As noted above, tke inclusion funvtor $G \colon {\mathcal{C}}_* \to {\matkcal{D}}_*$ lreserves flnite prosjcts. Since we wofk eidh pointta spaces, the univqrsal covxring space ${\oathval{C}}_* \ny \tilde{X} \tl X$ on a space $X \in {\mathfal{D}}_*$ kas tve unique lapping property, see e.g. [@Chevalleb II§8 Prop. 1], in ojhes wmrds, $F(\tijde{X})=\tllde{X} \to X$ is a universal arrpw from | beta$ to ${\mathcal{C}}$ does preserve finite products; part proof of Thm. 3]. In the $$\xymatrix{ {\mathcal{C}}=\mathsf{PsLocComp} \ar@<0.5ex>[r]^-{F=\beta} \mathsf{CompHaus}={\mathcal{D}}\,, \ar@<0.5ex>[l]^-{G} }$$ shows that every algebraic structure on pseudocompact and locally compact topological space $X$ ascends uniquely along the canonical map \colon X \to \beta X$ (which is the unit of the adjunction), as in A(b). is in Munkres [@Munkres Cor. 82.2] (see also May [@May Chap. 3§8]) that a topological space has universal covering space if and only if it path connected, locally path and semi-locally simply connected[^8]. Write for category of such and for the category simply connected topological spaces. Note that ${\mathcal{C}}$ and ${\mathcal{D}}$ are closed under finite products in ${\mathsf{Top}}$; see [@Chevalley II§7 and II§8 4]. ${\mathsf{Top}}_*$ the category of spaces and denote by ${\mathcal{C}}_*$ and subcategories of ${\mathsf{Top}}_*$ whose objects are the ones ${\mathcal{C}}$ and respectively. As noted above, the inclusion $F \colon {\mathcal{C}}_* \to {\mathcal{D}}_*$ preserves finite products. we work with pointed spaces, the universal covering space ${\mathcal{C}}_* \ni \tilde{X} \to X$ of $X \in {\mathcal{D}}_*$ has unique mapping property, e.g. II§8 1], other words, \to X$ is a universal arrow from | beta$ to ${\mathcal{C}}$ does preservE finite proDucts; ThiS is PaRt of ProoF of [@Glicksberg THM. 3]. In cOnclusion, [Theorem [\[thm:maiN\]]{}]{}(a) appLiES to tHE sItuatIon $$\xymaTRiX{
{\MAthCaL{C}}=\MatHsF{pslocCoMp} \aR@<0.5ex>[r]^-{F=\beTa} & \mathsf{CoMpHAuS}={\mathcal{D}}\,, \ar@<0.5eX>[L]^-{G}
}$$ And shows thAt eVery algebraiC stRucturE oN a pSEudocOmpAct anD localLY compaCt topologIcAL space $x$ Ascends UNIqUely Along the canonical MAp $\ETa_X \colon X \to \betA X$ (whicH iS ThE UNit Of tHe adjunctiOn), As assERted in THEoREM a(b).
IT Is proved in MunKres [@Munkres cOr. 82.2] (sEe also maY [@MaY chap. 3§8]) thAt a toPoLOgiCal space has A uniVersal covEring sPAce if anD Only if iT is patH coNneCted, LOcAlLy pAtH ConNEcTed, ANd sEmi-localLy SiMply cOnneCTED[^8]. writE ${\maThcaL{D}}$ for The category of All Such SPacEs, and ${\MathcAl{C}}=\mAtHsf{SCOnn}$ for The caTeGory of simply conNectEd topologIcaL sPacEs. note tHAt ${\mathCal{c}}$ anD ${\mathcaL{D}}$ are clOSed UnDER FiNite products in ${\mathSf{tOP}}$; sEe e.g. [@ChevAlley Ii§7 prOp. 1 ANd II§8 Prop. 4].
leT ${\maThsf{tOP}}_*$ be thE catEGoRy of poinTed topOLoGiCal spacEs And denOtE by ${\MatHcal{C}}_*$ ANd ${\maThcal{D}}_*$ The full sUbcatEGories of ${\mathsf{tOp}}_*$ whose objectS ArE THe ONes iN ${\maThcal{C}}$ and ${\maThcaL{d}}$, resPectIVeLy. AS Noted Above, ThE InCLusion functor $F \colon {\MaThcal{C}}_* \To {\matHcal{D}}_*$ preserveS finite proDUCTs. Since wE worK WiTH pointed spaces, The unIversal covERing spacE ${\mathCal{C}}_* \ni \tiLde{X} \to X$ of A SPace $X \in {\mAthCal{d}}_*$ haS thE UNiQue mapping proPERty, sEe E.g. [@ChevaLleY II§8 Prop. 1], In oTheR woRds, $f(\tIlde{X})=\tildE{X} \to X$ is a UnIvErSaL arRow frOM | beta$ to ${\mathcal{C}}$ d oes preser ve fi nit e p ro duct s; t his is part of proo f of [@Glicksberg Thm. 3].In conc l us ion,[Theore m [ \ [ thm :m ai n\] ]{ } ]{ }(a)app lies to the situa tio n$$\xymatrix{ {\mathca l{C }}=\mathsf{P sLo cComp} \ ar@ < 0.5ex >[r ]^-{F =\beta } & \ma thsf{Comp Ha u s}={\m a thcal{D } } \, , \a r@<0.5ex>[l]^-{G} } $$ and shows t hat ev er y a l g ebr aic structure o n a p s eudocom p ac t a ndl ocally compac t topologic a l s pace $ X$ as c ends u nique ly alo ng the cano nica l map $\e ta_X \ c olon X\ to \bet a X$ ( whi chis t h eun itof the ad jun c tio n), as a ss er ted i n Th e o r e m A( b).
It is p roved in Munk res [@M u nkr es Co r. 82 .2](s ee al so May [@Ma yChap. 3§8]) tha t atopologic alsp ace h as au nivers alcov ering s pace if and o n l y i f it is path conne ct e d ,locallypath c o nn ec t ed, andse mi- loca l l y sim plyc on nected[^ 8]. Wr i te $ {\mathc al {D}}$fo r t hecateg o ry o f allsuch spa ces,a nd ${\mathcal{ C }}=\mathsf{SC o nn } $ f o r th e c ategory ofsimp l y co nnec t ed to p ologi cal s pa c es . Note that ${\mathc al {C}}$and $ {\mathcal{D}} $ are clos e d under fi nite pr o ducts in ${\ma thsf{ Top}}$; se e e.g. [@ Cheva lley II§ 7 Prop. 1 a nd II§8Pro p.4].
L e t $ {\mathsf{Top} } _ *$ b ethe cat ego ry of p oin ted to pol og ical spac es and d en ot eby ${ \math c al{C}}_* $and $ {\m athca l {D}}_* $ the ful lsu b cat egories of $ {\ma th sf {Top }}_ *$ whos e ob j ect s are t he ones i n $ { \mat hc al {C}}$ a nd ${\mathcal {D }}$, respe ct ive ly. As n oted abo ve, the inclusion funct o r $F \c olo n {\m athc al{C}}_*\to {\mat hca l {D}}_* $ pres erves f ini t e prod u c ts . S in ce we work w ith poin te d sp aces, t he universal cover i ngspace ${\math cal {C}} _ * \ ni\ ti l de{ X} \to X $ of a space $X \in {\mat hc a l{ D}}_*$ has the u nique m appingprope r ty, see e.g. [@C hevalleyII §8 P r o p.1], in oth er words , $F(\til d e{X}) = \t ilde{ X}\to X$ i s a univ ersala rro w fro m | beta$ to_${\mathcal{C}}$ does_preserve finite products; this_is part_of_proof of_[@Glicksberg_Thm. 3]. In conclusion,_[Theorem [\[thm:main\]]{}]{}(a) applies to_the situation $$\xymatrix{
_ {\mathcal{C}}=\mathsf{PsLocComp}_\ar@<0.5ex>[r]^-{F=\beta}_& \mathsf{CompHaus}={\mathcal{D}}\,, \ar@<0.5ex>[l]^-{G}
}$$ and shows that every algebraic structure on a pseudocompact_and_locally compact_topological_space_$X$ ascends uniquely along the_canonical map $\eta_X \colon X_\to \beta_X$ (which is the unit of the adjunction),_as_asserted in Theorem A(b).
It_is proved in Munkres [@Munkres Cor. 82.2] (see also May [@May Chap. 3§8])_that a topological space has a_universal covering space_if_and_only if it is_path connected, locally path connected, and_semi-locally simply connected[^8]. Write ${\mathcal{D}}$ for_the category of all such spaces, and_${\mathcal{C}}=\mathsf{SConn}$ for the category of simply_connected topological spaces. Note that_${\mathcal{C}}$ and_${\mathcal{D}}$ are closed under finite_products in ${\mathsf{Top}}$;_see e.g. [@Chevalley_II§7 Prop. 1 and II§8 Prop. 4].
Let_${\mathsf{Top}}_*$ be the category of pointed_topological spaces and_denote by ${\mathcal{C}}_*$ and ${\mathcal{D}}_*$ the_full_subcategories of ${\mathsf{Top}}_*$_whose_objects_are the_ones in ${\mathcal{C}}$_and_${\mathcal{D}}$, respectively._As_noted above, the inclusion functor $F_\colon_{\mathcal{C}}_* \to {\mathcal{D}}_*$ preserves finite products. Since_we work with pointed_spaces,_the universal covering space_${\mathcal{C}}_* \ni \tilde{X} \to X$_of a space $X \in {\mathcal{D}}_*$_has the_unique mapping_property, see e.g. [@Chevalley II§8 Prop. 1], in other words, $F(\tilde{X})=\tilde{X} \to X$ is_a universal arrow from |
, for instance. The displacement field is, then, $\Delta \mathbf{X}$=$\Delta \widetilde{\mathbf{X}}$ +$\Delta \mathbf{b}$. While the first moment of the displacement field can be free of noise if $\left\langle \mathbf{b} \right\rangle_t=0$, its second moment can be expressed as:
$$\begin{array}{ccl}
\left\langle (\Delta \mathbf{X})^2\right\rangle_t &=& \left\langle \widetilde{\mathbf{v}}^2\right\rangle_t \Delta t^2 + 2\left\langle \mathbf{b}^2\right\rangle_t + 2\left\langle \widetilde{\mathbf{a}}.\widetilde{\mathbf{v}}\right\rangle_t \Delta t^3 +o(\Delta t^3),\\\\
\end{array}
\label{eq:dx2}$$
using a $2^{nd}$ order Taylor expansion $\Delta \widetilde{\mathbf{X}}=\widetilde{\mathbf{v}}\Delta t+\frac{1}{2}\widetilde{\mathbf{a}}\Delta t^2+o(\Delta t^3)$. If we bin the Lagrangian displacement measurements by their position in space, and subtract the mean displacement in each bin (free of noise as shown above), we obtain an Eulerian field of fluctuating displacement that is of interest to characterize the flow structures in time or space. [As seen in equation \[eq:dx2\], $\langle (\Delta \mathbf{X})^2 \rangle_t$ has a first contribution proportional to the second order moment of the real fluctuating velocity $\left\langle \widetilde{\mathbf{v}}^2\right\rangle_t$, a second one equal to twice the noise variance, plus a correction ($\Delta t^3$) that depends on the cross correlation of the fluctuating velocity and acceleration.]{} The new method to remove the noise from the velocity statistics consists, simply, in calculating [$\left\langle \Delta \mathbf{X}\right\rangle_t$ and]{} $\left\langle (\Delta \mathbf{X})^2\right\rangle_t$ for multiple experiments where images of the particles in the flow are collected at increasing values of $\Delta t$. [At a first step, these statistics can be computed on a grid, $X_\text{grid}$, to obtain the Eulerian values $\left\langle | , for instance. The displacement field is, then, $ \Delta \mathbf{X}$=$\Delta \widetilde{\mathbf{X}}$ + $ \Delta \mathbf{b}$. While the first moment of the displacement sphere can be spare of noise if $ \left\langle \mathbf{b } \right\rangle_t=0 $, its second moment can be expressed as:
$ $ \begin{array}{ccl }
\left\langle (\Delta \mathbf{X})^2\right\rangle_t & = & \left\langle \widetilde{\mathbf{v}}^2\right\rangle_t \Delta t^2 + 2\left\langle \mathbf{b}^2\right\rangle_t + 2\left\langle \widetilde{\mathbf{a}}.\widetilde{\mathbf{v}}\right\rangle_t \Delta t^3 + o(\Delta t^3),\\\\
\end{array }
\label{eq: dx2}$$
use a $ 2^{nd}$ order Taylor expansion $ \Delta \widetilde{\mathbf{X}}=\widetilde{\mathbf{v}}\Delta t+\frac{1}{2}\widetilde{\mathbf{a}}\Delta t^2+o(\Delta t^3)$. If we bin the Lagrangian displacement measurement by their position in space, and subtract the mean supplanting in each bin (free of noise as show above), we obtain an Eulerian field of fluctuate translation that is of interest to characterize the flow structures in time or outer space. [ As seen in equation \[eq: dx2\ ], $ \langle (\Delta \mathbf{X})^2 \rangle_t$ has a beginning contribution proportional to the second order moment of the real fluctuating velocity $ \left\langle \widetilde{\mathbf{v}}^2\right\rangle_t$, a second one adequate to twice the noise variability, plus a correction ($ \Delta t^3 $) that depends on the cross correlation coefficient of the fluctuating velocity and acceleration. ] { } The new method to remove the noise from the velocity statistic dwell, simply, in calculating [ $ \left\langle \Delta \mathbf{X}\right\rangle_t$ and ] { } $ \left\langle (\Delta \mathbf{X})^2\right\rangle_t$ for multiple experiments where images of the particles in the flow are collected at increasing values of $ \Delta t$. [ At a inaugural step, these statistics can be calculate on a power system, $ X_\text{grid}$, to obtain the Eulerian values $ \left\langle | , fog instance. The displacemtnt field is, then, $\Delta \kathbf{S}$=$\Delta \wkdetilde{\mathbf{X}}$ +$\Delta \mathbf{u}$. Whule tye first moment of the displacelent fieod cen be free of nomae if $\lcyt\lanfpe \mctibf{b} \right\ranglg_t=0$, its secong moment can ba dx'ressed as:
$$\begin{array}{ccl}
\left\langlq (\Delta \mwthbf{X})^2\right\randle_t &=& \leff\langle \widetilde{\mathbf{v}}^2\right\rangls_t \Delte t^2 + 2\left\langle \mathbf{b}^2\right\rangle_t + 2\left\pangpe \widetilde{\mathbf{w}}.\widetilde{\mqthbs{c}}\right\rangle_g \Delta t^3 +o(\Delta t^3),\\\\
\end{array}
\label{eq:dx2}$$
using a $2^{nd}$ oraer Tcylor expanwiin $\Fglta \widetilve{\mathff{X}}=\widetilde{\mathbf{v}}\Dalta t+\ftac{1}{2}\widetilde{\mabhbf{a}}\Veltq t^2+o(\Delta t^3)$. If we bin the Lagrangian difplacemend jeasurements by tyeur povitimn iv spxce, aid aubtraft vhe mean diaplacement un each bin (free of njpxe as shown zbove), re obtain an Eulerian field of fluctuatinc djsplacement that is of unterest to characterlze the fjow structures in time or space. [As seen in equatimn \[eq:vx2\], $\launjd (\Dflta \mathbf{X})^2 \rangle_t$ has a first contribution lrppprtional to thc second order momrnh ps the real floctuatiuf belocity $\left\langlf \widetylde{\mqthbf{v}}^2\rigrt\ramgle_t$, a second one equal to twice the njuse variance, plus c correction ($\Delja t^3$) tnat depends on the crosr codrelation ov the fludguating velocity anc dcceleraukon.]{} The new methoq to remote thz noise wrom the vqlocity stwtistlws consists, simply, in cclculdting [$\left\pangle \Delta \mathbf{X}\right\rangle_v$ and]{} $\left\lanbla (\Dvlta \mathyf{X})^2\rigmt\rangle_t$ for mtltiple experikents wkere ioages of tge partmcles in the flow are colngcted at incrxasing vajues of $\Eelta t$. [Xt a first stel, these snauistics cab be computed on a grks, $X_\text{grid}$, to ibtqin the Euleriam vxluqs $\lxft\lagcle | , for instance. The displacement field is, \mathbf{X}$=$\Delta +$\Delta \mathbf{b}$. the first moment be of noise if \mathbf{b} \right\rangle_t=0$, its moment can be expressed as: $$\begin{array}{ccl} (\Delta \mathbf{X})^2\right\rangle_t &=& \left\langle \widetilde{\mathbf{v}}^2\right\rangle_t \Delta t^2 + 2\left\langle \mathbf{b}^2\right\rangle_t + 2\left\langle \widetilde{\mathbf{a}}.\widetilde{\mathbf{v}}\right\rangle_t t^3 +o(\Delta t^3),\\\\ \end{array} \label{eq:dx2}$$ using a $2^{nd}$ order Taylor expansion $\Delta \widetilde{\mathbf{X}}=\widetilde{\mathbf{v}}\Delta t^2+o(\Delta If bin Lagrangian displacement measurements by their position in space, and subtract the mean displacement in each bin of noise as shown above), we obtain an field of fluctuating displacement is of interest to characterize flow in time space. seen equation \[eq:dx2\], $\langle \mathbf{X})^2 \rangle_t$ has a first contribution proportional to the second order moment of the real fluctuating velocity \widetilde{\mathbf{v}}^2\right\rangle_t$, a equal to the variance, a correction ($\Delta depends on the cross correlation of and acceleration.]{} The new method to remove the from the statistics consists, simply, in calculating [$\left\langle \mathbf{X}\right\rangle_t$ and]{} $\left\langle (\Delta \mathbf{X})^2\right\rangle_t$ for multiple experiments images of the particles in the flow are collected at increasing values of $\Delta t$. first step, these statistics be computed on grid, to the values $\left\langle | , for instance. The displacemenT field is, thEn, $\DelTa \mAthBf{x}$=$\DelTa \wiDetilde{\mathbf{X}}$ +$\dElta \Mathbf{b}$. While the first moMent oF tHE disPLaCemenT field cAN bE FRee Of NoIse If $\LEfT\langLe \mAthbf{b} \rIght\rangle_T=0$, itS sEcond moment cAN bE expressed As:
$$\bEgin{array}{ccl}
\LefT\langlE (\DEltA \MathbF{X})^2\rIght\rAngle_t &=& \LEft\lanGle \widetiLdE{\Mathbf{V}}^2\Right\raNGLe_T \DelTa t^2 + 2\left\langle \mathBF{b}^2\RIght\rangle_t + 2\lefT\langlE \wIDeTILde{\MatHbf{a}}.\widetiLdE{\mathBF{v}}\right\RAnGLE_T \DeLTa t^3 +o(\Delta t^3),\\\\
\end{Array}
\label{eQ:Dx2}$$
uSing a $2^{nD}$ oRdeR taylor ExpanSiON $\DeLta \widetildE{\matHbf{X}}=\widetIlde{\maTHbf{v}}\DelTA t+\frac{1}{2}\wIdetilDe{\mAthBf{a}}\DELtA t^2+O(\DeLtA T^3)$. If WE bIn tHE LaGrangian DiSpLacemEnt mEASURemeNts By thEir poSition in space, And SubtRAct The meAn disPlacEmEnt in Each biN (free Of Noise as shown aboVe), we Obtain an EUleRiAn fIeLd of fLUctuatIng DisPlacemeNt that iS Of iNtERESt To characterize the fLoW STrUctures iN time oR SpAcE. [as seen in EqUatIon \[eQ:DX2\], $\langLe (\DeLTa \Mathbf{X})^2 \rAngle_t$ HAs A fIrst conTrIbutioN pRopOrtIonal TO the Second Order momEnt of THe real fluctuatINg velocity $\lefT\LaNGLe \WIdetIldE{\mathbf{v}}^2\rigHt\raNGle_t$, A secONd One EQual tO twicE tHE nOIse variance, plus a corReCtion ($\DElta t^3$) That depends on The cross coRRELation of The fLUcTUating velocity And acCeleration.]{} tHe new metHod to Remove thE noise froM THe velociTy sTatIstIcs CONsIsts, simply, in cALCulaTiNg [$\left\lAngLe \Delta \MatHbf{x}\riGht\RaNgle_t$ and]{} $\lEft\langlE (\DElTa \MaThbF{X})^2\rigHT\rangle_t$ FoR muLtIplE expeRIments Where ImagEs Of THe pArticleS In THE floW aRe CollEctEd At incReasINg vAlues of $\delta t$. [At a FirST steP, tHeSe statiStics can be comPuTed on a grid, $x_\tExt{Grid}$, to OBTain the EUlerian values $\left\langle | , for instance. The displa cement fie ld is , t hen ,$\De lta\mathbf{X}$=$\ D elta \widetilde{\mathbf{X} }$ +$ \D e lta\ ma thbf{ b}$. Wh i le t hefi rs t m om e nt of t hedisplac ement fiel d c an be free ofn oi se if $\le ft\ langle \math bf{ b} \ri gh t\r a ngle_ t=0 $, it s seco n d mome nt can be e x presse d as:
$ $ \ be gin{ array}{ccl}
\ l ef t \langle (\Delt a \mat hb f {X } ) ^2\ rig ht\rangle_ t&=& \ l eft\lan g le \ w ide t ilde{\mathbf{ v}}^2\right \ ran gle_t\D elt a t^2 + 2\le ft \ lan gle \mathbf {b}^ 2\right\r angle_ t + 2\le f t\langl e \wid eti lde {\ma t hb f{ a}} .\ w ide t il de{ \ mat hbf{v}}\ ri gh t\ran gle_ t \ D elta t^ 3 +o (\Del ta t^3),\\\\
\e n d{a rray}
\lab el{e q: dx2}$ $
usi ng a$2 ^{nd}$ order Ta ylor expansio n $ \D elt a\wide t ilde{\ mat hbf {X}}=\w idetild e {\m at h b f {v }}\Delta t+\frac{1 }{ 2 } \w idetilde {\math b f{ a} } \Delta t ^2 +o( \Del t a t^3) $. I f w e bin th e Lagr a ng ia n displ ac ementme asu rem entsb y th eir po sition i n spa c e, and subtrac t the mean dis p la c e me n t in ea ch bin (fre e of nois e as sh own above ), we o b ta i n an Eulerian field o f fluc tuati ng displaceme nt that is o f interes t to ch a racterize theflowstructures in timeor sp ace. [As seen ine q uation \ [eq :dx 2\] , $ \ l an gle (\Delta \ m a thbf {X })^2 \r ang le_t$ h asa f irs t c on tribution proport io na lto th e sec o nd order m ome nt of ther eal fl uctua ting v el o cit y $\lef t \l a n gle\w id etil de{ \m athbf {v}} ^ 2\r ight\ra ngle_t$,a s e cond o ne equalto twice theno ise varian ce , p lus ac o rrection ($\Delta t^3$) that de p ends on th e cro ss c orrelatio n o f theflu c tuatin g velo cityan d a c c elera t i on .]{ }The new me t h odto re mo ve t he nois e from the velocit y st atistics cons ist s, s i m pl y,i nc alc ul a tin g [$\left\langle\Delta \ma th b f{ X}\right\r a ngl e_ t$ and] {} $\le ft\la n gle (\D elta \mat hbf{X})^2 \r ight \ r ang le_t$ formultiple experime n ts wh e re imag esof the p art icles in th e fl ow ar e coll ec ted at incr ea sing val ues of $\Delta t$. [Ata firs t ste p,these sta tis t ics can be c ompu ted on a g rid , $ X_\te xt{ g rid}$ , to ob tai n theEule r ian value s $ \le f t \l angle | , for_instance. The_displacement field is, then,_$\Delta \mathbf{X}$=$\Delta_\widetilde{\mathbf{X}}$_+$\Delta \mathbf{b}$._While_the first moment_of the displacement_field can be free_of noise if_$\left\langle_\mathbf{b} \right\rangle_t=0$, its second moment can be expressed as:
$$\begin{array}{ccl}
\left\langle (\Delta_\mathbf{X})^2\right\rangle_t_&=& \left\langle_\widetilde{\mathbf{v}}^2\right\rangle_t_\Delta_t^2 + 2\left\langle \mathbf{b}^2\right\rangle_t +_2\left\langle \widetilde{\mathbf{a}}.\widetilde{\mathbf{v}}\right\rangle_t \Delta t^3 +o(\Delta_t^3),\\\\
_ \end{array}
\label{eq:dx2}$$
using a $2^{nd}$ order Taylor_expansion_$\Delta \widetilde{\mathbf{X}}=\widetilde{\mathbf{v}}\Delta t+\frac{1}{2}\widetilde{\mathbf{a}}\Delta_t^2+o(\Delta t^3)$. If we bin the Lagrangian displacement measurements_by their position in space, and_subtract the mean_displacement_in_each bin (free of_noise as shown above), we obtain_an Eulerian field of fluctuating displacement_that is of interest to characterize the_flow structures in time or space._[As seen in equation \[eq:dx2\],_$\langle (\Delta_\mathbf{X})^2 \rangle_t$ has a first_contribution proportional to_the second_order moment of_the real fluctuating velocity $\left\langle \widetilde{\mathbf{v}}^2\right\rangle_t$,_a second one_equal to twice the noise variance,_plus_a correction ($\Delta_t^3$)_that_depends on_the cross correlation_of_the fluctuating_velocity_and acceleration.]{} The new method to_remove_the noise from the velocity statistics consists,_simply, in calculating [$\left\langle_\Delta_\mathbf{X}\right\rangle_t$ and]{} $\left\langle (\Delta_\mathbf{X})^2\right\rangle_t$ for multiple experiments where_images of the particles in the_flow are_collected at_increasing values of $\Delta t$. [At a first step, these statistics_can be computed on a grid,_$X_\text{grid}$, to obtain the_Eulerian values_$\left\langle |
Pr[E(y_i,Z)]^2\Pr[Y=y_i],$$ since $Z$ and $Z'$ are i.i.d. Therefore, by Jensen’s inequality, we obtain $$\label{eqn:Jensen-1}
\Pr[E(Y,Z)]^2\leq \sum_{i=1}^n\Pr[E(y_i,Z)]^2\Pr[Y=y_i]=\Pr[E(Y,Z)\cap E(Y,Z')].$$ We also have $$\Pr[E(Y,Z)\cap E(Y,Z')]=\sum_{i=1}^m\sum_{j=1}^m\Pr[E(Y,z_i)\cap E(Y,z_j)]\Pr[Z=z_i]\Pr[Z=z_j],$$ and $$\Pr[E(Y,Z)\cap E(Y,Z')\cap E(Y',Z)\cap E(Y',Z')]=
\sum_{i=1}^n\sum_{j=1}^n \Pr[E(Y,z_i)\cap E(Y,z_j)]^2\Pr[Z=z_i]\Pr[Z=z_j].$$ Once again, by Jensen’s inequality, we obtain $$\label{eqn:Jensen-2}
\Pr[E(Y,Z)\cap E(Y,Z')]^2\leq \Pr[E(Y,Z)\cap E(Y,Z')\cap E(Y',Z)\cap E(Y',Z')].$$ By combining \[eqn:Jensen-1\] and \[eqn:Jensen-2\], we obtain the desired conclusion.
Next, we explain how to use the above decoupling lemma for our purpose. For this discussion, recall \[eqn:laplace-expansion\]. Fix a non-trivial partition $[n] = U_1 \sqcup U_2$. Let $Y:= (x_i)_{i \in U_1} $ and $Z:= (x_i)_{i \in U_2}$. Let $E_{\alpha, \textbf{c}} := E_{\alpha, \textbf{c}}(Y,Z)$ denote the event that $$Q_{\alpha, \textbf{c}}(Y,Z):=\alpha - \sum_{2 | Pr[E(y_i, Z)]^2\Pr[Y = y_i],$$ since $ Z$ and $ Z'$ are i.i.d. Therefore, by Jensen ’s inequality, we obtain $ $ \label{eqn: Jensen-1 }
\Pr[E(Y, Z)]^2\leq \sum_{i=1}^n\Pr[E(y_i, Z)]^2\Pr[Y = y_i]=\Pr[E(Y, Z)\cap E(Y, Z')].$$ We also have $ $ \Pr[E(Y, Z)\cap E(Y, Z')]=\sum_{i=1}^m\sum_{j=1}^m\Pr[E(Y, z_i)\cap E(Y, z_j)]\Pr[Z = z_i]\Pr[Z = z_j],$$ and $ $ \Pr[E(Y, Z)\cap E(Y, Z')\cap E(Y',Z)\cap E(Y',Z')]=
\sum_{i=1}^n\sum_{j=1}^n \Pr[E(Y, z_i)\cap E(Y, z_j)]^2\Pr[Z = z_i]\Pr[Z = z_j].$$ Once again, by Jensen ’s inequality, we receive $ $ \label{eqn: Jensen-2 }
\Pr[E(Y, Z)\cap E(Y, Z')]^2\leq \Pr[E(Y, Z)\cap E(Y, Z')\cap E(Y',Z)\cap E(Y',Z')].$$ By compound \[eqn: Jensen-1\ ] and \[eqn: Jensen-2\ ], we obtain the desired conclusion.
Next, we excuse how to use the above decoupling lemma for our purpose. For this discussion, recall \[eqn: laplace - expansion\ ]. pay back a non - trivial partition $ [ n ] = U_1 \sqcup U_2$. permit $ Y:= (x_i)_{i \in U_1 } $ and $ Z:= (x_i)_{i \in U_2}$. Let $ E_{\alpha, \textbf{c } }: = E_{\alpha, \textbf{c}}(Y, Z)$ denote the consequence that $ $ Q_{\alpha, \textbf{c}}(Y, Z):=\alpha - \sum_{2 | Pr[E(j_i,Z)]^2\Pr[Y=y_i],$$ since $Z$ and $Z'$ ave i.i.d. Therefore, by Jenven’s ihequalith, we obtain $$\label{eqn:Jensen-1}
\Pr[X(Y,Z)]^2\lwq \sun_{i=1}^n\Pr[E(y_i,Z)]^2\Pr[Y=y_i]=\Pr[E(Y,Z)\cap E(Y,Z')].$$ We apso have $$\Pr[E(B,Z)\cap E(Y,Z')]=\sum_{i=1}^m\sum_{j=1}^m\Pr[E(Y,z_l)\eap E(G,d_j)]\Pr[Z=v_i]\'r[Z=z_j],$$ and $$\Pr[E(Y,Z)\gap E(Y,Z')\cap A(Y',Z)\cap E(Y',Z')]=
\sum_{i=1}^t\sjm_{l=1}^n \Pr[E(Y,z_i)\cap E(Y,z_j)]^2\Pr[Z=z_i]\Pr[Z=z_j].$$ Once agayn, by Jrnden’s inequaliti, we pftaih $$\label{eqn:Jensen-2}
\Pr[E(Y,Z)\cap E(Y,Z')]^2\leq \Pr[S(Y,Z)\cap T(Y,Z')\cap E(Y',Z)\cap E(Y',Z')].$$ Ny combining \[eqn:Jensen-1\] and \[eqn:Uensen-2\], we obtain tje desired xoncjysion.
Next, we explain how to use thg above decoupling lemma for our purppse. For thus didwussion, recell \[eqg:laplace-expakxion\]. Fhx a nom-trivial partibion $[i] = U_1 \sqcup U_2$. Let $Y:= (x_i)_{i \in U_1} $ and $Z:= (x_i)_{i \in U_2}$. Let $E_{\alpva, \textbf{c}} := E_{\alpha, \reztbf{c}}(I,Z)$ detote rhe evtnt tgat $$Q_{\apphe, \textbf{c}}(Y,Z):=\zlpha - \sum_{2 | Pr[E(y_i,Z)]^2\Pr[Y=y_i],$$ since $Z$ and $Z'$ are i.i.d. Jensen’s we obtain \Pr[E(Y,Z)]^2\leq \sum_{i=1}^n\Pr[E(y_i,Z)]^2\Pr[Y=y_i]=\Pr[E(Y,Z)\cap E(Y,Z')].$$ E(Y,z_j)]\Pr[Z=z_i]\Pr[Z=z_j],$$ $$\Pr[E(Y,Z)\cap E(Y,Z')\cap E(Y',Z)\cap \sum_{i=1}^n\sum_{j=1}^n \Pr[E(Y,z_i)\cap E(Y,z_j)]^2\Pr[Z=z_i]\Pr[Z=z_j].$$ again, by Jensen’s inequality, we obtain \Pr[E(Y,Z)\cap E(Y,Z')]^2\leq \Pr[E(Y,Z)\cap E(Y,Z')\cap E(Y',Z)\cap E(Y',Z')].$$ By combining \[eqn:Jensen-1\] and \[eqn:Jensen-2\], we obtain desired conclusion. Next, we explain how to use the above decoupling lemma for purpose. this recall Fix a non-trivial partition $[n] = U_1 \sqcup U_2$. Let $Y:= (x_i)_{i \in U_1} $ and (x_i)_{i \in U_2}$. Let $E_{\alpha, \textbf{c}} := E_{\alpha, denote the event that \textbf{c}}(Y,Z):=\alpha - \sum_{2 | Pr[E(y_i,Z)]^2\Pr[Y=y_i],$$ since $Z$ and $Z'$ are i.I.d. ThereforE, by JeNseN’s iNeQualIty, wE obtain $$\label{eqN:jensEn-1}
\Pr[E(Y,Z)]^2\leq \sum_{i=1}^n\Pr[E(y_i,Z)]^2\PR[Y=y_i]=\PR[E(y,z)\cap e(y,Z')].$$ we alsO have $$\Pr[e(y,Z)\CAP E(Y,z')]=\sUm_{I=1}^m\sUm_{J=1}^M\PR[E(Y,z_i)\Cap e(Y,z_j)]\Pr[Z=Z_i]\Pr[Z=z_j],$$ and $$\pr[E(y,Z)\Cap E(Y,Z')\cap E(Y',Z)\CAp e(Y',Z')]=
\sum_{i=1}^n\suM_{j=1}^n \pr[E(Y,z_i)\cap E(Y,z_J)]^2\Pr[z=z_i]\Pr[Z=Z_j].$$ oncE Again, By JEnsen’S inequALity, we Obtain $$\labEl{EQn:JensEN-2}
\Pr[E(Y,Z)\cAP e(Y,z')]^2\leq \pr[E(Y,Z)\cap E(Y,Z')\cap E(Y',Z)\CAp e(y',Z')].$$ By combining \[eQn:JensEn-1\] ANd \[EQN:JeNseN-2\], we obtain tHe DesirED concluSIoN.
nEXt, wE Explain how to uSe the above dECouPling lEmMa fOR our puRpose. foR ThiS discussion, RecaLl \[eqn:laplAce-expANsion\]. FiX A non-triVial paRtiTioN $[n] = U_1 \sQCuP U_2$. let $y:= (x_I)_{I \in u_1} $ AnD $Z:= (x_I)_{I \in u_2}$. Let $E_{\alpHa, \TeXtbf{c}} := e_{\alpHA, \TEXtbf{C}}(Y,Z)$ DenoTe the Event that $$Q_{\alpHa, \tExtbF{C}}(Y,Z):=\Alpha - \Sum_{2 | Pr[E(y_i,Z)]^2\Pr[Y=y_i],$ $ since $Z $ and $Z '$ar e i. i.d. Therefore, by Jens en’s inequality, we ob tain$$ \ labe l {e qn:Je nsen-1} \P r [ E(Y ,Z )] ^2\ le q \ sum_{ i=1 }^n\Pr[ E(y_i,Z)]^ 2\P r[ Y=y_i]=\Pr[E ( Y, Z)\cap E(Y ,Z' )].$$ We als o h ave $$ \P r[E ( Y,Z)\ cap E(Y, Z')]=\ s um_{i= 1}^m\sum_ {j = 1}^m\P r [E(Y,z_ i ) \c ap E (Y,z_j)]\Pr[Z=z_i ] \P r [Z=z_j],$$ and $$\Pr [E ( Y, Z ) \ca p E (Y,Z')\cap E (Y',Z ) \cap E( Y ', Z ' ) ]=\ sum_{i=1}^n\s um_{j=1}^n\ Pr[ E(Y,z_ i) \ca p E(Y,z _j)]^ 2\ P r[Z =z_i]\Pr[Z= z_j] .$$ Onceagain, by Jens e n’s ine qualit y,weobta i n$$ \la be l {eq n :J ens e n-2 }
\Pr[E( Y, Z) \capE(Y, Z ' ) ] ^2\l eq\Pr[ E(Y,Z )\cap E(Y,Z') \ca p E( Y ',Z )\cap E(Y' ,Z') ]. $$ By combi ning\[ eqn:Jensen-1\]and\[eqn:Jen sen -2 \], w e obt a in the de sir ed conc lusion.
Ne xt , w eexplain how to use t h e a bove dec ouplin g l em m a for ou rpur pose . For t hisd is cussion, recal l \ [e qn:lapl ac e-expa ns ion \]. Fixa non -trivi al parti tion$ [n] = U_1 \sqc u p U_2$. Let $ Y := ( x_ i )_{i \i n U_1} $ an d $Z : = (x _i)_ { i\in U_2}$ . Let $ E _{ \ alpha, \textbf{c}}:= E_{\a lpha, \textbf{c}}( Y,Z)$ deno t e the even t th a t$ $Q_{\alpha, \t extbf {c}}(Y,Z): = \alpha - \sum _{2 | Pr[E(y_i,Z)]^2\Pr[Y=y_i],$$ since_$Z$ and_$Z'$ are i.i.d. Therefore,_by Jensen’s_inequality,_we obtain_$$\label{eqn:Jensen-1}
\Pr[E(Y,Z)]^2\leq_\sum_{i=1}^n\Pr[E(y_i,Z)]^2\Pr[Y=y_i]=\Pr[E(Y,Z)\cap E(Y,Z')].$$ We_also have $$\Pr[E(Y,Z)\cap_E(Y,Z')]=\sum_{i=1}^m\sum_{j=1}^m\Pr[E(Y,z_i)\cap E(Y,z_j)]\Pr[Z=z_i]\Pr[Z=z_j],$$ and $$\Pr[E(Y,Z)\cap_E(Y,Z')\cap E(Y',Z)\cap E(Y',Z')]=
\sum_{i=1}^n\sum_{j=1}^n_\Pr[E(Y,z_i)\cap_E(Y,z_j)]^2\Pr[Z=z_i]\Pr[Z=z_j].$$ Once again, by Jensen’s inequality, we obtain $$\label{eqn:Jensen-2}
\Pr[E(Y,Z)\cap E(Y,Z')]^2\leq \Pr[E(Y,Z)\cap E(Y,Z')\cap E(Y',Z)\cap E(Y',Z')].$$_By_combining \[eqn:Jensen-1\]_and_\[eqn:Jensen-2\],_we obtain the desired conclusion.
Next,_we explain how to use_the above_decoupling lemma for our purpose. For this discussion,_recall_\[eqn:laplace-expansion\]. Fix a_non-trivial partition $[n] = U_1 \sqcup U_2$. Let $Y:=_(x_i)_{i \in U_1} $ and $Z:=_(x_i)_{i \in U_2}$._Let_$E_{\alpha,_\textbf{c}} := E_{\alpha, \textbf{c}}(Y,Z)$_denote the event that $$Q_{\alpha, \textbf{c}}(Y,Z):=\alpha_- \sum_{2 |
As already mentioned, SN1992A is one of the best ever observed Ia SNe and therefore enters the zero point determination of the Ia Hubble diagram with a high weight. The GCS of NGC1380 has been analysed by Kissler-Patig et al. [@kissler:97] and Della Valle et al. [@della:98]. Interestingly, it turned out that the GCS could be separated into an elongated metal-rich bulge and a spherical, metal-poor halo population.
Due to the excellent data quality (very long exposed NTT data in very good seeing conditions) it was possible to trace the luminosity function 1mag below the TOM, and thus the GCS of NGC1380 is the deepest observed GCS from ground-based data. Table \[tab:92A\] lists the apparent brightnesses and the widths of the $t_5$-functions for $B$, $V$, and $I$. The metallicity corrections are negligible in this case.
Filter TOM $\sigma_t$
-------- ---------------- ---------------
$B$ $24.38\pm0.09$ $0.89\pm0.10$
$V$ $23.69\pm0.11$ $0.95\pm0.10$
$R$ $23.17\pm0.10$ $0.98\pm0.10$
: Listed are the apparent TOMs and $\sigma_t$-values of the $t_5$-functions representing the GCS of NGC1380 in the bands $V$, $B$, and $R$ according to Della Valle et al. [@della:98][]{data-label="tab:92A"}
### NGC1316, host galaxy to SN1980N and SN1981D
The fact that NGC1316 hosted two Ia’s (photometry can be found in Hamuy et al. [@hamuy:91]), gives it a particular character. However, it is not the perfect standard it could be. First, SN1981D is not very well observed, in particular the $V$-peak maximum has a large error, making the colour and the corresponding corrections uncertain. Secondly, NGC1316 is not a well behaved elliptical galaxy, but is classified as a peculiar S0 galaxy, suspected to have been formed in its present shape | As already mentioned, SN1992A is one of the best ever observed Ia SNe and consequently insert the zero point decision of the Ia Hubble diagram with a gamey weight. The GCS of NGC1380 has been analysed by Kissler - Patig et al. [ @kissler:97 ] and Della Valle et al. [ @della:98 ]. Interestingly, it flex out that the GCS could be separated into an elongated metallic element - rich bulge and a spherical, alloy - poor halo population.
Due to the excellent datum quality (very long exposed NTT data in very good seeing conditions) it was possible to hound the luminosity function 1mag below the TOM, and thus the GCS of NGC1380 is the deepest observe GCS from ground - based data. Table \[tab:92A\ ] list the apparent brightnesses and the widths of the $ t_5$-functions for $ B$, $ V$, and $ I$. The metallicity corrections are negligible in this case.
Filter TOM $ \sigma_t$
-------- ---------------- ---------------
$ B$ $ 24.38\pm0.09 $ $ 0.89\pm0.10 $
$ V$ $ 23.69\pm0.11 $ $ 0.95\pm0.10 $
$ R$ $ 23.17\pm0.10 $ $ 0.98\pm0.10 $
: Listed are the apparent TOMs and $ \sigma_t$-values of the $ t_5$-functions map the GCS of NGC1380 in the bands $ V$, $ B$, and $ R$ according to Della Valle et al. [ @della:98][]{data - label="tab:92A " }
# # # NGC1316, host galaxy to SN1980N and SN1981D
The fact that NGC1316 hosted two Ia ’s (photometry can be found in Hamuy et al. [ @hamuy:91 ]), gives it a particular character. However, it is not the perfect standard it could be. First, SN1981D is not very well observed, in particular the $ V$-peak maximum has a bombastic error, making the color and the correspond corrections uncertain. Secondly, NGC1316 is not a well behave elliptical galaxy, but is classified as a peculiar S0 galax, suspected to have been formed in its present shape |
As wlready mentioned, SN1992A is one of the besj wver ouserved Ia SNe xnd therefore enters the zerl point determination of the Ka Hubble diagram witi a high weight. Vge GCS of NGC1380 mas bzei analysed by Klssler-Patig et al. [@kissler:97] dna Bella Valle et al. [@della:98]. Interestingly, yt turnrd out that the DCS botld gv weparated into an elongated mstal-rici bulge and a slherical, metal-poor halo pooulahion.
Due to the excfllent data quajuty (very lone exposed NTT data in bery good seeing conditions) it das ppssible to teacf the luminowity sunction 1mag below tve TOM, snd thus the GGS of NGX1380 is the deepest obsecved GCS from ground-fased datd. Cable \[tab:92A\] lists the appqrent brichtndwser ahd tge widhhs of the $t_5$-fhnctions foe $B$, $V$, and $I$. The metsljpvity correctjons awe negligible in this case.
Filter TKM $\sigma_t$
-------- ---------------- ---------------
$B$ $24.38\pm0.09$ $0.89\pm0.10$
$V$ $23.69\pm0.11$ $0.95\pm0.10$
$R$ $23.17\pm0.10$ $0.98\pm0.10$
: Listed are the apparent TOKs anv $\rigna_b$-valjws of the $t_5$-functions representing the GCS of NGS1380 im nhe bands $V$, $B$, and $R$ accordinb ho Qella Valle et xl. [@della:98][]{dzta-label="tab:92A"}
### NGC1316, hlst galwxy ti SN1980N and SN1981D
Yhe fact that NGC1316 hosted twi Ia’s (photomvtry can be found in Hcmuy et al. [@haouy:91]), bives it a particular characcer. Hosever, it is not the ldrfect standard kt boulg be. First, SN1981D is not very well obsxrved, in pargicukar thq $V$-peak madimum has a large error, lakiny the colour anf the corresponding corrections uncertain. Secpngly, NGC1316 is uot a eell behaved qlliptical galcxy, but ns clarsified as a pecuniar S0 galavy, suspected dl have been hormed in its prewent shxoe | As already mentioned, SN1992A is one of ever Ia SNe therefore enters the Ia diagram with a weight. The GCS NGC1380 has been analysed by Kissler-Patig al. [@kissler:97] and Della Valle et al. [@della:98]. Interestingly, it turned out that GCS could be separated into an elongated metal-rich bulge and a spherical, metal-poor population. to excellent quality (very long exposed NTT data in very good seeing conditions) it was possible to trace luminosity function 1mag below the TOM, and thus GCS of NGC1380 is deepest observed GCS from ground-based Table lists the brightnesses the of the $t_5$-functions $B$, $V$, and $I$. The metallicity corrections are negligible in this case. Filter TOM $\sigma_t$ -------- ---------------- $B$ $24.38\pm0.09$ $23.69\pm0.11$ $0.95\pm0.10$ $23.17\pm0.10$ : are the apparent $\sigma_t$-values of the $t_5$-functions representing the in the bands $V$, $B$, and $R$ according Della Valle al. [@della:98][]{data-label="tab:92A"} ### NGC1316, host galaxy SN1980N and SN1981D The fact that NGC1316 hosted Ia’s (photometry can be found in Hamuy et al. [@hamuy:91]), gives it a particular character. is not the perfect it could be. SN1981D not well in particular $V$-peak maximum has a large error, making the colour and the corrections uncertain. Secondly, NGC1316 is not a well behaved elliptical is as a peculiar galaxy, suspected to have formed its present shape |
As already mentioned, SN1992A is onE of the best Ever oBseRveD IA SNe And tHerefore enters THe zeRo point determination of The Ia huBBle dIAgRam wiTh a high WEiGHT. ThE GcS Of NgC1380 HAs Been aNalYsed by KIssler-PatiG et Al. [@Kissler:97] and DeLLa valle et al. [@dEllA:98]. InterestingLy, iT turneD oUt tHAt the gCS Could Be sepaRAted inTo an elongAtED metal-RIch bulgE ANd A sphErical, metal-poor haLO pOPulation.
Due to tHe exceLlENt DATa qUalIty (very lonG eXposeD nTT data IN vERY GooD Seeing conditiOns) it was posSIblE to traCe The LUminosIty fuNcTIon 1Mag below the tOM, aNd thus the gCS of Ngc1380 is the dEEpest obServed gCS FroM groUNd-BaSed DaTA. TaBLe \[Tab:92a\] LisTs the appArEnT brigHtneSSES And tHe wIdthS of thE $t_5$-functions foR $B$, $V$, And $I$. tHe mEtallIcity CorrEcTions Are negLigibLe In this case.
FilteR TOM $\Sigma_t$
-------- ---------------- ---------------
$B$ $24.38\pm0.09$ $0.89\Pm0.10$
$V$ $23.69\Pm0.11$ $0.95\Pm0.10$
$R$ $23.17\Pm0.10$ $0.98\Pm0.10$
: LisTEd are tHe aPpaRent TOMS and $\sigMA_t$-vAlUES Of The $t_5$-functions repreSeNTInG the GCS oF NGC1380 in THe BaNDs $V$, $B$, and $R$ AcCorDing TO della vallE Et Al. [@della:98][]{dAta-labEL="tAb:92a"}
### NGC1316, hosT gAlaxy tO Sn1980N aNd Sn1981D
The FAct tHat NGC1316 Hosted twO Ia’s (pHOtometry can be fOUnd in Hamuy et aL. [@HaMUY:91]), gIVes iT a pArticular chAracTEr. HoWeveR, It Is nOT the pErfecT sTAnDArd it could be. First, SN1981d iS not veRy welL observed, in paRticular thE $v$-PEak maximUm haS A lARge error, making The coLour and the COrresponDing cOrrectioNs uncertaIN. secondly, nGC1316 Is nOt a WelL BEhAved ellipticaL GAlaxY, bUt is claSsiFied as a PecUliAr S0 GalAxY, suspecteD to have bEeN fOrMeD in Its prESent shapE |
As already mentioned, SN1 992A is on e ofthe be st eve r ob served Ia SNea nd t herefore enters the ze ro po in t det e rm inati on of t h eI a Hu bb le di ag r am with ahigh we ight. TheGCS o f NGC1380 ha s b een analys edby Kissler-P ati g et a l. [@ k issle r:9 7] an d Dell a Valle et al. [ @d e lla:98 ] . Inter e s ti ngly , it turned out t h at the GCS couldbe sep ar a te d int o a n elongate dmetal - rich bu l ge a n d a spherical, me tal-poor ha l o p opulat io n.Due to theex c ell ent data qu alit y (very l ong ex p osed NT T data i n very go odseei n gco ndi ti o ns) it wa s po ssible t otr ace t he l u m i n osit y f unct ion 1 mag below the TO M, a n d t hus t he GC S of N GC138 0 is t he de ep est observed GC S fr om ground -ba se d d at a. Ta b le \[t ab: 92A \] list s the a p par en t b ri ghtnesses and thewi d t hs of the$t_5$- f un ct i ons for$B $,$V$, a nd $I $. T h emetallic ity co r re ct ions ar eneglig ib leinthisc ase.
F ilter T OM $\ s igma_t$
--- - -- - - - - ---- --- ------- --- ---- - ---- --- $B $ $24.3 8\ p m0 . 09$ $0.89\pm0.10$
$V$ $23.69\pm0.11 $ $0.95\ p m 0 .10$
$R$ $23.17\pm0.10 $ $ 0.98\pm0.1 0 $
: L isted are the apparent T OMs and$\s igm a_t $-v a l ue s of the $t_5 $ - func ti ons rep res entingthe GC S o f N GC 1380 in t he bands $ V$ ,$B $,and $ R $ accord in g t oDel la Va l le etal. [ @del la :9 8 ][] {data-l a be l = "tab :9 2A "}
### N GC131 6, h o stgalaxyto SN1980 N a n d SN 19 81 D
Thefact that NGC 13 16 hostedtw o I a’s (p h o tometrycan be found in Hamuy e t al. [@ ham uy:91 ]),gives ita p articu lar charac ter. H oweve r, it i s not t he pe rf ect standa r d it coul dbe.First,SN1981D is not ver y we ll observed,inpart i c ul art he $V$ -p e akm a ximum has a lar ge error,ma k in g the colo u r a nd the co rrespon dingc orrecti ons uncer tain. Sec on dly, N GC1 316 is not a wellbehaved e l lipti c al gala xy, but i scla ssifi ed asa pe culia r S0 g al axy, s uspec te d to hav e been formed in its pr esentshape |
As already_mentioned, SN1992A_is one of the_best ever_observed_Ia SNe_and_therefore enters the_zero point determination_of the Ia Hubble_diagram with a_high_weight. The GCS of NGC1380 has been analysed by Kissler-Patig et al. [@kissler:97] and Della Valle_et al._[@della:98]. Interestingly,_it_turned_out that the GCS could_be separated into an elongated_metal-rich bulge_and a spherical, metal-poor halo population.
Due to the_excellent_data quality (very_long exposed NTT data in very good seeing conditions)_it was possible to trace the_luminosity function 1mag_below_the_TOM, and thus the_GCS of NGC1380 is the deepest_observed GCS from ground-based data. Table \[tab:92A\]_lists the apparent brightnesses and the widths_of the $t_5$-functions for $B$, $V$,_and $I$. The metallicity corrections_are negligible_in this case.
_Filter _ _ _TOM _ _ $\sigma_t$
-------- ---------------- ---------------
__ $B$___ _$24.38\pm0.09$ _$0.89\pm0.10$
_ __$V$ $23.69\pm0.11$__ $0.95\pm0.10$
$R$ _ $23.17\pm0.10$__ $0.98\pm0.10$
:_Listed are the apparent TOMs_and $\sigma_t$-values of the $t_5$-functions representing_the GCS_of NGC1380_in the bands $V$, $B$, and $R$ according to Della Valle et al._[@della:98][]{data-label="tab:92A"}
### NGC1316, host galaxy to SN1980N_and SN1981D
The fact that_NGC1316 hosted_two_Ia’s (photometry can_be_found in_Hamuy et al. [@hamuy:91]), gives it a particular_character. However,_it is not the perfect standard_it could be. First,_SN1981D_is not very well observed, in_particular the $V$-peak maximum has a_large error, making the colour_and_the_corresponding corrections uncertain. Secondly, NGC1316_is not a well behaved elliptical_galaxy, but is_classified as a peculiar S0 galaxy, suspected_to_have been formed in its present_shape |
V. Loo, L. Lanco, A. Lemaitre, I. Sagnes, O. Krebs, P. Voisin, and P. Senellart, Appl. Phys. Lett. [**97**]{}, 241110 (2010).
R. J. Barbour, P. A. Dalgarno, A. Curran, K. M. Nowak, H. J. Baker, D. R. Hall, N. G. Stoltz, P. M. Petroff, and R. J. Warburton, J. Appl. Phys. [**110**]{}, 053107 (2011).
We assume that a single layer of quantum dots is put inside a cavity whose linear dimension is $2.8$ $\mu$m as in Ref..
N. Perret, D. Morris, L. Franchomme-Fossé, R. Côté, S. Fafard, V. Aimez, and J. Beauvais, Phys. Rev. B [**62**]{}, 5092 (2000).
A. Faribault, P. Calabrese and J.-S. Caux, J. Math. Phys. [**50**]{}, 095212 (2009).
A. Faribault, O. El Araby, C. Sträter and V. Gritsev, Phys. Rev. B [**83**]{}, 235124 (2011).
O. El Araby, V. Gritsev and A. Faribault Phys. Rev. B [**85**]{}, 115130 (2012).
O. Tsyplyatyev, J. von Delft and D. Loss, Phys. Rev. B [**82**]{}, 092203 (2010).
N. A. Slavnov, Theoretical and Mathematical Physics [**79**]{} 502-508 (1989).
J. Links, H. Zhou, R. H. McKenzie and M. D. Gold, J. Phys. A: Math. [**36**]{} R63 (2003).
E. A. Yuzbashyan, V. B. Kuznetsov and B. L. Altshuler, Phys. Rev B [**72**]{}, 144524 (2005).
E. A. Yuzbashyan, O. Tsyplyatyev, and B. L. Altshuler, Phys. Rev. Let. [**96**]{}, 0 | V. Loo, L. Lanco, A. Lemaitre, I. Sagnes, O. Krebs, P. Voisin, and P. Senellart, Appl. Phys. Lett. [ * * 97 * * ] { }, 241110 (2010).
R. J. Barbour, P. A. Dalgarno, A. Curran, K. M. Nowak, H. J. Baker, D. R. Hall, N. G. Stoltz, P. M. Petroff, and R. J. Warburton, J. Appl. Phys. [ * * 110 * * ] { }, 053107 (2011).
We assume that a single layer of quantum dots is place inside a pit whose linear dimension is $ 2.8 $ $ \mu$m as in Ref..
N. Perret, D. Morris, L. Franchomme - Fossé, R. Côté, S. Fafard, V. Aimez, and J. Beauvais, Phys. Rev. B [ * * 62 * * ] { }, 5092 (2000).
A. Faribault, P. Calabrese and J.-S. Caux, J. Math. Phys. [ * * 50 * * ] { }, 095212 (2009).
A. Faribault, O. El Araby, C. Sträter and V. Gritsev, Phys. Rev. B [ * * 83 * * ] { }, 235124 (2011).
O. El Araby, V. Gritsev and A. Faribault Phys. Rev. B [ * * 85 * * ] { }, 115130 (2012).
O. Tsyplyatyev, J. von Delft and D. Loss, Phys. Rev. B [ * * 82 * * ] { }, 092203 (2010).
N. A. Slavnov, Theoretical and Mathematical Physics [ * * 79 * * ] { } 502 - 508 (1989).
J. Links, H. Zhou, R. H. McKenzie and M. D. Gold, J. Phys. deoxyadenosine monophosphate: Math. [ * * 36 * * ] { } R63 (2003).
E. A. Yuzbashyan, V. B. Kuznetsov and B. L. Altshuler, Phys. Rev B [ * * 72 * * ] { }, 144524 (2005).
E. A. Yuzbashyan, O. Tsyplyatyev, and B. L. Altshuler, Phys. Rev. get. [ * * 96 * * ] { }, 0 |
V. Llo, L. Lanco, A. Lemaitre, I. Ragnes, O. Krebs, K. Coisin, and P. Senellaft, Appl. Phys. Lett. [**97**]{}, 241110 (2010).
R. J. Barblue, P. A. Dalgarno, A. Curran, K. M. Nowak, H. U. Baker, E. R. Iall, N. G. Stoltz, '. M. Petroff, ans R. J. Xarburton, J. Appk. Phys. [**110**]{}, 053107 (2011).
Wa assume that d riugle layer of quantum dots is put infide a vagity whose lingar dpmqnsikn is $2.8$ $\mu$m as in Ref..
N. Perret, D. Mordis, L. Fganchomme-Fossé, R. Côyé, S. Fafard, V. Aimez, and J. Bfauvwis, Phys. Rev. B [**62**]{}, 5092 (2000).
A. Faribault, K. Cajqbrese and J.-R. Caux, J. Math. Phys. [**50**]{}, 095212 (2009).
A. Faribault, O. El Araby, C. Sträter xnd V. Gritsev, Pyyw. Rfe. B [**83**]{}, 235124 (2011).
O. El Eraby, N. Gritsev and A. Faribdult Phus. Rev. B [**85**]{}, 115130 (2012).
O. Txypnyaryev, J. von Delft and V. Loss, Phys. Rev. B [**82**]{}, 092203 (2010).
G. A. Slavnmv, Theoretical and Narhemajical Phyrucs [**79**]{} 502-508 (1989).
N. Kihks, H. Xhon, R. H. McKenaie and M. D. Gold, J. Phys. A: Math. [**36**]{} W63 (2003).
E. A. Yuzbashyzn, V. B. Ktznetsov and B. L. Altshuler, Phys. Rev B [**72**]{}, 144524 (2005).
E. Z. Yuzbashyan, O. Tsyplyattev, and B. L. Altshuler, Phys. Rev. Let. [**96**]{}, 0 | V. Loo, L. Lanco, A. Lemaitre, I. Krebs, Voisin, and Senellart, Appl. Phys. J. P. A. Dalgarno, Curran, K. M. H. J. Baker, D. R. Hall, G. Stoltz, P. M. Petroff, and R. J. Warburton, J. Appl. Phys. [**110**]{}, (2011). We assume that a single layer of quantum dots is put inside cavity linear is $\mu$m as in Ref.. N. Perret, D. Morris, L. Franchomme-Fossé, R. Côté, S. Fafard, V. Aimez, J. Beauvais, Phys. Rev. B [**62**]{}, 5092 (2000). Faribault, P. Calabrese and Caux, J. Math. Phys. [**50**]{}, (2009). Faribault, O. Araby, Sträter V. Gritsev, Phys. B [**83**]{}, 235124 (2011). O. El Araby, V. Gritsev and A. Faribault Phys. Rev. B [**85**]{}, 115130 O. Tsyplyatyev, Delft and Loss, Rev. [**82**]{}, 092203 (2010). Slavnov, Theoretical and Mathematical Physics [**79**]{} Links, H. Zhou, R. H. McKenzie and M. Gold, J. A: Math. [**36**]{} R63 (2003). E. Yuzbashyan, V. B. Kuznetsov and B. L. Altshuler, Rev B [**72**]{}, 144524 (2005). E. A. Yuzbashyan, O. Tsyplyatyev, and B. L. Altshuler, Phys. [**96**]{}, 0 |
V. Loo, L. Lanco, A. Lemaitre, I. SagneS, O. Krebs, P. VoIsin, aNd P. senElLart, appl. phys. Lett. [**97**]{}, 241110 (2010).
R. J. BarbOUr, P. A. dalgarno, A. Curran, K. M. Nowak, h. J. BakEr, d. r. HalL, n. G. stoltZ, P. M. PetrOFf, AND R. J. waRbUrtOn, j. apPl. PhyS. [**110**]{}, 053107 (2011).
We Assume tHat a single LayEr Of quantum dotS Is Put inside a CavIty whose lineAr dImensiOn Is $2.8$ $\mU$M as in ref..
n. PerrEt, D. MorRIs, L. FraNchomme-FoSsÉ, r. Côté, S. fAfard, V. AIMEz, And J. beauvais, Phys. Rev. B [**62**]{}, 5092 (2000).
A. fArIBault, P. CalabresE and J.-S. caUX, J. mATh. PHys. [**50**]{}, 095212 (2009).
a. Faribault, o. EL ArabY, c. SträteR AnD v. gRitSEv, Phys. Rev. B [**83**]{}, 235124 (2011).
O. El araby, V. GritsEV anD A. FariBaUlt pHys. Rev. b [**85**]{}, 115130 (2012).
O. TsyPlYAtyEv, J. von Delft And D. loss, Phys. REv. B [**82**]{}, 092203 (2010).
N. A. SLAvnov, ThEOreticaL and MaTheMatIcal pHySiCs [**79**]{} 502-508 (1989).
J. liNKs, H. zHoU, R. H. mCKeNzie and M. d. GOlD, J. PhyS. A: MaTH. [**36**]{} r63 (2003).
e. a. YuzBasHyan, v. B. KuzNetsov and B. L. AlTshUler, pHys. rev B [**72**]{}, 144524 (2005).
E. a. YuzbAshyAn, o. TsypLyatyeV, and B. l. ALtshuler, Phys. Rev. let. [**96**]{}, 0 |
V. Loo, L. Lanco, A. Lema itre, I. S agnes , O . K re bs,P. V oisin, and P.S enel lart, Appl. Phys. Lett . [** 97 * *]{} , 2 41110 (2010) .
R . J. B ar bou r, P. A. D alg arno, A . Curran,K.M. Nowak, H. J . B aker, D. R . H all, N. G. S tol tz, P. M . P e troff , a nd R. J. Wa r burton , J. Appl .P hys. [ * *110**] { } ,0531 07 (2011).
We as s um e that a single layer o f q u a ntu m d ots is put i nside a cavit y w h o s e l i near dimensio n is $2.8$$ \mu $m asin Re f ..
N. Perr et , D. Morris, L. Fra nchomme-F ossé,R . Côté, S. Fafa rd, V. Ai mez , an d J .Bea uv a is, Ph ys. Rev . B [**6 2* *] {}, 5 092( 2 0 0 0).
A. Far ibaul t, P. Calabre seandJ .-S . Cau x, J. Mat h. Phys . [**5 0**]{ }, 095212 (2009).
A. Faribaul t,O. El A raby, C. Str äte r a nd V. G ritsev, Phy s. R e v. B [**83**]{}, 235 12 4 (2 011).
O . El A r ab y, V. Grits ev an d A. F ariba ultP hy s. Rev.B [**8 5 ** ]{ }, 1151 30 (2012 ).
O . T syply a tyev , J. v on Delft andD . Loss, Phys.R ev. B [**82** ] {} , 09 2 203(20 10).
N. A. Sla v nov, The o re tic a l and Math em a ti c al Physics [**79**] {} 502-5 08 (1 989).
J. Lin ks, H. Zho u , R. H. Mc Kenz i ea nd M. D. Gold, J. P hys. A: Ma t h. [**36 **]{} R63 (20 03).
E.A . Yuzbash yan , V . B . K u z ne tsov and B. L . Alts hu ler, Ph ys. Rev B[** 72* *]{ },14 4524 (200 5).
E.A. Y uz ba shy an, O . Tsyplya ty ev, a ndB. L. Altshu ler,Phys .Re v . L et. [** 9 6* * ] {},0 |
V. Loo,_L. Lanco,_A. Lemaitre, I. Sagnes,_O. Krebs,_P._Voisin, and_P._Senellart, Appl. Phys._Lett. [**97**]{}, 241110_(2010).
R. J. Barbour, P._A. Dalgarno, A._Curran,_K. M. Nowak, H. J. Baker, D. R. Hall, N. G. Stoltz, P. M._Petroff,_and R._J._Warburton,_J. Appl. Phys. [**110**]{}, 053107_(2011).
We assume that a single_layer of_quantum dots is put inside a cavity whose_linear_dimension is $2.8$_$\mu$m as in Ref..
N. Perret, D. Morris, L. Franchomme-Fossé,_R. Côté, S. Fafard, V. Aimez,_and J. Beauvais,_Phys._Rev._B [**62**]{}, 5092 (2000).
A._Faribault, P. Calabrese and J.-S. Caux,_J. Math. Phys. [**50**]{}, 095212 (2009).
A._Faribault, O. El Araby, C. Sträter and_V. Gritsev, Phys. Rev. B [**83**]{},_235124 (2011).
O. El Araby, V._Gritsev and_A. Faribault Phys. Rev. B_[**85**]{}, 115130 (2012).
O._Tsyplyatyev, J._von Delft and_D. Loss, Phys. Rev. B [**82**]{},_092203 (2010).
N. A._Slavnov, Theoretical and Mathematical Physics [**79**]{}_502-508_(1989).
J. Links, H._Zhou,_R._H. McKenzie_and M. D._Gold,_J. Phys._A:_Math. [**36**]{} R63 (2003).
E. A. Yuzbashyan,_V._B. Kuznetsov and B. L. Altshuler, Phys._Rev B [**72**]{}, 144524_(2005).
E._A. Yuzbashyan, O. Tsyplyatyev,_and B. L. Altshuler, Phys._Rev. Let. [**96**]{}, 0 |
0 \to L(C_0) \to E \to F \to 0$$ such that $E$ is semi-stable (see the proof of Proposition \[prop:1\]). Thus $Z$ is not empty. Therefore, to compute the Picard group of $\overline{M}_{H_x}(r;c_1,c_2)$, it is enough to consider $Pic(M^0 \cup (Z^0 \cap M_0^{\mu}) )$, where $Z^0$ is the subscheme of $M_H(r;c_1,c_2)$ consisting of stable sheaves which are defined by the exact sequence.
We set $V_i=\cal O_X(-nH_x)^{\oplus N_i}$, $(i=1,2$). Let $Quot_{V_1/X/\Bbb C}^{\gamma_1}$ ( resp. $Quot_{V_2/X/\Bbb C}^{\gamma _2}$) be a quot-scheme parametrizing all quotients $V_1\to F_1$ (resp. $V_2 \to F_2$) such that $\gamma(F_1)=(r_1,C_0+\frac{d_1}{r_1}f,0)$ (resp. $\gamma(F_2)=(r_2,\frac{d_2}{r_2}f,0)$). Let $Q_i$ ($i=1,2$) be the open subscheme of $Quot_{V_i/X/\Bbb C}^{\gamma_i}$ consisting all quotients $V_i \to F_i$ which satisfy
1. $F_i$ is $\mu$-semi-stable with respect to $H_x$,
2. $F_{i|\pi^{-1}(\eta)}$ is a semi-stable vector bundle, where $\eta$ is the generic point of $C$,
3. $H^0(X,V_i(nH_x)) \cong H^0(X,F_i(nH_x))$, $H^j(X,F_i(nH_x))=0,j>0$.
Let $V_i \otimes \cal O_{Q_i \times X} \to \cal F_i$ be the universal quotient, | 0 \to L(C_0) \to E \to F \to 0$$ such that $ E$ is semi - stable (see the proof of Proposition \[prop:1\ ]). Thus $ Z$ is not empty. consequently, to calculate the Picard group of $ \overline{M}_{H_x}(r;c_1,c_2)$, it is adequate to consider $ Pic(M^0 \cup (Z^0 \cap M_0^{\mu }) ) $, where $ Z^0 $ is the subscheme of $ M_H(r;c_1,c_2)$ consisting of stable bundle which are defined by the exact sequence.
We arrange $ V_i=\cal O_X(-nH_x)^{\oplus N_i}$, $ (i=1,2 $). Let $ Quot_{V_1 / X/\Bbb C}^{\gamma_1}$ (resp. $ Quot_{V_2 / X/\Bbb C}^{\gamma _ 2}$) be a quot - scheme parametrizing all quotient $ V_1\to F_1 $ (resp. $ V_2 \to F_2 $) such that $ \gamma(F_1)=(r_1,C_0+\frac{d_1}{r_1}f,0)$ (resp. $ \gamma(F_2)=(r_2,\frac{d_2}{r_2}f,0)$). Let $ Q_i$ ($ i=1,2 $) be the open subscheme of $ Quot_{V_i / X/\Bbb C}^{\gamma_i}$ consist all quotients $ V_i \to F_i$ which satisfy
1. $ F_i$ is $ \mu$-semi - static with respect to $ H_x$,
2. $ F_{i|\pi^{-1}(\eta)}$ is a semi - stable vector bundle, where $ \eta$ is the generic point of $ C$,
3. $ H^0(X, V_i(nH_x) ) \cong H^0(X, F_i(nH_x))$, $ H^j(X, F_i(nH_x))=0,j>0$.
Let $ V_i \otimes \cal O_{Q_i \times X } \to \cal F_i$ be the universal quotient, |
0 \to L(C_0) \to E \to F \to 0$$ such tmat $E$ is semi-stayoe (see the pdoof of Oroposition \[prop:1\]). Thus $Z$ is nlt emptt. Therefore, to compute the Picagd group if $\oterline{M}_{H_x}(r;c_1,c_2)$, it is enounk to dlnsibec $Pic(M^0 \cup (Z^0 \cak M_0^{\mu}) )$, where $Z^0$ is the subswhdmz of $M_H(r;c_1,c_2)$ consisting of stable sheades whivh are defined bi the qxacf sequence.
We set $V_i=\cal O_X(-nH_x)^{\oplus H_i}$, $(i=1,2$). Leu $Quot_{V_1/X/\Bbb C}^{\gamma_1}$ ( resp. $Quot_{V_2/X/\Bbb C}^{\gamma _2}$) bf a euot-scheme parametgizing all wuotywnts $V_1\to F_1$ (rdsp. $V_2 \to F_2$) such that $\gzmma(F_1)=(r_1,C_0+\frac{d_1}{r_1}f,0)$ (resp. $\gamma(F_2)=(r_2,\frac{a_2}{r_2}f,0)$). Lzt $Q_i$ ($i=1,2$) be jkw oogn subscheme of $Qlot_{V_i/X/\Bbb C}^{\gamma_i}$ convisting all quotients $V_i \to F_i$ which satisfy
1. $F_i$ is $\mu$-semi-stable wyth respewt to $H_x$,
2. $F_{i|\pi^{-1}(\eta)}$ iw q semh-statle xwctur guidls, wherf $\eva$ is the gsneric poinr of $C$,
3. $H^0(X,V_i(nH_x)) \conb R^0(Q,G_i(nH_x))$, $H^j(X,F_i(nG_x))=0,j>0$.
Let $V_y \otimes \cal O_{Q_i \times X} \to \cal F_i$ be tve hniversal quotient, | 0 \to L(C_0) \to E \to F such $E$ is (see the proof is empty. Therefore, to the Picard group $\overline{M}_{H_x}(r;c_1,c_2)$, it is enough to consider \cup (Z^0 \cap M_0^{\mu}) )$, where $Z^0$ is the subscheme of $M_H(r;c_1,c_2)$ consisting stable sheaves which are defined by the exact sequence. We set $V_i=\cal O_X(-nH_x)^{\oplus $(i=1,2$). $Quot_{V_1/X/\Bbb ( $Quot_{V_2/X/\Bbb C}^{\gamma _2}$) be a quot-scheme parametrizing all quotients $V_1\to F_1$ (resp. $V_2 \to F_2$) such $\gamma(F_1)=(r_1,C_0+\frac{d_1}{r_1}f,0)$ (resp. $\gamma(F_2)=(r_2,\frac{d_2}{r_2}f,0)$). Let $Q_i$ ($i=1,2$) be the subscheme of $Quot_{V_i/X/\Bbb C}^{\gamma_i}$ all quotients $V_i \to F_i$ satisfy $F_i$ is with to 2. $F_{i|\pi^{-1}(\eta)}$ is semi-stable vector bundle, where $\eta$ is the generic point of $C$, 3. $H^0(X,V_i(nH_x)) \cong H^0(X,F_i(nH_x))$, $H^j(X,F_i(nH_x))=0,j>0$. Let \otimes \cal X} \to F_i$ the quotient, |
0 \to L(C_0) \to E \to F \to 0$$ such that $E$ is seMi-stable (seE the pRooF of prOposItioN \[prop:1\]). Thus $Z$ is noT EmptY. Therefore, to compute the picarD gROup oF $\OvErlinE{M}_{H_x}(r;c_1,c_2)$, IT iS ENouGh To ConSiDEr $pic(M^0 \cUp (Z^0 \Cap M_0^{\mu}) )$, wHere $Z^0$ is the SubScHeme of $M_H(r;c_1,c_2)$ cONsIsting of stAblE sheaves whicH arE definEd By tHE exacT seQuencE.
We set $v_I=\cal O_X(-NH_x)^{\oplus N_I}$, $(i=1,2$). lEt $Quot_{v_1/x/\Bbb C}^{\gaMMA_1}$ ( rEsp. $QUot_{V_2/X/\Bbb C}^{\gamma _2}$) be a QUoT-Scheme parametrIzing aLl QUoTIEntS $V_1\tO F_1$ (resp. $V_2 \to F_2$) SuCh thaT $\Gamma(F_1)=(r_1,c_0+\FrAC{D_1}{R_1}f,0)$ (rESp. $\gamma(F_2)=(r_2,\frac{D_2}{r_2}f,0)$). Let $Q_i$ ($i=1,2$) be THe oPen subScHemE Of $Quot_{v_i/X/\BbB C}^{\GAmmA_i}$ consistinG all Quotients $v_i \to F_i$ WHich satISfy
1. $F_i$ is $\Mu$-semi-StaBle With REsPeCt tO $H_X$,
2. $f_{i|\pI^{-1}(\EtA)}$ is A SemI-stable vEcToR bundLe, whERE $\ETa$ is The GeneRic poInt of $C$,
3. $H^0(X,V_i(nH_x)) \ConG H^0(X,F_I(NH_x))$, $h^j(X,F_i(NH_x))=0,j>0$.
LEt $V_i \OtImes \cAl O_{Q_i \tImes X} \To \Cal F_i$ be the univeRsal Quotient, |
0 \to L(C_0) \to E \to F\to 0$$ su ch th at$E$ i s se mi-s table (see the proo f of Proposition \[pro p:1\] ). Thus $Z $ isnot emp t y. T her ef or e,to co mpute th e Picar d group of $\ ov erline{M}_{H _ x} (r;c_1,c_2 )$, it is enoug h t o cons id er$ Pic(M ^0\cup(Z^0 \ c ap M_0 ^{\mu}) ) $, where$ Z^0$ is t he sub scheme of $M_H(r; c _1 , c_2)$ consisti ng ofst a bl e she ave s which ar edefin e d by th e e x a c t s e quence.
We s et $V_i=\ca l O_ X(-nH_ x) ^{\ o plus N _i}$, $ ( i=1 ,2$). Let $ Quot _{V_1/X/\ Bbb C} ^ {\gamma _ 1}$ ( r esp. $ Quo t_{ V_2/ X /\ Bb b C }^ { \ga m ma _2 } $)be a quo t- sc hemepara m e t r izin g a ll q uotie nts $V_1\to F _1$ (re s p.$V_2\to F _2$) s uch t hat $\ gamma (F _1)=(r_1,C_0+\f rac{ d_1}{r_1} f,0 )$ (r es p. $\ g amma(F _2) =(r _2,\fra c{d_2}{ r _2} f, 0 ) $ ). Let $Q_i$ ($i=1,2 $) b ethe open subsc h em eo f $Quot_ {V _i/ X/\B b b C}^{ \gam m a_ i}$ cons isting al lquotien ts $V_i\t o F _i$ whic h sat isfy
1. $F_i $ is$ \mu$-semi-stab l e with respec t t o $H _ x$,
2. $F_{i|\pi ^{-1 } (\et a)}$ is as emi-s table v e ct o r bundle, where $\e ta $ is t he ge neric point o f $C$,
3. $ H^0(X,V_ i(nH _ x) ) \cong H^0(X,F _i(nH _x))$, $H^ j (X,F_i(n H_x)) =0,j>0$.
Let $V_ i \otimes\ca l O _{Q _i\ t im es X} \to \ca l F_i$ b e the u niv ersal q uot ien t, |
0 \to_L(C_0) \to_E \to F \to_0$$ such_that_$E$ is_semi-stable_(see the proof_of Proposition \[prop:1\])._Thus $Z$ is not_empty. Therefore, to_compute_the Picard group of $\overline{M}_{H_x}(r;c_1,c_2)$, it is enough to consider $Pic(M^0 \cup (Z^0 \cap_M_0^{\mu})_)$, where_$Z^0$_is_the subscheme of $M_H(r;c_1,c_2)$ consisting_of stable sheaves which are_defined by_the exact sequence.
We set $V_i=\cal O_X(-nH_x)^{\oplus N_i}$, $(i=1,2$)._Let_$Quot_{V_1/X/\Bbb C}^{\gamma_1}$ (_resp. $Quot_{V_2/X/\Bbb C}^{\gamma _2}$) be a quot-scheme parametrizing all_quotients $V_1\to F_1$ (resp. $V_2 \to_F_2$) such that_$\gamma(F_1)=(r_1,C_0+\frac{d_1}{r_1}f,0)$_(resp._$\gamma(F_2)=(r_2,\frac{d_2}{r_2}f,0)$). Let $Q_i$ ($i=1,2$)_be the open subscheme of $Quot_{V_i/X/\Bbb_C}^{\gamma_i}$ consisting all quotients $V_i \to_F_i$ which satisfy
1. $F_i$ is $\mu$-semi-stable_with respect to $H_x$,
2. $F_{i|\pi^{-1}(\eta)}$_is a semi-stable vector bundle,_where $\eta$_is the generic point of_$C$,
3. $H^0(X,V_i(nH_x))_\cong H^0(X,F_i(nH_x))$,_$H^j(X,F_i(nH_x))=0,j>0$.
Let $V_i \otimes_\cal O_{Q_i \times X} \to \cal_F_i$ be the_universal quotient, |
h$ and $\ell$ are odd integers).[@Matsumura] This suggests that some structural change takes place at $T=T_{\rm Q}$ from the tetragonal phase at high temperatures.[@Tanaka; @Hirota] A buckling of sheets of B and C atoms was proposed,[@Tanaka] and the non-resonant intensities by the buckling has recently been evaluated; about $0.01$ ${\rm \AA}$ shift of B and/or C atoms may be sufficient to give rise to such large intensities.[@Adachi] It is not clear in experiments whether the intensity on this spot is resonantly enhanced at the $L_{\rm III}$ edge, since the non-resonant part is so large that it may mask the resonant behavior. On the other hand, the resonant enhancement of RXS intensities has clearly been observed on the superlattice spot $(00\frac{\ell}{2})$.
In this paper, we study the mechanism of the RXS spectra at the $L_{\rm III}$ edge in Phase II of DyB$_2$C$_2$. Since the $5d$ states are so extended in space that they are sensitive to lattice distortion caused by the buckling of sheets of B and C atoms. Then the question arises whether the direct influence of the lattice distortion on the $5d$ states is larger than the influence of the anisotropic $4f$ charge distribution associated with the quadrupole order through the $5d$-$4f$ Coulomb interaction. Lovesey and Knight[@Lovesey] have discussed the mechanism from the symmetry viewpoint, and have pointed out that the RXS intensities on $(00\frac{\ell}{2})$ and $(h0\frac{\ell}{2})$ spots come from lowering the local symmetry probably due to lattice distortion. The argument based on symmetry alone is powerful in some respect, but does not shed light on this issue. In the transition-metal compounds, the corresponding question has already been answered by [*a*b initio]{} calculations as mentioned above. However, such [*a*b initio]{} calculations are difficult in rare-earth compounds. We resort to a model calculation by treating the $5d$ states as a band and the $4f$ states as localized states. The buckling of sheets of B and C atoms causes modulations of the $5d$ bands and of the $4f$ states. We analyze such effects of lattice distortion on the basis of the point charge model,[@ | h$ and $ \ell$ are odd integers).[@Matsumura ] This suggests that some structural change take home at $ T = T_{\rm Q}$ from the tetragonal phase at high temperatures.[@Tanaka; @Hirota ] A buckling of sheets of B and C atom was proposed,[@Tanaka ] and the non - resonant intensities by the buckling has recently been evaluate; approximately $ 0.01 $ $ { \rm \AA}$ shift of B and/or C atoms may be sufficient to give rise to such large intensities.[@Adachi ] It is not absolved in experiments whether the intensity on this spotlight is resonantly enhance at the $ L_{\rm III}$ edge, since the non - resonant part is so large that it may dissemble the resonant behavior. On the other bridge player, the resonant enhancement of RXS intensities has clearly been detect on the superlattice spot $ (00\frac{\ell}{2})$.
In this paper, we study the mechanism of the RXS spectra at the $ L_{\rm III}$ edge in Phase II of DyB$_2$C$_2$. Since the $ 5d$ country are so extended in space that they are sensitive to lattice distortion caused by the buckling of sheets of B and C atoms. Then the question arises whether the direct influence of the lattice distortion on the $ 5d$ states is larger than the influence of the anisotropic $ 4f$ care distribution associated with the quadrupole order through the $ 5d$-$4f$ Coulomb interaction. Lovesey and Knight[@Lovesey ] have hash out the mechanism from the isotropy vantage point, and have pointed out that the RXS intensities on $ (00\frac{\ell}{2})$ and $ (h0\frac{\ell}{2})$ spot come from lowering the local symmetry probably due to wicket distortion. The argument based on symmetry alone is powerful in some respect, but does not shed light on this issue. In the transition - metal compound, the corresponding doubt has already been answered by [ * a*b initio ] { } calculations as mentioned above. However, such [ * a*b initio ] { } calculations are difficult in rare - ground compounds. We resort to a model calculation by treat the $ 5d$ states as a band and the $ 4f$ states as localized state. The buckling of sheets of B and hundred atoms causes transition of the $ 5d$ bands and of the $ 4f$ state of matter. We analyze such effects of wicket distortion on the basis of the point bang model,[@ | h$ ajd $\ell$ are odd integers).[@Mxtsumura] This sogtests vhat soje strucgural change takes place at $V=T_{\rm Q}$ frim the tetragonal phasd at high temperarurew.[@Ranaka; @Hirota] A bughlinf of vieets of B and G atoms was proposed,[@Tanakd] xnb the non-resonant intensities by the bucklimg has recently feen qvalhated; about $0.01$ ${\rm \AA}$ shift of B and/od C atoks may be sufgicient to give rise to sufh lwrge intensities.[@Adwchi] It is bot soear in expefiments whttker the intgnsity on this spot is resonantlh enhcnced at thg $M_{\rl III}$ edge, smnce tre non-resonakn part hs so lsrge that it msy kasj the resonant behavimr. On the other hagd, the revouant enhancement of RZS intetsithes fqs zlezrky been lbsxrved on ths superlattuce spot $(00\frac{\ell}{2})$.
In uhif paper, we stusy the mqchanism of the RXS spectra at the $L_{\rm PII}$ sdge in Phase II of DyB$_2$X$_2$. Since the $5d$ states wre so exeended in space that they are sensitive to lattica disvoftiin zquded by the buckling of sheets of B and C atomf. Tnek the question avises whether the cigevj influence of the lcftjce distortion on hhe $5d$ sjates us larger tham the influence of the anisitropic $4f$ chcrgw distribution asslciated witk the auadripole order through the $5d$-$4f$ Doulomb intfraction. Muvesey and Knighg[@Lonesef] have discussed the mechagism from the symmetfy voewpoigt, and havf poikded out that the RDS injensithes on $(00\fraf{\ell}{2})$ and $(h0\frac{\ell}{2})$ spots come from lowering thg lmcan symmetxy pronably due to laetice distortipn. The crgumevt based oh symmevry alone is powerful in vlme respect, uut does got whed light uv this issue. Im the traufitiin-metal compounds, bhe cudresponding quewtiin has already neev agsaeced br [*a*b initio]{} cdlcuuatkpns ar mentioned xbovr. However, such [*a*b inhtio]{} calculations are cinficult ib rare-eawth compounds. We resort to a mofel celculavion bu tteating the $5d$ states as a band znd the $4f$ stwtes as locajizee states. The buckling of sheets of B and C atoms canses modulations of thg $5d$ bands and of the $4y$ xtates. We aialyze such effacts of lattice distirtion on the basls of the point charge model,[@ | h$ and $\ell$ are odd integers).[@Matsumura] This some change takes at $T=T_{\rm Q}$ high @Hirota] A buckling sheets of B C atoms was proposed,[@Tanaka] and the intensities by the buckling has recently been evaluated; about $0.01$ ${\rm \AA}$ shift B and/or C atoms may be sufficient to give rise to such large It not in whether the intensity on this spot is resonantly enhanced at the $L_{\rm III}$ edge, since the part is so large that it may mask resonant behavior. On the hand, the resonant enhancement of intensities clearly been on superlattice $(00\frac{\ell}{2})$. In this we study the mechanism of the RXS spectra at the $L_{\rm III}$ edge in Phase II of Since the are so in that are sensitive to caused by the buckling of sheets C atoms. Then the question arises whether the influence of lattice distortion on the $5d$ states larger than the influence of the anisotropic $4f$ distribution associated with the quadrupole order through the $5d$-$4f$ Coulomb interaction. Lovesey and Knight[@Lovesey] have mechanism from the symmetry and have pointed that RXS on and $(h0\frac{\ell}{2})$ come from lowering the local symmetry probably due to lattice distortion. argument based on symmetry alone is powerful in some respect, not light on this In the transition-metal compounds, corresponding has already been answered initio]{} as such initio]{} are difficult in rare-earth We resort to a model by treating the $5d$ the $4f$ states as localized states. The buckling sheets of B and C atoms causes of the $5d$ bands and of the $4f$ states. We analyze such of lattice the basis of the point charge model,[@ | h$ and $\ell$ are odd integers).[@MatsUmura] This sUggesTs tHat SoMe stRuctUral change takeS PlacE at $T=T_{\rm Q}$ from the tetragoNal phAsE At hiGH tEmperAtures.[@TANaKA; @hirOtA] A BucKlINg Of sheEts Of B and C Atoms was prOpoSeD,[@Tanaka] and thE NoN-resonant iNteNsities by the BucKling hAs RecENtly bEen EvaluAted; abOUt $0.01$ ${\rm \AA}$ Shift of B aNd/OR C atomS May be suFFIcIent To give rise to such lARgE Intensities.[@AdaChi] It iS nOT cLEAr iN exPeriments wHeTher tHE intensITy ON THis SPot is resonantLy enhanced aT The $l_{\rm III}$ EdGe, sINce the Non-reSoNAnt Part is so larGe thAt it may maSk the rESonant bEHavior. ON the otHer HanD, the REsOnAnt EnHAncEMeNt oF rXS IntensitIeS hAs cleArly BEEN ObseRveD on tHe supErlattice spot $(00\FraC{\ell}{2})$.
iN thIs papEr, we sTudy ThE mechAnism oF the RxS Spectra at the $L_{\rm iII}$ eDge in PhasE II Of dyB$_2$c$_2$. SInce tHE $5d$ statEs aRe sO extendEd in spaCE thAt THEY aRe sensitive to lattiCe DIStOrtion caUsed by THe BuCKling of sHeEts Of B aND c atomS. TheN ThE questioN ariseS WhEtHer the dIrEct infLuEncE of The laTTice DistorTion on thE $5d$ staTEs is larger than THe influence of THe ANIsOTropIc $4f$ Charge distrIbutIOn asSociATeD wiTH the qUadruPoLE oRDer through the $5d$-$4f$ CoulOmB interActioN. Lovesey and KnIght[@LoveseY] HAVe discusSed tHE mEChanism from the SymmeTry viewpoiNT, and have PointEd out thaT the RXS inTENsities oN $(00\frAc{\eLl}{2})$ aNd $(h0\FRAc{\Ell}{2})$ spots come fROM lowErIng the lOcaL symmetRy pRobAblY duE tO lattice dIstortioN. THe ArGuMenT baseD On symmetRy AloNe Is pOwerfUL in somE respEct, bUt DoES noT shed liGHt ON This IsSuE. In tHe tRaNsitiOn-meTAl cOmpoundS, the correSpoNDing QuEsTion has Already been anSwEred by [*a*b inItIo]{} cAlculaTIOns as menTioned above. However, such [*a*B Initio]{} cAlcUlatiOns aRe difficuLt iN rare-eArtH CompouNds. We rEsort To A moDEL calcULAtIon By Treating thE $5D$ StaTes as A bAnd aNd the $4f$ sTates as localized stATes. the buckling of SheEts oF b AnD C aTOmS CauSeS ModULAtions of the $5d$ banDs and of the $4F$ sTAtEs. We analyzE SucH eFfects oF latticE distORtion on The basis oF the point ChArge MODel,[@ | h$ and $\ell$ are odd inte gers).[@Ma tsumu ra] Th is sug gest s that some st r uctu ral change takes place at $ T= T _{\r m Q }$ fr om thet et r a gon al p has ea thightem peratur es.[@Tanak a;@H irota] A buc k li ng of shee tsof B and C a tom s waspr opo s ed,[@ Tan aka]and th e non-r esonant i nt e nsitie s by the b uc klin g has recently be e ne valuated; abou t $0.0 1$ ${ \ r m \ AA} $ shift of B and/ o r C ato m sm a y be sufficient to give riset o s uch la rg e i n tensit ies.[ @A d ach i] It is no t cl ear in ex perime n ts whet h er theintens ity on thi s s po t i sr eso n an tly enh anced at t he $L_{ \rmI I I } $ ed ge, sin ce th e non-resonan t p arti s s o lar ge th at i tmay m ask th e res on ant behavior. O n th e other h and ,the r esona n t enha nce men t of RX S inten s iti es h a sclearly been obser ve d on the sup erlatt i ce s p ot $(00\ fr ac{ \ell } { 2})$.
In th is paper , we s t ud ythe mec ha nism o fthe RX S spe c traat the $L_{\rm III} $ edge in Phase II of DyB$_2$ C $_ 2 $ .S ince th e $5d$ stat es a r e so ext e nd edi n spa ce th at th e y are sensitive tola tticedisto rtion causedby the buc k l i ng of sh eets of B and C atoms. Then the quest i on arise s whe ther the direct i n f luence o f t helat tic e di stortion on t h e $5d $statesislargertha n t heinf lu ence of t he aniso tr op ic $ 4f$ char g e distri bu tio nass ociat e d with thequad ru po l e o rder th r ou g h the $ 5d $-$4 f$Co ulomb int e rac tion. L ovesey an d K n ight [@ Lo vesey]have discusse dthe mechan is m f rom th e symmetry viewpoint, and have po i nted ou t t hat t he R XS intens iti es on$(0 0 \frac{ \ell}{ 2})$an d $ ( h 0\fra c { \e ll} {2 })$ spotsc o mefromlo weri ng thelocal symmetry pro b abl y due to latt ice dis t o rt ion . T h e a rg u men t based on symmet ry alone i sp ow erful in s o mere spect,but doe s not shed li ght on th is issue. I n th e tra nsition-me tal comp ounds, th e corr e sp ondin g q uestio nhas alre ady be e n a nswer ed by[* a*b in itio] {} calcula tions as mentioned abov e. How ever, su ch [*a*bini t io] {} calcul atio ns are dif fic ult in r are - earth com p ou nds . We r esor t to a mod e lcal c u la tion by tre a t i ngthe $ 5d$ states asa band and the $4 f $ states as lo cali z e d s tat e s. T he buckling of s hee ts o f B andCatoms cause s modula ti o ns of the $ 5d$ ba nds and o ft he $4f $ st ate s. We ana lyz es uch eff ec ts of lat tice d istort ion on theb a sis of the point char g e mode l ,[@ | h$ and_$\ell$ are_odd integers).[@Matsumura] This suggests_that some_structural_change takes_place_at $T=T_{\rm Q}$_from the tetragonal_phase at high temperatures.[@Tanaka;_@Hirota] A buckling_of_sheets of B and C atoms was proposed,[@Tanaka] and the non-resonant intensities by the_buckling_has recently_been_evaluated;_about $0.01$ ${\rm \AA}$ shift_of B and/or C atoms_may be_sufficient to give rise to such large intensities.[@Adachi]_It_is not clear_in experiments whether the intensity on this spot is_resonantly enhanced at the $L_{\rm III}$_edge, since the_non-resonant_part_is so large that_it may mask the resonant behavior._On the other hand, the resonant_enhancement of RXS intensities has clearly been_observed on the superlattice spot $(00\frac{\ell}{2})$.
In_this paper, we study the_mechanism of_the RXS spectra at the_$L_{\rm III}$ edge_in Phase_II of DyB$_2$C$_2$._Since the $5d$ states are so_extended in space_that they are sensitive to lattice_distortion_caused by the_buckling_of_sheets of_B and C_atoms._Then the_question_arises whether the direct influence of_the_lattice distortion on the $5d$ states is_larger than the influence_of_the anisotropic $4f$ charge_distribution associated with the quadrupole_order through the $5d$-$4f$ Coulomb interaction._Lovesey and_Knight[@Lovesey] have_discussed the mechanism from the symmetry viewpoint, and have pointed out_that the RXS intensities on $(00\frac{\ell}{2})$_and $(h0\frac{\ell}{2})$ spots come_from lowering_the_local symmetry probably_due_to lattice_distortion. The argument based on symmetry alone_is powerful_in some respect, but does not_shed light on this_issue._In the transition-metal compounds, the corresponding_question has already been answered by_[*a*b initio]{} calculations as mentioned_above._However,_such [*a*b initio]{} calculations are_difficult in rare-earth compounds. We resort_to a model_calculation by treating the $5d$ states as_a_band and the $4f$ states as_localized_states. The buckling of sheets of_B_and_C atoms causes modulations of_the $5d$ bands and of the_$4f$ states. We analyze such effects of lattice distortion_on the basis_of the point charge model,[@ |
by @jordan04, mass segregation and stellar evolution effects are the leading causes of the blue/red GC size differences. In this scenario little change of relative GC sizes with galactocentric distances is expected.
By comparing the average half-light radii of the red and blue GC in NGC5866 we do not find any significant difference between the two subpopulations. Furthermore, the size versus galactocentric radius comparison, shown in Figure \[reffrad\], reveals a slight tendency of outer GC to have larger half-light radii.
Due to the small number of GCs available, the absence of a clear size difference between the red and blue GCs, and the uncertainties in the half-light radius versus $R_{gc}$ correlation, these observations cannot be used to clearly support either the mass segregation or the projection scenarios.
Comparison with models
----------------------
Figure \[spot\] shows a comparison between the observed colors of NGC5866 GCs and the predictions from R05. Each panel in the figure shows a Simple Stellar Population (SSP) model with a defined \[Fe/H\], and for ages 1 $\leq$ t (Gyr) $\leq$ 14.
At first glance, these panels will give little information about the age and metallicity properties of the observed GCs, due to the strong overlap of models with different physical properties. This is not related to the particular choice of SSP models. In order to reduce model systematics, we have also considered a data to models comparison using the @bc03 SSP models. The new set of models essentially agrees with the R05 one, and the overlap between models at different \[Fe/H\] is still present. Moreover, since the R05 models are optimized to match in detail the observational features of Color Magnitude Diagrams of Galactic and Magellanic Clouds stellar clusters, in this section we will take the set of stellar synthesis models from R05 as reference ones.
If no constraint is put on models, the conclusions that can be drawn from the comparison shown in Figure \[spot\] are mainly: $i$) SSP models, i.e. single age and single metallicity stellar systems, reproduce nicely the observational color-color properties of GCs; $ii$) except one single object, no old super metal-rich (t $>$ 7 Gyr, \[Fe/H\]$>$\[Fe/H\]$_{\sun}$) GC seems to be present; $iii$) the reddest GCs | by @jordan04, mass segregation and stellar evolution effects are the lead campaign of the blue / red GC size differences. In this scenario little variety of relative GC sizes with galactocentric distances is expect.
By comparing the average half - unaccented radii of the red and aristocratic GC in NGC5866 we do not find any significant remainder between the two subpopulations. Furthermore, the size versus galactocentric radius comparison, shown in Figure \[reffrad\ ], reveals a slight tendency of out GC to have larger half - light radii.
Due to the little number of GCs available, the absence of a clear size difference between the red and gloomy GCs, and the uncertainties in the half - light radius versus $ R_{gc}$ correlation, these observation cannot be used to clearly support either the mass segregation or the projection scenario.
Comparison with models
----------------------
Figure \[spot\ ] shows a comparison between the observed colors of NGC5866 GCs and the predictions from R05. Each panel in the figure shows a Simple Stellar Population (SSP) model with a specify \[Fe / H\ ], and for ages 1 $ \leq$ t (Gyr) $ \leq$ 14.
At beginning glance, these panel will give little information about the old age and metallicity properties of the observed GCs, due to the strong lap of models with different physical properties. This is not related to the particular choice of SSP models. In order to reduce model systematics, we have also consider a datum to models comparison using the @bc03 SSP models. The new set of models essentially agrees with the R05 one, and the lap between models at unlike \[Fe / H\ ] is still present. Moreover, since the R05 models are optimized to match in contingent the observational features of Color Magnitude Diagrams of Galactic and Magellanic Clouds stellar clusters, in this section we will claim the set of stellar deduction models from R05 as reference ones.
If no constraint is put on mannequin, the conclusions that can be drawn from the comparison shown in Figure \[spot\ ] are mainly: $ i$) SSP models, i.e. single age and single metallicity leading systems, reproduce nicely the experimental color - color properties of gigahertz; $ ii$) except one single object, no old super metal - rich (t $ > $ 7 Gyr, \[Fe / H\]$>$\[Fe / H\]$_{\sun}$) GC seems to be present; $ iii$) the reddest GCs | by @jordan04, mass segregation and stellar evolution effecfs are tfe leading causes of the blux/red GC suze differences. In thir scenaril little chaige of relative JD sizes with fwlaccorentric distancgs is expectad.
By comparing tfe average half-light radii of the red and blie GC in NGC5866 we qo npe fihd any significant difference betwsen the two subpopulstions. Furthermore, the sizf vegsus galactocentrif radius conparywon, shown in Figure \[reffrad\], reveala a slight tendency of outer GC to hcve larger yaof-llcht radii.
Dux to tre small numnvr of GWs avaikable, the absekce oh a xlear size difference between the red agd blue GWs, and the uncertaibtues it tha haud-lieht redihs verdus $R_{gc}$ corremation, thesw observations cannpt ve used to clsarly fu[port either the mass segregation or tht pronection scenarios.
Comparuson with models
----------------------
Figurg \[spot\] shors a comparison between the observed colors of NGW5866 GCs xnd tme pfwdlctions from R05. Each panel in the figure shows z Xikple Stellar Pjpulation (SXP) mpqel with a dewined \[Ys/H\], and for ages 1 $\leq$ t (Gyr) $\jeq$ 14.
Ar first gjancr, these panels will give lirtle informanion about the age and metallicitv propgrties of the observed GCs, duz to tge strong ogerlap of oodels with diffdrekt [hysical properties. This if not releted co the pxrtivular shoice of DSP models. In order to rfduce mmdel systelatics, we have also considered e data to modgls cokparison usinn the @bc03 SSP moqels. The new sgt of modzls esrentially zgrees xith the R05 oge, and the ovatlap between kodels ae didferwnt \[Fe/H\] ks still presemt. Moreovvr, sinxe the R05 models arc optkjized to match nu eetail the obsetvagiogap heatuwas of Color Kagnktuar Diaerams of Gaoxctiv and Magellanic Clogds atellar clusters, im bhis sectuon we wyll take the xet of stellar synuhesis movels ftom R05 as reference ones.
If no consfraint is pub on models, tre cinclusions tkat can be drawn from the comparison shoxn in Figure \[spot\] are mqinly: $i$) SSP models, n.e. single age and fingle medallicity stellar sywtems, reproduce nlcely the observationam colos-colog properties of GCs; $ii$) except one single object, no old super metal-rich (t $>$ 7 Gyr, \[He/R\]$>$\[Fe/H\]$_{\sun}$) GC aeemx to te 'resent; $iii$) vhe reddest GCs | by @jordan04, mass segregation and stellar evolution the causes of blue/red GC size change relative GC sizes galactocentric distances is By comparing the average half-light radii the red and blue GC in NGC5866 we do not find any significant between the two subpopulations. Furthermore, the size versus galactocentric radius comparison, shown in \[reffrad\], a tendency outer GC to have larger half-light radii. Due to the small number of GCs available, the of a clear size difference between the red blue GCs, and the in the half-light radius versus correlation, observations cannot used clearly either the mass or the projection scenarios. Comparison with models ---------------------- Figure \[spot\] shows a comparison between the observed colors NGC5866 GCs predictions from Each in figure shows a Population (SSP) model with a defined ages 1 $\leq$ t (Gyr) $\leq$ 14. At glance, these will give little information about the and metallicity properties of the observed GCs, due the strong overlap of models with different physical properties. This is not related to the of SSP models. In to reduce model we also a to models using the @bc03 SSP models. The new set of models essentially with the R05 one, and the overlap between models at is present. Moreover, since R05 models are optimized match detail the observational features Magnitude of Clouds clusters, this section we will the set of stellar synthesis from R05 as reference put on models, the conclusions that can be from the comparison shown in Figure \[spot\] mainly: $i$) SSP models, i.e. single age and single metallicity stellar systems, nicely the properties of GCs; $ii$) except one single object, old super metal-rich (t 7 Gyr, \[Fe/H\]$>$\[Fe/H\]$_{\sun}$) GC seems to be present; $iii$) reddest | by @jordan04, mass segregation anD stellar evOlutiOn eFfeCtS are The lEading causes of THe blUe/red GC size differences. in thiS sCEnarIO lIttle Change oF ReLATivE Gc sIzeS wITh GalacTocEntric dIstances is ExpEcTed.
By compariNG tHe average hAlf-Light radii of The Red and BlUe Gc In NGC5866 We dO not fInd any SIgnifiCant diffeReNCe betwEEn the twO SUbPopuLations. FurthermorE, ThE Size versus galaCtocenTrIC rADIus ComParison, shoWn In FigURe \[reffrAD\], rEVEAls A Slight tendencY of outer GC tO HavE largeR hAlf-LIght raDii.
DuE tO The Small number Of GCS availablE, the abSEnce of a CLear sizE diffeRenCe bEtweEN tHe Red AnD BluE gCS, anD The UncertaiNtIeS in thE halF-LIGHt raDiuS verSus $R_{gC}$ correlation, tHesE obsERvaTions CannoT be uSeD to clEarly sUpporT eIther the mass segRegaTion or the ProJeCtiOn ScenaRIos.
ComParIsoN with moDels
----------------------
FigURe \[sPoT\] SHOwS a comparison betweeN tHE ObServed coLors of ngC5866 gCS And the prEdIctIons FROm R05. EaCh paNEl In the figUre shoWS a siMple SteLlAr PopuLaTioN (SSp) modeL With A definEd \[Fe/H\], and For agES 1 $\leq$ t (Gyr) $\leq$ 14.
At fIRst glance, thesE PaNELs WIll gIve Little inforMatiON aboUt thE AgE anD MetalLicitY pROpERties of the observed Gcs, Due to tHe strOng overlap of mOdels with dIFFErent phySicaL PrOPerties. This is nOt relAted to the pARticular ChoicE of SSP moDels. In ordER To reduce ModEl sYstEmaTICs, We have also conSIDereD a Data to mOdeLs compaRisOn uSinG thE @bC03 SSP modelS. The new sEt Of MoDeLs eSsentIAlly agreEs WitH tHe R05 One, anD The oveRlap bEtweEn MoDEls At diffeREnT \[fE/H\] is StIlL preSenT. MOreovEr, siNCe tHe R05 modeLs are optiMizED to mAtCh In detaiL the observatiOnAl features Of colOr MagnITUde DiagrAms of Galactic and MagellaNIc CloudS stEllar ClusTers, in thiS seCtion wE wiLL take tHe set oF stelLaR syNTHesis MODeLs fRoM R05 as refereNCE onEs.
If nO cOnstRaint is Put on models, the concLUsiOns that can be dRawN froM THe ComPArISon ShOWn iN fIgure \[spot\] are maiNly: $i$) SSP modElS, I.e. Single age aND siNgLe metalLicity sTellaR Systems, Reproduce Nicely the ObServATIonAl color-colOr properTies of GCs; $II$) excePT oNe sinGle Object, No Old Super Metal-rICh (t $>$ 7 gyr, \[Fe/h\]$>$\[Fe/H\]$_{\suN}$) Gc seems To be pReSent; $iii$) tHe reddest GCs | by @jordan04, mass segreg ation andstell arevo lu tion eff ects are the l e adin g causes of the blue/r ed GC s i ze d i ff erenc es. Int hi s sce na ri o l it t le chan geof rela tive GC si zes w ith galactoc e nt ric distan ces is expected .
By com pa rin g theave ragehalf-l i ght ra dii of th er ed and blue GC i nNGC5 866 we do not fin d a n y significantdiffer en c eb e twe enthe two su bp opula t ions. F u rt h e r mor e , the size ve rsus galact o cen tric r ad ius compar ison, s h own in Figure\[re ffrad\],reveal s a slig h t tende ncy of ou ter GCt oha vela r ger ha lf- l igh t radii.
Du e tothes m a l l nu mbe r of GCsavailable, th e a bsen c e o f a c learsize d iffer ence b etwee nthe red and blu e GC s, and th e u nc ert ai nties in the ha lf- light r adius v e rsu s$ R _ {g c}$ correlation, t he s e o bservati ons ca n no tb e used t ocle arly s uppor t ei t he r the ma ss seg r eg at ion orth e proj ec tio n s cenar i os.
Compa rison wi th mo d els
---------- - -----------
F ig u r e\ [spo t\] shows a co mpar i sonbetw e en th e obse rvedco l or s of NGC5866 GCs and t he pre dicti ons from R05. Each pane l i n the fi gure sh o ws a Simple St ellar Populatio n (SSP) m odelwith a d efined \[ F e /H\], an d f orage s 1 $ \l eq$ t (Gyr) $ \ l eq$14 .
At f irs t glanc e,the sepan el s will gi ve littl ein fo rm ati on ab o ut the a ge an dmet allic i ty pro perti es o fth e ob servedG Cs , dueto t he s tro ng over lapo f m odels w ith diffe ren t phy si ca l prope rties. This i snot relate dtothe pa r t icular c hoice of SSP models. In order t o r educe mod el system ati cs, we ha v e also consi dered a da t a to m o d el s c om parison us i n g t he @b c0 3 SS P model s. The new set ofm ode ls essentiall y a gree s wi tht he R05 o n e,a n d the overlap b etween mod el s a t differen t \[ Fe /H\] is stillprese n t. More over, sin ce the R0 5mode l s ar e optimize d to mat ch in det a il th e o bserv ati onal f ea tur es of Color Mag nitud e Diag ra ms ofGalac ti c and Ma gellanic Clouds stellar clust ers,inthis sect ion wewill take the set of st ell arsynth esi s mode ls f r om R0 5 as r efer e nce ones.
I f n o co nstraint is p u t o n mod els , the c oncl usions that can b e drawn from th e co m p ari son show nin Figure \[sp ot\ ]a r e mainly :$i$) SSP mo dels, i. e. singl e ageand si ngle me t a ll i city s tell arsystems,rep ro d uce nic el yt he obs erva ti onal c olor-c o lorp r operties of GCs; $ii$ ) excep t on e sin gl e objec t , no old super metal-rich (t $> $ 7Gyr,\[Fe/H\ ]$ >$\[Fe /H\ ]$ _{\sun}$)G C seems t o bepresent ;$iii $)the re ddes t GCs | by_@jordan04, mass_segregation and stellar evolution_effects are_the_leading causes_of_the blue/red GC_size differences. In_this scenario little change_of relative GC_sizes_with galactocentric distances is expected.
By comparing the average half-light radii of the red and_blue_GC in_NGC5866_we_do not find any significant_difference between the two subpopulations._Furthermore, the_size versus galactocentric radius comparison, shown in Figure_\[reffrad\],_reveals a slight_tendency of outer GC to have larger half-light radii.
Due_to the small number of GCs_available, the absence_of_a_clear size difference between_the red and blue GCs, and_the uncertainties in the half-light radius_versus $R_{gc}$ correlation, these observations cannot be_used to clearly support either the_mass segregation or the projection_scenarios.
Comparison with_models
----------------------
Figure \[spot\] shows a comparison_between the observed_colors of_NGC5866 GCs and_the predictions from R05. Each panel_in the figure_shows a Simple Stellar Population (SSP)_model_with a defined_\[Fe/H\],_and_for ages_1 $\leq$ t_(Gyr)_$\leq$ 14.
At_first_glance, these panels will give little_information_about the age and metallicity properties of_the observed GCs, due_to_the strong overlap of_models with different physical properties._This is not related to the_particular choice_of SSP_models. In order to reduce model systematics, we have also considered_a data to models comparison using_the @bc03 SSP models._The new_set_of models essentially_agrees_with the_R05 one, and the overlap between models_at different_\[Fe/H\] is still present. Moreover, since_the R05 models are_optimized_to match in detail the observational_features of Color Magnitude Diagrams of_Galactic and Magellanic Clouds stellar_clusters,_in_this section we will take_the set of stellar synthesis models_from R05 as_reference ones.
If no constraint is put on_models,_the conclusions that can be drawn_from_the comparison shown in Figure \[spot\]_are_mainly:_$i$) SSP models, i.e. single_age and single metallicity stellar systems,_reproduce nicely the observational color-color properties of GCs; $ii$)_except one single_object, no old super metal-rich_(t_$>$_7 Gyr, \[Fe/H\]$>$\[Fe/H\]$_{\sun}$) GC seems to be present; $iii$) the_reddest GCs |
tilde{E}(\tilde{T})-\tilde{E}_{bp}$ for the parameter values $\tilde{\omega}_i$ presented in Fig. \[fig1\], \[fig2\]. \[fig3\]](fig3_2.eps "fig:"){width="80.00000%"}
The dependencies obtained enable us to find the heat capacity of a TI bipolaron gas: $C_v(\tilde{T})=d\tilde{E}/d\tilde{T}$. With the use of (20) $C_v(\tilde{T})$ for $\tilde{T}\leq\tilde{T}_c$ is expressed as:
$$\label{22}
C_v(\tilde{T})=\frac{\tilde{T}^{3/2}}{2C_{bp}}
\left[\frac{\tilde{\omega}^2}{\tilde{T}^2}F_{1/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+
6\left(\frac{\tilde{\omega}}{\tilde{T}}\right)F_{3/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+
5F_{5/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)\right],$$
$$F_{1/2}(\alpha)=\frac{2}{\sqrt{\pi}}\int^{\infty}_0\frac{1}{\sqrt{x}}\ \frac{dx}{e^{x+\alpha}-1}$$
Expression (22) yields a well-known exponential dependence of the heat capacity at low temperatures $C_v\backsim\text{exp}(-\omega_0/T)$, caused by the availability of the energy gap $\omega_0$.
![ Temperature dependencies of the heat capacity for various values of the parameters $\omega_i$: $\omega_0=0$; $\ \tilde{T}_{C_0}=25,2$; $\ C_v(\tilde{T}_{c1})=2$; $\omega_1=0,2$; $\ \tilde{T}_{C_1}=27,3$; $\ C_v(\tilde{T}_{c1}-0)=2,16$; $\ C_v(\tilde{T}_{c1}+0)=1,9$; $\omega_2=1$; $\ \ \tilde{T}_{C_2}=30$; $\ C_v(\tilde{T}_{c2}-0)=2,46 | tilde{E}(\tilde{T})-\tilde{E}_{bp}$ for the parameter values $ \tilde{\omega}_i$ presented in Fig. \[fig1\ ], \[fig2\ ]. \[fig3\]](fig3_2.eps " fig:"){width="80.00000% " }
The dependencies prevail enable us to receive the heat capability of a TI bipolaron flatulence: $ C_v(\tilde{T})=d\tilde{E}/d\tilde{T}$. With the use of (20) $ C_v(\tilde{T})$ for $ \tilde{T}\leq\tilde{T}_c$ is expressed as:
$ $ \label{22 }
C_v(\tilde{T})=\frac{\tilde{T}^{3/2}}{2C_{bp } }
\left[\frac{\tilde{\omega}^2}{\tilde{T}^2}F_{1/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+
6\left(\frac{\tilde{\omega}}{\tilde{T}}\right)F_{3/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+
5F_{5/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)\right],$$
$ $ F_{1/2}(\alpha)=\frac{2}{\sqrt{\pi}}\int^{\infty}_0\frac{1}{\sqrt{x}}\ \frac{dx}{e^{x+\alpha}-1}$$
Expression (22) yield a well - know exponential dependence of the heat capacity at low temperature $ C_v\backsim\text{exp}(-\omega_0 / T)$, caused by the availability of the department of energy gap $ \omega_0$.
! [ Temperature dependencies of the heat capacity for various values of the parameters $ \omega_i$: $ \omega_0=0 $; $ \ \tilde{T}_{C_0}=25,2 $; $ \ C_v(\tilde{T}_{c1})=2 $; $ \omega_1=0,2 $; $ \ \tilde{T}_{C_1}=27,3 $; $ \ C_v(\tilde{T}_{c1}-0)=2,16 $; $ \ C_v(\tilde{T}_{c1}+0)=1,9 $; $ \omega_2=1 $; $ \ \ \tilde{T}_{C_2}=30 $; $ \ C_v(\tilde{T}_{c2}-0)=2,46 | tilfe{E}(\tilde{T})-\tilde{E}_{bp}$ for tht parameter values $\tilde{\mmega}_i$ presentdd in Fig. \[fig1\], \[fig2\]. \[fig3\]](fig3_2.eps "fmg:"){wieth="80.00000%"}
Tht dependencies obtakned enabpe us to finv the heat capacmfy of a TI billlarmi gas: $C_v(\tilde{T})=d\jilde{E}/d\tilde{D}$. With the use ow (20) $C_v(\tilde{T})$ for $\tilde{T}\leq\tilde{T}_c$ is evpressec ws:
$$\label{22}
C_v(\tilde{J})=\frac{\uilqe{T}^{3/2}}{2C_{gi}}
\ltft[\frac{\tilde{\omega}^2}{\tilde{T}^2}F_{1/2}\left(\frac{\tjlde{\omeja}}{\tilde{T}}\right)+
6\legt(\frac{\tilde{\omega}}{\tilde{T}}\righh)F_{3/2}\levt(\frac{\tilde{\omega}}{\tipde{T}}\right)+
5F_{5/2}\lgrt(\fwqc{\tilde{\omega}}{\gilde{T}}\righu)\rnght],$$
$$F_{1/2}(\alpha)=\ftac{2}{\sqrt{\pi}}\int^{\infty}_0\frac{1}{\sqrt{x}}\ \frac{db}{e^{x+\al'ha}-1}$$
Expressiin (22) ylglds a well-kiown eqponential deivndence of the heat capacity at loq temperatures $C_v\backvim\text{exp}(-\omega_0/T)$, cwused by dhz availability of the ebergy gap $\omeeq_0$.
![ Tdmptravurs depejdeicies of ths heat capaxity for various vakuqw of the parajeters $\oiega_i$: $\omega_0=0$; $\ \tilde{T}_{C_0}=25,2$; $\ C_v(\tilde{T}_{c1})=2$; $\omega_1=0,2$; $\ \uilde{F}_{C_1}=27,3$; $\ C_v(\tilde{T}_{c1}-0)=2,16$; $\ C_v(\tilde{T}_{x1}+0)=1,9$; $\omega_2=1$; $\ \ \tilde{T}_{C_2}=30$; $\ C_v(\jilde{T}_{c2}-0)=2,46 | tilde{E}(\tilde{T})-\tilde{E}_{bp}$ for the parameter values $\tilde{\omega}_i$ presented \[fig1\], \[fig3\]](fig3_2.eps "fig:"){width="80.00000%"} dependencies obtained enable capacity a TI bipolaron $C_v(\tilde{T})=d\tilde{E}/d\tilde{T}$. With the of (20) $C_v(\tilde{T})$ for $\tilde{T}\leq\tilde{T}_c$ is as: $$\label{22} C_v(\tilde{T})=\frac{\tilde{T}^{3/2}}{2C_{bp}} \left[\frac{\tilde{\omega}^2}{\tilde{T}^2}F_{1/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+ 6\left(\frac{\tilde{\omega}}{\tilde{T}}\right)F_{3/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+ 5F_{5/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)\right],$$ $$F_{1/2}(\alpha)=\frac{2}{\sqrt{\pi}}\int^{\infty}_0\frac{1}{\sqrt{x}}\ \frac{dx}{e^{x+\alpha}-1}$$ Expression (22) yields a well-known dependence of the heat capacity at low temperatures $C_v\backsim\text{exp}(-\omega_0/T)$, caused by the availability the gap ![ dependencies of the heat capacity for various values of the parameters $\omega_i$: $\omega_0=0$; $\ \tilde{T}_{C_0}=25,2$; $\ $\omega_1=0,2$; $\ \tilde{T}_{C_1}=27,3$; $\ C_v(\tilde{T}_{c1}-0)=2,16$; $\ C_v(\tilde{T}_{c1}+0)=1,9$; $\omega_2=1$; \ \tilde{T}_{C_2}=30$; $\ C_v(\tilde{T}_{c2}-0)=2,46 | tilde{E}(\tilde{T})-\tilde{E}_{bp}$ for thE parameter ValueS $\tiLde{\OmEga}_i$ PresEnted in Fig. \[fig1\], \[fIG2\]. \[fig3\]](Fig3_2.eps "fig:"){width="80.00000%"}
The depenDenciEs OBtaiNEd EnablE us to fiND tHE HeaT cApAciTy OF a tI bipOlaRon gas: $C_V(\tilde{T})=d\tiLde{e}/d\Tilde{T}$. With thE UsE of (20) $C_v(\tilde{t})$ foR $\tilde{T}\leq\tiLde{t}_c$ is exPrEssED as:
$$\laBel{22}
c_v(\tilDe{T})=\fraC{\Tilde{T}^{3/2}}{2c_{bp}}
\left[\frAc{\TIlde{\omEGa}^2}{\tilde{t}^2}f_{1/2}\LeFt(\frAc{\tilde{\omega}}{\tilde{t}}\RiGHt)+
6\left(\frac{\tildE{\omega}}{\TiLDe{t}}\RIghT)F_{3/2}\lEft(\frac{\tilDe{\Omega}}{\TIlde{T}}\riGHt)+
5f_{5/2}\LEFt(\fRAc{\tilde{\omega}}{\tIlde{T}}\right)\rIGht],$$
$$f_{1/2}(\alpha)=\FrAc{2}{\sQRt{\pi}}\inT^{\inftY}_0\fRAc{1}{\sQrt{x}}\ \frac{dx}{e^{X+\alpHa}-1}$$
ExpressIon (22) yieLDs a well-KNown expOnentiAl dEpeNdenCE oF tHe hEaT CapACiTy aT Low TemperatUrEs $c_v\bacKsim\TEXT{Exp}(-\oMegA_0/T)$, caUsed bY the availabilIty Of thE EneRgy gaP $\omegA_0$.
![ TemPeRaturE depenDenciEs Of the heat capaciTy foR various vAluEs Of tHe ParamETers $\omEga_I$: $\omEga_0=0$; $\ \tildE{T}_{C_0}=25,2$; $\ C_v(\tiLDe{T}_{C1})=2$; $\oMEGA_1=0,2$; $\ \tIlde{T}_{C_1}=27,3$; $\ C_v(\tilde{T}_{c1}-0)=2,16$; $\ C_v(\tIlDE{t}_{c1}+0)=1,9$; $\Omega_2=1$; $\ \ \tilDe{T}_{C_2}=30$; $\ C_v(\TIlDe{t}_{C2}-0)=2,46 | tilde{E}(\tilde{T})-\tilde {E}_{bp}$for t hepar am eter val ues $\tilde{\o m ega} _i$ presented in Fig.\[fig 1\ ] , \[ f ig 2\].\[fig3\ ] ]( f i g3_ 2. ep s " fi g :" ){wid th= "80.000 00%"}
The de pe ndencies obt a in ed enableusto find thehea t capa ci tyo f a T I b ipola ron ga s : $C_v (\tilde{T }) = d\tild e {E}/d\t i l de {T}$ . With the use of (2 0 ) $C_v(\tilde{ T})$ f or $\ t i lde {T} \leq\tilde {T }_c$i s expre s se d a s:$$\label{22}C_v(\tilde{ T })= \frac{ \t ild e {T}^{3 /2}}{ 2C _ {bp }}
\left[\f rac{ \tilde{\o mega}^ 2 }{\tild e {T}^2}F _{1/2} \le ft( \fra c {\ ti lde {\ o meg a }} {\t i lde {T}}\rig ht )+
6\le ft(\ f r a c {\ti lde {\om ega}} {\tilde{T}}\r igh t)F_ { 3/2 }\lef t(\fr ac{\ ti lde{\ omega} }{\ti ld e{T}}\right)+
5 F_{5 /2}\left( \fr ac {\t il de{\o m ega}}{ \ti lde {T}}\ri ght)\ri g ht] ,$ $ $$ F_{1/2}(\alpha)=\f ra c { 2} {\sqrt{\ pi}}\i n t^ {\ i nfty}_0\ fr ac{ 1}{\ s q rt{x} }\ \ f ra c{dx}{e^ {x+\al p ha }- 1}$$
E xp ressio n(22 ) y ields a we ll-kno wn expon entia l dependence of the heat capa c it y at lowtem peratures $ C_v\ b acks im\t e xt {ex p }(-\o mega_ 0/ T )$ , caused by the avai la bility of t he energy gap $\omega_0 $ .
![ Temp erat u re dependencies o f the heat capa c ity forvario us value s of thep a rameters $\ ome ga_ i$: $ \o mega_0=0$; $\ \ tild e{ T}_{C_0 }=2 5,2$; $ \ C _v( \ti lde {T }_{c1})=2 $; $\ome ga _1 =0 ,2 $;$\ \t i lde{T}_{ C_ 1}= 27 ,3$ ; $\C _v(\ti lde{T }_{c 1} -0 ) =2, 16$; $\ C_ v ( \til de {T }_{c 1}+ 0) =1,9$ ; $\ o meg a_2=1$; $\ \ \ti lde { T}_{ C_ 2} =30$; $ \ C_v(\tilde{ T} _{c2}-0)=2 ,4 6 | tilde{E}(\tilde{T})-\tilde{E}_{bp}$ for_the parameter_values $\tilde{\omega}_i$ presented in_Fig. \[fig1\], \[fig2\]._\[fig3\]](fig3_2.eps_"fig:"){width="80.00000%"}
The dependencies_obtained_enable us to_find the heat_capacity of a TI_bipolaron gas: $C_v(\tilde{T})=d\tilde{E}/d\tilde{T}$._With_the use of (20) $C_v(\tilde{T})$ for $\tilde{T}\leq\tilde{T}_c$ is expressed as:
$$\label{22}
C_v(\tilde{T})=\frac{\tilde{T}^{3/2}}{2C_{bp}}
\left[\frac{\tilde{\omega}^2}{\tilde{T}^2}F_{1/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+
6\left(\frac{\tilde{\omega}}{\tilde{T}}\right)F_{3/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)+
5F_{5/2}\left(\frac{\tilde{\omega}}{\tilde{T}}\right)\right],$$
$$F_{1/2}(\alpha)=\frac{2}{\sqrt{\pi}}\int^{\infty}_0\frac{1}{\sqrt{x}}\ \frac{dx}{e^{x+\alpha}-1}$$
Expression (22) yields a_well-known_exponential dependence_of_the_heat capacity at low temperatures_$C_v\backsim\text{exp}(-\omega_0/T)$, caused by the availability_of the_energy gap $\omega_0$.
![ Temperature dependencies of the heat_capacity_for various values_of the parameters $\omega_i$: $\omega_0=0$; $\ \tilde{T}_{C_0}=25,2$; $\ C_v(\tilde{T}_{c1})=2$;_$\omega_1=0,2$; $\ \tilde{T}_{C_1}=27,3$; $\ C_v(\tilde{T}_{c1}-0)=2,16$; $\_C_v(\tilde{T}_{c1}+0)=1,9$; $\omega_2=1$; $\_\_\tilde{T}_{C_2}=30$;_$\ C_v(\tilde{T}_{c2}-0)=2,46 |
We identify satellite galaxies in a suite of eight simulations of the formation of $L_*$ galaxies in the $\Lambda$CDM scenario. This series has been presented by Abadi, Navarro & Steinmetz (2006), and follow the same numerical scheme originally introduced by @steinmetzandnavarro02. The “primary” galaxies in these simulations have been analyzed in detail in several recent papers, which the interested reader may wish to consult for details [@abadi03a; @abadi03b; @meza03; @meza05; @navarro04]. We give a brief outline below for completeness.
Each simulation follows the evolution of a small region of the universe chosen so as to encompass the mass of an $L_{*}$ galaxy system. This region is chosen from a large periodic box and resimulated at higher resolution preserving the tidal fields from the whole box. The simulation includes the gravitational effects of dark matter, gas and stars, and follows the hydrodynamical evolution of the gaseous component using the Smooth Particle Hydrodynamics (SPH) technique [@steinmetz96]. We adopt the following cosmological parameters for the $\Lambda$CDM scenario: $H_0=65$ km/s/Mpc, $\sigma_8=0.9$, $\Omega_{\Lambda}=0.7$, $\Omega_{\rm CDM}=0.255$, $\Omega_{\rm b}=0.045$, with no tilt in the primordial power spectrum.
All re-simulations start at redshift $z_{\rm init}=50$, have force resolution of order $1$ kpc, and the mass resolution is chosen so that each galaxy is represented on average, at $z=0$, with $\sim 50,000$ dark matter/gas particles. Gas is turned into stars at rates consistent with the empirical Schmidt-like law of @kennicutt98. Because of this, star formation proceeds efficiently only in high-density regions at the center of dark halos, and the stellar components of primary and satellite galaxies are strongly segregated spatially from the dark matter.
Each re-simulation follows a single $\sim L_*$ galaxy in detail, and resolves as well a number of smaller, self-bound systems of stars, gas, and dark matter we shall call generically “satellites”. We shall hereafter refer to the main galaxy indistinctly as “primary” or “host”. The resolved satellites span a range of luminosities, down | We identify satellite galaxies in a suite of eight simulations of the constitution of $ L_*$ galaxy in the $ \Lambda$CDM scenario. This series has been presented by Abadi, Navarro & Steinmetz (2006), and follow the like numerical scheme originally introduce by @steinmetzandnavarro02. The “ primary ” galaxies in these simulation have been analyzed in detail in respective recent papers, which the interested reader may wish to confer for details [ @abadi03a; @abadi03b; @meza03; @meza05; @navarro04 ]. We give a abbreviated outline below for completeness.
Each model succeed the evolution of a small area of the universe chosen so as to embrace the mass of an $ L_{*}$ galax system. This region is chosen from a large periodic box and resimulated at higher settlement preserving the tidal fields from the whole box. The simulation includes the gravitational effects of dark matter, gas and stars, and follows the hydrodynamical evolution of the gaseous component using the Smooth Particle Hydrodynamics (SPH) technique [ @steinmetz96 ]. We adopt the follow cosmological parameters for the $ \Lambda$CDM scenario: $ H_0=65 $ km / s / Mpc, $ \sigma_8=0.9 $, $ \Omega_{\Lambda}=0.7 $, $ \Omega_{\rm CDM}=0.255 $, $ \Omega_{\rm b}=0.045 $, with no rock in the primordial power spectrum.
All re - model start at redshift $ z_{\rm init}=50 $, have force settlement of order $ 1 $ kpc, and the mass resolution is chosen so that each galaxy is represented on median, at $ z=0 $, with $ \sim 50,000 $ dark matter / gas particles. accelerator is turned into star at rates consistent with the empirical Schmidt - like law of @kennicutt98. Because of this, asterisk formation proceeds efficiently only in eminent - density regions at the center of dark halos, and the leading components of primary and satellite galaxies are strongly segregated spatially from the black matter.
Each rhenium - simulation watch a single $ \sim L_*$ galax in detail, and resolves as well a issue of smaller, self - tie down systems of stars, gas, and dark matter we shall call generically “ satellites ”. We shall hereafter mention to the main galaxy dimly as “ primary ” or “ host ”. The resolved satellites span a range of luminosities, down |
We ldentify satellite galaxles in a suite oy eight simulztions ow the formation of $L_*$ galaxied un tht $\Lambda$CDM scenariu. This segies has veen presented ug Abadi, Navardl & Scemnmetz (2006), and folkow the sake numerical swhdmz originally introduced by @steinmetzwndnavatrl02. The “primary” dalaqiqs ih these simulations have been analgzed in detail in seferal recent papers, which hhe lnterested reader lay wish to confylt for detakls [@abadi03a; @abadi03b; @meza03; @meza05; @navarro04]. We give a brief ojtlinz below for cimppgteness.
Each wimulwtion follows the evmlution of a small renion mf rhe universe chosen sm as to encompass jhe mass ox cn $L_{*}$ galaxy system. Thus regimn iv chuwen frkm a large peciodic box znd resimulqted at higher resokueppn preservinf the eiqal fields from the whole box. The simuldtikn includes the gravitarional effects of darn matter, das and stars, and follows the hydrodynamical evolgtion uf uhc eqsfous component using the Smooth Particle Hydrjsymakics (SPH) technlque [@steinmetz96]. We sdlpy the followine cosmolofical parameters flr the $\Jambdq$CDM scenwrio: $H_0=65$ km/s/Mpc, $\sigma_8=0.9$, $\Omega_{\Lambda}=0.7$, $\Omega_{\rm CDM}=0.255$, $\Imega_{\rm b}=0.045$, with no cilt in the 'rimorcial lower spectrum.
All re-simblatiohs start at redshift $x_{\rm init}=50$, have fofce rasolution of order $1$ kpc, anq the masw rexolutiov is choseg so that fach ndlaxy is representfd on aeerage, at $x=0$, with $\sim 50,000$ dark matter/gas partmrles. Gas is tornad pnto starf at vates consistene with the empnrical Sehmidt-uike law or @kennirutt98. Because of this, star vormation prmceeds esficuentoy only kn high-density regions cu the centwr of dark halos, akd thg atellar componeuus if primary and xatdllytv gelaxiqv are strongny sdgrdbated spatially nroo thr dark matter.
Each re-vimumation follows a sonnle $\sim L_*$ galaxy yn detail, and resolves as well w numuer of smalker, self-bound systems of stars, gaa, and darn mwtter we shajl cqll genericakly “satellites”. We shall hereafter refer to the main galaxt indistinctly as “ptimsry” or “host”. The wesolved vatellites span a rabge of luminositits, down | We identify satellite galaxies in a suite simulations the formation $L_*$ galaxies in has presented by Abadi, & Steinmetz (2006), follow the same numerical scheme originally by @steinmetzandnavarro02. The “primary” galaxies in these simulations have been analyzed in detail several recent papers, which the interested reader may wish to consult for details @abadi03b; @meza05; We a brief outline below for completeness. Each simulation follows the evolution of a small region of universe chosen so as to encompass the mass an $L_{*}$ galaxy system. region is chosen from a periodic and resimulated higher preserving tidal fields from whole box. The simulation includes the gravitational effects of dark matter, gas and stars, and follows the evolution of component using Smooth Hydrodynamics technique [@steinmetz96]. We following cosmological parameters for the $\Lambda$CDM $\sigma_8=0.9$, $\Omega_{\Lambda}=0.7$, $\Omega_{\rm CDM}=0.255$, $\Omega_{\rm b}=0.045$, with no in the power spectrum. All re-simulations start at $z_{\rm init}=50$, have force resolution of order $1$ and the mass resolution is chosen so that each galaxy is represented on average, at $\sim 50,000$ dark matter/gas Gas is turned stars rates with empirical Schmidt-like of @kennicutt98. Because of this, star formation proceeds efficiently only in regions at the center of dark halos, and the stellar primary satellite galaxies are segregated spatially from the matter. re-simulation follows a single galaxy detail, well number smaller, self-bound systems of gas, and dark matter we call generically “satellites”. We main galaxy indistinctly as “primary” or “host”. The satellites span a range of luminosities, down |
We identify satellite galaxiEs in a suite Of eigHt sImuLaTionS of tHe formation of $L_*$ GAlaxIes in the $\Lambda$CDM scenaRio. ThIs SErieS HaS been PresentED bY aBadI, NAvArrO & STEiNmetz (2006), And Follow tHe same numeRicAl Scheme originALlY introduceD by @SteinmetzandNavArro02. ThE “pRimARy” galAxiEs in tHese siMUlatioNs have beeN aNAlyzed IN detail IN SeVeraL recent papers, whicH ThE Interested readEr may wIsH To CONsuLt fOr details [@aBaDi03a; @abADi03b; @meza03; @MEzA05; @NAVarRO04]. We give a brief Outline beloW For CompleTeNesS.
each siMulatIoN FolLows the evolUtioN of a small Region OF the uniVErse choSen so aS to EncOmpaSS tHe MasS oF An $L_{*}$ GAlAxy SYstEm. This reGiOn Is choSen fROM A LargE peRiodIc box And resimulateD at HighER reSolutIon prEserViNg the Tidal fIelds FrOm the whole box. ThE simUlation inCluDeS thE gRavitATional EffEctS of dark Matter, gAS anD sTARS, aNd follows the hydrodYnAMIcAl evolutIon of tHE gAsEOus compoNeNt uSing THE SmooTh PaRTiCle HydroDynamiCS (SpH) TechniqUe [@SteinmEtZ96]. We AdoPt the FOlloWing coSmologicAl parAMeters for the $\LaMBda$CDM scenariO: $h_0=65$ kM/S/mpC, $\SigmA_8=0.9$, $\OmEga_{\Lambda}=0.7$, $\OmEga_{\rM cDM}=0.255$, $\OMega_{\RM b}=0.045$, WitH No tilT in thE pRImORdial power spectrum.
ALl Re-simuLatioNs start at redsHift $z_{\rm iniT}=50$, HAVe force rEsolUTiON of order $1$ kpc, and The maSs resolutiON is choseN so thAt each gaLaxy is repRESented on AveRagE, at $Z=0$, wiTH $\SiM 50,000$ dark matter/gaS PArtiClEs. Gas is TurNed into StaRs aT raTes CoNsistent wIth the emPiRiCaL SChmIdt-liKE law of @keNnIcuTt98. becAuse oF This, stAr forMatiOn PrOCeeDs efficIEnTLY onlY iN hIgh-dEnsItY regiOns aT The Center oF dark haloS, anD The sTeLlAr compoNents of primarY aNd satellitE gAlaXies arE STrongly sEgregated spatially from tHE dark maTteR.
Each Re-siMulation fOllOws a siNglE $\Sim L_*$ gaLaxy in DetaiL, aNd rESOlves AS WeLl a NuMber of smalLER, seLf-bouNd SystEms of stArs, gas, and dark matteR We sHall call generIcaLly “sATElLitES”. WE ShaLl HEreAFTer refer to the maIn galaxy inDiSTiNctly as “priMAry” Or “Host”. The ResolveD sateLLites spAn a range oF luminosiTiEs, doWN |
We identify satellite gal axies in a suit e o f e ig ht s imul ations of thef orma tion of $L_*$ galaxies in t he $\La m bd a$CDM scenar i o. T his s er ies h a sbeenpre sentedby Abadi,Nav ar ro & Steinme t z(2006), an d f ollow the sa menumeri ca l s c hemeori ginal ly int r oduced by @stei nm e tzandn a varro02 . Th e “p rimary” galaxiesi nt hese simulatio ns hav eb ee n ana lyz ed in deta il in s e veral r e ce n t pap e rs, which the interested rea der ma ywis h to co nsult f o r d etails [@ab adi0 3a; @abad i03b;@ meza03; @meza05 ; @nav arr o04 ]. W e g iv e a b r ief ou tli n e b elow for c om plete ness . E achsim ulat ion f ollows the ev olu tion ofa sma ll re gion o f the unive rse c ho sen so as to en comp ass the m ass o f a n$L_{* } $ gala xysys tem. Th is regi o n i sc h o se n from a large per io d i cbox andresimu l at ed at highe rres olut i o n pre serv i ng the tid al fie l ds f rom the w hole b ox . T hesimul a tion inclu des thegravi t ational effect s of dark matt e r, g as andsta rs, and fol lows thehydr o dy nam i cal e volut io n o f the gaseous compon en t usin g the Smooth Parti cle Hydrod y n a mics (SP H) t e ch n ique [@steinme tz96] . We adopt the foll owing cosmolo gical par a m eters fo r t he$\L amb d a $C DM scenario:$ H _0=6 5$ km/s/M pc, $\sigm a_8 =0. 9$, $\ Om ega_{\Lam bda}=0.7 $, $ \O me ga_ {\rmC DM}=0.25 5$ , $ \O meg a_{\r m b}=0. 045$, wit hno til t in th e p r i mord ia lpowe r s pe ctrum .
A l l r e-simul ations st art at r ed sh ift $z_ {\rm init}=50 $, have forc eres olutio n of order $1$ kpc, and the massr esoluti onis ch osen so thateac h gala xyi s repr esente d onav era g e , at$ z =0 $,wi th $\sim 5 0 , 000 $ dar kmatt er/gasparticles. Gas ist urn ed into stars at rat e s c ons i st e ntwi t h t h e empirical Schm idt-like l aw of @kennicut t 98. B ecauseof this , sta r format ion proce eds effic ie ntly o nly in high-d ensity r egions at the c e nt er of da rk hal os , a nd th e stel l arcompo nentsof prima ry an dsatellit e galaxies are strongly segre gated sp atially f rom the dark mat ter.
Each re- sim ula tionfol l ows a sin g le $\ s im L_ *$ g a laxy in d e ta il, a nd resolves a s w ell a nu mbe r of sm alle r, self-bound sys t ems of stars,gas, a nddar k mat te r we shall cal l g en e r ically “ sa tellites”.We shall h e reaft er ref er tothe mai n ga l axy in dist inc tly as “p rim ar y ” or “h os t” . The r esol ve d sate llites span a range of lumino sitie s , down |
We identify_satellite galaxies_in a suite of_eight simulations_of_the formation_of_$L_*$ galaxies in_the $\Lambda$CDM scenario._This series has been_presented by Abadi,_Navarro_& Steinmetz (2006), and follow the same numerical scheme originally introduced by @steinmetzandnavarro02. The_“primary”_galaxies in_these_simulations_have been analyzed in detail_in several recent papers, which_the interested_reader may wish to consult for details [@abadi03a;_@abadi03b;_@meza03; @meza05; @navarro04]._We give a brief outline below for completeness.
Each simulation_follows the evolution of a small_region of the_universe_chosen_so as to encompass_the mass of an $L_{*}$ galaxy_system. This region is chosen from_a large periodic box and resimulated at_higher resolution preserving the tidal fields_from the whole box. The_simulation includes_the gravitational effects of dark_matter, gas and_stars, and_follows the hydrodynamical_evolution of the gaseous component using_the Smooth Particle_Hydrodynamics (SPH) technique [@steinmetz96]. We adopt_the_following cosmological parameters_for_the_$\Lambda$CDM scenario:_$H_0=65$ km/s/Mpc, $\sigma_8=0.9$,_$\Omega_{\Lambda}=0.7$,_$\Omega_{\rm CDM}=0.255$,_$\Omega_{\rm_b}=0.045$, with no tilt in the_primordial_power spectrum.
All re-simulations start at redshift $z_{\rm_init}=50$, have force resolution_of_order $1$ kpc, and_the mass resolution is chosen_so that each galaxy is represented_on average,_at $z=0$,_with $\sim 50,000$ dark matter/gas particles. Gas is turned into stars_at rates consistent with the empirical_Schmidt-like law of @kennicutt98._Because of_this,_star formation proceeds_efficiently_only in_high-density regions at the center of dark_halos, and_the stellar components of primary and_satellite galaxies are strongly_segregated_spatially from the dark matter.
Each re-simulation_follows a single $\sim L_*$ galaxy_in detail, and resolves as_well_a_number of smaller, self-bound systems_of stars, gas, and dark matter_we shall call_generically “satellites”. We shall hereafter refer to_the_main galaxy indistinctly as “primary” or_“host”._The resolved satellites span a range_of_luminosities,_down |
\in \mathcal{D}_{m+1}(U)$ as the current $\p T(\omega):= T(d \omega)$. The **support** $supp \: T $ of a current $T$ is the complement of the union of all open sets $W$ such that $T (\omega) = 0$ for $\omega \in \mathcal{D}^n(U)$ with $supp \; \omega \subset W$. For any open $W \subset U$ and $T \in \mathcal{D}_{m}(U)$ we write $T \llcorner W$ for the current in $\mathcal{D}_{m}(W)$ we get from **restricting** $T$ to $\mathcal{D}_m(W)$. For any compactly supported current $T \in \mathcal{D}_{m}(U)$ we define its **push-forward** $f_\sharp T$ by $f_\sharp T(\omega):= T(f^*\omega)$, where $f^*\omega$ denotes the usual pull-back of the $m$-form $\omega$.\
We let ${\bf{M}}_U (T) := \sup_{|\omega| \le 1, \mbox{{\tiny{supp}}} \omega \subset U}T(\omega)$ be the **mass** of the current $T$. To define the **flat (pseudo-)metric** on $\mathcal{D}_{m}(U)$ we consider open subsets $W \subset \overline{W} \subset U\subset \mathbb{R}^n$. Then $$\label{fll}
\db_W(C_1,C_2):=\mbox{inf}\{{\bf{M}}_W (S)+{\bf{M}}_W (R)\, |\, C_1 - C_2 = S + \p R, S \in \mathcal{D}_m(U), R \in \mathcal{D}_{m+1}(U)) \}.$$ The family of these (pseudo-)metrics $\db_W$ generate the **flat norm topology** on $\mathcal{D}_m(U)$.\
Finally, we define some important subclasses of currents. We call a current $T \in \mathcal{D}_m(U)$ **integer multiplicity rectifiable** or **rectifiable** for short, if for any $\ve >0$ and any compact set $K \subset U$ there exists a compactly supported $m$- | \in \mathcal{D}_{m+1}(U)$ as the current $ \p T(\omega):= T(d \omega)$. The * * support * * $ supp \: T $ of a current $ T$ is the complement of the union of all open set $ W$ such that $ metric ton (\omega) = 0 $ for $ \omega \in \mathcal{D}^n(U)$ with $ supp \; \omega \subset W$. For any open $ W \subset U$ and $ T \in \mathcal{D}_{m}(U)$ we compose $ T \llcorner W$ for the current in $ \mathcal{D}_{m}(W)$ we get down from * * restricting * * $ T$ to $ \mathcal{D}_m(W)$. For any compactly support current $ T \in \mathcal{D}_{m}(U)$ we define its * * push button - forward * * $ f_\sharp T$ by $ f_\sharp T(\omega):= T(f^*\omega)$, where $ f^*\omega$ denotes the usual puff - binding of the $ m$-form $ \omega$.\
We let $ { \bf{M}}_U (T): = \sup_{|\omega| \le 1, \mbox{{\tiny{supp } } } \omega \subset U}T(\omega)$ be the * * mass * * of the current $ T$. To define the * * flat (pseudo-)metric * * on $ \mathcal{D}_{m}(U)$ we consider exposed subsets $ W \subset \overline{W } \subset U\subset \mathbb{R}^n$. Then $ $ \label{fll }
\db_W(C_1,C_2):=\mbox{inf}\{{\bf{M}}_W (S)+{\bf{M}}_W (R)\, |\, C_1 - C_2 = S + \p R, S \in \mathcal{D}_m(U), R \in \mathcal{D}_{m+1}(U) ) \}.$$ The family of these (pseudo-)metrics $ \db_W$ beget the * * flat norm topology * * on $ \mathcal{D}_m(U)$.\
Finally, we specify some important subclasses of currents. We call a current $ T \in \mathcal{D}_m(U)$ * * integer multiplicity rectifiable * * or * * rectifiable * * for light, if for any $ \ve > 0 $ and any compact set $ K \subset U$ there exists a compactly supported $ m$- | \in \mathcal{D}_{m+1}(U)$ as the currekt $\p T(\omega):= T(d \omega)$. Thx **suppodt** $supp \: T $ of a current $T$ is the colpoemenu of the union of aul open svts $W$ sucy thet $T (\omega) = 0$ for $\omega \ik \matggal{D}^n(B)$ xith $supp \; \omegs \subset W$. For any open $F \ruyset U$ and $T \in \mathcal{D}_{m}(U)$ we write $T \llcotnfr W$ for the correnu ig $\mafhcal{D}_{m}(W)$ we get from **restricting** $T$ to $\matical{D}_m(W)$. For any compactly supported currejt $T \in \mathcal{D}_{m}(U)$ we define its **pusr-dorward** $f_\shafp T$ by $f_\sharp T(\omega):= J(f^*\omega)$, where $f^*\omega$ denotes the usuak pull-back od tjg $m$-form $\omege$.\
We len ${\bf{M}}_U (T) := \sup_{|\omega| \le 1, \mbox{{\tony{supp}}} \omega \xubvet U}T(\omega)$ be the **mass** mf the current $T$. Tj define dhz **flat (pseudo-)metric** ob $\nathcdl{D}_{m}(G)$ we xonridtr ppsn subdeta $W \subset \overline{W} \wubset U\subset \mathnb{W}^b$. Then $$\label{fml}
\db_W(C_1,S_2):=\mfox{inf}\{{\bf{M}}_W (S)+{\bf{M}}_W (R)\, |\, C_1 - C_2 = S + \p R, S \in \mdthdal{D}_m(U), R \in \mathcal{D}_{m+1}(U)) \}.$$ The family of these (kseudo-)metrycs $\db_W$ generate the **flat norm topology** on $\mathcan{D}_m(U)$.\
Fmnxllv, we dddije some important subclasses of currents. We cwml a current $T \in \iathcal{D}_m(U)$ **onhebgr multipliciti rectiyjagle** or **rectifiable** for shjrt, id for any $\ve >0$ and any compact set $K \subswt U$ there eqistw a compactly supplrted $m$- | \in \mathcal{D}_{m+1}(U)$ as the current $\p T(\omega):= The $supp \: $ of a of union of all sets $W$ such $T (\omega) = 0$ for $\omega \mathcal{D}^n(U)$ with $supp \; \omega \subset W$. For any open $W \subset U$ $T \in \mathcal{D}_{m}(U)$ we write $T \llcorner W$ for the current in $\mathcal{D}_{m}(W)$ get **restricting** to For any compactly supported current $T \in \mathcal{D}_{m}(U)$ we define its **push-forward** $f_\sharp T$ by $f_\sharp T(f^*\omega)$, where $f^*\omega$ denotes the usual pull-back of $m$-form $\omega$.\ We let (T) := \sup_{|\omega| \le 1, \omega U}T(\omega)$ be **mass** the $T$. To define **flat (pseudo-)metric** on $\mathcal{D}_{m}(U)$ we consider open subsets $W \subset \overline{W} \subset U\subset \mathbb{R}^n$. Then $$\label{fll} \db_W(C_1,C_2):=\mbox{inf}\{{\bf{M}}_W (R)\, |\, C_2 = + R, \in \mathcal{D}_m(U), R \}.$$ The family of these (pseudo-)metrics **flat norm topology** on $\mathcal{D}_m(U)$.\ Finally, we define important subclasses currents. We call a current $T \mathcal{D}_m(U)$ **integer multiplicity rectifiable** or **rectifiable** for short, for any $\ve >0$ and any compact set $K \subset U$ there exists a compactly | \in \mathcal{D}_{m+1}(U)$ as the current $\p t(\omega):= T(d \omEga)$. ThE **suPpoRt** $Supp \: t $ of a Current $T$ is the cOMpleMent of the union of all opeN sets $w$ sUCh thAT $T (\Omega) = 0$ For $\omegA \In \MAThcAl{d}^n(u)$ wiTh $SUpP \; \omegA \suBset W$. FoR any open $W \sUbsEt u$ and $T \in \mathcAL{D}_{M}(U)$ we write $T \LlcOrner W$ for the CurRent in $\MaThcAL{D}_{m}(W)$ wE geT from **RestriCTing** $T$ tO $\mathcal{D}_M(W)$. fOr any cOMpactly SUPpOrteD current $T \in \mathcaL{d}_{m}(u)$ We define its **pusH-forwaRd** $F_\ShARP T$ bY $f_\sHarp T(\omega):= t(f^*\Omega)$, WHere $f^*\omEGa$ DENOteS The usual pull-bAck of the $m$-foRM $\omEga$.\
We lEt ${\Bf{M}}_u (t) := \sup_{|\omEga| \le 1, \MbOX{{\tiNy{supp}}} \omega \SubsEt U}T(\omega)$ Be the **mASs** of the CUrrent $T$. to defiNe tHe **fLat (pSEuDo-)MetRiC** On $\mAThCal{d}_{M}(U)$ wE consideR oPeN subsEts $W \SUBSEt \ovErlIne{W} \SubseT U\subset \mathbB{R}^n$. then $$\LAbeL{fll}
\dB_W(C_1,C_2):=\mBox{iNf}\{{\Bf{M}}_W (S)+{\Bf{M}}_W (R)\, |\, C_1 - c_2 = S + \p R, S \In \Mathcal{D}_m(U), R \in \maThcaL{D}_{m+1}(U)) \}.$$ The faMilY oF thEsE (pseuDO-)metriCs $\dB_W$ gEnerate The **flat NOrm ToPOLOgY** on $\mathcal{D}_m(U)$.\
FinalLy, WE DeFine some ImportANt SuBClasses oF cUrrEnts. wE Call a CurrENt $t \in \mathcAl{D}_m(U)$ **iNTeGeR multipLiCity reCtIfiAblE** or **reCTifiAble** foR short, if For anY $\Ve >0$ and any compacT Set $K \subset U$ thERe EXIsTS a coMpaCtly supportEd $m$- | \in \mathcal{D}_{m+1}(U)$ as the cu rrent $\ p T (\ omeg a):= T(d \omega)$. The**support** $supp \: T $ of a curr e nt $T$is thec om p l eme nt o f t he un ion o f a ll open sets $W$suc hthat $T (\om e ga ) = 0$ for $\ omega \in \m ath cal{D} ^n (U) $ with $s upp \ ; \ome g a \sub set W$. F or any op e n $W \s u b se t U$ and $T \in \mat h ca l {D}_{m}(U)$ we write $ T \ l l cor ner W$ for th ecurre n t in $\ m at h c a l{D } _{m}(W)$ we g et from **r e str icting ** $T $ to $\ mathc al { D}_ m(W)$. Foranycompactly suppo r ted cur r ent $T \in \ mat hca l{D} _ {m }( U)$ w e de f in e i t s * *push-fo rw ar d** $ f_\s h a r p T$by$f_\ sharp T(\omega):=T(f ^*\o m ega )$, w here$f^* \o mega$ denot es th eusual pull-back ofthe $m$-f orm $ \om eg a$.\W e let${\ bf{ M}}_U ( T) := \ s up_ {| \ o m eg a| \le 1, \mbox{{\ ti n y {s upp}}} \ omega\ su bs e t U}T(\o me ga) $ be t he ** mass * *of the c urrent $T $. To def in e the** fla t ( pseud o -)me tric** on $\ma thcal { D}_{m}(U)$ wec onsider opens ub s e ts $W \ sub set \overli ne{W } \su bset U\ sub s et \m athbb {R } ^n $ . Then $$\label{fll }\db_W( C_1,C _2):=\mbox{in f}\{{\bf{M } } _ W (S)+{\ bf{M } }_ W (R)\, |\, C_1 - C_ 2 = S + \p R, S \in \mat hcal{D}_ m(U), R \ i n \mathca l{D }_{ m+1 }(U ) ) \ }.$$ The fami l y ofth ese (ps eud o-)metr ics $\ db_ W$ge nerate th e **flat n or mto pol ogy** on $\mat hc al{ D} _m( U)$.\ Finall y, we def in es ome import a nt s ubcl as se s of cu rr ents. Wec all a curr ent $T \i n \ m athc al {D }_m(U)$ **integer mu lt iplicity r ec tif iable* * or **rec tifiable** for short, i f for an y $ \ve > 0$ a nd any co mpa ct set $K \subse t U$ t hereex ist s a com p a ct lysu pported $m $ - | \in_\mathcal{D}_{m+1}(U)$ as_the current $\p T(\omega):=_T(d \omega)$._The_**support** $supp_\:_T $ of_a current $T$_is the complement of_the union of_all_open sets $W$ such that $T (\omega) = 0$ for $\omega \in \mathcal{D}^n(U)$ with_$supp_\; \omega_\subset_W$._For any open $W \subset_U$ and $T \in_\mathcal{D}_{m}(U)$ we_write $T \llcorner W$ for the current in_$\mathcal{D}_{m}(W)$_we get from_**restricting** $T$ to $\mathcal{D}_m(W)$. For any compactly supported current_$T \in \mathcal{D}_{m}(U)$ we define_its **push-forward** $f_\sharp_T$_by_$f_\sharp T(\omega):= T(f^*\omega)$, where_$f^*\omega$ denotes the usual pull-back of_the $m$-form $\omega$.\
We let ${\bf{M}}_U (T)_:= \sup_{|\omega| \le 1, \mbox{{\tiny{supp}}} \omega \subset_U}T(\omega)$ be the **mass** of the_current $T$. To define the_**flat (pseudo-)metric**_on $\mathcal{D}_{m}(U)$ we consider open_subsets $W \subset_\overline{W} \subset_U\subset \mathbb{R}^n$. Then_$$\label{fll}
\db_W(C_1,C_2):=\mbox{inf}\{{\bf{M}}_W (S)+{\bf{M}}_W (R)\, |\, C_1 -_C_2 = S_+ \p R, S \in \mathcal{D}_m(U),_R_\in \mathcal{D}_{m+1}(U)) \}.$$_The_family_of these_(pseudo-)metrics $\db_W$ generate_the_**flat norm_topology**_on $\mathcal{D}_m(U)$.\
Finally, we define some important_subclasses_of currents. We call a current $T_\in \mathcal{D}_m(U)$ **integer multiplicity_rectifiable**_or **rectifiable** for short,_if for any $\ve >0$_and any compact set $K \subset_U$ there_exists a_compactly supported $m$- |
-by-name, the pioneering paper [@pag41] by Plotkin already verifies completeness of the $\lambda$-calculus in Thm. 6 (p. 153). The proof is by simulation relations. It is extended by de Groote to the call-by-name $\lambda\mu$-calculus [@owr18]. Fujita establishes completeness by the inverse translation [@cmv29][@eiq93]. He manages to deal with reduction using an idea similar to ours. Also Hofmann-Streicher [@iba58] and Selinger [@fck86] show completeness with respect to categorical models.
1ex
Acknowledgement {#acknowledgement.unnumbered}
===============
We are grateful to Kazunori Tobisawa for letting the author notice the importance of Cor. \[lly89\], especially in the context of the coding of cooperative multitasking \[sdp73\].
Properties of the type system for the target calculus {#ada92}
=====================================================
We collect several results of the intersection type system for the target calculus defined in §\[ehd63\]. The argument follows the standard one that uses filter domains [@jix62][@ftj70]. So we omit proof mostly.
We need subject reduction/expansion for the type system in Def. \[amg62\]. The results in the literature are not directly applicable, though our system is a subsystem of [@jix62] if we forget about sorts. The reason is that the types of terms are constrained relative to their sorts. For example, a type of term $K$ must be of the form $\neg\underline\sigma$, i.e, $\underline\sigma\rightarrow\dbot$. The right hand of the arrow is restricted to $\dbot$. We must ensure that the subject reduction/expansion remain to hold under the constraint. We extend the standard arguments to many-sorted languages.
We extend the types in Def. \[rle89\] by introducing $\underline\tau\;\mathrel{::=}\;\bigcap\tau$.
3ex
\[jbv06\] A [*$T$-filter*]{} $d$ is a non-empty set of types $\underline\tau$ satisfying the ordinary filter conditions: (i) $\underline\tau\cap\underline\tau'\in d$ whenever $\underline\tau,\underline\tau'\in d$;(ii) $\underline\tau | -by - name, the pioneering paper [ @pag41 ] by Plotkin already verifies completeness of the $ \lambda$-calculus in Thm. 6 (p. 153). The proof is by model relative. It is extended by de Groote to the call - by - name $ \lambda\mu$-calculus [ @owr18 ]. Fujita establish completeness by the inverse translation [ @cmv29][@eiq93 ]. He manages to distribute with decrease using an idea similar to ours. Also Hofmann - Streicher [ @iba58 ] and Selinger [ @fck86 ] show completeness with regard to categorical models.
1ex
Acknowledgement { # acknowledgement.unnumbered }
= = = = = = = = = = = = = = =
We are grateful to Kazunori Tobisawa for lease the writer notice the importance of Cor. \[lly89\ ], especially in the context of the coding of cooperative multitasking \[sdp73\ ].
Properties of the character system for the target calculus { # ada92 }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
We gather several results of the overlap type organization for the target tartar defined in § \[ehd63\ ]. The argument follows the standard one that uses filter domain [ @jix62][@ftj70 ]. So we omit proof mostly.
We need capable decrease / expansion for the character system in Def. \[amg62\ ]. The results in the literature are not directly applicable, though our system is a subsystem of [ @jix62 ] if we forget about sorts. The reason is that the types of footing are restrain relative to their sort. For case, a type of term $ K$ must be of the form $ \neg\underline\sigma$, i.e, $ \underline\sigma\rightarrow\dbot$. The correct bridge player of the arrow is restricted to $ \dbot$. We must ensure that the capable reduction / expansion remain to hold under the constraint. We extend the standard arguments to many - sorted languages.
We extend the types in Def. \[rle89\ ] by introducing $ \underline\tau\;\mathrel{::=}\;\bigcap\tau$.
3ex
\[jbv06\ ] A [ * $ T$-filter * ] { } $ d$ is a non - empty set of character $ \underline\tau$ satisfy the ordinary filter conditions: (i) $ \underline\tau\cap\underline\tau'\in d$ whenever $ \underline\tau,\underline\tau'\in d$;(ii) $ \underline\tau | -by-nwme, the pioneering paper [@pag41] by Plotkin alreadb verifjes compueteness of the $\lambda$-calculns ib Thm. 6 (p. 153). The proof is by simjlation rvlations. Ut iw extended ug de Groote to bhe cclo-by-name $\lambda\ku$-calculus [@owr18]. Fujita esdaclnshes completeness by the inverse trwnslatipn [@cmv29][@eiq93]. He manwges eo dsal with reduction using an idea sjmilar uo ours. Also Hofmamn-Streicher [@iba58] and Selingfr [@ffk86] show completeneds with resksct ro categoricxl models.
1eq
Ccknowledgejent {#acknowledgement.unnumbered}
===============
Wd are grateful jo Kweunori Tobisewa fog letting the author totice yhe importance of Coe. \[lly89\], especially in thx context of the codyng of compzrative multitasking \[sep73\].
Propetties of gye gypt sbstsm for thx target camculus {#ada92}
=====================================================
Ww collect several rtsujnx of the intsrsectyog type system for the target calculus dtfines in §\[ehd63\]. The argument fillows the standard oje that ufes filter domains [@jix62][@ftj70]. So we omit proof mostly.
Fe nexd suymcct fwdkction/expansion for the type system in Def. \[amg62\]. Fht rvsults in the litcrature are not diteftki applicable, tfough our system is a subsydtem of [@jix62] uf we fordet sbout sorts. The reason is tyat the typef of terms are conscrained relacive tp theor sorts. For example, a cype or term $K$ mudt be of ffe form $\neg\underuinv\sigka$, i.e, $\underline\sigma\rightawrow\dbot$. Vhe rnght hana of the awrow is redtricbad to $\dbot$. We must ensute thad the subjfct reduction/expansion remain to hold under tne cotstraint. We ewtend the standwrd arguments jo many-soxted lxnguages.
We extend the types yn Def. \[rle89\] by hjtroducing $\uiderline\twu\;\marhreo{::=}\;\bigcap\gxu$.
3ex
\[jbv06\] A [*$T$-filyer*]{} $d$ is c non-wmpty set of types $\unasrline\tau$ satisyvibg the ordinary fiutew bonvitiogv: (i) $\underlina\tau\zap\jmderlkne\tau'\in d$ chdnevrr $\underline\tau,\undernine\fau'\in d$;(ii) $\underlinr\twu | -by-name, the pioneering paper [@pag41] by Plotkin completeness the $\lambda$-calculus Thm. 6 (p. simulation It is extended de Groote to call-by-name $\lambda\mu$-calculus [@owr18]. Fujita establishes completeness the inverse translation [@cmv29][@eiq93]. He manages to deal with reduction using an idea to ours. Also Hofmann-Streicher [@iba58] and Selinger [@fck86] show completeness with respect to models. Acknowledgement =============== are grateful to Kazunori Tobisawa for letting the author notice the importance of Cor. \[lly89\], especially the context of the coding of cooperative multitasking Properties of the type for the target calculus {#ada92} We several results the type for the target defined in §\[ehd63\]. The argument follows the standard one that uses filter domains [@jix62][@ftj70]. So we omit mostly. We reduction/expansion for type in \[amg62\]. The results literature are not directly applicable, though a subsystem of [@jix62] if we forget about The reason that the types of terms are relative to their sorts. For example, a type term $K$ must be of the form $\neg\underline\sigma$, i.e, $\underline\sigma\rightarrow\dbot$. The right hand of the restricted to $\dbot$. We ensure that the reduction/expansion to under constraint. We the standard arguments to many-sorted languages. We extend the types in \[rle89\] by introducing $\underline\tau\;\mathrel{::=}\;\bigcap\tau$. 3ex \[jbv06\] A [*$T$-filter*]{} $d$ is set types $\underline\tau$ satisfying ordinary filter conditions: (i) d$ $\underline\tau,\underline\tau'\in d$;(ii) $\underline\tau | -by-name, the pioneering paper [@pAg41] by PlotkiN alreAdy VerIfIes cOmplEteness of the $\laMBda$-cAlculus in Thm. 6 (p. 153). The proof iS by siMuLAtioN ReLatioNs. It is eXTeNDEd bY dE GRooTe TO tHe calL-by-Name $\lamBda\mu$-calcuLus [@OwR18]. Fujita estabLIsHes completEneSs by the inverSe tRanslaTiOn [@cMV29][@eiq93]. HE maNages To deal WIth redUction usiNg AN idea sIMilar to OURs. also hofmann-Streicher [@iBA58] aND Selinger [@fck86] shOw compLeTEnESS wiTh rEspect to caTeGoricAL models.
1EX
ACKNOwlEDgement {#acknowLedgement.unNUmbEred}
===============
We ArE grATeful tO KazuNoRI ToBisawa for leTtinG the authoR noticE The impoRTance of cor. \[lly89\], EspEciAlly IN tHe ConTeXT of THe CodINg oF cooperaTiVe MultiTaskING \[SDp73\].
PrOpeRtieS of thE type system foR thE tarGEt cAlculUs {#ada92}
=====================================================
we coLlEct seVeral rEsultS oF the intersectioN typE system foR thE tArgEt CalcuLUs defiNed In §\[eHd63\]. The arGument fOLloWs THE StAndard one that uses fIlTER dOmains [@jiX62][@ftj70]. So WE oMiT Proof mosTlY.
We Need SUBject ReduCTiOn/expansIon for THe TyPe systeM iN Def. \[amG62\]. THe rEsuLts in THe liTeratuRe are not DirecTLy applicable, thOUgh our system iS A sUBSySTem oF [@jiX62] if we forget AbouT SortS. The REaSon IS that The tyPeS Of TErms are constrained rElAtive tO theiR sorts. For examPle, a type of TERM $K$ must be Of thE FoRM $\neg\underline\sIgma$, i.E, $\underline\SIgma\righTarroW\dbot$. The Right hand OF The arrow Is rEstRicTed TO $\DbOt$. We must ensurE THat tHe Subject RedUction/eXpaNsiOn rEmaIn To hold undEr the conStRaInT. WE exTend tHE standarD aRguMeNts To manY-Sorted LanguAges.
we ExTEnd The typeS In dEF. \[rle89\] By InTrodUciNg $\UnderLine\TAu\;\mAthrel{::=}\;\bIgcap\tau$.
3eX
\[jbV06\] a [*$T$-fiLtEr*]{} $D$ is a non-Empty set of typEs $\Underline\tAu$ SatIsfyinG THe ordinaRy filter conditions: (i) $\undeRLine\tau\Cap\UnderLine\Tau'\in d$ wheNevEr $\undeRliNE\tau,\unDerlinE\tau'\iN d$;(Ii) $\uNDErlinE\TAu | -by-name, the pioneering p aper [@pag 41] b y P lot ki n al read y verifies com p lete ness of the $\lambda$- calcu lu s inT hm . 6 ( p. 153) . T h e pr oo fisby si mulat ion relati ons. It is ex te nded by de G r oo te to thecal l-by-name $\ lam bda\mu $- cal c ulus[@o wr18] . Fuji t a esta blishes c om p letene s s by th e in vers e translation [@c m v2 9 ][@eiq93]. Hemanage st od e alwit h reductio nusing an idea si m i l art o ours. AlsoHofmann-Str e ich er [@i ba 58] and Se linge r[ @fc k86] show c ompl eteness w ith re s pect to categor ical m ode ls.
1e x
Ac kno wl e dge m en t { # ack nowledge me nt .unnu mber e d } ==== === ==== ====
We are grate ful toK azu noriTobis awafo r let ting t he au th or notice the i mpor tance ofCor .\[l ly 89\], especi all y i n the c ontexto f t he c o di ng of cooperativemu l t it asking \ [sdp73 \ ].
P ropertie softhet y pe sy stem fo r the ta rget c a lc ul us {#ad a9 2}
=== == === === ===== = ==== ====== ======== ===== = ============
W e collect sev e ra l re s ults of the inters ecti o n ty pe s y st emf or th e tar ge t c a lculus defined in § \[ ehd63\ ]. Th e argument fo llows thes t a ndard on e th a tu ses filter dom ains[@jix62][@ f tj70]. S o weomit pro of mostly .
We need su bje ctred u c ti on/expansionf o r th etype sy ste m in De f.\[a mg6 2\] .The resul ts in th eli te ra tur e are not dire ct lyap pli cable , thoug h our sys te mi s a subsys t em o f [@ ji x6 2] i f w eforge t ab o utsorts.The reaso n i s tha tth e types of terms are c onstrained r ela tive t o their so rts. For example, a typ e of ter m $ K$ mu st b e of thefor m $\ne g\u n derlin e\sigm a$, i .e , $ \ u nderl i n e\ sig ma \rightarro w \ dbo t$. T he rig ht hand of the arrow is r e str icted to $\db ot$ . We m us t e n su r e t ha t th e subject reducti on/expansi on re main to ho l d u nd er theconstra int.W e exten d the sta ndard arg um ents t o m any-sorted languag es.
We e x tendt he type s i n Def. \ [rl e89\] by in t rod ucing $\und er line\t au\;\ ma threl{:: =}\;\bigcap\tau$.
3ex
\[jbv 06\]A [ *$T$-filt er* ] {}$d$ is anon- empty setoftyp es $\ und e rline \tau $ s ati s fying the ordinaryf il ter c on ditions: (i ) $ \un derli ne\ t au\cap \und erline\tau'\in d$ whenever $\und erli n e \ta u,\ u nder li ne\tau'\in d$; (ii )$ \ underlin e\ tau | -by-name, the_pioneering paper_[@pag41] by Plotkin already_verifies completeness_of_the $\lambda$-calculus_in_Thm. 6 (p. 153). The_proof is by_simulation relations. It is_extended by de Groote_to_the call-by-name $\lambda\mu$-calculus [@owr18]. Fujita establishes completeness by the inverse translation [@cmv29][@eiq93]. He manages_to_deal with_reduction_using_an idea similar to ours._Also Hofmann-Streicher [@iba58] and Selinger_[@fck86] show_completeness with respect to categorical models.
1ex
Acknowledgement {#acknowledgement.unnumbered}
===============
We are_grateful_to Kazunori Tobisawa_for letting the author notice the importance of Cor. \[lly89\],_especially in the context of the_coding of cooperative_multitasking \[sdp73\].
Properties_of_the type system for_the target calculus {#ada92}
=====================================================
We collect several_results of the intersection type system_for the target calculus defined in §\[ehd63\]._The argument follows the standard one_that uses filter domains [@jix62][@ftj70]._So we_omit proof mostly.
We need subject_reduction/expansion for the_type system_in Def. \[amg62\]. The_results in the literature are not_directly applicable, though_our system is a subsystem of_[@jix62]_if we forget_about_sorts._The reason_is that the_types_of terms_are_constrained relative to their sorts. For_example,_a type of term $K$ must be_of the form $\neg\underline\sigma$,_i.e,_$\underline\sigma\rightarrow\dbot$. The right hand_of the arrow is restricted_to $\dbot$. We must ensure that_the subject_reduction/expansion remain_to hold under the constraint. We extend the standard arguments to_many-sorted languages.
We extend the types in_Def. \[rle89\] by introducing $\underline\tau\;\mathrel{::=}\;\bigcap\tau$.
3ex
\[jbv06\]_A [*$T$-filter*]{}_$d$_is a non-empty_set_of types_$\underline\tau$ satisfying the ordinary filter conditions: (i)_$\underline\tau\cap\underline\tau'\in d$_whenever $\underline\tau,\underline\tau'\in d$;(ii) $\underline\tau |
Metelmann and A. A. Clerk, Quantum-limited amplification via reservoir engineering, Phys. Rev. Lett. **112**, 133904 (2014).
L. Mercier de Lépinay, E. Damskägg, C. F. Ockeloen-Korppi, M. A. Sillanpää, Realization of directional amplification in a microwave optomechanical device, arXiv:1811.06036.
A. Nunnenkamp, V. Sudhir, A. K. Feofanov, A. Roulet, and T. J. Kippenberg, Quantum-limited amplification and parametric instability in the reversed dissipation regime of cavity optomechanics, Phys. Rev. Lett. **113**, 023604 (2014).
S. Manipatruni, J. T. Robinson, and M. Lipson, Optical nonreciprocity in optomechanical structures, Phys. Rev. Lett. **102**, 213903 (2009).
A. S. Zheng, G. Y. Zhang, H. Y. Chen, T. T. Mei and J. B. Liu, Nonreciprocal light propagation in coupled microcavities system beyond weak-excitation approximation, Sci. Rep. **7**, 14001 (2017).
L. N. Song, Z. H. Wang, and Y. Li, Enhancing optical nonreciprocity by an atomic ensemble in two coupled cavities, Opt. Commun. **415**, 39-42 (2018).
S. C. Zhang, Y. Q. Hu, G. W. Lin, Y. P. Niu, K. Y. Xia, J. B. Gong and S. Q. Gong, Thermal-motion-induced non-reciprocal quantum optical system, Nat. Photon. **12**, 744 (2018).
F. Ruesink, M.-A. Miri, A. Alù, and E. Verhagen, Nonreciprocity and magnetic-free isolation based on optomechanical interactions, Nat. Commun. **7**, 13662 (2016).
S. R. K. Rodriguez, V. Goblot, N. Carlon Zambon, A. Amo, and J. Bloch, Nonreciprocity and zero reflection in nonlinear cavities with tailored loss, Phys. Rev. A **99**, 013850 ( | Metelmann and A. A. Clerk, Quantum - limited amplification via reservoir engineering, Phys. Rev. Lett. * * 112 * *, 133904 (2014).
L. Mercier de Lépinay, E. Damskägg, C. F. Ockeloen - Korppi, M. A. Sillanpää, Realization of directional amplification in a microwave optomechanical device, arXiv:1811.06036.
A. Nunnenkamp, V. Sudhir, A. K. Feofanov, A. Roulet, and T. J. Kippenberg, Quantum - limited amplification and parametric instability in the converse profligacy regime of cavity optomechanics, Phys. Rev. Lett. * * 113 * *, 023604 (2014).
S. Manipatruni, J. T. Robinson, and M. Lipson, ocular nonreciprocity in optomechanical structure, Phys. Rev. Lett. * * 102 * *, 213903 (2009).
A. S. Zheng, G. Y. Zhang, H. Y. Chen, T. T. Mei and J. B. Liu, Nonreciprocal light propagation in copulate microcavities organization beyond weak - excitement estimate, Sci. Rep. * * 7 * *, 14001 (2017).
L. N. Song, Z. H. Wang, and Y. Li, Enhancing optical nonreciprocity by an nuclear ensemble in two coupled cavity, Opt. Commun. * * 415 * *, 39 - 42 (2018).
S. C. Zhang, Y. Q. Hu, G. W. Lin, Y. P. Niu, K. Y. Xia, J. B. Gong and S. Q. Gong, Thermal - apparent motion - induced non - reciprocal quantum optical system, Nat. Photon. * * 12 * *, 744 (2018).
F. Ruesink, M.-A. Miri, A. Alù, and E. Verhagen, Nonreciprocity and magnetic - free isolation establish on optomechanical interactions, Nat. Commun. * * 7 * *, 13662 (2016).
S. R. K. Rodriguez, V. Goblot, N. Carlon Zambon, A. Amo, and J. Bloch, Nonreciprocity and zero reflection in nonlinear cavities with sew loss, Phys. Rev. A * * 99 * *, 013850 ( | Mehelmann and A. A. Clerk, Quxntum-limited amkluficatmon via reservokr engineering, Phys. Rev. Lett. **112**, 133904 (2014).
L. Meecier de Lépinay, E. Damsyägg, C. F. Obkeloen-Koeppi, N. A. Sillan'ää, Realization kn dirzcvional amplificstion in a microwave optmmdckanical device, arXiv:1811.06036.
A. Nunnenkamp, V. Stdhir, A. K. Feofanov, A. Roolet, sgd T. J. Kippenberg, Quantum-limited ampliricatioi and parametriv instability in the reverded fissipation regime of cavity iptoiwchanics, Phyr. Rev. Lett. **113**, 023604 (2014).
S. Manipattuni, J. T. Robinson, and M. Lipson, Ooticak nonreciptodihi in optomecianicaj structures, Phys. Ree. Lett. **102**, 213903 (2009).
A. S. Zheng, G. N. Zhaig, H. Y. Chen, T. T. Mei and J. B. Liu, Nonreciprocwl light [rkpagation in coupoee mictocavhtier syrtej ueyknd wewk-eecitation alproximatiob, Sci. Rep. **7**, 14001 (2017).
L. N. Sonb, S. H. Wang, and Y. Li, Enragcing optical nonreciprocity by an atompc ehsemble in two coupled xavities, Opt. Commun. **415**, 39-42 (2018).
S. C. Zhand, Y. Q. Hu, G. W. Lin, Y. P. Niu, K. Y. Xia, J. B. Gong and S. Q. Gong, Ghexnal-moguoj-induced non-reciprocal quantum optical system, Hau. Pmoton. **12**, 744 (2018).
F. Ruesinh, M.-A. Miri, A. Apù, sgd E. Verhagen, Nonreejpdocity and magnetif-free ifolatuon based on pptomechanical interactions, Nat. Commun. **7**, 13662 (2016).
S. R. K. Rodriguez, V. Goblot, N. Ccrlon Eambon, A. Amo, and J. Bloch, Nonrzciprodity and zego reflecfkon in nonlinear canitias with uxilored loss, Phys. Rev. A **99**, 013850 ( | Metelmann and A. A. Clerk, Quantum-limited amplification engineering, Rev. Lett. 133904 (2014). L. C. Ockeloen-Korppi, M. A. Realization of directional in a microwave optomechanical device, arXiv:1811.06036. Nunnenkamp, V. Sudhir, A. K. Feofanov, A. Roulet, and T. J. Kippenberg, Quantum-limited and parametric instability in the reversed dissipation regime of cavity optomechanics, Phys. Rev. **113**, (2014). Manipatruni, T. Robinson, and M. Lipson, Optical nonreciprocity in optomechanical structures, Phys. Rev. Lett. **102**, 213903 (2009). S. Zheng, G. Y. Zhang, H. Y. Chen, T. Mei and J. Liu, Nonreciprocal light propagation in microcavities beyond weak-excitation Sci. **7**, (2017). L. N. Z. H. Wang, and Y. Li, Enhancing optical nonreciprocity by an atomic ensemble in two coupled cavities, Commun. **415**, S. C. Y. Hu, W. Lin, Y. K. Y. Xia, J. B. Gong Gong, Thermal-motion-induced non-reciprocal quantum optical system, Nat. Photon. 744 (2018). Ruesink, M.-A. Miri, A. Alù, and Verhagen, Nonreciprocity and magnetic-free isolation based on optomechanical Nat. Commun. **7**, 13662 (2016). S. R. K. Rodriguez, V. Goblot, N. Carlon Zambon, A. J. Bloch, Nonreciprocity and reflection in nonlinear with loss, Rev. **99**, 013850 | Metelmann and A. A. Clerk, QuantuM-limited amPlifiCatIon ViA resErvoIr engineering, PHYs. ReV. Lett. **112**, 133904 (2014).
L. Mercier de Lépinay, e. DamsKäGG, C. F. OCKeLoen-KOrppi, M. A. sIlLANpäÄ, REaLizAtIOn Of dirEctIonal amPlificatioN in A mIcrowave optoMEcHanical devIce, ArXiv:1811.06036.
A. NunnenKamP, V. SudhIr, a. K. FEOfanoV, A. ROulet, And T. J. KIPpenbeRg, Quantum-LiMIted amPLificatION aNd paRametric instabiliTY iN The reversed disSipatiOn REgIME of CavIty optomecHaNics, PHYs. Rev. LeTT. **113**, 023604 (2014).
S. mANIpaTRuni, J. T. RobinsoN, and M. Lipson, oPtiCal nonReCipROcity iN optoMeCHanIcal structuRes, PHys. Rev. LetT. **102**, 213903 (2009).
A. S. ZheNG, G. Y. ZhanG, h. Y. Chen, T. t. Mei anD J. B. liu, nonrECiPrOcaL lIGht PRoPagATioN in couplEd MiCrocaVitiES SYStem BeyOnd wEak-exCitation approXimAtioN, sci. rep. **7**, 14001 (2017).
L. N. song, Z. h. WanG, aNd Y. Li, enhancIng opTiCal nonreciprociTy by An atomic eNseMbLe iN tWo couPLed cavItiEs, OPt. CommuN. **415**, 39-42 (2018).
S. C. ZhanG, y. Q. HU, G. w. lIN, Y. p. Niu, K. Y. Xia, J. B. Gong and s. Q. gONg, thermal-mOtion-iNDuCeD Non-recipRoCal QuanTUM optiCal sYStEm, Nat. PhoTon. **12**, 744 (2018).
F. RuESiNk, m.-A. Miri, A. alÙ, and E. VErHagEn, NOnrecIProcIty and Magnetic-Free iSOlation based on OPtomechanical INtERAcTIons, nat. commun. **7**, 13662 (2016).
S. R. K. RoDrigUEz, V. GObloT, n. CArlON ZambOn, A. AmO, aND J. bLoch, Nonreciprocity aNd Zero reFlectIon in nonlineaR cavities wITH Tailored Loss, pHyS. rev. A **99**, 013850 ( | Metelmann and A. A. Clerk , Quantum- limit edamp li fica tion via reservoir engi neering, Phys. Rev. Le tt. * *1 1 2**, 13 3904(2014).
L . Mer ci er de L é pi nay,E.Damskäg g, C. F. O cke lo en-Korppi, M . A . Sillanpä ä,Realizationofdirect io nal ampli fic ation in am icrowa ve optome ch a nicald evice,a r Xi v:18 11.06036.
A. Nun n en k amp, V. Sudhir , A. K .F eo f a nov , A . Roulet,an d T.J . Kippe n be r g , Qu a ntum-limitedamplificati o n a nd par am etr i c inst abili ty inthe reverse d di ssipation regim e of cav i ty opto mechan ics , P hys. Re v. Le tt . ** 1 13 **, 023 604 (201 4) .
S. M anip a t r u ni,J.T. R obins on, and M. Li pso n, O p tic al no nreci proc it y inoptome chani ca l structures, P hys. Rev. Let t.** 102 ** , 213 9 03 (20 09) .
A. S. Z heng, G . Y. Z h a n g, H. Y. Chen, T. T. M e i a nd J. B. Liu,N on re c iprocalli ght pro p a gatio n in co upled mi crocav i ti es system b eyondwe ak- exc itati o n ap proxim ation, S ci. R e p. **7**, 1400 1 (2017).
L.N .S o ng , Z.H.Wang, and Y . Li , Enh anci n gopt i cal n onrec ip r oc i ty by an atomic ens em ble in twocoupled cavit ies, Opt.C o m mun. **4 15** , 3 9 -42 (2018).
S . C.Zhang, Y.Q . Hu, G. W. L in, Y. P . Niu, K. Y . Xia, J . B . G ong an d S. Q. Gong, The r m al-m ot ion-ind uce d non-r eci pro cal qu an tum optic al syste m, N at .Pho ton.* *12**, 7 44 (2 01 8).
F.R uesink , M.- A. M ir i, A.Alù, an d E . Verh ag en , No nre ci proci ty a n d m agnetic -free iso lat i on b as ed on opt omechanical i nt eractions, N at. Commu n . **7**,13662 (2016).
S. R. K. Rodrigu ez, V. G oblo t, N. Car lon Zambo n,A . Amo, and J . Blo ch , N o n recip r o ci tyan d zero ref l e cti on in n onli near ca vities with tailor e d l oss, Phys. Re v.A ** 9 9 ** , 0 1 38 5 0 ( | Metelmann_and A._A. Clerk, Quantum-limited amplification_via reservoir_engineering,_Phys. Rev._Lett._**112**, 133904 (2014).
L._Mercier de Lépinay,_E. Damskägg, C. F._Ockeloen-Korppi, M. A._Sillanpää,_Realization of directional amplification in a microwave optomechanical device, arXiv:1811.06036.
A. Nunnenkamp, V. Sudhir, A._K._Feofanov, A._Roulet,_and_T. J. Kippenberg, Quantum-limited amplification_and parametric instability in the_reversed dissipation_regime of cavity optomechanics, Phys. Rev. Lett. **113**,_023604_(2014).
S. Manipatruni, J._T. Robinson, and M. Lipson, Optical nonreciprocity in optomechanical_structures, Phys. Rev. Lett. **102**, 213903_(2009).
A. S. Zheng,_G._Y._Zhang, H. Y. Chen,_T. T. Mei and J. B._Liu, Nonreciprocal light propagation in coupled_microcavities system beyond weak-excitation approximation, Sci. Rep._**7**, 14001 (2017).
L. N. Song, Z._H. Wang, and Y. Li,_Enhancing optical_nonreciprocity by an atomic ensemble_in two coupled_cavities, Opt._Commun. **415**, 39-42_(2018).
S. C. Zhang, Y. Q. Hu,_G. W. Lin,_Y. P. Niu, K. Y. Xia,_J._B. Gong and_S._Q._Gong, Thermal-motion-induced_non-reciprocal quantum optical_system,_Nat. Photon._**12**,_744 (2018).
F. Ruesink, M.-A. Miri, A._Alù,_and E. Verhagen, Nonreciprocity and magnetic-free isolation_based on optomechanical interactions,_Nat._Commun. **7**, 13662 (2016).
S._R. K. Rodriguez, V. Goblot,_N. Carlon Zambon, A. Amo, and_J. Bloch,_Nonreciprocity and_zero reflection in nonlinear cavities with tailored loss, Phys. Rev. A_**99**, 013850 ( |
ta- hypernuclei. For anti-flavour (positive strangeness, beauty or negative charm) the same formula as above holds, but with certain changes for the hyperfine splitting constants, $c_F \to c_{\bar F}$ and $\bar c_F \to \bar c_{\bar F}$ in the last term $\Delta M_{HFS} $. $c_{\bar F}$ ($\bar c_{\bar F}$) is obtained from $c_F$ ($\bar c_F$) by means of substitution $\mu\to -\mu$: $$c_{\bar F} =1-{\Theta_I\over 2\mu \Theta_F}(\mu +1); \qquad
\bar c_{\bar F} =1+{\Theta_I\over \mu^2\Theta_F}(\mu +1).$$ This change is crucially important for the link between rotator and bound state models of the $SU(3)$ quantization [@ksh2], but often it was not made in the literature. The mass of the $\Theta^+$ hyperon within this approach equals to about $1570\,MeV$ ($e=4.1, \; F_K/F_\pi =1.22$).
$ A$ $\bar \omega_s$ $ \epsilon^{tot}_s $ $\bar \omega_c$ $\epsilon^{tot}_c$ $\bar \omega_b$ $\epsilon^{tot}_b$
------ ----------------- ---------------------- ----------------- -------------------- ----------------- --------------------
$1$ $591$ — $1750$ — $4940$ —
$3$ $564$ $76$ $1710$ $46$ $4890$ $57$
$5$ $558$ $108$ $1710$ $71$ $4880$ $82$
$7$ $559$ $120$ $1710$ $85$ $4880$ $100$
$9$ $550$ $152$ $1710$ $100$ $4900$ $100$
$11$ $547$ $173$ $1710$ $115$ $4900$ $110$
$13$ $546$ $196$ $1720 | ta- hypernuclei. For anti - flavour (positive strangeness, beauty or negative appeal) the like formula as above holds, but with sealed changes for the hyperfine splitting constant, $ c_F \to c_{\bar F}$ and $ \bar c_F \to \bar c_{\bar F}$ in the last term $ \Delta M_{HFS } $. $ c_{\bar F}$ ($ \bar c_{\bar F}$) is obtain from $ c_F$ ($ \bar c_F$) by means of substitution $ \mu\to -\mu$: $ $ c_{\bar F } = 1-{\Theta_I\over 2\mu \Theta_F}(\mu +1); \qquad
\bar c_{\bar F } = 1+{\Theta_I\over \mu^2\Theta_F}(\mu +1).$$ This variety is crucially authoritative for the link between rotator and bind state models of the $ SU(3)$ quantization [ @ksh2 ], but much it was not make in the literature. The mass of the $ \Theta^+$ hyperon within this approach equals to approximately $ 1570\,MeV$ ($ e=4.1, \; F_K / F_\pi = 1.22 $).
$ A$ $ \bar \omega_s$ $ \epsilon^{tot}_s $ $ \bar \omega_c$ $ \epsilon^{tot}_c$ $ \bar \omega_b$ $ \epsilon^{tot}_b$
------ ----------------- ---------------------- ----------------- -------------------- ----------------- --------------------
$ 1 $ $ 591 $ — $ 1750 $ — $ 4940 $ —
$ 3 $ $ 564 $ $ 76 $ $ 1710 $ $ 46 $ $ 4890 $ $ 57 $
$ 5 $ $ 558 $ $ 108 $ $ 1710 $ $ 71 $ $ 4880 $ $ 82 $
$ 7 $ $ 559 $ $ 120 $ $ 1710 $ $ 85 $ $ 4880 $ $ 100 $
$ 9 $ $ 550 $ $ 152 $ $ 1710 $ $ 100 $ $ 4900 $ $ 100 $
$ 11 $ $ 547 $ $ 173 $ $ 1710 $ $ 115 $ $ 4900 $ $ 110 $
$ 13 $ $ 546 $ $ 196 $ $ 1720 | ta- jypernuclei. For anti-flavuur (positive sttabgenesv, beaufy or neeative charm) the same formule as abovt holds, but with ceftain chajges for the yyperfine splittinn conabants, $r_F \to c_{\bar F}$ anc $\bar c_F \tm \bar c_{\bar F}$ it ghz last term $\Delta M_{HFS} $. $c_{\bar F}$ ($\bar c_{\far F}$) ix lbtained from $s_F$ ($\bsw c_F$) by means of substitution $\mu\to -\mu$: $$d_{\bar F} =1-{\Uheta_I\over 2\mu \Thets_F}(\mu +1); \qquad
\bar c_{\bar F} =1+{\Thetw_I\ovfr \mu^2\Theta_F}(\mu +1).$$ Thid change is crusually importxnt for tht nink betwegn rotator and bound state modelr of che $SU(3)$ quanjnzatlmn [@ksh2], but iften it was not made in dhe litrrature. The maxs mf rhe $\Theta^+$ hyperon witiin this approach eqoals to abmuc $1570\,MeV$ ($e=4.1, \; F_K/F_\pi =1.22$).
$ A$ $\bqr \omgga_s$ $ \eowilun^{tkt}_x $ $\bar \omxga_c$ $\epsimon^{tot}_c$ $\bqr \omega_b$ $\epsilon^{uot}_f$
------ ----------------- ---------------------- ----------------- -------------------- ----------------- --------------------
$1$ $591$ — $1750$ — $4940$ —
$3$ $564$ $76$ $1710$ $46$ $4890$ $57$
$5$ $558$ $108$ $1710$ $71$ $4880$ $82$
$7$ $559$ $120$ $1710$ $85$ $4880$ $100$
$9$ $550$ $152$ $1710$ $100$ $4900$ $100$
$11$ $547$ $173$ $1710$ $115$ $4900$ $110$
$13$ $546$ $196$ $1720 | ta- hypernuclei. For anti-flavour (positive strangeness, beauty charm) same formula above holds, but hyperfine constants, $c_F \to F}$ and $\bar \to \bar c_{\bar F}$ in the term $\Delta M_{HFS} $. $c_{\bar F}$ ($\bar c_{\bar F}$) is obtained from $c_F$ c_F$) by means of substitution $\mu\to -\mu$: $$c_{\bar F} =1-{\Theta_I\over 2\mu \Theta_F}(\mu +1); \bar F} \mu^2\Theta_F}(\mu This change is crucially important for the link between rotator and bound state models of the quantization [@ksh2], but often it was not made the literature. The mass the $\Theta^+$ hyperon within this equals about $1570\,MeV$ \; =1.22$). A$ $\bar \omega_s$ \epsilon^{tot}_s $ $\bar \omega_c$ $\epsilon^{tot}_c$ $\bar \omega_b$ $\epsilon^{tot}_b$ ------ ----------------- ---------------------- ----------------- -------------------- ----------------- -------------------- $1$ $591$ $1750$ — $3$ $564$ $1710$ $4890$ $5$ $558$ $108$ $4880$ $82$ $7$ $559$ $120$ $1710$ $9$ $550$ $152$ $1710$ $100$ $4900$ $100$ $11$ $173$ $1710$ $4900$ $110$ $13$ $546$ $196$ $1720 | ta- hypernuclei. For anti-flavoUr (positive StranGenEss, BeAuty Or neGative charm) the SAme fOrmula as above holds, but wIth ceRtAIn chANgEs for The hypeRFiNE SplItTiNg cOnSTaNts, $c_F \To c_{\Bar F}$ and $\Bar c_F \to \bar C_{\baR F}$ In the last terM $\deLta M_{HFS} $. $c_{\baR F}$ ($\bAr c_{\bar F}$) is obtAinEd from $C_F$ ($\Bar C_f$) by meAns Of subStitutIOn $\mu\to -\Mu$: $$c_{\bar F} =1-{\ThEtA_i\over 2\mU \theta_F}(\mU +1); \QQuAd
\baR c_{\bar F} =1+{\Theta_I\over \mU^2\thETa_F}(\mu +1).$$ This changE is cruCiALlY IMpoRtaNt for the liNk BetweEN rotatoR AnD BOUnd STate models of tHe $SU(3)$ quantizATioN [@ksh2], buT oFteN It was nOt madE iN The Literature. THe maSs of the $\ThEta^+$ hypERon withIN this apProach EquAls To abOUt $1570\,mev$ ($e=4.1, \; F_k/F_\PI =1.22$).
$ A$ $\bAR \oMegA_S$ $ \epSilon^{tot}_S $ $\bAr \Omega_C$ $\epsILON^{Tot}_c$ $\Bar \OmegA_b$ $\epsIlon^{tot}_b$
------ ----------------- ---------------------- ----------------- -------------------- ----------------- --------------------
$1$ $591$ — $1750$ — $4940$ —
$3$ $564$ $76$ $1710$ $46$ $4890$ $57$
$5$ $558$ $108$ $1710$ $71$ $4880$ $82$
$7$ $559$ $120$ $1710$ $85$ $4880$ $100$
$9$ $550$ $152$ $1710$ $100$ $4900$ $100$
$11$ $547$ $173$ $1710$ $115$ $4900$ $110$
$13$ $546$ $196$ $1720 | ta- hypernuclei. For anti- flavour (p ositi vestr an gene ss,beauty or nega t ivecharm) the same formul a asab o ve h o ld s, bu t withc er t a inch an ges f o rthe h ype rfine s plitting c ons ta nts, $c_F \t o c _{\bar F}$ an d $\bar c_F\to \barc_ {\b a r F}$ in thelast t e rm $\D elta M_{H FS } $. $c _ {\bar F } $ ( $\ba r c_{\bar F}$) is ob t ained from $c_ F$ ($\ ba r c _ F $)bymeans of s ub stitu t ion $\m u \t o - \mu $ : $$c_{\bar F } =1-{\Thet a _I\ over 2 \m u \ T heta_F }(\mu + 1 );\qquad
\bar c_{ \bar F} = 1+{\Th e ta_I\ov e r \mu^2 \Theta _F} (\m u +1 ) .$ $Thi sc han g eisc ruc ially im po rt ant f or t h e l inkbet ween rota tor and bound st atem ode ls of the$SU( 3) $ qua ntizat ion [ @k sh2], but often itwas not m ade i n t he lite r ature. Th e m ass ofthe $\T h eta ^+ $ h yp eron within this a pp r o ac h equals to ab o ut $ 1 570\,MeV $($e =4.1 , \; F_ K/F_ \ pi =1.22$) .
$ A$ $\bar\o mega_s $ $ \e psilo n ^{to t}_s $ $\bar \ome g a_c$ $\epsil o n^{tot}_c$ $ \b a r \ o mega _b$ $\epsilo n^{t o t}_b $
- -- --- ----- ----- -- - -- - - ----------------- -- --- -- ----- ---------- -- ---------- - - - ----- -- ---- - -- - ------- ------ ----- --------- $1$ $591$ — $17 5 0 $ — $4940$ — $3$ $564$ $ 76$ $1 710$ $ 46$ $4 8 9 0$ $ 57 $
$ 5$ $5 58$ $10 8 $ $1710$ $71 $ $48 80$ $82$
$ 7 $ $5 59$ $120$ $ 1710$ $ 85$ $488 0 $ $1 00$
$ 9$ $550$ $152$ $1 7 10 $ $100$ $49 00 $ $ 100 $ $11$ $547$ $173$ $ 1 7 10$ $115$ $ 49 00$ $ 110 $
$ 13$ $ 546 $ $196$ $1720 | ta- hypernuclei._For anti-flavour_(positive strangeness, beauty or_negative charm)_the_same formula_as_above holds, but_with certain changes_for the hyperfine splitting_constants, $c_F \to_c_{\bar_F}$ and $\bar c_F \to \bar c_{\bar F}$ in the last term $\Delta M_{HFS}_$._$c_{\bar F}$_($\bar_c_{\bar_F}$) is obtained from $c_F$_($\bar c_F$) by means of_substitution $\mu\to_-\mu$: $$c_{\bar F} =1-{\Theta_I\over 2\mu \Theta_F}(\mu +1); \qquad
\bar_c_{\bar_F} =1+{\Theta_I\over \mu^2\Theta_F}(\mu_+1).$$ This change is crucially important for the link_between rotator and bound state models_of the $SU(3)$_quantization_[@ksh2],_but often it was_not made in the literature. The_mass of the $\Theta^+$ hyperon within_this approach equals to about $1570\,MeV$ ($e=4.1,_\; F_K/F_\pi =1.22$).
$ A$_ $\bar \omega_s$ _ $_\epsilon^{tot}_s $ $\bar_\omega_c$ _$\epsilon^{tot}_c$ _ $\bar \omega_b$_ $\epsilon^{tot}_b$
------ -----------------_---------------------- ----------------- --------------------_----------------- --------------------
$1$ __$591$ ___ _ __ _—_ __ _ __ $1750$ _ _ —_ _ _ _ $4940$ _ _ __ _—
_ $3$_ $564$ _ _ _ $76$ __ _ _ $1710$ ___ _ $46$ _ _ __ $4890$ __ ___$57$
$5$ _ $558$ _ $108$_ _ ___ $1710$ _ _ _ $71$ _ _ $4880$ _ $82$
$7$_ $559$ _ __ $120$ ____ _ _ _ _ $1710$ _ _ __ $85$ _ $4880$ _ __ _$100$
$9$ _ $550$ _ $152$ __ _ _ _ _ _ _$1710$ _ _ __$100$ ____ _ $4900$ _ _ __ $100$
$11$_ $547$_ _ $173$ _ $1710$_ _ _ __ $115$ _ _ __ $4900$ __$110$
_ $13$ __$546$ _____ $196$ _ __ _ _$1720 |
s in Mathematics Vol. 227, Springer-Verlag, New York, 2004.
M. Reineke, Quivers, desingularizations and canonical bases. Studies in memory of Issai Schur (Chevaleret/Rehovot, 2000), 325–344, Progr. Math., 210, Birkhäuser Boston, Boston, MA, 2003. [math.AG/0104284]{}
C. M. Ringel, Representations of $K$-species and bimodules, J. Algebra 41 (1976), no. 2, 269–302.
C. M. Ringel, Tame algebras and integral quadratic forms,\
Lecture Notes in Mathematics, 1099. Springer-Verlag, Berlin, 1984.
J. J. Rotman, An introduction to homological algebra, Pure and Applied Mathematics, 85. Academic Press, Inc. (Harcourt Brace Jovanovich, Publishers), New York-London, 1979.
R. Thomason, Une formule de Lefschetz en $K$-théorie équivariante algébrique,\
Duke Math. J. 68 (1992), no. 3, 447–462.
S. Zelikson, Auslander-Reiten quivers and the Coxeter complex, Algebr. Represent. Theory 8 (2005), no. 1, 35–55. [math.QA/0208098]{}
G. Zwara, Unibranch orbit closures in module varieties, Ann. Scient. Éc. Norm. Sup., 4e série, t. 35, 2002, p. 877–895.
[^1]: This condition on the weights is not as restrictive as it looks. If $Z\subseteq Y$ is invariant under rescaling (i.e. is the affine cone over a projective variety), then we can extend the action of $T$ to $T\times {{\mathbb G}_m}$ where ${{\mathbb G}_m}$ acts by dilation, and now all the weights live in $T^* \times \{1\}$. If $Z$ is not already rescaling-invariant, we can replace it by the limit subscheme $Z' := \lim_{t\to 0} t\cdot Z$, and compute the more refined multidegree $[Z'] \in {\mathrm{Sym}}^\bullet((T\times {{\mathbb | s in Mathematics Vol. 227, Springer - Verlag, New York, 2004.
M. Reineke, Quivers, desingularizations and canonical bases. Studies in memory of Issai Schur (Chevaleret / Rehovot, 2000), 325–344, Progr. Math. , 210, Birkhäuser Boston, Boston, MA, 2003. [ math. AG/0104284 ] { }
C. M. Ringel, Representations of $ K$-species and bimodules, J. Algebra 41 (1976), no. 2, 269–302.
C. M. Ringel, Tame algebra and built-in quadratic equation forms,\
Lecture Notes in Mathematics, 1099. Springer - Verlag, Berlin, 1984.
J. J. Rotman, An introduction to homological algebra, Pure and Applied Mathematics, 85. Academic Press, Inc. (Harcourt Brace Jovanovich, Publishers), New York - London, 1979.
R. Thomason, Une formule de Lefschetz en $ K$-théorie équivariante algébrique,\
Duke Math. J. 68 (1992), no. 3, 447–462.
S. Zelikson, Auslander - Reiten shaking and the Coxeter complex, Algebr. Represent. Theory 8 (2005), no. 1, 35–55. [ mathematics. QA/0208098 ] { }
G. Zwara, Unibranch orbit closures in module assortment, Ann. Scient. Éc. Norm. Sup. , 4e série, t. 35, 2002, p. 877–895.
[ ^1 ]: This condition on the weight is not as restrictive as it looks. If $ Z\subseteq Y$ is changeless under rescaling (i.e. is the affine cone over a projective variety), then we can extend the action of $ T$ to $ T\times { { \mathbb G}_m}$ where $ { { \mathbb G}_m}$ acts by dilation, and now all the weights live in $ T^ * \times \{1\}$. If $ Z$ is not already rescale - invariant, we can replace it by the limit subscheme $ omega': = \lim_{t\to 0 } t\cdot Z$, and compute the more refined multidegree $ [ Z' ] \in { \mathrm{Sym}}^\bullet((T\times { { \mathbb | s ij Mathematics Vol. 227, Sprinner-Verlag, New Yotk, 2004.
M. Reiieke, Qujvers, deringularizations and canonicel bqses. Wtudies in memory of Irsai Schug (Chevaleeet/Rthovot, 2000), 325–344, Progr. Mavg., 210, Birkmäbser Glstou, Uoston, MA, 2003. [math.SG/0104284]{}
C. M. Ringal, Representathovs of $K$-species and bimodules, J. Algebrw 41 (1976), no. 2, 269–302.
C. M. Ringel, Tame algtbrws ahd integral quadratic forms,\
Lecture Notes pn Mathematics, 1099. Slringer-Verlag, Berlin, 1984.
J. J. Rltmaj, An introduction ho homologixal wogebra, Pure xnd Applied Mathematica, 85. Academic Press, Inc. (Harcourt Crace Jovanovicy, Pubphshers), New Bork-Logdon, 1979.
R. Thomason, Une xormule de Lefschetz cn $K$-tiéoriw équivariante algébriqne,\
Duke Math. J. 68 (1992), no. 3, 447–462.
F. Zeliksot, Cuslander-Reiten quivees and jhe Cmxetde cumpmee, Amgebr. Gepcesent. Theody 8 (2005), no. 1, 35–55. [mqth.QA/0208098]{}
G. Zwara, Unibramcr orbit closurss in ioqule varieties, Ann. Scient. Éc. Norm. Sup., 4e véris, t. 35, 2002, p. 877–895.
[^1]: This condition on the weights is noj as restryctive as it looks. If $Z\subseteq Y$ is invariant unger rxszalnkn (i.e. us the affine cone over a projective variety), thqh ee can extend thc action of $T$ to $T\yilex {{\mathbb G}_m}$ whgre ${{\matkgb G}_m}$ acts by dilatiln, and gow aol the weyghtx live in $T^* \times \{1\}$. If $Z$ is not already eescaling-invariant, we can repuace it bu the limit subscheme $Z' := \lij_{t\to 0} t\cdot Z$, and cojoute the more rewinvd mgltidegrtd $[Z'] \in {\mathrm{Sym}}^\btllet((T\timxs {{\machbb | s in Mathematics Vol. 227, Springer-Verlag, New M. Quivers, desingularizations canonical bases. Studies (Chevaleret/Rehovot, 325–344, Progr. Math., Birkhäuser Boston, Boston, 2003. [math.AG/0104284]{} C. M. Ringel, Representations $K$-species and bimodules, J. Algebra 41 (1976), no. 2, 269–302. C. M. Ringel, algebras and integral quadratic forms,\ Lecture Notes in Mathematics, 1099. Springer-Verlag, Berlin, 1984. J. An to algebra, Pure and Applied Mathematics, 85. Academic Press, Inc. (Harcourt Brace Jovanovich, Publishers), New York-London, 1979. Thomason, Une formule de Lefschetz en $K$-théorie équivariante Duke Math. J. 68 no. 3, 447–462. S. Zelikson, quivers the Coxeter Algebr. Theory (2005), no. 1, [math.QA/0208098]{} G. Zwara, Unibranch orbit closures in module varieties, Ann. Scient. Éc. Norm. Sup., 4e série, t. 2002, p. This condition the is as restrictive as If $Z\subseteq Y$ is invariant under the affine cone over a projective variety), then can extend action of $T$ to $T\times {{\mathbb where ${{\mathbb G}_m}$ acts by dilation, and now the weights live in $T^* \times \{1\}$. If $Z$ is not already rescaling-invariant, we can by the limit subscheme := \lim_{t\to 0} Z$, compute more multidegree $[Z'] {\mathrm{Sym}}^\bullet((T\times {{\mathbb | s in Mathematics Vol. 227, Springer-verlag, New YOrk, 2004.
M. REinEke, quIverS, desIngularizationS And cAnonical bases. Studies in MemorY oF issaI scHur (ChEvalereT/reHOVot, 2000), 325–344, prOgR. MaTh., 210, bIrKhäusEr BOston, BoSton, MA, 2003. [math.aG/0104284]{}
C. m. RIngel, RepreseNTaTions of $K$-spEciEs and bimodulEs, J. algebrA 41 (1976), nO. 2, 269–302.
C. M. rIngel, tamE algeBras anD IntegrAl quadratIc FOrms,\
LeCTure NotES In mathEmatics, 1099. Springer-VeRLaG, berlin, 1984.
J. J. Rotman, an intrOdUCtION to HomOlogical alGeBra, PuRE and AppLIeD mATheMAtics, 85. Academic press, Inc. (HarCOurT Brace joVanOVich, PuBlishErS), new york-London, 1979.
R. thomAson, Une foRmule dE lefscheTZ en $K$-théOrie éqUivAriAnte ALgÉbRiqUe,\
dUke mAtH. J. 68 (1992), nO. 3, 447–462.
s. ZeLikson, AuSlAnDer-ReIten QUIVErs aNd tHe CoXeter Complex, Algebr. repReseNT. ThEory 8 (2005), nO. 1, 35–55. [math.qA/0208098]{}
G. ZWaRa, UniBranch Orbit ClOsures in module vArieTies, Ann. ScIenT. ÉC. NoRm. sup., 4e sÉRie, t. 35, 2002, p. 877–895.
[^1]: THis ConDition oN the weiGHts Is NOT As Restrictive as it looKs. iF $z\sUbseteq Y$ Is invaRIaNt UNder rescAlIng (I.e. is THE affiNe coNE oVer a projEctive VArIeTy), then wE cAn exteNd The ActIon of $t$ To $T\tImes {{\maThbb G}_m}$ whEre ${{\maTHbb G}_m}$ acts by dilATion, and now all THe WEIgHTs liVe iN $T^* \times \{1\}$. If $Z$ iS not ALreaDy reSCaLinG-InvarIant, wE cAN rEPlace it by the limit suBsCheme $Z' := \Lim_{t\tO 0} t\cdot Z$, and comPute the morE REFined mulTideGReE $[z'] \in {\mathrm{Sym}}^\buLlet((T\Times {{\mathbB | s in Mathematics Vol. 227, Springer- Verla g,New Y ork, 200 4.
M. Reineke , Qui vers, desingularizatio ns an dc anon i ca l bas es. Stu d ie s inme mo ryof Is sai S chu r (Chev aleret/Reh ovo t, 2000), 325– 3 44 , Progr. M ath ., 210, Birk häu ser Bo st on, Bosto n,MA, 2 003. [ m ath.AG /0104284] {}
C. M. Ringel, R ep rese ntations of $K$-s p ec i es and bimodul es, J. A l ge b r a 4 1 ( 1976), no. 2 , 269 – 302.
C . M . R ing e l, Tame algeb ras and int e gra l quad ra tic forms, \
Lec tu r e N otes in Mat hema tics, 109 9. Spr i nger-Ve r lag, Be rlin,198 4.
J.J .Ro tma n, Ani nt rod u cti on to ho mo lo gical alg e b r a , Pu reandAppli ed Mathematic s,85.A cad emicPress , In c. (Har courtBrace J ovanovich, Publ ishe rs), NewYor k- Lon do n, 19 7 9.
R. Th oma son, Un e formu l e d eL e f sc hetz en $K$-théori eé q ui variante algéb r iq ue , \
Duke M at h.J. 6 8 (1992 ), n o .3, 447–4 62.
S . Z el ikson,Au slande r- Rei ten quiv e rs a nd the Coxeter comp l ex, Algebr. Re p resent. Theor y 8 ( 20 0 5),no. 1, 35–55.[mat h .QA/ 0208 0 98 ]{}
G. Z wara, U n ib r anch orbit closures i n modu le va rieties, Ann. Scient. Ã ‰ c . Norm. S up., 4e série, t. 35,2002, p. 877–89 5 .
[^1]: This conditi on on the w eights i s n otasres t r ic tive as it lo o k s. I f$Z\subs ete q Y$ is in var ian t u nd er rescal ing (i.e .is t he af finec one over a pr oj ect ive v a riety) , the n we c an ext end the ac t i on o f$T $ to $T \t imes{{\m a thb b G}_m} $ where $ {{\ m athb bG} _m}$ ac ts by dilatio n, and now a ll th e weig h t s live i n $T^* \times \{1\}$.I f $Z$ i s n ot al read y rescali ng- invari ant , we ca n repl ace i tbyt h e lim i t s ubs ch eme $Z' := \ lim _{t\t o0} t \cdot Z $, and compute the mor e refined mul tid egre e $[ Z'] \i n {\ ma t hrm { S ym}}^\bullet((T \times {{\ ma t hb b | s in_Mathematics Vol._227, Springer-Verlag, New York,_2004.
M. Reineke,_Quivers,_desingularizations and_canonical_bases. Studies in_memory of Issai_Schur (Chevaleret/Rehovot, 2000), 325–344,_Progr. Math., 210,_Birkhäuser_Boston, Boston, MA, 2003. [math.AG/0104284]{}
C. M. Ringel, Representations of $K$-species and bimodules, J. Algebra_41_(1976), no._2,_269–302.
C._M. Ringel, Tame algebras and_integral quadratic forms,\
Lecture Notes in_Mathematics, 1099._Springer-Verlag, Berlin, 1984.
J. J. Rotman, An introduction to_homological_algebra, Pure and_Applied Mathematics, 85. Academic Press, Inc. (Harcourt Brace Jovanovich,_Publishers), New York-London, 1979.
R. Thomason, Une_formule de Lefschetz_en_$K$-théorie_équivariante algébrique,\
Duke Math. J._68 (1992), no. 3, 447–462.
S. Zelikson,_Auslander-Reiten quivers and the Coxeter complex,_Algebr. Represent. Theory 8 (2005), no. 1,_35–55. [math.QA/0208098]{}
G. Zwara, Unibranch orbit closures_in module varieties, Ann. Scient._Éc. Norm._Sup., 4e série, t. 35,_2002, p. 877–895.
[^1]:_This condition_on the weights_is not as restrictive as it_looks. If $Z\subseteq_Y$ is invariant under rescaling (i.e._is_the affine cone_over_a_projective variety),_then we can_extend_the action_of_$T$ to $T\times {{\mathbb G}_m}$ where_${{\mathbb_G}_m}$ acts by dilation, and now all_the weights live in_$T^*_\times \{1\}$. If_$Z$ is not already rescaling-invariant,_we can replace it by the_limit subscheme_$Z' :=_\lim_{t\to 0} t\cdot Z$, and compute the more refined multidegree $[Z']_\in {\mathrm{Sym}}^\bullet((T\times {{\mathbb |
uklm\], we obtain that $\p(c^*)\le\p\cuk(a^*)$. Since $\p(d)\le\p\cuk(b)$, we get that $(\p\cuk(a^*))^*\le (\p(c^*))^*\lle
\p(d)\le\p\cuk(b)$. Therefore, $(\p\cuk(a^*))^*\lle\p\cuk(b)$. So, (DLC3) is fulfilled.
For verifying (DLC4), let $b\in\BBBB$. Then there exists $a\in\BBBB$ such that $b\le\p(a)$. By (BC1), there exists $a_1\in\BBBB$ with $a\llx a_1$. Then $b\le\p(a)\le\p\cuk(a_1)$. Thus, $\p\cuk$ satisfies condition (DLC4).
\[assocuk\] Let $\p_i:(A_i,\rho_i,\BBBB_i)\lra
(A_{i+1},\rho_{i+1},\BBBB_{i+1})$, where $i=1,2$, be two functions between CLCAs. Then:
\(a) $(\p_2\cuk\circ\p_1)\cuk=(\p_2\circ\p_1)\cuk$;
\(b) If $\p_1$ and $\p_2$ are monotone functions, then $(\p_2\circ\p_1\cuk)\cuk=(\p_2\circ\p_1)\cuk$;
\(c) If $\p_1$ and $\p_2$ satisfy conditions (DLC1)-(DLC5) then the function $\p_2\circ\p_1$ satisfies conditions (DLC1)-(DLC4) and even condition (DLC3S).
We will write, for $i=1,2$, $``\ll_i$“ instead of $``\ll_{\rho_i}$”. We also set $\p=\p_2\circ\p_1$.
\(a) Let $a\in A_1$. Then $(\p_2\cuk\circ\p_1)\cuk(a)=\bv\{\p_ | uklm\ ], we obtain that $ \p(c^*)\le\p\cuk(a^*)$. Since $ \p(d)\le\p\cuk(b)$, we get that $ (\p\cuk(a^*))^*\le (\p(c^*))^*\lle
\p(d)\le\p\cuk(b)$. Therefore, $ (\p\cuk(a^*))^*\lle\p\cuk(b)$. So, (DLC3) is satisfy.
For verifying (DLC4), lease $ b\in\BBBB$. Then there exists $ a\in\BBBB$ such that $ b\le\p(a)$. By (BC1), there exists $ a_1\in\BBBB$ with $ a\llx a_1$. Then $ b\le\p(a)\le\p\cuk(a_1)$. Thus, $ \p\cuk$ satisfies circumstance (DLC4).
\[assocuk\ ] Let $ \p_i:(A_i,\rho_i,\BBBB_i)\lra
(A_{i+1},\rho_{i+1},\BBBB_{i+1})$, where $ i=1,2 $, be two functions between CLCAs. Then:
\(a) $ (\p_2\cuk\circ\p_1)\cuk=(\p_2\circ\p_1)\cuk$;
\(b) If $ \p_1 $ and $ \p_2 $ are monotone functions, then $ (\p_2\circ\p_1\cuk)\cuk=(\p_2\circ\p_1)\cuk$;
\(c) If $ \p_1 $ and $ \p_2 $ satisfy circumstance (DLC1)-(DLC5) then the function $ \p_2\circ\p_1 $ satisfies conditions (DLC1)-(DLC4) and even circumstance (DLC3S).
We will spell, for $ i=1,2 $, $ '` \ll_i$ “ instead of $ '` \ll_{\rho_i}$ ”. We besides set $ \p=\p_2\circ\p_1$.
\(a) Let $ a\in A_1$. Then $ (\p_2\cuk\circ\p_1)\cuk(a)=\bv\{\p _ | ukll\], we obtain that $\p(c^*)\le\p\cuy(a^*)$. Since $\p(d)\le\p\cok(v)$, we gxt that $(\p\cuk(a^*))^*\le (\p(c^*))^*\lle
\p(d)\le\p\cuk(b)$. Therefore, $(\p\cnk(a^*))^*\loe\p\cuj(b)$. So, (DLC3) is fulfilled.
Wor verifjing (DLC4), oet $u\in\BBBB$. Then thecs exists $a\in\BGNB$ sueh that $b\le\p(a)$. By (BC1), there axists $a_1\in\BBBB$ wktk $a\llx a_1$. Then $b\le\p(a)\le\p\cuk(a_1)$. Thus, $\p\cuk$ satisfoed condition (DLS4).
\[asspsuk\] Mvt $\p_i:(A_i,\rho_i,\BBBB_i)\lra
(A_{i+1},\rho_{i+1},\BBBB_{i+1})$, shere $i=1,2$, be two functoons between CLCAs. Then:
\(a) $(\p_2\fuk\clrc\p_1)\cuk=(\p_2\circ\p_1)\cuk$;
\(b) Lf $\p_1$ and $\p_2$ qre iinotone funcgions, then $(\p_2\circ\p_1\cuk)\cui=(\p_2\circ\p_1)\cuk$;
\(c) If $\p_1$ and $\p_2$ satisfy zondicions (DLC1)-(DLX5) rhej the functiin $\p_2\cprc\p_1$ satisfies condithons (DLV1)-(DLC4) and even gondivion (DLC3S).
We will write, foc $i=1,2$, $``\ll_i$“ instead of $``\lj_{\rho_i}$”. We dlao set $\p=\p_2\circ\p_1$.
\(a) Lwt $a\in D_1$. Than $(\p_2\zyk\ckrc\k_1)\cul(a)=\gv\{\p_ | uklm\], we obtain that $\p(c^*)\le\p\cuk(a^*)$. Since $\p(d)\le\p\cuk(b)$, that (\p(c^*))^*\lle \p(d)\le\p\cuk(b)$. $(\p\cuk(a^*))^*\lle\p\cuk(b)$. So, (DLC3) let Then there exists such that $b\le\p(a)$. (BC1), there exists $a_1\in\BBBB$ with $a\llx Then $b\le\p(a)\le\p\cuk(a_1)$. Thus, $\p\cuk$ satisfies condition (DLC4). \[assocuk\] Let $\p_i:(A_i,\rho_i,\BBBB_i)\lra (A_{i+1},\rho_{i+1},\BBBB_{i+1})$, where $i=1,2$, two functions between CLCAs. Then: \(a) $(\p_2\cuk\circ\p_1)\cuk=(\p_2\circ\p_1)\cuk$; \(b) If $\p_1$ and $\p_2$ are functions, $(\p_2\circ\p_1\cuk)\cuk=(\p_2\circ\p_1)\cuk$; If and $\p_2$ satisfy conditions (DLC1)-(DLC5) then the function $\p_2\circ\p_1$ satisfies conditions (DLC1)-(DLC4) and even condition (DLC3S). will write, for $i=1,2$, $``\ll_i$“ instead of $``\ll_{\rho_i}$”. also set $\p=\p_2\circ\p_1$. \(a) $a\in A_1$. Then $(\p_2\cuk\circ\p_1)\cuk(a)=\bv\{\p_ | uklm\], we obtain that $\p(c^*)\le\p\cuk(a^*)$. since $\p(d)\le\p\Cuk(b)$, wE geT thAt $(\P\cuk(A^*))^*\le (\p(C^*))^*\lle
\p(d)\le\p\cuk(b)$. THErefOre, $(\p\cuk(a^*))^*\lle\p\cuk(b)$. So, (DLC3) iS fulfIlLEd.
FoR VeRifyiNg (DLC4), leT $B\iN\bbBB$. thEn TheRe EXiSts $a\iN\BBbB$ such tHat $b\le\p(a)$. By (bC1), tHeRe exists $a_1\in\BbbB$ With $a\llx a_1$. THen $B\le\p(a)\le\p\cuk(a_1)$. thuS, $\p\cuk$ sAtIsfIEs conDitIon (DLc4).
\[assocUK\] Let $\p_i:(a_i,\rho_i,\BBBb_i)\LRa
(A_{i+1},\rhO_{I+1},\BBBB_{i+1})$, wHERe $I=1,2$, be tWo functions betweeN cLcas. Then:
\(a) $(\p_2\cuk\cirC\p_1)\cuk=(\p_2\CiRC\p_1)\CUK$;
\(b) IF $\p_1$ aNd $\p_2$ are monoToNe funCTions, thEN $(\p_2\CIRC\p_1\cUK)\cuk=(\p_2\circ\p_1)\cuk$;
\(C) If $\p_1$ and $\p_2$ satISfy ConditIoNs (Dlc1)-(DLC5) thEn the FuNCtiOn $\p_2\circ\p_1$ satIsfiEs conditiOns (DLC1)-(dlC4) and evEN conditIon (DLC3s).
We WilL wriTE, fOr $I=1,2$, $``\ll_I$“ iNSteAD oF $``\ll_{\RHo_i}$”. we also seT $\p=\P_2\cIrc\p_1$.
\(a) let $a\IN a_1$. tHen $(\p_2\Cuk\Circ\P_1)\cuk(a)=\Bv\{\p_ | uklm\], we obtain that $\p (c^*)\le\p \cuk( a^* )$. S ince $\p (d)\le\p\cuk(b ) $, w e get that $(\p\cuk(a^ *))^* \l e (\p ( c^ *))^* \lle
\p ( d) \ l e\p \c uk (b) $. Th erefo re, $(\p\c uk(a^*))^* \ll e\ p\cuk(b)$. S o ,(DLC3) isful filled.
For ve rifyin g(DL C 4), l et$b\in \BBBB$ . Thenthere exi st s $a\in \ BBBB$ s u c hthat $b\le\p(a)$. By( BC 1 ), there exist s $a_1 \i n \B B B B$wit h $a\llx a _1 $. Th e n $b\le \ p( a ) \ le\ p \cuk(a_1)$. T hus, $\p\cu k $ s atisfi es co n dition (DLC 4) .
\ [assocuk\]Let$\p_i:(A_ i,\rho _ i,\BBBB _ i)\lra(A_{i+ 1}, \rh o_{i + 1} ,\ BBB B_ { i+1 } )$ , w h ere $i=1,2$ ,be twofunc t i o n s be twe en C LCAs. Then:
\(a)$(\ p_2\ c uk\ circ\ p_1)\ cuk= (\ p_2\c irc\p_ 1)\cu k$ ;
\(b) If $\p_ 1$ a nd $\p_2$ ar emon ot one f u nction s,the n $(\p_ 2\circ\ p _1\ cu k ) \ cu k=(\p_2\circ\p_1)\ cu k $ ;
\(c) If $\p_1 $ a nd $\p_2$ s at isf y co n d ition s (D L C1 )-(DLC5) thent he f unction $ \p_2\c ir c\p _1$ sati s fies condi tions (D LC1)- ( DLC4) and even condition (DL C 3S ) .
W e wi llwrite, for$i=1 , 2$,$``\ l l_ i$“ inste ad of $ ` `\ l l_{\rho_i}$”. We al so set $ \p=\p _2\circ\p_1$.
\(a) Let $ a \in A_1$ . Th e n$ (\p_2\cuk\circ \p_1) \cuk(a)=\b v \{\p_ | uklm\], we_obtain that_$\p(c^*)\le\p\cuk(a^*)$. Since $\p(d)\le\p\cuk(b)$, we_get that_$(\p\cuk(a^*))^*\le_(\p(c^*))^*\lle
\p(d)\le\p\cuk(b)$. Therefore,_$(\p\cuk(a^*))^*\lle\p\cuk(b)$._So, (DLC3) is_fulfilled.
For verifying (DLC4),_let $b\in\BBBB$. Then there_exists $a\in\BBBB$ such_that_$b\le\p(a)$. By (BC1), there exists $a_1\in\BBBB$ with $a\llx a_1$. Then $b\le\p(a)\le\p\cuk(a_1)$. Thus, $\p\cuk$ satisfies_condition_(DLC4).
\[assocuk\] Let_$\p_i:(A_i,\rho_i,\BBBB_i)\lra
(A_{i+1},\rho_{i+1},\BBBB_{i+1})$,_where_$i=1,2$, be two functions between_CLCAs. Then:
\(a) $(\p_2\cuk\circ\p_1)\cuk=(\p_2\circ\p_1)\cuk$;
\(b) If $\p_1$_and $\p_2$_are monotone functions, then $(\p_2\circ\p_1\cuk)\cuk=(\p_2\circ\p_1)\cuk$;
\(c) If $\p_1$ and_$\p_2$_satisfy conditions (DLC1)-(DLC5)_then the function $\p_2\circ\p_1$ satisfies conditions (DLC1)-(DLC4) and even_condition (DLC3S).
We will write, for $i=1,2$,_$``\ll_i$“ instead of_$``\ll_{\rho_i}$”._We_also set $\p=\p_2\circ\p_1$.
\(a) Let_$a\in A_1$. Then $(\p_2\cuk\circ\p_1)\cuk(a)=\bv\{\p_ |
_{\theta(\omega_1),\theta(\omega_2)}$. A ResNet consists of a series of blocks. One block is given in Figure \[fig:schematic\] with two linear transformations, two activation functions, and one short cut. Detailed description of ResNet is included in the Supporting Information. Parameters of the surface roughness ($\theta(\omega_{1})$,$\theta (\omega_{2})$) are fed as input, and the EDL $\sigma$ is extracted as the output function over the entire parameter space. Sigmoid function is chosen as the activation function here.
The loss function we use is the MSE between the actual EDL $\sigma_{\theta(\omega_1),\theta(\omega_2)}$ given by the diffusion equation model and the predicted EDL $\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])$ given by the ResNet $$\label{eqn:MSE}
MSE= \frac{1}{M}\sum_{j=1}^{M} \left(\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}-\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])\right)^2,$$ where $\theta$ represents the parameter set in the ResNet, $j$ is the $j$-th sample, and M is the size of training data set.
Define the relative $L^{\infty}$ error of EDL as $$Error = \max_{1\leq j\leq M} \dfrac{\left|\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}-\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])\right|} {\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}},$$ which will be used to quantify the approximation accuracy of DL.
Quasi-Monte Carlo Sampling
--------------------------
Compared to uniform sampling and Monte-Carlo sampling, quasi-Monte Carlo sampling provides the best compromise between accuracy and efficiency. It overcomes the curse of dimensionality and has high accuracy [@Russel1998]. For the simulations in our work, at least three orders of magnitude reduction in the size of data set is found for quasi-Monte Carlo sampling without loss of accuracy (see the Supporting Information).
Results and Discussion
======================
Accuracy check and training data set
| _ { \theta(\omega_1),\theta(\omega_2)}$. A ResNet consists of a series of blocks. One block is give in Figure \[fig: schematic\ ] with two analogue transformations, two activation function, and one unretentive cut. Detailed description of ResNet is included in the Supporting Information. parameter of the airfoil roughness ($ \theta(\omega_{1})$,$\theta (\omega_{2})$) are prey as input, and the EDL $ \sigma$ is extracted as the output signal function over the entire argument space. Sigmoid function is chosen as the activation function here.
The loss function we practice is the MSE between the actual EDL $ \sigma_{\theta(\omega_1),\theta(\omega_2)}$ given by the diffusion equation mannequin and the predicted EDL $ \sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])$ given by the ResNet $ $ \label{eqn: MSE }
MSE= \frac{1}{M}\sum_{j=1}^{M } \left(\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}-\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])\right)^2,$$ where $ \theta$ represent the parameter put in the ResNet, $ j$ is the $ j$-th sample, and M is the size of training data set.
Define the proportional $ L^{\infty}$ error of EDL as $ $ Error = \max_{1\leq j\leq M } \dfrac{\left|\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}-\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])\right| } { \sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}},$$ which will be used to quantify the approximation accuracy of DL.
Quasi - Monte Carlo Sampling
--------------------------
Compared to uniform sampling and Monte - Carlo sample distribution, quasi - Monte Carlo sample distribution leave the best compromise between accuracy and efficiency. It overcomes the curse of dimensionality and has high accuracy [ @Russel1998 ]. For the simulations in our workplace, at least three order of magnitude reduction in the size of datum rig is find for quasi - Monte Carlo sampling without loss of accuracy (determine the Supporting Information).
Results and Discussion
= = = = = = = = = = = = = = = = = = = = = =
Accuracy check and training data jell | _{\theha(\omega_1),\theta(\omega_2)}$. A ResNtt consists of a series mf blodks. One clock is given in Figure \[fig:dcyematuc\] with two linear travsformatilns, two qctitation functions, and one short gut. Dzteiled descriptipn of ResNat is included iv che Supporting Information. Parameterf of thr durface roughngss ($\tnqta(\ojvgc_{1})$,$\theta (\omega_{2})$) are fed as input, ahd the TDL $\sigma$ is extravted as the output functioj ovfr the entire paraleter space. Sigiiid function is chosen as the actibation function here.
The loss fuvctiou we use is tye LVE between vhe acnual EDL $\sigma_{\theta(\omaga_1),\thets(\omega_2)}$ given bn the didfusion equation moden and the predicteq EDL $\sigka(\cheta(\omega_1)[j],\theta(\omega_2)[h])$ tiven by dhe FwsNdt $$\mauel{sqn:MSE}
MAE= \frac{1}{M}\suj_{j=1}^{M} \left(\signa_{\theta(\omega_1)[j],\theta(\okedq_2)[j]}-\sigma(\theta(\ojega_1)[j],\treea(\omega_2)[j])\right)^2,$$ where $\theta$ represents tht parzmeter set in the ResNer, $j$ is the $j$-th sample, and M is the size of training data set.
Define the relative $L^{\infvy}$ erxir of WDP as $$Error = \max_{1\leq j\leq M} \dfrac{\left|\sigma_{\thetw(\kmtga_1)[m],\theta(\omega_2)[j]}-\sigmc(\theta(\omega_1)[j],\thets(\olebw_2)[j])\right|} {\sigma_{\jheta(\omzfa_1)[n],\theta(\omega_2)[j]}},$$ which will bg used to quantyfy yhe approximation accuracy if DL.
Quasi-Moute Carlo Sampling
--------------------------
Com'ared to uniyorm ssmplimg and Monte-Carlo samplnng, quzsi-Monte Caglo sampljvg provides the cesn cokpromise between accuracy wnd efficmency. It ovefcomgs the surse of dlmenslmnality and has hihh aceuracf [@Russel1998]. Flr the simulations in our work, ev least three osdegs of maguitude reduction in the size of dcta set ns fouvd for quaai-Monte Carlo sampjing without nlss of accurecy (see tre Syppoeting Ivwormation).
Resulys and Discussion
======================
Acxuracy check and tvainivf data set
| _{\theta(\omega_1),\theta(\omega_2)}$. A ResNet consists of a series One is given Figure \[fig:schematic\] with functions, one short cut. description of ResNet included in the Supporting Information. Parameters the surface roughness ($\theta(\omega_{1})$,$\theta (\omega_{2})$) are fed as input, and the EDL $\sigma$ extracted as the output function over the entire parameter space. Sigmoid function is as activation here. loss function we use is the MSE between the actual EDL $\sigma_{\theta(\omega_1),\theta(\omega_2)}$ given by the diffusion model and the predicted EDL $\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])$ given by ResNet $$\label{eqn:MSE} MSE= \frac{1}{M}\sum_{j=1}^{M} where $\theta$ represents the parameter in ResNet, $j$ the sample, M is the of training data set. Define the relative $L^{\infty}$ error of EDL as $$Error = \max_{1\leq j\leq M} {\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}},$$ which used to the accuracy DL. Quasi-Monte Carlo Compared to uniform sampling and Monte-Carlo sampling provides the best compromise between accuracy and It overcomes curse of dimensionality and has high [@Russel1998]. For the simulations in our work, at three orders of magnitude reduction in the size of data set is found for quasi-Monte without loss of accuracy the Supporting Information). and ====================== check training data | _{\theta(\omega_1),\theta(\omega_2)}$. A ResNEt consists Of a seRieS of BlOcks. one bLock is given in FIGure \[Fig:schematic\] with two linEar trAnSFormATiOns, twO activaTIoN FUncTiOnS, anD oNE sHort cUt. DEtailed DescriptioN of reSNet is includED iN the SupporTinG Information. parAmeterS oF thE SurfaCe rOughnEss ($\theTA(\omega_{1})$,$\Theta (\omegA_{2})$) aRE fed as INput, and THE EdL $\siGma$ is extracted as tHE oUTput function ovEr the eNtIRe PARamEteR space. SigmOiD funcTIon is chOSeN AS The ACtivation funcTion here.
The LOss FunctiOn We uSE is the mSE beTwEEn tHe actual EDL $\SigmA_{\theta(\omeGa_1),\thetA(\Omega_2)}$ giVEn by the DiffusIon EquAtioN MoDeL anD tHE prEDiCteD eDL $\Sigma(\theTa(\OmEga_1)[j],\tHeta(\OMEGA_2)[j])$ giVen By thE ResNEt $$\label{eqn:MSE}
mSE= \Frac{1}{m}\Sum_{J=1}^{M} \lefT(\sigmA_{\theTa(\Omega_1)[J],\theta(\Omega_2)[J]}-\sIgma(\theta(\omega_1)[j],\ThetA(\omega_2)[j])\riGht)^2,$$ WhEre $\ThEta$ rePResentS thE paRameter Set in thE resneT, $J$ IS tHe $j$-th sample, and M is tHe SIZe Of trainiNg data SEt.
deFIne the reLaTivE $L^{\inFTY}$ erroR of Edl aS $$Error = \maX_{1\leq j\lEQ M} \DfRac{\left|\SiGma_{\theTa(\OmeGa_1)[j],\Theta(\OMega_2)[J]}-\sigma(\Theta(\omeGa_1)[j],\thETa(\omega_2)[j])\right|} {\sIGma_{\theta(\omega_1)[J],\ThETA(\oMEga_2)[j]}},$$ WhiCh will be useD to qUAntiFy thE ApProXImatiOn accUrACy OF DL.
Quasi-Monte Carlo SAmPling
--------------------------
COmparEd to uniform saMpling and MONTE-Carlo saMpliNG, qUAsi-Monte Carlo sAmpliNg provides THe best coMpromIse betweEn accuracY ANd efficiEncY. It OveRcoMES tHe curse of dimeNSIonaLiTy and haS hiGh accurAcy [@rusSel1998]. for ThE simulatiOns in our WoRk, At LeAst Three ORders of mAgNitUdE reDuctiON in the Size oF datA sEt IS foUnd for qUAsI-mOnte caRlO samPliNg WithoUt loSS of AccuracY (see the SuPpoRTing inFoRmation).
results and DisCuSsion
======================
AccurAcY chEck and TRAining daTa set
| _{\theta(\omega_1),\theta( \omega_2)} $. ARes Net c onsi stsof a series of bloc ks. One block is given in F ig u re \ [ fi g:sch ematic\ ] w i t h t wo l ine ar tr ansfo rma tions,two activa tio nfunctions, a n done shortcut . Detailed d esc riptio nofR esNet is incl uded i n the S upporting I n format i on. Par a m et ersof the surface ro u gh n ess ($\theta(\ omega_ {1 } )$ , $ \th eta (\omega_{ 2} )$) a r e fed a s i n p u t,a nd the EDL $\ sigma$ is e x tra cted a sthe output func ti o n o ver the ent ireparameter space . Sigmoi d functi on ischo sen ast he a cti va t ion fu nct i onhere.
T he l oss f unct i o n we u seis t he MS E between the ac tual EDL $\si gma_{ \the ta (\ome ga_1), \thet a( \omega_2)}$ giv en b y the dif fus io n e qu ation modeland th e predi cted ED L $\ si g m a (\ theta(\omega_1)[j] ,\ t h et a(\omega _2)[j] ) $gi v en by th eRes Net$ $ \labe l{eq n :M SE}
MS E= \fr a c{ 1} {M}\sum _{ j=1}^{ M} \l eft (\sig m a_{\ theta( \omega_1 )[j], \ theta(\omega_2 ) [j]}-\sigma(\ t he t a (\ o mega _1) [j],\theta( \ome g a_2) [j]) \ ri ght ) ^2,$$ wher e$ \t h eta$ represents the p aramet er se t in the ResN et, $j$ is t h e $j$-th sam p le , and M is thesizeof trainin g data se t.
D efine th e relativ e $L^{\inf ty} $ e rro r o f ED L as $$Error= \ma x_ {1\leqj\l eq M} \ dfr ac{ \le ft| \s igma_{\th eta(\ome ga _1 )[ j] ,\t heta( \ omega_2) [j ]}- \s igm a(\th e ta(\om ega_1 )[j] ,\ th e ta( \omega_ 2 )[ j ] )\ri gh t| } {\ sig ma _{\th eta( \ ome ga_1)[j ],\theta( \om e ga_2 )[ j] }},$$ w hich will beus ed to quan ti fythe ap p r oximatio n accuracy of DL.
Quas i -MonteCar lo Sa mpli ng
------ --- ------ --- - ------ -
Com pared t o u n i forms a mp lin gand Monte- C a rlo samp li ng,quasi-M onte Carlo samplin g pr ovides the be stcomp r o mi seb et w een a c cur a c y and efficienc y. It over co m es the curse ofdi mension ality a nd ha s high a ccuracy [ @Russel19 98 ]. F o r th e simulati ons in o ur work,a t lea s tthree or ders o fmag nitud e redu c tio n inthe si ze of da ta se tis found for quasi-Monte Carlosampli ng wi tho ut loss o f a c cur acy (seetheSupporting In for matio n).
Resu ltsa nd Di s cussi on
= = ========= = == === = = == =
Accuracy c h eck andtra i ning d ataset
| _{\theta(\omega_1),\theta(\omega_2)}$. A_ResNet consists_of a series of_blocks. One_block_is given_in_Figure \[fig:schematic\] with_two linear transformations,_two activation functions, and_one short cut._Detailed_description of ResNet is included in the Supporting Information. Parameters of the surface roughness_($\theta(\omega_{1})$,$\theta_(\omega_{2})$) are_fed_as_input, and the EDL $\sigma$_is extracted as the output_function over_the entire parameter space. Sigmoid function is chosen_as_the activation function_here.
The loss function we use is the MSE between_the actual EDL $\sigma_{\theta(\omega_1),\theta(\omega_2)}$ given by_the diffusion equation_model_and_the predicted EDL $\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])$_given by the ResNet $$\label{eqn:MSE}
_MSE= \frac{1}{M}\sum_{j=1}^{M} \left(\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}-\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])\right)^2,$$ where $\theta$ represents_the parameter set in the ResNet, $j$_is the $j$-th sample, and M_is the size of training_data set.
Define_the relative $L^{\infty}$ error of_EDL as $$Error_= _\max_{1\leq j\leq M}_\dfrac{\left|\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}-\sigma(\theta(\omega_1)[j],\theta(\omega_2)[j])\right|} {\sigma_{\theta(\omega_1)[j],\theta(\omega_2)[j]}},$$ which will be used_to quantify the_approximation accuracy of DL.
Quasi-Monte Carlo Sampling
--------------------------
Compared_to_uniform sampling and_Monte-Carlo_sampling,_quasi-Monte Carlo_sampling provides the_best_compromise between_accuracy_and efficiency. It overcomes the curse_of_dimensionality and has high accuracy [@Russel1998]. For_the simulations in our_work,_at least three orders_of magnitude reduction in the_size of data set is found_for quasi-Monte_Carlo sampling_without loss of accuracy (see the Supporting Information).
Results and Discussion
======================
Accuracy check_and training data set
|
quark, for which $B_3(\tau_t,\tau_{t/Z})\approx B_1(\tau_{t/Z})\approx-0.024$ is very small. As in the case of the $a\to\gamma\gamma$ decay discussed in Section \[sec:agaga\], the main effect of electroweak radiative corrections would be to renormalize the gauge couplings. In the present case the coupling $\alpha$ associated with the photon is evaluated at $q^2=0$, while the coupling $\alpha(m_Z)/(s_w^2 c_w^2)$ associated with the $Z$ boson should be evaluated at $q^2=m_Z^2$ as indicated. The $Z\to\gamma a$ branching fraction is obtained by dividing this partial decay rate by the $Z$-boson total width $\Gamma_Z$. This yields $$\label{ZgaaBR}
\mbox{Br}(Z\to\gamma a) = 8.17\cdot 10^{-4}\,\big| C_{\gamma Z}^{\rm eff} \big|^2
\left( 1 - \frac{m_a^2}{m_Z^2} \right)^3
\left[ \frac{1\,\mbox{TeV}}{\Lambda} \right]^2.$$ By requiring the $Z$-boson total width to agree with the direct measurement $\Gamma_Z=(2.495\pm 0.0023)$GeV performed at LEP [@ALEPH:2005ab], an upper bound on the Wilson coefficient $|C_{\gamma Z}^{\rm eff}|$ can be extracted. At 95% CL we find $\mbox{Br}(Z\to\mbox{BSM})<0.0018$ and $$\label{eq:cZgammabound}
\big| C_{\gamma Z}^{\rm eff} \big| < 1.48\,\bigg[ \frac{\Lambda}{1\,\mbox{TeV}} \bigg] \,.$$ This bound is obtained by neglecting the ALP mass and gets weaker when $m_a$ approaches the kinematic threshold at $m_a=m_Z$.
![\[fig:Zga\] Parameter space excluded by measurements of $\mbox{Br}(Z\to\gamma\gamma)$ and $\mbox{Br}(Z\to\gamma\gamma\gamma)$ (left) and measurements of $\mbox{Br}(Z\ | quark, for which $ B_3(\tau_t,\tau_{t / Z})\approx B_1(\tau_{t / Z})\approx-0.024 $ is very small. As in the case of the $ a\to\gamma\gamma$ decay discussed in Section \[sec: agaga\ ], the main consequence of electroweak radiative correction would be to renormalize the gauge couplings. In the present lawsuit the coupling $ \alpha$ associated with the photon is evaluate at $ q^2=0 $, while the yoke $ \alpha(m_Z)/(s_w^2 c_w^2)$ associated with the $ Z$ boson should be evaluated at $ q^2 = m_Z^2 $ as indicated. The $ Z\to\gamma a$ branching fraction is obtained by separate this partial decay rate by the $ Z$-boson sum width $ \Gamma_Z$. This give $ $ \label{ZgaaBR }
\mbox{Br}(Z\to\gamma a) = 8.17\cdot 10^{-4}\,\big| C_{\gamma Z}^{\rm eff } \big|^2
\left (1 - \frac{m_a^2}{m_Z^2 } \right)^3
\left [ \frac{1\,\mbox{TeV}}{\Lambda } \right]^2.$$ By requiring the $ Z$-boson total width to agree with the direct measurement $ \Gamma_Z=(2.495\pm 0.0023)$GeV performed at LEP [ @ALEPH:2005ab ], an upper oblige on the Wilson coefficient $ |C_{\gamma Z}^{\rm eff}|$ can be extracted. At 95% CL we find $ \mbox{Br}(Z\to\mbox{BSM})<0.0018 $ and $ $ \label{eq: cZgammabound }
\big| C_{\gamma Z}^{\rm eff } \big| < 1.48\,\bigg [ \frac{\Lambda}{1\,\mbox{TeV } } \bigg ] \,.$$ This adhere is obtained by neglecting the ALP mass and gets weak when $ m_a$ approaches the kinematic threshold at $ m_a = m_Z$.
! [ \[fig: Zga\ ] Parameter space excluded by measurements of $ \mbox{Br}(Z\to\gamma\gamma)$ and $ \mbox{Br}(Z\to\gamma\gamma\gamma)$ (leftover) and measurements of $ \mbox{Br}(Z\ | quwrk, for which $B_3(\tau_t,\tau_{t/Z})\xpprox B_1(\tau_{t/Z})\apkrix-0.024$ is tery smzll. As iv the case of the $a\to\gamma\galmq$ decqy discussed in Sectiov \[sec:agaga\], the maib efhect of electrowxzk radiative dlrreetmons would be tp renormalhze the gauge wojppings. In the present case the couplyng $\alpna$ associated wijh tht protoh is evaluated at $q^2=0$, while the coupming $\alkha(m_Z)/(s_w^2 c_w^2)$ associayed with the $Z$ boson shoulf be evaluated at $q^2=m_Z^2$ ws indicatee. Thq $Z\to\gamma a$ cranching fraction is kbtained by dividing this partixl deeay rate by tye $X$-toson total widtr $\Gamma_Z$. This yields $$\label{ZbaaBR}
\mbox{Br}(D\to\gakma a) = 8.17\cdot 10^{-4}\,\big| C_{\gamma Z}^{\cm eff} \big|^2
\left( 1 - \frac{m_a^2}{m_Z^2} \rnght)^3
\left[ \frac{1\,\mboz{TwV}}{\Lamtda} \sighg]^2.$$ By resumrihg the $Z$-bkson total width to atree with the direcu mqqsurement $\Gamja_Z=(2.495\pm 0.0023)$DeD performed at LEP [@ALEPH:2005ab], an upper boutd kn the Wilson coefficiebt $|C_{\gamma Z}^{\rm eff}|$ can be extrasted. At 95% CL we find $\mbox{Br}(Z\to\mbox{BSM})<0.0018$ and $$\label{eq:cSgammebuunb}
\bie| C_{\hamma Z}^{\rm eff} \big| < 1.48\,\bigg[ \frac{\Lambda}{1\,\mbox{TeV}} \bigd] \,.$$ Uhix bound is obtcined by neglectonh yre ALP mass avd gets wsaker when $m_a$ apprlaches jhe kibematic trresnold at $m_a=m_Z$.
![\[fig:Zga\] Parametee space excllded by measurements oy $\mbox{Br}(Z\to\gcmma\gakma)$ amd $\mbox{Br}(Z\to\gamma\gamma\gcmma)$ (lsft) and meadurements uf $\mbox{Br}(Z\ | quark, for which $B_3(\tau_t,\tau_{t/Z})\approx B_1(\tau_{t/Z})\approx-0.024$ is very in case of $a\to\gamma\gamma$ decay discussed effect electroweak radiative corrections be to renormalize gauge couplings. In the present case coupling $\alpha$ associated with the photon is evaluated at $q^2=0$, while the coupling c_w^2)$ associated with the $Z$ boson should be evaluated at $q^2=m_Z^2$ as indicated. $Z\to\gamma branching is by dividing this partial decay rate by the $Z$-boson total width $\Gamma_Z$. This yields $$\label{ZgaaBR} \mbox{Br}(Z\to\gamma = 8.17\cdot 10^{-4}\,\big| C_{\gamma Z}^{\rm eff} \big|^2 \left( - \frac{m_a^2}{m_Z^2} \right)^3 \left[ \right]^2.$$ By requiring the $Z$-boson width agree with direct $\Gamma_Z=(2.495\pm performed at LEP an upper bound on the Wilson coefficient $|C_{\gamma Z}^{\rm eff}|$ can be extracted. At 95% CL we $\mbox{Br}(Z\to\mbox{BSM})<0.0018$ and C_{\gamma Z}^{\rm \big| 1.48\,\bigg[ \bigg] \,.$$ This obtained by neglecting the ALP mass when $m_a$ approaches the kinematic threshold at $m_a=m_Z$. Parameter space by measurements of $\mbox{Br}(Z\to\gamma\gamma)$ and $\mbox{Br}(Z\to\gamma\gamma\gamma)$ and measurements of $\mbox{Br}(Z\ | quark, for which $B_3(\tau_t,\tau_{t/Z})\apProx B_1(\tau_{t/Z})\ApproX-0.024$ is VerY sMall. as in The case of the $a\tO\GammA\gamma$ decay discussed in sectiOn \[SEc:agAGa\], The maIn effecT Of ELEctRoWeAk rAdIAtIve coRreCtions wOuld be to reNorMaLize the gauge COuPlings. In thE prEsent case the CouPling $\aLpHa$ aSSociaTed With tHe photON is evaLuated at $q^2=0$, WhILe the cOUpling $\aLPHa(M_Z)/(s_w^2 C_w^2)$ associated with tHE $Z$ BOson should be evAluateD aT $Q^2=m_z^2$ AS inDicAted. The $Z\to\GaMma a$ bRAnching FRaCTIOn iS Obtained by divIding this paRTiaL decay RaTe bY The $Z$-boSon toTaL WidTh $\Gamma_Z$. ThiS yieLds $$\label{ZGaaBR}
\mBOx{Br}(Z\to\GAmma a) = 8.17\cdOt 10^{-4}\,\big| C_{\GamMa Z}^{\Rm efF} \BiG|^2
\lEft( 1 - \FrAC{m_a^2}{M_z^2} \rIghT)^3
\LefT[ \frac{1\,\mboX{TEV}}{\lambdA} \rigHT]^2.$$ bY RequIriNg thE $Z$-bosOn total width tO agRee wITh tHe dirEct meAsurEmEnt $\GaMma_Z=(2.495\pm 0.0023)$geV peRfOrmed at LEP [@ALEPH:2005Ab], an Upper bounD on ThE WiLsOn coeFFicienT $|C_{\gAmmA Z}^{\rm eff}|$ Can be exTRacTeD. aT 95% cL We find $\mbox{Br}(Z\to\mboX{Bsm})<0.0018$ AnD $$\label{eq:CZgammABoUnD}
\Big| C_{\gammA Z}^{\Rm eFf} \biG| < 1.48\,\BIgg[ \frAc{\LaMBdA}{1\,\mbox{TeV}} \Bigg] \,.$$ ThIS bOuNd is obtAiNed by nEgLecTinG the Alp masS and geTs weaker When $m_A$ Approaches the kINematic threshOLd AT $M_a=M_z$.
![\[fig:zga\] parameter spAce eXCludEd by MEaSurEMents Of $\mboX{BR}(z\tO\Gamma\gamma)$ and $\mbox{Br}(z\tO\gamma\Gamma\Gamma)$ (left) and mEasurementS OF $\Mbox{Br}(Z\ | quark, for which $B_3(\ta u_t,\tau_{ t/Z}) \ap pro xB_1( \tau _{t/Z})\approx - 0.02 4$ is very small. As i n the c a se o f t he $a \to\gam m a\ g a mma $de cay d i sc ussed in Sectio n \[sec:ag aga \] , the main e f fe ct of elec tro weak radiati vecorrec ti ons would be to r enorma l ize th e gauge c ou p lings. In thep r es entcase the coupling $\ a lpha$ associat ed wit ht he p hot onis evaluat ed at $ q ^2=0$,w hi l e the coupling $\al pha(m_Z)/(s _ w^2 c_w^2 )$ as s ociate d wit ht he$Z$ boson s houl d be eval uateda t $q^2= m _Z^2$ a s indi cat ed. The $Z \t o\g am m a a $ b ran c hin g fracti on i s obt aine d b y div idi ng t his p artial decayrat e by the $Z$- boson tot al widt h $\Ga mma_Z $. This yields $$ \lab el{ZgaaBR }
\mb ox {Br}( Z \to\ga mma a) = 8.17 \cdot 1 0 ^{- 4} \ , \ bi g| C_{\gamma Z}^{\ rm e ff } \big|^ 2
\ le ft ( 1 - \fr ac {m_ a^2} { m _Z^2} \ri g ht )^3
\left [ \ fr ac{1\,\ mb ox{TeV }} {\L amb da} \ r ight ]^2.$$ By requ iring the $Z$-bosont otal width to ag r e ew iththe direct mea sure m ent$\Ga m ma _Z= ( 2.495 \pm 0 .0 0 23 ) $GeV performed at L EP [@ALE PH:20 05ab], an upp er bound o n t he Wilso n co e ff i cient $|C_{\ga mma Z }^{\rm eff } |$ can b e ext racted.At 95% CL w e find $ \mb ox{ Br} (Z\ t o \m box{BSM})<0.0 0 1 8$ a nd $$\lab el{ eq:cZga mma bou nd}
\ big| C_{\ gamma Z} ^{ \r mef f}\big| < 1.48\, \b igg [\fr ac{\L a mbda}{ 1\,\m box{ Te V} } \b igg] \, . $$ T hisbo un d is ob ta inedby n e gle cting t he ALP ma ssa nd g et sweakerwhen $m_a$ ap pr oaches the k ine matict h resholdat $m_a=m_Z$.
![\[fig: Z ga\] Pa ram eterspac e exclude d b y meas ure m ents o f $\mb ox{Br }( Z\t o \ gamma \ g am ma) $and $\mbox { B r}( Z\to\ ga mma\ gamma\g amma)$ (left) andm eas urements of $ \mb ox{B r } (Z \ | quark,_for which_$B_3(\tau_t,\tau_{t/Z})\approx B_1(\tau_{t/Z})\approx-0.024$ is very_small. As_in_the case_of_the $a\to\gamma\gamma$ decay_discussed in Section \[sec:agaga\],_the main effect of_electroweak radiative corrections_would_be to renormalize the gauge couplings. In the present case the coupling $\alpha$ associated_with_the photon_is_evaluated_at $q^2=0$, while the coupling_$\alpha(m_Z)/(s_w^2 c_w^2)$ associated with the_$Z$ boson_should be evaluated at $q^2=m_Z^2$ as indicated. The_$Z\to\gamma_a$ branching fraction_is obtained by dividing this partial decay rate by_the $Z$-boson total width $\Gamma_Z$. This_yields $$\label{ZgaaBR}
__\mbox{Br}(Z\to\gamma_a) = 8.17\cdot 10^{-4}\,\big|_C_{\gamma Z}^{\rm eff} \big|^2
_ \left( 1 - \frac{m_a^2}{m_Z^2} \right)^3_
\left[ \frac{1\,\mbox{TeV}}{\Lambda} \right]^2.$$_By requiring the $Z$-boson total width_to agree with the direct_measurement $\Gamma_Z=(2.495\pm_0.0023)$GeV performed at LEP [@ALEPH:2005ab],_an upper bound_on the_Wilson coefficient $|C_{\gamma_Z}^{\rm eff}|$ can be extracted. At_95% CL we_find $\mbox{Br}(Z\to\mbox{BSM})<0.0018$ and $$\label{eq:cZgammabound}
_\big|_C_{\gamma Z}^{\rm eff}_\big|_<_1.48\,\bigg[ \frac{\Lambda}{1\,\mbox{TeV}}_\bigg] \,.$$ This_bound_is obtained_by_neglecting the ALP mass and gets_weaker_when $m_a$ approaches the kinematic threshold at_$m_a=m_Z$.
![\[fig:Zga\] Parameter space excluded_by_measurements of $\mbox{Br}(Z\to\gamma\gamma)$ and_$\mbox{Br}(Z\to\gamma\gamma\gamma)$ (left) and measurements of_$\mbox{Br}(Z\ |
Thus, it should be used as a good additional tracer for AGNs in low signal-to-noise ratio surveys.
Summary\[sub:Summary\]
----------------------
{width="0.3\paperwidth"}{width="0.3\paperwidth"}\
{width="0.3\paperwidth"}{width="0.3\paperwidth"}
Figure \[Flo:summary\] shows how the different types of galaxies (according to K06) appear in the high-redshift diagrams (top panels) and how the high redshift classifications appear back in one of the K06 diagnostic diagrams (bottom panels). In all panels, SFG are plotted in blue, Seyfert 2 in green (except in the bottom-right panel where green points stand for all types of AGNs), LINERs in cyan, and composites in magenta. The L10 diagnostic (left panels) implies several regions where different types of galaxies get mixed. Seyfert 2 region and LINERs region are now quite well defined, but we see composites falling in the SFGs and LINERs regions. Most of the composites fall in the region of the L10 diagnostic called SF-LIN/comp(marked by the dashed contour corresponding to Eqs. \[eq: New blue composite 1\] and \[eq: New blue composite 2\]). SFGs and Seyfert 2 are now separated quite well, but still there is a small region of the L10 diagnostic, called SF/Sy2, where they get mixed. In the bottom-left panel, it seems that most of the SF/Sy2 galaxies belong to the K06 SFG region, and that a large number of SFG/comp galaxies belong to K06 SFG region. LIN/comp galaxies seem to appear half/half in the K06 composites and LINERs regions.
We now the compare K06 and DEW classifications (right panels). We see that all K06 LINERs are correctly classified as AGNs in the DEW diagnostic. Most of K06 Seyfert 2 galaxies lie in the DEW AGN region as well, so that is quite satisfying. However composites are shared in DEW SFG and AGN regions, which confirms that composites are sort of hybrids between AGNs and SFGs, also in terms of stellar populations. Thus they obviously cannot be isolated in the DEW diagnostic. | Thus, it should be used as a good additional tracer for AGNs in depleted signal - to - randomness ratio surveys.
Summary\[sub: Summary\ ]
----------------------
! [ image](143fig1){width="0.3\paperwidth"}{width="0.3\paperwidth"}\
! [ image](143fig3){width="0.3\paperwidth"}{width="0.3\paperwidth " }
calculate \[Flo: summary\ ] prove how the different types of galaxy (according to K06) appear in the eminent - redshift diagrams (top panel) and how the high redshift classification look back in one of the K06 diagnostic diagrams (bottom panels). In all panel, SFG are plotted in blue, Seyfert 2 in fleeceable (except in the bottom - right panel where green point stand for all types of AGNs), liner in cyan, and composites in magenta. The L10 diagnostic (leftover panels) incriminate several regions where different character of galaxies get mixed. Seyfert 2 region and LINERs area are nowadays quite well defined, but we see composites falling in the SFGs and LINERs regions. Most of the composites fall in the region of the L10 diagnostic call SF - LIN / comp(marked by the dashed contour corresponding to Eqs. \[eq: New aristocratic composite 1\ ] and \[eq: New blue composite 2\ ]). SFGs and Seyfert 2 are nowadays separated quite well, but still there is a small area of the L10 diagnostic, called SF / Sy2, where they get mixed. In the bottom - left panel, it seems that most of the SF / Sy2 galaxies belong to the K06 SFG region, and that a large number of SFG / comprehensive examination galaxies belong to to K06 SFG region. LIN / comp galaxies seem to appear half / half in the K06 complex and LINERs regions.
We now the compare K06 and DEW classification (right panels). We see that all K06 LINERs are correctly classify as AGNs in the DEW diagnostic. Most of K06 Seyfert 2 galaxies lie in the DEW AGN area as well, so that is quite satisfying. However composites are shared in DEW SFG and AGN region, which confirms that composite are sort of hybrids between AGNs and SFGs, also in terms of stellar populations. therefore they obviously cannot be isolated in the DEW diagnostic. | Thks, it should be used as x good additionco tracxr for ZGNs in uow signal-to-noise ratio survxys.
Symmart\[sub:Summary\]
----------------------
{widgh="0.3\paperwifth"}{wmdth="0.3\paperwidth"}\
{width="0.3\papsvwidtk"}{width="0.3\pakerwidth"}
Figuse \[Flo:summary\] svods how the different types of galaxief (accorcijg to K06) appear in uhe higg-gebshift diagrams (top panels) and gow the high redshify classifications appear bwck ln one of the K06 diwgnostic diqgraiw (bottom pandls). In all panels, SFG zre plotted in blue, Seyfert 2 in greeu (except in tye hmttom-right 'anel rhere green ipints vtand fpr all types on AGNv), LUNERs in cyan, and com'osites in magenta. Tre L10 diagtoatic (left panels) umplies sevaral eegkona xhede difverxnt types or galaxies tet mixed. Seyfert 2 tedppn and LINERa regijn are now quite well defined, but we see bompksites falling in the SDGs and LINERs regiond. Most of the composites fall in the region of the L10 diagnmstic zaloeq WF-PIN/comp(marked by the dashed contour corresponqjnb no Eqs. \[eq: New blue composite 1\] ajd \[rz: New blue coopositz 2\]). AFGs and Seyfert 2 wre now sepaeated quiue wekl, but still there is a smaol region of rhe L10 diagnostic, cclled SF/Sy2, wkere tney grt mixed. In the bottom-lzft pahel, it seemd that moag of the SF/Sy2 gauaxpes telong to the K06 SFG region, and that a lcrge numcer pf SFG/somp galaxles bcnong to K06 SFG regiln. LIU/comp galaxies deem to appear half/half in the K06 composites amd LITERs regnons.
We now the compwre K06 and DEW elassifieationr (right pahels). We see that ajl K06 LINERs asg correctly cnassifieq as AGNw in thd DEW diagnostiv. Most of K06 Seyfert 2 galaxies lie in bhe DGW AGN region as ctll, so that is quiye ratysvymng. Hjfever composhtes ard shardd in DTC SNG xnd SGN regions, which cotfirjs that composites ave sort od hybridf between AGNx and SFGs, also in termv oh stelkar populations. Thus they obviousmy cannot be isolated in the DEW diagnoscic. | Thus, it should be used as a tracer AGNs in signal-to-noise ratio surveys. \[Flo:summary\] how the different of galaxies (according K06) appear in the high-redshift diagrams panels) and how the high redshift classifications appear back in one of the diagnostic diagrams (bottom panels). In all panels, SFG are plotted in blue, Seyfert in (except the panel where green points stand for all types of AGNs), LINERs in cyan, and composites in The L10 diagnostic (left panels) implies several regions different types of galaxies mixed. Seyfert 2 region and region now quite defined, we composites falling in SFGs and LINERs regions. Most of the composites fall in the region of the L10 diagnostic called by the corresponding to \[eq: blue 1\] and \[eq: composite 2\]). SFGs and Seyfert 2 quite well, but still there is a small of the diagnostic, called SF/Sy2, where they get In the bottom-left panel, it seems that most the SF/Sy2 galaxies belong to the K06 SFG region, and that a large number of belong to K06 SFG LIN/comp galaxies seem appear in K06 and LINERs We now the compare K06 and DEW classifications (right panels). We that all K06 LINERs are correctly classified as AGNs in diagnostic. of K06 Seyfert galaxies lie in the AGN as well, so that satisfying. composites DEW and regions, which confirms that are sort of hybrids between and SFGs, also in they obviously cannot be isolated in the DEW | Thus, it should be used as a good Additional TraceR foR AGns In loW sigNal-to-noise ratiO SurvEys.
Summary\[sub:Summary\]
----------------------
{wIDth="0.3\pAPeRwidtH"}{wIDTh="0.3\pApErWidTh"}\
{WidTh="0.3\paperWidth"}{WiDth="0.3\paperwidtH"}
fiGure \[Flo:sumMarY\] shows how the DifFerent TyPes OF galaXieS (accoRding tO k06) appeaR in the higH-rEDshift DIagrams (TOP pAnelS) and how the high redSHiFT classificatioNs appeAr BAcK IN onE of The K06 diagnoStIc diaGRams (botTOm PANEls). iN all panels, SFG Are plotted iN BluE, SeyfeRt 2 In gREen (excEpt in ThE BotTom-right panEl whEre green pOints sTAnd for aLL types oF AGNs), LiNErs iN cyaN, AnD cOmpOsITes IN mAgeNTa. THe L10 diagnOsTiC (left PaneLS) IMPlieS seVeraL regiOns where diffeRenT typES of GalaxIes geT mixEd. seyfeRt 2 regiOn and lInERs region are noW quiTe well defIneD, bUt wE sEe comPOsites FalLinG in the SfGs and LinERS rEGIOnS. Most of the compositEs FALl In the regIon of tHE L10 DiAGnostic cAlLed sF-LIn/COmp(maRked BY tHe dashed ContouR CoRrEspondiNg To Eqs. \[eQ: NEw bLue CompoSIte 1\] aNd \[eq: NeW blue comPositE 2\]). sFGs and Seyfert 2 ARe now separateD QuITE wELl, buT stIll there is a SmalL RegiOn of THe l10 diAGnostIc, calLeD sF/sY2, where they get mixed. IN tHe bottOm-lefT panel, it seems That most of THE sF/Sy2 galaXies BElONg to the K06 SFG regIon, anD that a largE Number of sFG/coMp galaxiEs belong tO k06 sFG regioN. LIn/coMp gAlaXIEs Seem to appear hALF/halF iN the K06 coMpoSites anD LInERS reGioNs.
we now the cOmpare K06 aNd dEw cLaSsiFicatIOns (right PaNelS). WE seE that ALl K06 LINeRs arE corReCtLY clAssifieD As agns in ThE DeW diAgnOsTic. MoSt of k06 seyFert 2 galAxies lie iN thE dEW AgN ReGion as wEll, so that is quItE satisfyinG. HOweVer comPOSites are Shared in DEW SFG and AGN regIOns, whicH coNfirmS thaT compositEs aRe sort Of hYBrids bEtween aGNs aNd sFGS, ALso in TERmS of StEllar populATIonS. Thus ThEy obViously Cannot be isolated in THe DeW diagnostic. | Thus, it should be used a s a good a dditi ona l t ra cerforAGNs in low si g nal- to-noise ratio surveys .
Su mm a ry\[ s ub :Summ ary\]
- - -- - - --- -- -- --- -- - -- -
{widt h=" 0. 3\paperwidth " }! [image](14 3fi g2){width="0 .3\ paperw id th" } \
 { width= "0.3\pape rw i dth"}! [ image]( 1 4 3f ig4) {width="0.3\paper w id t h"}
Figure \[ Flo:su mm a ry \ ] sh ows how the d if feren t typeso fg a l axi e s (accordingto K06) app e arin the h igh - redshi ft di ag r ams (top panel s) a nd how th e high redshif t classi ficati ons ap pear ba ck in o n e o f t heK 06diagnost ic d iagra ms ( b o t t om p ane ls). In a ll panels, SF G a re p l ott ed in blue , Se yf ert 2 in gr een ( ex cept in the bot tom- right pan elwh ere g reenp ointssta ndfor all typeso f A GN s ) , L INERs in cyan, and c o m po sites in magen t a. T h e L10 di ag nos tic( l eft p anel s )impliessevera l r eg ions wh er e diff er ent ty pes o f gal axiesget mixe d. Se y fert 2 regiona nd LINERs reg i on a re nowqui te well def ined , but wes ee co m posit es fa ll i ng in the SFGs and LIN ER s regi ons.Most of the c ompositesf a l l in the reg i on of the L10 dia gnost ic calledS F-LIN/co mp(ma rked bythe dashe d contourcor res pon din g to Eqs. \[eq: N e w blu ecomposi te1\] and \[ eq: Ne w b lu e composi te 2\]). S FG san d S eyfer t 2 are n ow se pa rat ed qu i te wel l, bu t st il lt her e is as ma l l reg io nof t heL1 0 dia gnos t ic, called SF/Sy2,whe r e th ey g et mixe d. In the bot to m-left pan el , i t seem s that mos t of the SF/Sy2 galaxie s belong to theK06SFG regio n,and th ata large numbe r ofSF G/c o m p gal a x ie s b el ong to K06 S FGregio n. LIN /comp g alaxies seem to ap p ear half/half in th e K0 6 co mpo s it e s a nd LIN E R s regions.
Wenow the co mp a re K06 and D E W c la ssifica tions ( right panels) . We seethat allK0 6 LI N E Rsare correc tly clas sified as AGNsi nthe D EWdiagno st ic. Most of K0 6 Se yfert 2 gal ax ies li e inth e DEW AG N region as well, so th at isquite sa tisfying. Ho w eve r composi tesare shared in DE W SFG an d AGNregi o ns , w h ich c onfi r ms that c o mp osi t e sare sort of h y bri ds be twe e n AGNs and SFGs, also in te r ms of stellarpopu l a tio ns. Thus t hey obviouslycan no t be isola te d in the DE W diagno st i c. | Thus,_it should_be used as a_good additional_tracer_for AGNs_in_low signal-to-noise ratio_surveys.
Summary\[sub:Summary\]
----------------------
{width="0.3\paperwidth"}{width="0.3\paperwidth"}\
{width="0.3\paperwidth"}{width="0.3\paperwidth"}
Figure \[Flo:summary\] shows how_the different types of_galaxies (according to_K06)_appear in the high-redshift diagrams (top panels) and how the high redshift classifications appear_back_in one_of_the_K06 diagnostic diagrams (bottom panels)._In all panels, SFG are_plotted in_blue, Seyfert 2 in green (except in the_bottom-right_panel where green_points stand for all types of AGNs), LINERs in_cyan, and composites in magenta. The_L10 diagnostic (left_panels)_implies_several regions where different_types of galaxies get mixed. Seyfert_2 region and LINERs region are_now quite well defined, but we see_composites falling in the SFGs and_LINERs regions. Most of the_composites fall_in the region of the_L10 diagnostic called_SF-LIN/comp(marked by_the dashed contour_corresponding to Eqs. \[eq: New blue composite_1\] and \[eq: New_blue composite 2\]). SFGs and Seyfert_2_are now separated_quite_well,_but still_there is a_small_region of_the_L10 diagnostic, called SF/Sy2, where they_get_mixed. In the bottom-left panel, it seems_that most of the_SF/Sy2_galaxies belong to the_K06 SFG region, and that_a large number of SFG/comp galaxies_belong to_K06 SFG_region. LIN/comp galaxies seem to appear half/half in the K06 composites_and LINERs regions.
We now the compare_K06 and DEW classifications_(right panels)._We_see that all_K06_LINERs are_correctly classified as AGNs in the DEW_diagnostic. Most_of K06 Seyfert 2 galaxies lie_in the DEW AGN_region_as well, so that is quite_satisfying. However composites are shared in_DEW SFG and AGN regions,_which_confirms_that composites are sort of_hybrids between AGNs and SFGs, also_in terms of_stellar populations. Thus they obviously cannot be_isolated_in the DEW diagnostic. |
$G^A$ from $Y$ onto $A$ with respect to $E_1^{\rho, u}$ and $F^A$. Since $A' \cap D={\mathbf C}1$, by Watatani [@Watatani:index Proposition 4.1], $F^A =E_1^{\sigma, v}$. Hence $G^A$ is a conditional expectation from $Y$ onto $A$ with respect to $E_1^{\rho, u}$ and $E_1^{\sigma, v}$. By the discussions in [@KT4:morita Section 2] and the proof of Rieffel [@Rieffel:rotation Proposition 2.1], there is an isomorphism $\Psi$ of $D$ onto $C$ defined by $$\Psi(d)={}_C {\langle}1_A \cdot d \,, \, 1_A {\rangle}$$ for any $d\in D$, where $A$ is a closed subspace of $Y$ and the unit element in $A$ is regarded as an element in $Y$. Then for any $a\in A$ $$\Psi(a)={}_A {\langle}1_A \cdot a \,, \, 1_A {\rangle}={}_A {\langle}a, \, 1_A {\rangle}=a.$$ Also, for any $d\in D$ $$\begin{aligned}
(E_1^{\rho, u}\circ\Psi)(d) & =E_1^{\rho, u}({}_C {\langle}1_A \cdot d \,, \, 1_A {\rangle})
={}_A {\langle}G^A (1_A \cdot d) \,, \, 1_A {\rangle}={}_A {\langle}1_A \cdot E_1^{\sigma, v}(d) \,, \, 1_A {\rangle}\\
& =E_1^{\sigma, v}(d).\end{aligned}$$ Thus, we obtain the following lemma:
\[lem:equation\]With the above notations, we suppose that $A' \cap C={\mathbf C}1$. Then $\Psi$ is an isomorphism of $D$ onto $C$ satisfying that $$\Psi|_A ={{\rm{id}}}_A, \quad E_1^{\rho, u}\circ\Psi=E_1^{\sigma, v}$$
Let $\widehat{\rho}$ and $\widehat{\sigma | $ G^A$ from $ Y$ onto $ A$ with respect to $ E_1^{\rho, u}$ and $ F^A$. Since $ A' \cap D={\mathbf C}1 $, by Watatani [ @Watatani: index Proposition 4.1 ], $ F^A = E_1^{\sigma, v}$. Hence $ G^A$ is a conditional anticipation from $ Y$ onto $ A$ with obedience to $ E_1^{\rho, u}$ and $ E_1^{\sigma, v}$. By the discussions in [ @KT4: morita Section 2 ] and the proof of Rieffel [ @Rieffel: rotation Proposition 2.1 ], there is an isomorphism $ \Psi$ of $ D$ onto $ C$ define by $ $ \Psi(d)={}_C { \langle}1_A \cdot d \, , \, 1_A { \rangle}$$ for any $ d\in D$, where $ A$ is a closed subspace of $ Y$ and the whole component in $ A$ is regarded as an element in $ Y$. Then for any $ a\in A$ $ $ \Psi(a)={}_A { \langle}1_A \cdot a \, , \, 1_A { \rangle}={}_A { \langle}a, \, 1_A { \rangle}=a.$$ Also, for any $ d\in D$ $ $ \begin{aligned }
(E_1^{\rho, u}\circ\Psi)(d) & = E_1^{\rho, u}({}_C { \langle}1_A \cdot d \, , \, 1_A { \rangle })
= { } _ A { \langle}G^A (1_A \cdot d) \, , \, 1_A { \rangle}={}_A { \langle}1_A \cdot E_1^{\sigma, v}(d) \, , \, 1_A { \rangle}\\
& = E_1^{\sigma, v}(d).\end{aligned}$$ therefore, we obtain the following lemma:
\[lem: equation\]With the above notations, we suppose that $ A' \cap C={\mathbf C}1$. Then $ \Psi$ is an isomorphism of $ D$ onto $ C$ meet that $ $ \Psi|_A = { { \rm{id}}}_A, \quad E_1^{\rho, u}\circ\Psi = E_1^{\sigma, v}$$
Let $ \widehat{\rho}$ and $ \widehat{\sigma | $G^A$ from $Y$ onto $A$ with resptct to $E_1^{\rho, u}$ and $F^A$. Sinre $A' \cal D={\mathbw C}1$, by Watatani [@Watatani:indee Priposiupon 4.1], $F^A =E_1^{\sigma, v}$. Hencd $G^A$ is a conditiinal wxpectatioi from $Y$ onto $Z$ witk cespect to $E_1^{\rho, u}$ and $E_1^{\sicma, v}$. By the dhszudsions in [@KT4:morita Section 2] and the proof pf Rieffel [@Rieffgl:rotseion Iriposition 2.1], there is an isomorlhism $\Pvi$ of $D$ onto $V$ defined by $$\Psi(d)={}_C {\langle}1_A \cdoh d \,, \, 1_A {\rangle}$$ for any $d\in D$, qherq $A$ is a closdd subspact mf $Y$ and tge unit element in $A$ is regardea as cn element un $Y$. Hven for any $a\in W$ $$\Psi(a)={}_A {\langle}1_A \cdot a \,, \, 1_A {\tangle}={}_A {\langle}s, \, 1_E {\rabgle}=a.$$ Also, for any $d\ii D$ $$\begin{aligned}
(E_1^{\rho, u}\circ\Psi)(g) & =E_1^{\rho, u}({}_C {\langle}1_A \xdit d \,, \, 1_A {\sanguw})
={}_A {\uanflx}G^A (1_A \cdoh d) \,, \, 1_A {\rangls}={}_A {\langle}1_A \xdot E_1^{\sigma, v}(d) \,, \, 1_A {\tagtle}\\
& =E_1^{\sigma, v}(d).\snd{alidnqd}$$ Thus, we obtain the following lemma:
\[lek:eqhation\]With the above norations, we suppose thwt $A' \cap S={\mathbf C}1$. Then $\Psi$ is an isomorphism of $D$ onto $C$ vatishykng tmat $$\Owi|_W ={{\rm{id}}}_A, \quad E_1^{\rho, u}\circ\Psi=E_1^{\sigma, v}$$
Let $\widehat{\wgo}$ akd $\widehat{\sigma | $G^A$ from $Y$ onto $A$ with respect u}$ $F^A$. Since \cap D={\mathbf C}1$, $F^A v}$. Hence $G^A$ a conditional expectation $Y$ onto $A$ with respect to u}$ and $E_1^{\sigma, v}$. By the discussions in [@KT4:morita Section 2] and the of Rieffel [@Rieffel:rotation Proposition 2.1], there is an isomorphism $\Psi$ of $D$ onto defined $$\Psi(d)={}_C \cdot \,, \, 1_A {\rangle}$$ for any $d\in D$, where $A$ is a closed subspace of $Y$ the unit element in $A$ is regarded as element in $Y$. Then any $a\in A$ $$\Psi(a)={}_A {\langle}1_A a \, 1_A {\langle}a, 1_A Also, for any D$ $$\begin{aligned} (E_1^{\rho, u}\circ\Psi)(d) & =E_1^{\rho, u}({}_C {\langle}1_A \cdot d \,, \, 1_A {\rangle}) ={}_A {\langle}G^A (1_A d) \,, {\rangle}={}_A {\langle}1_A E_1^{\sigma, \,, 1_A {\rangle}\\ & Thus, we obtain the following lemma: notations, we suppose that $A' \cap C={\mathbf C}1$. $\Psi$ is isomorphism of $D$ onto $C$ satisfying $$\Psi|_A ={{\rm{id}}}_A, \quad E_1^{\rho, u}\circ\Psi=E_1^{\sigma, v}$$ Let $\widehat{\rho}$ $\widehat{\sigma | $G^A$ from $Y$ onto $A$ with respect to $e_1^{\rho, u}$ and $F^A$. since $a' \caP D={\mAtHbf C}1$, By WaTatani [@Watatani:INdex proposition 4.1], $F^A =E_1^{\sigma, v}$. HeNce $G^A$ Is A CondITiOnal eXpectatIOn FROm $Y$ OnTo $a$ wiTh REsPect tO $E_1^{\rHo, u}$ and $E_1^{\Sigma, v}$. By thE diScUssions in [@KT4:mORiTa Section 2] aNd tHe proof of RieFfeL [@RieffEl:RotATion PRopOsitiOn 2.1], therE Is an isOmorphism $\psI$ Of $D$ ontO $c$ defineD BY $$\PSi(d)={}_C {\Langle}1_A \cdot d \,, \, 1_A {\rangLE}$$ fOR any $d\in D$, where $A$ Is a cloSeD SuBSPacE of $y$ and the uniT eLemenT In $A$ is reGArDED As aN Element in $Y$. TheN for any $a\in A$ $$\pSi(a)={}_a {\langlE}1_A \CdoT A \,, \, 1_A {\rangLe}={}_A {\laNgLE}a, \, 1_A {\Rangle}=a.$$ Also, For aNy $d\in D$ $$\begIn{aligNEd}
(E_1^{\rho, u}\CIrc\Psi)(d) & =e_1^{\rho, u}({}_C {\LanGle}1_a \cdoT D \,, \, 1_A {\RaNglE})
={}_A {\LAngLE}G^a (1_A \cDOt d) \,, \, 1_a {\rangle}={}_A {\LaNgLe}1_A \cdOt E_1^{\sIGMA, V}(d) \,, \, 1_A {\rAngLe}\\
& =E_1^{\sIgma, v}(D).\end{aligned}$$ ThUs, wE obtAIn tHe folLowinG lemMa:
\[Lem:eqUation\]with tHe Above notations, wE supPose that $A' \Cap c={\mAthBf c}1$. Then $\pSi$ is an IsoMorPhism of $d$ onto $C$ sATisFyING ThAt $$\Psi|_A ={{\rm{id}}}_A, \quad E_1^{\rhO, u}\CIRc\psi=E_1^{\sigmA, v}$$
Let $\wIDeHaT{\Rho}$ and $\wiDeHat{\SigmA | $G^A$ from $Y$ onto $A$ w ith respec t to$E_ 1^{ \r ho,u}$and $F^A$. Sin c e $A ' \cap D={\mathbf C}1$ , byWa t atan i [ @Wata tani:in d ex P rop os it ion 4 . 1] , $F^ A = E_1^{\s igma, v}$. He nc e $G^A$ is a co nditionalexp ectation fro m $ Y$ ont o$A$ withres pectto $E_ 1 ^{\rho , u}$ and $ E _1^{\s i gma, v} $ . B y th e discussions in[ @K T 4:morita Secti on 2]an d t h e pr oof of Rieffe l[@Rie f fel:rot a ti o n Pro p osition 2.1], there is a n is omorph is m $ \ Psi$ o f $D$ o n to$C$ defined by$$\Psi(d) ={}_C{ \langle } 1_A \cd ot d \ ,,\,1_A{ \r an gle }$ $ fo r a ny$ d\i n D$, wh er e$A$ i s ac l o s ed s ubs pace of $ Y$ and the un itelem e ntin $A $ isrega rd ed as an el ement i n $Y$. Then for any $a\in A$ $$ \P si( a) ={}_A {\lang le} 1_A \cdota \,, \ , 1_ A{ \ r an gle}={}_A {\langle }a , \, 1_A {\r angle} = a. $$ Also, fo rany $d\ i n D$ $ $\be g in {aligned }
(E_1 ^ {\ rh o, u}\c ir c\Psi) (d ) & =E _1^{\ r ho,u}({}_ C {\lang le}1_ A \cdot d \,, \ , 1_A {\rangle } )= { }_ A {\l ang le}G^A (1_A \cd o t d) \,, \, 1_ A {\ra ngle} ={ } _A {\langle}1_A \cdotE_ 1^{\si gma,v}(d) \,, \,1_A {\rang l e } \\
& =E_ 1^{\ s ig m a, v}(d).\end{ align ed}$$ Thus , we obta in th e follow ing lemma :
\[lem:e qua tio n\] Wit h th e above notat i o ns,we suppos e t hat $A' \c apC={ \ma th bf C}1$.Then $\P si $is a n i somor p hism of$D $ o nt o $ C$ sa t isfyin g tha t $$ \P si | _A={{\rm{ i d} } } _A,\q ua d E_ 1^{ \r ho, u }\ci r c\P si=E_1^ {\sigma,v}$ $
Le t$\ widehat {\rho}$ and $ \w idehat{\si gm a | $G^A$_from $Y$_onto $A$ with respect_to $E_1^{\rho,_u}$_and $F^A$._Since_$A' \cap D={\mathbf_C}1$, by Watatani_[@Watatani:index Proposition 4.1], $F^A_=E_1^{\sigma, v}$. Hence_$G^A$_is a conditional expectation from $Y$ onto $A$ with respect to $E_1^{\rho, u}$ and_$E_1^{\sigma,_v}$. By_the_discussions_in [@KT4:morita Section 2] and_the proof of Rieffel [@Rieffel:rotation_Proposition 2.1],_there is an isomorphism $\Psi$ of $D$ onto_$C$_defined by $$\Psi(d)={}_C_{\langle}1_A \cdot d \,, \, 1_A {\rangle}$$ for any_$d\in D$, where $A$ is a_closed subspace of_$Y$_and_the unit element in_$A$ is regarded as an element_in $Y$. Then for any $a\in_A$ $$\Psi(a)={}_A {\langle}1_A \cdot a \,, \,_1_A {\rangle}={}_A {\langle}a, \, 1_A {\rangle}=a.$$_Also, for any $d\in D$_$$\begin{aligned}
(E_1^{\rho, u}\circ\Psi)(d)_& =E_1^{\rho, u}({}_C {\langle}1_A \cdot_d \,, \,_1_A {\rangle})
={}_A_{\langle}G^A (1_A \cdot_d) \,, \, 1_A {\rangle}={}_A {\langle}1_A_\cdot E_1^{\sigma, v}(d)_\,, \, 1_A {\rangle}\\
& =E_1^{\sigma, v}(d).\end{aligned}$$_Thus,_we obtain the_following_lemma:
\[lem:equation\]With_the above_notations, we suppose_that_$A' \cap_C={\mathbf_C}1$. Then $\Psi$ is an isomorphism_of_$D$ onto $C$ satisfying that $$\Psi|_A ={{\rm{id}}}_A,_\quad E_1^{\rho, u}\circ\Psi=E_1^{\sigma, v}$$
Let_$\widehat{\rho}$_and $\widehat{\sigma |
[@CB4], use polynomial space. This approach would give a chance to do, also in polylogarithmic space, maximization of foldings relative to various parameters similar to those discussed in Theorem \[thm3\].
[*Acknowledgements.*]{} The author is grateful to Tim Riley for bringing to the author’s attention folklore arguments that solve the precise word problem for presentation $\langle \, a, b \, \| \, a=1, b=1 \, \rangle$ in polynomial time, the article [@SL1], and the similarity between the [precise word problem]{} for presentation $\langle \, a, b \, \| \, a=1, b=1 \, \rangle$ and the problem on efficient planar folding of RNA strands. The author thanks the referee for a number of useful remarks and suggestions.
[\[75\]]{} A. V. Anisimov, [*The group languages*]{}, Kibernetika (Kiev) [**4**]{}(1971), 18–24.
R. Armoni, A. Ta-Shma, A. Wigderson and S. Zhou, [*A $(\log n )^{4/3}$ space algorithm for $(s,t)$ connectivity in undirected graphs*]{}, J. Assoc. Comput. Mach. [**47**]{}(2000), 294–311.
S. Arora and B. Barak, [*Computational complexity – a modern approach*]{}, Cambridge Univ. Press, 2009.
D. Barrington, P. Kadau, K. Lange and P. McKenzie, [*On the complexity of some problems on groups input as multiplication tables*]{}, J. Comput. System Sci. [**63**]{}(2001), 186–200.
J.-C. Birget, A. Yu. Ol’shanskii, E. Rips, and M. V. Sapir, [*Isoperimetric functions of groups and computational complexity of the word problem*]{}, [Ann. Math.]{} [**156**]{}(2002), 467–518.
W. W. Boone, [*On certain simple undecidable problems in group theory, V, VI*]{}, Indag. Math. [**19**]{}(1957), 22–27, 227–232.
W. W. Boone, [*The word problem*]{}, Ann. Math. [**70**]{}(1959), 207–265.
V. V. Borisov, [*Simple | [ @CB4 ], use polynomial space. This approach would give a probability to do, besides in polylogarithmic space, maximization of foldings proportional to various parameter similar to those discussed in Theorem \[thm3\ ].
[ * recognition. * ] { } The author is grateful to Tim Riley for bringing to the writer ’s attention folklore arguments that clear the precise word trouble for presentation $ \langle \, a, b \, \| \, a=1, b=1 \, \rangle$ in polynomial time, the article [ @SL1 ], and the similarity between the [ precise word problem ] { } for presentation $ \langle \, a, b \, \| \, a=1, b=1 \, \rangle$ and the problem on effective planar folding of RNA strands. The author thanks the reviewer for a number of useful remarks and suggestions.
[ \[75\ ] ] { } A. V. Anisimov, [ * The group languages * ] { }, Kibernetika (Kiev) [ * * 4**]{}(1971), 18–24.
R. Armoni, A. Ta - Shma, A. Wigderson and S. Zhou, [ * A $ (\log n) ^{4/3}$ space algorithm for $ (second, t)$ connectivity in undirected graphs * ] { }, J. Assoc. Comput. Mach. [ * * 47**]{}(2000), 294–311.
S. Arora and B. Barak, [ * Computational complexity – a modern approach * ] { }, Cambridge Univ. Press, 2009.
D. Barrington, P. Kadau, K. Lange and P. McKenzie, [ * On the complexity of some problems on group input as multiplication tables * ] { }, J. Comput. System Sci. [ * * 63**]{}(2001), 186–200.
J.-C. Birget, A. Yu. Ol’shanskii, E. Rips, and M. V. Sapir, [ * Isoperimetric functions of groups and computational complexity of the word problem * ] { }, [ Ann. Math. ] { } [ * * 156**]{}(2002), 467–518.
W. W. Boone, [ * On certain simple undecidable problems in group theory, V, VI * ] { }, Indag. Math. [ * * 19**]{}(1957), 22–27, 227–232.
W. W. Boone, [ * The word problem * ] { }, Ann. Math. [ * * 70**]{}(1959), 207–265.
V. V. Borisov, [ * Simple | [@CB4], use polynomial space. Thls approach woulb give e chancs to do, xlso in polylogarithmic spacx, mazimizqtion of foldings relagive to vwrious pqramtters similar to vgose discusses in Chxorem \[thm3\].
[*Acknowlgdgements.*]{} Tha author is grdtdfbl to Tim Riley for bringing to the wuthor’s ahtention folkljre swgumsnts that solve the precise word pdoblem hor presentatiom $\langle \, a, b \, \| \, a=1, b=1 \, \ranhle$ in polynomial hime, the arjjclq [@SL1], and the rimilarity between the [precise word problem]{} for presengatiou $\langle \, a, b \, \| \, a=1, b=1 \, \rengle$ wnd the problem on exficieny planar foldikg of RNQ strands. The author vhanks the referee fjr a numbar of useful remarkw qnd soggesdionr.
[\[75\]]{} A. X. Ahixijov, [*Thf gcoup languafes*]{}, Kibernerika (Kiev) [**4**]{}(1971), 18–24.
R. Armoni, A. Ns-Shma, A. Wigdsrson wnq S. Zhou, [*A $(\log n )^{4/3}$ space algorithm for $(s,u)$ conhectivity in undirected graphs*]{}, J. Assoc. Compuj. Mach. [**47**]{}(2000), 294–311.
S. Wrora and B. Barak, [*Computational complexity – a modarn a'pfoaem*]{}, Zqmhridge Univ. Press, 2009.
D. Barrington, P. Kadau, K. Langq amd P. McKenzie, [*On the complecihy jf some problgms on ydohps input as multiolicatijn tavles*]{}, J. Coiput. System Sci. [**63**]{}(2001), 186–200.
J.-C. Birget, A. Yu. Ol’shanskii, V. Rips, and M. V. Sapir, [*Idoperimetrie funcjions pf groups and computatiunal complexity of the wkfd problem*]{}, [Ann. Mxth.]{} [**156**]{}(2002), 467–518.
F. W. Boont, [*On certain simplq undecideble 'roblems in broup eheory, V, VL*]{}, Indag. Math. [**19**]{}(1957), 22–27, 227–232.
W. W. Boone, [*The cord [roblem*]{}, Anj. Math. [**70**]{}(1959), 207–265.
V. V. Borisov, [*Simple | [@CB4], use polynomial space. This approach would chance do, also polylogarithmic space, maximization parameters to those discussed Theorem \[thm3\]. [*Acknowledgements.*]{} author is grateful to Tim Riley bringing to the author’s attention folklore arguments that solve the precise word problem presentation $\langle \, a, b \, \| \, a=1, b=1 \, \rangle$ in time, article and similarity between the [precise word problem]{} for presentation $\langle \, a, b \, \| \, a=1, \, \rangle$ and the problem on efficient planar of RNA strands. The thanks the referee for a of remarks and [\[75\]]{} V. [*The group languages*]{}, (Kiev) [**4**]{}(1971), 18–24. R. Armoni, A. Ta-Shma, A. Wigderson and S. Zhou, [*A $(\log n )^{4/3}$ space for $(s,t)$ undirected graphs*]{}, Assoc. Mach. 294–311. S. Arora Barak, [*Computational complexity – a modern Press, 2009. D. Barrington, P. Kadau, K. Lange P. McKenzie, the complexity of some problems on input as multiplication tables*]{}, J. Comput. System Sci. 186–200. J.-C. Birget, A. Yu. Ol’shanskii, E. Rips, and M. V. Sapir, [*Isoperimetric functions of computational complexity of the problem*]{}, [Ann. Math.]{} 467–518. W. [*On simple undecidable in group theory, V, VI*]{}, Indag. Math. [**19**]{}(1957), 22–27, 227–232. W. Boone, [*The word problem*]{}, Ann. Math. [**70**]{}(1959), 207–265. V. V. | [@CB4], use polynomial space. This aPproach wouLd givE a cHanCe To do, Also In polylogarithMIc spAce, maximization of foldiNgs reLaTIve tO VaRious ParametERs SIMilAr To ThoSe DIsCusseD in theorem \[Thm3\].
[*AcknowlEdgEmEnts.*]{} The authoR Is Grateful to tim riley for brinGinG to the AuThoR’S atteNtiOn folKlore aRGumentS that solvE tHE preciSE word prOBLeM for Presentation $\langlE \, A, b \, \| \, A=1, B=1 \, \rangle$ in polynOmial tImE, ThE ARtiCle [@sL1], and the siMiLaritY Between THe [PRECisE Word problem]{} foR presentatiON $\laNgle \, a, b \, \| \, A=1, b=1 \, \RanGLe$ and tHe proBlEM on Efficient plAnar Folding of rNA strANds. The aUThor thaNks the RefEreE for A NuMbEr oF uSEfuL ReMarKS anD suggestIoNs.
[\[75\]]{} a. V. AniSimoV, [*tHE GrouP laNguaGes*]{}, KiBernetika (Kiev) [**4**]{}(1971), 18–24.
r. ArMoni, a. ta-SHma, A. WIgderSon aNd s. Zhou, [*a $(\log n )^{4/3}$ sPace aLgOrithm for $(s,t)$ connEctiVity in undIreCtEd gRaPhs*]{}, J. ASSoc. ComPut. macH. [**47**]{}(2000), 294–311.
S. Arora And B. BarAK, [*CoMpUTATiOnal complexity – a modErN APpRoach*]{}, CamBridge uNiV. PREss, 2009.
D. BarrInGtoN, P. KaDAU, K. LanGe anD p. MCKenzie, [*ON the coMPlExIty of soMe ProbleMs On gRouPs inpUT as mUltiplIcation tAbles*]{}, j. comput. System ScI. [**63**]{}(2001), 186–200.
j.-C. Birget, A. Yu. Ol’SHaNSKiI, e. RipS, anD M. V. Sapir, [*IsoPeriMEtriC funCTiOns OF grouPs and CoMPuTAtional complexity of ThE word pRobleM*]{}, [Ann. Math.]{} [**156**]{}(2002), 467–518.
W. W. BooNe, [*On certaiN SIMple undeCidaBLe PRoblems in group TheorY, V, VI*]{}, Indag. MATh. [**19**]{}(1957), 22–27, 227–232.
W. W. BoonE, [*The wOrd problEm*]{}, Ann. Math. [**70**]{}(1959), 207–265.
v. v. borisov, [*SImpLe | [@CB4], use polynomial sp ace. Thisappro ach wo ul d gi ve a chance to do, also in polylogarithmic sp ace,ma x imiz a ti on of foldin g sr e lat iv etova r io us pa ram eters s imilar totho se discussed i n T heorem \[t hm3 \].
[*Ackno wle dgemen ts .*] { } The au thoris gra t eful t o Tim Ril ey for br i nging t o th e au thor’s attentionf ol k lore arguments thatso l ve t hepre cise wordpr oblem for pre s en t a t ion $\langle \, a , b \, \|\, a=1, b =1\ , \ra ngle$ i n po lynomial ti me,the artic le [@S L 1], and the sim ilarit y b etw eent he [ pre ci s e w o rd pr o ble m]{} for p re senta tion $ \ l angl e \ , a, b \, \| \, a=1 , b=1\ , \rang le$ a nd t he prob lem on effi ci ent planar fold ingof RNA st ran ds . T he auth o r than ksthe refere e for a num be r o fuseful remarks and s u g ge stions.
[\[75 \ ]] {} A. V. An is imo v, [ * T he gr oupl an guages*] {}, Ki b er ne tika (K ie v) [** 4* *]{ }(1 971), 18–2 4.
R. Armoni, A. T a -Shma, A. Wigd e rson and S. Z h ou , [* A $(\ log n )^{4/3}$ spa c e al gori t hm fo r $(s, t)$ c on n ec t ivity in undirected g raphs* ]{},J. Assoc. Com put. Mach. [ * *47**]{} (200 0 ), 294–311.
S. A roraand B. Bar a k, [*Com putat ional co mplexity– a modern ap pro ach *]{ } , C ambridge Univ . Pres s, 2009.
D. Barrin gto n,P.Kad au , K. Lang e and P. M cK en zi e,[*Ont he compl ex ity o f s ome p r oblems on g roup sin p utas mult i pl i c atio nta bles *]{ }, J. C ompu t . S ystem S ci. [**63 **] { }(20 01 ), 186–20 0.
J.-C. Bir ge t, A. Yu.Ol ’sh anskii , E. Rips, and M. V. Sapir, [*Iso p erimetr icfunct ions of group s a nd com put a tional compl exity o f t h e word p ro ble m* ]{}, [Ann. M ath .]{}[* *156 **]{}(2 002), 467–518.
W. W.Boone, [*On c ert ains i mp leu nd e cid ab l e p r o blems in grouptheory, V, V I *] {}, Indag. Mat h. [**19* *]{}(19 57),2 2–27, 2 27–232.
W. W. Boo ne , [* T h e w ord proble m*]{}, A nn. Math. [**70 * *] {}(19 59) , 207– 26 5.
V. V . Bori s ov, [*Si mple | [@CB4],_use polynomial_space. This approach would_give a_chance_to do,_also_in polylogarithmic space,_maximization of foldings_relative to various parameters_similar to those_discussed_in Theorem \[thm3\].
[*Acknowledgements.*]{} The author is grateful to Tim Riley for bringing to the author’s_attention_folklore arguments_that_solve_the precise word problem for_presentation $\langle \, a, b_\, _\| \, a=1, b=1 \,__\rangle$ in polynomial_time, the article [@SL1], and the similarity between the_[precise word problem]{} for presentation $\langle \,_a, b \,__\|_ \, a=1,_ b=1 \, \rangle$ and_the problem on efficient planar folding_of RNA strands. The author thanks the_referee for a number of useful_remarks and suggestions.
[\[75\]]{} A. V._Anisimov, [*The_group languages*]{}, Kibernetika (Kiev) [**4**]{}(1971),_18–24.
R. Armoni, A._Ta-Shma, A._Wigderson and S._Zhou, [*A $(\log n )^{4/3}$ space_algorithm for $(s,t)$_connectivity in undirected graphs*]{}, J. Assoc._Comput._Mach. [**47**]{}(2000), 294–311.
S._Arora_and_B. Barak,_[*Computational complexity –_a_modern approach*]{},_Cambridge_Univ. Press, 2009.
D. Barrington, P. Kadau,_K._Lange and P. McKenzie, [*On the complexity_of some problems on_groups_input as multiplication tables*]{},_J. Comput. System Sci. [**63**]{}(2001),_186–200.
J.-C. Birget, A. Yu. Ol’shanskii, E._Rips, and_M. V._Sapir, [*Isoperimetric functions of groups and computational complexity of the word_problem*]{}, [Ann. Math.]{} [**156**]{}(2002), 467–518.
W. W._Boone, [*On certain simple_undecidable problems_in_group theory, V,_VI*]{},_Indag. Math._[**19**]{}(1957), 22–27, 227–232.
W. W. Boone, [*The word_problem*]{}, Ann._Math. [**70**]{}(1959), 207–265.
V. V. Borisov, [*Simple |
\rho^{(S)}_s\bb{p} (\bm{n} \cdot \hat{\bm{\sigma}}) \nonumber \\
\qquad \qquad\qquad\qquad\qquad\qquad\qquad - (-1)^s \sinh{(\omega)} \frac{E_p - m}{E_p} \{ \bm{n} \cdot \hat{\bm{\sigma}}, \rho^{(S)}_s\bb{p} \}\Big],\end{aligned}$$ where $\{\,\,,\,\,\}$ denotes anti-commutators, and which, in the limit $E_p - m \simeq E_p$, can be subtly simplified as to give a transformation law in the form of $\rho^{\prime(S)}_s\bb{p^\prime} =\hat{O} \, \rho^{(S)}_s\bb{p} \, \hat{O}^\dagger$, where $\hat{O}$ is the unitary operator $$\hat{O} = \frac{1}{\sqrt{\cosh{(\omega)}}} \left[ \cosh{\left(\frac{\omega}{2} \right)} \hat{I}_2 - \sinh{\left(\frac{\omega}{2} \right)} (\bm{n} \cdot \hat{\bm{\sigma}}) \left({\bm{e}}_{p} \cdot \hat{\bm{\sigma}} \right) \right].$$ In fact, such transformation under a Lorentz boost is the same as that one obtained for states belonging to the [*irrep*]{} $(+, \frac{1}{2})$ of the Poincaré group, which can be recast in terms of a momentum dependent rotation and which is the basis of several results in relativistic quantum information[^6].
Considering the generic two-particle state (\[2partDM\]) in a reference frame $\mathcal{S}$, the transformed density matrix describing the state in an inertial frame $\mathcal{S}^{\prime}$, related to $\mathcal{S}$ by a Lorentz boost, $\Lambda$, is given by $$\begin{aligned}
\rho\bb{\bm{p}, \bm{q}} \rightarrow \rho^\prime\bb{\bm{p}^\prime, \bm{q}^\prime} &=& \frac{1}{\nu} \big( \, \hat{S}^{A}[\Lambda] \otimes \hat{S}^{B}[\Lambda] \,\big) \,\rho\bb{\bm{p}, \bm{q}} \, \big( \, (\ | \rho^{(S)}_s\bb{p } (\bm{n } \cdot \hat{\bm{\sigma } }) \nonumber \\
\qquad \qquad\qquad\qquad\qquad\qquad\qquad - (-1)^s \sinh{(\omega) } \frac{E_p - m}{E_p } \ { \bm{n } \cdot \hat{\bm{\sigma } }, \rho^{(S)}_s\bb{p } \}\Big],\end{aligned}$$ where $ \{\,\,,\,\,\}$ denotes anti - commutators, and which, in the limit $ E_p - m \simeq E_p$, can be subtly simplified as to render a transformation jurisprudence in the form of $ \rho^{\prime(S)}_s\bb{p^\prime } = \hat{O } \, \rho^{(S)}_s\bb{p } \, \hat{O}^\dagger$, where $ \hat{O}$ is the unitary operator $ $ \hat{O } = \frac{1}{\sqrt{\cosh{(\omega) } } } \left [ \cosh{\left(\frac{\omega}{2 } \right) } \hat{I}_2 - \sinh{\left(\frac{\omega}{2 } \right) } (\bm{n } \cdot \hat{\bm{\sigma } }) \left({\bm{e}}_{p } \cdot \hat{\bm{\sigma } } \right) \right].$$ In fact, such transformation under a Lorentz rise is the same as that one obtained for states belong to the [ * irrep * ] { } $ (+, \frac{1}{2})$ of the Poincaré group, which can be recast in terms of a momentum dependent rotation and which is the footing of respective results in relativistic quantum information[^6 ].
Considering the generic two - atom state (\[2partDM\ ]) in a reference frame of reference $ \mathcal{S}$, the transformed density matrix describing the state in an inertial frame $ \mathcal{S}^{\prime}$, related to $ \mathcal{S}$ by a Lorentz rise, $ \Lambda$, is given by $ $ \begin{aligned }
\rho\bb{\bm{p }, \bm{q } } \rightarrow \rho^\prime\bb{\bm{p}^\prime, \bm{q}^\prime } & = & \frac{1}{\nu } \big (\, \hat{S}^{A}[\Lambda ] \otimes \hat{S}^{B}[\Lambda ] \,\big) \,\rho\bb{\bm{p }, \bm{q } } \, \big (\, (\ | \rhl^{(S)}_s\bb{p} (\bm{n} \cdot \hat{\bm{\sigoa}}) \nonumber \\
\qquce \qquav\qquad\qsuad\qquaa\qquad\qquad - (-1)^s \sinh{(\omega)} \frar{E_p - m}{E_p} \{ \bm{n} \cdot \hat{\bm{\sigma}}, \rfo^{(S)}_s\bb{p} \}\Bpg],\end{aligbed}$$ xhere $\{\,\,,\,\,\}$ denotes aifi-commubctors, wnd chmch, in the limij $E_p - m \simex E_p$, can be suttuy simplified as to give a transformaeion lae ln the form of $\rho^{\kriie(S)}_s\gb{p^\prime} =\hat{O} \, \rho^{(S)}_s\bb{p} \, \hat{O}^\dagged$, where $\hat{O}$ is the initary operator $$\hat{O} = \fraf{1}{\sqrh{\cosh{(\omega)}}} \left[ \codh{\left(\frac{\onega}{2} \eight)} \hat{I}_2 - \rinh{\left(\frac{\omega}{2} \righj)} (\bm{n} \cdot \hat{\bm{\sigma}}) \left({\bm{e}}_{p} \caot \hct{\bm{\sigma}} \rugyt) \ghght].$$ In facv, such transformatlpn undar a Lotentz boost is thx sane as that one obtainxd for states belongyng to tha [*nrrep*]{} $(+, \frac{1}{2})$ of the Pounxaré gtoup, fhicf cav bt rxcaat in herjs of a mojentum depebdent rotation and ehybn is the basjs of federal results in relativistic quantum itfodmation[^6].
Considering the teneric two-particle sjate (\[2partDI\]) in a reference frame $\mathcal{S}$, the transformed dansitb oatxlw derxrlbing the state in an inertial frame $\mathcal{S}^{\[dike}$, related to $\mabhcal{S}$ by a Lorenta hopft, $\Lambda$, is eiven yg $$\gegin{aligned}
\rho\bb{\bl{p}, \bm{q}} \tightaerow \rho^\pwime\nb{\bm{p}^\prime, \bm{q}^\prime} &=& \frac{1}{\nu} \big( \, \hat{S}^{A}[\Lcmbea] \otimes \hat{S}^{B}[\Lamyda] \,\big) \,\rho\by{\bm{p}, \bk{q}} \, \bog( \, (\ | \rho^{(S)}_s\bb{p} (\bm{n} \cdot \hat{\bm{\sigma}}) \nonumber \\ \qquad (-1)^s \frac{E_p - \{ \bm{n} \cdot denotes and which, in limit $E_p - \simeq E_p$, can be subtly simplified to give a transformation law in the form of $\rho^{\prime(S)}_s\bb{p^\prime} =\hat{O} \, \rho^{(S)}_s\bb{p} \hat{O}^\dagger$, where $\hat{O}$ is the unitary operator $$\hat{O} = \frac{1}{\sqrt{\cosh{(\omega)}}} \left[ \cosh{\left(\frac{\omega}{2} \right)} - \right)} \cdot \left({\bm{e}}_{p} \cdot \hat{\bm{\sigma}} \right) \right].$$ In fact, such transformation under a Lorentz boost is the same that one obtained for states belonging to the $(+, \frac{1}{2})$ of the group, which can be recast terms a momentum rotation which the basis of results in relativistic quantum information[^6]. Considering the generic two-particle state (\[2partDM\]) in a reference frame $\mathcal{S}$, the density matrix state in inertial $\mathcal{S}^{\prime}$, to $\mathcal{S}$ by boost, $\Lambda$, is given by $$\begin{aligned} \rho^\prime\bb{\bm{p}^\prime, \bm{q}^\prime} &=& \frac{1}{\nu} \big( \, \hat{S}^{A}[\Lambda] \otimes \,\big) \,\rho\bb{\bm{p}, \, \big( \, (\ | \rho^{(S)}_s\bb{p} (\bm{n} \cdot \hat{\bm{\sigma}}) \Nonumber \\
\qqUad \qqUad\QquAd\QquaD\qquAd\qquad\qquad - (-1)^s \sINh{(\omEga)} \frac{E_p - m}{E_p} \{ \bm{n} \cdot \hat{\Bm{\sigMa}}, \RHo^{(S)}_s\BB{p} \}\big],\enD{aligneD}$$ WhERE $\{\,\,,\,\,\}$ deNoTeS anTi-COmMutatOrs, And whicH, in the limiT $E_p - M \sImeq E_p$, can be sUBtLy simplifiEd aS to give a tranSfoRmatioN lAw iN The foRm oF $\rho^{\pRime(S)}_s\BB{p^\primE} =\hat{O} \, \rho^{(S)}_S\bB{P} \, \hat{O}^\dAGger$, wheRE $\HaT{O}$ is The unitary operatoR $$\HaT{o} = \frac{1}{\sqrt{\cosh{(\oMega)}}} \leFt[ \COsH{\LEft(\FraC{\omega}{2} \righT)} \hAt{I}_2 - \siNH{\left(\frAC{\oMEGA}{2} \riGHt)} (\bm{n} \cdot \hat{\bM{\sigma}}) \left({\bM{E}}_{p} \cDot \hat{\Bm{\SigMA}} \right) \Right].$$ in FAct, Such transfoRmatIon under a lorentZ Boost is THe same aS that oNe oBtaIned FOr StAteS bELonGInG to THe [*iRrep*]{} $(+, \frac{1}{2})$ Of ThE PoinCaré GROUP, whiCh cAn be RecasT in terms of a moMenTum dEPenDent rOtatiOn anD wHich iS the baSis of SeVeral results in rElatIvistic quAntUm InfOrMatioN[^6].
considEriNg tHe generIc two-paRTicLe STATe (\[2PartDM\]) in a reference FrAME $\mAthcal{S}$, tHe tranSFoRmED density MaTriX desCRIbing The sTAtE in an ineRtial fRAmE $\mAthcal{S}^{\PrIme}$, relAtEd tO $\maThcal{s}$ By a LOrentz Boost, $\LamBda$, is GIven by $$\begin{aliGNed}
\rho\bb{\bm{p}, \bm{Q}} \RiGHTaRRow \rHo^\pRime\bb{\bm{p}^\prIme, \bM{Q}^\priMe} &=& \frAC{1}{\nU} \biG( \, \Hat{S}^{A}[\lambdA] \oTImES \hat{S}^{B}[\Lambda] \,\big) \,\rho\bB{\bM{p}, \bm{q}} \, \bIg( \, (\ | \rho^{(S)}_s\bb{p} (\bm{n } \cdot \h at{\b m{\ sig ma }})\non umber \\
\qqua d \qq uad\qquad\qquad\qquad\ qquad \q q uad- ( -1)^s \sinh{ ( \o m e ga) }\f rac {E _ p- m}{ E_p } \{ \b m{n} \cdot \h at {\bm{\sigma} } ,\rho^{(S)} _s\ bb{p} \}\Big ],\ end{al ig ned } $$ wh ere $\{\ ,\,,\, \ ,\}$ d enotes an ti - commut a tors, a n d w hich , in the limit $E _ p- m \simeq E_p$ , canbe su b t lysim plified as t o giv e a tran s fo r m a tio n law in the f orm of $\rh o ^{\ prime( S) }_s \ bb{p^\ prime }= \ha t{O} \, \rh o^{( S)}_s\bb{ p} \,\ hat{O}^ \ dagger$ , wher e $ \ha t{O} $ i sthe u n ita r yope r ato r $$\hat {O }= \fr ac{1 } { \ s qrt{ \co sh{( \omeg a)}}} \left[\co sh{\ l eft (\fra c{\om ega} {2 } \ri ght)}\hat{ I} _2 - \sinh{\lef t(\f rac{\omeg a}{ 2} \r ig ht)}( \bm{n} \c dot \hat{\ bm{\sig m a}} )\ l e ft ({\bm{e}}_{p} \cdo t\ h at {\bm{\si gma}}\ ri gh t ) \right ]. $$In f a c t, su ch t r an sformati on und e raLorentz b oost i sthe sa me as that one o btainedfor s t ates belonging to the [*irre p *] { } $ ( +, \ fra c{1}{2})$ o f th e Poi ncar é g rou p , whi ch ca nb er ecast in terms of a m omentu m dep endent rotati on and whi c h is the b asis of several result s inrelativist i c quantu m inf ormation [^6].
Co n s ideringthe ge ner ict w o- particle stat e (\[2 pa rtDM\]) in a refe ren cefra me$\ mathcal{S }$, thetr an sf or med dens i ty matri xdes cr ibi ng th e state in a n in er ti a l f rame $\ m at h c al{S }^ {\ prim e}$ ,relat ed t o $\ mathcal {S}$ by a Lo r entz b oo st, $\L ambda$, is gi ve n by $$\be gi n{a ligned } \rho\bb{ \bm{p}, \bm{q}} \righta r row \rh o^\ prime \bb{ \bm{p}^\p rim e, \bm {q} ^ \prime } &=&\frac {1 }{\ n u } \bi g ( \ , \ ha t{S}^{A}[\ L a mbd a] \o ti mes\hat{S} ^{B}[\Lambda] \,\b i g)\,\rho\bb{\bm {p} , \b m { q} } \ , \ b ig( \ , (\ | \rho^{(S)}_s\bb{p}_(\bm{n} \cdot_\hat{\bm{\sigma}}) \nonumber \\
\qquad \qquad\qquad\qquad\qquad\qquad\qquad_- (-1)^s_\sinh{(\omega)}_\frac{E_p -_m}{E_p}_\{ \bm{n} \cdot_\hat{\bm{\sigma}}, \rho^{(S)}_s\bb{p} \}\Big],\end{aligned}$$_where $\{\,\,,\,\,\}$ denotes anti-commutators,_and which, in_the_limit $E_p - m \simeq E_p$, can be subtly simplified as to give a_transformation_law in_the_form_of $\rho^{\prime(S)}_s\bb{p^\prime} =\hat{O} \, \rho^{(S)}_s\bb{p}_\, \hat{O}^\dagger$, where $\hat{O}$ is_the unitary_operator $$\hat{O} = \frac{1}{\sqrt{\cosh{(\omega)}}} \left[ \cosh{\left(\frac{\omega}{2} \right)} \hat{I}_2_-_\sinh{\left(\frac{\omega}{2} \right)} (\bm{n}_\cdot \hat{\bm{\sigma}}) \left({\bm{e}}_{p} \cdot \hat{\bm{\sigma}} \right) \right].$$ In fact,_such transformation under a Lorentz boost_is the same_as_that_one obtained for states_belonging to the [*irrep*]{} $(+, \frac{1}{2})$_of the Poincaré group, which can_be recast in terms of a momentum_dependent rotation and which is the_basis of several results in_relativistic quantum_information[^6].
Considering the generic two-particle state_(\[2partDM\]) in a_reference frame_$\mathcal{S}$, the transformed_density matrix describing the state in_an inertial frame_$\mathcal{S}^{\prime}$, related to $\mathcal{S}$ by a_Lorentz_boost, $\Lambda$, is_given_by_$$\begin{aligned}
\rho\bb{\bm{p}, \bm{q}}_\rightarrow \rho^\prime\bb{\bm{p}^\prime, \bm{q}^\prime}_&=&_\frac{1}{\nu} \big(_\,_\hat{S}^{A}[\Lambda] \otimes \hat{S}^{B}[\Lambda] \,\big) \,\rho\bb{\bm{p}, \bm{q}}_\,_\big( \, (\ |
its action contains higher derivatives. Combining the Horndeski gravity with the mimetic idea, the mimetic constraint will eliminate the dynamics and degrade the dynamical component to a constraint. This is consistent with the result of Ref. [@Arroja:2015yvd], in which the perturbation equations of the mimetic Horndeski gravity have been studied. In general case, $\alpha+\beta\neq 0$, one can see from Eq. (\[equationshift1\]) that the divergence of $N^i$ is proportional to the time derivative of the curvature perturbation and has no relation with its spatial derivatives. This is the origin of the gradient instability in the model considered in this section. Substitute Eq. (\[equationshift1\]) to the action (\[mimeticbox2\]) and using the background equation we get the quadratic actions for the cosmological perturbations of this model, \[quadratic\] S\^[(2)]{}=d\^4x a\^3\[\_[ij]{}\_[ij]{}-\_l\_[ij]{}\_l\_[ij]{}+\^2+\_i\_i\] . So, to have no ghost, we need $f+2\beta>0$, and $(3\alpha+\beta-f)/(\alpha+\beta)>0$. This is possible by appropriate choices of the function $f$ and the parameters $\alpha$ and $\beta$. However, if we requires the tensor perturbation being free from the gradient instability, $f$ should be positive, and this will cause the gradient instability in the scalar perturbation. The quadratic action (\[quadratic\]) tells us that this difficulty cannot be cured by simply adding more higher derivatives in the original action. This is because the spatial derivative terms of the scalar and tensor perturbations have nothing to do with higher derivative terms of the mimetic field, they are totally come from the three dimensional curvature $^{(3)}R$, as pointed out in Ref. [@Ijjas:2016pad] for the model(\[mimeticbox\]). This is easily seen in the unitary gauge. As we mentioned before, higher derivative terms of the mimetic field are identical to the extrinsic curvature terms which by definitions depend on the time derivatives of $\zeta$ and $\gamma_{ij}$ (through $\dot h_{ij}$ in the definitions of $E_{ij}$) and the divergence of the shift vector $N^i$. The result (\[equationshift1\]) showed that this divergence also has nothing to do with the spatial derivatives of the scalar and tensor perturbations. So, the $\ | its action contains higher derivatives. compound the Horndeski graveness with the mimetic idea, the mimetic constraint will rule out the moral force and degrade the dynamical component to a restraint. This is coherent with the result of Ref. [ @Arroja:2015yvd ], in which the disturbance equations of the mimetic Horndeski gravity have been learn. In general case, $ \alpha+\beta\neq 0 $, one can visualize from Eq. (\[equationshift1\ ]) that the divergence of $ N^i$ is proportional to the time derivative of the curvature perturbation and has no relation with its spatial derived function. This is the origin of the gradient instability in the model consider in this section. Substitute Eq. (\[equationshift1\ ]) to the action (\[mimeticbox2\ ]) and using the background equality we get the quadratic natural process for the cosmological perturbations of this model, \[quadratic\ ] S\^[(2)]{}=d\^4x a\^3\[\_[ij]{}\_[ij]{}-\_l\_[ij]{}\_l\_[ij]{}+\^2+\_i\_i\ ] . therefore, to have no ghost, we need $ f+2\beta>0 $, and $ (3\alpha+\beta - f)/(\alpha+\beta)>0$. This is possible by appropriate choices of the function $ f$ and the parameters $ \alpha$ and $ \beta$. However, if we requires the tensor perturbation being free from the gradient instability, $ f$ should be positive, and this will cause the gradient instability in the scalar perturbation. The quadratic action (\[quadratic\ ]) tells us that this trouble cannot be bring around by simply lend more eminent derivatives in the original action. This is because the spatial derivative terms of the scalar and tensor perturbations have nothing to do with higher derivative term of the mimetic field, they are totally come from the three dimensional curvature $ ^{(3)}R$, as pointed out in Ref. [ @Ijjas:2016pad ] for the model(\[mimeticbox\ ]). This is easily seen in the one gauge. As we mentioned before, higher derivative terms of the mimetic field are identical to the extrinsic curvature terms which by definition depend on the time derivatives of $ \zeta$ and $ \gamma_{ij}$ (through $ \dot h_{ij}$ in the definitions of $ E_{ij}$) and the divergence of the chemise vector $ N^i$. The result (\[equationshift1\ ]) showed that this divergence also has nothing to do with the spatial derivatives of the scalar and tensor perturbations. indeed, the $ \ | itd action contains higher derivatives. Combining the Hkrndeski gravity with the mimetic idxa, tye minetic constraint will dliminate the dynqmicw and degravs the dnuamiczp cok'onent to a conxtraint. Thhs is consistett wnth the result of Ref. [@Arroja:2015yvd], in wrich thr oerturbation ezuatpogs or the mimetic Horndeski gravity habe been studied. In grneral case, $\alpha+\beta\neq 0$, lne fan see from Eq. (\[eqkationshift1\]) thae the divergevce of $N^i$ pv proportiknal to the time derivative of ghe cbrvature petcyrbwjion and has no rvlation with lns spathal derovatives. This ls thx orugin of the gradient mnstability in the mjdel conshdzred in this section. Wuvstitote Ex. (\[eqjqtiunsgiht1\]) fo the acvion (\[mimetidbox2\]) and usung the background tquwnoon we get tge quaqrwtic actions for the cosmological pertugbatjons of this model, \[quadeatic\] S\^[(2)]{}=d\^4x a\^3\[\_[ij]{}\_[ij]{}-\_l\_[ij]{}\_l\_[ij]{}+\^2+\_l\_i\] . So, to rave no ghost, we need $f+2\beta>0$, and $(3\alpha+\beta-f)/(\alpha+\beda)>0$. Thms is pjrwihle by appropriate choices of the function $f$ whd tme parameters $\aliha$ and $\beta$. Howevrr, ig we requires jhe tensor perturbation beinh free srom rhe graditnt imstability, $f$ should be posirive, and thif will cause the grcdient instayility in tne scalar perturbation. Che quzdratic actlon (\[quadrzgic\]) tells us thag tmis difficulty cannot be cureq by simpoy abding mofe hogher qerivatived in bve original action. This iv because hhe spatial derivative terms of the scalar anc denvor pertbrbatipns have nothyng to do with higher derixative terjs of tie mimetic fyeld, they are jotally come hrom the ehrew dinensionxu curvature $^{(3)}R$, ss pointeb out in Ref. [@Ijjas:2016pad] fpr jhs model(\[mimeticbix\]). Uhus is easily sren in tje ughtary gauge. Ds wd mdmtiondd before, hlghdr drrivative terms of tve mjmetic field are icektical to the extwinsic curvatire terms which by defiiitionr deprnd on the time derivatives of $\zefa$ and $\galma_{lj}$ (through $\doe h_{im}$ in the definntions of $E_{ij}$) and the divergence of the whift vector $N^i$. Thw result (\[equationshnfu1\]) showed that this divergenwe also has nothing ro do with the spstial derivatives of fhe scdlar wnd tensor perturbations. So, the $\ | its action contains higher derivatives. Combining the with mimetic idea, mimetic constraint will the component to a This is consistent the result of Ref. [@Arroja:2015yvd], in the perturbation equations of the mimetic Horndeski gravity have been studied. In general $\alpha+\beta\neq 0$, one can see from Eq. (\[equationshift1\]) that the divergence of $N^i$ proportional the derivative the curvature perturbation and has no relation with its spatial derivatives. This is the origin of gradient instability in the model considered in this Substitute Eq. (\[equationshift1\]) to action (\[mimeticbox2\]) and using the equation get the actions the perturbations of this \[quadratic\] S\^[(2)]{}=d\^4x a\^3\[\_[ij]{}\_[ij]{}-\_l\_[ij]{}\_l\_[ij]{}+\^2+\_i\_i\] . So, to have no ghost, we need $f+2\beta>0$, and $(3\alpha+\beta-f)/(\alpha+\beta)>0$. This is possible appropriate choices function $f$ the $\alpha$ $\beta$. However, if the tensor perturbation being free from $f$ should be positive, and this will cause gradient instability the scalar perturbation. The quadratic action tells us that this difficulty cannot be cured simply adding more higher derivatives in the original action. This is because the spatial derivative the scalar and tensor have nothing to with derivative of mimetic field, are totally come from the three dimensional curvature $^{(3)}R$, as pointed in Ref. [@Ijjas:2016pad] for the model(\[mimeticbox\]). This is easily seen unitary As we mentioned higher derivative terms of mimetic are identical to the terms by the derivatives $\zeta$ and $\gamma_{ij}$ (through h_{ij}$ in the definitions of and the divergence of result (\[equationshift1\]) showed that this divergence also has to do with the spatial derivatives of scalar and tensor perturbations. So, the $\ | its action contains higher deRivatives. COmbinIng The hoRndeSki gRavity with the mIMetiC idea, the mimetic constraInt wiLl ELimiNAtE the dYnamics ANd DEGraDe ThE dyNaMIcAl comPonEnt to a cOnstraint. THis Is Consistent wiTH tHe result of ref. [@arroja:2015yvd], in wHicH the peRtUrbATion eQuaTions Of the mIMetic HOrndeski gRaVIty havE Been stuDIEd. in geNeral case, $\alpha+\betA\NeQ 0$, One can see from EQ. (\[equatIoNShIFT1\]) thAt tHe divergenCe Of $N^i$ iS ProportIOnAL TO thE Time derivativE of the curvaTUre PerturBaTioN And has No relAtIOn wIth its spatiAl deRivatives. this is THe origiN Of the grAdient InsTabIlitY In ThE moDeL ConSIdEreD In tHis sectiOn. suBstitUte EQ. (\[EQUAtioNshIft1\]) tO the aCtion (\[mimeticbOx2\]) aNd usINg tHe bacKgrouNd eqUaTion wE get thE quadRaTic actions for thE cosMological PerTuRbaTiOns of THis modEl, \[qUadRatic\] S\^[(2)]{}=d\^4X a\^3\[\_[ij]{}\_[ij]{}-\_l\_[IJ]{}\_l\_[iJ]{}+\^2+\_i\_I\] . sO, To Have no ghost, we need $f+2\BeTA>0$, AnD $(3\alpha+\beTa-f)/(\alpHA+\bEtA)>0$. this is poSsIblE by aPPRopriAte cHOiCes of the FunctiON $f$ AnD the parAmEters $\aLpHa$ aNd $\bEta$. HoWEver, If we reQuires thE tensOR perturbation bEIng free from thE GrADIeNT insTabIlity, $f$ shoulD be pOSitiVe, anD ThIs wILl cauSe the GrADiENt instability in the sCaLar perTurbaTion. The quadraTic action (\[qUADRatic\]) telLs us THaT This difficulty CannoT be cured by SImply addIng moRe higher DerivativES In the oriGinAl aCtiOn. THIS iS because the spATIal dErIvative TerMs of the ScaLar And TenSoR perturbaTions havE nOtHiNg To dO with HIgher derIvAtiVe TerMs of tHE mimetIc fieLd, thEy ArE TotAlly comE FrOM The tHrEe DimeNsiOnAl curVatuRE $^{(3)}R$, aS pointeD out in Ref. [@ijjAS:2016pad] FoR tHe model(\[Mimeticbox\]). ThiS iS easily seeN iN thE unitaRY Gauge. As wE mentioned before, higher dERivativE teRms of The mImetic fieLd aRe idenTicAL to the ExtrinSic cuRvAtuRE Terms WHIcH by DeFinitions dEPEnd On the TiMe deRivativEs of $\zeta$ and $\gamma_{ij}$ (THroUgh $\dot h_{ij}$ in thE deFiniTIOnS of $e_{Ij}$) ANd tHe DIveRGEnce of the shift vEctor $N^i$. The ReSUlT (\[equationsHIft1\]) ShOwed thaT this diVergeNCe also hAs nothing To do with tHe SpatIAL deRivatives oF the scalAr and tensOR pertURbAtionS. So, The $\ | its action contains highe r derivati ves.Com bin in g th e Ho rndeski gravit y wit h the mimetic idea, th e mim et i c co n st raint will e l im i n ate t he dy na m ic s and de grade t he dynamic alco mponent to a co nstraint.Thi s is consist ent withth e r e sultofRef.[@Arro j a:2015 yvd], inwh i ch the perturb a t io n eq uations of the mi m et i c Horndeski gr avityha v eb e enstu died. In g en eralc ase, $\ a lp h a + \be t a\neq 0$, one can see fr o m E q. (\[ eq uat i onshif t1\]) t h atthe diverge nceof $N^i$is pro p ortiona l to the timeder iva tive of t hecu r vat u re pe r tur bation a nd h as no rel a t i o n wi thitsspati al derivative s.This isthe o rigin ofth e gra dientinsta bi lity in the mod el c onsidered in t his s ectio n . Subs tit ute Eq. (\ [equati o nsh if t 1 \ ]) to the action (\[ mi m e ti cbox2\]) and u s in gt he backg ro und equ a t ion w e ge t t he quadr atic a c ti on s for t he cosmo lo gic alpertu r bati ons of this mo del,\ [quadratic\] S \ ^[(2)]{}=d\^4 x a \ ^ 3\ [ \_[i j]{ }\_[ij]{}-\ _l\_ [ ij]{ }\_l \ _[ ij] { }+\^2 +\_i\ _i \ ]. So, to have no gho st , we n eed $ f+2\beta>0$,and $(3\al p h a +\beta-f )/(\ a lp h a+\beta)>0$. T his i s possible by appro priat e choice s of thef u nction $ f$and th e p a r am eters $\alpha $ and$\ beta$.How ever, i f w e r equ ire sthe tenso r pertur ba ti on b ein g fre e from th egra di ent inst a bility , $f$ sho ul db e p ositive , a n d thi swi ll c aus ethe g radi e ntinstabi lity in t hes cala rpe rturbat ion. The quad ra tic action ( \[q uadrat i c \]) tell s us that this difficul t y canno t b e cur ed b y simplyadd ing mo reh igherderiva tives i n t h e orig i n al ac ti on. This i s bec auseth e sp atial d erivative terms of the scalar and t ens or p e r tu rba t io n s h av e no t h ing to do withhigher der iv a ti ve terms o f th emimetic field, they are tot ally come from the t hree d ime nsional cu rvature$^{(3)}R$ , as p o in ted o utin Ref .[@I jjas: 2016pa d ] f or th e mode l( \[mime ticbo x\ ]). This is easily seen in theunitar y gau ge. As we me nti o ned before,high er derivat ive te rms o f t h e mim etic fi eld are i dent i cal to th e e xtr i n si c curvature t e rms whic h b y defin itio ns depend on thet ime derivative s of $ \ze ta$ and$\ gamma_{ij}$ (t hro ug h $\dot h_ {i j}$ in thedefiniti on s of $ E_{ij} $) and the di v e rg e nce of the sh ift vecto r $ N^ i $. There su l t (\[e quat io nshift 1\]) s h owed t hat this diverge nce a l s o has not hingto do wit h the spatial d erivativesof the sca lar a nd tens or pertu rba ti ons. So, t h e $\ | its_action contains_higher derivatives. Combining the_Horndeski gravity_with_the mimetic_idea,_the mimetic constraint_will eliminate the_dynamics and degrade the_dynamical component to_a_constraint. This is consistent with the result of Ref. [@Arroja:2015yvd], in which the perturbation_equations_of the_mimetic_Horndeski_gravity have been studied. In_general case, $\alpha+\beta\neq 0$, one_can see_from Eq. (\[equationshift1\]) that the divergence of $N^i$_is_proportional to the_time derivative of the curvature perturbation and has no_relation with its spatial derivatives. This_is the origin_of_the_gradient instability in the_model considered in this section. Substitute_Eq. (\[equationshift1\]) to the action (\[mimeticbox2\])_and using the background equation we get_the quadratic actions for the cosmological_perturbations of this model, \[quadratic\]_S\^[(2)]{}=d\^4x a\^3\[\_[ij]{}\_[ij]{}-\_l\_[ij]{}\_l\_[ij]{}+\^2+\_i\_i\] ._So, to have no ghost,_we need $f+2\beta>0$,_and $(3\alpha+\beta-f)/(\alpha+\beta)>0$._This is possible_by appropriate choices of the function_$f$ and the_parameters $\alpha$ and $\beta$. However, if_we_requires the tensor_perturbation_being_free from_the gradient instability,_$f$_should be_positive,_and this will cause the gradient_instability_in the scalar perturbation. The quadratic action_(\[quadratic\]) tells us that_this_difficulty cannot be cured_by simply adding more higher_derivatives in the original action. This_is because_the spatial_derivative terms of the scalar and tensor perturbations have nothing to_do with higher derivative terms of_the mimetic field, they_are totally_come_from the three_dimensional_curvature $^{(3)}R$,_as pointed out in Ref. [@Ijjas:2016pad] for_the model(\[mimeticbox\])._This is easily seen in the_unitary gauge. As we_mentioned_before, higher derivative terms of the_mimetic field are identical to the_extrinsic curvature terms which by_definitions_depend_on the time derivatives of_$\zeta$ and $\gamma_{ij}$ (through $\dot h_{ij}$_in the definitions_of $E_{ij}$) and the divergence of the_shift_vector $N^i$. The result (\[equationshift1\]) showed_that_this divergence also has nothing to_do_with_the spatial derivatives of the_scalar and tensor perturbations. So, the_$\ |
2^{h'}
= H_1(ID^*)^{s_2} g_2^{r' + h'}
= H_1(ID^*)^{s_2} g_2^{r^*}
\end{aligned}$$ where $h^* = H_2(ID^* \| M^*)$ and $r^* = r' + h'$. To extract the CDH value from the forged signature of ${\mathcal{F}}$ by using Forking Lemma, the simulator of Yuan [*et al.*]{} should obtain two valid signatures $\sigma_1^* = (U_1^*, V_1^*, W_1^*)$ and $\sigma_2^* = (U_2^*,
V_2^*, W_2^*)$ on the same identity and message pair $(ID^*, M^*)$ such that $U_1^* = U_2^*$ and $h_1^* \neq h_2^*$ after replaying ${\mathcal{F}}$ with the same random tape but different choices of the hash oracle $H_2$. Let $\sigma_1^* =
(U_1^*, V_1^*, W_1^*)$ and $\sigma_2^* = (U_2^*, V_2^*, W_2^*)$ be the two valid signatures obtained from ${\mathcal{F}}$ by using Forking Lemma and $\sigma'_1
= (U'_1, V'_1, W'_1)$ and $\sigma'_2 = (U'_2, V'_2, W'_2)$ be the original signatures before the re-randomization of ${\mathcal{F}}$. If $h_1^* \neq h_2^*$, then $h'_1 \neq h'_2$ except negligible probability since $H'$ is a collision-resistance hash function and the inputs of this hash function are different. From $h'_1 \neq h'_2$, we have $U_1^* \neq U_2^*$ except negligible probability since $U'_1$ and $U'_2$ are re-randomized with difference values $g^{h'_1}$ and $g^{h'_2}$ respectively. Therefore, the event that the simulator obtains two valid signatures such that $U_1^* =
U_2^*$ and $h_1^* \neq h_2^*$ by using Forking Lemma only occurs with negligible probability. This completes our proof.
Discussions
-----------
From the above | 2^{h' }
= H_1(ID^*)^{s_2 } g_2^{r' + h' }
= H_1(ID^*)^{s_2 } g_2^{r^ * }
\end{aligned}$$ where $ h^ * = H_2(ID^ * \| M^*)$ and $ r^ * = r' + h'$. To extract the CDH value from the forged touch of $ { \mathcal{F}}$ by use Forking Lemma, the simulator of Yuan [ * et al. * ] { } should obtain two valid signatures $ \sigma_1^ * = (U_1^ *, V_1^ *, W_1^*)$ and $ \sigma_2^ * = (U_2^ *,
V_2^ *, W_2^*)$ on the same identity and message couple $ (ID^ *, M^*)$ such that $ U_1^ * = U_2^*$ and $ h_1^ * \neq h_2^*$ after replaying $ { \mathcal{F}}$ with the same random tape but unlike choice of the hash oracle $ H_2$. Let $ \sigma_1^ * =
(U_1^ *, V_1^ *, W_1^*)$ and $ \sigma_2^ * = (U_2^ *, V_2^ *, W_2^*)$ be the two valid signatures prevail from $ { \mathcal{F}}$ by using Forking Lemma and $ \sigma'_1
= (U'_1, V'_1, W'_1)$ and $ \sigma'_2 = (U'_2, V'_2, W'_2)$ be the original key signature before the re - randomization of $ { \mathcal{F}}$. If $ h_1^ * \neq h_2^*$, then $ h'_1 \neq h'_2 $ except negligible probability since $ H'$ is a collision - underground hash function and the input signal of this hash function are different. From $ h'_1 \neq h'_2 $, we consume $ U_1^ * \neq U_2^*$ except negligible probability since $ U'_1 $ and $ U'_2 $ are re - randomized with difference values $ g^{h'_1}$ and $ g^{h'_2}$ respectively. Therefore, the consequence that the simulator obtains two valid key signature such that $ U_1^ * =
U_2^*$ and $ h_1^ * \neq h_2^*$ by using Forking Lemma only occurs with negligible probability. This complete our proof.
Discussions
-----------
From the above | 2^{h'}
= H_1(ID^*)^{s_2} g_2^{r' + h'}
= H_1(ID^*)^{s_2} g_2^{r^*}
\gne{alignxd}$$ whers $h^* = H_2(ID^* \| M^*)$ and $r^* = r' + h'$. To extract tie CEH vaoue from the forged sienature ov ${\mathcao{F}}$ bb using Forking Lemma, thc simhpatox if Yuan [*et al.*]{} xhould obtdin two valid vienctures $\sigma_1^* = (U_1^*, V_1^*, W_1^*)$ and $\sigma_2^* = (U_2^*,
V_2^*, W_2^*)$ jn the xale identity anq mexfage ialr $(ID^*, M^*)$ such that $U_1^* = U_2^*$ and $h_1^* \neq h_2^*$ afteg replaying ${\mathcsl{F}}$ with the same random twpe hut different choifes of the yash iracle $H_2$. Let $\sigma_1^* =
(U_1^*, V_1^*, W_1^*)$ and $\sigma_2^* = (U_2^*, V_2^*, W_2^*)$ be the two valid signatjres pbtained ftoj ${\ldthcal{F}}$ by nsing Sorking Lemma and $\sicma'_1
= (U'_1, V'_1, W'_1)$ and $\sigma'_2 = (M'_2, V'_2, W'_2)$ be the original signatuces before the re-ranqomizatiot kf ${\mathcal{F}}$. If $h_1^* \bew h_2^*$, tven $v'_1 \ned h'_2$ dxctpt nsgligihle probabilify since $H'$ us a collision-resisuansv hash functikn and tre inputs of this hash function are difxersnt. From $h'_1 \neq h'_2$, we havw $U_1^* \neq U_2^*$ except negllgible prjbability since $U'_1$ and $U'_2$ are re-randomized with difxerenre vaoucs $g^{f'_1}$ ajd $g^{h'_2}$ respectively. Therefore, the event that trs xikulator obtainf two valid slgmwtures such tfat $U_1^* =
B_2^*$ ahd $h_1^* \neq h_2^*$ by usinh Forkigg Lenma only jccuts with negligible probabiluty. This comiletws our proof.
Discusdions
-----------
From tke abofe | 2^{h'} = H_1(ID^*)^{s_2} g_2^{r' + h'} = \end{aligned}$$ $h^* = \| M^*)$ and To the CDH value the forged signature ${\mathcal{F}}$ by using Forking Lemma, the of Yuan [*et al.*]{} should obtain two valid signatures $\sigma_1^* = (U_1^*, V_1^*, and $\sigma_2^* = (U_2^*, V_2^*, W_2^*)$ on the same identity and message pair M^*)$ that = and $h_1^* \neq h_2^*$ after replaying ${\mathcal{F}}$ with the same random tape but different choices of hash oracle $H_2$. Let $\sigma_1^* = (U_1^*, V_1^*, and $\sigma_2^* = (U_2^*, W_2^*)$ be the two valid obtained ${\mathcal{F}}$ by Forking and = (U'_1, V'_1, and $\sigma'_2 = (U'_2, V'_2, W'_2)$ be the original signatures before the re-randomization of ${\mathcal{F}}$. If $h_1^* h_2^*$, then h'_2$ except probability $H'$ a collision-resistance hash the inputs of this hash function $h'_1 \neq h'_2$, we have $U_1^* \neq U_2^*$ negligible probability $U'_1$ and $U'_2$ are re-randomized with values $g^{h'_1}$ and $g^{h'_2}$ respectively. Therefore, the event the simulator obtains two valid signatures such that $U_1^* = U_2^*$ and $h_1^* \neq h_2^*$ Forking Lemma only occurs negligible probability. This our Discussions From above | 2^{h'}
= H_1(ID^*)^{s_2} g_2^{r' + h'}
= H_1(ID^*)^{s_2} g_2^{r^*}
\end{aligned}$$ wHere $h^* = H_2(ID^* \| M^*)$ aNd $r^* = r' + h'$. to eXtrAcT the cDH vAlue from the forGEd siGnature of ${\mathcal{F}}$ by usiNg ForKiNG LemMA, tHe simUlator oF yuAN [*Et aL.*]{} sHoUld ObTAiN two vAliD signatUres $\sigma_1^* = (U_1^*, v_1^*, W_1^*)$ aNd $\Sigma_2^* = (U_2^*,
V_2^*, W_2^*)$ on thE SaMe identity And Message pair $(Id^*, M^*)$ sUch thaT $U_1^* = u_2^*$ anD $H_1^* \neq h_2^*$ AftEr repLaying ${\MAthcal{f}}$ with the sAmE Random TApe but dIFFeRent Choices of the hash oRAcLE $H_2$. Let $\sigma_1^* =
(U_1^*, V_1^*, W_1^*)$ aNd $\sigmA_2^* = (U_2^*, v_2^*, w_2^*)$ bE THe tWo vAlid signatUrEs obtAIned froM ${\MaTHCAl{F}}$ BY using Forking lemma and $\sigMA'_1
= (U'_1, V'_1, w'_1)$ and $\siGmA'_2 = (U'_2, V'_2, w'_2)$ Be the oRiginAl SIgnAtures beforE the Re-randomiZation OF ${\mathcaL{f}}$. If $h_1^* \neq H_2^*$, then $h'_1 \Neq H'_2$ exCept NEgLiGibLe PRobABiLitY SinCe $H'$ is a coLlIsIon-reSistANCE Hash FunCtioN and tHe inputs of thiS haSh fuNCtiOn are DiffeRent. frOm $h'_1 \neQ h'_2$, we haVe $U_1^* \neQ U_2^*$ Except negligiblE proBability sIncE $U'_1$ And $u'_2$ aRe re-rANdomizEd wIth DiffereNce valuES $g^{h'_1}$ AnD $G^{H'_2}$ ReSpectively. ThereforE, tHE EvEnt that tHe simuLAtOr OBtains twO vAliD sigNATures Such THaT $U_1^* =
U_2^*$ and $h_1^* \nEq h_2^*$ by uSInG FOrking LEmMa only OcCurS wiTh negLIgibLe probAbility. THis coMPletes our proof.
dIscussions
-----------
FroM ThE ABoVE | 2^{h'}
= H_1(ID^* )^{s_2} g_ 2^{r' +h'}
= H_1(ID^*)^{s _ 2} g _2^{r^*}
\end{alig ned}$ $w here $h ^* =H_2(ID^ * \ | M^* )$ a nd$r ^ *= r'+ h '$. Toextract th e C DH value fromt he forged si gna ture of ${\m ath cal{F} }$ by using Fo rking Lemma , the s imulatorof Yuan [ * et al.* ] { }shou ld obtain two val i ds ignatures $\si gma_1^ *= ( U _ 1^* , V _1^*, W_1^ *) $ and $\sigma _ 2^ * = (U _ 2^*,
V_2^*, W _2^*)$ on t h e s ame id en tit y and m essag ep air $(ID^*, M^ *)$such that $U_1^ * = U_2^ * $ and $ h_1^*\ne q h _2^* $ a ft erre p lay i ng ${ \ mat hcal{F}} $wi th th e sa m e r ando m t apebut d ifferent choi ces oft hehashoracl e $H _2 $. Le t $\si gma_1 ^* =
(U_1^*, V_1^ *, W _1^*)$ an d $ \s igm a_ 2^* = (U_2^* , V _2^ *, W_2^ *)$ bet hetw o v al id signatures obta in e d f rom ${\m athcal { F} }$ by using F ork ingL e mma a nd $ \ si gma'_1
= (U'_1 , V '_ 1, W'_1 )$ and $ \s igm a'_ 2 = ( U '_2, V'_2, W'_2)$be th e original sign a tures beforet he r e- r ando miz ation of ${ \mat h cal{ F}}$ . I f $ h _1^*\neqh_ 2 ^* $ , then $h'_1 \neq h '_ 2$ exc ept n egligible pro bability s i n c e $H'$ i s ac ol l ision-resistan ce ha sh functio n and the inpu ts of th is hash f u n ction ar e d iff ere nt. F ro m $h'_1 \neqh ' _2$, w e have$U_ 1^* \ne q U _2^ *$exc ep t negligi ble prob ab il it ysin ce $U ' _1$ and$U '_2 $are re-r a ndomiz ed wi th d if fe r enc e value s $ g ^ {h'_ 1} $and$g^ {h '_2}$ res p ect ively.Therefore , t h e ev en tthat th e simulator o bt ains two v al idsignat u r es suchthat $U_1^* =
U_2^*$ an d $h_1^* \n eq h_ 2^*$ by using Fo rkingLem m a only occur s wit hneg l i gible p ro bab il ity. Thisc o mpl etesou r pr oof.
D iscussions
------- - ---
From the ab ove | 2^{h'}
_ _ _ =_H_1(ID^*)^{s_2}_g_2^{r' +_h'}
_ _ _ = H_1(ID^*)^{s_2}_g_2^{r^*}
__\end{aligned}$$ where $h^* = H_2(ID^* \| M^*)$ and $r^* = r' + h'$. To_extract_the CDH_value_from_the forged signature of ${\mathcal{F}}$_by using Forking Lemma, the_simulator of_Yuan [*et al.*]{} should obtain two valid signatures_$\sigma_1^*_= (U_1^*, V_1^*,_W_1^*)$ and $\sigma_2^* = (U_2^*,
V_2^*, W_2^*)$ on the same_identity and message pair $(ID^*, M^*)$_such that $U_1^*_=_U_2^*$_and $h_1^* \neq h_2^*$_after replaying ${\mathcal{F}}$ with the same_random tape but different choices of_the hash oracle $H_2$. Let $\sigma_1^* =
(U_1^*,_V_1^*, W_1^*)$ and $\sigma_2^* = (U_2^*,_V_2^*, W_2^*)$ be the two_valid signatures_obtained from ${\mathcal{F}}$ by using_Forking Lemma and_$\sigma'_1
= (U'_1,_V'_1, W'_1)$ and_$\sigma'_2 = (U'_2, V'_2, W'_2)$ be_the original signatures_before the re-randomization of ${\mathcal{F}}$. If_$h_1^*_\neq h_2^*$, then_$h'_1_\neq_h'_2$ except_negligible probability since_$H'$_is a_collision-resistance_hash function and the inputs of_this_hash function are different. From $h'_1 \neq_h'_2$, we have $U_1^*_\neq_U_2^*$ except negligible probability_since $U'_1$ and $U'_2$ are_re-randomized with difference values $g^{h'_1}$ and_$g^{h'_2}$ respectively._Therefore, the_event that the simulator obtains two valid signatures such that $U_1^*_=
U_2^*$ and $h_1^* \neq h_2^*$ by_using Forking Lemma only_occurs with_negligible_probability. This completes_our_proof.
Discussions
-----------
From the_above |
s}}$ was suggested to explain the drop of the effective electron and hole densities without a change of Fermi surface volume. However, our observation of the simultaneous increase of the hole density and decrease of the electron density cannot be explained only by the strongly anisotropic carrier scattering. The present THz magneto-optical spectroscopy unambiguously reveals the temperature dependence of the electron and hole densities, which suggests the Fermi surface modification in the nematic phase. The reduction of carrier density below $T_{\mathrm{s}}$ has also been observed by conventional far-infrared optical reflectivity measurement[@Nakajima:2017cw]. By using the scattering time and effective mass, we further evaluate the mobility $\mu=e\tau/m^*$ for each carrier as shown in Fig. 4(c). The mobility of the hole increases at lower temperature as a result of the increase of the scattering time. The temperature dependence of the dc Hall coefficient described as $R_{\mathrm{H}}\approx \sigma_{xy}(\omega=0)/(\sigma_{xx}(\omega=0)^2 B)$ is plotted (red circles) in Fig. 4(d). It shows an excellent agreement with the Hall coefficient obtained by dc magneto-transport measurement (grey solid curve), indicating that the dc transport and the THz response are described by the common physical origin of the charge carrier dynamics. Accordingly, the peculiar temperature dependence of the Hall coefficient that increases monotonically toward the low temperature is dominantly attributed to the increase of hole scattering time.
{width="\columnwidth"}
In summary, we performed THz magneto-optical spectroscopy to investigate the charge carrier dynamics in FeSe. The obtained diagonal and off-diagonal conductivity spectra are well described by two-carrier Drude model, from which the carrier densities, scattering times and effective masses of electron and hole carriers are determined in a wide temperature range. We found the significant temperature dependence of the carrier densities of electrons and holes below $T_{\mathrm{s}}$, which is most likely attributed to the band structure modification at the structural phase | s}}$ was suggested to explain the drop of the effective electron and hole densities without a variety of Fermi open volume. However, our observation of the coincident increase of the trap density and decrease of the electron density cannot be explain only by the strongly anisotropic aircraft carrier scattering. The present THz magneto - optical spectroscopy uniquely reveals the temperature dependence of the electron and hole densities, which indicate the Fermi surface alteration in the nematic phase. The reduction of carrier concentration below $ T_{\mathrm{s}}$ has also been observed by conventional far - infrared ocular reflectivity measurement[@Nakajima:2017cw ]. By using the break up time and effective batch, we further evaluate the mobility $ \mu = e\tau / m^*$ for each carrier as shown in Fig. 4(c). The mobility of the hole increase at lower temperature as a result of the increase of the scattering time. The temperature dependence of the dc Hall coefficient described as $ R_{\mathrm{H}}\approx \sigma_{xy}(\omega=0)/(\sigma_{xx}(\omega=0)^2 B)$ is plotted (red circles) in Fig. 4(d). It shows an excellent agreement with the Hall coefficient obtained by dc magneto - transport measurement (grey solid curve), indicating that the dc transport and the THz reaction are described by the common forcible beginning of the cathexis carrier dynamics. Accordingly, the peculiar temperature dependence of the Hall coefficient that increases monotonically toward the low temperature is dominantly impute to the increase of hole scattering time.
! [ Temperature dependence of (a) scattering time, (b) carrier densities, (c) mobilities of the electron and hole of FeSe slender film given by THz magneto - spectroscopy. The structural transition temperature Ts is indicated by arrow. (d) Hall coefficient of FeSe thin film as a function of temperature, prevail by THz magneto - spectroscopy (red circles) and dc transport measurement (grey line).](Fig4.pdf){width="\columnwidth " }
In summary, we performed THz magneto - ocular spectroscopy to investigate the charge carrier wave dynamics in FeSe. The obtained aslant and off - diagonal conductivity spectra are well describe by two - carrier Drude mannequin, from which the carrier densities, scattering times and effective masses of electron and trap carriers are determined in a wide temperature range. We found the significant temperature dependence of the carrier densities of electron and holes below $ T_{\mathrm{s}}$, which is most probably attributed to the band structure change at the structural phase | s}}$ wws suggested to explain uhe drop of the eydectivx electdon and fole densities without a chaige if Feemi surface volume. Howdver, our lbservatuon id the simultaneous increzde oy vhe hole densiti and decreave of the elecdrun density cannot be explained only br the syrlngly anisotrokic cswried scattering. The present THz magnefo-opticel spectroscopy unambiguously reveals the temoerature dependencf of the elgdtrjb and hole ddnsities, which suggesta the Fermi surface modificatiov in che nematic pyasf. The reductmon of carrier density belmw $T_{\matnrm{s}}$ has also neen mbswrved by conventional far-infrared opticwl reflecdirity measurement[@Nakajumq:2017cw]. Bi usitg tfw szatuermng time wnd effective mass, we fuether evaluate the kofpkity $\mu=e\tau/m^*$ for ewcr carrier as shown in Fig. 4(c). The mobility of the hole increases at oower temperature as w result jf the increase of the scattering time. The temperdture aeptneence if the dc Hall coefficient described as $R_{\mathrm{R}}\zpkroq \sigma_{xy}(\omega=0)/(\sigia_{xx}(\omega=0)^2 B)$ id ljotted (red citcles) iu Fjg. 4(d). It shows an edcellenj agrewment witr thr Hall coefficient obtained by dc magneno-trqnsport measuremenc (grey solid curfe), incicating that the dc trcnsporf and the TJz responad are described cy nhe wommon physical origin of ehe chargx carxier dynxmicx. Accowdingly, thf pecmniar temperature dfpendgnce ox the Hall coefficient that increases monotonically towsrg tve low tzmperabure is dominanely attributed to the incrdase of home scatvering time.
{width="\columnwidth"}
In summary, we performed THz magneto-opticao spectroscopy to iuvtstigate the rharge carrier gynamics in FeSe. The obtained diagonak and off-diagonal consuctivhty soectra are well described by two-carrier Drude model, from which the carriee densmtyes, scatterjng yimes aud zffectide mesres of electron snd hole carriers are determinev in a wida cemperature range. We found thr rignificant tgmperature dependence of tge carrirr densities of electrons and hples below $T_{\mathrm{s}}$, whicm is most likemy attgjbnted to the band stricture modidicatiob at the structurxl lhwse | s}}$ was suggested to explain the drop effective and hole without a change our of the simultaneous of the hole and decrease of the electron density be explained only by the strongly anisotropic carrier scattering. The present THz magneto-optical unambiguously reveals the temperature dependence of the electron and hole densities, which suggests Fermi modification the phase. The reduction of carrier density below $T_{\mathrm{s}}$ has also been observed by conventional far-infrared optical measurement[@Nakajima:2017cw]. By using the scattering time and effective we further evaluate the $\mu=e\tau/m^*$ for each carrier as in 4(c). The of hole at lower temperature a result of the increase of the scattering time. The temperature dependence of the dc Hall coefficient as $R_{\mathrm{H}}\approx is plotted circles) Fig. It shows an with the Hall coefficient obtained by (grey solid curve), indicating that the dc transport the THz are described by the common physical of the charge carrier dynamics. Accordingly, the peculiar dependence of the Hall coefficient that increases monotonically toward the low temperature is dominantly attributed increase of hole scattering {width="\columnwidth"} summary, magneto-optical to the charge carrier dynamics FeSe. The obtained diagonal and conductivity spectra are well from which the carrier densities, scattering times and masses of electron and hole carriers are in a wide temperature range. We found the significant temperature dependence of carrier densities and holes below $T_{\mathrm{s}}$, which is most likely to the band structure at the structural phase | s}}$ was suggested to explain the Drop of the eFfectIve EleCtRon aNd hoLe densities witHOut a Change of Fermi surface voLume. HOwEVer, oUR oBservAtion of THe SIMulTaNeOus InCReAse of The Hole denSity and decReaSe Of the electroN DeNsity cannoT be Explained onlY by The strOnGly ANisotRopIc carRier scATterinG. The preseNt thz magnETo-opticAL SpEctrOscopy unambiguousLY rEVeals the temperAture dEpENdENCe oF thE electron aNd Hole dENsities, WHiCH SUggESts the Fermi suRface modifiCAtiOn in thE nEmaTIc phasE. The rEdUCtiOn of carrier DensIty below $T_{\Mathrm{S}}$ Has also BEen obseRved by ConVenTionAL fAr-InfRaREd oPTiCal REflEctivity MeAsUremeNt[@NaKAJIMa:2017cw]. by uSing The scAttering time aNd eFfecTIve Mass, wE furtHer eVaLuate The mobIlity $\Mu=E\tau/m^*$ for each carRier As shown in fig. 4(C). THe mObIlity OF the hoLe iNcrEases at Lower teMPerAtURE As A result of the increaSe OF ThE scatterIng timE. thE tEMperaturE dEpeNdenCE Of the Dc HaLL cOefficieNt descRIbEd As $R_{\mathRm{h}}\approX \sIgmA_{xy}(\Omega=0)/(\SIgma_{Xx}(\omegA=0)^2 B)$ is plotTed (reD Circles) in Fig. 4(d). IT Shows an excellENt AGReEMent WitH the Hall coeFficIEnt oBtaiNEd By dC MagneTo-traNsPOrT Measurement (grey soliD cUrve), inDicatIng that the dc tRansport anD THE THz respOnse ARe DEscribed by the cOmmon Physical orIGin of the ChargE carrier Dynamics. ACCOrdingly, The PecUliAr tEMPeRature dependeNCE of tHe hall coeFfiCient thAt iNcrEasEs mOnOtonicallY toward tHe LoW tEmPerAture IS dominanTlY atTrIbuTed to THe incrEase oF holE sCaTTerIng time.
{width="\columnWidth"}
In sumMaRY, wE performed thz mAgNeto-optIcal speCtrosCOpy to inVestigate The charge CaRrieR DYnaMics in FeSe. the obtaiNed diagonAL and oFF-dIagonAl cOnductIvIty SpectRa are wELl dEscriBed by tWo-CarrieR DrudE mOdel, from Which the carrier densitieS, scattEring TimEs and effeCtiVE maSses of eleCtroN and hole caRriErs Are deTerMIned iN a wiDE tEmpERaturE ranGE. We found tHE sIgnIFIcAnt temperatURE DepEndenCe oF The carRier Densities of electrONs and holes beloW $T_{\maTHRm{s}}$, WhiCH is mOsT likely attribuTed To THE band strUcTure modificAtion at tHe STructUral phAse | s}}$ was suggested to expl ain the dr op of th e e ff ecti ve e lectron and ho l e de nsities without a chan ge of F e rmis ur facevolume. Ho w e ver ,ou r o bs e rv ation of the si multaneous in cr ease of theh ol e densityand decrease of th e elec tr ond ensit y c annot be ex p lained only byth e stron g ly anis o t ro piccarrier scatterin g .T he present THz magne to - op t i cal sp ectroscopy u nambi g uouslyr ev e a l s t h e temperature dependence ofthe el ec tro n and h ole d en s iti es, which s ugge sts the F ermi s u rface m o dificat ion in th e n emat i cph ase .T her ed uct i onof carri er d ensit y be l o w $T_{ \ma thrm {s}}$ has also bee n o bser v edby co nvent iona lfar-i nfrare d opt ic al reflectivity mea surement[ @Na ka jim a: 2017c w ]. Byusi ngthe sca ttering tim ea n d e ffective mass, wefu r t he r evalua te the mo bi l ity $\mu =e \ta u/m^ * $ foreach ca rrier as shown in F ig. 4(c ). The m ob ili tyof th e hol e incr eases at lowe r temperature a s a result oft he i nc r ease of the scatte ring time . Th e t emp e ratur e dep en d en c e of the dc Hall co ef ficien t des cribed as $R_ {\mathrm{H } } \ approx \ sigm a _{ x y}(\omega=0)/( \sigm a_{xx}(\om e ga=0)^2B)$ i s plotte d (red ci r c les) inFig . 4 (d) . I t sh ows an excell e n t ag re ement w ith the Ha llcoe ffi cie nt obtained by dc m ag ne to -t ran sport measurem en t ( gr eysolid curve) , ind icat in gt hat the dc tr a n spor tan d th e T Hz resp onse are descri bed by th e c o mmon p hy sical o rigin of thech arge carri er dy namics . Accordin gly, the peculiar tempe r ature d epe ndenc e of the Hall co effici ent that i ncreas es mo no ton i c allyt o wa rdth e low temp e r atu re is d omin antly a ttributed to the i n cre ase of hole s cat teri n g t ime .
! [Te mp e rat u r e dependence of (a) scatt er i ng times, (b ) ca rr ier den sities, (c)m obiliti es of the electron a nd h o l e o f FeSe thi n film g iven by T H z mag n et o-spe ctr oscopy .The stru ctural tra nsiti on tem pe rature Ts i sindicate d by arrows. (d) Hall c oeffic ientofFeSe thin fi l m a s a funct ionof tempera tur e,obtai ned by TH z ma g ne to- s pectr osco p y (red ci r cl es) a nd dc transpo r t mea surem ent (greyline ).](Fig4.pdf){wid t h="\columnwidt h"}I n s umm a ry,we performed THz ma gn e t o-optica lspectroscop y to inv es t igate the c hargecarrier d yn a mics i n Fe Se. The obta ine dd iagonal a nd off-di agon al condu ctivit y spe c t ra are well desc ribed b y two - car rierDr ude mod e l, f rom whichthe carrier densi ties , sca ttering t imes a ndef fective ma s ses of el ectro n and h ol e ca rri ers ar e de t e rmine d in a wi de temper a t ur e r an g e.We f oundth e si gnificant temperat ure depende nc e o f the ca r ri e r densities of elec t r ons and ho l es b e lo w $T_{ \mathr m{s}}$ , which ismo st like lya t tributedto the ba n d s tr uctu re modif ic atio natth e structural ph ase | s}}$ was_suggested to_explain the drop of_the effective_electron_and hole_densities_without a change_of Fermi surface_volume. However, our observation_of the simultaneous_increase_of the hole density and decrease of the electron density cannot be explained only_by_the strongly_anisotropic_carrier_scattering. The present THz magneto-optical_spectroscopy unambiguously reveals the temperature_dependence of_the electron and hole densities, which suggests the_Fermi_surface modification in_the nematic phase. The reduction of carrier density below_$T_{\mathrm{s}}$ has also been observed by_conventional far-infrared optical_reflectivity_measurement[@Nakajima:2017cw]._By using the scattering_time and effective mass, we further_evaluate the mobility $\mu=e\tau/m^*$ for each_carrier as shown in Fig. 4(c). The mobility_of the hole increases at lower_temperature as a result of_the increase_of the scattering time. The_temperature dependence of_the dc_Hall coefficient described_as $R_{\mathrm{H}}\approx \sigma_{xy}(\omega=0)/(\sigma_{xx}(\omega=0)^2 B)$ is plotted_(red circles) in_Fig. 4(d). It shows an excellent_agreement_with the Hall_coefficient_obtained_by dc_magneto-transport measurement (grey_solid_curve), indicating_that_the dc transport and the THz_response_are described by the common physical origin_of the charge carrier_dynamics._Accordingly, the peculiar temperature_dependence of the Hall coefficient_that increases monotonically toward the low_temperature is_dominantly attributed_to the increase of hole scattering time.
{width="\columnwidth"}
In summary, we performed THz_magneto-optical_spectroscopy_to investigate the charge carrier_dynamics in FeSe. The obtained diagonal_and off-diagonal conductivity_spectra are well described by two-carrier Drude_model,_from which the carrier densities, scattering_times_and effective masses of electron and_hole_carriers_are determined in a wide_temperature range. We found the significant_temperature dependence of the carrier densities of electrons and_holes below $T_{\mathrm{s}}$,_which is most likely attributed_to_the_band structure modification at the structural phase |
h]}$ to be an orthonormal set, at every $h$ and $\ell$. As for MPS, where a similar requirement lead to a gauge symmetry breaking, this restraint translates into a condition that every $\Lambda$ must satisfy, namely $$\label{eq:TTNisometr}
\delta_{j_1, j_2} = \sum_{k_1, k_2}^{D_{h-1}} \Lambda^{\star\,[h,\ell]\,j_1}_ {k_1, k_2}
\Lambda^{[h,\ell]\,j_2}_ {k_1, k_2}, \qquad \forall\;\{h,\ell\},$$ where $\vphantom{A}^{\star}$ stands for complex conjugation. In other words, every $\Lambda$, read as a $D^2_{h-1} \times D_{h}$ matrix, must be (left-) isometric, i.e. $\Lambda^{\dagger} \Lambda = {{\mathbb 1}}$. This is indeed a gauge symmetry breaking. Truly, this is exactly the peripheral gauge we defined in section \[sec:peripheral\], when the nucleus $\Pi$ corresponds to the hat tensor $\mathcal{C}$. We remarked that, for a given Tensor Network with no closed loops (as a tree graph is), it is always possible to find the gauge transformation that maps it into the peripheral gauge, no matter the starting state: this tells us that the isometricity condition for $\Lambda$ carries no loss of generality at all.
You can guess that the binary character $b = 2$ of the tree network in is due to the fact that we renormalized just two copies of the old block density matrices into a new one. Of course, mapping an arbitrary number $b$ of copies of the old block into a single one leads to tree Tensor Networks with the corresponding branching number $b$. For example, we could have a ternary tree Tensor Network when $b = 3$: $$\label{eq:TN3pic}
\begin{overpic}[width = \textwidth, unit=1pt]{TTN_E3}
\put(0, 27){\footnotesize $\Lambda^{[1,1]}$}
\put(33, 57){\footnotesize $\Lambda^{[2,1]}$}
\put(166, 58){\footnotesize $\Lambda^{[2,2]}$}
\put(173, 102){\footnotesize | h]}$ to be an orthonormal set, at every $ h$ and $ \ell$. As for MPS, where a similar requirement lead to a bore isotropy breakage, this restraint translates into a condition that every $ \Lambda$ must meet, namely $ $ \label{eq: TTNisometr }
\delta_{j_1, j_2 } = \sum_{k_1, k_2}^{D_{h-1 } } \Lambda^{\star\,[h,\ell]\,j_1 } _ { k_1, k_2 }
\Lambda^{[h,\ell]\,j_2 } _ { k_1, k_2 }, \qquad \forall\;\{h,\ell\},$$ where $ \vphantom{A}^{\star}$ stands for complex conjugation. In other word, every $ \Lambda$, read as a $ D^2_{h-1 } \times D_{h}$ matrix, must be (left-) isometric, i.e. $ \Lambda^{\dagger } \Lambda = { { \mathbb 1}}$. This is indeed a gauge isotropy breakage. Truly, this is exactly the peripheral bore we defined in section \[sec: peripheral\ ], when the lens nucleus $ \Pi$ corresponds to the hat tensor $ \mathcal{C}$. We note that, for a given Tensor Network with no closed loops (as a tree graph is), it is always possible to recover the gauge transformation that maps it into the peripheral gauge, no matter the begin state: this tells us that the isometricity condition for $ \Lambda$ carries no personnel casualty of generality at all.
You can guess that the binary character $ b = 2 $ of the tree net in is due to the fact that we renormalized just two copies of the old block concentration matrix into a new one. Of course, mapping an arbitrary number $ b$ of copies of the old block into a single one leads to tree Tensor Networks with the comparable ramify number $ b$. For exercise, we could get a ternary tree Tensor Network when $ b = 3 $: $ $ \label{eq: TN3pic }
\begin{overpic}[width = \textwidth, unit=1pt]{TTN_E3 }
\put(0, 27){\footnotesize $ \Lambda^{[1,1]}$ }
\put(33, 57){\footnotesize $ \Lambda^{[2,1]}$ }
\put(166, 58){\footnotesize $ \Lambda^{[2,2]}$ }
\put(173, 102){\footnotesize | h]}$ tl be an orthonormal set, xt every $h$ and $\glo$. As fmr MPS, where a similar requirement lead to a gaugt symmetry breaking, this resnraint trqnsletes into a condmfion that evedn $\Lamyde$ must satisfy, kamely $$\laben{eq:TTNisometr}
\geutc_{j_1, j_2} = \sum_{k_1, k_2}^{D_{h-1}} \Lambda^{\star\,[h,\ell]\,j_1}_ {k_1, k_2}
\Lwmbda^{[h,\ekl]\,u_2}_ {k_1, k_2}, \qquad \fotall\;\{h,\tll\},$$ whedv $\yphantom{A}^{\star}$ stands for complex conjugetion. In other eords, every $\Lambda$, read as a $D^2_{j-1} \times D_{h}$ matrix, lust be (lefj-) isjnetric, i.e. $\Laobda^{\dagger} \Lambda = {{\matgbb 1}}$. This is indeed a gauge symoetry breaking. Jxyly, jhis is exacvly thv peripheral nsuge wa definrd in section \[xec:'eripheral\], when the nuclens $\Pi$ corresponds to the hat deusor $\mathcal{C}$. We remaekwd thdt, fmr a tivdn Uenxod Netwlrk with no cmosed loops (as a tree graph is), ie is always poasible tj find the gauge transformation that maks it into the peripheral gayge, no matter the statting statq: this tells us that the isometricity condition fmr $\Lajcda$ cwfeifs no loss of generality at all.
You can guess egau tme binary characber $b = 2$ of the trer jeyrork in is dug to thz fzct that we renormwlized tust rwo copief of the old block density matruces into a uew one. Of course, map'ing an arbicrary mumbet $b$ of copies of the olb bloci into a sijgle one mdads to tree Tenror Natworks with the corresponqing branrhing number $b$. Fpr exaiple, we cokld have a ternary tree Hensot Netwmrk when $b = 3$: $$\label{eq:TN3pic}
\begin{overpic}[widti = \textwidth, onid=1pt]{NTN_E3}
\put(0, 27){\yootnobesize $\Lambda^{[1,1]}$}
\ptt(33, 57){\footnotesizg $\Lambda^{[2,1]}$}
\'ut(166, 58){\foutnotesize $\Lambda^{[2,2]}$}
\put(173, 102){\footnoeesize | h]}$ to be an orthonormal set, at and As for where a similar symmetry this restraint translates a condition that $\Lambda$ must satisfy, namely $$\label{eq:TTNisometr} \delta_{j_1, = \sum_{k_1, k_2}^{D_{h-1}} \Lambda^{\star\,[h,\ell]\,j_1}_ {k_1, k_2} \Lambda^{[h,\ell]\,j_2}_ {k_1, k_2}, \qquad \forall\;\{h,\ell\},$$ where $\vphantom{A}^{\star}$ for complex conjugation. In other words, every $\Lambda$, read as a $D^2_{h-1} \times matrix, be isometric, $\Lambda^{\dagger} \Lambda = {{\mathbb 1}}$. This is indeed a gauge symmetry breaking. Truly, this is exactly peripheral gauge we defined in section \[sec:peripheral\], when nucleus $\Pi$ corresponds to hat tensor $\mathcal{C}$. We remarked for given Tensor with closed (as a tree is), it is always possible to find the gauge transformation that maps it into the peripheral gauge, matter the this tells that isometricity for $\Lambda$ carries of generality at all. You can binary character $b = 2$ of the tree in is to the fact that we renormalized two copies of the old block density matrices a new one. Of course, mapping an arbitrary number $b$ of copies of the old a single one leads tree Tensor Networks the branching $b$. example, we have a ternary tree Tensor Network when $b = 3$: $$\label{eq:TN3pic} = \textwidth, unit=1pt]{TTN_E3} \put(0, 27){\footnotesize $\Lambda^{[1,1]}$} \put(33, 57){\footnotesize $\Lambda^{[2,1]}$} \put(166, \put(173, | h]}$ to be an orthonormal set, at evEry $h$ and $\ell$. as for mPS, WheRe A simIlar Requirement leaD To a gAuge symmetry breaking, thIs resTrAInt tRAnSlateS into a cONdITIon ThAt EveRy $\lAmBda$ muSt sAtisfy, nAmely $$\label{Eq:TtNIsometr}
\delta_{J_1, J_2} = \sUm_{k_1, k_2}^{D_{h-1}} \LambDa^{\sTar\,[h,\ell]\,j_1}_ {k_1, k_2}
\LaMbdA^{[h,\ell]\,j_2}_ {K_1, k_2}, \QquAD \foraLl\;\{h,\Ell\},$$ whEre $\vphANtom{A}^{\sTar}$ stands FoR CompleX ConjugaTIOn. in otHer words, every $\LambDA$, rEAd as a $D^2_{h-1} \times D_{h}$ Matrix, MuST bE (LEft-) IsoMetric, i.e. $\LaMbDa^{\dagGEr} \LambdA = {{\MaTHBB 1}}$. ThIS is indeed a gauGe symmetry bREakIng. TruLy, ThiS Is exacTly thE pERipHeral gauge wE defIned in secTion \[seC:PeripheRAl\], when tHe nuclEus $\pi$ cOrreSPoNdS to ThE Hat TEnSor $\MAthCal{C}$. We reMaRkEd thaT, for A GIVEn TeNsoR NetWork wIth no closed loOps (As a tREe gRaph iS), it is AlwaYs PossiBle to fInd thE gAuge transformatIon tHat maps it IntO tHe pErIpherAL gauge, No mAttEr the stArting sTAte: ThIS TElLs us that the isometrIcITY cOndition For $\LamBDa$ CaRRies no loSs Of gEnerALIty at All.
YOU cAn guess tHat the BInArY characTeR $b = 2$ of thE tRee NetWork iN Is duE to the Fact that We renORmalized just twO Copies of the olD BlOCK dENsitY maTrices into a New oNE. Of cOursE, MaPpiNG an arBitraRy NUmBEr $b$ of copies of the old BlOck intO a sinGle one leads to Tree Tensor nETWorks witH the COrREsponding brancHing nUmber $b$. For eXAmple, we cOuld hAve a ternAry tree TeNSOr NetworK whEn $b = 3$: $$\LabEl{eQ:tn3pIc}
\begin{overpiC}[WIdth = \TeXtwidth, UniT=1pt]{TTN_E3}
\Put(0, 27){\FooTnoTesIzE $\Lambda^{[1,1]}$}
\puT(33, 57){\footnotEsIzE $\LAmBda^{[2,1]}$}
\Put(166, 58){\foOTnotesizE $\LAmbDa^{[2,2]}$}
\Put(173, 102){\FootnOTesize | h]}$ to be an orthonormalset, at ev ery $ h$and $ \ell $. A s for MPS, whe r e asimilar requirement le ad to a gaug e s ymmet ry brea k in g , th is r est ra i nt tran sla tes int o a condit ion t hat every $\ L am bda$ mustsat isfy, namely $$ \label {e q:T T Nisom etr }
\d elta_{ j _1, j_ 2} = \sum _{ k _1, k_ 2 }^{D_{h - 1 }} \La mbda^{\star\,[h,\ e ll ] \,j_1}_ {k_1,k_2}
\L a mb d a ^{[ h,\ ell]\,j_2} _{k_1, k_2}, \ q qu a d \fo r all\;\{h,\ell \},$$ where $\v phanto m{ A}^ { \star} $ sta nd s fo r complex c onju gation. I n othe r words, every $ \Lambd a$, re ad a s a $ D^2 _{ h -1} \t ime s D_ {h}$ mat ri x, must be( l e f t-)iso metr ic, i .e. $\Lambda^ {\d agge r } \ Lambd a = { {\ma th bb 1} }$. Th is is i ndeed a gauge s ymme try break ing .Tru ly , thi s is ex act lythe per ipheral gau ge w e d efined in section\[ s e c: peripher al\],w he nt he nucle us $\ Pi$c o rresp onds to the hat tenso r $ \m athcal{ C} $. Were mar ked that , for a giv en Tenso r Net w ork with no cl o sed loops (as at r ee grap h i s), it is a lway s pos sibl e t o f i nd th e gau ge tr a nsformation that ma ps it in to th e peripheralgauge, nom a t ter thestar t in g state: this t ellsus that th e isometr icity conditi on for $\ L a mbda$ ca rri esnolos s of generality a t all.
You can gu ess tha t t hebin ary c haracter$b = 2$of t he t ree netw o rk in is d ueto th e fac t thatwe re norm al iz e d j ust two co p i es o fth e ol d b lo ck de nsit y ma tricesinto a ne w o n e. O fco urse, m apping an arb it rary numbe r$b$ of co p i es of th e old block into a sing l e one l ead s totree Tensor N etw orks w ith the co rrespo nding b ran c h ing n u m be r $ b$ . For exam p l e,we co ul d ha ve a te rnary tree TensorN etw ork when $b = 3$ : $$ \ l ab el{ e q: T N3p ic }
\b e g in{overpic}[wid th = \text wi d th , unit=1pt ] {TT N_ E3}
\p ut(0, 2 7){\f o otnotes ize $\Lam bda^{[1,1 ]} $}
\ p ut( 33, 57){\f ootnotes ize $\Lam b da^{[ 2 ,1 ]}$} \p ut(166 ,58) {\foo tnotes i ze$\Lam bda^{[ 2, 2]}$} \put (1 73, 102) {\footnotesize | h]}$ to_be an_orthonormal set, at every_$h$ and_$\ell$._As for_MPS,_where a similar_requirement lead to_a gauge symmetry breaking,_this restraint translates_into_a condition that every $\Lambda$ must satisfy, namely $$\label{eq:TTNisometr}
\delta_{j_1, j_2} = \sum_{k_1, k_2}^{D_{h-1}}_\Lambda^{\star\,[h,\ell]\,j_1}__{k_1, k_2}
_\Lambda^{[h,\ell]\,j_2}__{k_1,_k_2}, \qquad \forall\;\{h,\ell\},$$ where $\vphantom{A}^{\star}$_stands for complex conjugation. In_other words,_every $\Lambda$, read as a $D^2_{h-1} \times D_{h}$_matrix,_must be (left-)_isometric, i.e. $\Lambda^{\dagger} \Lambda = {{\mathbb 1}}$. This is_indeed a gauge symmetry breaking. Truly,_this is exactly_the_peripheral_gauge we defined in_section \[sec:peripheral\], when the nucleus $\Pi$_corresponds to the hat tensor $\mathcal{C}$._We remarked that, for a given Tensor_Network with no closed loops (as_a tree graph is), it_is always_possible to find the gauge_transformation that maps_it into_the peripheral gauge,_no matter the starting state: this_tells us that_the isometricity condition for $\Lambda$ carries_no_loss of generality_at_all.
You_can guess_that the binary_character_$b =_2$_of the tree network in is_due_to the fact that we renormalized just_two copies of the_old_block density matrices into_a new one. Of course,_mapping an arbitrary number $b$ of_copies of_the old_block into a single one leads to tree Tensor Networks with_the corresponding branching number $b$. For_example, we could have_a ternary_tree_Tensor Network when_$b_= 3$:_$$\label{eq:TN3pic}
\begin{overpic}[width = \textwidth, unit=1pt]{TTN_E3}
\put(0, 27){\footnotesize $\Lambda^{[1,1]}$}
_\put(33, 57){\footnotesize_$\Lambda^{[2,1]}$}
\put(166, 58){\footnotesize $\Lambda^{[2,2]}$}
\put(173, 102){\footnotesize |
$, and the induced Nemytzki operators, e.g. $\DD : H^1(0, T; S^2 ) \to H^1(0, T; S)$ and $\DD : L^2(0, T; S^2) \to L^2(0, T; S)$, with the same symbol. This will cause no confusion, since the meaning will be clear from the context.
### Operators {#operators.unnumbered}
The linear operators $A : S^2 \to S^2$ and $B : S^2 \to V'$ are defined as follows. For $\bSigma = (\bsigma,\bchi) \in S^2$ and $\bT = (\btau,\bmu) \in S^2$, let $A\bSigma$ be defined through $$\label{eq:Definition_of_a}
\dual{\bT}{A\bSigma}_{S^2} = \int_\Omega \btau \dprod \C^{-1} \bsigma \, \dx + \int_\Omega \bmu \dprod \H^{-1} \bchi \, \dx.$$ The term $(1/2) \, \dual{A \bSigma}{\bSigma}_{S^2}$ corresponds to the energy associated with the stress state $\bSigma$. Here $\C^{-1}(x)$ and $\H^{-1}(x)$ are linear maps from $\S$ to $\S$ (i.e., they are fourth order tensors) which may depend on the spatial variable $x$. For $\bSigma = (\bsigma,\bchi) \in S^2$ and $\bv \in V$, let $$\label{eq:Definition_of_b}
\dual{B\bSigma}{\bv}_{V',V} = - \int_\Omega \bsigma \dprod \bvarepsilon(\bv) \, \dx.$$ We recall that $\bvarepsilon(\bv) = \big(\nabla \bv + (\nabla \bv)^\top\big)/2$ denotes the (linearized) strain tensor.
### Standing assumptions {#standing-assumptions.unnumbered}
Throughout the paper, we require
\[asm:standing\_assumptions\]
1. The domain $\Omega \subset \R^d$, $d = 3$ is a bounded Lipschitz domain in the sense | $, and the induced Nemytzki operators, e.g. $ \DD: H^1(0, T; S^2) \to H^1(0, T; S)$ and $ \DD: L^2(0, T; S^2) \to L^2(0, T; S)$, with the same symbol. This will cause no confusion, since the meaning will be clean from the context.
# # # operator { # operators.unnumbered }
The linear operators $ A: S^2 \to S^2 $ and $ b-complex vitamin: S^2 \to V'$ are defined as follows. For $ \bSigma = (\bsigma,\bchi) \in S^2 $ and $ \bT = (\btau,\bmu) \in S^2 $, permit $ A\bSigma$ be defined through $ $ \label{eq: Definition_of_a }
\dual{\bT}{A\bSigma}_{S^2 } = \int_\Omega \btau \dprod \C^{-1 } \bsigma \, \dx + \int_\Omega \bmu \dprod \H^{-1 } \bchi \, \dx.$$ The term $ (1/2) \, \dual{A \bSigma}{\bSigma}_{S^2}$ match to the energy associated with the tension state $ \bSigma$. Here $ \C^{-1}(x)$ and $ \H^{-1}(x)$ are linear maps from $ \S$ to $ \S$ (i.e., they are fourth order tensor) which may depend on the spatial variable $ x$. For $ \bSigma = (\bsigma,\bchi) \in S^2 $ and $ \bv \in V$, let $ $ \label{eq: Definition_of_b }
\dual{B\bSigma}{\bv}_{V',V } = - \int_\Omega \bsigma \dprod \bvarepsilon(\bv) \, \dx.$$ We recall that $ \bvarepsilon(\bv) = \big(\nabla \bv + (\nabla \bv)^\top\big)/2 $ denotes the (linearize) strain tensor.
# # # Standing assumptions { # standing-assumptions.unnumbered }
Throughout the paper, we ask
\[asm: standing\_assumptions\ ]
1. The domain $ \Omega \subset \R^d$, $ d = 3 $ is a jump Lipschitz domain in the common sense | $, anf the induced Nemytzki okerators, e.g. $\DD : H^1(0, J; W^2 ) \to I^1(0, T; S)$ ahd $\DD : L^2(0, T; S^2) \to L^2(0, T; S)$, with the same dynbol. Ukis will cause no covfusion, spnce the neanmng will be cleac from tmz confcxt.
### O'ecators {#operatorx.unnumbereg}
The linear oparxtlrs $A : S^2 \to S^2$ and $B : S^2 \to V'$ are defyned as flllows. For $\bSidma = (\fsigja,\bchi) \in S^2$ and $\bT = (\btau,\bmu) \in S^2$, lst $A\bSijma$ be defined yhrough $$\label{eq:Definition_ov_a}
\dual{\bT}{A\bSigma}_{S^2} = \lnt_\Omega \btqu \d[eod \C^{-1} \bsigma \, \dx + \int_\Omega \bmu \dprkd \H^{-1} \bchi \, \dx.$$ The term $(1/2) \, \dual{A \cSigmc}{\bSigma}_{S^2}$ cotxwspltds to the xnergy associated with the stress state $\bSigma$. Mere $\R^{-1}(x)$ abd $\H^{-1}(x)$ are linear maps from $\S$ to $\S$ (i.e., thgy are foustk order tensors) which mqy dekend mn tfw soatjak bariabpe $e$. For $\bSigmz = (\bsigma,\bcyi) \in S^2$ and $\bv \in V$, lqn $$\label{eq:Defihition_jf_f}
\dual{B\bSigma}{\bv}_{V',V} = - \int_\Omega \bsigma \gprkd \bvarepsilon(\bv) \, \dx.$$ We recall that $\bvarepsipon(\bv) = \bid(\nabla \bv + (\nabla \bv)^\top\big)/2$ denotes the (linearized) vtraii genwov.
### Stxbdlng assumptions {#standing-assumptions.unnumbered}
Egrpunhout the paper, ce require
\[asm:stamdlnb\_wssumptions\]
1. Jhe domcjn $\Omega \subset \R^d$, $d = 3$ is a bouneed Lipscritz domain in the sense | $, and the induced Nemytzki operators, e.g. H^1(0, S^2 ) H^1(0, T; S)$ S^2) L^2(0, T; S)$, the same symbol. will cause no confusion, since the will be clear from the context. ### Operators {#operators.unnumbered} The linear operators $A S^2 \to S^2$ and $B : S^2 \to V'$ are defined as follows. $\bSigma (\bsigma,\bchi) S^2$ $\bT = (\btau,\bmu) \in S^2$, let $A\bSigma$ be defined through $$\label{eq:Definition_of_a} \dual{\bT}{A\bSigma}_{S^2} = \int_\Omega \btau \dprod \bsigma \, \dx + \int_\Omega \bmu \dprod \H^{-1} \, \dx.$$ The term \, \dual{A \bSigma}{\bSigma}_{S^2}$ corresponds to energy with the state Here and $\H^{-1}(x)$ are maps from $\S$ to $\S$ (i.e., they are fourth order tensors) which may depend on the spatial $x$. For (\bsigma,\bchi) \in and \in let $$\label{eq:Definition_of_b} \dual{B\bSigma}{\bv}_{V',V} \int_\Omega \bsigma \dprod \bvarepsilon(\bv) \, \dx.$$ $\bvarepsilon(\bv) = \big(\nabla \bv + (\nabla \bv)^\top\big)/2$ denotes (linearized) strain ### Standing assumptions {#standing-assumptions.unnumbered} Throughout the we require \[asm:standing\_assumptions\] 1. The domain $\Omega \subset $d = 3$ is a bounded Lipschitz domain in the sense | $, and the induced Nemytzki operAtors, e.g. $\DD : H^1(0, t; S^2 ) \to H^1(0, t; S)$ aNd $\Dd : L^2(0, t; S^2) \to l^2(0, T; S)$, wIth the same symbOL. ThiS will cause no confusion, sInce tHe MEaniNG wIll be Clear frOM tHE ConTeXt.
### opeRaTOrS {#operAtoRs.unnumBered}
The liNeaR oPerators $A : S^2 \to s^2$ AnD $B : S^2 \to V'$ are dEfiNed as follows. for $\BSigma = (\BsIgmA,\Bchi) \iN S^2$ aNd $\bT = (\bTau,\bmu) \IN S^2$, let $A\BSigma$ be dEfINed thrOUgh $$\labeL{EQ:DEfinItion_of_a}
\dual{\bT}{A\bSIGmA}_{s^2} = \int_\Omega \btau \dProd \C^{-1} \bSiGMa \, \DX + \Int_\omeGa \bmu \dprod \h^{-1} \bChi \, \dx.$$ tHe term $(1/2) \, \dUAl{a \BsIgmA}{\BSigma}_{S^2}$ corresPonds to the eNErgY assocIaTed WIth the StresS sTAte $\BSigma$. Here $\C^{-1}(X)$ and $\h^{-1}(x)$ are lineAr maps FRom $\S$ to $\S$ (I.E., they arE fourtH orDer TensORs) WhIch MaY DepENd On tHE spAtial varIaBlE $x$. For $\BSigMA = (\BSIgma,\BchI) \in S^2$ And $\bv \In V$, let $$\label{eq:defInitIOn_oF_b}
\duaL{B\bSiGma}{\bV}_{V',v} = - \int_\OMega \bsIgma \dPrOd \bvarepsilon(\bv) \, \Dx.$$ We Recall thaT $\bvArEpsIlOn(\bv) = \bIG(\nabla \Bv + (\nAblA \bv)^\top\bIg)/2$ denotES thE (lINEArIzed) strain tensor.
### StAnDINg AssumptiOns {#staNDiNg-ASsumptioNs.UnnUmbeRED}
ThroUghoUT tHe paper, wE requiRE
\[aSm:StandinG\_aSsumptIoNs\]
1. THe dOmain $\oMega \Subset \r^d$, $d = 3$ is a boUnded lIpschitz domain IN the sense | $, and the induced Nemytzk i operator s, e. g.$\D D: H^ 1(0, T; S^2 ) \toH ^1(0 , T; S)$ and $\DD : L^ 2(0,T; S^2) \t o L^2 (0, T;S )$ , wit hth e s am e s ymbol . T his wil l cause no co nf usion, since th e meaningwil l be clear f rom the c on tex t .
## # O perat ors {# o perato rs.unnumb er e d}
Th e linear o pe rato rs $A : S^2 \to S ^ 2$ and $B : S^2 \ to V'$ a r ed e fin edas follows .For $ \ bSigma= ( \ b s igm a ,\bchi) \in S ^2$ and $\b T =(\btau ,\ bmu ) \in S ^2$,le t $A \bSigma$ be def ined thro ugh $$ \ label{e q :Defini tion_o f_a }
\ d ua l{ \bT }{ A \bS i gm a}_ { S^2 } = \int _\ Om ega \ btau \ d p rod\C^ {-1} \bsi gma \, \dx +\in t_\O m ega \bmu \dpr od \ H^ {-1}\bchi\, \d x. $$ The term $(1 /2)\, \dual{ A \ bS igm a} {\bSi g ma}_{S ^2} $ c orrespo nds tot heen e r g yassociated with th es t re ss state $\bSi g ma $. Here $\C ^{ -1} (x)$ a nd $\ H^{- 1 }( x)$ arelinear ma ps from $ \S $ to $ \S $ ( i.e ., th e y ar e four th order tens o rs) which mayd epend on thes pa t i al vari abl e $x$. For$\bS i gma= (\ b si gma , \bchi ) \in S ^ 2$ and $\bv \in V$, le t$$\lab el{eq :Definition_o f_b}
\ d u a l{B\bSig ma}{ \ bv } _{V',V} = - \i nt_\O mega \bsig m a \dprod \bva repsilon (\bv) \,\ d x.$$ Werec all th at$ \ bv arepsilon(\bv ) = \b ig (\nabla \b v + (\n abl a \ bv) ^\t op \big)/2$denotesth e(l in ear ized) strain t en sor .
## # Sta n ding a ssump tion s{# s tan ding-as s um p t ions .u nn umbe red }
Thro ugho u t t he pape r, we req uir e
\[ as m: standin g\_assumption s\ ]
1. The d oma in $\O m e ga \subs et \R^d$, $d = 3$ is ab oundedLip schit z do main in t hesense | $, and_the induced_Nemytzki operators, e.g. $\DD :_H^1(0, T;_S^2_) \to_H^1(0,_T; S)$ and_$\DD : L^2(0,_T; S^2) \to L^2(0,_T; S)$, with_the_same symbol. This will cause no confusion, since the meaning will be clear from_the_context.
### Operators_{#operators.unnumbered}
The_linear_operators $A : S^2 \to_S^2$ and $B : S^2_\to V'$_are defined as follows. For $\bSigma = (\bsigma,\bchi)_\in_S^2$ and $\bT_= (\btau,\bmu) \in S^2$, let $A\bSigma$ be defined through_$$\label{eq:Definition_of_a}
\dual{\bT}{A\bSigma}_{S^2} =_\int_\Omega \btau \dprod_\C^{-1}_\bsigma_\, \dx + \int_\Omega_\bmu \dprod \H^{-1} \bchi \, \dx.$$_The term $(1/2) \, \dual{A \bSigma}{\bSigma}_{S^2}$_corresponds to the energy associated with the_stress state $\bSigma$. Here $\C^{-1}(x)$ and_$\H^{-1}(x)$ are linear maps from_$\S$ to_$\S$ (i.e., they are fourth_order tensors) which_may depend_on the spatial_variable $x$. For $\bSigma = (\bsigma,\bchi)_\in S^2$ and_$\bv \in V$, let $$\label{eq:Definition_of_b}
__ \dual{B\bSigma}{\bv}_{V',V} =_-_\int_\Omega_\bsigma \dprod_\bvarepsilon(\bv) \, \dx.$$_We_recall that_$\bvarepsilon(\bv)_= \big(\nabla \bv + (\nabla \bv)^\top\big)/2$_denotes_the (linearized) strain tensor.
### Standing assumptions {#standing-assumptions.unnumbered}
Throughout_the paper, we require
\[asm:standing\_assumptions\]
1.__The domain $\Omega \subset_\R^d$, $d = 3$ is_a bounded Lipschitz domain in the_sense |
Electronic Structure of disordered alloys, surfaces and interfaces.”, Kluwer Academic Publishers (1997). A. Chattopadhyay, S. Das Sarma and A.J. Millis, Phys. Rev. Lett. [**87**]{}, 227202 (2001). In the case of $S=1/2$, this corresponds to a singlet state $E_{S}=-\frac{3}{4}J$ with degeneracy 1 and a triplet state $E_{T}=+\frac{1}{4} J$ with degeneracy 3. In GaAs the volume of the unit cell is $v=a_{0}^{3}/4$ where $a_{0}=0.565$ nm. For simplicity, our calculations are performed on a simple cubic lattice, thus the lattice spacing we take is $a_{1}=(a_{0}^{3}/4)^{1/3}$. Assuming a larger $n_{h}$ of order $0.3-0.4$ leads to a value of $J$ weakly affected (only within $10 \%$). J. Okabayashi [*et al.*]{}, Phys. Rev. B [**58**]{}, R4211 (1998).
---
abstract: |
Large scale tensors, including large scale Hankel tensors, have many applications in science and engineering. In this paper, we propose an inexact curvilinear search optimization method to compute Z- and H-eigenvalues of $m$th order $n$ dimensional Hankel tensors, where $n$ is large. Owing to the fast Fourier transform, the computational cost of each iteration of the new method is about $\mathcal{O}(mn\log(mn))$. Using the Cayley transform, we obtain an effective curvilinear search scheme. Then, we show that every limiting point of iterates generated by the new algorithm is an eigen-pair of Hankel tensors. Without the assumption of a second-order sufficient condition, we analyze the linear convergence rate of iterate sequence by the Kurdyka-[Ł]{}ojasiewicz property. Finally, numerical experiments for Hankel tensors, whose dimension may up to one million, are reported to show the efficiency of the proposed curvilinear search method.
[**Key words:**]{} Cayley transform, curvilinear search, eigenvalue, fast Fourier transform, Hankel tensor, Kurdyka-[Ł]{}ojasiewicz property, large scale tensor.
[**AMS subject classifications (2010):**]{} 15A18, 15A69, 65F15, 65 | Electronic Structure of disordered alloys, surfaces and interfaces. ”, Kluwer Academic Publishers (1997). A. Chattopadhyay, S. Das Sarma and A.J. Millis, Phys. Rev. Lett. [ * * 87 * * ] { }, 227202 (2001). In the subject of $ S=1/2 $, this correspond to a singlet state $ E_{S}=-\frac{3}{4}J$ with degeneracy 1 and a triplet country $ E_{T}=+\frac{1}{4 } J$ with degeneracy 3. In GaAs the volume of the whole cellular telephone is $ v = a_{0}^{3}/4 $ where $ a_{0}=0.565 $ nm. For simplicity, our calculations are performed on a simple cubic wicket, thus the lattice spacing we take is $ a_{1}=(a_{0}^{3}/4)^{1/3}$. simulate a larger $ n_{h}$ of order $ 0.3 - 0.4 $ leads to a value of $ J$ weakly affected (only within $ 10 \%$). J. Okabayashi [ * et al. * ] { }, Phys. Rev. B [ * * 58 * * ] { }, R4211 (1998).
---
abstract: |
bombastic scale tensors, including big scale Hankel tensors, have many applications in science and technology. In this paper, we aim an inexact curvilinear search optimization method to compute Z- and H - eigenvalue of $ m$th order $ n$ dimensional Hankel tensors, where $ n$ is large. Owing to the fast Fourier transform, the computational cost of each iteration of the new method is about $ \mathcal{O}(mn\log(mn))$. Using the Cayley transform, we obtain an effective curvilinear search scheme. Then, we show that every limiting point of iterates generated by the new algorithm is an eigen - couple of Hankel tensor. Without the assumption of a second - decree sufficient circumstance, we analyze the linear convergence rate of iterate sequence by the Kurdyka-[Ł]{}ojasiewicz property. Finally, numerical experiments for Hankel tensor, whose dimension may up to one million, are reported to show the efficiency of the proposed curvilinear search method.
[ * * Key actor's line :* * ] { } Cayley transform, curvilinear search, eigenvalue, fast Fourier transform, Hankel tensor, Kurdyka-[Ł]{}ojasiewicz property, large scale tensor.
[ * * AMS subject categorization (2010 ): * * ] { } 15A18, 15A69, 65F15, 65 | Eleftronic Structure of disurdered alloys, surfacev and jnterfacds.”, Kluwer Academic Publisherd (1997). A. Chqttopadhyay, S. Das Sarmx and A.J. Lillis, Pyys. Cev. Lett. [**87**]{}, 227202 (2001). In tis case of $S=1/2$, tgls coxrxsponds to a sikglet state $E_{S}=-\frac{3}{4}J$ with geeeueracy 1 and a triplet state $E_{T}=+\frac{1}{4} J$ with drgfneracy 3. In GaWs tnq vomlmt of the unit cell is $v=a_{0}^{3}/4$ where $a_{0}=0.565$ nj. For spmplicity, our calvulations are performed on a slmple cubic latticf, thus the oattyxe spacing wd take is $a_{1}=(a_{0}^{3}/4)^{1/3}$. Assuming z larger $n_{h}$ of order $0.3-0.4$ leads to x valbe of $J$ weajlt avxected (only withpn $10 \%$). J. Okabayashi [*et dl.*]{}, Phys. Rev. B [**58**]{}, R4211 (1998).
---
absbract: |
Large scale tensors, including large ssale Hankal tensors, have mant qppliwatimns kb sziehcx ahd englnexring. In thjs paper, we propose an inexact ctgfilinear seadch opeiiization method to compute Z- and H-eigennaluss of $m$th order $n$ dimenwional Hankel tensors, where $n$ ys large. Owing to the fast Fourier transform, the wompuvagiouql corr lf each iteration of the new method is about $\iztncsl{O}(mn\log(mn))$. Usikg the Cayley tranxflrk, we obtain an effeccjvs curvilinear searfh scheie. Thwn, we shor thst every limiting point of uterates genvratwd by the new algoxithm is an zigen-psir og Hankel tensors. Withouc the zssumption lf a secoha-order sufficieng cpngition, wt analyze the linewr converjence rate ow itgrate sqquence by the Kurdyka-[Ł]{}ojasiewicz pgoperjy. Findlly, numerlcal experiments for Hankel tensors, whose dimgnshon may up co one million, are weported to shpw the zfficidncy of thv proposev curvilineaw search methmf.
[**Key worvs:**]{} Cayley trabsfoem, curvkuinear search, rigenvaluv, yast Fouruer transform, Hankcl tevaor, Kurdyka-[Ł]{}ojasntwixz property, larbe rcaje txnsor.
[**AMS subjewt cuasroficagions (2010):**]{} 15A18, 15A69, 65N15, 65 | Electronic Structure of disordered alloys, surfaces and Academic (1997). A. S. Das Sarma Lett. 227202 (2001). In case of $S=1/2$, corresponds to a singlet state $E_{S}=-\frac{3}{4}J$ degeneracy 1 and a triplet state $E_{T}=+\frac{1}{4} J$ with degeneracy 3. In GaAs volume of the unit cell is $v=a_{0}^{3}/4$ where $a_{0}=0.565$ nm. For simplicity, our are on simple lattice, thus the lattice spacing we take is $a_{1}=(a_{0}^{3}/4)^{1/3}$. Assuming a larger $n_{h}$ of order $0.3-0.4$ to a value of $J$ weakly affected (only $10 \%$). J. Okabayashi al.*]{}, Phys. Rev. B [**58**]{}, (1998). abstract: | scale including scale Hankel tensors, many applications in science and engineering. In this paper, we propose an inexact curvilinear search optimization method compute Z- of $m$th $n$ Hankel where $n$ is to the fast Fourier transform, the each iteration of the new method is about Using the transform, we obtain an effective curvilinear scheme. Then, we show that every limiting point iterates generated by the new algorithm is an eigen-pair of Hankel tensors. Without the assumption second-order sufficient condition, we the linear convergence of sequence the property. Finally, experiments for Hankel tensors, whose dimension may up to one million, reported to show the efficiency of the proposed curvilinear search words:**]{} transform, curvilinear search, fast Fourier transform, Hankel Kurdyka-[Ł]{}ojasiewicz large scale tensor. [**AMS (2010):**]{} 15A69, | Electronic Structure of disoRdered alloYs, surFacEs aNd InteRfacEs.”, Kluwer AcademIC PubLishers (1997). A. Chattopadhyay, S. das SaRmA And A.j. miLlis, PHys. Rev. LETt. [**87**]{}, 227202 (2001). iN The CaSe Of $S=1/2$, ThIS cOrresPonDs to a siNglet state $e_{S}=-\fRaC{3}{4}J$ with degeneRAcY 1 and a triplEt sTate $E_{T}=+\frac{1}{4} J$ wIth DegeneRaCy 3. IN gaAs tHe vOlume Of the uNIt cell Is $v=a_{0}^{3}/4$ where $A_{0}=0.565$ nM. for simPLicity, oUR CaLculAtions are performeD On A Simple cubic latTice, thUs THe LATtiCe sPacing we taKe Is $a_{1}=(a_{0}^{3}/4)^{1/3}$. ASSuming a LArGER $N_{h}$ oF Order $0.3-0.4$ leads to a Value of $J$ weaKLy aFfecteD (oNly WIthin $10 \%$). J. okabaYaSHi [*eT al.*]{}, Phys. Rev. B [**58**]{}, r4211 (1998).
---
absTract: |
LargE scale TEnsors, iNCluding Large sCalE HaNkel TEnSoRs, hAvE ManY ApPliCAtiOns in sciEnCe And enGineERING. In tHis PapeR, we prOpose an inexacT cuRvilINeaR searCh optImizAtIon meThod to CompuTe z- and H-eigenvalueS of $m$Th order $n$ dImeNsIonAl hankeL TensorS, whEre $N$ is largE. Owing tO The FaST fOuRier transform, the coMpUTAtIonal cosT of eacH ItErATion of thE nEw mEthoD IS abouT $\matHCaL{O}(mn\log(mN))$. Using THe caYley traNsForm, we ObTaiN an EffecTIve cUrviliNear searCh schEMe. Then, we show thAT every limitinG PoINT oF IterAteS generated bY the NEw alGoriTHm Is aN Eigen-Pair oF HANkEL tensors. Without the aSsUmptioN of a sEcond-order sufFicient conDITIon, we anaLyze THe LInear convergenCe ratE of iterate SEquence bY the KUrdyka-[Ł]{}oJasiewicz PROperty. FiNalLy, nUmeRicAL ExPeriments for HANKel tEnSors, whoSe dImensioN maY up To oNe mIlLion, are rePorted to ShOw ThE eFfiCiencY Of the proPoSed CuRviLineaR Search MethoD.
[**Key WoRdS:**]{} cayLey tranSFoRM, CurvIlInEar sEarCh, EigenValuE, FasT FourieR transforM, HaNKel tEnSoR, KurdykA-[Ł]{}ojasiewicz pRoPerty, large ScAle Tensor.
[**ams subject Classifications (2010):**]{} 15A18, 15A69, 65F15, 65 | Electronic Structure of di sordered a lloys , s urf ac es a nd i nterfaces.”, K l uwer Academic Publishers ( 1997) .A . Ch a tt opadh yay, S. Da s Sar ma a ndA. J .Milli s,Phys. R ev. Lett.[** 87 **]{}, 22720 2 ( 2001). Inthe case of $S= 1/2 $, thi scor r espon dsto asingle t state $E_{S}=- \f r ac{3}{ 4 }J$ wit h de gene racy 1 and a trip l et state $E_{T}=+ \frac{ 1} { 4} J $ w ith degenerac y3. In GaAs th e v o l u meo f the unit ce ll is $v=a_ { 0}^ {3}/4$ w her e $a_{0 }=0.5 65 $ nm . For simpl icit y, our ca lculat i ons are perform ed ona s imp le c u bi clat ti c e,t hu s t h e l attice s pa ci ng we tak e i s $a_ {1} =(a_ {0}^{ 3}/4)^{1/3}$. As sumi n g a larg er $n _{h} $of or der $0 .3-0. 4$ leads to a val ue o f $J$ wea kly a ffe ct ed (o n ly wit hin $1 0 \%$). J. Oka b aya sh i [ *e t al.*]{}, Phys. R ev . B[**58**] {}, R4 2 11 ( 1 998).
- --
ab stra c t : |
L a rg e scaletensor s ,in cluding l arge s ca leHan kel t e nsor s, hav e many a pplic a tions in scien c e and enginee r in g . I n thi s p aper, we pr opos e aninex a ct cu r vilin ear s ea r ch optimization method t o comp ute Z - and H-eigen values of$ m $ th order $n$ di m ensional Hanke l ten sors, wher e $n$ islarge . Owingto the fa s t Fourier tr ans for m,t h ecomputational c ostof each i ter ation o f t henew me th od is abo ut $\mat hc al {O }( mn\ log(m n ))$. Usi ng th eCay ley t r ansfor m, we obt ai na n e ffectiv e c u r vili ne ar sea rch s cheme . Th e n,we show that eve ryl imit in gpoint o f iterates ge ne rated by t he ne w algo r i thm is a n eigen-pair of Hankelt ensors. Wi thout the assumpti onof a s eco n d-orde r suff icien tcon d i tion, w eana ly ze the lin e a r c onver ge ncerate of iterate sequenceb y t he Kurdyka-[Ł ]{} ojas i e wi czp ro p ert y. Fin a l ly, numerical e xperiments f o rHankel ten s ors ,whose d imensio n may up to o ne millio n, are re po rted t o s how the ef ficiency of the p r opose d c urvil ine ar sea rc h m ethod .
[** Key w ords:* *] {} Cay ley t ra nsform,curvilinear search, eig envalu e, fa stFourier t ran s for m, Hankel ten sor, Kurdy ka- [Ł] {}oja sie w icz p rope r ty , l a rge s cale tensor.
[* * A MS subject cl a s s ifi catio ns( 2010): **]{ } 15A18, 15A69, 6 5 F15, 65 | Electronic Structure_of disordered_alloys, surfaces and interfaces.”,_Kluwer Academic_Publishers_(1997). A._Chattopadhyay,_S. Das Sarma_and A.J. Millis,_Phys. Rev. Lett. [**87**]{},_227202 (2001). In_the_case of $S=1/2$, this corresponds to a singlet state $E_{S}=-\frac{3}{4}J$ with degeneracy 1 and_a_triplet state_$E_{T}=+\frac{1}{4}_J$_with degeneracy 3. In GaAs_the volume of the unit_cell is_$v=a_{0}^{3}/4$ where $a_{0}=0.565$ nm. For simplicity, our calculations are_performed_on a simple_cubic lattice, thus the lattice spacing we take is_$a_{1}=(a_{0}^{3}/4)^{1/3}$. Assuming a larger $n_{h}$ of_order $0.3-0.4$ leads_to_a_value of $J$ weakly_affected (only within $10 \%$). J._Okabayashi [*et al.*]{}, Phys. Rev. B_[**58**]{}, R4211 (1998).
---
abstract: |
_ Large scale tensors, including large_scale Hankel tensors, have many_applications in_science and engineering. In this_paper, we propose_an inexact_curvilinear search optimization_method to compute Z- and H-eigenvalues_of $m$th order_$n$ dimensional Hankel tensors, where $n$_is_large. Owing to_the_fast_Fourier transform,_the computational cost_of_each iteration_of_the new method is about $\mathcal{O}(mn\log(mn))$._Using_the Cayley transform, we obtain an effective_curvilinear search scheme. Then,_we_show that every limiting_point of iterates generated by_the new algorithm is an eigen-pair_of Hankel_tensors. Without_the assumption of a second-order sufficient condition, we analyze the linear_convergence rate of iterate sequence by_the Kurdyka-[Ł]{}ojasiewicz property. Finally,_numerical experiments_for_Hankel tensors, whose_dimension_may up_to one million, are reported to show_the efficiency_of the proposed curvilinear search method.
_ [**Key_words:**]{}_Cayley transform, curvilinear search, eigenvalue, fast_Fourier transform, Hankel tensor, Kurdyka-[Ł]{}ojasiewicz property,_large scale tensor.
__[**AMS_subject classifications (2010):**]{} 15A18, 15A69,_65F15, 65 |
, if it is not, the energy is always raised. It follows that the DMC energy is always less than or equal to the VMC energy with the same trial wave function, and always greater than or equal to the exact ground-state energy.
The fixed-node DMC algorithm described above is extremely inefficient and a vastly superior algorithm can be obtained by introducing an importance sampling transformation [@grimm_1971; @kalos_1974]. Consider the mixed distribution, $$\begin{aligned}
\label{eq:f}
f({\bf R},t) = \Psi_{\rm T}({\bf R}) \Phi({\bf R},t) \;,\end{aligned}$$ which has the same sign everywhere if and only if the nodal surface of $\Phi({\bf R},t)$ equals that of $\Psi_{\rm T}({\bf R})$. Substituting in equation (\[eq:imaginary\_time\_se\]) for $\Phi$ we obtain $$\begin{aligned}
\label{eq:importance_sampled_imaginary_time_se}
-\frac{\partial f}{\partial t} = -\frac{1}{2} \nabla_{\bf R}^2 f + \nabla_{\bf
R} \cdot [{\bf v}f] + [E_{\rm L}-E_{\rm T}]f \;,\end{aligned}$$ where the $3N$-dimensional drift velocity is defined as $$\begin{aligned}
\label{eq:drift_velocity}
{\bf v}({\bf R}) = \Psi_{\rm T}^{-1}({\bf R}) \nabla_{\bf R} \Psi_{\rm T}({\bf
R}) \;.\end{aligned}$$ The three terms on the right-hand side of equation (\[eq:importance\_sampled\_imaginary\_time\_se\]) correspond to diffusion, drift and branching processes, respectively. The importance sampling transformation has several consequences. First, the density of configurations is increased where $|\Psi_{\rm T}|$ is large, so that the more important parts of the wave function are sampled more often. Second, the rate of branching is now controlled by the local energy which is normally a much smoother function than the potential energy. This is particularly important for the Coulomb interaction, which diverges when particles are coincident. The importance sampling transformation, together with an algorithm that imposes $f({\bf R},t)
\ge 0$, ensures that $\Psi_{\rm T}$ and $\Phi({\bf | , if it is not, the energy is always raised. It follows that the DMC department of energy is constantly less than or equal to the VMC energy with the like trial wave routine, and always greater than or adequate to the accurate ground - state energy.
The fixed - node DMC algorithm identify above is extremely inefficient and a vastly ranking algorithm can be prevail by introducing an importance sampling transformation [ @grimm_1971; @kalos_1974 ]. Consider the mixed distribution, $ $ \begin{aligned }
\label{eq: f }
f({\bf R},t) = \Psi_{\rm T}({\bf R }) \Phi({\bf R},t) \;,\end{aligned}$$ which has the like sign everywhere if and only if the nodal surface of $ \Phi({\bf R},t)$ equal that of $ \Psi_{\rm T}({\bf R})$. Substituting in equation (\[eq: imaginary\_time\_se\ ]) for $ \Phi$ we receive $ $ \begin{aligned }
\label{eq: importance_sampled_imaginary_time_se }
-\frac{\partial f}{\partial t } = -\frac{1}{2 } \nabla_{\bf R}^2 f + \nabla_{\bf
roentgen } \cdot [ { \bf v}f ] + [ E_{\rm L}-E_{\rm T}]f \;,\end{aligned}$$ where the $ 3N$-dimensional drift velocity is defined as $ $ \begin{aligned }
\label{eq: drift_velocity }
{ \bf v}({\bf R }) = \Psi_{\rm T}^{-1}({\bf R }) \nabla_{\bf R } \Psi_{\rm T}({\bf
R }) \;.\end{aligned}$$ The three terms on the right - bridge player side of equation (\[eq: importance\_sampled\_imaginary\_time\_se\ ]) correspond to diffusion, drift and branching processes, respectively. The importance sampling transformation has several consequences. First, the density of configurations is increased where $ |\Psi_{\rm T}|$ is large, so that the more important parts of the wave function are sampled more frequently. Second, the rate of branching is now master by the local department of energy which is normally a a lot smoother function than the potential energy. This is particularly important for the Coulomb interaction, which diverges when particles are coincident. The importance sample transformation, together with an algorithm that imposes $ f({\bf R},t)
\ge 0 $, ensures that $ \Psi_{\rm T}$ and $ \Phi({\bf | , if it is not, the energy is always raised. Nr follmws thzt the DOC energy is always less thai or equao to the VMC energy wigh the sale trial wavt function, and alxzys greater tgwn ox xqual to the exsct ground-vtate energy.
Tha wired-node DMC algorithm described abovq is exyrfmely inefficignt amq a bastly superior algorithm can be ogtained by introducimg an importance sampling hrandformation [@grimm_1971; @kwlos_1974]. Considgd trw mixed distfibution, $$\btgnn{aligned}
\lagel{eq:f}
f({\bf R},t) = \Psi_{\rm T}({\bf R}) \Phi({\bf R},t) \;,\eud{aligned}$$ wyixh jds the same sign everywhere lg and mnly if the nodal surnace mf $\Phi({\bf R},t)$ equals that mf $\Psi_{\rm T}({\bf R})$. Subftituting iu equation (\[eq:imaginart\_tume\_se\]) for $\Phi$ qe ubtzii $$\bsgin{allgnxd}
\label{eq:imlortance_sampled_imaginary_time_se}
-\grwb{\lartial f}{\parfial t} = -\srac{1}{2} \nabla_{\bf R}^2 f + \nabla_{\bf
R} \cdot [{\bf v}f] + [E_{\rj L}-E_{\rm T}]f \;,\end{aligned}$$ whwre the $3N$-dimensional frift veljcity is defined as $$\begin{aligned}
\label{eq:drift_velochty}
{\bf x}({\bf R}) = \Psl_{\rm T}^{-1}({\bf R}) \nabla_{\bf R} \Psi_{\rm T}({\bf
R}) \;.\end{aligned}$$ The fhtev terms on the rinht-hand side of eqiahipg (\[eq:importancg\_sampleb\_jmzginary\_time\_se\]) corrfspond jo difdusion, dryft snd branching processes, respectively. Thv importance sampling cransformatiun hss seferal consequences. Firsc, the sensity of fonfigurafkons is increasea wmera $|\Psi_{\rm U}|$ is large, so that the more impprtant oartx of tre wave fujctiok are sampled more lften. Sacond, the gate of branching is now controlled by the lovan etergy whnch is normally a mtch smoother fonction tkan thd potentiam energb. This is pawticularly im[lrtant for tie Coulomf inreraxtion, wfkch diverges wnen partiblts are coibcident. The importence aampling transfirmqtion, together eitf ag wljoritrk that imposas $f({\cf F},y)
\ge 0$, dnsures thab $\Pri_{\rm T}$ and $\Phi({\bf | , if it is not, the energy raised. follows that DMC energy is to VMC energy with same trial wave and always greater than or equal the exact ground-state energy. The fixed-node DMC algorithm described above is extremely inefficient a vastly superior algorithm can be obtained by introducing an importance sampling transformation @kalos_1974]. the distribution, \label{eq:f} f({\bf R},t) = \Psi_{\rm T}({\bf R}) \Phi({\bf R},t) \;,\end{aligned}$$ which has the same sign everywhere and only if the nodal surface of $\Phi({\bf equals that of $\Psi_{\rm R})$. Substituting in equation (\[eq:imaginary\_time\_se\]) $\Phi$ obtain $$\begin{aligned} -\frac{\partial t} -\frac{1}{2} \nabla_{\bf R}^2 + \nabla_{\bf R} \cdot [{\bf v}f] + [E_{\rm L}-E_{\rm T}]f \;,\end{aligned}$$ where the $3N$-dimensional drift velocity is as $$\begin{aligned} v}({\bf R}) \Psi_{\rm R}) R} \Psi_{\rm T}({\bf The three terms on the right-hand (\[eq:importance\_sampled\_imaginary\_time\_se\]) correspond to diffusion, drift and branching processes, The importance transformation has several consequences. First, the of configurations is increased where $|\Psi_{\rm T}|$ is so that the more important parts of the wave function are sampled more often. Second, of branching is now by the local which normally much function than potential energy. This is particularly important for the Coulomb interaction, which when particles are coincident. The importance sampling transformation, together with that $f({\bf R},t) \ge ensures that $\Psi_{\rm T}$ $\Phi({\bf | , if it is not, the energy is alwayS raised. It fOllowS thAt tHe dMC eNergY is always less tHAn or Equal to the VMC energy witH the sAmE TriaL WaVe funCtion, anD AlWAYs gReAtEr tHaN Or Equal To tHe exact Ground-statE enErGy.
The fixed-noDE DmC algorithM deScribed above Is eXtremeLy IneFFicieNt aNd a vaStly suPErior aLgorithm cAn BE obtaiNEd by intRODuCing An importance samplINg TRansformation [@gRimm_1971; @kaLoS_1974]. coNSIdeR thE mixed distRiButioN, $$\Begin{alIGnED}
\LAbeL{Eq:f}
f({\bf R},t) = \Psi_{\rm t}({\bf R}) \Phi({\bf R},t) \;,\ENd{aLigned}$$ WhIch HAs the sAme siGn EVerYwhere if and Only If the nodaL surfaCE of $\Phi({\bF r},t)$ equalS that oF $\PsI_{\rm t}({\bf R})$. sUbStItuTiNG in EQuAtiON (\[eq:ImaginarY\_tImE\_se\]) foR $\Phi$ WE OBTain $$\BegIn{alIgned}
\Label{eq:importAncE_samPLed_ImagiNary_tIme_sE}
-\fRac{\paRtial f}{\PartiAl T} = -\frac{1}{2} \nabla_{\bf R}^2 f + \nAbla_{\Bf
R} \cdot [{\bf V}f] + [E_{\Rm l}-E_{\rM T}]F \;,\end{aLIgned}$$ wHerE thE $3N$-dimenSional dRIft VeLOCItY is defined as $$\begin{aLiGNEd}
\Label{eq:dRift_veLOcItY}
{\Bf v}({\bf R}) = \PsI_{\rM T}^{-1}({\bF R}) \naBLA_{\bf R} \PSi_{\rm t}({\Bf
r}) \;.\end{aligNed}$$ The THrEe Terms on ThE right-HaNd sIde Of equATion (\[Eq:impoRtance\_saMpled\_IMaginary\_time\_se\]) COrrespond to diFFuSIOn, DRift And Branching prOcesSEs, reSpecTIvEly. tHe impOrtanCe SAmPLing transformation hAs SeveraL consEquences. First, The density OF COnfiguraTionS Is INcreased where $|\PSi_{\rm T}|$ Is large, so tHAt the morE impoRtant parTs of the waVE Function Are SamPleD moRE OfTen. Second, the rATE of bRaNching iS noW controLleD by The LocAl Energy whiCh is normAlLy A mUcH smOotheR Function ThAn tHe PotEntiaL Energy. this iS parTiCuLArlY importANt FOR the coUlOmb iNteRaCtion, WhicH DivErges whEn particlEs aRE coiNcIdEnt. The iMportance sampLiNg transforMaTioN, togetHER with an aLgorithm that imposes $f({\bf R},T)
\Ge 0$, ensurEs tHat $\PsI_{\rm T}$ And $\Phi({\bf | , if it is not, the energy is always rais ed. It f ollo ws t hat the DMC en e rgyis always less than or equa lt o th e V MC en ergy wi t ht h e s am etri al wa ve fu nct ion, an d always g rea te r than or eq u al to the ex act ground-stat e e nergy.
The fixed -no de DM C algo r ithm d escribedab o ve ise xtremel y in effi cient and a vastl y s u perior algorit hm can b e o b t ain edby introdu ci ng an importa n ce s a mpl i ng transforma tion [@grim m _19 71; @k al os_ 1 974].Consi de r th e mixed dis trib ution, $$ \begin { aligned }
\label {eq:f}
f( {\b f R} , t) = \P si _ {\r m T }({ \ bfR}) \Phi ({ \b f R}, t) \ ; , \ e nd{a lig ned} $$ wh ich has the s ame sig n ev erywh ere i f an donlyif the noda lsurface of $\Ph i({\ bf R},t)$ eq ua lsth at of $\Psi_ {\r m T }({\bfR})$. S u bst it u t i ng in equation (\[eq :i m a gi nary\_ti me\_se \ ]) f o r $\Phi$ w e o btai n $$\be gin{ a li gned}
\l abel{e q :i mp ortance _s ampled _i mag ina ry_ti m e_se }
-\fr ac{\part ial f } {\partial t} = -\frac{1}{2}\ na b l a_ { \bfR}^ 2 f + \nabl a_{\ b f
R } \c d ot [{ \ bf v} f] +[E _ {\ r m L}-E_{\rm T}]f \; ,\ end{al igned }$$ where the $3N$-dime n s i onal dri ft v e lo c ity is defined as $ $\begin{al i gned}
\l abel{ eq:drift _velocity } {\bf v}( {\b f R })= \ P s i_ {\rm T}^{-1}( { \ bf R }) \nabla _{\ bf R} \ Psi _{\ rmT}( {\ bf
R}) \; .\end{al ig ne d} $$ Th e thr e e termson th erig ht-ha n d side of e quat io n( \[e q:impor t an c e \_sa mp le d\_i mag in ary\_ time \ _se \]) cor respond t o d i ffus io n, driftand branching p rocesses,re spe ctivel y . The imp ortance sampling transf o rmation ha s sev eral conseque nce s. Fir st, the de nsityof co nf igu r a tions i sinc re ased where $ |\P si_{\ rm T}| $ is la rge, so that the m o reimportant par tsof t h e w ave fu n cti on are s ampled more oft en. Second ,t he rate of b r anc hi ng is n ow cont rolle d by the local en ergy whic his n o r mal ly a muchsmoother function thant he pote nti al ene rg y.Thisis par t icu larly impor ta nt for theCo ulomb in teraction, which diverg es whe n par tic les are c oin c ide nt. The i mpor tance samp lin g t ransf orm a tion, tog e th erw ith a n al g orithm th a timp o s es $f({\bf R} , t )
\g e 0$, en s ures t hat$\Psi_{\rm T}$ an d $\Phi({\bf | , if_it is_not, the energy is_always raised._It_follows that_the_DMC energy is_always less than_or equal to the_VMC energy with_the_same trial wave function, and always greater than or equal to the exact ground-state_energy.
The_fixed-node DMC_algorithm_described_above is extremely inefficient and_a vastly superior algorithm can_be obtained_by introducing an importance sampling transformation [@grimm_1971; @kalos_1974]._Consider_the mixed distribution,_$$\begin{aligned}
\label{eq:f}
f({\bf R},t) = \Psi_{\rm T}({\bf R}) \Phi({\bf R},t) \;,\end{aligned}$$_which has the same sign everywhere_if and only_if_the_nodal surface of $\Phi({\bf_R},t)$ equals that of $\Psi_{\rm T}({\bf_R})$. Substituting in equation (\[eq:imaginary\_time\_se\]) for_$\Phi$ we obtain $$\begin{aligned}
\label{eq:importance_sampled_imaginary_time_se}
-\frac{\partial f}{\partial t} =_-\frac{1}{2} \nabla_{\bf R}^2 f + \nabla_{\bf
_R} \cdot [{\bf v}f] +_[E_{\rm L}-E_{\rm_T}]f \;,\end{aligned}$$ where the $3N$-dimensional_drift velocity is_defined as_$$\begin{aligned}
\label{eq:drift_velocity}
{\bf v}({\bf R})_= \Psi_{\rm T}^{-1}({\bf R}) \nabla_{\bf R}_\Psi_{\rm T}({\bf
R}) \;.\end{aligned}$$_The three terms on the right-hand_side_of equation (\[eq:importance\_sampled\_imaginary\_time\_se\])_correspond_to_diffusion, drift_and branching processes,_respectively._The importance_sampling_transformation has several consequences. First, the_density_of configurations is increased where $|\Psi_{\rm T}|$_is large, so that_the_more important parts of_the wave function are sampled_more often. Second, the rate of_branching is_now controlled_by the local energy which is normally a much smoother function_than the potential energy. This is_particularly important for the_Coulomb interaction,_which_diverges when particles_are_coincident. The_importance sampling transformation, together with an algorithm_that imposes_$f({\bf R},t)
\ge 0$, ensures that $\Psi_{\rm_T}$ and $\Phi({\bf |
A(1) \cong R \times_{R/I} R, \ (a,i) \mapsto (a, a+i),$$ the fiber product of the two copies of the natural homomorphism $R \to R/I$. Hence, if $R$ is a reduced ring, then so is $A(1)$.
Let us note the following.
\[lemma 3.1\] Let $(R,\m)$ be a $($not necessarily Noetherian$)$ local ring. Assume that $I \ne R$ or $\alpha \in \m$. Then $A(\alpha)$ is a local ring with maximal ideal $\m \times I$.
Let $(a,x) \in A(\alpha) \setminus (\m \times I)$. Then, $a+\alpha x \not\in \m$, since $a \not\in \m$ but $\alpha x \in \m$. Therefore, setting $b = a^{-1}$ and $y = -(a+\alpha x)^{-1}{\cdot}xb$, we get $(a,x)(b,y)=1$ in $A(\alpha)$. Hence, $A(\alpha)$ is a local ring, because $\m \times I$ is an ideal of $A(\alpha)$.
When $I=R$, $A(-1)$ is not a local ring, even if $(R,\m)$ is a local ring. In fact, assume that $A(-1)$ is a local ring. Then, because $\m \times R$ is a maximal ideal of $A(-1)$ and $(1,1) \not\in \m \times R$, we have $(1,1)(b,y)= (1,0)$ for some $(b,y) \in A(-1)$, so that $b=1$ and $y+b + (-1){\cdot}1{\cdot}y=0$. This is absurd.
In what follows, let $(R,\m)$ be a one-dimensional Cohen-Macaulay local ring with a canonical fractional ideal $K$. We set $S=R[K]$ and ${\mathfrak{c}}= R:S$. Let $T$ be a birational module-finite extension of $R$ (hence $R \subseteq T \subseteq \overline{R}$), and assume that $K \subseteq T$ but $R \ne T$. We set $I = R: | A(1) \cong R \times_{R / I } R, \ (a, i) \mapsto (a, a+i),$$ the fiber product of the two copy of the lifelike homomorphism $ R \to R / I$. therefore, if $ R$ is a reduced band, then so is $ A(1)$.
Let us notice the pursuit.
\[lemma 3.1\ ] Let $ (R,\m)$ be a $ ($ not necessarily Noetherian$)$ local ring. Assume that $ I \ne R$ or $ \alpha \in \m$. Then $ A(\alpha)$ is a local band with maximal ideal $ \m \times I$.
Let $ (a, x) \in A(\alpha) \setminus (\m \times I)$. Then, $ a+\alpha x \not\in \m$, since $ a \not\in \m$ but $ \alpha x \in \m$. Therefore, setting $ boron = a^{-1}$ and $ y = -(a+\alpha x)^{-1}{\cdot}xb$, we get $ (a, x)(b, y)=1 $ in $ A(\alpha)$. Hence, $ A(\alpha)$ is a local ring, because $ \m \times I$ is an ideal of $ A(\alpha)$.
When $ I = R$, $ A(-1)$ is not a local ring, even if $ (R,\m)$ is a local ring. In fact, assume that $ A(-1)$ is a local hoop. Then, because $ \m \times R$ is a maximal ideal of $ A(-1)$ and $ (1,1) \not\in \m \times R$, we have $ (1,1)(b, y)= (1,0)$ for some $ (b, y) \in A(-1)$, so that $ b=1 $ and $ y+b + (-1){\cdot}1{\cdot}y=0$. This is absurd.
In what follows, let $ (R,\m)$ embody a one - dimensional Cohen - Macaulay local ring with a canonical fractional ideal $ K$. We set $ S = R[K]$ and $ { \mathfrak{c}}= roentgen: S$. Let $ T$ be a birational module - finite extension of $ R$ (hence $ R \subseteq deoxythymidine monophosphate \subseteq \overline{R}$), and assume that $ K \subseteq T$ but $ R \ne T$. We set $ I = R: | A(1) \clng R \times_{R/I} R, \ (a,i) \mapsuo (a, a+i),$$ the fiber producv of ths two cooies of the natural homomorpiism $R \to R/I$. Hence, if $R$ is a reauced rinh, then si is $Q(1)$.
Let us novs the followihn.
\[lemmc 3.1\] Let $(R,\m)$ be a $($npt necessasily Noetheriat$)$ uoeal ring. Assume that $I \ne R$ or $\alpha \in \m$. Tnej $A(\alpha)$ is a jocak rinf with maximal ideal $\m \times I$.
Let $(z,x) \in A(\elpha) \setminus (\k \times I)$. Then, $a+\alpha x \noh\in \l$, since $a \not\in \m$ hut $\alpha x \in \i$. Therefore, sdtting $b = a^{-1}$ and $y = -(a+\alkha x)^{-1}{\cdot}xb$, we get $(a,x)(b,y)=1$ in $A(\alphx)$. Henee, $A(\alpha)$ iw q llwal ring, berause $\i \times I$ is an idean of $A(\akpha)$.
When $I=R$, $A(-1)$ ls nov a oocal ring, even if $(R,\m)$ is a local ring. Ig fact, asvuje that $A(-1)$ is a loxao rinc. Than, bdxaure $\j \vimss R$ id a maximal iseal of $A(-1)$ abd $(1,1) \not\in \m \times R$, wq have $(1,1)(b,y)= (1,0)$ for some $(f,y) \in A(-1)$, so that $b=1$ and $y+b + (-1){\cdot}1{\cdot}y=0$. This ps agsurd.
In what follows, ler $(R,\m)$ be a one-dimensiojal Cohen-Iacaulay local ring with a canonical fractional igeal $I$. We sct $S=F[J]$ wnd ${\mathfrak{c}}= R:S$. Let $T$ be a birational module-sjnotv extension of $R$ (mence $R \subseteq T \skbxgteq \overline{R}$), and assuje that $K \subseteq T$ but $T \ne T$. We set $I = R: | A(1) \cong R \times_{R/I} R, \ (a,i) a+i),$$ fiber product the two copies \to Hence, if $R$ a reduced ring, so is $A(1)$. Let us note following. \[lemma 3.1\] Let $(R,\m)$ be a $($not necessarily Noetherian$)$ local ring. Assume $I \ne R$ or $\alpha \in \m$. Then $A(\alpha)$ is a local ring maximal $\m I$. $(a,x) \in A(\alpha) \setminus (\m \times I)$. Then, $a+\alpha x \not\in \m$, since $a \not\in \m$ $\alpha x \in \m$. Therefore, setting $b = and $y = -(a+\alpha we get $(a,x)(b,y)=1$ in $A(\alpha)$. $A(\alpha)$ a local because \times is an ideal $A(\alpha)$. When $I=R$, $A(-1)$ is not a local ring, even if $(R,\m)$ is a local ring. In assume that a local Then, $\m R$ is a of $A(-1)$ and $(1,1) \not\in \m have $(1,1)(b,y)= (1,0)$ for some $(b,y) \in A(-1)$, that $b=1$ $y+b + (-1){\cdot}1{\cdot}y=0$. This is absurd. what follows, let $(R,\m)$ be a one-dimensional Cohen-Macaulay ring with a canonical fractional ideal $K$. We set $S=R[K]$ and ${\mathfrak{c}}= R:S$. Let $T$ birational module-finite extension of (hence $R \subseteq \subseteq and that \subseteq T$ $R \ne T$. We set $I = R: | A(1) \cong R \times_{R/I} R, \ (a,i) \mapsto (a, a+i),$$ The fiber prOduct Of tHe tWo CopiEs of The natural homoMOrphIsm $R \to R/I$. Hence, if $R$ is a redUced rInG, Then SO iS $A(1)$.
Let Us note tHE fOLLowInG.
\[lEmmA 3.1\] LET $(R,\M)$ be a $($nOt nEcessarIly NoetherIan$)$ LoCal ring. AssumE ThAt $I \ne R$ or $\alPha \In \m$. Then $A(\alphA)$ is A local RiNg wITh maxImaL ideaL $\m \timeS i$.
Let $(a,x) \In A(\alpha) \sEtMInus (\m \tIMes I)$. TheN, $A+\AlPha x \Not\in \m$, since $a \not\in \M$ BuT $\Alpha x \in \m$. ThereFore, seTtINg $B = A^{-1}$ And $Y = -(a+\aLpha x)^{-1}{\cdot}xB$, wE get $(a,X)(B,y)=1$ in $A(\alPHa)$. hENCe, $A(\ALpha)$ is a local rIng, because $\m \TImeS I$ is an IdEal OF $A(\alphA)$.
When $i=R$, $a(-1)$ Is nOt a local rinG, eveN if $(R,\m)$ is a lOcal riNG. In fact, ASsume thAt $A(-1)$ is a LocAl rIng. THEn, BeCauSe $\M \TimES R$ Is a MAxiMal ideal Of $a(-1)$ aNd $(1,1) \not\In \m \tIMES r$, we hAve $(1,1)(B,y)= (1,0)$ foR some $(B,y) \in A(-1)$, so that $b=1$ aNd $y+B + (-1){\cdoT}1{\CdoT}y=0$. ThiS is abSurd.
in What fOllows, Let $(R,\m)$ Be A one-dimensional coheN-Macaulay LocAl RinG wIth a cANonicaL frActIonal idEal $K$. We sET $S=R[k]$ aND ${\MAtHfrak{c}}= R:S$. Let $T$ be a birAtIONaL module-fInite eXTeNsIOn of $R$ (henCe $r \suBsetEQ t \subsEteq \OVeRline{R}$), anD assumE ThAt $k \subsetEq t$ but $R \nE T$. we sEt $I = r: | A(1) \cong R \times_{R/I}R, \ (a,i) \map sto (a ,a+i) ,$$the fiber prod u ct o f the two copies of th e nat ur a l ho m om orphi sm $R \ t oR / I$. H en ce, i f $ R$ is areduced ring, the n s ois $A(1)$.
L et us note t hefollowing.
\[l emma 3 .1 \]L et $( R,\ m)$ b e a $( $ not ne cessarily N o etheri a n$)$ lo c a lring . Assume that $I\ ne R$ or $\alpha\in \m $. Th e n $A (\a lpha)$ isalocal ring wi t hm a x ima l ideal $\m \t imes I$.
L e t $ (a,x)\i n A ( \alpha ) \se tm i nus (\m \times I)$ . Then, $ a+\alp h a x \no t \in \m$ , sinc e $ a \ not\ i n\m $ b ut $\a l ph a x \in \m$. Th er ef ore,sett i n g $b = a^ {-1} $ and $y = -(a+\al pha x)^ { -1} {\cdo t}xb$ , we g et $( a,x)(b ,y)=1 $in $A(\alpha)$. Hen ce, $A(\a lph a) $ i sa loc a l ring , b eca use $\m \times I$is a n i deal of $A(\alpha) $. Wh en $I=R$ , $A(- 1 )$ i s not a l oc alring , evenif $ ( R, \m)$ isa loca l r in g. In f ac t, ass um e t hat $A(- 1 )$ i s a lo cal ring . The n , because $\m\ times R$ is a ma x i ma l ide alof $A(-1)$and$ (1,1 ) \n o t\ in\ m \ti mes R $, we have $(1,1)(b,y)= ( 1, 0)$ fo r som e $(b,y) \inA(-1)$, so t h at $b=1$ and $y + b + (-1){\cdot }1{\c dot}y=0$.T his is a bsurd .
In wh at follow s , let $(R ,\m )$bea o n e -d imensional Co h e n-Ma ca ulay lo cal ring w ith acan oni ca l fractio nal idea l$K $. W e s et $S = R[K]$ an d${\ ma thf rak{c } }= R:S $. Le t $T $be a b iration a lm o dule -f in iteext en sionof $ R $ ( hence $ R \subset eqT \su bs et eq \ove rline{R}$), a nd assume th at $K \subs e t eq T$ bu t $R \ne T$. We set $I= R: | A(1) \cong_R \times_{R/I}_R, \ (a,i) \mapsto_(a, a+i),$$_the_fiber product_of_the two copies_of the natural_homomorphism $R \to R/I$._Hence, if $R$_is_a reduced ring, then so is $A(1)$.
Let us note the following.
\[lemma 3.1\] Let $(R,\m)$_be_a $($not_necessarily_Noetherian$)$_local ring. Assume that $I_\ne R$ or $\alpha \in_\m$. Then_$A(\alpha)$ is a local ring with maximal ideal_$\m_\times I$.
Let $(a,x)_\in A(\alpha) \setminus (\m \times I)$. Then, $a+\alpha x_\not\in \m$, since $a \not\in \m$_but $\alpha x_\in_\m$._Therefore, setting $b =_a^{-1}$ and $y = -(a+\alpha x)^{-1}{\cdot}xb$,_we get $(a,x)(b,y)=1$ in $A(\alpha)$. Hence,_$A(\alpha)$ is a local ring, because $\m_\times I$ is an ideal of_$A(\alpha)$.
When $I=R$, $A(-1)$ is not_a local_ring, even if $(R,\m)$ is_a local ring._In fact,_assume that $A(-1)$_is a local ring. Then, because_$\m \times R$_is a maximal ideal of $A(-1)$_and_$(1,1) \not\in \m_\times_R$,_we have_$(1,1)(b,y)= (1,0)$ for_some_$(b,y) \in_A(-1)$,_so that $b=1$ and $y+b +_(-1){\cdot}1{\cdot}y=0$._This is absurd.
In what follows, let $(R,\m)$_be a one-dimensional Cohen-Macaulay_local_ring with a canonical_fractional ideal $K$. We set_$S=R[K]$ and ${\mathfrak{c}}= R:S$. Let $T$_be a_birational module-finite_extension of $R$ (hence $R \subseteq T \subseteq \overline{R}$), and assume_that $K \subseteq T$ but $R_\ne T$. We set_$I =_R: |
prime }_n(x,y) d \nu(x,y) +
\int_{[0,1]^d} \varphi^{\prime }_n(u) d\mu(u)=\inf(\ref{duallike})
\end{split}$$ which ends the existence proof.
[99]{} A. Belloni, R.L. Winkler, On Multivariate Quantiles Under Partial Orders, The Annals of Statistics, **39** (2), 1125-1179 (2011).
Y. Brenier, Polar factorization and monotone rearrangement of vector-valued functions, Comm. Pure Appl. Math. 44 ** 4**, 375–417 (1991).
G. Carlier, A. Galichon, Exponential convergence for a convexifying equation, ESAIM, Control, Optimisation and Calculus of Variations, **18** (3), 611-620 (2012).
G. Carlier, V. Chernozhukov, A. Galichon, Vector quantile regression: an optimal transport approach, The Annals of Statistics, **44** (3), 1165-1192 (2016)
I. Ekeland, A. Galichon, M. Henry, Comonotonic measures of multivariate risks, Math. Finance, **22** (1), 109-132 (2012).
I. Ekeland, R. Temam, *Convex Analysis and Variational Problems*, Classics in Mathematics, Society for Industrial and Applied Mathematics, Philadelphia, (1999).
A. Galichon, *Optimal Transport Methods in Economics*, Princeton University Press.
A. Galichon, M. Henry, Dual theory of choice with multivariate risks, J. Econ Theory, **47** (4), 1501-516 (2012).
M. Hallin, D. Paindaveine, M. Siman, Multivariate quantiles and multiple-output regression quantiles: From $L^1$ optimization to halfspace depth, The Annals of Statistics, **38** (2), 635-669 (2010).
R. Koenker, G. Bassett, Regression Quantiles, Econometrica, **46**, 33-50 (1978).
J. Komlos, A generalization of a problem of Steinhaus, *Acta Mathematica Academiae Scientiarum Hungaricae*, **18 | prime } _ n(x, y) d \nu(x, y) +
\int_{[0,1]^d } \varphi^{\prime } _ n(u) d\mu(u)=\inf(\ref{duallike })
\end{split}$$ which ends the existence proof.
[ 99 ] { } A. Belloni, R.L. Winkler, On Multivariate Quantiles Under Partial Orders, The Annals of Statistics, * * 39 * * (2), 1125 - 1179 (2011).
Y. Brenier, pivotal factorization and monotone rearrangement of vector - respect functions, Comm. Pure Appl. Math. 44 * * 4 * *, 375–417 (1991).
G. Carlier, A. Galichon, Exponential convergence for a convexifying equation, ESAIM, Control, Optimisation and Calculus of Variations, * * 18 * * (3), 611 - 620 (2012).
G. Carlier, V. Chernozhukov, A. Galichon, Vector quantile arrested development: an optimal transport approach, The Annals of Statistics, * * 44 * * (3), 1165 - 1192 (2016)
I. Ekeland, A. Galichon, M. Henry, Comonotonic standard of multivariate risks, Math. Finance, * * 22 * * (1), 109 - 132 (2012).
I. Ekeland, R. Temam, * Convex Analysis and Variational Problems *, Classics in Mathematics, Society for Industrial and Applied Mathematics, Philadelphia, (1999).
A. Galichon, * Optimal Transport Methods in Economics *, Princeton University Press.
A. Galichon, M. Henry, double theory of choice with multivariate risk, J. Econ Theory, * * 47 * * (4), 1501 - 516 (2012).
M. Hallin, D. Paindaveine, M. Siman, Multivariate quantiles and multiple - output regression quantiles: From $ L^1 $ optimization to halfspace astuteness, The Annals of Statistics, * * 38 * * (2), 635 - 669 (2010).
R. Koenker, G. Bassett, Regression Quantiles, Econometrica, * * 46 * *, 33 - 50 (1978).
J. Komlos, A generalization of a problem of Steinhaus, * Acta Mathematica Academiae Scientiarum Hungaricae *, * * 18 | prile }_n(x,y) d \nu(x,y) +
\int_{[0,1]^d} \varphl^{\prime }_n(u) d\mu(u)=\iny(\eef{duanlike})
\ehd{split}$$ dhich ends the existence prolf.
[99]{} A. Beoloni, R.L. Winkler, On Muutivariatv Quantilws Uider Partial Ordxds, The Annals lf Scavistics, **39** (2), 1125-1179 (2011).
Y. Brgnier, Polar xactorization dna lonotone rearrangement of vector-valted funvtlons, Comm. Pure Appk. Matg. 44 ** 4**, 375–417 (1991).
G. Carlier, A. Galichon, Exponentjal connergence for a comvexifying equation, ESAIM, Fontgol, Optimisation ajd Calculus of Dqriations, **18** (3), 611-620 (2012).
G. Carlier, V. Chernozhuiov, A. Galichon, Vector quantile fegrexsion: an okcumap transport epproabh, The Annals of Stathstics, **44** (3), 1165-1192 (2016)
I. Ekeland, A. Ganicyon, M. Henry, Comonotonmc measures of multidariate rhsis, Math. Finance, **22** (1), 109-132 (2012).
U. Ekenand, R. Tdnam, *Cohvxx Znalysls end Variatiknal Problens*, Classics in Mathtmaepvs, Society fkr Indtserial and Applied Mathematics, Philadelpvia, (1999).
A. Galichon, *Optimal Trabsport Methods in Ecojomics*, Prynceton University Press.
A. Galichon, M. Henry, Dual tveory uf emjkxe with multivariate risks, J. Econ Theory, **47** (4), 1501-516 (2012).
M. Hwmlon, D. Paindaveine, M. Siman, Muktlvstiate quantiler and mulfiple-output regresdion quwntilws: From $L^1$ optomization to halfspace depty, The Annals if Statistics, **38** (2), 635-669 (2010).
R. Koenker, G. Yassetj, Regrrssion Quantiles, Economztrica, **46**, 33-50 (1978).
J. Komlos, A generamkzation of a proclek mf Steinhaus, *Acta Mathematyca Acadeniae Scientkaruk Hungwricae*, **18 | prime }_n(x,y) d \nu(x,y) + \int_{[0,1]^d} \varphi^{\prime \end{split}$$ ends the proof. [99]{} A. Quantiles Partial Orders, The of Statistics, **39** 1125-1179 (2011). Y. Brenier, Polar factorization monotone rearrangement of vector-valued functions, Comm. Pure Appl. Math. 44 ** 4**, 375–417 G. Carlier, A. Galichon, Exponential convergence for a convexifying equation, ESAIM, Control, Optimisation Calculus Variations, (3), (2012). G. Carlier, V. Chernozhukov, A. Galichon, Vector quantile regression: an optimal transport approach, The Annals Statistics, **44** (3), 1165-1192 (2016) I. Ekeland, A. M. Henry, Comonotonic measures multivariate risks, Math. Finance, **22** 109-132 I. Ekeland, Temam, Analysis Variational Problems*, Classics Mathematics, Society for Industrial and Applied Mathematics, Philadelphia, (1999). A. Galichon, *Optimal Transport Methods in Economics*, Princeton Press. A. Henry, Dual of with risks, J. Econ (4), 1501-516 (2012). M. Hallin, D. Multivariate quantiles and multiple-output regression quantiles: From $L^1$ to halfspace The Annals of Statistics, **38** (2), (2010). R. Koenker, G. Bassett, Regression Quantiles, Econometrica, 33-50 (1978). J. Komlos, A generalization of a problem of Steinhaus, *Acta Mathematica Academiae Scientiarum | prime }_n(x,y) d \nu(x,y) +
\int_{[0,1]^d} \varphi^{\prIme }_n(u) d\mu(u)=\iNf(\ref{DuaLliKe})
\End{sPlit}$$ Which ends the exIStenCe proof.
[99]{} A. Belloni, R.L. WinklEr, On MUlTIvarIAtE QuanTiles UnDEr pARtiAl orDerS, THE ANnals Of STatistiCs, **39** (2), 1125-1179 (2011).
Y. Brenier, polAr FactorizatioN AnD monotone rEarRangement of vEctOr-valuEd FunCTions, comM. Pure appl. MaTH. 44 ** 4**, 375–417 (1991).
G. CarlIer, A. GalicHoN, exponeNTial conVERgEnce For a convexifying eQUaTIon, ESAIM, ControL, OptimIsATiON And calCulus of VarIaTions, **18** (3), 611-620 (2012).
g. carlier, v. chERNOzhUKov, A. Galichon, VEctor quantiLE reGressiOn: An oPTimal tRanspOrT AppRoach, The AnnAls oF StatistiCs, **44** (3), 1165-1192 (2016)
I. EkeLAnd, A. GalIChon, M. HeNry, ComOnoTonIc meASuReS of MuLTivARiAte RIskS, Math. FinAnCe, **22** (1), 109-132 (2012).
i. EkelAnd, R. tEMAM, *ConVex analYsis aNd Variational proBlemS*, claSsics In MatHemaTiCs, SocIety foR InduStRial and Applied MAtheMatics, PhiLadElPhiA, (1999).
A. galicHOn, *OptiMal traNsport MEthods iN ecoNoMICS*, PRinceton University prESS.
A. galichon, m. Henry, dUaL tHEory of chOiCe wIth mULTivarIate RIsKs, J. Econ THeory, **47** (4), 1501-516 (2012).
M. hAlLiN, D. PaindAvEine, M. SImAn, MUltIvariATe quAntileS and multIple-oUTput regression QUantiles: From $L^1$ OPtIMIzATion To hAlfspace depTh, ThE annaLs of sTaTisTIcs, **38** (2), 635-669 (2010).
R. KOenkeR, G. bAsSEtt, Regression QuantiLeS, EconoMetriCa, **46**, 33-50 (1978).
J. Komlos, A genEralizatioN OF A problem Of StEInHAus, *Acta MathemaTica ACademiae ScIEntiarum hungaRicae*, **18 | prime }_n(x,y) d \nu(x,y)+
\int_{[0 ,1]^d } \ var ph i^{\ prim e }_n(u) d\mu( u )=\i nf(\ref{duallike})
\en d{spl it } $$ w h ic h end s the e x is t e nce p ro of.
[ 99 ]{} A . B elloni, R.L. Wink ler ,On Multivari a te Quantiles Un der PartialOrd ers, T he An n als o f S tatis tics,* *39**(2), 1125 -1 1 79 (20 1 1).
Y. B re nier , Polar factoriza t io n and monotonerearra ng e me n t of ve ctor-value dfunct i ons, Co m m. P u reA ppl. Math. 44 ** 4**, 37 5 –41 7 (199 1) .
G . Carl ier,A. Gal ichon, Expo nent ial conve rgence for a c o nvexify ing eq uat ion , ES A IM ,Con tr o l,O pt imi s ati on and C al cu lus o f Va r i a t ions , * *18* * (3) , 611-620 (20 12) .
G . Ca rlier , V.Cher no zhuko v, A.Galic ho n, Vector quant ileregressio n:an op ti mal t r anspor t a ppr oach, T he Anna l s o fS t a ti stics, **44** (3), 1 1 6 5- 1192 (20 16)
I . E ke l and, A.Ga lic hon, M . Hen ry,C om onotonic measu r es o f multi va riateri sks , M ath.F inan ce, ** 22** (1) , 109 - 132 (2012).
I . Ekeland, R.T em a m ,* Conv exAnalysis an d Va r iati onal Pr obl e ms*,Class ic s i n Mathematics, Socie ty for I ndust rial and Appl ied Mathem a t i cs, Phil adel p hi a , (1999).
A.Galic hon, *Opti m al Trans portMethodsin Econom i c s*, Prin cet onUni ver s i ty Press.
A. G a l icho n, M. Hen ry, Dual t heo ryofcho ic e with mu ltivaria te r is ks , J . Eco n Theory, * *47 ** (4 ), 15 0 1-516(2012 ).
M. H a lli n, D. P a in d a vein e, M . Si man ,Multi vari a tequantil es and mu lti p le-o ut pu t regre ssion quantil es : From $L^ 1$ op timiza t i on to ha lfspace depth, The Anna l s of St ati stics , ** 38** (2), 63 5-669(20 1 0).
R . Koen ker,G. Ba s s ett,R e gr ess io n Quantile s , Ec onome tr ica, **46** , 33-50 (1978).
J . Ko mlos, A gener ali zati o n o f a pr o ble mo f S t e inhaus, *Acta M athematica A c ad emiae Scie n tia ru m Hunga ricae*, **18 | prime }_n(x,y)_d \nu(x,y)_+
\int_{[0,1]^d} \varphi^{\prime }_n(u) d\mu(u)=\inf(\ref{duallike})
\end{split}$$_which ends_the_existence proof.
[99]{}_A._Belloni, R.L. Winkler,_On Multivariate Quantiles_Under Partial Orders, The_Annals of Statistics,_**39**_(2), 1125-1179 (2011).
Y. Brenier, Polar factorization and monotone rearrangement of vector-valued functions, Comm. Pure_Appl._Math. 44_** 4**,_375–417_(1991).
G. Carlier, A. Galichon, Exponential_convergence for a convexifying equation,_ESAIM, Control,_Optimisation and Calculus of Variations, **18** (3), 611-620_(2012).
G._Carlier, V. Chernozhukov,_A. Galichon, Vector quantile regression: an optimal transport approach,_The Annals of Statistics, **44** (3),_1165-1192 (2016)
I. Ekeland,_A._Galichon,_M. Henry, Comonotonic measures_of multivariate risks, Math. Finance, **22**_(1), 109-132 (2012).
I. Ekeland, R. Temam,_*Convex Analysis and Variational Problems*, Classics in_Mathematics, Society for Industrial and Applied_Mathematics, Philadelphia, (1999).
A. Galichon, *Optimal_Transport Methods_in Economics*, Princeton University Press.
A._Galichon, M. Henry,_Dual theory_of choice with_multivariate risks, J. Econ Theory, **47**_(4), 1501-516 (2012).
M._Hallin, D. Paindaveine, M. Siman, Multivariate_quantiles_and multiple-output regression_quantiles:_From_$L^1$ optimization_to halfspace depth,_The_Annals of_Statistics,_**38** (2), 635-669 (2010).
R. Koenker, G._Bassett,_Regression Quantiles, Econometrica, **46**, 33-50 (1978).
J. Komlos,_A generalization of a_problem_of Steinhaus, *Acta Mathematica_Academiae Scientiarum Hungaricae*, **18 |
${\mathbb{R}}^n$ into disjoint Borel sets $\{X_i\}_1^\infty$ such that $0<\nu(X_i)<\infty$ for all $i\geq 1$. Suppose we can find such a partition. Then the measure $$\rho:=\sum_{i=1}^\infty \frac{1}{2^i\nu(X_i)}\nu{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}X_i$$ is a Borel probability measure such that $\nu\ll\rho\ll \nu$. Theorem \[t:big\] identifies a Borel set $A_\rho$ such that $\rho^1_{{rect}}= \rho{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}A_\rho$ and $\rho^1_{{pu}}=\rho{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}({\mathbb{R}}^n\setminus A_\rho)$. Since $\nu\ll\rho$, it follows that $$\nu^1_{{rect}}=\nu{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}A_\rho\quad\text{and}\quad\nu^1_{{pu}}= \nu{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}({\mathbb{R}}^n\setminus A_\rho),$$ as well (which we leave as a simple exercise for the reader). In this sense, because we can solve the identification problem for any finite Borel measure $\rho$ on ${\mathbb{R}}^n$, we can also solve the problem for any $\sigma$-finite Borel measure $\nu$ on ${\mathbb{R}}^n$.
For every Radon measure $\mu$ on ${\mathbb{R}}^n$ and dyadic cube $Q\subset{\mathbb{R}}^n$, one can define a variant $\beta^{**}_2(\mu,Q)$ of $\beta^*(\mu,Q)$ without the truncation weight by $$\beta_2^{**}(\mu,Q) := \inf_\ell \sup_{R\in\Delta^*(Q)} \beta_2(\mu,3R,\ell).$$ Also define the associated density-normalized Jones function $J^{**}_2(\mu,\cdot):{\mathbb{R}}^ | $ { \mathbb{R}}^n$ into disjoint Borel sets $ \{X_i\}_1^\infty$ such that $ 0<\nu(X_i)<\infty$ for all $ i\geq 1$. Suppose we can find such a partition. Then the standard $ $ \rho:=\sum_{i=1}^\infty \frac{1}{2^i\nu(X_i)}\nu{\hbox { { \vrule height.22cm}{\leaders\hrule\hskip.2 cm } } } X_i$$ is a Borel probability standard such that $ \nu\ll\rho\ll \nu$. Theorem \[t: big\ ] identifies a Borel set $ A_\rho$ such that $ \rho^1_{{rect}}= \rho{\hbox { { \vrule height.22cm}{\leaders\hrule\hskip.2 cm } } } A_\rho$ and $ \rho^1_{{pu}}=\rho{\hbox { { \vrule height.22cm}{\leaders\hrule\hskip.2 cm } } } ({ \mathbb{R}}^n\setminus A_\rho)$. Since $ \nu\ll\rho$, it follows that $ $ \nu^1_{{rect}}=\nu{\hbox { { \vrule height.22cm}{\leaders\hrule\hskip.2 cm } } } A_\rho\quad\text{and}\quad\nu^1_{{pu}}= \nu{\hbox { { \vrule height.22cm}{\leaders\hrule\hskip.2 cm } } } ({ \mathbb{R}}^n\setminus A_\rho),$$ equally well (which we leave as a simple exercise for the lector). In this sense, because we can solve the designation problem for any finite Borel measure $ \rho$ on $ { \mathbb{R}}^n$, we can besides solve the problem for any $ \sigma$-finite Borel measure $ \nu$ on $ { \mathbb{R}}^n$.
For every Radon measure $ \mu$ on $ { \mathbb{R}}^n$ and dyadic cube $ Q\subset{\mathbb{R}}^n$, one can specify a variant $ \beta^{**}_2(\mu, Q)$ of $ \beta^*(\mu, Q)$ without the truncation weight by $ $ \beta_2^{**}(\mu, Q): = \inf_\ell \sup_{R\in\Delta^*(Q) } \beta_2(\mu,3R,\ell).$$ Also define the associated density - normalize Jones function $ J^{**}_2(\mu,\cdot):{\mathbb{R}}^ | ${\mahhbb{R}}^n$ into disjoint Bortl sets $\{X_i\}_1^\infty$ socy that $0<\nu(X_i)<\ihfty$ for all $i\geq 1$. Suppose we can fiid sych a partition. Then the mexsure $$\rho:=\dum_{i=1}^\inftt \frec{1}{2^i\nu(X_i)}\nu{\hbox{ {\vrnme heigmc.22cm}{\lezfers\krnle\hskip.2cm} }}X_i$$ ix a Borel [robability medsjrz such that $\nu\ll\rho\ll \nu$. Theorem \[t:bid\] identofles a Borel sej $A_\rhp$ sucg that $\rho^1_{{rect}}= \rho{\hbox{ {\vrule heighf.22cm}{\leadtrs\hrule\hskip.2cm} }}A_\rno$ and $\rho^1_{{pu}}=\rho{\hbox{ {\vrule jeigjt.22cm}{\leaders\hrule\hsnip.2cm} }}({\mathbv{R}}^n\sqrminus A_\rho)$. Rince $\nu\ll\gko$, it folloss that $$\nu^1_{{rect}}=\nu{\hbox{ {\vrule heigft.22cm}{\lzaders\hrule\ysjip.2fk} }}A_\rho\quad\txxt{and}\zuad\nu^1_{{pu}}= \nu{\hnpx{ {\vrune heignt.22cm}{\leaders\hruke\hvkip.2cm} }}({\mathbb{R}}^n\setminus E_\rho),$$ as well (which wg leave as a simple exercise doe the reager). Kb tfis sxnss, becakse we can somve the idebtification problem fjg any finite Gorel iewsure $\rho$ on ${\mathbb{R}}^n$, we can also solve ths problem for any $\sigma$-dinite Borel measure $\ju$ on ${\matrbb{R}}^n$.
For every Radon measure $\mu$ on ${\mathbb{R}}^n$ and dfadic zubt $W\subsdr{\mwthbb{R}}^n$, one can define a variant $\beta^{**}_2(\mu,Q)$ of $\bqfa^*(\ku,A)$ without the bruncation weight ny $$\brja_2^{**}(\mu,Q) := \inf_\ell \rup_{R\in\Bslfa^*(Q)} \beta_2(\mu,3R,\ell).$$ Alsl defing the qssociateq demsity-normalized Jones functuon $J^{**}_2(\mu,\cdot):{\mcthvb{R}}^ | ${\mathbb{R}}^n$ into disjoint Borel sets $\{X_i\}_1^\infty$ such for $i\geq 1$. we can find measure \frac{1}{2^i\nu(X_i)}\nu{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} is a Borel measure such that $\nu\ll\rho\ll \nu$. Theorem identifies a Borel set $A_\rho$ such that $\rho^1_{{rect}}= \rho{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}A_\rho$ and {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}({\mathbb{R}}^n\setminus A_\rho)$. Since $\nu\ll\rho$, it follows that $$\nu^1_{{rect}}=\nu{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}A_\rho\quad\text{and}\quad\nu^1_{{pu}}= {\vrule }}({\mathbb{R}}^n\setminus as (which we leave as a simple exercise for the reader). In this sense, because we can the identification problem for any finite Borel measure on ${\mathbb{R}}^n$, we can solve the problem for any Borel $\nu$ on For Radon $\mu$ on ${\mathbb{R}}^n$ dyadic cube $Q\subset{\mathbb{R}}^n$, one can define a variant $\beta^{**}_2(\mu,Q)$ of $\beta^*(\mu,Q)$ without the truncation weight by $$\beta_2^{**}(\mu,Q) \inf_\ell \sup_{R\in\Delta^*(Q)} define the density-normalized function | ${\mathbb{R}}^n$ into disjoint Borel Sets $\{X_i\}_1^\inftY$ such ThaT $0<\nu(x_i)<\InftY$ for All $i\geq 1$. Suppose WE can Find such a partition. Then The meAsURe $$\rhO:=\SuM_{i=1}^\infTy \frac{1}{2^i\NU(X_I)}\NU{\hbOx{ {\VrUle HeIGhT.22cm}{\leAdeRs\hrule\Hskip.2cm} }}X_i$$ iS a BOrEl probabilitY MeAsure such tHat $\Nu\ll\rho\ll \nu$. THeoRem \[t:biG\] iDenTIfies A BoRel seT $A_\rho$ sUCh that $\Rho^1_{{rect}}= \rhO{\hBOx{ {\vrulE Height.22cM}{\LEaDers\Hrule\hskip.2cm} }}A_\rho$ aND $\rHO^1_{{pu}}=\rho{\hbox{ {\vrulE heighT.22cM}{\LeADErs\HruLe\hskip.2cm} }}({\mAtHbb{R}}^n\SEtminus a_\RhO)$. sINce $\NU\ll\rho$, it folloWs that $$\nu^1_{{recT}}=\Nu{\hBox{ {\vruLe HeiGHt.22cm}{\leAders\HrULe\hSkip.2cm} }}A_\rho\qUad\tExt{and}\quaD\nu^1_{{pu}}= \nU{\Hbox{ {\vruLE height.22Cm}{\leadErs\HruLe\hsKIp.2Cm} }}({\MatHbB{r}}^n\sETmInuS a_\rhO),$$ as well (wHiCh We leaVe as A SIMPle eXerCise For thE reader). In this SenSe, beCAusE we caN solvE the IdEntifIcatioN probLeM for any finite BoRel mEasure $\rho$ On ${\mAtHbb{r}}^n$, We can ALso solVe tHe pRoblem fOr any $\siGMa$-fInITE boRel measure $\nu$ on ${\mathBb{r}}^N$.
foR every RaDon meaSUrE $\mU$ On ${\mathbb{r}}^n$ And DyadIC Cube $Q\SubsET{\mAthbb{R}}^n$, oNe can dEFiNe A varianT $\bEta^{**}_2(\mu,Q)$ Of $\BetA^*(\mu,q)$ withOUt thE truncAtion weiGht by $$\BEta_2^{**}(\mu,Q) := \inf_\ell \suP_{r\in\Delta^*(Q)} \beta_2(\MU,3R,\ELL).$$ ALSo deFinE the associaTed dENsitY-norMAlIzeD jones FunctIoN $j^{**}_2(\mU,\Cdot):{\mathbb{R}}^ | ${\mathbb{R}}^n$ into dis joint Bore l set s $ \{X _i \}_1 ^\in fty$ such that $0<\ nu(X_i)<\infty$ for al l $i\ ge q 1$. Su ppose we can fi n d su ch a pa rt i ti on. T hen the me asure $$\r ho: =\ sum_{i=1}^\i n ft y \frac{1} {2^ i\nu(X_i)}\n u{\ hbox{{\ vru l e hei ght .22cm }{\lea d ers\hr ule\hskip .2 c m} }}X _ i$$ isa Bo relprobability measu r es uch that $\nu\ ll\rho \l l \ n u $.The orem \[t:b ig \] id e ntifies aB o r els et $A_\rho$ s uch that $\ r ho^ 1_{{re ct }}= \rho{\ hbox{ { \ vru le height.2 2cm} {\leaders \hrule \ hskip.2 c m} }}A_ \rho$and $\ rho^ 1 _{ {p u}} =\ r ho{ \ hb ox{ {\v rule hei gh t. 22cm} {\le a d e r s\hr ule \hsk ip.2c m} }}({\mathb b{R }}^n \ set minus A_\r ho)$ .Since $\nu\ ll\rh o$ , it follows th at $ $\nu^1_{{ rec t} }=\ nu {\hbo x { {\vr ule he ight.22 cm}{\le a der s\ h r u le \hskip.2cm} }}A_\r ho \ q ua d\text{a nd}\qu a d\ nu ^ 1_{{pu}} =\nu {\hb o x { {\v rule he ight.22c m}{\le a de rs \hrule\ hs kip.2c m} }} ({\ mathb b {R}} ^n\set minus A_ \rho) , $$ as well (wh i ch we leave a s a s im p le e xer cise for th e re a der) . In th iss ense, beca us e w e can solve the iden ti ficati on pr oblem for any finite Bo r e l measure $\r h o$ on ${\mathbb{R }}^n$ , we can a l so solve theproblemfor any $ \ s igma$-fi nit e B ore l m e a su re $\nu$ on $ { \ math bb {R}}^n$ .
For eve ryRad onmea su re $\mu$on ${\ma th bb {R }} ^n$ andd yadic cu be $Q \s ubs et{\m a thbb{R }}^n$ , on eca n de fine av ar i a nt $ \b et a^{* *}_ 2( \mu,Q )$ o f $\ beta^*( \mu,Q)$ w ith o ut t he t runcati on weight by$$ \beta_2^{* *} (\m u,Q) : = \inf_\el l \sup_{R\in\Delta^*(Q) } \beta_ 2(\ mu,3R ,\el l).$$ Als o d efinethe associ ated d ensit y- nor m a lized J on esfu nction $J^ { * *}_ 2(\mu ,\ cdot ):{\mat hbb{R}}^ | ${\mathbb{R}}^n$_into disjoint_Borel sets $\{X_i\}_1^\infty$ such_that $0<\nu(X_i)<\infty$_for_all $i\geq_1$._Suppose we can_find such a_partition. Then the measure_$$\rho:=\sum_{i=1}^\infty \frac{1}{2^i\nu(X_i)}\nu{\hbox{ {\vrule_height.22cm}{\leaders\hrule\hskip.2cm}_}}X_i$$ is a Borel probability measure such that $\nu\ll\rho\ll \nu$. Theorem \[t:big\] identifies a_Borel_set $A_\rho$_such_that_$\rho^1_{{rect}}= \rho{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}A_\rho$_and $\rho^1_{{pu}}=\rho{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}({\mathbb{R}}^n\setminus_A_\rho)$. Since_$\nu\ll\rho$, it follows that $$\nu^1_{{rect}}=\nu{\hbox{ {\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}A_\rho\quad\text{and}\quad\nu^1_{{pu}}=_\nu{\hbox{_{\vrule height.22cm}{\leaders\hrule\hskip.2cm} }}({\mathbb{R}}^n\setminus_A_\rho),$$ as well (which we leave as a simple_exercise for the reader). In this_sense, because we_can_solve_the identification problem for_any finite Borel measure $\rho$ on_${\mathbb{R}}^n$, we can also solve the_problem for any $\sigma$-finite Borel measure $\nu$_on ${\mathbb{R}}^n$.
For every Radon measure $\mu$_on ${\mathbb{R}}^n$ and dyadic cube_$Q\subset{\mathbb{R}}^n$, one_can define a variant $\beta^{**}_2(\mu,Q)$_of $\beta^*(\mu,Q)$ without_the truncation_weight by $$\beta_2^{**}(\mu,Q)_:= \inf_\ell \sup_{R\in\Delta^*(Q)} \beta_2(\mu,3R,\ell).$$ Also define_the associated density-normalized_Jones function $J^{**}_2(\mu,\cdot):{\mathbb{R}}^ |
such that ML hypothesis $\bar{i}_n = X$ for every $n \geq {{{T}}}$. That is, $$\begin{aligned}
T = \min\{n': \bar{i}_n = X ~ \forall n \geq n'\}.\end{aligned}$$ Notice that ${{{T}}}$ is a random variable. Under Assumption \[steadinf\], it was shown in [@chernoff1959sequential] (Lemma 1) that there exist constants $b,K > 0$ such that for every $i\in \mathcal{X}$ and any strategy $g \in \mathcal{G}$, we have ${{\mathbb P}}_i^g[{{{T}}} > n] \leq Ke^{-bn}$. Let $$\begin{aligned}
N' \doteq \left\lceil -\frac{1}{b}\log\frac{\epsilon_N}{2K} \right\rceil.\end{aligned}$$ This ensures that ${{\mathbb P}}_i^g[{{{T}}} > N'] \leq \epsilon_N/2$. Fix a hypothesis $i$. Define the following event for each $n \geq N'$ $$\begin{aligned}
\mathscr{Z}_n = \{\bar{i}_k = X, N' \leq k \leq n \}.\end{aligned}$$ Clearly, the events $\mathscr{Z}_n$ are decreasing with $n$. Also, we have $$\begin{aligned}
\{ T \leq N'\} \subseteq \mathscr{Z}_n,\end{aligned}$$ for every $n \geq N'$.
Due to the threshold structure of the inference strategy $\bar{f}^N$, proving that $\psi_N(i) \geq 1 - \epsilon_N$ is equivalent to showing that $$\begin{aligned}
{{\mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{N+1}) - \mathcal{C}_i({{\rho}}_1) < \theta_N(i)] \leq \epsilon_N.\end{aligned}$$ To do so, we will use a Chernoff-bound based approach similar to the approach in Appendix \[detthmproof\]. We have $$\begin{aligned}
&{{\mathbb P}}_i^g[\mathcal{C}_{i}({{\rho}}_{N+1})- \mathcal{C}_{i}({{\rho}}_1) < \theta_N(i)]\\
&= {{\mathbb P}}_ | such that ML hypothesis $ \bar{i}_n = X$ for every $ n \geq { { { T}}}$. That is, $ $ \begin{aligned }
T = \min\{n': \bar{i}_n = X ~ \forall n \geq n'\}.\end{aligned}$$ Notice that $ { { { T}}}$ is a random variable star. Under Assumption \[steadinf\ ], it was usher in [ @chernoff1959sequential ] (Lemma 1) that there exist constants $ b, K > 0 $ such that for every $ i\in \mathcal{X}$ and any scheme $ g \in \mathcal{G}$, we have $ { { \mathbb P}}_i^g[{{{T } } } > n ] \leq Ke^{-bn}$. get $ $ \begin{aligned }
N' \doteq \left\lceil -\frac{1}{b}\log\frac{\epsilon_N}{2 K } \right\rceil.\end{aligned}$$ This guarantee that $ { { \mathbb P}}_i^g[{{{T } } } > N' ] \leq \epsilon_N/2$. Fix a hypothesis $ i$. specify the following event for each $ normality \geq N'$ $ $ \begin{aligned }
\mathscr{Z}_n = \{\bar{i}_k = X, N' \leq k \leq n \}.\end{aligned}$$ Clearly, the events $ \mathscr{Z}_n$ are decrease with $ n$. besides, we have $ $ \begin{aligned }
\ { T \leq N'\ } \subseteq \mathscr{Z}_n,\end{aligned}$$ for every $ n \geq N'$.
Due to the brink structure of the inference strategy $ \bar{f}^N$, prove that $ \psi_N(i) \geq 1 - \epsilon_N$ is equivalent to showing that $ $ \begin{aligned }
{ { \mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{N+1 }) - \mathcal{C}_i({{\rho}}_1) < \theta_N(i) ] \leq \epsilon_N.\end{aligned}$$ To do so, we will practice a Chernoff - bound based access similar to the access in Appendix \[detthmproof\ ]. We have $ $ \begin{aligned }
& { { \mathbb P}}_i^g[\mathcal{C}_{i}({{\rho}}_{N+1})- \mathcal{C}_{i}({{\rho}}_1) < \theta_N(i)]\\
& = { { \mathbb P } } _ | sufh that ML hypothesis $\bav{i}_n = X$ for every $n \geq {{{V}}}$. That js, $$\begin{xligned}
T = \min\{n': \bar{i}_n = X ~ \forell b \geq n'\}.\end{aligned}$$ Notice thxt ${{{T}}}$ is a random carieble. Under Assum'fion \[stccdinf\], lt wav shown in [@cherkoff1959sequenthal] (Lemma 1) thad ghzre exist constants $b,K > 0$ such that fjr everu $l\in \mathcal{X}$ agd amr stdategy $g \in \mathcal{G}$, we have ${{\mathbg P}}_i^g[{{{T}}} > n] \leq Ke^{-bn}$. Lrt $$\begin{aligned}
N' \doteq \lefh\lcell -\frac{1}{b}\log\frac{\epsllon_N}{2K} \righj\dceyo.\end{aligned}$$ Ghis ensurts that ${{\mathgb P}}_i^g[{{{T}}} > N'] \leq \epsilon_N/2$. Fix a hhpothzsis $i$. Defibe thf following xvent sor each $n \gca N'$ $$\becin{aligmed}
\mathscr{Z}_n = \{\nar{i}_k = X, N' \leq k \leq n \}.\end{alijned}$$ Clearly, the evegts $\mathswr{V}_n$ are decreasing wity $b$. Alsm, we havd $$\beein{zlmgnsd}
\{ T \lfq I'\} \subseteq \jathscr{Z}_n,\ene{aligned}$$ for every $m \dva N'$.
Due to ths threfhjld structure of the inference strategy $\bad{f}^N$, proving that $\psi_N(i) \teq 1 - \epsilon_N$ is equlvalent tj showing that $$\begin{aligned}
{{\mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{T+1}) - \mavhzal{E}_l({{\vho}}_1) < \rhfta_N(i)] \leq \epsilon_N.\end{aligned}$$ To do so, we will hst a Chernoff-bound based apprpafh fimilar to thg approcdh in Appendix \[detthlproof\]. Re hace $$\begin{ajignrd}
&{{\mathbb P}}_i^g[\mathcal{C}_{i}({{\rho}}_{N+1})- \mqthcal{C}_{i}({{\rho}}_1) < \rheta_N(i)]\\
&= {{\mathbb P}}_ | such that ML hypothesis $\bar{i}_n = X$ $n {{{T}}}$. That $$\begin{aligned} T = \forall \geq n'\}.\end{aligned}$$ Notice ${{{T}}}$ is a variable. Under Assumption \[steadinf\], it was in [@chernoff1959sequential] (Lemma 1) that there exist constants $b,K > 0$ such that every $i\in \mathcal{X}$ and any strategy $g \in \mathcal{G}$, we have ${{\mathbb P}}_i^g[{{{T}}} n] Ke^{-bn}$. $$\begin{aligned} \doteq \left\lceil -\frac{1}{b}\log\frac{\epsilon_N}{2K} \right\rceil.\end{aligned}$$ This ensures that ${{\mathbb P}}_i^g[{{{T}}} > N'] \leq \epsilon_N/2$. Fix a hypothesis Define the following event for each $n \geq $$\begin{aligned} \mathscr{Z}_n = \{\bar{i}_k X, N' \leq k \leq \}.\end{aligned}$$ the events are with Also, we have \{ T \leq N'\} \subseteq \mathscr{Z}_n,\end{aligned}$$ for every $n \geq N'$. Due to the threshold structure of inference strategy that $\psi_N(i) 1 \epsilon_N$ equivalent to showing {{\mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{N+1}) - \mathcal{C}_i({{\rho}}_1) < \theta_N(i)] do so, we will use a Chernoff-bound based similar to approach in Appendix \[detthmproof\]. We have &{{\mathbb P}}_i^g[\mathcal{C}_{i}({{\rho}}_{N+1})- \mathcal{C}_{i}({{\rho}}_1) < \theta_N(i)]\\ &= {{\mathbb P}}_ | such that ML hypothesis $\bar{i}_n = x$ for every $n \Geq {{{T}}}$. THat Is, $$\bEgIn{alIgneD}
T = \min\{n': \bar{i}_n = X ~ \foRAll n \Geq n'\}.\end{aligned}$$ Notice thAt ${{{T}}}$ is A rANdom VArIable. under AsSUmPTIon \[StEaDinF\], iT WaS showN in [@ChernofF1959sequentiaL] (LeMmA 1) that there exISt Constants $b,k > 0$ suCh that for eveRy $i\In \mathCaL{X}$ aND any sTraTegy $g \In \mathCAl{G}$, we hAve ${{\mathbb p}}_i^G[{{{t}}} > n] \leq KE^{-Bn}$. Let $$\beGIN{aLignEd}
N' \doteq \left\lceil -\FRaC{1}{B}\log\frac{\epsiloN_N}{2K} \rigHt\RCeIL.\End{AliGned}$$ This enSuRes thAT ${{\mathbb p}}_I^g[{{{t}}} > n'] \LEq \ePSilon_N/2$. Fix a hypOthesis $i$. DefINe tHe follOwIng EVent foR each $N \gEQ N'$ $$\bEgin{aligned}
\MathScr{Z}_n = \{\bar{i}_K = X, N' \leq K \Leq n \}.\end{ALigned}$$ CLearly, The EveNts $\mAThScR{Z}_n$ ArE DecREaSinG WitH $n$. Also, we HaVe $$\Begin{AligNED}
\{ t \Leq N'\} \SubSeteQ \mathScr{Z}_n,\end{alignEd}$$ fOr evERy $n \Geq N'$.
DUe to tHe thReShold StructUre of ThE inference stratEgy $\bAr{f}^N$, proviNg tHaT $\psI_N(I) \geq 1 - \ePSilon_N$ Is eQuiValent tO showinG ThaT $$\bEGIN{aLigned}
{{\mathbb P}}_i^g[\matHcAL{c}_i&({{\Rho}}_{N+1}) - \mathCal{C}_i({{\rHO}}_1) < \tHeTA_N(i)] \leq \epSiLon_n.\end{ALIgned}$$ to do SO, wE will use A ChernOFf-BoUnd baseD aPproacH sImiLar To the APproAch in APpendix \[dEtthmPRoof\]. We have $$\begiN{Aligned}
&{{\mathbb p}}_I^g[\MAThCAl{C}_{i}({{\Rho}}_{n+1})- \mathcal{C}_{i}({{\rHo}}_1) < \thETa_N(i)]\\
&= {{\MathBB P}}_ | such that ML hypothesis $ \bar{i}_n= X$for ev er y $n \ge q {{{T}}}$. Th a t is , $$\begin{aligned}
T= \mi n\ { n':\ ba r{i}_ n = X ~ \f o r all n \ geq n ' \} .\end {al igned}$ $ Notice t hat $ {{{T}}}$ isa r andom vari abl e. Under Ass ump tion \ [s tea d inf\] , i t was shown in [@c hernoff19 59 s equent i al] (Le m m a1) t hat there exist c o ns t ants $b,K > 0$ suchth a tf o r e ver y $i\in \m at hcal{ X }$ anda ny s t rat e gy $g \in \ma thcal{G}$,w e h ave ${ {\ mat h bb P}} _i^g[ {{ { T}} } > n] \leq Ke^ {-bn}$. L et $$\ b egin{al i gned}
N ' \dot eq\le ft\l c ei l-\f ra c {1} { b} \lo g \fr ac{\epsi lo n_ N}{2K } \r i g h t \rce il. \end {alig ned}$$ This e nsu rest hat ${{\ mathb b P} }_ i^g[{ {{T}}} > N' ]\leq \epsilon_N /2$. Fix a hy pot he sis $ i$. D e fine t hefol lowingevent f o r e ac h $ n\geq N'$ $$\begin{ al i g ne d}
\math scr{Z} _ n=\ {\bar{i} _k =X, N ' \leqk \l e qn \}.\en d{alig n ed }$ $ Clear ly , theev ent s $ \math s cr{Z }_n$ a re decre asing with $n$. Also , we have $$\b e gi n { al i gned }
\ { T \leq N' \} \ s ubse teq\ ma ths c r{Z}_ n,\en d{ a li g ned}$$ for every $n \ geq N' $.
D ue to the thr eshold str u c t ure of t he i n fe r ence strategy$\bar {f}^N$, pr o ving tha t $\p si_N(i)\geq 1 -\ e psilon_N $ i s e qui val e n tto showing th a t $$\ be gin{ali gne d}
{{\m ath bbP}} _i^ g[ \mathcal{ C}_i&({{ \r ho }} _{ N+1 }) -\ mathcal{ C} _i( {{ \rh o}}_1 ) < \th eta_N (i)] \ le q \e psilon_ N .\ e n d{al ig ne d}$$ To d o so, wew ill use aChernoff- bou n d ba se dapproac h similar toth e approach i n A ppendi x \[detthm proof\]. We have $$\beg i n{align ed}
&{{\ math bb P}}_i^ g[\ mathca l{C } _{i}({ {\rho} }_{N+ 1} )-\ m athca l { C} _{i }( {{\rho}}_1 ) < \ theta _N (i)] \\
&= { {\mathbb P}}_ | such_that ML_hypothesis $\bar{i}_n = X$_for every_$n_\geq {{{T}}}$._That_is, $$\begin{aligned}
T =_\min\{n': \bar{i}_n =_X ~ \forall n_\geq n'\}.\end{aligned}$$ Notice_that_${{{T}}}$ is a random variable. Under Assumption \[steadinf\], it was shown in [@chernoff1959sequential] (Lemma_1)_that there_exist_constants_$b,K > 0$ such that_for every $i\in \mathcal{X}$ and_any strategy_$g \in \mathcal{G}$, we have ${{\mathbb P}}_i^g[{{{T}}} >_n]_\leq Ke^{-bn}$. Let_$$\begin{aligned}
N' \doteq \left\lceil -\frac{1}{b}\log\frac{\epsilon_N}{2K} \right\rceil.\end{aligned}$$ This ensures that ${{\mathbb_P}}_i^g[{{{T}}} > N'] \leq \epsilon_N/2$. Fix_a hypothesis $i$._Define_the_following event for each_$n \geq N'$ $$\begin{aligned}
\mathscr{Z}_n = \{\bar{i}_k_= X, N' \leq k \leq_n \}.\end{aligned}$$ Clearly, the events $\mathscr{Z}_n$ are_decreasing with $n$. Also, we have_$$\begin{aligned}
\{ T \leq N'\} \subseteq_\mathscr{Z}_n,\end{aligned}$$ for_every $n \geq N'$.
Due to_the threshold structure_of the_inference strategy $\bar{f}^N$,_proving that $\psi_N(i) \geq 1 -_\epsilon_N$ is equivalent_to showing that $$\begin{aligned}
{{\mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{N+1}) -_\mathcal{C}_i({{\rho}}_1)_< \theta_N(i)] \leq_\epsilon_N.\end{aligned}$$_To_do so,_we will use_a_Chernoff-bound based_approach_similar to the approach in Appendix_\[detthmproof\]._We have $$\begin{aligned}
&{{\mathbb P}}_i^g[\mathcal{C}_{i}({{\rho}}_{N+1})- \mathcal{C}_{i}({{\rho}}_1) < \theta_N(i)]\\
&=_{{\mathbb P}}_ |
{\left(\kappa \underline{R}/{ \begingroup
\def\mathaccent#R##2{ \rel@kern{0.8} \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.2}} \rel@kern{-0.2} } \macc@depth\@ne
\let\math@bgroup\@empty \let\math@egroup\macc@set@skewchar
\mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax} \macc@set@skewchar\relax
\let\mathaccentV\macc@nested@a
\macc@nested@a\relax111{R} \endgroup
} \right)^{1/\alpha}}, {1}/{2} \right\} \right)^{i} (M -m).\end{aligned}$$ The procedure requires thus the computation of at most $N$ tests where $N$ is the smallest integer such that $$\left( \max \left\{1 - {\left(\kappa \underline{R}/{ \begingroup
\def\mathaccent#R##2{ \rel@kern{0.8} \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.2}} \rel@kern{-0.2} } \macc@depth\@ne
\let\math@bgroup\@empty \let\math@egroup\macc@set@skewchar
\mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax} \macc@set@skewchar\relax
\let\mathaccentV\macc@nested@a
\macc@nested@a\relax111{R} \endgroup
} \right)^{1/\alpha}}, {1}/{2} \right\} \right)^{N} (M-m) \leq \eta$$ that is $$N\geq \frac{\log \left( (M- m )/\eta \right)}{ -\log \left[ \max \left\{1 - {\ | { \left(\kappa \underline{R}/ { \begingroup
\def\mathaccent#R##2 { \rel@kern{0.8 } \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.2 } } \rel@kern{-0.2 } } \macc@depth\@ne
\let\math@bgroup\@empty \let\math@egroup\macc@set@skewchar
\mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax } \macc@set@skewchar\relax
\let\mathaccentV\macc@nested@a
\macc@nested@a\relax111{R } \endgroup
} \right)^{1/\alpha } }, { 1}/{2 } \right\ } \right)^{i } (M -m).\end{aligned}$$ The procedure requires thus the computation of at about $ N$ examination where $ N$ is the smallest integer such that $ $ \left (\max \left\{1 - { \left(\kappa \underline{R}/ { \begingroup
\def\mathaccent#R##2 { \rel@kern{0.8 } \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.2 } } \rel@kern{-0.2 } } \macc@depth\@ne
\let\math@bgroup\@empty \let\math@egroup\macc@set@skewchar
\mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax } \macc@set@skewchar\relax
\let\mathaccentV\macc@nested@a
\macc@nested@a\relax111{R } \endgroup
} \right)^{1/\alpha } }, { 1}/{2 } \right\ } \right)^{N } (M - m) \leq \eta$$ that is $ $ N\geq \frac{\log \left (( M- m) /\eta \right) } { -\log \left [ \max \left\{1 - { \ | {\levt(\kappa \underline{R}/{ \beglngroup
\def\mathcxcent#R##2{ \rem@kern{0.8} \overline{\rel@kern{-0.8}\macc@nucleus\rxl@keen{0.2}} \rel@kern{-0.2} } \macc@depth\@nd
\let\matj@bgroup\@enpty \oet\math@egroup\macc@scc@skewdmar
\kethsurround\z@ \frpzen@everymdth{\mathgroup\mawc@erlup\relax} \macc@set@skewchar\relax
\let\iathaccrnhV\macc@nested@a
\macb@nqstes@a\relax111{R} \endgroup
} \right)^{1/\alpha}}, {1}/{2} \rifht\} \rigit)^{i} (M -m).\end{alignrd}$$ The procedure requires hhus the computation ov at most $N$ tesew where $N$ is the smalltsc integer soch that $$\left( \max \left\{1 - {\left(\kappx \unberline{R}/{ \bgyunggmup
\def\matiaccenn#R##2{ \rel@kern{0.8} \overnine{\rel@lern{-0.8}\macc@nucleux\ren@keen{0.2}} \rel@kern{-0.2} } \macc@vepth\@ne
\let\math@bgroop\@empty \led\mcth@egroup\macc@set@skewcyae
\majhsursouna\z@ \ffoztn@etergmath{\mwthjroup\macc@grkup\relax} \mqcc@set@skewchar\relax
\jvy\mathaccentV\jacc@neftqd@a
\macc@nested@a\relax111{R} \endgroup
} \right)^{1/\anphz}}, {1}/{2} \right\} \right)^{N} (M-m) \lwq \eta$$ that is $$N\geq \ftac{\log \lefe( (M- m )/\eta \right)}{ -\log \left[ \max \left\{1 - {\ | {\left(\kappa \underline{R}/{ \begingroup \def\mathaccent#R##2{ \rel@kern{0.8} \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.2}} \rel@kern{-0.2} \let\math@bgroup\@empty \mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax} \let\mathaccentV\macc@nested@a \macc@nested@a\relax111{R} \endgroup (M The procedure requires the computation of most $N$ tests where $N$ is smallest integer such that $$\left( \max \left\{1 - {\left(\kappa \underline{R}/{ \begingroup \def\mathaccent#R##2{ \rel@kern{0.8} \rel@kern{-0.2} } \macc@depth\@ne \let\math@bgroup\@empty \let\math@egroup\macc@set@skewchar \mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax} \macc@set@skewchar\relax \let\mathaccentV\macc@nested@a \macc@nested@a\relax111{R} \endgroup } \right)^{1/\alpha}}, \right\} (M-m) \eta$$ is $$N\geq \frac{\log \left( (M- m )/\eta \right)}{ -\log \left[ \max \left\{1 - {\ | {\left(\kappa \underline{R}/{ \begingRoup
\def\matHacceNt#R##2{ \Rel@KeRn{0.8} \ovErliNe{\rel@kern{-0.8}\macc@nUCleuS\rel@kern{0.2}} \rel@kern{-0.2} } \macc@depTh\@ne
\lEt\MAth@bGRoUp\@empTy \let\maTH@eGROup\MaCc@Set@SkEWcHar
\maThsUrround\Z@ \frozen@eveRymAtH{\mathgroup\maCC@gRoup\relax} \mAcc@Set@skewchar\rElaX
\let\maThAccENtV\maCc@nEsted@A
\macc@nESted@a\rElax111{R} \endgRoUP
} \right)^{1/\ALpha}}, {1}/{2} \rigHT\} \RiGht)^{i} (m -m).\end{aligned}$$ The prOCeDUre requires thuS the coMpUTaTIOn oF at Most $N$ tests WhEre $N$ iS The smalLEsT INTegER such that $$\left( \Max \left\{1 - {\left(\KAppA \underLiNe{R}/{ \BEgingrOup
\deF\mAThaCcent#R##2{ \rel@keRn{0.8} \ovErline{\rel@Kern{-0.8}\maCC@nucleuS\Rel@kern{0.2}} \Rel@kerN{-0.2} } \maCc@dEpth\@NE
\lEt\MatH@bGRouP\@EmPty \LEt\mAth@egrouP\mAcC@set@sKewcHAR
\MAthsUrrOund\Z@ \frozEn@everymath{\maThgRoup\MAcc@Group\Relax} \Macc@SeT@skewChar\reLax
\leT\mAthaccentV\macc@nEsteD@a
\macc@nesTed@A\rElaX111{R} \EndgrOUp
} \righT)^{1/\alPha}}, {1}/{2} \Right\} \riGht)^{N} (M-m) \lEQ \etA$$ tHAT Is $$n\geq \frac{\log \left( (M- m )/\eTa \RIGhT)}{ -\log \left[ \Max \lefT\{1 - {\ | {\left(\kappa \underline {R}/{ \be gingr oup
\d ef\m atha ccent#R##2{ \rel @kern{0.8} \overlin e{\re l@ k ern{ - 0. 8}\ma cc@nucl e us \ r el@ ke rn {0. 2} } \re l@k ern{-0. 2} } \ma cc@ de pth\@ne
\l e t\ math@bgrou p\@ empty \let\m ath @egrou p\ mac c @set@ ske wchar
\ma t hsurro und\z@ \f ro z en@eve r ymath{\ m a th grou p\macc@group\rela x }\macc@set@skew char\r el a x\le t\m athaccentV \m acc@n e sted@a \ m a c c@n e sted@a\relax1 11{R} \end g rou p
} \r ig ht) ^ {1/\al pha}} ,{ 1}/ {2} \right \} \ right)^{i } (M - m ).\end{ a ligned} $$ The pr oce dure re qu ire st hus th e c o mpu tation o fat most $N$ t e s ts w her e $N $ isthe smallestint eger suc h tha t $$\ left (\max\left\ {1 -{\ left(\kappa \u nder line{R}/{ \ be gin gr oup
\def\m ath acc ent#R## 2{ \ r el@ ke r n { 0. 8} \overline{\r el @ k er n{-0.8}\ macc@n u cl eu s \rel@ker n{ 0.2 }} \rel@ kern { -0 .2} } \macc@ d ep th \@ne
\l et\mat h@ bgr oup \@emp t y \l et\mat h@egroup \macc @ set@skewchar
\mathsurround \ z@ \ fr o zen@ eve rymath{\mat hgro u p\ma cc@g r ou p\r e lax} \mac c@ s et @ skewchar\relax
\l et \matha ccent V\macc@nested @a
\macc @ n e sted@a\r elax 1 11 { R} \endgroup} \ri ght)^{1/\a l pha}}, { 1}/{2 } \righ t\} \righ t ) ^{N} (M- m) \ leq \e t a $$ that is $$N\ g e q \f ra c{\log\le ft( (M- m)/\ eta \ ri ght)}{ - \log \l ef t[ \ ma x \ left\ { 1 - {\ | {\left(\kappa_ \underline{R}/{_ \begingroup
\def\mathaccent#R##2{_ __\rel@kern{0.8} __ \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.2}} _ \rel@kern{-0.2}_ } \macc@depth\@ne
_ \let\math@bgroup\@empty \let\math@egroup\macc@set@skewchar
__\mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax} \macc@set@skewchar\relax
\let\mathaccentV\macc@nested@a
\macc@nested@a\relax111{R} \endgroup
} \right)^{1/\alpha}}, {1}/{2} \right\}_\right)^{i}_(M -m).\end{aligned}$$_The_procedure_requires thus the computation of_at most $N$ tests where_$N$ is_the smallest integer such that $$\left( \max \left\{1_-_{\left(\kappa \underline{R}/{_ \begingroup
\def\mathaccent#R##2{ \rel@kern{0.8} _ \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.2}} _\rel@kern{-0.2} }__\macc@depth\@ne
_ \let\math@bgroup\@empty \let\math@egroup\macc@set@skewchar
_\mathsurround\z@ \frozen@everymath{\mathgroup\macc@group\relax} \macc@set@skewchar\relax
\let\mathaccentV\macc@nested@a
_ \macc@nested@a\relax111{R} \endgroup
} \right)^{1/\alpha}}, {1}/{2}_ \right\} \right)^{N} (M-m) \leq_\eta$$ that is $$N\geq \frac{\log \left(_(M- m )/\eta \right)}{_ -\log_ \left[ \max \left\{1 -_{\ |
is $\sim$100mJy, suggesting the presence of a hot dust component, which as discussed in the previous section is a clear sign of a hot dusty torus of an AGN [@Laurent2000]. Similarly, one can draw the same conclusion by observing the combination of the f$_{15\mu
m}$/f$_{6.7\mu m}$ and f$_{6.7\mu m}$/f$_{6\mu m}$ flux ratios. In IRAS19254-7245S, the low f$_{6.7\mu m}$/f$_{6\mu m}$ indicates weak UIB emission while f$_{15\mu m}$/f$_{6.7\mu m}$$\sim$2.7, a value somewhat lower than other well studied starburst galaxies such as Arp220 [f$_{15\mu m}$/f$_{6.7\mu
m}$$\sim$3.9, @Charmandaris1999b] or the extremely strong starburst region in the Cartwheel [f$_{15\mu m}$/f$_{6.7\mu
m}$$\sim$5.2, @Charmandaris1999a]. This effect can be understood since the hot continuum produced by an AGN at short MIR wavelengths would cause the flux in the 6–10$\mu$m range to increase more relative to the increase observed between 12–16$\mu$m and as result it would be added the UIB emission sampled by the LW2 filter.
Could the large difference in the MIR brightness between the north and south component in IRAS19254-7245 be related to the additional contribution of the AGN? Studies of the dynamical evolution of this system suggest that the starburst time scale is much shorter than the dynamical age of the merger [@Mihos1998]. Even though we can not quantify accurately the fraction of MIR luminosity due to the AGN activity, it appears that the southern component of IRAS19254-7245 has reached an AGN dominant phase, however short this may be, after an initial phase of strong starburst activity (see @Laurent2000 and @Genzel1998 for details on the MIR AGN/starburst fraction of this and other galaxies). The MIR properties of the northern nucleus are similar to a normal spiral galaxy which indicates that even if a starburst did occur in it at some point, it has by now subsided and the star formation is progressing in a more quiescent rate.
| is $ \sim$100mJy, suggesting the presence of a hot dust part, which as discourse in the previous section is a clean signboard of a hot dusty torus of an AGN [ @Laurent2000 ]. Similarly, one can guide the like conclusion by observing the combination of the f$_{15\mu
m}$/f$_{6.7\mu m}$ and f$_{6.7\mu m}$/f$_{6\mu m}$ flux ratios. In IRAS19254 - 7245S, the low f$_{6.7\mu m}$/f$_{6\mu m}$ argue weak UIB emission while f$_{15\mu m}$/f$_{6.7\mu m}$$\sim$2.7, a value reasonably humble than other well studied starburst galaxies such as Arp220 [ f$_{15\mu m}$/f$_{6.7\mu
m}$$\sim$3.9, @Charmandaris1999b ] or the extremely strong starburst area in the Cartwheel [ f$_{15\mu m}$/f$_{6.7\mu
m}$$\sim$5.2, @Charmandaris1999a ]. This effect can be understood since the hot continuum produce by an AGN at short MIR wavelengths would induce the flux in the 6–10$\mu$m image to increase more relative to the addition observed between 12–16$\mu$m and as result it would be added the UIB emission sample by the LW2 filter.
Could the large difference in the MIR brightness between the north and south component in IRAS19254 - 7245 be related to the additional contribution of the AGN? Studies of the dynamical evolution of this system suggest that the starburst time scale is much shorter than the dynamical historic period of the merger [ @Mihos1998 ]. Even though we cannot quantify accurately the fraction of MIR luminosity due to the AGN activity, it appear that the southern component of IRAS19254 - 7245 has reached an AGN dominant phase, however light this may be, after an initial phase of strong starburst activity (see @Laurent2000 and @Genzel1998 for details on the MIR AGN / starburst fraction of this and early galaxies). The MIR properties of the northern nucleus are similar to a normal spiral galaxy which indicates that even if a starburst did occur in it at some point, it has by now subsided and the asterisk formation is progress in a more quiescent rate. | is $\sim$100mJy, suggesting the pvesence of a hot dust cmmponeht, which as discussed in the previoud wectiin is a clear sign of x hot dusny torus if ai AGN [@Laurent2000]. Similarly, one cah drac vhe same concluxion by obverving the cokbknction of the f$_{15\mu
m}$/f$_{6.7\mu m}$ and f$_{6.7\mu m}$/f$_{6\mu i}$ flux tahios. In IRAS19254-7245S, jhe lpr f$_{6.7\mh m}$/f$_{6\mu m}$ indicates weak UIB emissikn whilt f$_{15\mu m}$/f$_{6.7\mu m}$$\sim$2.7, a falue somewhat lower than ltheg well studied stagburst galazies wuch as Arp220 [w$_{15\mu m}$/f$_{6.7\mu
m}$$\spk$3.9, @Charmandzris1999b] or the extremely strong sgarbuxst region un thf Cartwheel [h$_{15\mu m}$/f$_{6.7\iu
m}$$\sim$5.2, @Charmandaris1999a]. This egfect can be ukdersvood since the hot continnum produced by an ADN at shost MIR wavelengths qoyld cduse the dlub ih vhe 6–10$\mu$m rwngx to increaae more relqtive to the increaxe ibserved betwsen 12–16$\mu$i wnd as result it would be added the UIB emjssion sampled by the LQ2 filter.
Could the larhe differqnce in the MIR brightness between the north and vouth zomkokcnt kb LRAS19254-7245 be related to the additional contributiog og nhe AGN? Studies on the dynamical evplktojn of this syrtem sbfgsst that the starbkrst tiie scqle is mush snorter than the dynamical ate of the megger [@Mihos1998]. Even though we can not quamtify accurately the fractiou of MJR luminosihy due to ghe AGN activity, it a[pears that the southern cjmponent if IXAS19254-7245 has feacned an AGN dominwnt pmdse, however short hhis laf be, after an initial phase of strong stacuurst activiti (sae @Naurent2000 cnd @Gekzel1998 for detailf on the MIR AYN/starbuxst frxction of nhis and mther galaxyes). The MIR pslperties of vhe northqrn buclwus are rimilar to a nprmal spigaj gaoaxy which indicatcs thxf even if a staxyutwt did occur im ig ae doke [oint, it has by vow xubsiaed and the rtar formation is progrevsinf in a more quiescrnb rate.
| is $\sim$100mJy, suggesting the presence of a component, as discussed the previous section a dusty torus of AGN [@Laurent2000]. Similarly, can draw the same conclusion by the combination of the f$_{15\mu m}$/f$_{6.7\mu m}$ and f$_{6.7\mu m}$/f$_{6\mu m}$ flux ratios. IRAS19254-7245S, the low f$_{6.7\mu m}$/f$_{6\mu m}$ indicates weak UIB emission while f$_{15\mu m}$/f$_{6.7\mu a somewhat than well studied starburst galaxies such as Arp220 [f$_{15\mu m}$/f$_{6.7\mu m}$$\sim$3.9, @Charmandaris1999b] or the extremely strong starburst in the Cartwheel [f$_{15\mu m}$/f$_{6.7\mu m}$$\sim$5.2, @Charmandaris1999a]. This can be understood since hot continuum produced by an at MIR wavelengths cause flux the 6–10$\mu$m range increase more relative to the increase observed between 12–16$\mu$m and as result it would be added the emission sampled LW2 filter. the difference the MIR brightness north and south component in IRAS19254-7245 the additional contribution of the AGN? Studies of dynamical evolution this system suggest that the starburst scale is much shorter than the dynamical age the merger [@Mihos1998]. Even though we can not quantify accurately the fraction of MIR luminosity the AGN activity, it that the southern of has an dominant phase, short this may be, after an initial phase of strong starburst (see @Laurent2000 and @Genzel1998 for details on the MIR AGN/starburst this other galaxies). The properties of the northern are to a normal spiral indicates even did in at some point, it by now subsided and the formation is progressing in | is $\sim$100mJy, suggesting the presEnce of a hot Dust cOmpOneNt, WhicH as dIscussed in the pREvioUs section is a clear sign oF a hot DuSTy toRUs Of an AgN [@LaureNT2000]. SIMIlaRlY, oNe cAn DRaW the sAme ConclusIon by obserVinG tHe combinatioN Of The f$_{15\mu
m}$/f$_{6.7\mu M}$ anD f$_{6.7\mu m}$/f$_{6\mu m}$ fluX raTios. In iRaS19254-7245S, THe low F$_{6.7\mu M}$/f$_{6\mu m}$ IndicaTEs weak uIB emissiOn WHile f$_{15\mU M}$/f$_{6.7\mu m}$$\siM$2.7, A VaLue sOmewhat lower than oTHeR Well studied staRburst GaLAxIES suCh aS Arp220 [f$_{15\mu m}$/f$_{6.7\mU
m}$$\Sim$3.9, @ChARmandarIS1999b] OR THe eXTremely strong Starburst reGIon In the CArTwhEEl [f$_{15\mu m}$/F$_{6.7\mu
m}$$\sIm$5.2, @cHarMandaris1999a]. ThIs efFect can be UndersTOod sincE The hot cOntinuUm pRodUced BY aN AgN aT sHOrt miR WavELenGths woulD cAuSe the Flux IN THE 6–10$\mu$m RanGe to IncreAse more relatiVe tO the INcrEase oBservEd beTwEen 12–16$\mu$M and as ResulT iT would be added thE UIB Emission sAmpLeD by ThE LW2 fiLTer.
CouLd tHe lArge difFerence IN thE Mir BRiGhtness between the nOrTH AnD south coMponenT In iRas19254-7245 be relatEd To tHe adDITionaL conTRiBution of The AGN? sTuDiEs of the DyNamicaL eVolUtiOn of tHIs syStem suGgest thaT the sTArburst time scaLE is much shorteR ThAN ThE DynaMicAl age of the mErgeR [@mihoS1998]. EveN ThOugH We can Not quAnTIfY Accurately the fractiOn Of MIR lUminoSity due to the AgN activity, IT APpears thAt thE SoUThern component Of IRAs19254-7245 has reacheD An AGN domInant Phase, howEver short THIs may be, aFteR an IniTiaL PHaSe of strong staRBUrst AcTivity (sEe @LAurent2000 aNd @GEnzEl1998 fOr dEtAils on the mIR AGN/stArBuRsT fRacTion oF This and oThEr gAlAxiEs). The miR propErtieS of tHe NoRTheRn nucleUS aRE SimiLaR tO a noRmaL sPiral GalaXY whIch indiCates that EveN If a sTaRbUrst did Occur in it at soMe Point, it has By Now SubsidED And the stAr formation is progressinG In a more QuiEscenT ratE.
| is $\sim$100mJy, suggesti ng the pre sence of aho t du st c omponent, whic h asdiscussed in the previ ous s ec t ioni sa cle ar sign of a ho tdu sty t o ru s ofanAGN [@L aurent2000 ].Si milarly, one ca n draw the sa me conclusio n b y obse rv ing the c omb inati on oft he f$_ {15\mu
m} $/ f $_{6.7 \ mu m}$a n df$_{ 6.7\mu m}$/f$_{6\ m um }$ flux ratios . In I RA S 19 2 5 4-7 245 S, the low f $_{6. 7 \mu m}$ / f$ _ { 6 \mu m}$ indicates weak UIB e m iss ion wh il e f $ _{15\m u m}$ /f $ _{6 .7\mu m}$$\ sim$ 2.7, a va lue so m ewhat l o wer tha n othe r w ell stu d ie dsta rb u rst ga lax i essuch asAr p2 20 [f $_{1 5 \ m u m}$ /f$ _{6. 7\mum}$$\sim$3.9, @C harm a nda ris19 99b]or t he extr emelystron gstarburst regio n in the Cart whe el [f $_ {15\m u m}$/f $_{ 6.7 \mu
m}$ $\sim$5 . 2,@C h a r ma ndaris1999a]. This e f f ec t can be under s to od since th ehot con t i nuumprod u ce d by anAGN at sh or t MIR w av elengt hs wo uld caus e the fluxin the 6 –10$\ m u$m range to i n crease more r e la t i ve to t heincrease ob serv e d be twee n 1 2–1 6 $\mu$ m and a s r e sult it would be ad de d theUIB e mission sampl ed by theL W 2 filter.
Co u ld the large diff erenc e in the M I R bright nessbetweenthe north a nd south co mpo nen t i n IR AS19254-7245b e rel at ed to t headditio nal co ntr ibu ti on of the AGN? St ud ie sof th e dyn a mical ev ol uti on of this system sugg estth at the starbu r st t imesc al e is mu ch shor tert han the dy namical a geo f th eme rger [@ Mihos1998]. E ve n though w ecan not q u a ntify ac curately the fraction o f MIR lu min osity due to the A GNactivi ty, it app ears t hat t he so u t hernc o mp one nt of IRAS19 2 5 4-7 245 h as rea ched an AGN dominant phas e , h owever shortthi s ma y be , a f te r an i n iti a l phase of stron g starburs ta ct ivity (see @La ur ent2000 and @G enzel 1 998 for detailson the MI RAGN/ s t arb urst fract ion of t his and o t her g a la xies) . T he MIR p rop ertie s of t h e n orthe rn nuc le us are simi la r to a n ormal spiral galaxy whi ch ind icate s t hat evenifa st arburst d id o ccur in it at so me po int , it h as b y n ows ubsid ed a n d the sta r f orm a t io n is progre s s i ngin amor e quies cent rate.
| is_$\sim$100mJy, suggesting_the presence of a_hot dust_component,_which as_discussed_in the previous_section is a_clear sign of a_hot dusty torus_of_an AGN [@Laurent2000]. Similarly, one can draw the same conclusion by observing the combination_of_the f$_{15\mu
m}$/f$_{6.7\mu_m}$_and_f$_{6.7\mu m}$/f$_{6\mu m}$ flux ratios._In IRAS19254-7245S, the low f$_{6.7\mu_m}$/f$_{6\mu m}$_indicates weak UIB emission while f$_{15\mu m}$/f$_{6.7\mu m}$$\sim$2.7,_a_value somewhat lower_than other well studied starburst galaxies such as Arp220_[f$_{15\mu m}$/f$_{6.7\mu
m}$$\sim$3.9, @Charmandaris1999b] or the extremely_strong starburst region_in_the_Cartwheel [f$_{15\mu m}$/f$_{6.7\mu
m}$$\sim$5.2, @Charmandaris1999a]._This effect can be understood since_the hot continuum produced by an_AGN at short MIR wavelengths would cause_the flux in the 6–10$\mu$m range_to increase more relative to_the increase_observed between 12–16$\mu$m and as_result it would_be added_the UIB emission_sampled by the LW2 filter.
Could the_large difference in_the MIR brightness between the north_and_south component in_IRAS19254-7245_be_related to_the additional contribution_of_the AGN?_Studies_of the dynamical evolution of this_system_suggest that the starburst time scale is_much shorter than the_dynamical_age of the merger_[@Mihos1998]. Even though we can_not quantify accurately the fraction of_MIR luminosity_due to_the AGN activity, it appears that the southern component of IRAS19254-7245_has reached an AGN dominant phase,_however short this may_be, after_an_initial phase of_strong_starburst activity_(see @Laurent2000 and @Genzel1998 for details on_the MIR_AGN/starburst fraction of this and other_galaxies). The MIR properties_of_the northern nucleus are similar to_a normal spiral galaxy which indicates_that even if a starburst_did_occur_in it at some point,_it has by now subsided and_the star formation_is progressing in a more quiescent rate.
|
frac{\alpha_A}{\alpha}$). Here, $u_A$ and $u_B$ are the utilities of users in class $A$ and class $B$, respectively, and $u$ represents the utility of the users if they all had loose delay requirements which means $\tilde{\gamma}^*_k=\gamma^*$ for all $k$. Fig. \[fig2\] shows the loss for the matched filter, the decorrelator, and the MMSE detector. We observe from the figure that for the matched filter both classes of users suffer significantly due to the presence of delay sensitive traffic. For example, when half of the users are delay sensitive, the utilities achieved by class $A$ and class $B$ users are, respectively, 50% and 60% of the utilities for the case of no delay constraints. For the decorrelator, only class $A$ users suffer and the reduction in utility is smaller than that of the matched filter. For the MMSE detector, the reduction in utility for class $A$ users is similar to that of the decorrelator, and the reduction in utility for class $B$ is negligible.
We repeat the experiment for a highly loaded network with $\alpha=0.9$ (see Fig. \[fig3\]). Since the matched filter cannot handle such a significant load, we have shown the plots for the decorrelator and MMSE detector only. We observe from Fig. \[fig3\] that because of the higher system load, the reduction in the utilities is more significant for the MMSE detector compared to the case of $\alpha=0.1$. It should be noted that for the decorrelator the reduction in utility of class $A$ users is independent of the system load. This is because the decorrelator completely removes the multiple-access interference.
It should be further noted that in Figs. \[fig2\] and \[fig3\] we have only plotted the ratio of the utilities (not the actual values). As discussed in Section \[multiclass\], the achieved utilities for the MMSE detector are larger than those of the decorrelator and the matched filter.
Conclusions
===========
We have proposed a game-theoretic approach for studying power control in multiple-access networks with (transmission) delay constraints. We have considered a non-cooperative game where each user seeks to choose a transmit power that maximizes its own utility while satisfying the user’s delay requirements. The utility function measures | frac{\alpha_A}{\alpha}$). Here, $ u_A$ and $ u_B$ are the utilities of users in class $ A$ and class $ B$, respectively, and $ u$ represent the utility program of the users if they all had lax delay necessity which means $ \tilde{\gamma}^*_k=\gamma^*$ for all $ k$. Fig. \[fig2\ ] shows the passing for the equal filter, the decorrelator, and the MMSE detector. We observe from the figure that for the matched filter both classes of drug user suffer significantly due to the bearing of delay sensitive traffic. For case, when half of the users are delay sensitive, the utility achieved by class $ A$ and class $ B$ drug user are, respectively, 50% and 60% of the utilities for the case of no delay constraints. For the decorrelator, merely class $ A$ users suffer and the reduction in utility is minor than that of the matched filter. For the MMSE detector, the reduction in utility for class $ A$ users is exchangeable to that of the decorrelator, and the reduction in utility for class $ B$ is negligible.
We repeat the experiment for a highly loaded network with $ \alpha=0.9 $ (see Fig. \[fig3\ ]). Since the matched filter cannot handle such a significant cargo, we have shown the plots for the decorrelator and MMSE detector entirely. We note from Fig. \[fig3\ ] that because of the higher system load, the decrease in the utilities is more meaning for the MMSE detector compared to the case of $ \alpha=0.1$. It should be noted that for the decorrelator the decrease in utility of class $ A$ users is independent of the system load. This is because the decorrelator wholly removes the multiple - access interference.
It should be further noted that in Figs. \[fig2\ ] and \[fig3\ ] we have only plotted the proportion of the utility (not the actual values). As discussed in Section \[multiclass\ ], the achieved utilities for the MMSE detector are larger than those of the decorrelator and the matched filter.
Conclusions
= = = = = = = = = = =
We have aim a game - theoretic approach for studying might control in multiple - entree networks with (transmission) delay constraints. We have considered a non - cooperative plot where each user seeks to choose a transmit power that maximizes its own utility program while satisfying the user ’s delay requirements. The utility program function measures | fraf{\alpha_A}{\alpha}$). Here, $u_A$ and $u_B$ are the utilities mf useds in clxss $A$ and class $B$, respectivepy, and $y$ represents the utiligy of the users id thty all had loose vslay requiremskts wkirh means $\tilde{\gsmma}^*_k=\gamma^*$ for all $k$. Fig. \[fkg2\] shows the loss for the matched fileer, the dfcorrelator, anq tht MISE svttctor. We observe from the figure fhat fog the matched filyer both classes of users dufffr significantly dke to the ptssegxe of delay rensitive urcffic. For esample, when half of the users afe dekay sensituvw, tjg utilities echievvd by class $A$ and clavs $B$ usrrs are, respecbivelb, 50% abd 60% of the utilities hor the case of no dglay constsannts. For the decorrelqtir, onny cnass $Q$ urera xurfer ajd vhe reductikn in utiliry is smaller than uhae of the matchsd fileew. For the MMSE detector, the reduction it ufility for class $A$ userw is similar to that lf the desorrelator, and the reduction in utility for class $B$ is vegoinibld.
Qe repeat the experiment for a highly loaded neesotk with $\alpha=0.9$ (sec Fig. \[fig3\]). Since thr layshed filter cxnnot kznsle such a signififant lowd, we have shorn tne plots for the decorrelatir and MMSE betwctor only. We obsexve from Fig. \[yig3\] thst bevause of the higher syscem lozd, the reduftion in ffe utilities is oorv sicnificanu for the MMSE detqctor com'ared to the casg of $\al[ha=0.1$. It shokld bc noted that for thf declrselator thf reduction in utility of class $A$ users is ince[engent of che syxtem load. Thif is because tke decorxelatof completemy remotes the multyple-access indgrference.
It siould be surtyer boted tfxt in Figs. \[fig2\] amd \[fig3\] we kcve only plotted the ratio pf jhs utilities (not tht qctual values). Ss aissudsxd in Vection \[multhclars\], gne acfieved utilltids fpr the MMSE detector are larger than those on the decirrelatow and the matvhed filter.
Concluslons
===========
Wx have proppseq a game-theoretic approach for studying porer control yn mmltikle-access nztworks with (transmission) delay constraiits. We have considered q non-cooperative gakt where each nser sqeks to cvoose a transmit powwr that maximizes its own utility whils sativfyinh the user’s delay requirements. The utility function measures | frac{\alpha_A}{\alpha}$). Here, $u_A$ and $u_B$ are the users class $A$ class $B$, respectively, of users if they had loose delay which means $\tilde{\gamma}^*_k=\gamma^*$ for all $k$. \[fig2\] shows the loss for the matched filter, the decorrelator, and the MMSE We observe from the figure that for the matched filter both classes of suffer due the of delay sensitive traffic. For example, when half of the users are delay sensitive, the utilities by class $A$ and class $B$ users are, 50% and 60% of utilities for the case of delay For the only $A$ suffer and the in utility is smaller than that of the matched filter. For the MMSE detector, the reduction in for class is similar that the and the reduction for class $B$ is negligible. We for a highly loaded network with $\alpha=0.9$ (see \[fig3\]). Since matched filter cannot handle such a load, we have shown the plots for the and MMSE detector only. We observe from Fig. \[fig3\] that because of the higher system reduction in the utilities more significant for MMSE compared the of $\alpha=0.1$. should be noted that for the decorrelator the reduction in utility class $A$ users is independent of the system load. This the completely removes the interference. It should be noted in Figs. \[fig2\] and have plotted the (not actual values). As discussed Section \[multiclass\], the achieved utilities the MMSE detector are decorrelator and the matched filter. Conclusions =========== We proposed a game-theoretic approach for studying power in multiple-access networks with (transmission) delay constraints. We have considered a non-cooperative where each to choose a transmit power that maximizes its utility while satisfying the delay requirements. The utility function measures | frac{\alpha_A}{\alpha}$). Here, $u_A$ and $u_b$ are the utiLitieS of UseRs In clAss $A$ And class $B$, respeCTiveLy, and $u$ represents the utiLity oF tHE useRS iF they All had lOOsE DElaY rEqUirEmENtS whicH meAns $\tildE{\gamma}^*_k=\gamMa^*$ fOr All $k$. Fig. \[fig2\] shOWs The loss for The Matched filteR, thE decorReLatOR, and tHe MmSE deTector. wE obserVe from the FiGUre thaT For the mATChEd fiLter both classes of USeRS suffer signifiCantly DuE To THE prEseNce of delay SeNsitiVE traffiC. foR EXAmpLE, when half of thE users are deLAy sEnsitiVe, The UTilitiEs achIeVEd bY class $A$ and cLass $b$ users are, RespecTIvely, 50% anD 60% Of the utIlitieS foR thE casE Of No DelAy COnsTRaIntS. for The decorReLaTor, onLy clASS $a$ UserS suFfer And thE reduction in uTilIty iS SmaLler tHan thAt of ThE matcHed filTer. FoR tHe MMSE detector, tHe reDuction in UtiLiTy fOr Class $a$ Users iS siMilAr to thaT of the dECorReLATOr, And the reduction in uTiLITy For class $b$ is negLIgIbLE.
We repeaT tHe eXperIMEnt foR a hiGHlY loaded nEtwork WItH $\aLpha=0.9$ (see fiG. \[fig3\]). SiNcE thE maTched FIlteR cannoT handle sUch a sIGnificant load, wE Have shown the pLOtS FOr THe deCorRelator and MmSE dETectOr onLY. WE obSErve fRom FiG. \[fIG3\] tHAt because of the higheR sYstem lOad, thE reduction in tHe utilitieS IS More signIficANt FOr the MMSE detecTor coMpared to thE Case of $\alPha=0.1$. It Should be Noted that FOR the decoRreLatOr tHe rEDUcTion in utility OF ClasS $A$ Users is IndEpendenT of The SysTem LoAd. This is bEcause thE dEcOrReLatOr comPLetely reMoVes ThE muLtiplE-Access InterFereNcE.
IT ShoUld be fuRThER NoteD tHaT in FIgs. \[FiG2\] and \[fIg3\] we HAve Only ploTted the raTio OF the UtIlIties (noT the actual valUeS). As discussEd In SEction \[MULticlass\], The achieved utilities for THe MMSE dEteCtor aRe laRger than tHosE of the DecORrelatOr and tHe matChEd fILTer.
CoNCLuSioNs
===========
we have propOSEd a Game-tHeOretIc approAch for studying poweR ConTrol in multiplE-acCess NETwOrkS WiTH (trAnSMisSIOn) delay constraiNts. We have cOnSIdEred a non-coOPerAtIve game Where eaCh useR Seeks to Choose a trAnsmit powEr That MAXimIzes its own Utility wHile satisFYing tHE uSer’s dElaY requiReMenTs. The UtilitY FunCtion MeasurEs | frac{\alpha_A}{\alpha}$).Here, $u_A $ and $u _B$ a re t he u tilities of us e rs i n class $A$ and class$B$,re s pect i ve ly, a nd $u$r ep r e sen ts t heut i li ty of th e users if they a llha d loose dela y r equirement s w hich means $ \ti lde{\g am ma} ^ *_k=\ gam ma^*$ for a l l $k$. Fig. \[f ig 2 \] sho w s the l o s sforthe matched filte r ,t he decorrelato r, and t h eM M SEdet ector. Weob serve from th e f i g u ret hat for the m atched filt e r b oth cl as ses of use rs su ff e r s ignificantl y du e to thepresen c e of de l ay sens itivetra ffi c. F o rex amp le , wh e nhal f of the use rs a re de lays e n s itiv e,theutili ties achieved by cla s s $ A$ an d cla ss $ B$ user s are, resp ec tively, 50% and 60% of the u til it ies f or th e caseofnodelay c onstrai n ts. F o r th e decorrelator, on ly c la ss $A$ u sers s u ff er and there duc tion i n uti lity is smaller thant ha tof thema tchedfi lte r.For t h e MM SE det ector, t he re d uction in util i ty for class$ A$ u se r s is si milar to th at o f the dec o rr ela t or, a nd th er ed u ction in utility fo rclass$B$ i s negligible.
We repea t t he exper imen t f o r a highly loa ded n etwork wit h $\alpha =0.9$ (see Fi g. \[fig3 \ ] ). Since th e m atc hed f il ter cannot ha n d le s uc h a sig nif icant l oad , w e h ave s hown theplots fo rth ede cor relat o r and MM SE de te cto r onl y . We o bserv e fr om F i g.\[fig3\ ] t h a t be ca us e of th ehighe r sy s tem load,the reduc tio n inth eutiliti es is more si gn ificant fo rthe MMSEd e tector c ompared to the case of$ \alpha= 0.1 $. It sho uld be no ted thatfor the de correl atorth e r e d uctio n in ut il ity of cla s s $A $ use rs isindepen dent of the system loa d. This is be cau se t h e d eco r re l ato rc omp l e tely removes th e multiple -a c ce ss interfe r enc e.
It sh ould be furt h er note d that in Figs. \[ fi g2\] a nd\[fig3\] w e have o nly plott e d the ra tio o f t he uti li tie s (no t thea ctu al va lues). A s disc ussed i n Sectio n \[multiclass\], the a chieve d uti lit ies for t heM MSE detector are larger th antho se of th e deco rrel a to r a n d the mat c hed filte r .
Co n c lu sions
===== = = = ===
Wehav e propo seda game-theoretica pproach for st udyi n g po wer cont ro l in multiple- acc es s networks w ith (transm ission)de l ay co nstrai nts. W e havec o ns i dereda no n-c ooperativ e g am e whereea ch user s eeks t o choo se a t r ansm i t power that maxi mizes i ts ow n ut ility w hile sa t isfy ing the us er’s delayrequir emen ts. T he util it y func tio nmeasures | frac{\alpha_A}{\alpha}$). Here,_$u_A$ and_$u_B$ are the utilities_of users_in_class $A$_and_class $B$, respectively,_and $u$ represents_the utility of the_users if they_all_had loose delay requirements which means $\tilde{\gamma}^*_k=\gamma^*$ for all $k$. Fig. \[fig2\] shows the_loss_for the_matched_filter,_the decorrelator, and the MMSE_detector. We observe from the_figure that_for the matched filter both classes of users_suffer_significantly due to_the presence of delay sensitive traffic. For example, when_half of the users are delay_sensitive, the utilities_achieved_by_class $A$ and class_$B$ users are, respectively, 50% and_60% of the utilities for the_case of no delay constraints. For the_decorrelator, only class $A$ users suffer_and the reduction in utility_is smaller_than that of the matched_filter. For the_MMSE detector,_the reduction in_utility for class $A$ users is_similar to that_of the decorrelator, and the reduction_in_utility for class_$B$_is_negligible.
We repeat_the experiment for_a_highly loaded_network_with $\alpha=0.9$ (see Fig. \[fig3\]). Since_the_matched filter cannot handle such a significant_load, we have shown_the_plots for the decorrelator_and MMSE detector only. We_observe from Fig. \[fig3\] that because of_the higher_system load,_the reduction in the utilities is more significant for the MMSE_detector compared to the case of_$\alpha=0.1$. It should be_noted that_for_the decorrelator the_reduction_in utility_of class $A$ users is independent of_the system_load. This is because the decorrelator_completely removes the multiple-access_interference.
It_should be further noted that in_Figs. \[fig2\] and \[fig3\] we have only plotted the_ratio of the utilities (not_the_actual_values). As discussed in Section_\[multiclass\], the achieved utilities for the_MMSE detector are_larger than those of the decorrelator and_the_matched filter.
Conclusions
===========
We have proposed a game-theoretic_approach_for studying power control in multiple-access_networks_with_(transmission) delay constraints. We have_considered a non-cooperative game where each_user seeks to choose a transmit power that maximizes_its own utility_while satisfying the user’s delay_requirements._The_utility function measures |
good model of citation influence could suggest a model of hypertext link importance. This could improve measures of the importance of web pages, such as PageRank [@Qi:2007:MSD:1244408.1244418].
Improved recommender systems: Researchers often need help identifying relevant work that they should read. Filtering out less relevant citations might help paper recommender systems [@MEET:MEET14504701330; @springerlink:10.1007/978-3-642-23535-1_35].
Related work {#related-work.unnumbered}
============
The idea that the mere counting of citations is dubious is not new [@chubin1975content]: The field of *citation context analysis* has a long history dating back to the early days of citation indexing. There is a wide variety of reasons for a researcher to cite a source and many ways of categorizing them. For instance, identified fifteen such reasons, including giving credit for related work, correcting a work, and criticizing previous work.
For articles in the field of high energy physics, distinguished four major classes of polar opposite pairs, conceptual–operational, organic–perfunctory, evolutionary–juxtapositional, and confirmative–negational. They found that the fraction of negational references, i.e., citations indicating that the cited source is wrong, is not negligible (14%).
presented one of the first automatic citation indexing systems (CiteSeer). It could parse citations and use them to compute similarities between documents.
might have implemented the first automated classification systems for citations. They used over 200 manually selected rules to classify citations in one of 35 categories.
Machine learning methods for automatic classification can be applied to the text of a citing document. distinguish categories of citations that can be identified via linguistic cues in the text. They are able to classify citations into one of four categories (weak, positive, contrast, neutral) with an average of 68%. For a classification in three categories (weakness, positive, neutral), they get an average of 71%. Their classifier relies on 892 manually selected cue phrases, such as whether the citation is a self-citation, the location of the citation in the text, and manually acquired verb clusters.
annotated a corpus of 43 open-access full-text biomedical articles. They built classifiers using Support Vector Machine (SVM) and Multinomial Na | good model of citation influence could suggest a exemplar of hypertext liaison importance. This could improve measures of the importance of web page, such as PageRank [ @Qi:2007: MSD:1244408.1244418 ].
Improved recommender systems: research worker often need help identify relevant work that they should read. trickle out less relevant citations might help newspaper recommender systems [ @MEET: MEET14504701330; @springerlink:10.1007/978 - 3 - 642 - 23535 - 1_35 ].
Related work { # related-work.unnumbered }
= = = = = = = = = = = =
The idea that the mere count of citations is dubious is not new [ @chubin1975content ]: The discipline of * citation context analysis * hold a farseeing history dating back to the early day of citation indexing. There is a wide variety show of reasons for a researcher to cite a source and many ways of categorizing them. For instance, identified fifteen such reasons, including giving credit for related work, correcting a work, and criticizing previous employment.
For articles in the field of gamey department of energy purgative, distinguished four major classes of polar opposite pairs, conceptual – operational, organic – perfunctory, evolutionary – juxtapositional, and collateral – negational. They found that the fraction of negational references, i.e., citations indicating that the cited source is wrong, is not negligible (14 %).
presented one of the first automatic citation indexing systems (CiteSeer). It could parse citations and practice them to compute similarity between document.
might have implemented the first automated classification system for citations. They used over 200 manually selected rule to classify citations in one of 35 categories.
Machine learning method acting for automatic classification can be apply to the textbook of a citing document. distinguish categories of citations that can be identify via linguistic cues in the textbook. They are able to classify citations into one of four categories (decrepit, positive, line, neutral) with an average of 68% . For a classification in three categories (helplessness, positive, neutral), they get an average of 71% . Their classifier relies on 892 manually selected clue phrases, such as whether the citation is a self - quotation, the location of the citation in the text, and manually acquired verb clusters.
annotated a corpus of 43 candid - access full - text biomedical articles. They built classifiers use Support Vector Machine (SVM) and Multinomial Na | gold model of citation infuuence could suytest a model of hypeftext link importance. This cluod imkgove measures of the kmportancv of web pagew, such as PefeRank [@Ql:2007:ISD:1244408.1244418].
Ijiroveb cecommender sysjems: Researcvers often neeg fepp identifying relevant work that trey shoilf read. Filterigg olt lesa relevant citations might help paler recmmmender systrms [@MEET:MEET14504701330; @springerlink:10.1007/978-3-642-23535-1_35].
Repatef work {#related-work.knnumbered}
============
Tye iqwa that the oere countpug of citatjons is dubious is not new [@chubiv1975contznt]: The fieod of *witation coitext wnalysis* has a long vistory dating back tp tie eqrly days of citation indexing. There is a wide vdrnety of reasons for a rwsearwher to zute a aonrcs and lanb ways of cztegorizing them. For instance, odqbtified fiftesn sucr weasons, including giving credit for reldtes work, correcting a worj, and criticizing pregious worh.
For articles in the field of high energy physics, distmneuiwhcd fuyr major classes of polar opposite pairs, concepehak–oierational, organig–perfunctory, evoluyilnsty–juxtapositioval, anb cknfirmative–negatiojal. Thei foune that tht fravtion of negational referenxes, i.e., citatpons indicating that tke cited souxce is wronb, is not negligible (14%).
prerentsd one of tje first zjtomatic citatiov ikdefing sysudms (CiteSeer). It cjuld parsx citctions avd uxe thei to compuhe similarities between focumgnts.
micht have ilplemented the first automated rkassificatiom vysnems for eitatipns. They used over 200 manually selectzd rulds to clasaify civations in oge of 35 categorhgs.
Machine leacning metrods for automagkc classificatoon can bv cpplied ti the text of a cibing akcument. distingbnsy categories of cigatyojs trdt can be idantiwiea via uinguisunc guer in the text. They are atle fo classify citatipnf into obe of fotr categories (weak, positive, conurast, ieutrau) witn ag average of 68%. For a classificafion in tjrec categories (reakkess, positive, ueutral), they get an average of 71%. Their clessifier relies on 892 manoally selected cue phtasts, such as whxther ehe citathon is a self-citatiob, the location of the citation in the fext, atd majually acquired verb clusters.
annotated a corpus of 43 open-access full-text biimedicel articles. Tgey nuilt cjcssifiews udmng Support Vectog Machine (SVM) and Multinomial Na | good model of citation influence could suggest of link importance. could improve measures pages, as PageRank [@Qi:2007:MSD:1244408.1244418]. recommender systems: Researchers need help identifying relevant work that should read. Filtering out less relevant citations might help paper recommender systems [@MEET:MEET14504701330; Related work {#related-work.unnumbered} ============ The idea that the mere counting of citations is is new The of *citation context analysis* has a long history dating back to the early days of citation There is a wide variety of reasons for researcher to cite a and many ways of categorizing For identified fifteen reasons, giving for related work, a work, and criticizing previous work. For articles in the field of high energy physics, distinguished four classes of pairs, conceptual–operational, evolutionary–juxtapositional, confirmative–negational. found that the negational references, i.e., citations indicating that is wrong, is not negligible (14%). presented one the first citation indexing systems (CiteSeer). It could citations and use them to compute similarities between might have implemented the first automated classification systems for citations. They used over 200 manually to classify citations in of 35 categories. learning for classification be applied the text of a citing document. distinguish categories of citations that be identified via linguistic cues in the text. They are classify into one of categories (weak, positive, contrast, with average of 68%. For in categories they an of 71%. Their classifier on 892 manually selected cue such as whether the location of the citation in the text, and acquired verb clusters. annotated a corpus of open-access full-text biomedical articles. They built classifiers using Support Vector Machine (SVM) Multinomial Na | good model of citation influeNce could suGgest A moDel Of HypeRtexT link importancE. this Could improve measures of The imPoRTancE Of Web paGes, such AS PAGERaNk [@qi:2007:mSD:1244408.1244418].
imPRoVed reComMender sYstems: ReseArcHeRs often need hELp IdentifyinG reLevant work thAt tHey shoUlD reAD. FiltEriNg out Less reLEvant cItations mIgHT help pAPer recoMMEnDer sYstems [@MEET:MEET14504701330; @sprINgERlink:10.1007/978-3-642-23535-1_35].
Related woRk {#relaTeD-WoRK.UnnUmbEred}
============
The ideA tHat thE Mere couNTiNG OF ciTAtions is dubioUs is not new [@cHUbiN1975conteNt]: the FIeld of *CitatIoN ConText analysiS* has A long histOry datINg back tO The earlY days oF ciTatIon iNDeXiNg. THeRE is A WiDe vARieTy of reasOnS fOr a reSearCHER To ciTe a SourCe and Many ways of catEgoRiziNG thEm. For InstaNce, iDeNtifiEd fiftEen suCh Reasons, includinG givIng credit For ReLatEd Work, cORrectiNg a WorK, and criTicizinG PreViOUS WoRk.
For articles in the FiELD oF high eneRgy phySIcS, dIStinguisHeD foUr maJOR clasSes oF PoLar opposIte paiRS, cOnCeptual–OpEratioNaL, orGanIc–perFUnctOry, evoLutionarY–juxtAPositional, and cONfirmative–negATiONAl. tHey fOunD that the fraCtioN Of neGatiONaL reFErencEs, i.e., cItATiONs indicating that the CiTed souRce is Wrong, is not negLigible (14%).
preSENTed one of The fIRsT Automatic citatIon inDexing systEMs (CiteSeEr). It cOuld parsE citationS ANd use theM to ComPutE siMILaRities between DOCumeNtS.
might hAve ImplemeNteD thE fiRst AuTomated clAssificaTiOn SyStEms For ciTAtions. ThEy UseD oVer 200 ManuaLLy seleCted rUles To ClASsiFy citatIOnS IN one Of 35 CaTegoRieS.
MAchinE leaRNinG methodS for automAtiC ClasSiFiCation cAn be applied to ThE text of a ciTiNg dOcumenT. DIstinguiSh categories of citations THat can bE idEntifIed vIa linguisTic Cues in The TExt. TheY are abLe to cLaSsiFY CitatIONs IntO oNe of four caTEGorIes (weAk, PosiTive, conTrast, neutral) with an AVerAge of 68%. For a clasSifIcatION iN thREe CAteGoRIes (WEAkness, positive, nEutral), they GeT An Average of 71%. THEir ClAssifieR relies On 892 manUAlly selEcted cue pHrases, sucH aS wheTHEr tHe citation Is a self-cItation, thE LocatIOn Of the CitAtion iN tHe tExt, anD manuaLLy aCquirEd verb ClUsters.
AnnotAtEd a corpuS of 43 open-access full-text biOmedicAl artIclEs. They buiLt cLAssIfiers usiNg SuPport VectoR MaChiNe (SVM) And mUltinOmiaL na | good model of citation in fluence co uld s ugg est a mod el o f hypertext li n k im portance. This could i mprov em easu r es of t he impo r ta n c e o fwe b p ag e s, such as PageRa nk [@Qi:20 07: MS D:1244408.12 4 44 18].
Impr ove d recommende r s ystems :Res e arche rsoften needh elp id entifying r e levant work th a t t heyshould read. Filt e ri n g out less rel evantci t at i o nsmig ht help pa pe r rec o mmender sy s t e ms[ @MEET:MEET145 04701330; @ s pri ngerli nk :10 . 1007/9 78-3- 64 2 -23 535-1_35].
Rel ated work {#rel a ted-wor k .unnumb ered}=== === ==== = =
T heid e a t h at th e me re count in gof ci tati o n s is d ubi ousis no t new [@chubi n19 75co n ten t]: T he fi eldof *cit ationconte xt analysis* hasa lo ng histor y d at ing b ack t o the e arl y d ays ofcitatio n in de x i n g. There is a wide v ar i e ty of reas ons fo r a r e searcher t o c itea sourc e an d m any ways of ca t eg or izing t he m. For i nst anc e, id e ntif ied fi fteen su ch re a sons, includin g giving credi t f o r r e late d w ork, correc ting a wo rk,a nd cr i ticiz ing p re v io u s work.
For articl es in th e fie ld of high en ergy physi c s , disting uish e df our major clas ses o f polar op p osite pa irs,conceptu al–operat i o nal, org ani c–p erf unc t o ry , evolutionar y – juxt ap osition al, and co nfi rma tiv e–n eg ational.They fou nd t ha tthe frac t ion of n eg ati on alrefer e nces,i.e., cit at io n s i ndicati n gt h at t he c ited so ur ce is wro n g,is notnegligibl e ( 1 4%).
pr esented one of the f ir st automat ic ci tation i ndexingsystems (CiteSeer). Itc ould pa rse cita tion s and use th em tocom p ute si milari tiesbe twe e n docu m e nt s.
m ight havei m ple mente dthefirst a utomated classific a tio n systems for ci tati o n s. Th e yu sed o v er2 0 0 manually sele cted rules t o c lassify ci t ati on s in on e of 35 cate g ories.
Machinelearningme thod s for automatic classif ication c a n bea pp liedtothe te xt of a ci ting d o cum ent.distin gu ish ca tegor ie s of cit ations that can be iden tified vialin guistic c ues inthe text. The y are able to cl assif y c i tatio ns i n to on e of f ourc ategories (w eak , po sitive, con t r a st, neut ral ) withan a verage of 68%. Fo r a classificat ioni n th ree cate go ries (weakness , p os i t ive, neu tr al), they g et an av er a ge of 71%.Theirclassif i e rr elieson 8 92manuallysel ec t ed cueph ra s es, su ch a swhethe r thec itat i o n is a self-cita tion, t he lo c ati on of t he cita t ionin the tex t, and manu ally a cqui red v erb clu st ers.
ann ot ated a cor p us of 43open- accessfu ll-t ext biome dica l artic les. T hey built cl a s si f ie rs usi ng S uppor tVect or Machin e (SVM) a ndM ultinom ia l N a | good_model of_citation influence could suggest_a model_of_hypertext link_importance._This could improve_measures of the_importance of web pages,_such as PageRank [@Qi:2007:MSD:1244408.1244418].
Improved_recommender_systems: Researchers often need help identifying relevant work that they should read. Filtering out_less_relevant citations_might_help_paper recommender systems [@MEET:MEET14504701330; @springerlink:10.1007/978-3-642-23535-1_35].
Related work_{#related-work.unnumbered}
============
The idea that the mere_counting of_citations is dubious is not new [@chubin1975content]: The field_of_*citation context analysis*_has a long history dating back to the early_days of citation indexing. There is_a wide variety_of_reasons_for a researcher to_cite a source and many ways_of categorizing them. For instance, identified_fifteen such reasons, including giving credit for_related work, correcting a work, and_criticizing previous work.
For articles in_the field_of high energy physics, distinguished_four major classes_of polar_opposite pairs, conceptual–operational,_organic–perfunctory, evolutionary–juxtapositional, and confirmative–negational. They found_that the fraction_of negational references, i.e., citations indicating_that_the cited source_is_wrong,_is not_negligible (14%).
presented one_of_the first_automatic_citation indexing systems (CiteSeer). It could_parse_citations and use them to compute similarities_between documents.
might have implemented_the_first automated classification systems_for citations. They used over_200 manually selected rules to classify citations_in one_of 35 categories.
Machine_learning methods for automatic classification can be applied to the text_of a citing document. distinguish categories_of citations that can_be identified_via_linguistic cues in_the_text. They_are able to classify citations into one_of four_categories (weak, positive, contrast, neutral) with_an average of 68%._For_a classification in three categories (weakness,_positive, neutral), they get an average_of 71%. Their classifier relies_on_892 manually_selected cue phrases, such as_whether the citation is a self-citation,_the location of_the citation in the text, and manually_acquired_verb clusters.
annotated a corpus of 43 open-access_full-text_biomedical articles. They built classifiers using_Support_Vector_Machine (SVM) and Multinomial Na |
=(\mu_1,\mu_2,\dots)$, $$\mu \le \lambda \qquad \Leftrightarrow \qquad
\mu_1 + \cdots+\mu_i \le \lambda_1+ \cdots +\lambda_i \quad \text{for all $i \ge 1$}.$$ Let ${\mathbb{C}}[\bx^{\pm 1}] = {\mathbb{C}}[x_1^{\pm 1}, \dots, x_n^{\pm 1}]$ be the ring of all Laurent polynomials in $n$ variables $\bx=(x_1,\dots,x_n)$. The Weyl group $W={\mathbb{Z}}_2 \wr {\mathfrak{S}}_n = {\mathbb{Z}}_2^n \rtimes {\mathfrak{S}}_n$ of type $BC_n$ acts naturally on ${\mathbb{Z}}^n$ and ${\mathbb{C}}[\bx^{\pm 1}]$, respectively. Denote by ${\mathbb{C}}[\bx^{\pm 1}]^W$ the subring of all $W$-invariants in ${\mathbb{C}}[\bx^{\pm 1}]$. Let $\Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3)$ be a function on ${\mathbb{T}}^n$ defined by $$\Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3)
= \prod_{1 \le i < j \le n}
|1-z_i z_j^{-1}|^{2 k_3} |1-z_i z_j|^{2 k_3}
\cdot
\prod_{1 \le j \le n}
|1-z_j|^{2k_1} |1-z_j^2|^{2k_2}.$$ Here we assume $k_1$, $k_2$, and $k_3$ are real numbers such that $$k_1+k_2>-1/2, \quad k_2 > -1/2, \quad k_3 \ge 0.$$ Define an inner product on ${\mathbb{C}}[\bx^{\pm 1}]^W$ by $$\langle f,g \rangle_{\Delta^{{\mathrm{HO}}}} =
\frac{1}{2^n n!} \int_{{\mathbb{T}}^n} f(\bz) g(\bz^{-1}) \ | =( \mu_1,\mu_2,\dots)$, $ $ \mu \le \lambda \qquad \Leftrightarrow \qquad
\mu_1 + \cdots+\mu_i \le \lambda_1 + \cdots + \lambda_i \quad \text{for all $ i \ge 1$}.$$ Let $ { \mathbb{C}}[\bx^{\pm 1 } ] = { \mathbb{C}}[x_1^{\pm 1 }, \dots, x_n^{\pm 1}]$ be the ring of all Laurent polynomials in $ n$ variables $ \bx=(x_1,\dots, x_n)$. The Weyl group $ W={\mathbb{Z}}_2 \wr { \mathfrak{S}}_n = { \mathbb{Z}}_2^n \rtimes { \mathfrak{S}}_n$ of character $ BC_n$ act naturally on $ { \mathbb{Z}}^n$ and $ { \mathbb{C}}[\bx^{\pm 1}]$, respectively. Denote by $ { \mathbb{C}}[\bx^{\pm 1}]^W$ the subring of all $ W$-invariants in $ { \mathbb{C}}[\bx^{\pm 1}]$. Let $ \Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3)$ be a affair on $ { \mathbb{T}}^n$ define by $ $ \Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3)
= \prod_{1 \le i < j \le n }
|1 - z_i z_j^{-1}|^{2 k_3 } |1 - z_i z_j|^{2 k_3 }
\cdot
\prod_{1 \le j \le n }
|1 - z_j|^{2k_1 } |1 - z_j^2|^{2k_2}.$$ Here we assume $ k_1 $, $ k_2 $, and $ k_3 $ are real number such that $ $ k_1+k_2>-1/2, \quad k_2 > -1/2, \quad k_3 \ge 0.$$ specify an inner merchandise on $ { \mathbb{C}}[\bx^{\pm 1}]^W$ by $ $ \langle f, g \rangle_{\Delta^{{\mathrm{HO } } } } =
\frac{1}{2^n n! } \int_{{\mathbb{T}}^n } f(\bz) g(\bz^{-1 }) \ | =(\mu_1,\mk_2,\dots)$, $$\mu \le \lambda \qquad \Leftrightarrow \qquad
\ku_1 + \cdkts+\mu_i \ld \lambda_1+ \cdots +\lambda_i \quad \vext{dor aol $i \ge 1$}.$$ Let ${\mathbb{C}}[\bx^{\om 1}] = {\mathhb{C}}[x_1^{\pm 1}, \eots, z_n^{\pm 1}]$ be tis ring of all Paurznv polynomials ik $n$ variablas $\bx=(x_1,\dots,x_n)$. Tve Wzyl group $W={\mathbb{Z}}_2 \wr {\mathfrak{S}}_n = {\maehbb{Z}}_2^n \ttlmes {\mathfrak{S}}_g$ of eype $BC_n$ acts naturally on ${\mathbb{Z}}^n$ ans ${\mathbu{C}}[\bx^{\pm 1}]$, respectovely. Denote by ${\mathbb{C}}[\bx^{\pl 1}]^W$ hhe subring of all $W$-invariantw in ${\nathbb{C}}[\bx^{\pm 1}]$. Let $\Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,i_3)$ be a function on ${\mathbb{T}}^n$ defkned yy $$\Delta^{{\matyrn{HO}}}(\he;k_1,k_2,k_3)
= \prid_{1 \le i < j \le n}
|1-z_i z_j^{-1}|^{2 k_3} |1-z_o z_j|^{2 k_3}
\ccot
\prod_{1 \le j \le n}
|1-z_j|^{2k_1} |1-z_j^2|^{2k_2}.$$ Here wg assume $k_1$, $k_2$, and $k_3$ are real nymvers vuch thag $$k_1+k_2>-1/2, \quzd k_2 > -1/2, \quaf k_3 \ge 0.$$ Defins an inner product on ${\mathbb{C}}[\bc^{\pi 1}]^W$ by $$\langle r,g \randlq_{\Delta^{{\mathrm{HO}}}} =
\frac{1}{2^n n!} \int_{{\mathbb{T}}^n} f(\bz) g(\ba^{-1}) \ | =(\mu_1,\mu_2,\dots)$, $$\mu \le \lambda \qquad \Leftrightarrow \qquad \cdots+\mu_i \lambda_1+ \cdots \quad \text{for all 1}] {\mathbb{C}}[x_1^{\pm 1}, \dots, 1}]$ be the of all Laurent polynomials in $n$ $\bx=(x_1,\dots,x_n)$. The Weyl group $W={\mathbb{Z}}_2 \wr {\mathfrak{S}}_n = {\mathbb{Z}}_2^n \rtimes {\mathfrak{S}}_n$ of type acts naturally on ${\mathbb{Z}}^n$ and ${\mathbb{C}}[\bx^{\pm 1}]$, respectively. Denote by ${\mathbb{C}}[\bx^{\pm 1}]^W$ the of $W$-invariants ${\mathbb{C}}[\bx^{\pm Let $\Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3)$ be a function on ${\mathbb{T}}^n$ defined by $$\Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3) = \prod_{1 \le i < j n} |1-z_i z_j^{-1}|^{2 k_3} |1-z_i z_j|^{2 k_3} \cdot \le j \le n} |1-z_j^2|^{2k_2}.$$ Here we assume $k_1$, and are real such $$k_1+k_2>-1/2, k_2 > -1/2, k_3 \ge 0.$$ Define an inner product on ${\mathbb{C}}[\bx^{\pm 1}]^W$ by $$\langle f,g \rangle_{\Delta^{{\mathrm{HO}}}} = \frac{1}{2^n n!} f(\bz) g(\bz^{-1}) | =(\mu_1,\mu_2,\dots)$, $$\mu \le \lambda \qquad \LeFtrightarrOw \qquAd
\mU_1 + \cdOtS+\mu_i \Le \laMbda_1+ \cdots +\lambdA_I \quaD \text{for all $i \ge 1$}.$$ Let ${\mathbB{C}}[\bx^{\pM 1}] = {\mAThbb{c}}[X_1^{\pM 1}, \dots, X_n^{\pm 1}]$ be tHE rING of AlL LAurEnT PoLynomIalS in $n$ varIables $\bx=(x_1,\dOts,X_n)$. the Weyl group $w={\MaThbb{Z}}_2 \wr {\matHfrAk{S}}_n = {\mathbb{Z}}_2^n \RtiMes {\matHfRak{s}}_N$ of tyPe $Bc_n$ actS naturALly on ${\mAthbb{Z}}^n$ anD ${\mAThbb{C}}[\bX^{\Pm 1}]$, respeCTIvEly. DEnote by ${\mathbb{C}}[\bx^{\pM 1}]^w$ tHE subring of all $W$-InvariAnTS iN ${\MAthBb{C}}[\Bx^{\pm 1}]$. Let $\DelTa^{{\MathrM{hO}}}(\bz;k_1,k_2,k_3)$ BE a FUNCtiON on ${\mathbb{T}}^n$ deFined by $$\DeltA^{{\MatHrm{HO}}}(\bZ;k_1,K_2,k_3)
= \pROd_{1 \le i < j \Le n}
|1-z_i Z_j^{-1}|^{2 K_3} |1-Z_i z_J|^{2 k_3}
\cdot
\prod_{1 \lE j \le N}
|1-z_j|^{2k_1} |1-z_j^2|^{2k_2}.$$ HeRe we asSUme $k_1$, $k_2$, anD $K_3$ are reaL numbeRs sUch That $$K_1+K_2>-1/2, \qUaD k_2 > -1/2, \qUaD K_3 \ge 0.$$ dEfIne AN inNer produCt On ${\MathbB{C}}[\bx^{\PM 1}]^w$ BY $$\lanGle F,g \raNgle_{\DElta^{{\mathrm{HO}}}} =
\fRac{1}{2^N n!} \inT_{{\MatHbb{T}}^n} F(\bz) g(\bZ^{-1}) \ | =(\mu_1,\mu_2,\dots)$, $$\ mu \le \la mbda\qq uad \ Left righ tarrow \qquad\mu_ 1 + \cdots+\mu_i \le \ lambd a_ 1 + \c d ot s +\l ambda_i \q u a d \ te xt {fo ra ll $i \ ge1$}.$$Let ${\mat hbb {C }}[\bx^{\pm1 }] = {\mathb b{C }}[x_1^{\pm1}, \dots ,x_n ^ {\pm1}] $ bethe ri n g of a ll Lauren tp olynom i als in$ n $vari ables $\bx=(x_1,\ d ot s ,x_n)$. The We yl gro up $W = { \ma thb b{Z}}_2 \w r{\mat h frak{S} } _n = {\m a thbb{Z}}_2^n\rtimes {\m a thf rak{S} }_ n$o f type $BC_ n$ act s naturally on${\mathbb {Z}}^n $ and ${ \ mathbb{ C}}[\b x^{ \pm 1}] $ ,re spe ct i vel y .Den o teby ${\ma th bb {C}}[ \bx^ { \ p m 1}] ^W$ the subr ing of all $W $-i nvar i ant s in${\ma thbb {C }}[\b x^{\pm 1}]$ .Let $\Delta^{{\ math rm{HO}}}( \bz ;k _1, k_ 2,k_3 ) $ be a fu nct ion on${\math b b{T }} ^ n $ d efined by $$\Delta ^{ { \ ma thrm{HO} }}(\bz ; k_ 1, k _2,k_3)
= \p r o d_{1\lei < j \le n }
| 1-z_i z _j ^{-1}| ^{ 2 k _3} |1-z _ i z_ j|^{2k_3}
\ c dot
\pro d _{1 \le j \l e n } |1- z_j|^{2k_1} |1- z _j^2 |^{2 k _2 }.$ $ Here we a ss u me $k_1$, $k_2$, and $ k_ 3$ are real numbers such that $$k_ 1 + k _2>-1/2, \qu a dk _2 > -1/2, \qu ad k_ 3 \ge 0.$$ Define a n inn er produ ct on ${\ m a thbb{C}} [\b x^{ \pm 1} ] ^ W$ by $$\langle f ,g \ ra ngle_{\ Del ta^{{\m ath rm{ HO} }}} =
\frac{1 }{2^n n! }\i nt _{ {\m athbb { T}}^n} f (\ bz) g (\b z^{-1 } ) \ | =(\mu_1,\mu_2,\dots)$, $$\mu_\le \lambda_\qquad \Leftrightarrow \qquad
\mu_1_+ \cdots+\mu_i_\le_\lambda_1+ \cdots_+\lambda_i_\quad \text{for all_$i \ge 1$}.$$_Let ${\mathbb{C}}[\bx^{\pm 1}] =_{\mathbb{C}}[x_1^{\pm 1}, \dots,_x_n^{\pm_1}]$ be the ring of all Laurent polynomials in $n$ variables $\bx=(x_1,\dots,x_n)$. The Weyl_group_$W={\mathbb{Z}}_2 \wr_{\mathfrak{S}}_n_=_{\mathbb{Z}}_2^n \rtimes {\mathfrak{S}}_n$ of type_$BC_n$ acts naturally on ${\mathbb{Z}}^n$_and ${\mathbb{C}}[\bx^{\pm_1}]$, respectively. Denote by ${\mathbb{C}}[\bx^{\pm 1}]^W$ the subring_of_all $W$-invariants in_${\mathbb{C}}[\bx^{\pm 1}]$. Let $\Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3)$ be a function on ${\mathbb{T}}^n$_defined by $$\Delta^{{\mathrm{HO}}}(\bz;k_1,k_2,k_3)
=_ \prod_{1_\le_i_< j \le n}
_ _ |1-z_i z_j^{-1}|^{2 k_3} |1-z_i z_j|^{2_k_3}
_\cdot
_\prod_{1 \le j \le_n}
_ _|1-z_j|^{2k_1} |1-z_j^2|^{2k_2}.$$ Here_we assume_$k_1$, $k_2$, and_$k_3$ are real numbers such that_$$k_1+k_2>-1/2, \quad k_2_> -1/2, \quad k_3 \ge 0.$$_Define_an inner product_on_${\mathbb{C}}[\bx^{\pm_1}]^W$ by_$$\langle f,g \rangle_{\Delta^{{\mathrm{HO}}}}_=_
\frac{1}{2^n n!}_\int_{{\mathbb{T}}^n}_f(\bz) g(\bz^{-1}) \ |
,25)[$E_6$]{} (132, 3)[$\alpha_2$]{} ( 61,43)[$\alpha_1$]{} ( 91,43)[$\alpha_3$]{} (121,43)[$\alpha_4$]{} (151,43)[$\alpha_5$]{} (181,43)[$\alpha_6$]{} (125, 5) ( 65,35) ( 95,35) (125,35) (155,35) (185,35) (125,35)[(0,-1)[30]{}]{} ( 65,35)[(1,0)[30]{}]{} ( 95,35)[(1,0)[30]{}]{} (125,35)[(1,0)[30]{}]{} (155,35)[(1,0)[30]{}]{}
Again, since $p=2$, we do not have to worry about the precise choice of a Chevalley basis in the underlying Lie algebra. Further note that $G({{\mathbb{F}}}_2)$ is a simple group; it will just be denoted by $E_6({{\mathbb{F}}}_2)$. (For $q$ an even power of $2$, the group $G({{\mathbb{F}}}_q)$ has a normal subgroup of index $3$). Now we essentially rely on the knowledge of the character table of $E_6({{\mathbb{F}}}_2)$, which has been determined by B. Fischer and is contained in the [GAP]{} library [@bre]. (We have $|{{\operatorname{Irr}}}(E_6({{\mathbb{F}}}_2))|=180$.) Although the group is bigger, the further discussion will actually be simpler than that for type $F_4$ in the previous section, since there are fewer cuspidal unipotent character sheaves.
\[r61\] By Lusztig [@L2d 20.3] and Shoji [@S3 4.6], there are two cuspidal character sheaves $A_1,A_2\in
\hat{G}^{\text{un}}$. Both $A_1$, $A_2$ have the same support, namely, the closure of the $G$-conjugacy class $C$ of an element $g_1=su\in G^F$ such that $s\in G^F$ is semisimple, with $C_G(s)^\circ$ isogneous to $\mbox{SL}_3(k)\times \mbox{SL | , 25)[$E_6 $ ] { } (132, 3)[$\alpha_2 $ ] { } (61,43)[$\alpha_1 $ ] { } (91,43)[$\alpha_3 $ ] { } (121,43)[$\alpha_4 $ ] { } (151,43)[$\alpha_5 $ ] { } (181,43)[$\alpha_6 $ ] { } (125, 5) (65,35) (95,35) (125,35) (155,35) (185,35) (125,35)[(0,-1)[30 ] { } ] { } (65,35)[(1,0)[30 ] { } ] { } (95,35)[(1,0)[30 ] { } ] { } (125,35)[(1,0)[30 ] { } ] { } (155,35)[(1,0)[30 ] { } ] { }
Again, since $ p=2 $, we do not have to worry about the precise choice of a Chevalley footing in the implicit in Lie algebra. Further note that $ G({{\mathbb{F}}}_2)$ is a simple group; it will barely be denote by $ E_6({{\mathbb{F}}}_2)$. (For $ q$ an even power of $ 2 $, the group $ G({{\mathbb{F}}}_q)$ have a normal subgroup of index $ 3 $). Now we essentially rely on the knowledge of the quality table of $ E_6({{\mathbb{F}}}_2)$, which has been determined by B. Fischer and is check in the [ GAP ] { } library [ @bre ]. (We accept $ |{{\operatorname{Irr}}}(E_6({{\mathbb{F}}}_2))|=180$.) Although the group is bigger, the further discussion will actually be simpler than that for type $ F_4 $ in the previous section, since there embody fewer cuspidal unipotent character bundle.
\[r61\ ] By Lusztig [ @L2d 20.3 ] and Shoji [ @S3 4.6 ], there are two cuspidal fictional character sheaves $ A_1,A_2\in
\hat{G}^{\text{un}}$. Both $ A_1 $, $ A_2 $ have the like support, namely, the blockage of the $ G$-conjugacy class $ C$ of an element $ g_1 = su\in G^F$ such that $ s\in G^F$ is semisimple, with $ C_G(s)^\circ$ isogneous to $ \mbox{SL}_3(k)\times \mbox{SL | ,25)[$E_6$]{} (132, 3)[$\wlpha_2$]{} ( 61,43)[$\alpha_1$]{} ( 91,43)[$\alpha_3$]{} (121,43)[$\alphx_4$]{} (151,43)[$\alpha_5$]{} (181,43)[$\alpha_6$]{} (125, 5) ( 65,35) ( 95,35) (125,35) (155,35) (185,35) (125,35)[(0,-1)[30]{}]{} ( 65,35)[(1,0)[30]{}]{} ( 95,35)[(1,0)[30]{}]{} (125,35)[(1,0)[30]{}]{} (155,35)[(1,0)[30]{}]{}
Again, sknce $p=2$, we do not have to worcy avout uke precise choice of a Chevalpey basiw in rhe underlbjng Lie algebdw. Fuxtier note that $G({{\kathbb{F}}}_2)$ is a simple grou[; kt will just be denoted by $E_6({{\mathbb{F}}}_2)$. (Fjr $q$ an egen power of $2$, tre ggotp $G({{\jathbb{F}}}_q)$ has a normal subgroup of jndex $3$). Nmw we essentislly rely on the knowledge of hhe character tablf of $E_6({{\mathbv{F}}}_2)$, wruch has been determined by B. Fischet and is contained in the [GAP]{} licrary [@yre]. (We have $|{{\operwjorname{Irr}}}(E_6({{\methbb{F}}}_2))|=180$.) Although thc group is bigber, the furthev disrussuon will actually be vimpler than that sor type $X_4$ nn the previous sectiin, sincg these afw fdwed ruslidal kni'otent charzcter sheavws.
\[r61\] By Lusztig [@L2d 20.3] snq Shoji [@S3 4.6], thede are tro cuspidal character sheaves $A_1,A_2\in
\hat{G}^{\ttxt{un}}$. Both $A_1$, $A_2$ have the same support, namely, the cposure of the $G$-conjugacy class $C$ of an element $g_1=su\in G^F$ suwh thet $s\iu G^F$ ir sfmisimple, with $C_G(s)^\circ$ isogneous to $\mbox{SL}_3(k)\tiiss \mnox{SL | ,25)[$E_6$]{} (132, 3)[$\alpha_2$]{} ( 61,43)[$\alpha_1$]{} ( 91,43)[$\alpha_3$]{} (181,43)[$\alpha_6$]{} 5) ( ( 95,35) (125,35) ( (125,35)[(1,0)[30]{}]{} (155,35)[(1,0)[30]{}]{} Again, $p=2$, we do have to worry about the precise of a Chevalley basis in the underlying Lie algebra. Further note that $G({{\mathbb{F}}}_2)$ a simple group; it will just be denoted by $E_6({{\mathbb{F}}}_2)$. (For $q$ an power $2$, group has a normal subgroup of index $3$). Now we essentially rely on the knowledge of the table of $E_6({{\mathbb{F}}}_2)$, which has been determined by Fischer and is contained the [GAP]{} library [@bre]. (We $|{{\operatorname{Irr}}}(E_6({{\mathbb{F}}}_2))|=180$.) the group bigger, further will actually be than that for type $F_4$ in the previous section, since there are fewer cuspidal unipotent character sheaves. By Lusztig and Shoji 4.6], are cuspidal character sheaves Both $A_1$, $A_2$ have the same closure of the $G$-conjugacy class $C$ of an $g_1=su\in G^F$ that $s\in G^F$ is semisimple, with isogneous to $\mbox{SL}_3(k)\times \mbox{SL | ,25)[$E_6$]{} (132, 3)[$\alpha_2$]{} ( 61,43)[$\alpha_1$]{} ( 91,43)[$\alpha_3$]{} (121,43)[$\alpha_4$]{} (151,43)[$\alphA_5$]{} (181,43)[$\alpha_6$]{} (125, 5) ( 65,35) ( 95,35) (125,35) (155,35) (185,35) (125,35)[(0,-1)[30]{}]{} ( 65,35)[(1,0)[30]{}]{} ( 95,35)[(1,0)[30]{}]{} (125,35)[(1,0)[30]{}]{} (155,35)[(1,0)[30]{}]{}
AgaiN, sincE $p=2$, wE do NoT havE to wOrry about the prECise Choice of a Chevalley basiS in thE uNDerlYInG Lie aLgebra. FURtHER noTe ThAt $G({{\MaTHbB{F}}}_2)$ is a SimPle grouP; it will jusT be DeNoted by $E_6({{\mathBB{F}}}_2)$. (for $q$ an even PowEr of $2$, the group $g({{\maThbb{F}}}_q)$ HaS a nORmal sUbgRoup oF index $3$). nOw we esSentially ReLY on the KNowledgE OF tHe chAracter table of $E_6({{\maTHbB{f}}}_2)$, which has been dEtermiNeD By b. fIscHer And is contaInEd in tHE [GAP]{} libRArY [@BRE]. (We HAve $|{{\operatornaMe{Irr}}}(E_6({{\mathbB{f}}}_2))|=180$.) AlThough ThE grOUp is biGger, tHe FUrtHer discussiOn wiLl actuallY be simPLer than THat for tYpe $F_4$ in The PreViouS SeCtIon, SiNCe tHErE arE FewEr cuspidAl UnIpoteNt chARACTer sHeaVes.
\[r61\] by LusZtig [@L2d 20.3] and ShojI [@S3 4.6], tHere ARe tWo cusPidal CharAcTer shEaves $A_1,a_2\in
\haT{G}^{\Text{un}}$. Both $A_1$, $A_2$ havE the Same suppoRt, nAmEly, ThE closURe of thE $G$-cOnjUgacy clAss $C$ of aN EleMeNT $G_1=Su\In G^F$ such that $s\in G^F$ iS sEMIsImple, witH $C_G(s)^\ciRC$ iSoGNeous to $\mBoX{SL}_3(K)\timES \Mbox{Sl | ,25)[$E_6$]{} (132, 3)[$\a lpha_2$]{} ( 61 ,43 )[$ \a lpha _1$] {} ( 91,43)[$\ a lpha _3$]{} (121,43)[$\alph a_4$] {} (151 , 43 )[$\a lpha_5$ ] {} ( 181 ,4 3) [$\ al p ha _6$]{ } ( 125, 5) ( 65,35)( 9 5, 35) (125,35) (1 55,35) (18 5,3 5) (125,35)[ (0, -1)[30 ]{ }]{ } ( 65 ,35 )[(1, 0)[30] { }]{} ( 95,35)[( 1, 0 )[30]{ } ]{} (12 5 , 35 )[(1 ,0)[30]{}]{} (155 , 35 ) [(1,0)[30]{}]{ }
Aga in , s i n ce$p= 2$, we dono t hav e to wor r ya b o utt he precise ch oice of a C h eva lley b as isi n theunder ly i ngLie algebra . Fu rther not e that $G({{\m a thbb{F} }}_2)$ is asimp l egr oup ;i t w i ll ju s t b e denote dby $E_6 ({{\ m a t h bb{F }}} _2)$ . (Fo r $q$ an even po wero f $ 2$, t he gr oup$G ({{\m athbb{ F}}}_ q) $ has a normalsubg roup of i nde x$3$ ). Noww e esse nti all y relyon thek now le d g e o f the character ta bl e of $E_6({{ \mathb b {F }} } _2)$, wh ic h h as b e e n det ermi n ed by B. F ischer an dis cont ai ned in t he[GA P]{}l ibra ry [@b re]. (We have $|{{\operatorn a me{Irr}}}(E_6 ( {{ \ m at h bb{F }}} _2))|=180$. ) Al t houg h th e g rou p is b igger ,t he further discussionwi ll act ually be simpler t han that f o r type $F_ 4$ i n t h e previous sec tion, since the r e are fe wer c uspidalunipotent c haracter sh eav es.
\ [ r 61 \] By Lusztig [ @L2d 2 0.3] an d S hoji [@ S34.6 ],the re are twocuspidal c ha ra ct ersheav e s $A_1,A _2 \in
\ hat {G}^{ \ text{u n}}$. Bot h$A _ 1$, $A_2$h av e thesa me sup por t, name ly,t heclosure of the $ G$- c onju ga cy class$C$ of an ele me nt $g_1=su \i n G ^F$ su c h that $s \in G^F$ is semisimple, with $C _G( s)^\c irc$ isogneou s t o $\mb ox{ S L}_3(k )\time s \mb ox {SL | ,25)[$E_6$]{} (132,_3)[$\alpha_2$]{} (_61,43)[$\alpha_1$]{} ( 91,43)[$\alpha_3$]{} (121,43)[$\alpha_4$]{}_(151,43)[$\alpha_5$]{} (181,43)[$\alpha_6$]{}_(125,_5) (_65,35)_( 95,35) (125,35)_(155,35) (185,35) (125,35)[(0,-1)[30]{}]{}_( 65,35)[(1,0)[30]{}]{} ( 95,35)[(1,0)[30]{}]{}_(125,35)[(1,0)[30]{}]{} (155,35)[(1,0)[30]{}]{}
Again, since_$p=2$,_we do not have to worry about the precise choice of a Chevalley basis_in_the underlying_Lie_algebra._Further note that $G({{\mathbb{F}}}_2)$ is_a simple group; it will_just be_denoted by $E_6({{\mathbb{F}}}_2)$. (For $q$ an even power_of $2$,_the group $G({{\mathbb{F}}}_q)$_has a normal subgroup of index $3$). Now we essentially_rely on the knowledge of the_character table of_$E_6({{\mathbb{F}}}_2)$,_which_has been determined by_B. Fischer and is contained in the_[GAP]{} library [@bre]. (We have $|{{\operatorname{Irr}}}(E_6({{\mathbb{F}}}_2))|=180$.) Although_the group is bigger, the further discussion_will actually be simpler than that_for type $F_4$ in the_previous section,_since there are fewer cuspidal_unipotent character sheaves.
\[r61\]_By Lusztig_[@L2d 20.3] and_Shoji [@S3 4.6], there are two_cuspidal character sheaves_$A_1,A_2\in
\hat{G}^{\text{un}}$. Both $A_1$, $A_2$ have the_same_support, namely, the_closure_of_the $G$-conjugacy_class $C$ of_an_element $g_1=su\in_G^F$_such that $s\in G^F$ is semisimple,_with_$C_G(s)^\circ$ isogneous to $\mbox{SL}_3(k)\times \mbox{SL |
] symmetry which can be extended to all sectors of the Lagrangian with an invisible axion solving the strong-CP problem [@Pal95];
5. Spontaneous CP violation in the electroweak sector [@Epele95]. There are several natural sources of explicit and spontaneous CP violations [@vicente];
6. The quark mass hierarchy [@austr];
7. Although the leptoquark-bilepton models do not conserve each generation lepton number $L_\ell$, the neutrinoless double beta decay is forbidden because of the conservation of the quantum number ${\cal F}\equiv L + B$, where $B$ is the barion number and $L=\sum_\ell L_\ell$ is the total lepton number. If this global symmetry is explicitly violated in the Higgs potential, there are contributions to the decay which depend less on the neutrino mass than they do in too many extensions of the standard model [@pt93]. The double beta decay with Majoron emission is possible as well [@fpshe1];
8. There is an electric charge quantization without any constraint on the Dirac, Majorana or Dirac–Majorana [@Esposito] character of the massive neutral fermions [@doff].
Representation contents are determined by embedding the electric charge operator $$\frac{{\cal Q}}{|e|} = (\Lambda_3 + \xi\Lambda_8 + \zeta\Lambda_{15}) + N
\label{opcar}$$ in the neutral generators $\Lambda_{3,8,15} = \lambda^{\rm SU(4)}_{3,8,15}/2$ of the largest weak isospin group SU(4) extension and $N$ is the new U(1)$_N$ charge equivalent to the electric charge average of the fermions contained in each flavor multiplet. If we consider the lightest leptons as the fermions which determine the approximate symmetry, and also independent flavor generations, then SU(4)$\times$U(1) is the largest non-symmetric gauge group of the electroweak sector. There is no room for the chiral semisimple group SU(5)$\times$U(1) if lepton electric charges are only $0$, $\pm 1$. The weak hypercharge of the $G_{321}$ standard model is $$\frac{Y}{2} = (\xi\Lambda_8 + \zeta\Lambda_{15})+N
\label{anaa | ] symmetry which can be extended to all sectors of the Lagrangian with an invisible axion clear the impregnable - CP problem [ @Pal95 ];
5. Spontaneous CP violation in the electroweak sector [ @Epele95 ]. There be several natural informant of explicit and spontaneous CP misdemeanor [ @vicente ];
6. The quark mass hierarchy [ @austr ];
7. Although the leptoquark - bilepton models do not conserve each genesis lepton number $ L_\ell$, the neutrinoless double beta decay is prevent because of the conservation of the quantum numeral $ { \cal F}\equiv L + B$, where $ B$ is the barion number and $ L=\sum_\ell L_\ell$ is the total lepton number. If this global symmetry is explicitly violate in the Higgs potential, there are contributions to the decay which count less on the neutrino mass than they do in excessively many extensions of the standard exemplar [ @pt93 ]. The double beta decay with Majoron emission is possible as well [ @fpshe1 ];
8. There is an electric mission quantization without any constraint on the Dirac, Majorana or Dirac – Majorana [ @Esposito ] character of the massive neutral fermions [ @doff ].
Representation subject are determined by embedding the electric charge operator $ $ \frac{{\cal Q}}{|e| } = (\Lambda_3 + \xi\Lambda_8 + \zeta\Lambda_{15 }) + N
\label{opcar}$$ in the neutral generators $ \Lambda_{3,8,15 } = \lambda^{\rm SU(4)}_{3,8,15}/2 $ of the largest weak isospin group SU(4) extension and $ N$ is the new U(1)$_N$ tear equivalent to the electric charge average of the fermions control in each relish multiplet. If we consider the lightest lepton as the fermions which determine the approximate symmetry, and also autonomous flavor generations, then SU(4)$\times$U(1) is the largest non - symmetric gauge group of the electroweak sector. There is no room for the chiral semisimple group SU(5)$\times$U(1) if lepton electric charges are only $ 0 $, $ \pm 1$. The weak hypercharge of the $ G_{321}$ standard model is $ $ \frac{Y}{2 } = (\xi\Lambda_8 + \zeta\Lambda_{15})+N
\label{anaa | ] sylmetry which can be extekded to all sectors of vhe Lagdangian dith an invisible axion solvmng rhe sugong-CP problem [@Pal95];
5. Spuntaneous CP violqtioi in the electroxsak secbjr [@Epspe95]. Tkece are several katural sousces of explicht aud spontaneous CP violations [@vicente];
6. The qusrn mass hierarcry [@auxer];
7. Zlthough the leptoquark-bilepton mosels do not conserve each generation lepton nulber $L_\ell$, the neutrinopess double betw decay is fofbidden bebcuse of the conservation of the quantum nuober ${\eal F}\equiv O + B$, avere $B$ is tie barpon number and $L=\sum_\eln L_\ell$ os the total lcpton nunber. If this global sbmmetry is explicitli violated iu the Higgs potential, tyere dre wontfubugiohs tk the fecey which delend less ob the neutrino mass trqn they do in too mwnr extensions of the standard model [@pt93]. Tht dougle beta decay with Majiron emission is posslble as wqll [@fpshe1];
8. There is an electric charge quantizatiot witiojt ckn covwtgaint on the Dirac, Majorana or Dirac–Majorana [@Efloxino] character of tme massive neutral ffrkyons [@doff].
Reprerentatnkn contents are detegmined fy emvedding tre ekectric charge operator $$\frax{{\cal Q}}{|e|} = (\Lamyda_3 + \xi\Lambda_8 + \zeta\Lalbda_{15}) + N
\labeu{opcsr}$$ in the neutral generators $\Lamgda_{3,8,15} = \lambda^{\gm SU(4)}_{3,8,15}/2$ of ffe largest weak ksoxphn group SU(4) extension and $G$ is the iew U(1)$_U$ charge equovalene to the epectrlw charge average ov the farmions cojtained in each flavor multiplev. If we consices tve lightzst leitons as the fewmions which dgtermine che aporoximate aymmetrb, and also igdependent fldgor generatimns, then SU(4)$\tumes$Y(1) is thd largest non-sykmetric gcbge group of the electroweal sgcfor. There is no roim for the chirsl remyspmpke croup SU(5)$\timev$U(1) iw ldlton dlectric chcrees sre only $0$, $\pm 1$. The wedk hgpercharge of the $B_{321}$ ftandard model if $$\frac{Y}{2} = (\xi\Lakbda_8 + \zeta\Lambda_{15})+N
\lwbel{aiaa | ] symmetry which can be extended to of Lagrangian with invisible axion solving Spontaneous violation in the sector [@Epele95]. There several natural sources of explicit and CP violations [@vicente]; 6. The quark mass hierarchy [@austr]; 7. Although the leptoquark-bilepton do not conserve each generation lepton number $L_\ell$, the neutrinoless double beta decay forbidden of conservation the quantum number ${\cal F}\equiv L + B$, where $B$ is the barion number and $L=\sum_\ell is the total lepton number. If this global is explicitly violated in Higgs potential, there are contributions the which depend on neutrino than they do too many extensions of the standard model [@pt93]. The double beta decay with Majoron emission is possible well [@fpshe1]; is an charge without constraint on the or Dirac–Majorana [@Esposito] character of the [@doff]. Representation contents are determined by embedding the charge operator Q}}{|e|} = (\Lambda_3 + \xi\Lambda_8 + + N \label{opcar}$$ in the neutral generators $\Lambda_{3,8,15} \lambda^{\rm SU(4)}_{3,8,15}/2$ of the largest weak isospin group SU(4) extension and $N$ is the new equivalent to the electric average of the contained each multiplet. we consider lightest leptons as the fermions which determine the approximate symmetry, and independent flavor generations, then SU(4)$\times$U(1) is the largest non-symmetric gauge the sector. There is room for the chiral group if lepton electric charges $0$, 1$. of $G_{321}$ model is $$\frac{Y}{2} = + \zeta\Lambda_{15})+N \label{anaa | ] symmetry which can be extendeD to all sectOrs of The lagRaNgiaN witH an invisible axIOn soLving the strong-CP probleM [@Pal95];
5. SPoNTaneOUs cP vioLation iN ThE ELecTrOwEak SeCToR [@EpelE95]. ThEre are sEveral natuRal SoUrces of expliCIt And spontanEouS CP violationS [@viCente];
6. THe QuaRK mass HieRarchY [@austr];
7. aLthougH the leptoQuARk-bilePTon modeLS Do Not cOnserve each generaTIoN Lepton number $L_\eLl$, the nEuTRiNOLesS doUble beta deCaY is foRBidden bECaUSE Of tHE conservation Of the quantuM NumBer ${\cal f}\eQuiV l + B$, wherE $B$ is tHe BAriOn number and $l=\sum_\Ell L_\ell$ is The totAL lepton NUmber. If This glObaL syMmetRY iS eXplIcITly VIoLatED in The Higgs PoTeNtial, TherE ARE ContRibUtioNs to tHe decay which dEpeNd leSS on The neUtrinO masS tHan thEy do in Too maNy Extensions of the StanDard model [@Pt93]. THe DouBlE beta DEcay wiTh MAjoRon emisSion is pOSsiBlE AS WeLl [@fpshe1];
8. There is an elEcTRIc Charge quAntizaTIoN wIThout any CoNstRainT ON the DIrac, mAjOrana or DIrac–MaJOrAnA [@EsposiTo] CharacTeR of The MassiVE neuTral feRmions [@doFf].
RepREsentation contENts are determiNEd BY EmBEddiNg tHe electric cHargE OperAtor $$\FRaC{{\caL q}}{|e|} = (\LamBda_3 + \xi\laMBdA_8 + \Zeta\Lambda_{15}) + N
\label{opcAr}$$ In the nEutraL generators $\LaMbda_{3,8,15} = \lambda^{\RM su(4)}_{3,8,15}/2$ of the laRgesT WeAK isospin group Su(4) exteNsion and $N$ iS The new U(1)$_N$ ChargE equivalEnt to the eLECtric chaRge AveRagE of THE fErmions contaiNED in eAcH flavor MulTiplet. IF we ConSidEr tHe Lightest lEptons as ThE fErMiOns Which DEtermine ThE apPrOxiMate sYMmetry, And alSo inDePeNDenT flavor GEnERAtioNs, ThEn SU(4)$\TimEs$u(1) is thE larGEst Non-symmEtric gaugE grOUp of ThE eLectrowEak sector. TherE iS no room for ThE chIral seMISimple grOup SU(5)$\times$U(1) if lepton elecTRic charGes Are onLy $0$, $\pm 1$. the weak hyPerCharge Of tHE $G_{321}$ stanDard moDel is $$\FrAc{Y}{2} = (\XI\lambdA_8 + \ZEtA\LaMbDa_{15})+N
\label{anAA | ] symmetry which can be ex tended toall s ect ors o f th e La grangian witha n in visible axion solvingthe s tr o ng-C P p roble m [@Pal 9 5] ;
5. Sp ont an e ou s CPvio lationin the ele ctr ow eak sector [ @ Ep ele95]. Th ere are several na turalso urc e s ofexp licit and s p ontane ous CP vi ol a tions[ @vicent e ] ;
6. The quark mass h i er a rchy [@austr];
7. Al t ho u g h t heleptoquark -b ilept o n model s d o n otc onserve eachgenerationl ept on num be r $ L _\ell$ , the n e utr inoless dou blebeta deca y is f o rbidden because of th e c ons erva t io nofth e qu a nt umn umb er ${\ca lF} \equi v L+ B $ , wh ere $B$ is t he barion num ber and $L= \sum_ \ellL_\e ll $ isthe to tal l ep ton number. Ifthis global s ymm et ryis expl i citlyvio lat ed in t he Higg s po te n t i al , there are contri bu t i on s to the decay wh ic h dependle sson t h e neut rino ma ss thanthey d o i ntoo man yextens io nsofthe s t anda rd mod el [@pt9 3]. T h e double betad ecay with Maj o ro n em i ssio n i s possibleas w e ll [ @fps h e1 ];8. T hereis an electric charge qua nt izatio n wit hout any cons traint ont h e Dirac,Majo r an a or Dirac–Majo rana[@Esposito ] charact er of the mas sive neut r a l fermio ns[@d off ].R ep resentation c o n tent sare det erm ined by em bed din g t he electric chargeop er at or $$ \frac { {\cal Q} }{ |e| }= ( \Lamb d a_3 +\xi\L ambd a_ 8+ \z eta\Lam b da _ { 15}) + N
\la bel {o pcar} $$ i n th e neutr al genera tor s $\L am bd a_{3,8, 15} = \lambda ^{ \rm SU(4)} _{ 3,8 ,15}/2 $ of the l argest weak isospin gro u p SU(4) ex tensi on a nd $N$ is th e newU(1 ) $_N$ c hargeequiv al ent t o the e le ctr ic charge av e r age of t he fer mions c ontained in each f l avo r multiplet.Ifwe c o n si der th e li gh t est l eptons as the f ermions wh ic h d etermine t h e a pp roximat e symme try,a nd also independ ent flavo rgene r a tio ns, then S U(4)$\ti mes$U(1)i s the la rgest no n-symm et ric gaug e grou p of theelectr ow eak se ctor. T here isno room for the chiralsemisi mplegro up SU(5)$ \ti m es$ U(1) if l epto n electric ch arg es ar e o n ly $0 $, $ \ pm 1$ . Theweak hyperchar g eoft h e$G_{321}$ s t a n dar d mod eli s $$\f rac{ Y}{2} = (\xi\Lamb d a_8 + \zeta\La mbda _ { 15} )+N \lab el {anaa | ] symmetry_which can_be extended to all_sectors of_the_Lagrangian with_an_invisible axion solving_the strong-CP problem [@Pal95];
5._ Spontaneous CP violation_in the electroweak_sector [@Epele95]._There are several natural sources of explicit and spontaneous CP violations [@vicente];
6. The quark_mass_hierarchy [@austr];
7. _Although_the_leptoquark-bilepton models do not conserve_each generation lepton number $L_\ell$,_the neutrinoless_double beta decay is forbidden because of the_conservation_of the quantum_number ${\cal F}\equiv L + B$, where $B$ is_the barion number and $L=\sum_\ell L_\ell$_is the total_lepton_number._If this global symmetry_is explicitly violated in the Higgs_potential, there are contributions to the_decay which depend less on the neutrino_mass than they do in too_many extensions of the standard_model [@pt93]. The_double beta decay with Majoron_emission is possible_as well [@fpshe1];
8._ There is_an electric charge quantization without any_constraint on the_Dirac, Majorana or Dirac–Majorana [@Esposito] character of_the_massive neutral fermions [@doff].
Representation_contents_are_determined by_embedding the electric_charge_operator $$\frac{{\cal_Q}}{|e|}_= (\Lambda_3 + \xi\Lambda_8 + \zeta\Lambda_{15})_+_N
\label{opcar}$$ in the neutral generators $\Lambda_{3,8,15} =_\lambda^{\rm SU(4)}_{3,8,15}/2$ of the_largest_weak isospin group SU(4)_extension and $N$ is the_new U(1)$_N$ charge equivalent to the_electric charge_average of_the fermions contained in each flavor multiplet. If we consider the_lightest leptons as the fermions which_determine the approximate symmetry,_and also_independent_flavor generations, then_SU(4)$\times$U(1)_is the_largest non-symmetric gauge group of the electroweak_sector. There_is no room for the chiral_semisimple group SU(5)$\times$U(1) if_lepton_electric charges are only $0$, $\pm_1$. The weak hypercharge of the_$G_{321}$ standard model is $$\frac{Y}{2}_=_(\xi\Lambda_8_+ \zeta\Lambda_{15})+N
\label{anaa |
P. F. Harrison, D. H. Perkins and W. G. Scott, Phys.Lett. B530, 167 (2002). R. G. Moorhouse, Phys. Rev. D77,053008 (2008)
---
abstract: 'We performed terahertz magneto-optical spectroscopy of FeSe thin film to elucidate the charge carrier dynamics. The measured diagonal (longitudinal) and off-diagonal (Hall) conductivity spectra are well reproduced by two-carrier Drude model, from which the carrier densities, scattering times and effective masses of electron and hole carriers are determined in a wide range of temperature. The hole density decreases below the structural transition temperature while electron density increases, which is attributed to the band structure modification in the electronic nematic phase. The scattering time of the hole carrier becomes substantially longer than that of the electron at lower temperature, which accounts for the increase of the positive dc Hall coefficient at low temperature.'
author:
- Naotaka Yoshikawa
- Masayuki Takayama
- Naoki Shikama
- Tomoya Ishikawa
- Fuyuki Nabeshima
- Atsutaka Maeda
- Ryo Shimano
bibliography:
-'refs\_FeSeFaraday.bib'
title: |
Charge carrier dynamics of FeSe thin film investigated by\
terahertz magneto-optical spectroscopy
---
=1
Since the discovery of iron-based superconductors (FeSCs), tremendous research efforts have been devoted to reveal the pairing mechanism of superconductivity. The elucidation of interplay between the nematic order, antiferromagnetic spin order, and superconductivity emergent in FeSCs has been believed to provide a clue to understand the emergent superconductivity. Among FeSCs, FeSe provides a unique playground to study the role of nematicity, because it lacks the long-range magnetic order in the nematic phase that appears below the tetragonal-orthorhombic structural transition temperature $T_{\mathrm{s}}\simeq \SI{90}{\kelvin}$, as evidenced by a significant electronic anisotropy from transport and nuclear magnetic resonance (NMR) spectral properties[@McQueen:2009hs; @Baek:2014gs; @Bohmer:2015fk]. While the superconducting transition temperature $T_{\mathrm{c}}$ of bulk FeSe is $\sim\SI{9}{K}$ at ambient pressure[@Hsu:2008ep], it shows a remarkable tunability. $ | P. F. Harrison, D. H. Perkins and W. G. Scott, Phys. Lett. B530, 167 (2002). R. G. Moorhouse, Phys. Rev. D77,053008 (2008)
---
abstract:' We performed terahertz magneto - optical spectroscopy of FeSe thin film to clarify the cathexis carrier dynamics. The measured aslant (longitudinal) and off - diagonal (Hall) conductivity spectrum are well reproduced by two - carrier Drude exemplar, from which the carrier densities, scatter times and effective masses of electron and trap mailman are determined in a wide range of temperature. The hole density decrease below the structural transition temperature while electron density increase, which is attributed to the band structure modification in the electronic nematic phase. The disperse time of the hole carrier becomes substantially long than that of the electron at lower temperature, which accounts for the increase of the positive dc Hall coefficient at low temperature.'
writer:
- Naotaka Yoshikawa
- Masayuki Takayama
- Naoki Shikama
- Tomoya Ishikawa
- Fuyuki Nabeshima
- Atsutaka Maeda
- Ryo Shimano
bibliography:
-'refs\_FeSeFaraday.bib'
title: |
Charge carrier dynamics of FeSe thin film investigated by\
terahertz magneto - optical spectroscopy
---
= 1
Since the discovery of iron - based superconductors (FeSCs), tremendous inquiry attempt have been devoted to unwrap the pair mechanism of superconductivity. The elucidation of interplay between the nematic order, antiferromagnetic spin orderliness, and superconductivity emergent in FeSCs has been believe to provide a clue to sympathize the emergent superconductivity. Among FeSCs, FeSe provides a unique playground to study the role of nematicity, because it lacks the farseeing - range magnetic order in the nematic phase that appears below the tetragonal - orthorhombic geomorphologic transition temperature $ T_{\mathrm{s}}\simeq \SI{90}{\kelvin}$, as attest by a significant electronic anisotropy from transport and nuclear magnetic resonance (NMR) spectral properties[@McQueen:2009hs; @Baek:2014gs; @Bohmer:2015fk ]. While the superconducting transition temperature $ T_{\mathrm{c}}$ of bulk FeSe is $ \sim\SI{9}{K}$ at ambient pressure[@Hsu:2008ep ], it indicate a remarkable tunability. $ | P. V. Harrison, D. H. Perkins akd W. G. Scott, Phys.Lett. B530, 167 (2002). R. G. Moorhoure, Phys. Rev. D77,053008 (2008)
---
abstract: 'We pxrfoemed uvrahertz magneto-opticxl spectrlscopy od FeWw thin film to elugndate bhe ckacge carrier dynsmics. The keasured diagotau (pongitudinal) and off-diagonal (Hall) cjnductifihy spectra are welk repdoduced by two-carrier Drude model, rrom whpch the carrier drnsities, scattering times wnd fffective masses ov electron qnd rile carriers are determined in a wjde range of temperature. The houe deusity decreqsws hglow the strncturaj transition temperadure whole electron dcnsitb inxreases, which is attrmbuted to the band sjructure mmdnfication in the elecrrinic temadic oyasd. Tge sdatterlng time of tge hole careier becomes substamtyqlly longer tgan thwt of the electron at lower temperature, wvicg accounts for the incrwase of the positive fc Hall cjefficient at low temperature.'
author:
- Naotaka Yoshinawa
- Jxsavmhk Twkayama
- Naoki Shikama
- Tomoya Ishikawa
- Fuyuki Gzbtshpma
- Atsutaka Maedc
- Ryo Shimano
bibkilgtwphy:
-'refs\_FeSeFxraday.yjb'
fitle: |
Charge cwrrier qynamucs of FeFe tnin film investigated by\
terahertz mcgnwto-optical spectrodcopy
---
=1
Since che dixcovety of iron-based supercouductods (FeSCs), trfmendous ddsearch efforts favv bean devotta to reveal the pwiring merhanixm of sjpervonduceivity. The eluclgation of interplaj betceen dhe nematif order, antiferromagnetic spin order, and supetcotdubtivity eiergekt in FeSCs has been believed to proride a clue to uhderstaid the emergqnt supercondgftivity. Amonj FeSCs, FqSe provudes a jvique playgroumd to stubv the rolw of nematicity, begause jt lacks the louy-rqnge magnetic otdef ig nhe nqkatic phase dhat aporars celow the tctrxgonsl-orthorhombic strucduram transition tempetabure $T_{\matyrm{s}}\simez \SI{90}{\kelvin}$, as evidenced by a sihnifirant euectrpnis anisotropy from transport ans nuclear mannetic resonagce (KMR) fpectral pxoperties[@McQueen:2009hs; @Baek:2014gs; @Bohmer:2015fk]. While the superconductibg transition tempetatmre $T_{\mathrm{c}}$ of btlk FeSe hs $\sim\SI{9}{K}$ at ambient pressure[@Hsu:2008ep], it shows a remarkable thnabilhty. $ | P. F. Harrison, D. H. Perkins and Scott, B530, 167 R. G. Moorhouse, abstract: performed terahertz magneto-optical of FeSe thin to elucidate the charge carrier dynamics. measured diagonal (longitudinal) and off-diagonal (Hall) conductivity spectra are well reproduced by two-carrier model, from which the carrier densities, scattering times and effective masses of electron hole are in wide range of temperature. The hole density decreases below the structural transition temperature while electron density which is attributed to the band structure modification the electronic nematic phase. scattering time of the hole becomes longer than of electron lower temperature, which for the increase of the positive dc Hall coefficient at low temperature.' author: - Naotaka Yoshikawa - Takayama - - Tomoya - Nabeshima Atsutaka Maeda - bibliography: -'refs\_FeSeFaraday.bib' title: | Charge carrier thin film investigated by\ terahertz magneto-optical spectroscopy --- Since the of iron-based superconductors (FeSCs), tremendous research have been devoted to reveal the pairing mechanism superconductivity. The elucidation of interplay between the nematic order, antiferromagnetic spin order, and superconductivity emergent has been believed to a clue to the superconductivity. FeSCs, provides a playground to study the role of nematicity, because it lacks the magnetic order in the nematic phase that appears below the transition $T_{\mathrm{s}}\simeq \SI{90}{\kelvin}$, as by a significant electronic from and nuclear magnetic resonance properties[@McQueen:2009hs; @Bohmer:2015fk]. transition $T_{\mathrm{c}}$ bulk FeSe is $\sim\SI{9}{K}$ ambient pressure[@Hsu:2008ep], it shows a tunability. $ | P. F. Harrison, D. H. Perkins and W. G. SCott, Phys.LeTt. B530, 167 (2002). R. G. mooRhoUsE, PhyS. Rev. d77,053008 (2008)
---
abstract: 'We perFOrmeD terahertz magneto-opticAl speCtROscoPY oF FeSe Thin filM To ELUciDaTe The ChARgE carrIer DynamicS. The measurEd dIaGonal (longituDInAl) and off-diAgoNal (Hall) conduCtiVity spEcTra ARe welL reProduCed by tWO-carriEr Drude moDeL, From whICh the caRRIeR denSities, scattering tIMeS And effective maSses of ElECtRON anD hoLe carriers ArE deteRMined in A WiDE RAngE Of temperature. the hole densITy dEcreasEs BelOW the stRuctuRaL TraNsition tempEratUre while eLectroN Density INcreaseS, which Is aTtrIbutED tO tHe bAnD StrUCtUre MOdiFication In ThE elecTronIC NEMatiC phAse. THe scaTtering time of The Hole CArrIer beComes SubsTaNtialLy longEr thaN tHat of the electroN at lOwer tempeRatUrE, whIcH accoUNts for The IncRease of The posiTIve Dc hALL cOefficient at low temPeRATuRe.'
author:
- naotakA yoShIKawa
- MasaYuKi TAkayAMA
- NaokI ShiKAmA
- Tomoya IShikawA
- fuYuKi NabesHiMa
- AtsuTaKa MAedA
- Ryo SHImanO
bibliOgraphy:
-'rEfs\_FesEFaraday.bib'
titLE: |
Charge carrieR DyNAMiCS of FESe Thin film invEstiGAted By\
teRAhErtZ MagneTo-optIcAL sPEctroscopy
---
=1
Since the dIsCovery Of iroN-based supercoNductors (FescS), TremendoUs reSEaRCh efforts have bEen deVoted to revEAl the paiRing mEchanism Of supercoNDUctivity. the EluCidAtiON Of Interplay betwEEN the NeMatic orDer, AntiferRomAgnEtiC spIn Order, and sUpercondUcTiViTy EmeRgent IN FeSCs haS bEen BeLieVed to PRovide A clue To unDeRsTAnd The emerGEnT SUperCoNdUctiVitY. AMong FESCs, fESe ProvideS a unique pLayGRounD tO sTudy the Role of nematicItY, because it LaCks The lonG-RAnge magnEtic order in the nematic phASe that aPpeArs beLow tHe tetragoNal-OrthorHomBIc struCtural TransItIon TEMperaTURe $t_{\maThRm{s}}\simeq \SI{90}{\KELviN}$, as evIdEnceD by a sigNificant electronic ANisOtropy from traNspOrt aND NuCleAR mAGneTiC ResONAnce (NMR) spectral Properties[@mcqUeEn:2009hs; @Baek:2014gs; @bOhmEr:2015Fk]. While The supeRcondUCting trAnsition tEmperaturE $T_{\MathRM{C}}$ of Bulk FeSe is $\Sim\SI{9}{K}$ at Ambient prESsure[@hSu:2008Ep], it sHowS a remaRkAblE tunaBility. $ | P. F. Harrison, D. H. Per kins and W . G.Sco tt, P hys. Lett . B530, 167 (2 0 02). R. G. Moorhouse, Phys . Rev .D 77,0 5 30 08 (2 008)
- - -a b str ac t: 'W ep er forme d t erahert z magneto- opt ic al spectrosc o py of FeSe t hin film to elu cid ate th echa r ge ca rri er dy namics . The m easured d ia g onal ( l ongitud i n al ) an d off-diagonal (H a ll ) conductivityspectr aa re w ell re produced b ytwo-c a rrier D r ud e m ode l , from whichthe carrier den sities ,sca t tering time sa ndeffective m asse s of elec tron a n d holec arriers are d ete rmi nedi nawid er ang e o f t e mpe rature.Th eholedens i t y decr eas es b elowthe structura l t rans i tio n tem perat urewh ile e lectro n den si ty increases, w hich is attri but ed to t he ba n d stru ctu remodific ation i n th ee l e ct ronic nematic phas e. T he scatter ing ti m eof the hole c arr ierb e comes sub s ta ntiallylonger th an that o fthe el ec tro n a t low e r te mperat ure, whi ch ac c ounts for thei ncrease of th e p o s it i ve d c H all coeffic ient at l ow t e mp era t ure.'
auth or :
- Naotaka Yoshikawa
- M asayuk i Tak ayama
- Naoki Shikama
- T o moya Ish ikaw a
- Fuyuki Nabeshi ma
-Atsutaka M a eda
- Ry o Shi mano
bib liography : -'refs\_ FeS eFa rad ay. b i b'
title: |
C harg ecarrier dy namicsofFeS e t hin f ilm inves tigatedby \ ter ahert z magneto -o pti ca l s pectr o scopy---
=1
Si nc e th e disco v er y of i ro n- base d s up ercon duct o rs(FeSCs) , tremend ous rese ar ch effort s have been d ev oted to re ve althe pa i r ing mech anism of superconductiv i ty. The el ucida tion of inter pla y betw een the ne maticorder ,ant i f errom a g ne tic s pin order, a ndsuper co nduc tivityemergent in FeSCsh asbeen believed to pro v i de ac lu e to u n der s t and the emergen t supercon du c ti vity. Amon g Fe SC s, FeSe provid es au nique p layground to study t he r o l e o f nematici ty, beca use it la c ks th e l ong-r ang e magn et icorder in th e ne matic phase t hat ap pears b elow the tetragonal-orthorhombi c stru ctura l t ransition te m per ature $T_ {\ma thrm{s}}\s ime q \ SI{90 }{\ k elvin }$,a sevi d enced bya signific a nt el e c tr onic anisot r o p y f rom t ran s port a nd n uclear magnetic r e sonance (NMR)spec t r alpro p erti es [@McQueen:2009 hs; @ B a ek:2014g s; @Bohmer:20 15fk]. W hi l e the super conduc ting tr a n si t ion te mper atu re $T_{\m ath rm { c}}$ of b ul k FeSeis $ \s im\SI{ 9}{K}$ at a m b ient pressure[@H su:20 0 8 ep],i t s howsaremarka b le t unability. $ | P._F. Harrison,_D. H. Perkins and_W. G._Scott,_Phys.Lett. B530,_167_(2002). R. G._Moorhouse, Phys. Rev._D77,053008 (2008)
---
abstract: 'We_performed terahertz magneto-optical_spectroscopy_of FeSe thin film to elucidate the charge carrier dynamics. The measured diagonal (longitudinal)_and_off-diagonal (Hall)_conductivity_spectra_are well reproduced by two-carrier_Drude model, from which the_carrier densities,_scattering times and effective masses of electron and_hole_carriers are determined_in a wide range of temperature. The hole density_decreases below the structural transition temperature_while electron density_increases,_which_is attributed to the_band structure modification in the electronic_nematic phase. The scattering time of_the hole carrier becomes substantially longer than_that of the electron at lower_temperature, which accounts for the_increase of_the positive dc Hall coefficient_at low temperature.'
author:
-_Naotaka Yoshikawa
-_Masayuki Takayama
- Naoki_Shikama
- Tomoya Ishikawa
- Fuyuki Nabeshima
- Atsutaka_Maeda
- Ryo Shimano
bibliography:
-'refs\_FeSeFaraday.bib'
title:_|
Charge carrier_dynamics_of FeSe thin_film_investigated_by\
_ terahertz_magneto-optical_spectroscopy
---
=1
Since the_discovery_of iron-based superconductors (FeSCs), tremendous research_efforts_have been devoted to reveal the pairing_mechanism of superconductivity. The_elucidation_of interplay between the_nematic order, antiferromagnetic spin order,_and superconductivity emergent in FeSCs has_been believed_to provide_a clue to understand the emergent superconductivity. Among FeSCs, FeSe provides_a unique playground to study the_role of nematicity, because_it lacks_the_long-range magnetic order_in_the nematic_phase that appears below the tetragonal-orthorhombic structural_transition temperature_$T_{\mathrm{s}}\simeq \SI{90}{\kelvin}$, as evidenced by a_significant electronic anisotropy from_transport_and nuclear magnetic resonance (NMR) spectral_properties[@McQueen:2009hs; @Baek:2014gs; @Bohmer:2015fk]. While the superconducting_transition temperature $T_{\mathrm{c}}$ of bulk_FeSe_is_$\sim\SI{9}{K}$ at ambient pressure[@Hsu:2008ep], it_shows a remarkable tunability. $ |
{T}^H_{_m}$. Furthermore, because the complex codimension of $\mathcal{Z}^H_{m}\setminus \mathcal{Z}_{m}$ is as least one in $\mathcal{Z}^H_{_m}$, the complex codimension of $\mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is also as least one in $\mathcal{T}^H_{_m}$.
First, it is not hard to see that the restriction map $\Phi_m$ is holomorphic. Indeed, we know that $i_m:\,\mathcal{T}\rightarrow \mathcal{T}_m$ is the lifting of $i\circ \pi_m$ and $\pi^H_{_m}|_{\mathcal{T}_m}:\, \mathcal{T}_m\rightarrow \mathcal{Z}_m$ is a holomorphic covering map, thus $i_m$ is also holomorphic. Since $\Phi=\Phi_m\circ i_m$ with both $\Phi$, $i_m$ holomorphic and $i_m$ locally invertible, we can conclude that $\Phi_m:\,\mathcal{T}_m\rightarrow D$ is a holomorphic map. Moreover, we have $\Phi_m(\mathcal{T}_m)=\Phi_m(i_m(\mathcal{T}))=\Phi(\mathcal{T})$ as $\Phi=i_m\circ \Phi_m$. In particular, as $\Phi: \,\mathcal{T}\rightarrow N_+\cap D$ is bounded, we get that $\Phi_m: \,\mathcal{T}\rightarrow N_+\cap D$ is also bounded in $N_+$ with the Euclidean metric. Thus $\Phi^H_{_m}$ is also bounded. Therefore applying Riemann extension theorem, we get
\[Riemannextension\]The map $\Phi^{H}_{_m}$ is a bounded holomorphic map from $\mathcal{T}^H_{_m}$ to $N_+\cap D$.
According to the above discussion, we know that the complement $\mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is the pre-image of $\mathcal{Z}^H_{m}\setminus \mathcal{Z}_{m}$ of the covering map $\pi_m^H$. So $\mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is a analytic subvariety of $\mathcal{T | { T}^H_{_m}$. Furthermore, because the complex codimension of $ \mathcal{Z}^H_{m}\setminus \mathcal{Z}_{m}$ is as least one in $ \mathcal{Z}^H_{_m}$, the complex codimension of $ \mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is also as least one in $ \mathcal{T}^H_{_m}$.
First, it is not unvoiced to examine that the restriction map $ \Phi_m$ is holomorphic. Indeed, we know that $ i_m:\,\mathcal{T}\rightarrow \mathcal{T}_m$ is the lifting of $ i\circ \pi_m$ and $ \pi^H_{_m}|_{\mathcal{T}_m}:\, \mathcal{T}_m\rightarrow \mathcal{Z}_m$ is a holomorphic application map, thus $ i_m$ is besides holomorphic. Since $ \Phi=\Phi_m\circ i_m$ with both $ \Phi$, $ i_m$ holomorphic and $ i_m$ locally invertible, we can conclude that $ \Phi_m:\,\mathcal{T}_m\rightarrow D$ is a holomorphic map. Moreover, we suffer $ \Phi_m(\mathcal{T}_m)=\Phi_m(i_m(\mathcal{T}))=\Phi(\mathcal{T})$ as $ \Phi = i_m\circ \Phi_m$. In particular, as $ \Phi: \,\mathcal{T}\rightarrow N_+\cap D$ is bounded, we catch that $ \Phi_m: \,\mathcal{T}\rightarrow N_+\cap D$ is also bound in $ N_+$ with the euclidian metric. Thus $ \Phi^H_{_m}$ is besides bounded. consequently put on Riemann extension theorem, we get
\[Riemannextension\]The map $ \Phi^{H}_{_m}$ is a bound holomorphic map from $ \mathcal{T}^H_{_m}$ to $ N_+\cap D$.
According to the above discussion, we acknowledge that the complement $ \mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is the pre - image of $ \mathcal{Z}^H_{m}\setminus \mathcal{Z}_{m}$ of the cover map $ \pi_m^H$. So $ \mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is a analytic subvariety of $ \mathcal{T | {T}^H_{_m}$. Furthermore, because the complex codimeuwion oh $\mathczl{Z}^H_{m}\setoinus \mathcal{Z}_{m}$ is as least lnw in $\nathcal{Z}^H_{_m}$, the complex codimenspon of $\marhcao{R}^H_{_m}\backslash\mathcal{T}_m$ is wlso es least one in $\mathcal{T}^H_{_k}$.
First, it is nmt hcrd to see that the restriction map $\[hi_m$ is hllomorphic. Indged, wt kgow fhat $i_m:\,\mathcal{T}\rightarrow \mathcal{T}_j$ is tht lifting of $i\circ \pi_m$ and $\pi^H_{_m}|_{\mathcal{T}_m}:\, \matjcal{H}_m\rightarrow \mathcwl{Z}_m$ is a hilomjephic coverivg map, thus $i_m$ is also holomorphic. Since $\Phi=\Phi_m\circ k_m$ wich both $\Phi$, $i_n$ hlnomorphic aid $i_m$ jocally invevnible, wa can cpnclude that $\Pmi_m:\,\mavhcao{T}_m\rightarrow D$ is a iolomorphic map. Morejver, we hdvz $\Phi_m(\mathcal{T}_m)=\Phi_m(i_m(\narhcal{J}))=\Phi(\mdthcxo{T})$ xs $\Khi=m_m\cjrc \Phl_m$. Mn particulzr, as $\Phi: \,\mqthcal{T}\rightarrow N_+\va[ D$ is bounded, we gee ehat $\Phi_m: \,\mathcal{T}\rightarrow N_+\cap D$ is dlsk bounded in $N_+$ with the Euclidean metric. Thud $\Phi^H_{_m}$ if also bounded. Therefore applying Riemann extensimn thxofem, wc geg
\[Eifmannextension\]The map $\Phi^{H}_{_m}$ is a bounded holoikrkhib map from $\mathcaj{T}^H_{_m}$ to $N_+\cal F$.
Avsording to thg above diacussion, we know tjat the compoement $\mauhcal{Y}^H_{_m}\backslash\mathcal{T}_m$ is thw pre-image oy $\mqthcal{Z}^H_{m}\setminus \lathcal{Z}_{m}$ oy the voverong map $\pi_m^H$. So $\mathcal{C}^H_{_m}\bacislash\mathcwl{T}_m$ is a xnalytic subvaridty ox $\mathcal{T | {T}^H_{_m}$. Furthermore, because the complex codimension of is least one $\mathcal{Z}^H_{_m}$, the complex as one in $\mathcal{T}^H_{_m}$. it is not to see that the restriction map is holomorphic. Indeed, we know that $i_m:\,\mathcal{T}\rightarrow \mathcal{T}_m$ is the lifting of $i\circ and $\pi^H_{_m}|_{\mathcal{T}_m}:\, \mathcal{T}_m\rightarrow \mathcal{Z}_m$ is a holomorphic covering map, thus $i_m$ is also Since i_m$ both $i_m$ holomorphic and $i_m$ locally invertible, we can conclude that $\Phi_m:\,\mathcal{T}_m\rightarrow D$ is a holomorphic map. we have $\Phi_m(\mathcal{T}_m)=\Phi_m(i_m(\mathcal{T}))=\Phi(\mathcal{T})$ as $\Phi=i_m\circ \Phi_m$. In particular, $\Phi: \,\mathcal{T}\rightarrow N_+\cap D$ bounded, we get that $\Phi_m: N_+\cap is also in with Euclidean metric. Thus is also bounded. Therefore applying Riemann extension theorem, we get \[Riemannextension\]The map $\Phi^{H}_{_m}$ is a bounded holomorphic from $\mathcal{T}^H_{_m}$ D$. According the discussion, know that the is the pre-image of $\mathcal{Z}^H_{m}\setminus \mathcal{Z}_{m}$ map $\pi_m^H$. So $\mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is a analytic subvariety $\mathcal{T | {T}^H_{_m}$. Furthermore, because the cOmplex codiMensiOn oF $\maThCal{Z}^h_{m}\seTminus \mathcal{Z}_{M}$ Is as Least one in $\mathcal{Z}^H_{_m}$, thE compLeX CodiMEnSion oF $\mathcaL{t}^H_{_M}\BAckSlAsH\maThCAl{t}_m$ is aLso As least One in $\mathcAl{T}^h_{_m}$.
first, it is not HArD to see that The Restriction mAp $\PHi_m$ is hOlOmoRPhic. INdeEd, we kNow thaT $I_m:\,\mathCal{T}\rightArROw \mathCAl{T}_m$ is tHE LiFtinG of $i\circ \pi_m$ and $\pi^H_{_M}|_{\MaTHcal{T}_m}:\, \mathcal{T}_M\rightArROw \MAThcAl{Z}_M$ is a holomoRpHic coVEring maP, ThUS $I_M$ is ALso holomorphiC. Since $\Phi=\PhI_M\ciRc i_m$ wiTh BotH $\phi$, $i_m$ hOlomoRpHIc aNd $i_m$ locally InveRtible, we cAn concLUde that $\pHi_m:\,\mathCal{T}_m\rIghTarRow D$ IS a HoLomOrPHic MAp. morEOveR, we have $\PHi_M(\mAthcaL{T}_m)=\PHI_M(I_M(\matHcaL{T}))=\PhI(\mathCal{T})$ as $\Phi=i_m\ciRc \PHi_m$. IN ParTiculAr, as $\PHi: \,\maThCal{T}\rIghtarRow N_+\cAp d$ is bounded, we get That $\phi_m: \,\mathcAl{T}\RiGhtArRow N_+\cAP D$ is alSo bOunDed in $N_+$ wIth the EUCliDeAN MEtRic. Thus $\Phi^H_{_m}$ is also BoUNDeD. TherefoRe applYInG RIEmann extEnSioN theOREm, we gEt
\[RiEMaNnextensIon\]The MAp $\phI^{H}_{_m}$ is a bOuNded hoLoMorPhiC map fROm $\maThcal{T}^h_{_m}$ to $N_+\cap d$.
AccoRDing to the above DIscussion, we knOW tHAT tHE comPleMent $\mathcal{t}^H_{_m}\bACkslAsh\mAThCal{t}_M$ is thE pre-iMaGE oF $\Mathcal{Z}^H_{m}\setminus \mAtHcal{Z}_{m}$ Of the Covering map $\pi_M^H$. So $\mathcaL{t}^h_{_M}\backslaSh\maTHcAL{T}_m$ is a analytic SubvaRiety of $\matHCal{T | {T}^H_{_m}$. Furthermore,because th e com ple x c od imen sion of $\mathcal{ Z }^H_ {m}\setminus \mathcal{ Z}_{m }$ is a s l eastone in$ \m a t hca l{ Z} ^H_ {_ m }$ , the co mplex c odimension of $ \mathcal{T}^ H _{ _m}\backsl ash \mathcal{T}_ m$is als oasl eastone in $ \mathc a l{T}^H _{_m}$.
Fi r st, it is noth a rd tosee that the rest r ic t ion map $\Phi_ m$ isho l om o r phi c.Indeed, we k now t h at $i_m : \, \ m a thc a l{T}\rightarr ow \mathcal { T}_ m$ isth e l i ftingof $i \c i rc\pi_m$ and$\pi ^H_{_m}|_ {\math c al{T}_m } :\, \ma thcal{ T}_ m\r ight a rr ow \m at h cal { Z} _m$ isa holomo rp hi c cov erin g m a p, t hus $i_ m$ is also holomor phi c. S i nce $\Ph i=\Ph i_m\ ci rc i_ m$ wit h bot h$\Phi$, $i_m$ h olom orphic an d $ i_ m$lo cally invert ibl e,we canconclud e th at $ \ Ph i_m:\,\mathcal{T}_ m\ r i gh tarrow D $ is a ho lo m orphic m ap . M oreo v e r, we hav e $ \Phi_m(\ mathca l {T }_ m)=\Phi _m (i_m(\ ma thc al{ T}))= \ Phi( \mathc al{T})$as $\ P hi=i_m\circ \P h i_m$. In part i cu l a r, as $ \Ph i: \,\mathc al{T } \rig htar r ow N_ + \capD$ is b o un d ed, we get that $\P hi _m: \, \math cal{T}\righta rrow N_+\c a p D$ is al so b o un d ed in $N_+$ wi th th e Euclidea n metric. Thus $\Phi^H _{_m}$ is a lso boun ded . T her efo r e a pplying Riema n n ext en sion th eor em, weget
\ [Ri ema nn extension \]The ma p$\ Ph i^ {H} _{_m} $ is a bo un ded h olo morph i c mapfrom$\ma th ca l {T} ^H_{_m} $ t o $N_+ \c ap D$.
A cc ordin g to the abovediscussio n,w e kn ow t hat the complement $ \m athcal{T}^ H_ {_m }\back s l ash\math cal{T}_m$ is the pre-im a ge of $ \ma thcal {Z}^ H_{m}\set min us \ma thc a l{Z}_{ m}$ of theco ver i n g map $ \p i_m ^H $. So $\ma t h cal {T}^H _{ _m}\ backsla sh\mathcal{T}_m$ i s aanalytic subv ari etyo f $ \ma t hc a l{T | {T}^H_{_m}$. Furthermore,_because the_complex codimension of $\mathcal{Z}^H_{m}\setminus_\mathcal{Z}_{m}$ is_as_least one_in_$\mathcal{Z}^H_{_m}$, the complex_codimension of $\mathcal{T}^H_{_m}\backslash\mathcal{T}_m$_is also as least_one in $\mathcal{T}^H_{_m}$.
First,_it_is not hard to see that the restriction map $\Phi_m$ is holomorphic. Indeed, we_know_that $i_m:\,\mathcal{T}\rightarrow_\mathcal{T}_m$_is_the lifting of $i\circ \pi_m$_and $\pi^H_{_m}|_{\mathcal{T}_m}:\, \mathcal{T}_m\rightarrow \mathcal{Z}_m$ is_a holomorphic_covering map, thus $i_m$ is also holomorphic. Since_$\Phi=\Phi_m\circ_i_m$ with both_$\Phi$, $i_m$ holomorphic and $i_m$ locally invertible, we can_conclude that $\Phi_m:\,\mathcal{T}_m\rightarrow D$ is a_holomorphic map. Moreover,_we_have_$\Phi_m(\mathcal{T}_m)=\Phi_m(i_m(\mathcal{T}))=\Phi(\mathcal{T})$ as $\Phi=i_m\circ \Phi_m$._In particular, as $\Phi: \,\mathcal{T}\rightarrow N_+\cap_D$ is bounded, we get that_$\Phi_m: \,\mathcal{T}\rightarrow N_+\cap D$ is also bounded_in $N_+$ with the Euclidean metric._Thus $\Phi^H_{_m}$ is also bounded._Therefore applying_Riemann extension theorem, we get
\[Riemannextension\]The_map $\Phi^{H}_{_m}$ is_a bounded_holomorphic map from_$\mathcal{T}^H_{_m}$ to $N_+\cap D$.
According to the_above discussion, we_know that the complement $\mathcal{T}^H_{_m}\backslash\mathcal{T}_m$ is_the_pre-image of $\mathcal{Z}^H_{m}\setminus_\mathcal{Z}_{m}$_of_the covering_map $\pi_m^H$. So_$\mathcal{T}^H_{_m}\backslash\mathcal{T}_m$_is a_analytic_subvariety of $\mathcal{T |
-qc\]]{}]{}. G. Compere, “[Symmetries and conservation laws in Lagrangian gauge theories with applications to the mechanics of black holes and to gravity in three dimensions]{},” [[ arXiv:0708.3153 \[hep-th\]]{}]{}.
R. M. Wald, “Black Hole Entropy is the Noether Charge,” [[ Phys. Rev.]{} [**D48**]{} (1993) 3427]{}, [[ \[arXiv:gr-qc/9307038\]]{}]{}.
V. Iyer and R. M. Wald, “Some Properties of Noether Charge and a Proposal for Dynamical Black Hole Entropy,” [[ Phys. Rev.]{} [**D50**]{} (1994) 846]{}, [[ \[arXiv:gr-qc/9403028\]]{}]{}.
T. Jacobson, G. Kang and R. C. Myers, “On Black Hole Entropy,” [[ Phys. Rev.]{} [**D49**]{} (1994) 6587]{}, [[ \[arXiv:gr-qc/9312023\]]{}]{}.
J. C. Breckenridge, R. C. Myers, A. W. Peet and C. Vafa, “D-branes and spinning black holes,” [[ Phys. Lett.]{} [**B391** ]{} (1997) 93-98]{}, [[ \[hep-th/9602065\]]{}]{}.
R. Kallosh, A. Rajaraman and W. K. Wong, “Supersymmetric rotating black holes and attractors,” [[ Phys. Rev.]{} [**D55** ]{} (1997) 3246-3249]{}, [[ \[hep-th/9611094\]]{}]{}.
R. C. Myers and M. J. Perry, “Black Holes in Higher Dimensional Space-Times,” [[ Annals of Physics]{} [**172** ]{} (1986) 304]{}.
D. Gaiotto, A. Strominger and X. Yin, “New connections between 4-D and 5-D black holes,” JHEP [**0602**]{} (2006) 024 \[hep-th/0503217\].
J. P. Gauntlett, R. C. Myers and P. K. Townsend, “Black holes of D = 5 supergravity,” Class. Quant. Grav. [** | -qc\ ] ] { } ] { }. G. Compere, “ [ Symmetries and conservation laws in Lagrangian gauge theories with application to the automobile mechanic of black holes and to gravity in three dimensions ] { }, ” [ [ arXiv:0708.3153 \[hep - th\ ] ] { } ] { }.
R. M. Wald, “ Black Hole Entropy is the Noether Charge, ” [ [ Phys. Rev. ] { } [ * * D48 * * ] { } (1993) 3427 ] { }, [ [ \[arXiv: gr - qc/9307038\ ] ] { } ] { }.
V. Iyer and R. M. Wald, “ Some Properties of Noether Charge and a Proposal for Dynamical Black Hole Entropy, ” [ [ Phys. Rev. ] { } [ * * D50 * * ] { } (1994) 846 ] { }, [ [ \[arXiv: gr - qc/9403028\ ] ] { } ] { }.
T. Jacobson, G. Kang and R. C. Myers, “ On Black Hole Entropy, ” [ [ Phys. Rev. ] { } [ * * D49 * * ] { } (1994) 6587 ] { }, [ [ \[arXiv: gr - qc/9312023\ ] ] { } ] { }.
J. C. Breckenridge, R. C. Myers, A. W. Peet and C. Vafa, “ five hundred - branes and spinning black hole, ” [ [ Phys. Lett. ] { } [ * * B391 * * ] { } (1997) 93 - 98 ] { }, [ [ \[hep - th/9602065\ ] ] { } ] { }.
R. Kallosh, A. Rajaraman and W. K. Wong, “ Supersymmetric rotate black hole and attractors, ” [ [ Phys. Rev. ] { } [ * * D55 * * ] { } (1997) 3246 - 3249 ] { }, [ [ \[hep - th/9611094\ ] ] { } ] { }.
R. C. Myers and M. J. Perry, “ Black Holes in Higher Dimensional Space - Times, ” [ [ Annals of Physics ] { } [ * * 172 * * ] { } (1986) 304 ] { }.
D. Gaiotto, A. Strominger and X. Yin, “ modern connections between 4 - D and 5 - five hundred black holes, ” JHEP [ * * 0602 * * ] { } (2006) 024 \[hep - th/0503217\ ].
J. P. Gauntlett, R. C. Myers and P. K. Townsend, “ Black holes of D = 5 supergravity, ” Class. Quant. Grav. [ * * | -qc\]]{}]{}. H. Compere, “[Symmetries and gonservation laws in Lajrangiah gauge gheories with applications tl rhe mtbhanics of black holer and to hravity un tiree dimensions]{},” [[ arXiv:0708.3153 \[hc'-th\]]{}]{}.
R. J. Walb, “Ulack Hole Entrppy is the Noether Charga,” [[ Pkys. Rev.]{} [**D48**]{} (1993) 3427]{}, [[ \[arXiv:gr-qc/9307038\]]{}]{}.
V. Iyer and R. M. Wald, “Spmf Properties os Notthqr Cgarge and a Proposal for Dynamical Black Iole Entropy,” [[ Pnys. Rev.]{} [**D50**]{} (1994) 846]{}, [[ \[arXiv:gr-qc/9403028\]]{}]{}.
T. Jafobsln, G. Kang and R. C. Lyers, “On Blqck Rile Entropy,” [[ Phys. Rev.]{} [**D49**]{} (1994) 6587]{}, [[ \[arXiv:gr-sc/9312023\]]{}]{}.
J. C. Breckenridge, R. C. Myers, A. W. Peet and E. Vafa, “D-brangs ajg spinning ulack roles,” [[ Phys. Lcnt.]{} [**B391** ]{} (1997) 93-98]{}, [[ \[hep-th/9602065\]]{}]{}.
T. Kallosh, A. Rajavaman ane W. K. Wong, “Supersymmetrmc rotating black hojes and adtxactors,” [[ Phys. Rev.]{} [**D55** ]{} (1997) 3246-3249]{}, [[ \[hep-tv/9611094\]]{}]{}.
R. C. Mfers qnd M. J. Kercy, “Glack Jolxs in Highed Dimensionql Space-Times,” [[ Annaks if Physics]{} [**172** ]{} (1986) 304]{}.
D. Gaioetj, A. Strominger and X. Yin, “New connections tetseen 4-D and 5-D black holew,” JHEP [**0602**]{} (2006) 024 \[hep-th/0503217\].
J. P. Gaunjlett, R. C. Myqrs and P. K. Townsend, “Black holes of D = 5 supergravitf,” Claar. Quckb. Grax. [** | -qc\]]{}]{}. G. Compere, “[Symmetries and conservation laws gauge with applications the mechanics of in dimensions]{},” [[ arXiv:0708.3153 R. M. Wald, Hole Entropy is the Noether Charge,” Phys. Rev.]{} [**D48**]{} (1993) 3427]{}, [[ \[arXiv:gr-qc/9307038\]]{}]{}. V. Iyer and R. M. Wald, Properties of Noether Charge and a Proposal for Dynamical Black Hole Entropy,” [[ Rev.]{} (1994) [[ T. Jacobson, G. Kang and R. C. Myers, “On Black Hole Entropy,” [[ Phys. Rev.]{} [**D49**]{} 6587]{}, [[ \[arXiv:gr-qc/9312023\]]{}]{}. J. C. Breckenridge, R. C. A. W. Peet and Vafa, “D-branes and spinning black [[ Lett.]{} [**B391** (1997) [[ R. Kallosh, A. and W. K. Wong, “Supersymmetric rotating black holes and attractors,” [[ Phys. Rev.]{} [**D55** ]{} (1997) 3246-3249]{}, \[hep-th/9611094\]]{}]{}. R. and M. Perry, Holes Higher Dimensional Space-Times,” of Physics]{} [**172** ]{} (1986) 304]{}. Strominger and X. Yin, “New connections between 4-D 5-D black JHEP [**0602**]{} (2006) 024 \[hep-th/0503217\]. J. Gauntlett, R. C. Myers and P. K. Townsend, holes of D = 5 supergravity,” Class. Quant. Grav. [** | -qc\]]{}]{}. G. Compere, “[Symmetries and coNservation Laws iN LaGraNgIan gAuge Theories with apPLicaTions to the mechanics of bLack hOlES and TO gRavitY in threE DiMENsiOnS]{},” [[ aRXiV:0708.3153 \[hEP-tH\]]{}]{}.
R. M. WaLd, “BLack HolE Entropy is The noEther Charge,” [[ PHYs. rev.]{} [**D48**]{} (1993) 3427]{}, [[ \[arXiv:gR-qc/9307038\]]{}]{}.
v. Iyer and R. M. WaLd, “SOme ProPeRtiES of NoEthEr ChaRge and A proposAl for DynaMiCAl BlacK hole EntROPy,” [[ phys. rev.]{} [**D50**]{} (1994) 846]{}, [[ \[arXiv:gr-qc/9403028\]]{}]{}.
T. JacOBsON, G. Kang and R. C. MyeRs, “On BlAcK hoLE entRopY,” [[ Phys. Rev.]{} [**D49**]{} (1994) 6587]{}, [[ \[aRXIv:gr-qC/9312023\]]{}]{}.
j. C. BreckENrIDGE, R. C. mYers, A. W. Peet and c. Vafa, “D-braneS And SpinniNg BlaCK holes,” [[ phys. LEtT.]{} [**b391** ]{} (1997) 93-98]{}, [[ \[heP-th/9602065\]]{}]{}.
R. Kallosh, a. RajAraman and w. K. Wong, “sUpersymMEtric roTating BlaCk hOles ANd AtTraCtORs,” [[ PHYs. rev.]{} [**d55** ]{} (1997) 3246-3249]{}, [[ \[Hep-Th/9611094\]]{}]{}.
R. C. MyerS aNd m. J. PerRy, “BlACK hOles In HIgheR DimeNsional Space-TImeS,” [[ AnnALs oF PhysIcs]{} [**172** ]{} (1986) 304]{}.
D. GAiotTo, a. StroMinger And X. YIn, “new connections bEtweEn 4-D and 5-D blAck HoLes,” jHeP [**0602**]{} (2006) 024 \[hep-TH/0503217\].
J. P. GauNtlEtt, r. C. Myers And P. K. ToWNseNd, “bLACk Holes of D = 5 supergraviTy,” cLAsS. Quant. GrAv. [** | -qc\]]{}]{}. G. Compere, “ [Symmetrie s and co nse rv atio n la ws in Lagrangi a n ga uge theories with appl icati on s tot he mech anics o f b l a ckho le s a nd to grav ity in thr ee dimensi ons ]{ },” [[ arXiv : 07 08.3153 \[ hep -th\]]{}]{}.
R . M. W al d,“ Black Ho le En tropyi s theNoether C ha r ge,” [ [ Phys.R e v. ]{}[**D48**]{} (1993 ) 3 4 27]{}, [[ \[ar Xiv:gr -q c /9 3 0 703 8\] ]{}]{}.
V .Iyera nd R. M . W a l d , “ S ome Propertie s of Noethe r Ch arge a nd aP roposa l for D y nam ical BlackHole Entropy, ” [[ P h ys. Rev . ]{} [** D50**] {}(19 94)8 46 ]{ },[[ \[a r Xi v:g r -qc /9403028 \] ]{ }]{}.
T. J a c obso n,G. K ang a nd R. C. Myer s,“OnB lac k Hol e Ent ropy ,” [[ P hys. R ev.]{ }[**D49**]{} (19 94)6587]{},[[\[ arX iv :gr-q c /93120 23\ ]]{ }]{}.
J. C. B r eck en r i d ge , R. C. Myers, A.W. P ee t and C. Vafa, “D -b r anes and s pin ning b lackhole s ,” [[ Phys . Lett . ]{ }[**B391 ** ]{} ( 19 97) 93 -98]{ } , [[ \[hep -th/9602 065\] ] {}]{}.
R. Kal l osh, A. Rajar a ma n an d W.K.Wong, “Supe rsym m etri c ro t at ing black hole sa nd attractors,” [[ Phy s. Rev.] {} [* *D55** ]{} (1 997) 3246- 3 2 4 9]{}, [[ \[h e p- t h/9611094\]]{} ]{}.
R. C. Mye r s and M. J. P erry, “B lack Hole s in Highe r D ime nsi ona l Sp ace-Times,” [ [ Anna ls of Phy sic s]{} [* *17 2** ]{ } ( 19 86) 304]{ }.
D. G ai ot to ,A.Strom i nger and X . Y in , “ New c o nnecti ons b etwe en 4 - D a nd 5-Db la c k hol es ,” JHE P [ ** 0602* *]{} (20 06) 024 \[hep-th /05 0 3217 \] .
J. P.Gauntlett, R. C . Myers an dP.K. Tow n s end, “Bl ack holes of D = 5 supe r gravity ,”Class . Qu ant. Grav . [** | -qc\]]{}]{}. G. Compere,_“[Symmetries and_conservation laws in Lagrangian_gauge theories_with_applications to_the_mechanics of black_holes and to_gravity in three dimensions]{},”_[[ arXiv:0708.3153 \[hep-th\]]{}]{}.
R._M._Wald, “Black Hole Entropy is the Noether Charge,” [[ Phys. Rev.]{} [**D48**]{} (1993) 3427]{},_[[_\[arXiv:gr-qc/9307038\]]{}]{}.
V. Iyer_and_R._M. Wald, “Some Properties of_Noether Charge and a Proposal_for Dynamical_Black Hole Entropy,” [[ Phys. Rev.]{} [**D50**]{} (1994)_846]{},_[[ \[arXiv:gr-qc/9403028\]]{}]{}.
T. Jacobson,_G. Kang and R. C. Myers, “On Black Hole_Entropy,” [[ Phys. Rev.]{} [**D49**]{} (1994)_6587]{}, [[ \[arXiv:gr-qc/9312023\]]{}]{}.
J. C. Breckenridge,_R. C. Myers,_A. W. Peet_and C. Vafa, “D-branes and_spinning black holes,” [[ Phys. Lett.]{} [**B391**_]{} (1997) 93-98]{}, [[ \[hep-th/9602065\]]{}]{}.
R. Kallosh, A. Rajaraman_and W. K. Wong, “Supersymmetric rotating black holes and_attractors,” [[ Phys. Rev.]{} [**D55** ]{}_(1997) 3246-3249]{}, [[ \[hep-th/9611094\]]{}]{}.
R. C. Myers and_M. J. Perry, “Black_Holes in Higher Dimensional Space-Times,”_[[ Annals of_Physics]{} [**172**_]{} (1986) 304]{}.
D. Gaiotto,_A. Strominger and X. Yin, “New connections between_4-D and 5-D_black holes,” JHEP [**0602**]{} (2006) 024_\[hep-th/0503217\].
J. P. Gauntlett,_R. C. Myers and P. K. Townsend,_“Black_holes_of D_= 5 supergravity,”_Class. Quant. Grav. _[** |
n$ is diagonalizing this matrix, which is possible since ${\mathbb{K}}_n[t^{\pm1}]$ is a PID. Since $c\neq1$ in $\pi_1({{\mathcal U}})/\pi_1({{\mathcal U}})^\prime$, it follows that $c \notin
\pi_1({{\mathcal U}})^{(n)}_r$ for all $n \geq 1$. Therefore $c\neq1$ in $\Gamma_n$ for all $n \geq 0$. Hence $1-c\neq0$ in ${\mathbb{Z}}\Gamma_n$ and is therefore invertible in ${\mathbb{K}}_n[t^{\pm1}]$. This allows us to multiply the last column in our presentation matrix by the unit $1-c$. Since our matrix is a presentation of a left module and since columns correspond to generators, we multiply columns on the right. The result of multiplying the last column (on the right) by the unit $1-c$ is the following: $$\left(
\begin{smallmatrix}
(aba^{-1}-baca^{-1})t+(1-b) & t-1 & (aba^{-1}-abaca^{-2})t^2+(baca^{-1}-b)t \\
(baca^{-1}-c)t+(b-c) & 1-ct^2 & (b-baca^{-1})t+(c-1) \\
\end{smallmatrix}
\right)$$ Next we add the first column times $1-t$ and the second column times $1-b$ to the last column. The result is the following: $$\left(
\begin{array}{ccc}
(aba^{-1}-baca^{-1})t+(1-b) & t-1 & 0 \\
(baca^{-1}-c)t+(b-c) & 1-ct^2 & 0 \\
\end{array}
\right)$$ This means that we have a free generator, which is expected since we are computing the homology relative to a basepoint.
Next we multiply the first row by $ct+c$ and add it to the second. Since our matrix is a presentation of a left module and since rows correspond to relations, we multiply rows on the left. The result of multiplying the first row (on the left) by $ct+c$ and adding it to the second is: $$\left( | n$ is diagonalizing this matrix, which is possible since $ { \mathbb{K}}_n[t^{\pm1}]$ is a PID. Since $ c\neq1 $ in $ \pi_1({{\mathcal U}})/\pi_1({{\mathcal U}})^\prime$, it follows that $ c \notin
\pi_1({{\mathcal U}})^{(n)}_r$ for all $ newton \geq 1$. consequently $ c\neq1 $ in $ \Gamma_n$ for all $ n \geq 0$. Hence $ 1 - c\neq0 $ in $ { \mathbb{Z}}\Gamma_n$ and is consequently invertible in $ { \mathbb{K}}_n[t^{\pm1}]$. This allows us to reproduce the last column in our presentation matrix by the unit of measurement $ 1 - c$. Since our matrix is a presentation of a left faculty and since column correspond to generators, we multiply columns on the right. The result of multiply the last column (on the right) by the unit $ 1 - c$ is the pursuit: $ $ \left (
\begin{smallmatrix }
(aba^{-1}-baca^{-1})t+(1 - b) & t-1 & (aba^{-1}-abaca^{-2})t^2+(baca^{-1}-b)t \\
(baca^{-1}-c)t+(b - c) & 1 - ct^2 & (b - baca^{-1})t+(c-1) \\
\end{smallmatrix }
\right)$$ Next we add the first column times $ 1 - t$ and the second column times $ 1 - b$ to the concluding column. The result is the pursuit: $ $ \left (
\begin{array}{ccc }
(aba^{-1}-baca^{-1})t+(1 - b) & t-1 & 0 \\
(baca^{-1}-c)t+(b - c) & 1 - ct^2 & 0 \\
\end{array }
\right)$$ This means that we consume a free generator, which is expected since we are computing the homology relative to a basepoint.
Next we breed the first row by $ ct+c$ and add it to the second. Since our matrix is a presentation of a left module and since rows correspond to relations, we multiply rows on the left. The result of multiplying the first row (on the left field) by $ ct+c$ and add it to the second is: $ $ \left ( | n$ id diagonalizing this matvix, which is possible smnce ${\mafhbb{K}}_n[t^{\po1}]$ is a PID. Since $c\neq1$ in $\pi_1({{\methcql U}})/\pu_1({{\mathcal U}})^\prime$, it foluows that $c \notin
\pi_1({{\mauhcal U}})^{(n)}_r$ for all $n \geq 1$. Bkerefkve $c\nzq1$ in $\Gamma_n$ for all $n \geq 0$. Hence $1-c\neq0$ it ${\oachbb{Z}}\Gamma_n$ and is therefore invertifle in ${\kahhbb{K}}_n[t^{\pm1}]$. This allprs ua to multiply the last column in ohr prestntation matrix by the unit $1-c$. Since our matrlx id a presentation ov a left moeule qnd since couumns corrts'ond to gengrators, we multiply columns on tfe riyht. The resolf lx multiplyiig the last column (on the sight) bu the unit $1-c$ ix tie fillowing: $$\left(
\begin{smanlmatrix}
(aba^{-1}-baca^{-1})t+(1-b) & t-1 & (aba^{-1}-abdcc^{-2})t^2+(baca^{-1}-b)t \\
(baca^{-1}-c)t+(b-c) & 1-ct^2 & (v-baca^{-1})j+(c-1) \\
\eng{smauomagris}
\rmghf)$$ Next we add the fjrst column times $1-t$ and the sevoge column timea $1-b$ to tre last column. The result is the followpng: $$\meft(
\begin{array}{ccc}
(aba^{-1}-bacq^{-1})t+(1-b) & t-1 & 0 \\
(baca^{-1}-c)t+(b-c) & 1-ct^2 & 0 \\
\end{arrwy}
\right)$$ This means that we have a free generator, fhich ks txicctea slnce we are computing the homology relative tj a bssepoint.
Next wc multiply the firxt rpr by $ct+c$ and xdd it to the second. Since lur mattix is a presenuatiom of a left module and sincw rows correfpond to relations, ce multiply xows om the left. The result of mulciplyihg the firsh row (on ffe left) by $ct+c$ avd sdging it uu the second is: $$\lqft( | n$ is diagonalizing this matrix, which is ${\mathbb{K}}_n[t^{\pm1}]$ a PID. $c\neq1$ in $\pi_1({{\mathcal $c \pi_1({{\mathcal U}})^{(n)}_r$ for $n \geq 1$. $c\neq1$ in $\Gamma_n$ for all $n 0$. Hence $1-c\neq0$ in ${\mathbb{Z}}\Gamma_n$ and is therefore invertible in ${\mathbb{K}}_n[t^{\pm1}]$. This allows to multiply the last column in our presentation matrix by the unit $1-c$. our is presentation a left module and since columns correspond to generators, we multiply columns on the right. The of multiplying the last column (on the right) the unit $1-c$ is following: $$\left( \begin{smallmatrix} (aba^{-1}-baca^{-1})t+(1-b) & & \\ (baca^{-1}-c)t+(b-c) 1-ct^2 (b-baca^{-1})t+(c-1) \end{smallmatrix} \right)$$ Next add the first column times $1-t$ and the second column times $1-b$ to the last column. The is the \begin{array}{ccc} (aba^{-1}-baca^{-1})t+(1-b) t-1 0 (baca^{-1}-c)t+(b-c) & 1-ct^2 \\ \end{array} \right)$$ This means that free generator, which is expected since we are the homology to a basepoint. Next we multiply first row by $ct+c$ and add it to second. Since our matrix is a presentation of a left module and since rows correspond we multiply rows on left. The result multiplying first (on left) by and adding it to the second is: $$\left( | n$ is diagonalizing this matriX, which is poSsiblE siNce ${\MaThbb{k}}_n[t^{\pM1}]$ is a PID. Since $c\nEQ1$ in $\pI_1({{\mathcal U}})/\pi_1({{\mathcal U}})^\priMe$, it fOlLOws tHAt $C \notiN
\pi_1({{\mathCAl u}})^{(N)}_R$ foR aLl $N \geQ 1$. THErEfore $C\neQ1$ in $\GammA_n$ for all $n \gEq 0$. HEnCe $1-c\neq0$ in ${\mathBB{Z}}\gamma_n$ and iS thErefore inverTibLe in ${\maThBb{K}}_N[T^{\pm1}]$. ThIs aLlows Us to muLTiply tHe last colUmN In our pREsentatION mAtriX by the unit $1-c$. Since oUR mATrix is a presentAtion oF a LEfT MOduLe aNd since colUmNs corREspond tO GeNERAtoRS, we multiply coLumns on the rIGht. the resUlT of MUltiplYing tHe LAst Column (on the RighT) by the uniT $1-c$ is thE FollowiNG: $$\left(
\beGin{smaLlmAtrIx}
(abA^{-1}-BaCa^{-1})T+(1-b) & t-1 & (AbA^{-1}-AbaCA^{-2})t^2+(BacA^{-1}-B)t \\
(bAca^{-1}-c)t+(b-c) & 1-cT^2 & (b-BaCa^{-1})t+(c-1) \\
\eNd{smALLMAtriX}
\riGht)$$ NExt we Add the first coLumN timES $1-t$ aNd the SeconD colUmN timeS $1-b$ to thE last CoLumn. The result is The fOllowing: $$\lEft(
\BeGin{ArRay}{ccC}
(Aba^{-1}-bacA^{-1})t+(1-b) & T-1 & 0 \\
(baCa^{-1}-c)t+(b-c) & 1-cT^2 & 0 \\
\end{arrAY}
\riGhT)$$ tHIs Means that we have a frEe GENeRator, whiCh is exPEcTeD Since we aRe ComPutiNG The hoMoloGY rElative tO a basePOiNt.
next we mUlTiply tHe FirSt rOw by $cT+C$ and Add it tO the secoNd. SinCE our matrix is a pREsentation of a LEfT MOdULe anD siNce rows corrEspoND to rElatIOnS, we MUltipLy rowS oN ThE Left. The result of multIpLying tHe firSt row (on the lefT) by $ct+c$ and aDDINg it to thE secONd IS: $$\left( | n$ is diagonalizing this m atrix, whi ch is po ssi bl e si nce${\mathbb{K}}_ n [t^{ \pm1}]$ is a PID. Sinc e $c\ ne q 1$ i n $ \pi_1 ({{\mat h ca l U}} )/ \p i_1 ({ { \m athca l U }})^\pr ime$, it f oll ow s that $c \n o ti n
\pi_1({{ \ma thcal U}})^{ (n) }_r$ f or al l $n \ geq 1$.Theref o re $c\ neq1$ in$\ G amma_n $ for al l $n \ge q 0$. Hence $1-c\ n eq 0 $ in ${\mathbb {Z}}\G am m a_ n $ an d i s therefor einver t ible in ${ \ m a thb b {K}}_n[t^{\pm 1}]$. Thisa llo ws usto mu l tiplythe l as t co lumn in our pre sentation matri x by the unit $1 -c$. S inc e o ur m a tr ix is a pre s en tat i onof a lef tmo duleands i n c e co lum ns c orres pond to gener ato rs,w e m ultip ly co lumn son th e righ t. Th eresult of multi plyi ng the la stco lum n(on t h e righ t)bythe uni t $1-c$ isth e f ol lowing: $$\left(
\ be g i n{ smallmat rix}
( a ba ^{ - 1}-baca^ {- 1}) t+(1 - b ) & t -1 & (a ba^{-1}- abaca^ { -2 }) t^2+(ba ca ^{-1}- b) t \ \
( baca^ { -1}- c)t+(b -c) & 1- ct^2& (b-baca^{-1}) t +(c-1) \\
\en d {s m a ll m atri x}\right)$$ N extw e ad d th e f irs t colu mn ti me s $ 1 -t$ and the secondco lumn t imes$1-b$ to thelast colum n . The resu lt i s t h e following: $ $\lef t(
\begin{ a rray}{cc c}
(a ba^{-1}- baca^{-1} ) t +(1-b) & t- 1 & 0\\( b ac a^{-1}-c)t+(b - c ) &1- ct^2 &0 \ \
\end{ arr ay}
\r igh t) $$ This m eans tha twe h av e a free generato r, wh ic h i s exp e cted s incewe a re c o mpu ting th e h o m olog yre lati veto a ba sepo i nt.
Nextwe multip lyt he f ir st row by $ct+c$ and a dd it to the s eco nd. Si n c e our ma trix is a presentationo f a lef t m odule and since ro wscorres pon d to re lation s, we m ult i p ly ro w s o n t he left. The r esu lt of m ulti plyingthe first row (ont heleft) by $ct+ c$anda d di ngi tt o t he sec o n d is: $$\left( | n$ is_diagonalizing this_matrix, which is possible_since ${\mathbb{K}}_n[t^{\pm1}]$_is_a PID._Since_$c\neq1$ in $\pi_1({{\mathcal_U}})/\pi_1({{\mathcal U}})^\prime$, it_follows that $c \notin
\pi_1({{\mathcal_U}})^{(n)}_r$ for all_$n_\geq 1$. Therefore $c\neq1$ in $\Gamma_n$ for all $n \geq 0$. Hence $1-c\neq0$ in_${\mathbb{Z}}\Gamma_n$_and is_therefore_invertible_in ${\mathbb{K}}_n[t^{\pm1}]$. This allows us_to multiply the last column_in our_presentation matrix by the unit $1-c$. Since our_matrix_is a presentation_of a left module and since columns correspond to_generators, we multiply columns on the_right. The result_of_multiplying_the last column (on_the right) by the unit $1-c$_is the following: $$\left(
\begin{smallmatrix}
(aba^{-1}-baca^{-1})t+(1-b) & t-1_& (aba^{-1}-abaca^{-2})t^2+(baca^{-1}-b)t \\
(baca^{-1}-c)t+(b-c) & 1-ct^2 & (b-baca^{-1})t+(c-1)_\\
\end{smallmatrix}
\right)$$ Next we add the first_column times $1-t$ and the_second column_times $1-b$ to the last_column. The result_is the_following: $$\left(
\begin{array}{ccc}
(aba^{-1}-baca^{-1})t+(1-b) &_t-1 & 0 \\
(baca^{-1}-c)t+(b-c) & 1-ct^2_& 0 \\
\end{array}
\right)$$_This means that we have a_free_generator, which is_expected_since_we are_computing the homology_relative_to a_basepoint.
Next_we multiply the first row by_$ct+c$_and add it to the second. Since_our matrix is a_presentation_of a left module_and since rows correspond to_relations, we multiply rows on the_left. The_result of_multiplying the first row (on the left) by $ct+c$ and adding_it to the second is: $$\left( |
Spurious Correlations
=======================================
There are enormous differences in the size of academic journals, and these differences swamp the patterns that Davis was seeking in his analysis. The JCR indexes journals that range in size from tiny (*Astronomy and Astrophysics Review* has published 13 articles over the previous five years) to huge (*The Journal of Biological Chemistry* has published 31,045 articles over the same period) with a coefficient of variation, $c_v$, equal to 1.910. Per-article citation intensity varies less, whether measured by Article Influence or by Impact Factor (AI: range 0–27.5, coefficient of variation$=1.785$; IF: range 0–63.3, coefficient of variation$=1.548$).
We can formalize these observations by decomposing Davis’ regression of Eigenfactor on Total Citations. Davis regresses
Log($EF_i$) Log($CT_i$),
where $EF_i$ is the Eigenfactor score for journal $i$ and $CT_i$ is the Total Citations received by journal $i$. We let $AI_i$ be the Article Influence for journal $i$, and $N_{i,5}$ is the total number of articles published over the last five years for journal $i$. Then by definition
$$\begin{aligned}
\log(EF_i)& = & \log(c_1 \times AI_i \times N_{i,5}) \nonumber \\
&=& \log c_1 + \log AI_i + \log N_{i,5}, \nonumber \end{aligned}$$
where $c_1$ is a scaling constant that normalizes the Article Influence scores so that the mean article in the JCR has an Article Influence score of 1.00. Similarly, letting $IF_i$ be the Impact Factor for journal $i$,
$$\begin{aligned}
\log(CT_i)&\approx& \log(c_2 \times IF_i \times N_{i,2}) \nonumber \\
&\approx& \log(c_2\, c_3 \times IF_i \times N_{i,5}) \nonumber \\
&=&\log c_2 \,c_3 + \log IF_i + \log N_{i,5} \nonumber\end{aligned}$$
where $c_2$ and $c | Spurious Correlations
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
There are enormous differences in the size of academic journals, and these differences deluge the design that Davis was seeking in his analysis. The JCR index journal that range in size from tiny (* Astronomy and Astrophysics Review * has print 13 articles over the previous five old age) to huge (* The Journal of Biological Chemistry * has published 31,045 articles over the like period) with a coefficient of variation, $ c_v$, equal to 1.910. Per - article citation volume varies less, whether measured by Article Influence or by Impact Factor (AI: range 0–27.5, coefficient of variation$=1.785 $; IF: range 0–63.3, coefficient of variation$=1.548 $).
We can formalize these observations by decompose Davis ’ regression of Eigenfactor on Total Citations. Davis regresses
Log($EF_i$) Log($CT_i$),
where $ EF_i$ is the Eigenfactor score for daybook $ i$ and $ CT_i$ is the Total Citations received by journal $ i$. We let $ AI_i$ be the Article Influence for journal $ i$, and $ N_{i,5}$ is the entire number of articles published over the last five years for diary $ i$. Then by definition
$ $ \begin{aligned }
\log(EF_i) & = & \log(c_1 \times AI_i \times N_{i,5 }) \nonumber \\
& = & \log c_1 + \log AI_i + \log N_{i,5 }, \nonumber \end{aligned}$$
where $ c_1 $ is a scaling constant that normalizes the Article Influence scores so that the mean article in the JCR has an Article Influence score of 1.00. Similarly, letting $ IF_i$ be the Impact Factor for journal $ i$,
$ $ \begin{aligned }
\log(CT_i)&\approx & \log(c_2 \times IF_i \times N_{i,2 }) \nonumber \\
& \approx & \log(c_2\, c_3 \times IF_i \times N_{i,5 }) \nonumber \\
& = & \log c_2 \,c_3 + \log IF_i + \log N_{i,5 } \nonumber\end{aligned}$$
where $ c_2 $ and $ c | Spkrious Correlations
=======================================
There are enormous dndferenres in fhe size of academic journals, and thxse eiffeeences swamp the pattefns that Favis waw setking in his analbais. The JCR ihfexev journals that range in vize from tiny (*Artxonomy and Astrophysics Review* has ptblishec 13 articles over the [revjous five years) to huge (*The Journam of Bimlogical Chemostry* has published 31,045 articpes lver the same perild) with a cieffyxient of varkation, $c_v$, tqbal to 1.910. Per-zrticle citation intensity varids lexs, whether mwaskted by Articoe Insluence or bn Impacd Factot (AI: range 0–27.5, cocfficment of variation$=1.785$; IF: rangx 0–63.3, coefficient of vatiation$=1.548$).
We wau formalize these obswrcatiots bf dezimpusihg Dzvis’ rfgrxssion of Ejgenfactor in Total Citations. Cadpx regresses
Lkg($EF_i$) Jod($CT_i$),
where $EF_i$ is the Eigenfactor score xor journal $i$ and $CT_i$ is tye Total Citations refeived by journal $i$. We let $AI_i$ be the Article Influence fos joucnxl $n$, and $V_{u,5}$ ls the total number of articles published ovew tne last five yeavs for journal $i$. Tnej ni definition
$$\beein{aliyhes}
\log(EF_i)& = & \log(c_1 \timfs AI_i \jimes B_{i,5}) \nonumbtr \\
&=& \kog c_1 + \log AI_i + \log N_{i,5}, \nonunber \end{aligued}$$
qhere $c_1$ is a scaliug constant chat nprmalozes the Article Influeuce sckres so thah the meah article in the GCR hds an Arukcle Influence scjre of 1.00. Smmilaxly, lettkng $OF_i$ be the Impach Facbmr for journal $i$,
$$\behin{alngned}
\nog(CT_i)&\apprlx& \log(c_2 \times IF_i \times N_{i,2}) \nonumber \\
&\approx& \lob(c_2\, c_3 \nimes IF_i \timex N_{i,5}) \nonumber \\
&=&\log c_2 \,c_3 + \log NF_i + \log N_{i,5} \nunumber\end{zligned}$$
xhere $c_2$ and $s | Spurious Correlations ======================================= There are enormous differences size academic journals, these differences swamp seeking his analysis. The indexes journals that in size from tiny (*Astronomy and Review* has published 13 articles over the previous five years) to huge (*The of Biological Chemistry* has published 31,045 articles over the same period) with a of $c_v$, to Per-article citation intensity varies less, whether measured by Article Influence or by Impact Factor (AI: range coefficient of variation$=1.785$; IF: range 0–63.3, coefficient of We can formalize these by decomposing Davis’ regression of on Citations. Davis Log($EF_i$) where is the Eigenfactor for journal $i$ and $CT_i$ is the Total Citations received by journal $i$. We let $AI_i$ be Article Influence $i$, and is total of articles published last five years for journal $i$. $$\begin{aligned} \log(EF_i)& = & \log(c_1 \times AI_i \times \nonumber \\ \log c_1 + \log AI_i + N_{i,5}, \nonumber \end{aligned}$$ where $c_1$ is a scaling that normalizes the Article Influence scores so that the mean article in the JCR has Influence score of 1.00. letting $IF_i$ be Impact for $i$, \log(CT_i)&\approx& \log(c_2 IF_i \times N_{i,2}) \nonumber \\ &\approx& \log(c_2\, c_3 \times IF_i \times \nonumber \\ &=&\log c_2 \,c_3 + \log IF_i + \log where and $c | Spurious Correlations
=======================================
There Are enormouS diffEreNceS iN the Size Of academic jourNAls, aNd these differences swamP the pAtTErns THaT DaviS was seeKInG IN hiS aNaLysIs. tHe jCR inDexEs journAls that ranGe iN sIze from tiny (*ASTrOnomy and AsTroPhysics RevieW* haS publiShEd 13 aRTicleS ovEr the PrevioUS five yEars) to hugE (*THE JournAL of BiolOGIcAl ChEmistry* has publishED 31,045 aRTicles over the sAme perIoD) WiTH A coEffIcient of vaRiAtion, $C_V$, equal tO 1.910. peR-ARTicLE citation inteNsity varies LEss, WhetheR mEasURed by ARticlE INFluEnce or by ImpAct FActor (AI: raNge 0–27.5, coeFFicient OF variatIon$=1.785$; IF: rAngE 0–63.3, coEffiCIeNt Of vArIAtiON$=1.548$).
WE caN ForMalize thEsE oBservAtioNS BY DecoMpoSing davis’ Regression of EIgeNfacTOr oN TotaL CitaTionS. DAvis rEgressEs
Log($eF_I$) Log($CT_i$),
where $EF_i$ Is thE EigenfacTor ScOre FoR jourNAl $i$ and $cT_i$ Is tHe Total citatioNS reCeIVED bY journal $i$. We let $AI_i$ bE tHE arTicle InfLuence FOr JoURnal $i$, and $n_{i,5}$ Is tHe toTAL numbEr of ARtIcles pubLished OVeR tHe last fIvE years FoR joUrnAl $i$. ThEN by dEfinitIon
$$\begin{AlignED}
\log(EF_i)& = & \log(c_1 \timES AI_i \times N_{i,5}) \noNUmBER \\
&=& \lOG c_1 + \loG AI_I + \log N_{i,5}, \nonumBer \eND{aliGned}$$
WHeRe $c_1$ IS a scaLing cOnSTaNT that normalizes the ARtIcle InFluenCe scores so thaT the mean arTICLe in the JcR haS An aRticle InfluencE scorE of 1.00. SimilarLY, letting $iF_i$ be The ImpacT Factor foR JOurnal $i$,
$$\bEgiN{alIgnEd}
\lOG(cT_I)&\approx& \log(c_2 \tiMES IF_i \TiMes N_{i,2}) \noNumBer \\
&\apprOx& \lOg(c_2\, C_3 \tiMes iF_I \times N_{i,5}) \nOnumber \\
&=&\lOg C_2 \,c_3 + \LoG If_i + \lOg N_{i,5} \nONumber\enD{aLigNeD}$$
whEre $c_2$ aND $c | Spurious Correlations
=== ========== ===== === === == ==== ==== =====
There a r e en ormous differences inthe s iz e ofa ca demic journa l s, a ndth es e d if f er ences sw amp the patternstha tDavis was se e ki ng in hisana lysis. The J CRindexe sjou r nalstha t ran ge ins ize fr om tiny ( *A s tronom y and As t r op hysi cs Review* has pu b li s hed 13 article s over t h ep r evi ous five year s) to h u ge (*Th e J o u r nal of Biological Chemistry* has publi sh ed3 1,045artic le s ov er the same per iod) with a coe f ficient of vari ation, $c _v$ , eq u al t o 1 .9 1 0.P er -ar t icl e citati on i ntens ityv a r i es l ess , wh ether measured byArt icle Inf luenc e orby I mp act F actor(AI:ra nge 0–27.5, coe ffic ient of v ari at ion $= 1.785 $ ; IF:ran ge0–63.3, coeffi c ien to f va riation$=1.548$).
W e ca n formal ize th e se o b servatio ns by dec o m posin g Da v is ’ regres sion o f E ig enfacto ron Tot al Ci tat ions. Davi s regr esses
Log($ E F_i$) Log($CT_ i $),
where $E F _i $ is theEig enfactor sc oref or j ourn a l$i$ and $ CT_i$ i s t h e Total Citations r ec eivedby jo urnal $i$. We let $AI_i $ b e the Ar ticl e I n fluence for jo urnal $i$, and$ N_{i,5}$ is t he total number o f articles pu bli she d o v e rthe last five y ears f or jour nal $i$. T hen by de fin it ion
$$\b egin{ali gn ed }\l og( EF_i) & = & \lo g( c_1 \ tim es AI _ i \tim es N_ {i,5 }) \ n onu mber \ \
& = & \lo gc_ 1 +\lo gAI_i+ \l o g N _{i,5}, \nonumbe r \ e nd{a li gn ed}$$
where $c_1$ i sa scalingco nst ant th a t normali zes the Article Influen c e score s s o tha t th e mean ar tic le inthe JCR ha s an A rticl eInf l u ences c or e o f1.00. Simi l a rly , let ti ng $ IF_i$ b e the Impact Facto r fo r journal $i$ ,
$$\b e g in {al i gn e d}\l o g(C T _ i)&\approx& \lo g(c_2 \tim es IF _i \timesN _{i ,2 }) \non umber \ \
&\a p prox& \ log(c_2\, c_3 \tim es IF_ i \ti mes N_{i,5 }) \nonu mber \\
& = &\log c_ 2 \,c _3+ \log I F_i + \l og N_{ i ,5} \non umber\ en d{alig ned}$ $
where $ c_2$ and $c | Spurious_Correlations
=======================================
There are_enormous differences in the_size of_academic_journals, and_these_differences swamp the_patterns that Davis_was seeking in his_analysis. The JCR_indexes_journals that range in size from tiny (*Astronomy and Astrophysics Review* has published 13_articles_over the_previous_five_years) to huge (*The Journal_of Biological Chemistry* has published_31,045 articles_over the same period) with a coefficient of_variation,_$c_v$, equal to_1.910. Per-article citation intensity varies less, whether measured by_Article Influence or by Impact Factor_(AI: range 0–27.5,_coefficient_of_variation$=1.785$; IF: range 0–63.3,_coefficient of variation$=1.548$).
We can formalize these_observations by decomposing Davis’ regression of_Eigenfactor on Total Citations. Davis regresses
Log($EF_i$) Log($CT_i$),
where_$EF_i$ is the Eigenfactor score for_journal $i$ and $CT_i$ is_the Total_Citations received by journal $i$._We let $AI_i$_be the_Article Influence for_journal $i$, and $N_{i,5}$ is the_total number of_articles published over the last five_years_for journal $i$._Then_by_definition
$$\begin{aligned}
\log(EF_i)& =_& \log(c_1 \times_AI_i_\times N_{i,5})_\nonumber_ \\
&=& \log c_1 + \log_AI_i_+ \log N_{i,5}, \nonumber \end{aligned}$$
where $c_1$ is_a scaling constant that_normalizes_the Article Influence scores_so that the mean article_in the JCR has an Article_Influence score_of 1.00._Similarly, letting $IF_i$ be the Impact Factor for journal $i$,
$$\begin{aligned}
\log(CT_i)&\approx& \log(c_2_\times IF_i \times N_{i,2}) \nonumber \\
&\approx&_\log(c_2\, c_3 \times IF_i_\times N_{i,5})_\nonumber_\\
&=&\log c_2 \,c_3_+_\log IF_i_+ \log N_{i,5} \nonumber\end{aligned}$$
where $c_2$ and $c |
the normalizer of $T$ in $G$. An element $\sigma\in
Aut(\mathfrak{g},\mathfrak{t})$ induces a symmetry of $\Phi$, and we denote the resulting group homomorphism by $$\tau: Aut(\mathfrak{g},\mathfrak{t}){\longrightarrow}Aut(\Phi), \ \sigma\mapsto (\sigma^{-1}_{|\mathfrak{t}})^*.$$
Recall also that the **Weyl group** of $\Phi$, denoted by $W\subset Aut(\Phi)$, is the group generated by the symmetries $s_{\alpha}$ (defined in (\[EQ\_symmetries\])), for $\alpha\in \Phi$.
For the following lemma see Theorem 7.8 [@Knapp] and section 3.15 [@DK].
\[Lemma\_3\] The map $\tau:Aut(\mathfrak{g},\mathfrak{t})\to Aut(\Phi)$ is surjective, and $$\tau^{-1}(W)=Ad(N_G(T))\subset
Aut(\mathfrak{g},\mathfrak{t}).$$ Therefore, $\tau$ induces an isomorphism between the groups $$Out(\mathfrak{g})\cong Aut(\Phi)/W.$$
If $\mathfrak{c}\subset \mathfrak{t}^*$ is an open Weyl chamber, then $Aut(\Phi)=Aut(\Phi,\mathfrak{c})\ltimes W$, where $$Aut(\Phi,\mathfrak{c}):=\{f\in Aut(\Phi)| f(i\mathfrak{c})=i\mathfrak{c}\}.$$ Moreover, $Aut(\Phi,\mathfrak{c})$ (hence also $Out(\mathfrak{g})$) is isomorphic to the symmetry group of the Dynkin diagram of $\Phi$.
The last part of the lemma allows us to compute $Out(\mathfrak{g})$ for all semisimple compact Lie algebras. First, it is enough to consider simple Lie algebras, since, if $\mathfrak{g}$ decomposes into simple components as $n_1\mathfrak{s}_1\oplus \ldots \oplus n_k \mathfrak{s}_k$, then $$Out(\mathfrak{g})\cong S_{n_1}\ltimes
Out(\mathfrak{s}_1)^{n_1}\times\ldots \times S_{n_k}\ltimes Out(\mathfrak{s}_k)^{n_k}.$$ Further, for the simple Lie algebras, a glimpse at their Dynkin diagrams reveals that the only ones with nontrivial outer automorphism group are: | the normalizer of $ T$ in $ G$. An element $ \sigma\in
Aut(\mathfrak{g},\mathfrak{t})$ induces a symmetry of $ \Phi$, and we denote the result group homomorphism by $ $ \tau: Aut(\mathfrak{g},\mathfrak{t}){\longrightarrow}Aut(\Phi), \ \sigma\mapsto (\sigma^{-1}_{|\mathfrak{t}})^*.$$
Recall besides that the * * Weyl group * * of $ \Phi$, denoted by $ W\subset Aut(\Phi)$, is the group generated by the isotropy $ s_{\alpha}$ (define in (\[EQ\_symmetries\ ]) ), for $ \alpha\in \Phi$.
For the following lemma see Theorem 7.8 [ @Knapp ] and part 3.15 [ @DK ].
\[Lemma\_3\ ] The map $ \tau: Aut(\mathfrak{g},\mathfrak{t})\to Aut(\Phi)$ is surjective, and $ $ \tau^{-1}(W)=Ad(N_G(T))\subset
Aut(\mathfrak{g},\mathfrak{t}).$$ Therefore, $ \tau$ induces an isomorphism between the group $ $ Out(\mathfrak{g})\cong Aut(\Phi)/W.$$
If $ \mathfrak{c}\subset \mathfrak{t}^*$ is an open Weyl chamber, then $ Aut(\Phi)=Aut(\Phi,\mathfrak{c})\ltimes W$, where $ $ Aut(\Phi,\mathfrak{c}):=\{f\in Aut(\Phi)| f(i\mathfrak{c})=i\mathfrak{c}\}.$$ furthermore, $ Aut(\Phi,\mathfrak{c})$ (hence also $ Out(\mathfrak{g})$) is isomorphic to the symmetry group of the Dynkin diagram of $ \Phi$.
The final part of the lemma allows us to compute $ Out(\mathfrak{g})$ for all semisimple compact Lie algebras. foremost, it is enough to consider simple Lie algebra, since, if $ \mathfrak{g}$ decomposes into simple components as $ n_1\mathfrak{s}_1\oplus \ldots \oplus n_k \mathfrak{s}_k$, then $ $ Out(\mathfrak{g})\cong S_{n_1}\ltimes
Out(\mathfrak{s}_1)^{n_1}\times\ldots \times S_{n_k}\ltimes Out(\mathfrak{s}_k)^{n_k}.$$ Further, for the simple Lie algebras, a glance at their Dynkin diagrams unwrap that the only ones with nontrivial outer automorphism group are: | thf normalizer of $T$ in $G$. Ak element $\sigma\iu
Qut(\matifrak{g},\mzthfrak{t})$ induces a symmetry of $\Phi$, aid ww denite the resulting grouo homomorihism by $$\rau: Eut(\mathfrak{g},\mathhdak{t}){\lonnxightzvrow}Abt(\'hi), \ \sigma\mapstp (\sigma^{-1}_{|\matvfrak{t}})^*.$$
Recall ansu chat the **Weyl group** of $\Phi$, denoted br $W\subsrt Aut(\Phi)$, is the grolp gensgaued by the symmetries $s_{\alpha}$ (defihed in (\[TQ\_symmetries\])), for $\akpha\in \Phi$.
For the followinh lelma see Theorem 7.8 [@Kjapp] and sextiog 3.15 [@DK].
\[Lemma\_3\] Thd map $\tau:Alc(\mathfrak{g},\mzthfrak{t})\to Aut(\Phi)$ is surjective, and $$\cau^{-1}(W)=Ad(N_G(T))\suvswt
Akj(\mathfrak{g},\mavhfrak{n}).$$ Therefore, $\tau$ inducas an ixomorphism beteeei thw groups $$Out(\mathfrak{g})\rong Aut(\Phi)/W.$$
If $\mathftak{c}\subset \mcthfrak{t}^*$ is an open Wwyo chakber, thev $Aug(\Phj)=Ant(\Pgi,\mathvrai{c})\ltimes W$, where $$Aut(\Pyi,\mathfrak{c}):=\{f\in Aut(\Pni)| d(i\mathfrak{c})=i\mzthfrah{c}\}.$$ Moreover, $Aut(\Phi,\mathfrak{c})$ (hence also $Ouu(\mathrrak{g})$) is isomorphic to rhe symmetry group of the Dynkyn diagram of $\Phi$.
The last part of the lemma allowv us vo conpmte $Uyt(\lathfrak{g})$ for all semisimple compact Lie algefdax. Nirst, it is enounh to consider simllf Kye algebras, sknce, iy $\mzthfrak{g}$ decomposed into fimplw componegts ss $n_1\mathfrak{s}_1\oplus \ldots \opous n_k \mathfgak{s}_j$, then $$Out(\mathfrak{y})\cong S_{n_1}\ltimzs
Out(\msthfrsk{s}_1)^{n_1}\times\ldots \times S_{n_y}\ltijes Out(\mathvrak{s}_k)^{n_k}.$$ Rjrther, for the skmpke Lie algtcras, a glimpse at their Dyikin biagrams revgals thwt the onlj ones with nontrivial okter cutommrphism grlup are: | the normalizer of $T$ in $G$. An Aut(\mathfrak{g},\mathfrak{t})$ a symmetry $\Phi$, and we by Aut(\mathfrak{g},\mathfrak{t}){\longrightarrow}Aut(\Phi), \ \sigma\mapsto Recall also that **Weyl group** of $\Phi$, denoted by Aut(\Phi)$, is the group generated by the symmetries $s_{\alpha}$ (defined in (\[EQ\_symmetries\])), for \Phi$. For the following lemma see Theorem 7.8 [@Knapp] and section 3.15 [@DK]. The $\tau:Aut(\mathfrak{g},\mathfrak{t})\to is and $$\tau^{-1}(W)=Ad(N_G(T))\subset Aut(\mathfrak{g},\mathfrak{t}).$$ Therefore, $\tau$ induces an isomorphism between the groups $$Out(\mathfrak{g})\cong Aut(\Phi)/W.$$ If $\mathfrak{c}\subset \mathfrak{t}^*$ an open Weyl chamber, then $Aut(\Phi)=Aut(\Phi,\mathfrak{c})\ltimes W$, where Aut(\Phi)| f(i\mathfrak{c})=i\mathfrak{c}\}.$$ Moreover, $Aut(\Phi,\mathfrak{c})$ also $Out(\mathfrak{g})$) is isomorphic to symmetry of the diagram $\Phi$. last part of lemma allows us to compute $Out(\mathfrak{g})$ for all semisimple compact Lie algebras. First, it is enough to simple Lie if $\mathfrak{g}$ into components $n_1\mathfrak{s}_1\oplus \ldots \oplus then $$Out(\mathfrak{g})\cong S_{n_1}\ltimes Out(\mathfrak{s}_1)^{n_1}\times\ldots \times S_{n_k}\ltimes the simple Lie algebras, a glimpse at their diagrams reveals the only ones with nontrivial outer group are: | the normalizer of $T$ in $G$. An elemEnt $\sigma\in
aut(\maThfRak{G},\mAthfRak{t})$ Induces a symmetRY of $\PHi$, and we denote the resultIng grOuP HomoMOrPhism By $$\tau: AuT(\MaTHFraK{g},\MaThfRaK{T}){\lOngriGhtArrow}AuT(\Phi), \ \sigma\mApsTo (\Sigma^{-1}_{|\mathfraK{T}})^*.$$
REcall also tHat The **Weyl group** Of $\PHi$, denoTeD by $w\SubseT AuT(\Phi)$, iS the grOUp geneRated by thE sYMmetriES $s_{\alpha}$ (DEFiNed iN (\[EQ\_symmetries\])), for $\aLPhA\In \Phi$.
For the folLowing LeMMa SEE ThEorEm 7.8 [@Knapp] and SeCtion 3.15 [@dk].
\[Lemma\_3\] THE mAP $\TAu:AUT(\mathfrak{g},\matHfrak{t})\to Aut(\pHi)$ iS surjeCtIve, ANd $$\tau^{-1}(W)=ad(N_G(T))\SuBSet
aut(\mathfrak{G},\matHfrak{t}).$$ TheRefore, $\TAu$ inducES an isomOrphisM beTweEn thE GrOuPs $$OUt(\MAthFRaK{g})\cONg AUt(\Phi)/W.$$
If $\MaThFrak{c}\SubsET \MAThfrAk{t}^*$ Is an Open WEyl chamber, theN $AuT(\Phi)=aUt(\PHi,\matHfrak{C})\ltiMeS W$, wheRe $$Aut(\PHi,\matHfRak{c}):=\{f\in Aut(\Phi)| f(i\MathFrak{c})=i\matHfrAk{C}\}.$$ MoReOver, $AUT(\Phi,\maThfRak{C})$ (hence aLso $Out(\mAThfRaK{G})$) IS iSomorphic to the symmEtRY GrOup of the dynkin DIaGrAM of $\Phi$.
ThE lAst Part OF The leMma aLLoWs us to coMpute $OUT(\mAtHfrak{g})$ fOr All semIsImpLe cOmpacT lie aLgebraS. First, it Is enoUGh to consider siMPle Lie algebraS, SiNCE, iF $\MathFraK{g}$ decomposeS intO SimpLe coMPoNenTS as $n_1\mAthfrAk{S}_1\OpLUs \ldots \oplus n_k \mathfRaK{s}_k$, theN $$Out(\mAthfrak{g})\cong S_{N_1}\ltimes
Out(\MATHfrak{s}_1)^{n_1}\tImes\LDoTS \times S_{n_k}\ltimeS Out(\mAthfrak{s}_k)^{n_K}.$$ further, fOr the Simple LiE algebras, A GLimpse at TheIr DYnkIn dIAGrAms reveals thaT THe onLy Ones witH noNtriviaL ouTer AutOmoRpHism group Are: | the normalizer of $T$ in$G$. An el ement $\ sig ma \inAut( \mathfrak{g},\ m athf rak{t})$ induces a sym metry o f $\P h i$ , and we den o te t here su lti ng gr oup h omo morphis m by $$\ta u:Au t(\mathfrak{ g }, \mathfrak{ t}) {\longrighta rro w}Aut( \P hi) , \ \s igm a\map sto (\ s igma^{ -1}_{|\ma th f rak{t} } )^*.$$R ec allalso that the **W e yl group** of $\P hi$, d en o te d by$W\ subset Aut (\ Phi)$ , is the gr o u p ge n erated by the symmetries $s_ {\alph a} $ ( d efined in ( \[ E Q\_ symmetries\ ])), for $\al pha\in \Phi$.For the follo win g l emma se eThe or e m 7 . 8[@K n app ] and se ct io n 3.1 5 [@ D K ] .
\[ Lem ma\_ 3\] T he map $\tau: Aut (\ma t hfr ak{g} ,\mat hfra k{ t})\t o Aut( \Phi) $is surjective,and$$\tau^{- 1}( W) =Ad (N _G(T) ) \subse t
A ut( \mathfr ak{g},\ m ath fr a k { t} ).$$ Therefore, $\ ta u $ i nduces a n isom o rp hi s m betwee nthe gro u p s $$O ut(\ m at hfrak{g} )\cong Au t( \Phi)/W .$ $
If$\ mat hfr ak{c} \ subs et \ma thfrak{t }^*$i s an open Weyl chamber, then $A u t (\ P hi)= Aut (\Phi,\math frak { c})\ ltim e sW$, where $$Au t( \ Ph i ,\mathfrak{c}):=\{f \i n Aut( \Phi) | f(i\mathfra k{c})=i\ma t h f rak{c}\} .$$M or e over, $Aut(\Ph i,\ma thfrak{c}) $ (hencealso$Out(\ma thfrak{g} ) $ ) is iso mor phi c t o t h e s ymmetry group o f th eDynkindia gram of $\ Phi $.
Th elast part of thele mm aal low s ust o comput e$Ou t( \ma thfra k {g})$for a ll s em is i mpl e compa c tL i e al ge br as.Fir st , itis e n oug h to co nsider si mpl e Lie a lg ebras,since, if $\m at hfrak{g}$de com posesi n to simpl e components as $n_1\ma t hfrak{s }_1 \oplu s \l dots \opl usn_k \m ath f rak{s} _k$, t hen $ $O ut( \ m athfr a k {g })\ co ng S_{n_1} \ l tim es
Ou t( \mat hfrak{s }_1)^{n_1}\times\l d ots \times S_{n_ k}\ ltim e s O ut( \ ma t hfr ak { s}_ k ) ^{n_k}.$$ Furth er, for th es im ple Lie al g ebr as , a gli mpse at thei r Dynkin diagrams revealsth at t h e on ly ones wi th nontr ivial out e r aut o mo rphis m g roup a re : | the_normalizer of_$T$ in $G$. An_element $\sigma\in
Aut(\mathfrak{g},\mathfrak{t})$_induces_a symmetry_of_$\Phi$, and we_denote the resulting_group homomorphism by $$\tau:_Aut(\mathfrak{g},\mathfrak{t}){\longrightarrow}Aut(\Phi), \ \sigma\mapsto_(\sigma^{-1}_{|\mathfrak{t}})^*.$$
Recall_also that the **Weyl group** of $\Phi$, denoted by $W\subset Aut(\Phi)$, is the group_generated_by the_symmetries_$s_{\alpha}$_(defined in (\[EQ\_symmetries\])), for $\alpha\in_\Phi$.
For the following lemma see_Theorem 7.8_[@Knapp] and section 3.15 [@DK].
\[Lemma\_3\] The map $\tau:Aut(\mathfrak{g},\mathfrak{t})\to_Aut(\Phi)$_is surjective, and_$$\tau^{-1}(W)=Ad(N_G(T))\subset
Aut(\mathfrak{g},\mathfrak{t}).$$ Therefore, $\tau$ induces an isomorphism between the groups_$$Out(\mathfrak{g})\cong Aut(\Phi)/W.$$
If $\mathfrak{c}\subset \mathfrak{t}^*$ is an_open Weyl chamber,_then_$Aut(\Phi)=Aut(\Phi,\mathfrak{c})\ltimes_W$, where $$Aut(\Phi,\mathfrak{c}):=\{f\in Aut(\Phi)|_f(i\mathfrak{c})=i\mathfrak{c}\}.$$ Moreover, $Aut(\Phi,\mathfrak{c})$ (hence also $Out(\mathfrak{g})$)_is isomorphic to the symmetry group_of the Dynkin diagram of $\Phi$.
The last_part of the lemma allows us_to compute $Out(\mathfrak{g})$ for all_semisimple compact_Lie algebras. First, it is_enough to consider_simple Lie_algebras, since, if_$\mathfrak{g}$ decomposes into simple components as_$n_1\mathfrak{s}_1\oplus \ldots \oplus_n_k \mathfrak{s}_k$, then $$Out(\mathfrak{g})\cong S_{n_1}\ltimes
Out(\mathfrak{s}_1)^{n_1}\times\ldots \times_S_{n_k}\ltimes_Out(\mathfrak{s}_k)^{n_k}.$$ Further, for_the_simple_Lie algebras,_a glimpse at_their_Dynkin diagrams_reveals_that the only ones with nontrivial_outer_automorphism group are: |
.1497$ $0.0016$ $88.7 $ $1.1 $ $1.6$ $1.57$
2013-09-07 987 Rozhen 2.0m $6542.50530$ $0.00018$ $5.97 $ $0.09 $ $0.1535$ $0.0012$ $88.3 $ $0.8 $ $0.9$ $0.67$
2013-09-07 987 Torun 0.6m $6542.50522$ $0.00052$ $5.89 $ $0.20 $ $0.1515$ $0.0029$ $87.9 $ $1.4 $ $3.5$ $2.33$
2013-10-06 1001 OSN 1.5m $6572.60532$ $0.00018$ $6.11 $ $0.06 $ $0.1465$ $0.0013$ $89.2 $ $0.7 $ $0.9$ $0.63$
2013-11-01 1013 Rozhen 2.0m $6598.40539$ $0.00017$ $6.05 $ $0.06 $ $0.1511$ $0.0010$ $88.9 $ $0.8 $ $0.8$ $0.68$
2013-11-03 1014 OSN 1.5m $6600.55546$ $0.00017$ $6.02 $ $0.05 $ $0.1503$ $0.0009$ $89.2 $ $0.8 $ $1.3$ $1.33$
2013-12-01 1027 OSN 1.5m $6628.50585$ $0.00031$ $6.13 $ $0.09 $ $0.1475$ $0.0022$ $89.2 $ $0.9 $ $1.8 | .1497 $ $ 0.0016 $ $ 88.7 $ $ 1.1 $ $ 1.6 $ $ 1.57 $
2013 - 09 - 07 987 Rozhen 2.0 m $ 6542.50530 $ $ 0.00018 $ $ 5.97 $ $ 0.09 $ $ 0.1535 $ $ 0.0012 $ $ 88.3 $ $ 0.8 $ $ 0.9 $ $ 0.67 $
2013 - 09 - 07 987 Torun 0.6 m $ 6542.50522 $ $ 0.00052 $ $ 5.89 $ $ 0.20 $ $ 0.1515 $ $ 0.0029 $ $ 87.9 $ $ 1.4 $ $ 3.5 $ $ 2.33 $
2013 - 10 - 06 1001 OSN 1.5 m $ 6572.60532 $ $ 0.00018 $ $ 6.11 $ $ 0.06 $ $ 0.1465 $ $ 0.0013 $ $ 89.2 $ $ 0.7 $ $ 0.9 $ $ 0.63 $
2013 - 11 - 01 1013 Rozhen 2.0 m $ 6598.40539 $ $ 0.00017 $ $ 6.05 $ $ 0.06 $ $ 0.1511 $ $ 0.0010 $ $ 88.9 $ $ 0.8 $ $ 0.8 $ $ 0.68 $
2013 - 11 - 03 1014 OSN 1.5 thousand $ 6600.55546 $ $ 0.00017 $ $ 6.02 $ $ 0.05 $ $ 0.1503 $ $ 0.0009 $ $ 89.2 $ $ 0.8 $ $ 1.3 $ $ 1.33 $
2013 - 12 - 01 1027 OSN 1.5 thousand $ 6628.50585 $ $ 0.00031 $ $ 6.13 $ $ 0.09 $ $ 0.1475 $ $ 0.0022 $ $ 89.2 $ $ 0.9 $ $ 1.8 | .1497$ $0.0016$ $88.7 $ $1.1 $ $1.6$ $1.57$
2013-09-07 987 Tozhen 2.0m $6542.50530$ $0.00018$ $5.97 $ $0.09 $ $0.1535$ $0.0012$ $88.3 $ $0.8 $ $0.9$ $0.67$
2013-09-07 987 Torun 0.6m $6542.50522$ $0.00052$ $5.89 $ $0.20 $ $0.1515$ $0.0029$ $87.9 $ $1.4 $ $3.5$ $2.33$
2013-10-06 1001 OSK 1.5m $6572.60532$ $0.00018$ $6.11 $ $0.06 $ $0.1465$ $0.0013$ $89.2 $ $0.7 $ $0.9$ $0.63$
2013-11-01 1013 Roahfn 2.0m $6598.40539$ $0.00017$ $6.05 $ $0.06 $ $0.1511$ $0.0010$ $88.9 $ $0.8 $ $0.8$ $0.68$
2013-11-03 1014 OSN 1.5m $6600.55546$ $0.00017$ $6.02 $ $0.05 $ $0.1503$ $0.0009$ $89.2 $ $0.8 $ $1.3$ $1.33$
2013-12-01 1027 OSG 1.5m $6628.50585$ $0.00031$ $6.13 $ $0.09 $ $0.1475$ $0.0022$ $89.2 $ $0.9 $ $1.8 | .1497$ $0.0016$ $88.7 $ $1.1 $ $1.6$ 987 2.0m $6542.50530$ $5.97 $ $0.09 $0.8 $0.9$ $0.67$ 2013-09-07 Torun 0.6m $6542.50522$ $5.89 $ $0.20 $ $0.1515$ $0.0029$ $ $1.4 $ $3.5$ $2.33$ 2013-10-06 1001 OSN 1.5m $6572.60532$ $0.00018$ $6.11 $ $ $0.1465$ $0.0013$ $89.2 $ $0.7 $ $0.9$ $0.63$ 2013-11-01 1013 Rozhen 2.0m $0.00017$ $ $ $0.0010$ $88.9 $ $0.8 $ $0.8$ $0.68$ 2013-11-03 1014 OSN 1.5m $6600.55546$ $0.00017$ $6.02 $ $0.05 $0.1503$ $0.0009$ $89.2 $ $0.8 $ $1.3$ $1.33$ 1027 OSN 1.5m $6628.50585$ $6.13 $ $0.09 $ $0.1475$ $89.2 $0.9 $ | .1497$ $0.0016$ $88.7 $ $1.1 $ $1.6$ $1.57$
2013-09-07 987 Rozhen 2.0m $6542.50530$ $0.00018$ $5.97 $ $0.09 $ $0.1535$ $0.0012$ $88.3 $ $0.8 $ $0.9$ $0.67$
2013-09-07 987 Torun 0.6m $6542.50522$ $0.00052$ $5.89 $ $0.20 $ $0.1515$ $0.0029$ $87.9 $ $1.4 $ $3.5$ $2.33$
2013-10-06 1001 OSN 1.5m $6572.60532$ $0.00018$ $6.11 $ $0.06 $ $0.1465$ $0.0013$ $89.2 $ $0.7 $ $0.9$ $0.63$
2013-11-01 1013 Rozhen 2.0m $6598.40539$ $0.00017$ $6.05 $ $0.06 $ $0.1511$ $0.0010$ $88.9 $ $0.8 $ $0.8$ $0.68$
2013-11-03 1014 OsN 1.5m $6600.55546$ $0.00017$ $6.02 $ $0.05 $ $0.1503$ $0.0009$ $89.2 $ $0.8 $ $1.3$ $1.33$
2013-12-01 1027 OSN 1.5m $6628.50585$ $0.00031$ $6.13 $ $0.09 $ $0.1475$ $0.0022$ $89.2 $ $0.9 $ $1.8 | .1497$ $0.0016$ $8 8.7 $ $1. 1$ $1. 6$ $1.57$
2 013- 09-07 987 Rozhen 2.0m $ 65 42.50 530$ $ 0. 0 0 018 $ $5. 97 $ $0. 09$ $0 .1535$ $0.0012$ $88.3$ $0.8 $ $ 0.9 $ $0 .6 7$ 2013 -09 -07 987 Toru n 0.6m $654 2 .50522$ $ 0.00 052$ $5.89 $ $ 0. 2 0 $ $0.1515 $ $ 0 .00 29$ $8 7.9 $ $1.4$ $ 3.5 $ $2.33$
2 013-10-06 10 01 O SN 1. 5 m $6 5 72. 60532$ $0 .000 18$ $6. 11 $ $0.06 $ $0.1 465$ $0 . 00 13 $ $ 89. 2 $ $0.7 $ $0.9$ $ 0 . 6 3 $
201 3-11 -01 1013 Rozh en2.0m $6 598.4 0539 $ $0. 00017$ $6 .0 5 $ $0.06 $ $0 .1511$ $ 0. 0010$ $88 .9 $ $0.8 $ $ 0. 8 $ $0.68$
2013-11-0 3 1 014 OS N 1.5m $660 0. 555 46$$0.00 017$ $6.02 $ $0.0 5 $ $0.15 03 $ $0. 0009$ $ 89.2 $ $0.8 $ $1.3$ $ 1 .33$
2013-1 2 -0 1 1027 OSN 1.5m $6 628. 5 05 85$ $0. 00031 $ $ 6 .13 $ $0.09 $ $0 .1475$ $0.0022$ $ 8 9 .2 $ $ 0.9$ $1.8 | .1497$ _ _ _ _$0.0016$_ __ _ _$88.7 $ _$1.1 $ __ $1.6$ $1.57$
2013-09-07 987 _Rozhen_2.0m ___ $6542.50530$ _ $0.00018$ $5.97_$ _ $0.09 $ $0.1535$ __ _ $0.0012$ _ $88.3_$ _$0.8_$_ $0.9$_ $0.67$
2013-09-07 _ 987 _Torun 0.6m _ $6542.50522$ $0.00052$_ $5.89 $ _ $0.20_$ $0.1515$_ _ _ _ $0.0029$ _ _ $87.9 $ _$1.4_$ __$3.5$_ _$2.33$
2013-10-06__ _1001_ OSN 1.5m __ $6572.60532$_ $0.00018$ __$6.11 $ _$0.06 $ _$0.1465$ _ _ _$0.0013$ $89.2_$ $0.7 $ _ $0.9$ _ $0.63$
__2013-11-01 __1013 _ Rozhen 2.0m _ _$6598.40539$ $0.00017$ _$6.05 $ _$0.06_$ $0.1511$ _ _ $0.0010$ ___ _$88.9 $ $0.8 $_ _$0.8$ $0.68$
2013-11-03 __ 1014 OSN 1.5m__ ___$6600.55546$ $0.00017$ _ $6.02 $ $0.05_$ $0.1503$ _ _ $0.0009$ ___ $89.2 $ $0.8_$ _ $1.3$_ $1.33$
2013-12-01 1027 _ OSN 1.5m _ $6628.50585$ $0.00031$ _$6.13 $ $0.09 $ _ $0.1475$ _ _$0.0022$_ ___$89.2_$ _$0.9 $ _ $1.8 |
the only deviation from regularity (see below).
For [NLTT 11748]{}, I recognize that $K_2$ is the radial velocity that was measured since the heavier object is the fainter one. So, $q\approx
0.15/0.71=0.21$, $K_2=271\,{\ensuremath{{\rm km\,s}^{-1}}}$, and $P=5.64\,$hr, which give ${\ensuremath{\Delta t_{\rm LT}}}=4.6\,$s. @sks+10 measured individual eclipse times to $\sim 10$s, making it hard to detect an effect like this. However, this was using 45 s exposures on a 2 m telescope, while the ingress/egress duration was only $\approx 20\,$s. Increasing to 4m or 8m will improve the S/N of individual exposures by a factor of 4–16, and using a cadence better matched to the orbit will help as well, driving eclipse time uncertainties to $\lesssim 1\,$s (as above). This is sufficient to detect [$\Delta t_{\rm LT}$]{}; below I discuss how well one can measure it and what constraints one can get from it.
Comparison With Eccentricity
----------------------------
The above discussion considered circular orbits. For eccentricity $e>0$ the situation changes. I note that the objects in Table \[tab:wd\] have orbits that are consistent with circular orbits, although quantitative limits for $e$ are not always given. This follows from their expected evolutionary histories, where common-envelope evolution [@nvypz00] should have circularized orbits. Nonetheless, in case our understanding of these systems is incorrect or some further evolution (such as interaction with another body) may have caused non-zero eccentricity, I consider the effect of a non-zero eccentricity on our detection of [$\Delta t_{\rm LT}$]{}.
First, there are changes to the expression for [$\Delta t_{\rm LT}$]{} [@fabrycky10]: $${\ensuremath{\Delta t_{\rm LT}}}=({\ensuremath{\Delta t_{\rm LT}}})_{e=0}\times\left( \frac{1-e^2}{1-e^2
\sin^2 \omega}\right)\approx
({\ensuremath{\Delta t_{\rm LT}}})_{e=0}\times\left(1-e^2\cos^2\omega+{\cal O}(e^4)\right)$$ where | the only deviation from regularity (see below).
For [ NLTT 11748 ] { }, I accredit that $ K_2 $ is the radial speed that was measured since the heavier aim is the fainter one. So, $ q\approx
0.15/0.71=0.21 $, $ K_2=271\,{\ensuremath{{\rm km\,s}^{-1}}}$, and $ P=5.64\,$hr, which grant $ { \ensuremath{\Delta t_{\rm LT}}}=4.6\,$s. @sks+10 measured individual eclipse times to $ \sim 10$s, have it hard to detect an effect like this. However, this was use 45 s exposures on a 2 m telescope, while the ingres / egress duration was only $ \approx 20\,$s. increase to 4 m or 8 megabyte will improve the S / N of individual exposure by a factor of 4–16, and using a meter better matched to the orbit will serve as well, driving eclipse clock time uncertainties to $ \lesssim 1\,$s (equally above). This is sufficient to detect [ $ \Delta t_{\rm LT}$ ] { }; below I discuss how well one can quantify it and what constraints one can get from it.
Comparison With Eccentricity
----------------------------
The above discussion considered circular orbits. For eccentricity $ e>0 $ the position changes. I note that the objects in Table \[tab: wd\ ] have orbits that are reproducible with circular orbits, although quantitative limits for $ e$ are not always given. This follows from their expected evolutionary histories, where common - envelope evolution [ @nvypz00 ] should have circularize orbits. Nonetheless, in casing our reason of these systems is incorrect or some further evolution (such as interaction with another soundbox) may have caused non - zero eccentricity, I consider the effect of a non - zero eccentricity on our detection of [ $ \Delta t_{\rm LT}$ ] { }.
foremost, there are changes to the expression for [ $ \Delta t_{\rm LT}$ ] { } [ @fabrycky10 ]: $ $ { \ensuremath{\Delta t_{\rm LT}}}=({\ensuremath{\Delta t_{\rm LT}}})_{e=0}\times\left (\frac{1 - e^2}{1 - e^2
\sin^2 \omega}\right)\approx
({ \ensuremath{\Delta t_{\rm LT}}})_{e=0}\times\left(1 - e^2\cos^2\omega+{\cal O}(e^4)\right)$$ where | thf only deviation from renularity (see below).
For [NNTT 11748]{}, I decognizd that $K_2$ is the radial velocmty rhat qas measured since the heavier lbject iw tht fainter one. So, $q\approx
0.15/0.71=0.21$, $K_2=271\,{\ensurslath{{\xm km\,s}^{-1}}}$, and $P=5.64\,$hr, wmich give ${\etsuremath{\Delta t_{\fm LT}}}=4.6\,$s. @sks+10 measured individual eclipsq times tl $\sim 10$s, making it nwrd fo detect an effect like this. Howeber, thiv was using 45 x exposures on a 2 m telesclpe, ahile the ingress/ehress duratuon rqs only $\apprux 20\,$s. Increasing to 4m ot 8m will improve the S/N of indivkdual exposures bt a xactor of 4–16, end uspng a cadence better katched to the orbit eiln hwlp as well, driving erlipse time uncertaigties to $\neassim 1\,$s (as above). Rhus is sufxicidbt go sevecf [$\Deltw t_{\cm LT}$]{}; below I discuss yow well one can messtgr it and whaf conserwints one can get from it.
Comparison Witv Edcentricity
----------------------------
The above diwcussion considered clrcular owbits. For eccentricity $e>0$ the situation changes. I tote vhxt uhc uvjfcts in Table \[tab:wd\] have orbits that are consiffemt with circular orbits, altnokgn quantitative limits fkr $e$ are not alwayd given. This follows srom their expected evolutionart histories, cheee common-envelope zvolution [@nvvpz00] shpuld nave circularized orbitr. Nohetheless, ij case oud understanding ow tmesa systems is incorrect or fome furtier erolution (sucn as igteraction with another body) may hwve ccused non-zero efcentricity, I consider the efferv of a non-zerp accvntricity on omr detection of [$\Delta t_{\rm LT}$]{}.
Fnrst, thexe are changes tk the eepression fow [$\Delta t_{\rm LD}$]{} [@vabrycky10]: $${\ensnremath{\Dejta r_{\rm OT}}}=({\ensurdoath{\Delta t_{\rm KT}}})_{e=0}\times\lvfu( \frac{1-e^2}{1-e^2
\win^2 \omega}\right)\apprpx
({\evauremath{\Delta t_{\xo LT}}})_{e=0}\times\left(1-e^2\cps^2\ooegw+{\cwl O}(q^4)\sight)$$ where | the only deviation from regularity (see below). 11748]{}, recognize that is the radial the object is the one. So, $q\approx $K_2=271\,{\ensuremath{{\rm km\,s}^{-1}}}$, and $P=5.64\,$hr, which give t_{\rm LT}}}=4.6\,$s. @sks+10 measured individual eclipse times to $\sim 10$s, making it hard detect an effect like this. However, this was using 45 s exposures on 2 telescope, the duration was only $\approx 20\,$s. Increasing to 4m or 8m will improve the S/N of individual by a factor of 4–16, and using a better matched to the will help as well, driving time to $\lesssim (as This sufficient to detect t_{\rm LT}$]{}; below I discuss how well one can measure it and what constraints one can get it. Comparison ---------------------------- The discussion circular For eccentricity $e>0$ changes. I note that the objects have orbits that are consistent with circular orbits, quantitative limits $e$ are not always given. This from their expected evolutionary histories, where common-envelope evolution should have circularized orbits. Nonetheless, in case our understanding of these systems is incorrect or evolution (such as interaction another body) may caused eccentricity, consider effect of non-zero eccentricity on our detection of [$\Delta t_{\rm LT}$]{}. First, there changes to the expression for [$\Delta t_{\rm LT}$]{} [@fabrycky10]: $${\ensuremath{\Delta t_{\rm \frac{1-e^2}{1-e^2 \sin^2 \omega}\right)\approx t_{\rm LT}}})_{e=0}\times\left(1-e^2\cos^2\omega+{\cal O}(e^4)\right)$$ where | the only deviation from regulArity (see beLow).
FoR [NLtT 11748]{}, I ReCognIze tHat $K_2$ is the radiaL VeloCity that was measured sinCe the HeAVier OBjEct is The fainTEr ONE. So, $Q\aPpRox
0.15/0.71=0.21$, $k_2=271\,{\eNSuRematH{{\rm Km\,s}^{-1}}}$, and $P=5.64\,$Hr, which givE ${\enSuRemath{\Delta t_{\RM Lt}}}=4.6\,$s. @sks+10 measuRed Individual ecLipSe timeS tO $\siM 10$S, makiNg iT hard To deteCT an effEct like thIs. hOwever, THis was uSINg 45 S expOsures on a 2 m telescoPE, wHIle the ingress/eGress dUrATiON Was OnlY $\approx 20\,$s. InCrEasinG To 4m or 8m wILl IMPRovE The S/N of indiviDual exposurES by A factoR oF 4–16, anD Using a CadenCe BEttEr matched to The oRbit will hElp as wELl, driviNG eclipsE time uNceRtaIntiES tO $\lEssSiM 1\,$S (as ABoVe). THIs iS sufficiEnT tO deteCt [$\DeLTA T_{\Rm LT}$]{}; BelOw I dIscusS how well one caN meAsurE It aNd whaT consTraiNtS one cAn get fRom it.
coMparison With EccEntrIcity
----------------------------
The aBovE dIscUsSion cONsiderEd cIrcUlar orbIts. For eCCenTrICITy $E>0$ the situation changEs. i NOtE that the ObjectS In taBLe \[tab:wd\] hAvE orBits THAt are ConsIStEnt with cIrculaR OrBiTs, althoUgH quantItAtiVe lImits FOr $e$ aRe not aLways givEn. ThiS Follows from theIR expected evolUTiONArY HistOriEs, where commOn-enVElopE evoLUtIon [@NVypz00] sHould HaVE cIRcularized orbits. NonEtHeless, In casE our understanDing of thesE SYStems is iNcorREcT Or some further eVolutIon (such as iNTeractioN with Another bOdy) may havE CAused non-ZerO ecCenTriCITy, i consider the eFFEct oF a Non-zero EccEntriciTy oN ouR deTecTiOn of [$\Delta T_{\rm LT}$]{}.
FirSt, ThErE aRe cHangeS To the expReSsiOn For [$\delta T_{\Rm LT}$]{} [@faBryckY10]: $${\ensUrEmATh{\DElta t_{\rm lt}}}=({\eNSUremAtH{\DElta T_{\rm lT}}})_{E=0}\timeS\lefT( \FraC{1-e^2}{1-e^2
\sin^2 \oMega}\right)\AppROx
({\enSuReMath{\DelTa t_{\rm LT}}})_{e=0}\times\LeFt(1-e^2\cos^2\omegA+{\cAl O}(E^4)\right)$$ WHEre | the only deviation from r egularity(seebel ow) .
For [NL TT 11748]{}, I reco gnize that $K_2$ is th e rad ia l vel o ci ty th at wasm ea s u red s in ceth e h eavie r o bject i s the fain ter o ne. So, $q\a p pr ox
0.15/0. 71= 0.21$, $K_2= 271 \,{\en su rem a th{{\ rmkm\,s }^{-1} } }$, an d $P=5.64 \, $ hr, wh i ch give $ {\ ensu remath{\Delta t_{ \ rm LT}}}=4.6\,$s. @sks+ 10 me a s ure d i ndividualec lipse times t o $ \ s i m 1 0 $s, making it hard to de t ect an ef fe ctl ike th is. H ow e ver , this wasusin g 45 s ex posure s on a 2 m teles cope,whi lethei ng re ss/ eg r ess du rat i onwas only $ \a pprox 20\ , $ s . Inc rea sing to 4 m or 8m willimp rove the S/Nof in divi du al ex posure s byafactor of 4–16, and using acad en cebe tterm atched to th e orbit will h e lpas w e ll , driving eclipseti m e u ncertain ties t o $ \l e sssim 1\ ,$ s ( as a b o ve).This is suffici ent to de te ct [$\D el ta t_{ \r m L T}$ ]{};b elow I dis cuss how well one can measur e it and whatc on s t ra i ntsone can get fr om i t .
C ompa r is onW ith E ccent ri c it y
------------------ -- ------ --
T he above disc ussion con s i d ered cir cula r o r bits. For ecce ntric ity $e>0$t he situa tionchanges. I note t h a t the ob jec tsinTab l e \ [tab:wd\] hav e orbi ts that a reconsist ent wi thcir cu lar orbit s, altho ug hqu an tit ative limits f or $e $are nota lwaysgiven . Th is f o llo ws from th e i r ex pe ct ed e vol ut ionar y hi s tor ies, wh ere commo n-e n velo pe e volutio n [@nvypz00]sh ould haveci rcu larize d orbits.Nonetheless, in case ou r unders tan dingof t hese syst ems is in cor r ect or somefurth er ev o l ution ( su chas interacti o n wi th an ot herbody) m ay have caused non - zer o eccentricit y,I co n s id ert he eff ec t of a non-zero eccen tricity on o u rdetectiono f [ $\ Delta t _{\rm L T}$]{ } .
Firs t, thereare chang es tot h e e xpressionfor [$\D elta t_{\ r m LT} $ ]{ } [@f abr ycky10 ]: $$ {\ens uremat h {\D eltat_{\rm L T}}}=( {\ens ur emath{\D elta t_{\rm LT}}})_{e=0 }\time s\lef t(\frac{1-e ^2} { 1-e ^2
\sin ^2 \ omega}\rig ht) \ap prox({\ e nsure math { \D elt a t_{\ rm L T }}})_{e=0 } \t ime s \ le ft(1-e^2\co s ^ 2 \om ega+{ \ca l O}(e^ 4)\r ight)$$ where | the_only deviation_from regularity (see below).
For_[NLTT 11748]{}, I_recognize_that $K_2$_is_the radial velocity_that was measured_since the heavier object_is the fainter_one._So, $q\approx
0.15/0.71=0.21$, $K_2=271\,{\ensuremath{{\rm km\,s}^{-1}}}$, and $P=5.64\,$hr, which give ${\ensuremath{\Delta t_{\rm LT}}}=4.6\,$s. @sks+10 measured individual_eclipse_times to_$\sim_10$s,_making it hard to detect_an effect like this. However,_this was_using 45 s exposures on a 2 m_telescope,_while the ingress/egress_duration was only $\approx 20\,$s. Increasing to 4m or_8m will improve the S/N of_individual exposures by_a_factor_of 4–16, and using_a cadence better matched to the_orbit will help as well, driving_eclipse time uncertainties to $\lesssim 1\,$s (as_above). This is sufficient to detect_[$\Delta t_{\rm LT}$]{}; below I_discuss how_well one can measure it_and what constraints_one can_get from it.
Comparison_With Eccentricity
----------------------------
The above discussion considered circular_orbits. For eccentricity_$e>0$ the situation changes. I note_that_the objects in_Table \[tab:wd\]_have_orbits that_are consistent with_circular_orbits, although_quantitative_limits for $e$ are not always_given._This follows from their expected evolutionary histories,_where common-envelope evolution [@nvypz00]_should_have circularized orbits. Nonetheless,_in case our understanding of_these systems is incorrect or some_further evolution_(such as_interaction with another body) may have caused non-zero eccentricity, I consider_the effect of a non-zero eccentricity_on our detection of_[$\Delta t_{\rm_LT}$]{}.
First,_there are changes_to_the expression_for [$\Delta t_{\rm LT}$]{} [@fabrycky10]: $${\ensuremath{\Delta t_{\rm LT}}}=({\ensuremath{\Delta_t_{\rm LT}}})_{e=0}\times\left(_\frac{1-e^2}{1-e^2
\sin^2 \omega}\right)\approx
({\ensuremath{\Delta t_{\rm LT}}})_{e=0}\times\left(1-e^2\cos^2\omega+{\cal_O}(e^4)\right)$$ where |
mathrm{bulge}}-M_B<0.4$ [@Sim86]. We ignore galaxies at the centers of simulated clusters since we have omitted the two central Coma galaxies from the analysis in this paper. Likewise, we exclude from the comparison galaxies that have been stripped-off their entire halo, because the only Coma galaxy that possibly lacks dark matter inside $3\, {r_\mathrm{eff}}$ has been excluded from the analysis in this paper as well (cf. Sec. \[sec:data\]). Isolated field spirals are drawn from objects with $M_{B,\mathrm{bulge}}-M_B>1.56$ in the semi-analytic models [@Sim86].
Simulated galaxies were chosen randomly from the catalogue of @deL07 in a way such that each of six luminosity intervals (between $M_B = -17$ and $M_B = -23$; width $\Delta M_B = 1.0$) contains roughly 50 galaxies. We use dust-corrected luminosities $M_B$ of the semi-analytic models.
Dark matter density
-------------------
Dark matter halos of simulated galaxies are reconstructed from tabulated virial velocities $v_\mathrm{vir}$, virial radii $r_\mathrm{vir}$, and maximum circular velocities $v_\mathrm{max}$ as follows. It is assumed that the halos can be approximated by an NFW-profile (cf. equation \[nfwdef\]), in which case the circular velocity profile reads $$\label{nfwcirc}
\left( \frac{v_\mathrm{circ}(r)}{v_\mathrm{vir}} \right)^2 =
\frac{1}{x} \frac{\ln(1+cx)-cx/(1+cx)}{\ln(1+c) - c/(1+c)}.$$ Here $x=r/r_\mathrm{vir}$ and the halo concentration is defined by $c=r_\mathrm{vir}/r_s$. The maximum circular velocity $v_\mathrm{max}$ of an NFW halo occurs at $r \approx 2 r_\mathrm{vir}/c$ [@Nav96], such that (with equation \[nfwcirc\]) $$4.63 \left( \frac{v_\mathrm{max}}{v_\mathrm{vir}} \right)^2 =
\frac{c}{\ln(1+c)-c/(1+c)}.$$ Using the tabulated $v_\mathrm | mathrm{bulge}}-M_B<0.4 $ [ @Sim86 ]. We ignore galaxies at the centers of simulated bunch since we have exclude the two central Coma galaxies from the psychoanalysis in this paper. besides, we exclude from the comparison galaxy that have been stripped - off their entire aura, because the only Coma galaxy that possibly miss dark matter inside $ 3\, { r_\mathrm{eff}}$ has been excluded from the analysis in this newspaper as well (cf. Sec. \[sec: data\ ]). Isolated field spirals are absorb from objects with $ M_{B,\mathrm{bulge}}-M_B>1.56 $ in the semi - analytic models [ @Sim86 ].
Simulated galaxy were chosen randomly from the catalogue of @deL07 in a way such that each of six luminosity time interval (between $ M_B = -17 $ and $ M_B = -23 $; width $ \Delta M_B = 1.0 $) contains roughly 50 galaxies. We use dust - adjust luminosities $ M_B$ of the semi - analytic models.
Dark matter density
-------------------
Dark matter halos of simulated galaxies are reconstructed from tabulated virial velocities $ v_\mathrm{vir}$, virial radii $ r_\mathrm{vir}$, and maximum circular velocities $ v_\mathrm{max}$ as follows. It is assumed that the halos can be approximate by an NFW - profile (cf. equality \[nfwdef\ ]), in which subject the round velocity profile reads $ $ \label{nfwcirc }
\left (\frac{v_\mathrm{circ}(r)}{v_\mathrm{vir } } \right)^2 =
\frac{1}{x } \frac{\ln(1+cx)-cx/(1+cx)}{\ln(1+c) - c/(1+c)}.$$ Here $ x = r / r_\mathrm{vir}$ and the halo concentration is defined by $ hundred = r_\mathrm{vir}/r_s$. The maximum circular velocity $ v_\mathrm{max}$ of an NFW halo occurs at $ r \approx 2 r_\mathrm{vir}/c$ [ @Nav96 ], such that (with equality \[nfwcirc\ ]) $ $ 4.63 \left (\frac{v_\mathrm{max}}{v_\mathrm{vir } } \right)^2 =
\frac{c}{\ln(1+c)-c/(1+c)}.$$ Using the tabulated $ v_\mathrm | matjrm{bulge}}-M_B<0.4$ [@Sim86]. We ignore galaxies at thg xenterv of sjmulated clusters since we have omitved rhe tqo central Coma galaxids from tje analywis mn this paper. Likewise, wc excmmde fxon the comparispn galaxiev that have bean scripped-off their entire halo, because the onky Coma galaxy trat kosfiblg lacks dark matter inside $3\, {r_\mathrj{eff}}$ hav been excludrd from the analysis in thls pwper as well (cf. Sef. \[sec:data\]). Isilatqe field spirxls are drawn from objgcts with $M_{B,\mathrm{bulge}}-M_B>1.56$ in the semi-cnalytic moeeos [@Dhm86].
Simulated galaqies were chosen randmmly frpm the catalogme of @deO07 in a way such that xach of six luminosijy intervans (between $M_B = -17$ and $M_V = -23$; whdth $\Delgq M_C = 1.0$) cpnfains goujhly 50 galaxjes. We use eust-corrected luminpsynoes $M_B$ of ths semi-wnwlytic models.
Dark matter density
-------------------
Dark mauter galos of simulated galazies are reconstructef from tafulated virial velocities $v_\mathrm{vir}$, virial radii $r_\matiro{vix}$, and oqxlmum circular velocities $v_\mathrm{max}$ as followf. Iu ix assumed that the halos vaj ng approximated by an NFS-profile (cf. equatiln \[nfwdgf\]), in qhich cast the circular velocity profile eeads $$\label{nywcurc}
\left( \frac{v_\mathrl{circ}(r)}{v_\mathxm{vir}} \tight)^2 =
\frac{1}{x} \frac{\ln(1+cx)-cx/(1+cx)}{\ln(1+c) - c/(1+c)}.$$ Gere $x=r/r_\matjrm{vir}$ ans the halo concengranion is defined by $c=r_\mathrm{vir}/w_s$. The maeimum circulxr vglocity $v_\mathrm{mad}$ of an NFW halo occurs wt $r \cpprof 2 r_\mathrm{gir}/c$ [@Nav96], such that (with equatioi \[nfwcirc\]) $$4.63 \legt( \frdc{v_\mathri{max}}{v_\kathrm{vir}} \rigrt)^2 =
\frac{c}{\ln(1+c)-c/(1+c)}.$$ Using che taculated $v_\mzthrm | mathrm{bulge}}-M_B<0.4$ [@Sim86]. We ignore galaxies at the simulated since we omitted the two analysis this paper. Likewise, exclude from the galaxies that have been stripped-off their halo, because the only Coma galaxy that possibly lacks dark matter inside $3\, has been excluded from the analysis in this paper as well (cf. Sec. Isolated spirals drawn objects with $M_{B,\mathrm{bulge}}-M_B>1.56$ in the semi-analytic models [@Sim86]. Simulated galaxies were chosen randomly from the catalogue @deL07 in a way such that each of luminosity intervals (between $M_B -17$ and $M_B = -23$; $\Delta = 1.0$) roughly galaxies. use dust-corrected luminosities of the semi-analytic models. Dark matter density ------------------- Dark matter halos of simulated galaxies are reconstructed from virial velocities radii $r_\mathrm{vir}$, maximum velocities as follows. It that the halos can be approximated (cf. equation \[nfwdef\]), in which case the circular profile reads \left( \frac{v_\mathrm{circ}(r)}{v_\mathrm{vir}} \right)^2 = \frac{1}{x} \frac{\ln(1+cx)-cx/(1+cx)}{\ln(1+c) c/(1+c)}.$$ Here $x=r/r_\mathrm{vir}$ and the halo concentration is by $c=r_\mathrm{vir}/r_s$. The maximum circular velocity $v_\mathrm{max}$ of an NFW halo occurs at $r \approx [@Nav96], such that (with \[nfwcirc\]) $$4.63 \left( \right)^2 \frac{c}{\ln(1+c)-c/(1+c)}.$$ the $v_\mathrm | mathrm{bulge}}-M_B<0.4$ [@Sim86]. We ignore gAlaxies at tHe cenTerS of SiMulaTed cLusters since we HAve oMitted the two central ComA galaXiES froM ThE analYsis in tHIs PAPer. liKeWisE, wE ExClude FroM the comParison galAxiEs That have been STrIpped-off thEir Entire halo, beCauSe the oNlY CoMA galaXy tHat poSsibly LAcks daRk matter iNsIDe $3\, {r_\matHRm{eff}}$ haS BEeN excLuded from the analySIs IN this paper as weLl (cf. SeC. \[sEC:dATA\]). IsOlaTed field spIrAls arE Drawn frOM oBJECts WIth $M_{B,\mathrm{buLge}}-M_B>1.56$ in the sEMi-aNalytiC mOdeLS [@Sim86].
SiMulatEd GAlaXies were choSen rAndomly frOm the cATalogue OF @deL07 in a Way sucH thAt eAch oF SiX lUmiNoSIty INtErvALs (bEtween $M_B = -17$ AnD $M_b = -23$; widtH $\DelTA m_b = 1.0$) ContAinS rouGhly 50 gAlaxies. We use dUst-CorrECteD lumiNositIes $M_b$ oF the sEmi-anaLytic MoDels.
Dark matter dEnsiTy
-------------------
Dark matTer HaLos Of SimulATed galAxiEs aRe reconStructeD FroM tABULaTed virial velocitieS $v_\MAThRm{vir}$, virIal radII $r_\MaTHrm{vir}$, anD mAxiMum cIRCular VeloCItIes $v_\mathRm{max}$ aS FoLlOws. It is AsSumed tHaT thE haLos caN Be apProximAted by an nFW-prOFile (cf. equation \[NFwdef\]), in which cASe THE cIRculAr vElocity profIle rEAds $$\lAbel{NFwCirC}
\Left( \fRac{v_\mAtHRm{CIrc}(r)}{v_\mathrm{vir}} \right)^2 =
\FrAc{1}{x} \fraC{\ln(1+cx)-Cx/(1+cx)}{\ln(1+c) - c/(1+c)}.$$ Here $X=r/r_\mathrm{vIR}$ ANd the halO conCEnTRation is defineD by $c=r_\Mathrm{vir}/r_S$. the maximUm cirCular velOcity $v_\matHRM{max}$ of an nFW HalO ocCurS AT $r \Approx 2 r_\mathrm{VIR}/c$ [@NaV96], sUch that (WitH equatiOn \[nFwcIrc\]) $$4.63 \LefT( \fRac{v_\mathrM{max}}{v_\matHrM{vIr}} \RiGht)^2 =
\Frac{c}{\LN(1+c)-c/(1+c)}.$$ UsinG tHe tAbUlaTed $v_\mAThrm | mathrm{bulge}}-M_B<0.4$ [@ Sim86]. We igno regal ax iesat t he centers ofs imul ated clusters since we have o m itte d t he tw o centr a lC o maga la xie sf ro m the an alysisin this pa per .Likewise, we ex clude from th e comparison ga laxies t hat havebee n str ipped- o ff the ir entire h a lo, be c ause th e on ly C oma galaxy that p o ss i bly lacks dark matte ri ns i d e $ 3\, {r_\mathr m{ eff}} $ has be e ne x c lud e d from the an alysis in t h ispaperas we l l (cf. Sec. \ [ sec :data\]). I sola ted field spira l s are d r awn fro m obje cts wi th $ M _{ B, \ma th r m{b u lg e}} - M_B >1.56$ i nth e sem i-an a l y t ic m ode ls [ @Sim8 6].
Simulate d g alax i eswerechose n ra nd omlyfrom t he ca ta logue of @deL07 ina way suc h t ha t e ac h ofs ix lum ino sit y inter vals (b e twe en $ M _B = -17$ and $M_B = - 2 3 $; width $ \Delta M_ B= 1.0$) c on tai ns r o u ghly50 g a la xies. We use d u st -c orrecte dlumino si tie s $ M_B$o f th e semi -analyti c mod e ls.
Dark matt e r density
--- - -- - - -- - ---- --- -
Dark mat terh alos ofs im ula t ed ga laxie sa re reconstructed fromta bulate d vir ial velocitie s $v_\math r m { vir}$, v iria l r a dii $r_\mathrm {vir} $, and max i mum circ ularvelociti es $v_\ma t h rm{max}$ as fo llo ws. I tis assumed th a t the h alos ca n b e appro xim ate d b y a nNFW-profi le (cf.eq ua ti on \[ nfwde f \]), inwh ich c ase thec ircula r vel ocit ypr o fil e reads $$ \ l abel {n fw circ }
\ le ft( \ frac { v_\ mathrm{ circ}(r)} {v_ \ math rm {v ir}} \r ight)^2 =
\f ra c{1}{x} \f ra c{\ ln(1+c x ) -cx/(1+c x)}{\ln(1+c) - c/(1+c)} . $$ Here $x =r/r_ \mat hrm{vir}$ an d thehal o conce ntrati on is d efi n e d by$ c =r _\m at hrm{vir}/r _ s $.The m ax imum circul ar velocity $v_\ma t hrm {max}$ of anNFW hal o oc cur s a t $r \ a ppr o x 2 r_\mathrm{vi r}/c$ [@Na v9 6 ], such that (wi th equati on \[nf wcirc \ ]) $$4. 63 \left( \frac{v_ \m athr m { max }}{v_\math rm{vir}} \right)^ 2 =
\ f ra c{c}{ \ln (1+c)- c/ (1+ c)}.$ $ Usin g th e tab ulated $ v_\mat hrm | mathrm{bulge}}-M_B<0.4$ [@Sim86]._We ignore_galaxies at the centers_of simulated_clusters_since we_have_omitted the two_central Coma galaxies_from the analysis in_this paper. Likewise,_we_exclude from the comparison galaxies that have been stripped-off their entire halo, because the_only_Coma galaxy_that_possibly_lacks dark matter inside $3\,_{r_\mathrm{eff}}$ has been excluded from_the analysis_in this paper as well (cf. Sec. \[sec:data\]). Isolated_field_spirals are drawn_from objects with $M_{B,\mathrm{bulge}}-M_B>1.56$ in the semi-analytic models [@Sim86].
Simulated_galaxies were chosen randomly from the_catalogue of @deL07_in_a_way such that each_of six luminosity intervals (between $M_B_= -17$ and $M_B = -23$;_width $\Delta M_B = 1.0$) contains roughly_50 galaxies. We use dust-corrected luminosities_$M_B$ of the semi-analytic models.
Dark_matter density
-------------------
Dark_matter halos of simulated galaxies_are reconstructed from_tabulated virial_velocities $v_\mathrm{vir}$, virial_radii $r_\mathrm{vir}$, and maximum circular velocities_$v_\mathrm{max}$ as follows._It is assumed that the halos_can_be approximated by_an_NFW-profile_(cf. equation_\[nfwdef\]), in which_case_the circular_velocity_profile reads $$\label{nfwcirc}
\left( \frac{v_\mathrm{circ}(r)}{v_\mathrm{vir}} \right)^2 =_
\frac{1}{x}_\frac{\ln(1+cx)-cx/(1+cx)}{\ln(1+c) - c/(1+c)}.$$ Here $x=r/r_\mathrm{vir}$ and the_halo concentration is defined_by_$c=r_\mathrm{vir}/r_s$. The maximum circular_velocity $v_\mathrm{max}$ of an NFW_halo occurs at $r \approx 2_r_\mathrm{vir}/c$ [@Nav96],_such that_(with equation \[nfwcirc\]) $$4.63 \left( \frac{v_\mathrm{max}}{v_\mathrm{vir}} \right)^2 =
\frac{c}{\ln(1+c)-c/(1+c)}.$$ Using the_tabulated $v_\mathrm |
}
\frac{x}{\vartheta} = \frac1{\alpha^2} \int_{\mathfrak{t}\in \mathbb{P}_\theta}\exp\left\{\frac{\mathfrak{u}(\mathfrak{t})-\theta}{\alpha}\right\}d\mathfrak{t}$$ depends on the details of the pattern $\mathfrak{u}(\mathfrak{t})$ in the vicinity of its maximum $\theta$. Nevertheless, these details seem not to be too essential; they determine mainly some cofactors of order unity, see also Ref. [@we1].
To justify the latter statements, first, Figure \[F8\] depicts two trajectories $\mathfrak{u}(\mathfrak{t})$ implementing the peak $\mathbb{P}_\theta$. The difference in their forms explains the partial independence of the variable $x$ and $\vartheta$. Second, Figure \[F9\] demonstrates the statistical properties of such trajectories. The shown patterns were obtained in the following way. A collection of random trajectories similar to ones shown in Fig. \[F8\] were generated based on Eq. with the discretization time step of 0.01. All the trajectories started from the point $\mathfrak{u}_u = 1$ and terminated when crossing the boundary $\mathfrak{u}_l = 0$ for the first time. Only the trajectories that passed through the layer $(\theta,\theta+1)$ with $\theta = 10$ without touching the upper boundary $\theta+1 = 11$ were taken into account. Then for each trajectory the time moment $\mathfrak{t}_\text{max}$ of attaining the corresponding maximum $\mathfrak{u}_\text{max}$ was fixed and the trajectory as a whole was shifted along the time axis that the point $\mathfrak{t}_\text{max}$ be located at the time origin $\mathfrak{t}=0$. In this way all the trajectories were rearranged that their maxima be located at the same point on the time axis. The total number of the trajectories constructed in this way was equal to $10^5$. Then the plane $\{\mathfrak{t,u}\}$ was partitioned into cells of $0.1\times 0.1$ size and the discretization points of individual trajectories fell into each cell were counted. Finally their numbers were renormalized to the obtained maximum. The left window in Fig. \[F9\] exhibits the obtained distribution of these values called the distribution pattern of $\mathfrak{u(t)}$. Actually this pattern visualizes the regular trend in the | }
\frac{x}{\vartheta } = \frac1{\alpha^2 } \int_{\mathfrak{t}\in \mathbb{P}_\theta}\exp\left\{\frac{\mathfrak{u}(\mathfrak{t})-\theta}{\alpha}\right\}d\mathfrak{t}$$ depends on the details of the pattern $ \mathfrak{u}(\mathfrak{t})$ in the vicinity of its maximum $ \theta$. however, these detail seem not to be too all-important; they decide mainly some cofactors of club one, see also Ref. [ @we1 ].
To justify the latter statements, foremost, Figure \[F8\ ] depicts two trajectories $ \mathfrak{u}(\mathfrak{t})$ implementing the peak $ \mathbb{P}_\theta$. The remainder in their forms explains the partial independence of the variable $ x$ and $ \vartheta$. Second, Figure \[F9\ ] demonstrates the statistical property of such trajectories. The shown patterns were prevail in the following way. A collection of random trajectory similar to one shown in Fig. \[F8\ ] were generated based on Eq. with the discretization meter step of 0.01. All the trajectories started from the point $ \mathfrak{u}_u = 1 $ and terminated when crossing the boundary $ \mathfrak{u}_l = 0 $ for the first time. Only the trajectories that passed through the layer $ (\theta,\theta+1)$ with $ \theta = 10 $ without touching the upper boundary $ \theta+1 = 11 $ were taken into account. Then for each trajectory the time here and now $ \mathfrak{t}_\text{max}$ of attaining the corresponding maximal $ \mathfrak{u}_\text{max}$ was situate and the trajectory as a whole was careen along the time axis that the distributor point $ \mathfrak{t}_\text{max}$ be located at the time origin $ \mathfrak{t}=0$. In this way all the trajectory were rearranged that their maxima be located at the same point on the time axis. The total number of the trajectories constructed in this way was equal to $ 10 ^ 5$. Then the airplane $ \{\mathfrak{t, u}\}$ was partitioned into cell of $ 0.1\times 0.1 $ size and the discretization point of individual trajectories fell into each cell were count. Finally their numbers were renormalized to the prevail maximum. The left window in Fig. \[F9\ ] exhibits the receive distribution of these values called the distribution form of $ \mathfrak{u(t)}$. Actually this pattern visualize the regular tendency in the | }
\fraf{x}{\vartheta} = \frac1{\alpha^2} \ikt_{\mathfrak{t}\in \majhvb{P}_\theva}\exp\lert\{\frac{\maghfrak{u}(\mathfrak{t})-\theta}{\alpha}\rijht\}d\nathfeak{t}$$ depends on the degails of nhe patteen $\methfrak{u}(\mathfrak{v})$ in the vicinjby of mts maximum $\theja$. Neverthelass, these detahlr deem not to be too essential; they dqterminr lainly some cosactpws or order unity, see also Ref. [@we1].
To jusfify tht latter statementx, first, Figure \[F8\] depicts twl trwjectories $\mathfran{u}(\mathfrak{t})$ impjwmenting the peak $\mathbb{P}_\theta$. The difference in their forms explxins che partial ibdeogndence of tie varpable $x$ and $\vartheta$. Vecond, Gigure \[F9\] demonsbratev tye statistical propervies of such trajectjries. The skown patterns were obrauned hn tve fuolodinf xay. A colpecvion of ransom trajectiries similar to onts fyown in Fig. \[F8\] sere gqnqrated based on Eq. with the discretizatpon fime step of 0.01. All the teajectories started ftom the poynt $\mathfrak{u}_u = 1$ and terminated when crossing the bounvafy $\nabhfrxj{u}_p = 0$ for the first time. Only the trajectories egau psssed through bhe layer $(\theta,\theya+1)$ wojh $\theta = 10$ witfout toucging the upper boujdary $\treta+1 = 11$ were tahen onto account. Then for each rrajectory tke rime moment $\mathfrck{t}_\text{max}$ oy attaoning the corresponding maxioum $\jathfrak{u}_\tedt{max}$ was wixed and the trxjebtorf as a whole was shifted ajong the vime cxis thag thg point $\mathfrak{t}_\hext{max}$ be located at thf timg orighn $\mathfran{t}=0$. In this way all the trajectocmes were reartanced that thzir mawima be located at the same ppint on the gime axis. Nhe total number of ehe trajectorhgs constructev in this way was equal gu $10^5$. Then the plsne $\{\mathfgah{t,u}\}$ qas partitioned inbo ceums of $0.1\times 0.1$ sivt abd the discretiaatkon pliits os individual drajdctuties well inuj cacf cekl were counted. Finanly fheir numbers were rcnormalizgd to the obtained maxomum. The left windlw in Fij. \[F9\] exhobijs the obtained distribution of these vapuef called the disbribotion pattexn of $\mathfrak{u(t)}$. Actually this pattern vmsualizes the regular jrend in the | } \frac{x}{\vartheta} = \frac1{\alpha^2} \int_{\mathfrak{t}\in \mathbb{P}_\theta}\exp\left\{\frac{\mathfrak{u}(\mathfrak{t})-\theta}{\alpha}\right\}d\mathfrak{t}$$ depends details the pattern in the vicinity these seem not to too essential; they mainly some cofactors of order unity, also Ref. [@we1]. To justify the latter statements, first, Figure \[F8\] depicts two $\mathfrak{u}(\mathfrak{t})$ implementing the peak $\mathbb{P}_\theta$. The difference in their forms explains the partial of variable and Second, Figure \[F9\] demonstrates the statistical properties of such trajectories. The shown patterns were obtained in following way. A collection of random trajectories similar ones shown in Fig. were generated based on Eq. the time step 0.01. the started from the $\mathfrak{u}_u = 1$ and terminated when crossing the boundary $\mathfrak{u}_l = 0$ for the first time. Only trajectories that the layer with = without touching the $\theta+1 = 11$ were taken into each trajectory the time moment $\mathfrak{t}_\text{max}$ of attaining corresponding maximum was fixed and the trajectory as whole was shifted along the time axis that point $\mathfrak{t}_\text{max}$ be located at the time origin $\mathfrak{t}=0$. In this way all the trajectories that their maxima be at the same on time The number of trajectories constructed in this way was equal to $10^5$. Then the $\{\mathfrak{t,u}\}$ was partitioned into cells of $0.1\times 0.1$ size and points individual trajectories fell each cell were counted. their were renormalized to the The window exhibits obtained of these values called distribution pattern of $\mathfrak{u(t)}$. Actually pattern visualizes the regular | }
\frac{x}{\vartheta} = \frac1{\alpha^2} \int_{\Mathfrak{t}\iN \mathBb{P}_\TheTa}\Exp\lEft\{\fRac{\mathfrak{u}(\maTHfraK{t})-\theta}{\alpha}\right\}d\mathFrak{t}$$ DePEnds ON tHe detAils of tHE pATTerN $\mAtHfrAk{U}(\MaThfraK{t})$ iN the vicInity of its MaxImUm $\theta$. NeverTHeLess, these dEtaIls seem not to Be tOo esseNtIal; THey deTerMine mAinly sOMe cofaCtors of orDeR Unity, sEE also ReF. [@WE1].
TO jusTify the latter statEMeNTs, first, Figure \[F8\] DepictS tWO tRAJecTorIes $\mathfraK{u}(\MathfRAk{t})$ implEMeNTINg tHE peak $\mathbb{P}_\tHeta$. The diffERenCe in thEiR foRMs explAins tHe PArtIal independEnce Of the variAble $x$ aND $\vartheTA$. Second, figure \[f9\] deMonStraTEs ThE stAtIStiCAl ProPErtIes of sucH tRaJectoRies. tHE SHown PatTernS were Obtained in the FolLowiNG waY. A colLectiOn of RaNdom tRajectOries SiMilar to ones showN in FIg. \[F8\] were geNerAtEd bAsEd on EQ. With thE diScrEtizatiOn time sTEp oF 0.01. ALL THe Trajectories starteD fROM tHe point $\mAthfraK{U}_u = 1$ AnD TerminatEd WheN croSSIng thE bouNDaRy $\mathfrAk{u}_l = 0$ foR ThE fIrst timE. ONly the TrAjeCtoRies tHAt paSsed thRough the Layer $(\THeta,\theta+1)$ with $\tHEta = 10$ without touCHiNG ThE UppeR boUndary $\theta+1 = 11$ Were TAken Into ACcOunT. then fOr eacH tRAjECtory the time moment $\mAtHfrak{t}_\Text{mAx}$ of attaining The correspONDIng maximUm $\maTHfRAk{u}_\text{max}$ was fIxed aNd the trajeCTory as a wHole wAs shifteD along the TIMe axis thAt tHe pOinT $\maTHFrAk{t}_\text{max}$ be lOCAted At The time OriGin $\mathFraK{t}=0$. IN thIs wAy All the traJectorieS wErE rEaRraNged tHAt their mAxIma Be LocAted aT The samE poinT on tHe TiME axIs. The toTAl NUMber Of ThE traJecToRies cOnstRUctEd in thiS way was eqUal TO $10^5$. TheN tHe Plane $\{\maThfrak{t,u}\}$ was paRtItioned intO cEllS of $0.1\timES 0.1$ Size and tHe discretization points oF IndividUal TrajeCtorIes fell inTo eAch celL weRE countEd. FinaLly thEiR nuMBErs weRE ReNorMaLized to the OBTaiNed maXiMum. THe left wIndow in Fig. \[F9\] exhibitS The Obtained distrIbuTion OF ThEse VAlUEs cAlLEd tHE Distribution patTern of $\mathFrAK{u(T)}$. Actually tHIs pAtTern visUalizes The reGUlar treNd in the | }
\frac{x}{\vartheta} = \f rac1{\alph a^2} \i nt_ {\ math frak {t}\in \mathbb { P}_\ theta}\exp\left\{\frac {\mat hf r ak{u } (\ mathf rak{t}) - \t h e ta} {\ al pha }\ r ig ht\}d \ma thfrak{ t}$$ depen dson the details of the patte rn$\mathfrak{u }(\ mathfr ak {t} ) $ inthe vici nity o f its m aximum $\ th e ta$. N e verthel e s s, the se details seem n o tt o be too essen tial;th e yd e ter min e mainly s om e cof a ctors o f o r d e r u n ity, see also Ref. [@we1 ] .
To jus ti fyt he lat ter s ta t eme nts, first, Fig ure \[F8\ ] depi c ts twot rajecto ries $ \ma thf rak{ u }( \m ath fr a k{t } )$ im p lem enting t he p eak $ \mat h b b { P}_\ the ta$. Thedifference in th eirf orm s exp lains the p artia l inde pende nc e of the variab le $ x$ and $\ var th eta $. Seco n d, Fig ure \[ F9\] de monstra t esth e s ta tistical propertie so f s uch traj ectori e s. T h e shownpa tte rnsw e re ob tain e din the f ollowi n gwa y. A co ll ection o f r and om tr a ject oriessimilarto on e s shown in Fig . \[F8\] wereg en e r at e d ba sed on Eq. wi th t h e di scre t iz ati o n tim e ste po f0 .01. All the trajec to ries s tarte d from the po int $\math f r a k{u}_u = 1$a nd terminated whe n cro ssing theb oundary$\mat hfrak{u} _l = 0$ f o r the fir sttim e.Onl y th e trajectorie s that p assed t hro ugh the la yer $( \th et a,\theta+ 1)$ with $ \t he ta =10$ w i thout to uc hin gthe uppe r bound ary $ \the ta +1 = 1 1$ were ta k e n in to a ccou nt. T hen f or e a chtraject ory the t ime mome nt $ \mathfr ak{t}_\text{m ax }$ of atta in ing the c o r respondi ng maximum $\mathfrak{u } _\text{ max }$ wa s fi xed and t hetrajec tor y as awholewas s hi fte d along t he ti me axis that t hepoint $ \mat hfrak{t }_\text{max}$ be l o cat ed at the tim e o rigi n $\ mat h fr a k{t }= 0 $.I n this way all t he traject or i es were rear r ang ed that t heir ma ximab e locat ed at the same poi nt ont h e t ime axis.The tota l numbero f the tr aject ori es con st ruc ted i n this way wasequalto $10^5 $. Th en the pla ne $\{\mathfrak{t,u}\}$ was p artit ion ed into c ell s of $0.1\tim es 0 .1$ size a ndthe disc ret i zatio n po i nt s o f indi vidu a l traject o ri esf e ll into eachc e l l w ere c oun t ed. Fi nall y their numbers w e re renormalize d to t heobt a ined m aximum. The le ftwi n d ow in Fi g. \[F9\] exh ibits th eo btain ed dis tribut ion oft h es e value s ca lle d the dis tri bu t ion pat te rn of $\m athf ra k{u(t) }$. Ac t uall y this pattern vis ualiz e s ther egu lar t re nd in t h e | }
\frac{x}{\vartheta} =_\frac1{\alpha^2} _\int_{\mathfrak{t}\in \mathbb{P}_\theta}\exp\left\{\frac{\mathfrak{u}(\mathfrak{t})-\theta}{\alpha}\right\}d\mathfrak{t}$$ depends on_the details_of_the pattern_$\mathfrak{u}(\mathfrak{t})$_in the vicinity_of its maximum_$\theta$. Nevertheless, these details_seem not to_be_too essential; they determine mainly some cofactors of order unity, see also Ref. [@we1].
To justify_the_latter statements,_first,_Figure \[F8\]_depicts two trajectories $\mathfrak{u}(\mathfrak{t})$ implementing_the peak $\mathbb{P}_\theta$. The difference_in their_forms explains the partial independence of the variable_$x$_and $\vartheta$. Second,_Figure \[F9\] demonstrates the statistical properties of such trajectories. The_shown patterns were obtained in the_following way. A_collection_of_random trajectories similar to_ones shown in Fig. \[F8\] were generated_based on Eq. with the discretization_time step of 0.01. All the trajectories_started from the point $\mathfrak{u}_u =_1$ and terminated when crossing_the boundary_$\mathfrak{u}_l = 0$ for the_first time. Only_the trajectories_that passed through_the layer $(\theta,\theta+1)$ with $\theta =_10$ without touching_the upper boundary $\theta+1 = 11$_were_taken into account._Then_for_each trajectory_the time moment_$\mathfrak{t}_\text{max}$_of attaining_the_corresponding maximum $\mathfrak{u}_\text{max}$ was fixed and_the_trajectory as a whole was shifted along_the time axis that_the_point $\mathfrak{t}_\text{max}$ be located_at the time origin $\mathfrak{t}=0$._In this way all the trajectories_were rearranged_that their_maxima be located at the same point on the time axis._The total number of the trajectories_constructed in this way_was equal_to_$10^5$. Then the_plane_$\{\mathfrak{t,u}\}$ was_partitioned into cells of $0.1\times 0.1$ size_and the_discretization points of individual trajectories fell_into each cell were_counted._Finally their numbers were renormalized to_the obtained maximum. The left window_in Fig. \[F9\] exhibits the obtained_distribution_of_these values called the distribution_pattern of $\mathfrak{u(t)}$. Actually this pattern_visualizes the regular_trend in the |
$, may be systematically affected by the echelle-order blaze residuals imprinted on the continuum in the UVES (slit-mode) observations. Differences in model atmospheres and hydrogen line-broadening theory used may also explain part of the offset (see Paper I).
![Comparison between our spectroscopic $T_{\rm eff}$-scale, the photometric values obtained from $v-y$ and $V-I$, and the results from six other studies. []{data-label="fig:other"}](0051fig4.ps){width="6.5cm"}
Surface gravities
-----------------
We derive surface gravities for the target stars using the customary relation between effective temperature, luminosity, mass, and surface gravity. Luminosities are calculated from the apparent visual magnitude $V$ and the @Alonso99 calibration for bolometric correction, which is given as a function of metallicity, \[Fe/H\][^6], and $T_{\rm eff}$. The metallicity is set to $-2.0$ for all stars. The distance modulus of NGC 6397 is assumed to be 12.57. Stellar masses are inferred from a 13.5Gyr isochrone of the cluster [@Richard05], which places the stars in the mass range $0.78\rm\,M_{\sun}-0.79\rm\,M_{\sun}$. With these values, the surface gravity found for TO stars at $T_{\rm eff}=6\,250\,$K is $\log{g}=3.96$ and $\log{g}=3.40$ for bRGB stars at $T_{\rm eff}=5\,450\,$K. We use a single, averaged value of $\log{g}$ for all stars for which we have derived the same $T_{\rm eff}$.\
The aim of this study is to draw conclusions about abundance differences between stars. We are therefore mainly interested in the accuracy with which we can determine surface gravities on the relative, rather than the absolute scale. In this respect, the effective temperatures have the largest, albeit small, influence on the surface gravity values. A rise in effective temperature of 100K corresponds to an increase in logarithmic surface gravity of approximately 0.03dex. This can be compared to an increase in stellar mass by $0.01\rm\,M_{\sun}$ that propagates into a rise in $\log{g}$ by 0.005dex. We set a constant metallicity, to avoid circular arguments as regards the existence | $, may be systematically affected by the echelle - order blaze residuals imprint on the continuum in the UVES (slit - mood) observations. Differences in model standard atmosphere and hydrogen line - broadening hypothesis used may also explain region of the offset (see Paper I).
! [ Comparison between our spectroscopic $ T_{\rm eff}$-scale, the photometric value obtained from $ v - y$ and $ vanadium - I$, and the result from six other studies. [ ] { data - label="fig: other"}](0051fig4.ps){width="6.5 cm " }
Surface gravities
-----------------
We deduce surface gravities for the target ace using the customary sexual intercourse between effective temperature, luminosity, mass, and open gravity. Luminosities are calculated from the apparent visual magnitude $ V$ and the @Alonso99 calibration for bolometric discipline, which is given as a function of metallicity, \[Fe / H\][^6 ], and $ T_{\rm eff}$. The metallicity is set to $ -2.0 $ for all stars. The distance modulus of NGC 6397 is assumed to be 12.57. Stellar multitude are inferred from a 13.5Gyr isochrone of the cluster [ @Richard05 ], which places the stars in the mass range $ 0.78\rm\,M_{\sun}-0.79\rm\,M_{\sun}$. With these values, the surface gravity found for TO star at $ T_{\rm eff}=6\,250\,$K is $ \log{g}=3.96 $ and $ \log{g}=3.40 $ for bRGB stars at $ T_{\rm eff}=5\,450\,$K. We use a individual, averaged value of $ \log{g}$ for all stars for which we have derived the same $ T_{\rm eff}$.\
The purpose of this study is to draw conclusions about abundance differences between asterisk. We are therefore mainly interested in the accuracy with which we can determine surface gravities on the relative, quite than the absolute scale. In this respect, the effective temperatures have the big, albeit small, influence on the open gravity values. A rise in effective temperature of 100 K corresponds to an increase in logarithmic surface graveness of approximately 0.03dex. This can be compared to an increase in stellar mass by $ 0.01\rm\,M_{\sun}$ that propagates into a rise in $ \log{g}$ by 0.005dex. We set a constant metallicity, to avoid round arguments as regards the existence | $, maj be systematically affegted by the echelle-ordec blaze residuaus imprinted on the continuul un tht UVES (slit-mode) obsdrvations. Differebces un model avjosphercf ans hydxojen line-broadenlng theory gsed may also axolcin part of the offset (see Paper I).
![Coiparisom hetween our spgctroxsopid $T_{\rm eff}$-scale, the photometric valhes obteined from $v-y$ amd $V-I$, and the results from six other studies. []{datw-label="fig:otyer"}](0051fyt4.ps){width="6.5cm"}
Sufface gravpcies
-----------------
We deribe surface gravities for the tafget xtars usint rhe wustomary rxlatiog between efnvctive demperayure, luminositn, masv, abd surface gravity. Lukinosities are calsulated fsoj the apparent viwuql macnitgde $X$ ana tge @Amonso99 falmbration fod bolometrix correction, which os tiven as a fuhction os metallicity, \[Fe/H\][^6], and $T_{\rm eff}$. The metalnicjty is set to $-2.0$ for all wtars. The distance mofulus of GGC 6397 is assumed to be 12.57. Stellar masses are inferreg froj a 13.5Ynv isuxhgone of the cluster [@Richard05], which places the ffats in the mass rcnge $0.78\rm\,M_{\sun}-0.79\rm\,M_{\sum}$. Aiyr these valuer, the surrace gravity found for TO starw at $T_{\rm tff}=6\,250\,$K os $\log{g}=3.96$ and $\log{g}=3.40$ for bRGB srars at $T_{\rm vff}=5\,450\,$K. We use a single, areraged valuz of $\lpg{g}$ fpr all stars for which ce havs derived tje same $T_{\do eff}$.\
The aim of ghix vtudy is to draw conclusiogs about ebundcnce difweremces bqtween stags. We are therefore mainpy injerestad in the wccuracy with which we can deteckine surface gsavpties on che rekative, rather than the absokute sccle. In this respvct, the ehfective tem[eratures hava the largest, albeit fmalo, indluence un the surface gravity najues. A rise in effectiye teolerature of 100K cirrtsponds to an invrexse ij koddrithmic surxace grxfity uf approximctdly 0.03cex. This can be compdred to an increase in sbellar maws by $0.01\rm\,I_{\sun}$ that prolagates into a rist in $\lmg{g}$ by 0.005drx. Re set a constant metallicity, fo avoid firgular argumenes aw regards thz existence | $, may be systematically affected by the residuals on the in the UVES atmospheres hydrogen line-broadening theory may also explain of the offset (see Paper I). between our spectroscopic $T_{\rm eff}$-scale, the photometric values obtained from $v-y$ and $V-I$, the results from six other studies. []{data-label="fig:other"}](0051fig4.ps){width="6.5cm"} Surface gravities ----------------- We derive surface for target using customary relation between effective temperature, luminosity, mass, and surface gravity. Luminosities are calculated from the apparent magnitude $V$ and the @Alonso99 calibration for bolometric which is given as function of metallicity, \[Fe/H\][^6], and eff}$. metallicity is to for stars. The distance of NGC 6397 is assumed to be 12.57. Stellar masses are inferred from a 13.5Gyr isochrone of cluster [@Richard05], the stars the range With these values, gravity found for TO stars at $\log{g}=3.96$ and $\log{g}=3.40$ for bRGB stars at $T_{\rm We use single, averaged value of $\log{g}$ for stars for which we have derived the same eff}$.\ The aim of this study is to draw conclusions about abundance differences between stars. therefore mainly interested in accuracy with which can surface on relative, rather the absolute scale. In this respect, the effective temperatures have the albeit small, influence on the surface gravity values. A rise temperature 100K corresponds to increase in logarithmic surface of 0.03dex. This can be an in $0.01\rm\,M_{\sun}$ propagates a rise in $\log{g}$ 0.005dex. We set a constant to avoid circular arguments | $, may be systematically affectEd by the echElle-oRdeR blAzE resIduaLs imprinted on tHE conTinuum in the UVES (slit-modE) obseRvATionS. diFfereNces in mODeL ATmoSpHeRes AnD HyDrogeN liNe-broadEning theorY usEd May also explaIN pArt of the ofFseT (see Paper I).
![CoMpaRison bEtWeeN Our spEctRoscoPic $T_{\rm EFf}$-scalE, the photoMeTRic valUEs obtaiNED fRom $v-Y$ and $V-I$, and the resulTS fROm six other studIes. []{datA-lABeL="FIg:oTheR"}](0051fig4.ps){widtH="6.5cM"}
SurfACe graviTIeS
-----------------
wE DerIVe surface gravIties for the TArgEt starS uSinG The cusTomarY rELatIon between eFfecTive tempeRature, LUminosiTY, mass, anD surfaCe gRavIty. LUMiNoSitIeS Are CAlCulATed From the aPpArEnt viSual MAGNItudE $V$ aNd thE @AlonSo99 calibration For BoloMEtrIc corRectiOn, whIcH is giVen as a FunctIoN of metallicity, \[FE/H\][^6], anD $T_{\rm eff}$. ThE meTaLliCiTy is sET to $-2.0$ for All StaRs. The diStance mODulUs OF ngC 6397 Is assumed to be 12.57. StellAr MASsEs are infErred fROm A 13.5GYR isochroNe Of tHe clUSTer [@RiCharD05], WhIch placeS the stARs In The mass RaNge $0.78\rm\,M_{\SuN}-0.79\rm\,m_{\suN}$. With THese Values, The surfaCe graVIty found for TO sTArs at $T_{\rm eff}=6\,250\,$K iS $\LoG{G}=3.96$ AnD $\Log{g}=3.40$ For BRGB stars at $t_{\rm eFF}=5\,450\,$K. We Use a SInGle, AVeragEd valUe OF $\lOG{g}$ for all stars for whiCh We have DerivEd the same $T_{\rm eFf}$.\
The aim of THIS study is To drAW cONclusions about AbundAnce differENces betwEen stArs. We are Therefore MAInly inteResTed In tHe aCCUrAcy with which wE CAn deTeRmine suRfaCe graviTieS on The RelAtIve, rather Than the aBsOlUtE sCalE. In thIS respect, ThE efFeCtiVe temPEraturEs havE the LaRgESt, aLbeit smALl, INFlueNcE oN the SurFaCe graVity VAluEs. A rise In effectiVe tEMperAtUrE of 100K corResponds to an iNcRease in logArIthMic surFACe gravitY of approximately 0.03dex. This CAn be comParEd to aN incRease in stEllAr mass By $0.01\rM\,m_{\sun}$ thAt propAgateS iNto A RIse in $\LOG{g}$ By 0.005dEx. we set a consTANt mEtallIcIty, tO avoid cIrcular arguments as REgaRds the existenCe | $, may be systematically a ffected by theech ell e- orde r bl aze residualsi mpri nted on the continuumin th eU VES( sl it-mo de) obs e rv a t ion s. D iff er e nc es in mo del atm ospheres a ndhy drogen line- b ro adening th eor y used may a lso expla in pa r t ofthe offs et (se e Paper I).
![C om p arison between o ur spe ctroscopic $T_{\r m e f f}$-scale, the photo me t ri c val ues obtainedfr om $v - y$ and$ V- I $ , an d the resultsfrom six ot h erstudie s. [] { data-l abel= "f i g:o ther"}](005 1fig 4.ps){wid th="6. 5 cm"}
S u rface g raviti es--- ---- - -- -- --- --
We de riv e su rface gr av it ies f or t h e t arge t s tars usin g the customa ryrela t ion betw een e ffec ti ve te mperat ure,lu minosity, mass, and surfacegra vi ty. L umino s itiesare ca lculate d fromt heap p a r en t visual magnitude $ V $ a nd the @ Alonso 9 9ca l ibration f orbolo m e triccorr e ct ion, whi ch isg iv en as a f un ctionof me tal licit y , \[ Fe/H\] [^6], an d $T_ { \rm eff}$. The metallicity i s s e t t o $-2 .0$ for all st ars. Thedist a nc e m o dulus of N GC 63 9 7 is assumed to be12 .57. S tella r masses areinferred f r o m a 13.5G yr i s oc h rone of the cl uster [@Richard 0 5], whic h pla ces thestars int h e mass r ang e $ 0.7 8\r m \ ,M _{\sun}-0.79\ r m \,M_ {\ sun}$.Wit h these va lue s,the s urface gr avity fo un dfo rTOstars at $T_{\ rm ef f} =6\ ,250\ , $K is$\log {g}= 3. 96 $ an d $\log { g} = 3 .40$ f or bRG B s ta rs at $T_ { \rm eff}=5 \,450\,$K . W e use a s ingle,averaged valu eof $\log{g }$ fo r alls t ars forwhich we have derived t h e same$T_ {\rmeff} $.\
The a imof thi s s t udy is to dr aw co nc lus i o ns ab o u tabu nd ance diffe r e nce s bet we en s tars. W e are therefore ma i nly interested i n t he a c c ur acy wi t h w hi c h w e can determine s urface gra vi t ie s on the r e lat iv e, rath er than thea bsolute scale. I n this re sp ect, t heeffectivetemperat ures have the l a rg est,alb eit sm al l,influ ence o n th e sur face g ra vity v alues .A rise i n effective temperature of 10 0K co rre sponds to an inc rease inloga rithmic su rfa cegravi tyo f app roxi m at ely 0.03d ex.T his can b e c omp a r ed to an incr e a s e i n ste lla r massby $ 0.01\rm\,M_{\sun} $ that propagat es i n t o a ri s e in $ \log{g}$ by 0. 005 de x . We setaconstant me tallicit y, to av oid ci rcular argume n t sa s rega rdsthe existenc e | $, may_be systematically_affected by the echelle-order_blaze residuals_imprinted_on the_continuum_in the UVES_(slit-mode) observations. Differences_in model atmospheres and_hydrogen line-broadening theory_used_may also explain part of the offset (see Paper I).
![Comparison between our spectroscopic $T_{\rm_eff}$-scale,_the photometric_values_obtained_from $v-y$ and $V-I$, and_the results from six other_studies. []{data-label="fig:other"}](0051fig4.ps){width="6.5cm"}
Surface_gravities
-----------------
We derive surface gravities for the target stars_using_the customary relation_between effective temperature, luminosity, mass, and surface gravity. Luminosities_are calculated from the apparent visual_magnitude $V$ and_the_@Alonso99_calibration for bolometric correction,_which is given as a function_of metallicity, \[Fe/H\][^6], and $T_{\rm eff}$._The metallicity is set to $-2.0$ for_all stars. The distance modulus of_NGC 6397 is assumed to_be 12.57._Stellar masses are inferred from_a 13.5Gyr isochrone_of the_cluster [@Richard05], which_places the stars in the mass_range $0.78\rm\,M_{\sun}-0.79\rm\,M_{\sun}$. With_these values, the surface gravity found_for_TO stars at_$T_{\rm_eff}=6\,250\,$K_is $\log{g}=3.96$_and $\log{g}=3.40$ for_bRGB_stars at_$T_{\rm_eff}=5\,450\,$K. We use a single, averaged_value_of $\log{g}$ for all stars for which_we have derived the_same_$T_{\rm eff}$.\
The aim of_this study is to draw_conclusions about abundance differences between stars._We are_therefore mainly_interested in the accuracy with which we can determine surface gravities_on the relative, rather than the_absolute scale. In this_respect, the_effective_temperatures have the_largest,_albeit small,_influence on the surface gravity values. A_rise in_effective temperature of 100K corresponds to_an increase in logarithmic_surface_gravity of approximately 0.03dex. This can_be compared to an increase in_stellar mass by $0.01\rm\,M_{\sun}$ that_propagates_into_a rise in $\log{g}$ by_0.005dex. We set a constant metallicity,_to avoid circular_arguments as regards the existence |
. The Clarendon Press, Oxford University Press, New York, 1995.
G.E. Murphy, A new construction of Young’s seminormal representation of the symmetric group, [*J. Algebra*]{} [**69**]{} (1981), 287–297.
V. Reiner, D. Stanton, and P. Webb, Springer’s regular elements over arbitrary fields, [*Math. Proc. Cambridge Philos. Soc.*]{} [**141**]{} (2006), 209–229.
V. Reiner, D. Stanton, and D. White, The cyclic sieving phenomenon, [*J. Combin. Theory Ser. A*]{} [**108**]{} (2004), 17–50.
A.J. Silberger and E.-W. Zink, The characters of the generalized Steinberg representations of finite general linear groups on the regular elliptic set, [*Trans. Amer. Math. Soc.*]{} [**352**]{} (2000), 3339–3356.
R.P. Stanley, Enumerative combinatorics, vol. 2, [*Cambridge Studies in Advanced Mathematics*]{} [**62**]{}, Cambridge University Press, Cambridge, 1999.
R. Steinberg, The representations of $GL(3,q),GL(4,q),PGL(3,q)$, and $PGL(4,q)$, [*[Canadian J. Math. ]{}*]{} [**[3]{}**]{}, (1951), 225–235.
A.M. Vershik and A. Okounkov, A new approach to representation theory of symmetric groups. II. [*J. Math. Sci. (N. Y.)*]{} [**131**]{} (2005), no. 2, 5471–5494.
A.V. Zelevinsky, Representations of finite classical groups: a Hopf algebra approach, [*Lecture Notes in Mathematics*]{} [**869**]{}. Springer-Verlag, Berlin-New York, 1981.
[^1]: Work partially supported by NSF grants DMS-1148634 and DMS-1001933.
[^2]: In checking this equivalence, it is useful to bear in mind that $f^{\lambda^t}(q) = q^{\binom{n}{2}} f^{\lambda}(q^{-1})$, along with the fact that if $\mu \subset \lambda$ with $|\mu|=|\lambda|-1$ and the unique cell of $\lambda/\mu$ lies in row $i$ and column $j$, then $n(\ | . The Clarendon Press, Oxford University Press, New York, 1995.
G.E. Murphy, A new construction of Young ’s seminormal representation of the symmetric group, [ * J. Algebra * ] { } [ * * 69 * * ] { } (1981), 287–297.
V. Reiner, D. Stanton, and P. Webb, Springer ’s even chemical element over arbitrary fields, [ * Math. Proc. Cambridge Philos. Soc. * ] { } [ * * 141 * * ] { } (2006), 209–229.
V. Reiner, D. Stanton, and D. White, The cyclic sieving phenomenon, [ * J. Combin. Theory Ser. A * ] { } [ * * 108 * * ] { } (2004), 17–50.
A.J. Silberger and E.-W. Zink, The characters of the generalized Steinberg representations of finite cosmopolitan linear groups on the regular egg-shaped bent, [ * Trans. Amer. Math. Soc. * ] { } [ * * 352 * * ] { } (2000), 3339–3356.
R.P. Stanley, Enumerative combinatorics, vol. 2, [ * Cambridge Studies in Advanced Mathematics * ] { } [ * * 62 * * ] { }, Cambridge University Press, Cambridge, 1999.
R. Steinberg, The representations of $ GL(3,q),GL(4,q),PGL(3,q)$, and $ PGL(4,q)$, [ * [ Canadian J. Math. ] { } * ] { } [ * * [ 3 ] { } * * ] { }, (1951), 225–235.
A.M. Vershik and A. Okounkov, A new approach to representation hypothesis of symmetric groups. II. [ * J. Math. Sci. (N. Y.) * ] { } [ * * 131 * * ] { } (2005), no. 2, 5471–5494.
A.V. Zelevinsky, Representations of finite classical groups: a Hopf algebra overture, [ * Lecture Notes in Mathematics * ] { } [ * * 869 * * ] { }. Springer - Verlag, Berlin - New York, 1981.
[ ^1 ]: Work partially supported by NSF grants DMS-1148634 and DMS-1001933.
[ ^2 ]: In checking this equivalence, it is utilitarian to bear in mind that $ f^{\lambda^t}(q) = q^{\binom{n}{2 } } f^{\lambda}(q^{-1})$, along with the fact that if $ \mu \subset \lambda$ with $ |\mu|=|\lambda|-1 $ and the unique cellular telephone of $ \lambda/\mu$ lies in row $ i$ and column $ j$, then $ n(\ | . Thf Clarendon Press, Oxford University Press, New Bork, 1995.
G.E. Murphy, X new construction of Young’s swminoemal representation of the symmvtric groyp, [*J. Qlgebra*]{} [**69**]{} (1981), 287–297.
T. Reiner, D. Stahbon, aud P. Webb, Springgr’s regular alements over drcicrary fields, [*Math. Proc. Cambridge Phijos. Soc.*]{} [**141**]{} (2006), 209–229.
V. Reiner, D. Stwntom, and D. White, The cyclic sieving phenomsnon, [*J. Bombin. Theory Ser. A*]{} [**108**]{} (2004), 17–50.
A.J. Silberger and E.-W. Zlnk, Hhe characters of hhe generaluzed Wteinberg reoresentations of finitg general linear groups on the rdgulax elliptic wer, [*Tgdns. Amer. Mavh. Soc.*]{} [**352**]{} (2000), 3339–3356.
R.P. Stanlen, Enumesative vombinatorics, yol. 2, [*Rambeidge Studies in Advaiced Mathematics*]{} [**62**]{}, Caibridge Utirersity Press, Cambridte, 1999.
R. Stginbesg, Tfw rdprtseitafions lf $JL(3,q),GL(4,q),PGL(3,q)$, znd $PGL(4,q)$, [*[Cabadian J. Math. ]{}*]{} [**[3]{}**]{}, (1951), 225–235.
A.M. Vqgxhik and A. Oiounkod, W new approach to representation theory of symmetric groups. II. [*J. Nath. Sci. (N. Y.)*]{} [**131**]{} (2005), no. 2, 5471–5494.
A.V. Zelevinshy, Representations of finite classical groups: a Hmpf ameebxq appfiafh, [*Lecture Notes in Mathematics*]{} [**869**]{}. Springer-Verlwf, Neglin-New York, 1981.
[^1]: Worh partially skpljrted by NSF erants DMA-1148634 and DMS-1001933.
[^2]: In checklng thif equuvalence, yt ix useful to bear in mind thqt $f^{\lambda^t}(q) = q^{\binom{n}{2}} f^{\lambda}(q^{-1})$, clong with tke facj that if $\mu \subset \lambda$ wich $|\mu|=|\lzmbda|-1$ and tje unique zell of $\lambda/\mu$ livs it row $i$ and column $j$, then $g(\ | . The Clarendon Press, Oxford University Press, 1995. Murphy, A construction of Young’s group, Algebra*]{} [**69**]{} (1981), V. Reiner, D. and P. Webb, Springer’s regular elements arbitrary fields, [*Math. Proc. Cambridge Philos. Soc.*]{} [**141**]{} (2006), 209–229. V. Reiner, D. and D. White, The cyclic sieving phenomenon, [*J. Combin. Theory Ser. A*]{} [**108**]{} 17–50. Silberger E.-W. The characters of the generalized Steinberg representations of finite general linear groups on the regular elliptic [*Trans. Amer. Math. Soc.*]{} [**352**]{} (2000), 3339–3356. R.P. Enumerative combinatorics, vol. 2, Studies in Advanced Mathematics*]{} [**62**]{}, University Cambridge, 1999. Steinberg, representations $GL(3,q),GL(4,q),PGL(3,q)$, and $PGL(4,q)$, J. Math. ]{}*]{} [**[3]{}**]{}, (1951), 225–235. A.M. Vershik and A. Okounkov, A new approach to representation theory symmetric groups. Math. Sci. Y.)*]{} (2005), 2, 5471–5494. A.V. of finite classical groups: a Hopf Notes in Mathematics*]{} [**869**]{}. Springer-Verlag, Berlin-New York, 1981. Work partially by NSF grants DMS-1148634 and DMS-1001933. In checking this equivalence, it is useful to in mind that $f^{\lambda^t}(q) = q^{\binom{n}{2}} f^{\lambda}(q^{-1})$, along with the fact that if $\mu \subset $|\mu|=|\lambda|-1$ and the unique of $\lambda/\mu$ lies row and $j$, $n(\ | . The Clarendon Press, Oxford UnIversity PrEss, NeW YoRk, 1995.
G.e. MUrphY, A neW construction oF younG’s seminormal representaTion oF tHE symMEtRic grOup, [*J. AlgEBrA*]{} [**69**]{} (1981), 287–297.
v. reiNeR, D. staNtON, aNd P. WeBb, SPringer’S regular elEmeNtS over arbitraRY fIelds, [*Math. PRoc. cambridge PhiLos. soc.*]{} [**141**]{} (2006), 209–229.
V. ReInEr, D. sTantoN, anD D. WhiTe, The cYClic siEving phenOmENon, [*J. CoMBin. TheoRY seR. A*]{} [**108**]{} (2004), 17–50.
A.J. silberger and E.-W. ZinK, thE Characters of thE generAlIZeD sTeiNbeRg represenTaTions OF finite GEnERAL liNEar groups on thE regular ellIPtiC set, [*TrAnS. AmER. Math. SOc.*]{} [**352**]{} (2000), 3339–3356.
R.P. STaNLey, enumerative CombInatorics, Vol. 2, [*CamBRidge StUDies in ADvanceD MaTheMatiCS*]{} [**62**]{}, CAmBriDgE uniVErSitY preSs, CambriDgE, 1999.
R. steinBerg, tHE REpreSenTatiOns of $gL(3,q),GL(4,q),PGL(3,q)$, and $pGL(4,Q)$, [*[CanADiaN J. MatH. ]{}*]{} [**[3]{}**]{}, (1951), 225–235.
A.M. VeRshiK aNd A. OkOunkov, a new aPpRoach to represenTatiOn theory oF syMmEtrIc GroupS. iI. [*J. MatH. ScI. (N. Y.)*]{} [**131**]{} (2005), No. 2, 5471–5494.
A.V. ZelEvinsky, rEprEsENTAtIons of finite classiCaL GRoUps: a Hopf AlgebrA ApPrOAch, [*LectuRe notEs in mAThemaTics*]{} [**869**]{}. sPrInger-VerLag, BerLIn-neW York, 1981.
[^1]: WoRk PartiaLlY suPpoRted bY nSF gRants DmS-1148634 and DMS-1001933.
[^2]: in cheCKing this equivaLEnce, it is usefuL To BEAr IN minD thAt $f^{\lambda^t}(q) = Q^{\binOM{n}{2}} f^{\lAmbdA}(Q^{-1})$, aLonG With tHe facT tHAt IF $\mu \subset \lambda$ with $|\Mu|=|\Lambda|-1$ And thE unique cell of $\Lambda/\mu$ liES IN row $i$ and ColuMN $j$, THen $n(\ | . The Clarendon Press, Oxf ord Univer sityPre ss, N ew Y ork, 1995.
G.E. M u rphy , A new construction o f You ng ’ s se m in ormal repres e nt a t ion o fthe s y mm etric gr oup, [* J. Algebra *]{ }[**69**]{} ( 1 98 1), 287–29 7.
V. Reiner,D.Stanto n, an d P. W ebb , Spr inger’ s regul ar elemen ts over a r bitrary f ie lds, [*Math. Proc. Ca m br i dge Philos. So c.*]{} [ * *1 4 1 **] {}(2006), 20 9– 229.V. Rein e r, D . St a nton, and D.White, Thec ycl ic sie vi ngp henome non,[* J . C ombin. Theo ry S er. A*]{} [**10 8 **]{} ( 2 004), 1 7–50.
A. J.Silb e rg er an dE .-W . Z ink , Th e charac te rs of t he g e n e r aliz edStei nberg representati ons off ini te ge neral lin ea r gro ups on there gular ellipticset, [*Trans. Am er . M at h. So c .*]{}[** 352 **]{} ( 2000),3 339 –3 3 5 6 .
R.P. Stanley, Enu me r a ti ve combi natori c s, v o l. 2, [* Ca mbr idge S tudie s in Ad vanced M athema t ic s* ]{} [** 62 **]{}, C amb rid ge Un i vers ity Pr ess, Cam bridg e , 1999.
R. St e inberg, The r e pr e s en t atio nsof $GL(3,q) ,GL( 4 ,q), PGL( 3 ,q )$, and $ PGL(4 ,q ) $, [*[Canadian J. Math .]{}*]{ } [** [3]{}**]{}, ( 1951), 225 – 2 3 5.
A.M. Ver s hi k and A. Okounk ov, A new appro a ch to re prese ntationtheory of s ymmetric gr oup s.II. [ *J . Math. Sci.( N . Y. )* ]{} [** 131 **]{} ( 200 5), no . 2 ,5471–5494 .
A.V.Ze le vi ns ky, Repr e sentatio ns of f ini te cl a ssical grou ps:aHo p f a lgebraa pp r o ach, [ *L ectu reNo tes i n Ma t hem atics*] {} [**869 **] { }. S pr in ger-Ver lag, Berlin-N ew York, 198 1.
[ ^1]: W o r k partia lly supported by NSF gr a nts DMS -11 48634 and DMS-1001 933 .
[^2 ]:I n chec king t his e qu iva l e nce,i t i s u se ful to bea r inmindth at $ f^{\lam bda^t}(q) = q^{\bi n om{ n}{2}} f^{\la mbd a}(q ^ { -1 })$ , a l ong w i tht h e fact that if$\mu \subs et \l ambda$ wit h $| \m u|=|\la mbda|-1 $ and the uni que cellof $\lamb da /\mu $ lie s in row $ i$ and c olumn $j$ , then $n (\ | . The_Clarendon Press,_Oxford University Press, New_York, 1995.
G.E._Murphy,_A new_construction_of Young’s seminormal_representation of the_symmetric group, [*J. Algebra*]{}_[**69**]{} (1981), 287–297.
V._Reiner,_D. Stanton, and P. Webb, Springer’s regular elements over arbitrary fields, [*Math. Proc. Cambridge_Philos._Soc.*]{} [**141**]{}_(2006),_209–229.
V._Reiner, D. Stanton, and D._White, The cyclic sieving phenomenon,_[*J. Combin._Theory Ser. A*]{} [**108**]{} (2004), 17–50.
A.J. Silberger and_E.-W._Zink, The characters_of the generalized Steinberg representations of finite general linear_groups on the regular elliptic set,_[*Trans. Amer. Math._Soc.*]{}_[**352**]{}_(2000), 3339–3356.
R.P. Stanley, Enumerative_combinatorics, vol. 2, [*Cambridge Studies in_Advanced Mathematics*]{} [**62**]{}, Cambridge University Press,_Cambridge, 1999.
R. Steinberg, The representations of $GL(3,q),GL(4,q),PGL(3,q)$,_and $PGL(4,q)$, [*[Canadian J. Math. ]{}*]{}_[**[3]{}**]{}, (1951), 225–235.
A.M. Vershik and_A. Okounkov,_A new approach to representation_theory of symmetric_groups. II._[*J. Math. Sci._(N. Y.)*]{} [**131**]{} (2005), no. 2,_5471–5494.
A.V. Zelevinsky, Representations_of finite classical groups: a Hopf_algebra_approach, [*Lecture Notes_in_Mathematics*]{}_[**869**]{}. Springer-Verlag,_Berlin-New York, 1981.
[^1]:_Work_partially supported_by_NSF grants DMS-1148634 and DMS-1001933.
[^2]: In_checking_this equivalence, it is useful to bear_in mind that $f^{\lambda^t}(q)_=_q^{\binom{n}{2}} f^{\lambda}(q^{-1})$, along with_the fact that if $\mu_\subset \lambda$ with $|\mu|=|\lambda|-1$ and the_unique cell_of $\lambda/\mu$_lies in row $i$ and column $j$, then $n(\ |
Subsets and Splits