id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.03823 | Anton Kutsenko A | Anton A Kutsenko | Complete left-tail asymptotic for branching processes with immigration | null | null | null | null | math.PR math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We derive a complete left-tail asymptotic series for the density of the {\it martingale limit} of a Galton-Watson process with immigration. We show that the series converges everywhere, not only for small arguments. This is the first complete result regarding the left tails of branching processes with immigration. A good, quickly computed approximation for the density will also be derived from the series. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:53:24 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Kutsenko",
"Anton A",
""
]
] |
2506.03824 | Maxim Masyukov | Maxim Masyukov, Aleksi Tamminen, Irina Nefedova, Andrey Generalov, Samu-Ville P\"alli, Roman Grigorev, Pouyan Rezapoor, Rui Silva, Juha Mallat, Juha Ala-Laurinaho and Zachary Taylor | Quasioptic, Calibrated, Full 2-port Measurements of Cryogenic Devices under Vacuum in the 220-330 GHz Band | null | null | null | null | physics.ins-det astro-ph.IM cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A quasi-optical (QO) test bench was designed, simulated, and calibrated for characterizing S-parameters of devices in the 220-330 GHz (WR-3.4) frequency range, from room temperature down to 4.8 K. The devices were measured through vacuum windows via focused beam radiation. A de-embedding method employing line-reflect-match (LRM) calibration was established to account for the effects of optical components and vacuum windows. The setup provides all four S-parameters with the reference plane located inside the cryostat, and achieves a return loss of 30 dB with an empty holder. System validation was performed with measurements of cryogenically cooled devices, such as bare silicon wafers and stainless-steel frequency-selective surface (FSS) bandpass filters, and superconducting bandpass FSS fabricated in niobium. A permittivity reduction of Si based on 4-GHz resonance shift was observed concomitant with a drop in temperature from 296 K to 4.8 K. The stainless steel FSS measurements revealed a relatively temperature invariant center frequency and return loss level of 263 GHz and 35 dB on average, respectively. Finally, a center frequency of 257 GHz was measured with the superconducting filters, with return loss improved by 7 dB on average at 4.8 K. To the best of our knowledge, this is the first reported attempt to scale LRM calibration to 330 GHz and use it to de-embed the impact of optics and cryostat from cryogenically cooled device S-parameters. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:53:27 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Masyukov",
"Maxim",
""
],
[
"Tamminen",
"Aleksi",
""
],
[
"Nefedova",
"Irina",
""
],
[
"Generalov",
"Andrey",
""
],
[
"Pälli",
"Samu-Ville",
""
],
[
"Grigorev",
"Roman",
""
],
[
"Rezapoor",
"Pouyan",
""
],
[
"Silva",
"Rui",
""
],
[
"Mallat",
"Juha",
""
],
[
"Ala-Laurinaho",
"Juha",
""
],
[
"Taylor",
"Zachary",
""
]
] |
2506.03825 | Stefan Purkhart | Stefan Purkhart, Hannah Collier, Laura A. Hayes, Astrid M. Veronig, Miho Janvier, S\"am Krucker | Spatiotemporal evolution of UV pulsations and their connection to 3D magnetic reconnection and particle acceleration | null | null | null | null | astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quasi-periodic pulsations (QPPs) are a common feature of impulsive solar flare emissions, yet their driving mechanism(s) remain unresolved. We study long-period (3.4 min) QPPs in the M3.7 flare on Feb 24, 2023, using SDO/AIA and Solar Orbiter/STIX to pinpoint their source and driver. We analyzed UV and HXR emissions from four subregions covering the flare ribbons and footpoints of the erupting filament. We find the QPP characteristics varied significantly across these regions. The strongest UV pulsations, which correlated strongly with HXR emissions, originated from a sequence of flare kernels at the expanding front of the southern flare ribbon. In contrast, another expanding ribbon front showed no pulsations. The spatial coincidence of UV and HXR sources points to non-thermal electrons as a common driver, with the HXR spectrum showing a soft-hard-soft evolution matching the pulsation timescale. We also observed UV pulsations from plasma injections into a separate filament channel, distinct from the main flare kernels. Our results suggest that the spatiotemporal evolution of the pulsations is driven by the propagation of magnetic reconnection in an asymmetric geometry. We propose that slipping reconnection along quasi-separatrix layers is key to generating and moving the UV kernels, potentially modulated by another time-varying process. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:54:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Purkhart",
"Stefan",
""
],
[
"Collier",
"Hannah",
""
],
[
"Hayes",
"Laura A.",
""
],
[
"Veronig",
"Astrid M.",
""
],
[
"Janvier",
"Miho",
""
],
[
"Krucker",
"Säm",
""
]
] |
2506.03826 | Tobias Schwarzinger | Tobias Schwarzinger, Gernot Steindl, Thomas Fr\"uhwirth, Thomas Preindl, Konrad Diwold, Katrin Ehrenm\"uller, Fajar J. Ekaputra | Signals as a First-Class Citizen When Querying Knowledge Graphs | null | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Cyber-Physical Systems (CPSs) tightly integrate computation with physical entities, often generating vast amounts of time series data from thousands of sensors. Although knowledge graphs offer a powerful means to contextualize these data, existing approaches to integrating knowledge graphs with time series data lack a concept to model the continuous temporal values inherent in CPSs. This gap can make expressing computations on the sensor data cumbersome. In this work, we propose the integration of knowledge graphs and signals, a proven concept for modeling temporal values. By treating signals as first-class citizens in query languages, we can enable seamless querying over knowledge graphs and signals. While the knowledge graph captures information on the CPS, signals represent its run-time data from sensors. We discuss the implications of such an approach and propose SigSPARQL, an extension to the SPARQL query language, to demonstrate these concepts. Furthermore, we evaluate the feasibility of implementing SigSPARQL with a prototype and demonstrate the applicability of the query language for a monitoring use case within a CPS. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:54:44 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Schwarzinger",
"Tobias",
""
],
[
"Steindl",
"Gernot",
""
],
[
"Frühwirth",
"Thomas",
""
],
[
"Preindl",
"Thomas",
""
],
[
"Diwold",
"Konrad",
""
],
[
"Ehrenmüller",
"Katrin",
""
],
[
"Ekaputra",
"Fajar J.",
""
]
] |
2506.03827 | Chun Yuan Yuan | Zhenhui Liu, Chunyuan Yuan, Ming Pang, Zheng Fang, Li Yuan, Xue Jiang, Changping Peng, Zhangang Lin, Zheng Luo, Jingping Shao | Multi-objective Aligned Bidword Generation Model for E-commerce Search Advertising | Accepted by SIGIR2025 | null | null | null | cs.CL cs.AI cs.IR | http://creativecommons.org/licenses/by/4.0/ | Retrieval systems primarily address the challenge of matching user queries with the most relevant advertisements, playing a crucial role in e-commerce search advertising. The diversity of user needs and expressions often produces massive long-tail queries that cannot be matched with merchant bidwords or product titles, which results in some advertisements not being recalled, ultimately harming user experience and search efficiency. Existing query rewriting research focuses on various methods such as query log mining, query-bidword vector matching, or generation-based rewriting. However, these methods often fail to simultaneously optimize the relevance and authenticity of the user's original query and rewrite and maximize the revenue potential of recalled ads.
In this paper, we propose a Multi-objective aligned Bidword Generation Model (MoBGM), which is composed of a discriminator, generator, and preference alignment module, to address these challenges. To simultaneously improve the relevance and authenticity of the query and rewrite and maximize the platform revenue, we design a discriminator to optimize these key objectives. Using the feedback signal of the discriminator, we train a multi-objective aligned bidword generator that aims to maximize the combined effect of the three objectives. Extensive offline and online experiments show that our proposed algorithm significantly outperforms the state of the art. After deployment, the algorithm has created huge commercial value for the platform, further verifying its feasibility and robustness. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:57:18 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Liu",
"Zhenhui",
""
],
[
"Yuan",
"Chunyuan",
""
],
[
"Pang",
"Ming",
""
],
[
"Fang",
"Zheng",
""
],
[
"Yuan",
"Li",
""
],
[
"Jiang",
"Xue",
""
],
[
"Peng",
"Changping",
""
],
[
"Lin",
"Zhangang",
""
],
[
"Luo",
"Zheng",
""
],
[
"Shao",
"Jingping",
""
]
] |
2506.03828 | Dhaval Patel Dr | Dhaval Patel, Shuxin Lin, James Rayfield, Nianjun Zhou, Roman Vaculin, Natalia Martinez, Fearghal O'donncha, Jayant Kalagnanam | AssetOpsBench: Benchmarking AI Agents for Task Automation in Industrial Asset Operations and Maintenance | 39 pages, 18 figures | null | null | null | cs.AI cs.MA | http://creativecommons.org/licenses/by/4.0/ | AI for Industrial Asset Lifecycle Management aims to automate complex operational workflows -- such as condition monitoring, maintenance planning, and intervention scheduling -- to reduce human workload and minimize system downtime. Traditional AI/ML approaches have primarily tackled these problems in isolation, solving narrow tasks within the broader operational pipeline. In contrast, the emergence of AI agents and large language models (LLMs) introduces a next-generation opportunity: enabling end-to-end automation across the entire asset lifecycle. This paper envisions a future where AI agents autonomously manage tasks that previously required distinct expertise and manual coordination. To this end, we introduce AssetOpsBench -- a unified framework and environment designed to guide the development, orchestration, and evaluation of domain-specific agents tailored for Industry 4.0 applications. We outline the key requirements for such holistic systems and provide actionable insights into building agents that integrate perception, reasoning, and control for real-world industrial operations. The software is available at https://github.com/IBM/AssetOpsBench. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:57:35 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Patel",
"Dhaval",
""
],
[
"Lin",
"Shuxin",
""
],
[
"Rayfield",
"James",
""
],
[
"Zhou",
"Nianjun",
""
],
[
"Vaculin",
"Roman",
""
],
[
"Martinez",
"Natalia",
""
],
[
"O'donncha",
"Fearghal",
""
],
[
"Kalagnanam",
"Jayant",
""
]
] |
2506.03829 | Manish Kumar | Indranil Biswas, Manish Kumar and A.J. Parameswaran | Isomorphism of \'etale fundamental groups lifts to isomorphism of stratified fundamental group | Final version | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is shown that if a finite generically smooth morphism $f\,:\,Y\,\longrightarrow\, X$ of smooth projective varieties induces an isomorphism of the \'etale fundamental groups, then the induced map of the stratified fundamental groups $\pi_1^{str}(f)\, :\, \pi_1^{str}(Y,\, y)\,\longrightarrow\, \pi_1^{str}(X,\, f(y))$ is also an isomorphism. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:57:54 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Biswas",
"Indranil",
""
],
[
"Kumar",
"Manish",
""
],
[
"Parameswaran",
"A. J.",
""
]
] |
2506.03831 | Ibrahim Ibrahimov | Ibrahim Ibrahimov, Zaink\'o Csaba, G\'abor Gosztolya | Conformer-based Ultrasound-to-Speech Conversion | accepted to Interspeech 2025 | null | null | null | cs.SD cs.MM eess.AS | http://creativecommons.org/licenses/by/4.0/ | Deep neural networks have shown promising potential for ultrasound-to-speech conversion task towards Silent Speech Interfaces. In this work, we applied two Conformer-based DNN architectures (Base and one with bi-LSTM) for this task. Speaker-specific models were trained on the data of four speakers from the Ultrasuite-Tal80 dataset, while the generated mel spectrograms were synthesized to audio waveform using a HiFi-GAN vocoder. Compared to a standard 2D-CNN baseline, objective measurements (MSE and mel cepstral distortion) showed no statistically significant improvement for either model. However, a MUSHRA listening test revealed that Conformer with bi-LSTM provided better perceptual quality, while Conformer Base matched the performance of the baseline along with a 3x faster training time due to its simpler architecture. These findings suggest that Conformer-based models, especially the Conformer with bi-LSTM, offer a promising alternative to CNNs for ultrasound-to-speech conversion. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:58:39 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Ibrahimov",
"Ibrahim",
""
],
[
"Csaba",
"Zainkó",
""
],
[
"Gosztolya",
"Gábor",
""
]
] |
2506.03832 | Omer Moussa | Omer Moussa, Mariya Toneva | Brain-tuned Speech Models Better Reflect Speech Processing Stages in the Brain | Proceedings of Interspeech 2025 | null | null | null | cs.CL cs.SD eess.AS q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | Pretrained self-supervised speech models excel in speech tasks but do not reflect the hierarchy of human speech processing, as they encode rich semantics in middle layers and poor semantics in late layers. Recent work showed that brain-tuning (fine-tuning models using human brain recordings) improves speech models' semantic understanding. Here, we examine how well brain-tuned models further reflect the brain's intermediate stages of speech processing. We find that late layers of brain-tuned models substantially improve over pretrained models in their alignment with semantic language regions. Further layer-wise probing reveals that early layers remain dedicated to low-level acoustic features, while late layers become the best at complex high-level tasks. These findings show that brain-tuned models not only perform better but also exhibit a well-defined hierarchical processing going from acoustic to semantic representations, making them better model organisms for human speech processing. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 10:59:11 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Moussa",
"Omer",
""
],
[
"Toneva",
"Mariya",
""
]
] |
2506.03833 | Stefano Pisoni | Raghavendra Dheeraj Peddinti, Stefano Pisoni, Egor Tiunov, Alessandro Marini, Leandro Aolita | Technical report on a quantum-inspired solver for simulating compressible flows | null | null | null | null | physics.flu-dyn physics.comp-ph quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This document presents a quantum-inspired solver for 2D Euler equations, accepted at the final phase of the Airbus-BWM Group Quantum Computing Challenge (ABQCC) 2024. We tackle the case study of Quantum Solvers for Predictive Aeroacoustic and Aerodynamic modeling tasks. We propose a tensor network based solver that scales polylogarithmically with the mesh size, in both runtime and memory. This provides a promising avenue for tackling the curse of dimensionality that plagues the direct numerical simulations in the field of computational fluid dynamics. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:01:45 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Peddinti",
"Raghavendra Dheeraj",
""
],
[
"Pisoni",
"Stefano",
""
],
[
"Tiunov",
"Egor",
""
],
[
"Marini",
"Alessandro",
""
],
[
"Aolita",
"Leandro",
""
]
] |
2506.03834 | Joonkyung Kim | Joonkyung Kim, Joonyeol Sim, Woojun Kim, Katia Sycara, and Changjoo Nam | Enhancing Safety of Foundation Models for Visual Navigation through Collision Avoidance via Repulsive Estimation | 16 pages, 6 figures | null | null | null | cs.RO cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose CARE (Collision Avoidance via Repulsive Estimation), a plug-and-play module that enhances the safety of vision-based navigation without requiring additional range sensors or fine-tuning of pretrained models. While recent foundation models using only RGB inputs have shown strong performance, they often fail to generalize in out-of-distribution (OOD) environments with unseen objects or variations in camera parameters (e.g., field of view, pose, or focal length). Without fine-tuning, these models may generate unsafe trajectories that lead to collisions, requiring costly data collection and retraining. CARE addresses this limitation by seamlessly integrating with any RGB-based navigation system that outputs local trajectories, dynamically adjusting them using repulsive force vectors derived from monocular depth maps. We evaluate CARE by combining it with state-of-the-art vision-based navigation models across multiple robot platforms. CARE consistently reduces collision rates (up to 100%) without sacrificing goal-reaching performance and improves collision-free travel distance by up to 10.7x in exploration tasks. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:02:15 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Kim",
"Joonkyung",
""
],
[
"Sim",
"Joonyeol",
""
],
[
"Kim",
"Woojun",
""
],
[
"Sycara",
"Katia",
""
],
[
"Nam",
"Changjoo",
""
]
] |
2506.03835 | Jianyuan Yin | Jianyuan Yin, Qianxiao Li | Learning task-specific predictive models for scientific computing | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider learning a predictive model to be subsequently used for a given downstream task (described by an algorithm) that requires access to the model evaluation. This task need not be prediction, and this situation is frequently encountered in machine-learning-augmented scientific computing. We show that this setting differs from classical supervised learning, and in general it cannot be solved by minimizing the mean square error of the model predictions as is frequently performed in the literature. Instead, we find that the maximum prediction error on the support of the downstream task algorithm can serve as an effective estimate for the subsequent task performance. With this insight, we formulate a task-specific supervised learning problem based on the given sampling measure, whose solution serves as a reliable surrogate model for the downstream task. Then, we discretize the empirical risk based on training data, and develop an iterative algorithm to solve the task-specific supervised learning problem. Three illustrative numerical examples on trajectory prediction, optimal control and minimum energy path computation demonstrate the effectiveness of the approach. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:02:53 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Yin",
"Jianyuan",
""
],
[
"Li",
"Qianxiao",
""
]
] |
2506.03836 | Susan Ilani Loubser | S. Ilani Loubser, Adebusola B. Alabi, Matt Hilton, Yin-Zhe Ma, Xin Tang, Narges Hatamkhani, Catherine Cress, Rosalind E. Skelton and S. Andile Nkosi | An independent estimate of H(z) at z = 0.5 from the stellar ages of brightest cluster galaxies | Accepted for publication in MNRAS | null | null | null | astro-ph.CO astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Several cosmological observations (e.g., Cosmic Microwave Background (CMB), Supernovae Type Ia, and local distance ladder measurements such as Cepheids) have been used to measure the global expansion rate of the Universe, i.e., the Hubble constant, $H_{0}$. However, these precision measurements have revealed tensions between different probes that are proving difficult to solve. Independent, robust techniques must be exploited to validate results or mitigate systematic effects. We use the Cosmic Chronometer (CC) method, which leverages the differential age evolution of passive galaxies, to measure $H(z)$, without any assumption of the underlying cosmology. Unlike previous CC studies, we used only brightest cluster galaxies (BCGs), the oldest and most massive galaxies in the Universe, to construct a pure and homogeneous sample. In this work we used a sample of 53 BCGs in massive, Sunyaev-Zel'dovich selected galaxy clusters (0.3 $< z <$ 0.7) with Southern African Large Telescope (SALT) spectroscopic observations. We used optical spectra to measure D4000$_{\rm n}$ of the BCGs to obtain a new direct measurement of $H(z) = 72.1 \pm 33.9(\rm stat) \pm 7.3$(syst) km s$^{-1}$ Mpc$^{-1}$ at $z=0.5$. By using BCGs, we significantly reduced the systematic errors to 10% by minimising the stellar mass and metallicity dependence of the method. The dominant uncertainty, and limitation for our study, is statistical, and we need larger, homogeneous samples of the oldest, most massive galaxies. By using the $Planck$+BAO prior of $\Omega_{m}$ and $\Omega_{\Lambda}$, the projected Hubble constant is $H_{0}$ = $54.6 \pm 25.7(\rm stat) \pm 5.5$(syst) km s$^{-1}$ Mpc$^{-1}$, consistent with both CMB and Cepheid measurements. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:06:29 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Loubser",
"S. Ilani",
""
],
[
"Alabi",
"Adebusola B.",
""
],
[
"Hilton",
"Matt",
""
],
[
"Ma",
"Yin-Zhe",
""
],
[
"Tang",
"Xin",
""
],
[
"Hatamkhani",
"Narges",
""
],
[
"Cress",
"Catherine",
""
],
[
"Skelton",
"Rosalind E.",
""
],
[
"Nkosi",
"S. Andile",
""
]
] |
2506.03837 | Ze-Feng Gao | Xiao-Qi Han, Ze-Feng Gao, Xin-De Wang, Zhenfeng Ouyang, Peng-Jie Guo, Zhong-Yi Lu | HTSC-2025: A Benchmark Dataset of Ambient-Pressure High-Temperature Superconductors for AI-Driven Critical Temperature Prediction | 7 pages, 2 figures | null | null | null | cond-mat.supr-con cond-mat.mtrl-sci cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The discovery of high-temperature superconducting materials holds great significance for human industry and daily life. In recent years, research on predicting superconducting transition temperatures using artificial intelligence~(AI) has gained popularity, with most of these tools claiming to achieve remarkable accuracy. However, the lack of widely accepted benchmark datasets in this field has severely hindered fair comparisons between different AI algorithms and impeded further advancement of these methods. In this work, we present the HTSC-2025, an ambient-pressure high-temperature superconducting benchmark dataset. This comprehensive compilation encompasses theoretically predicted superconducting materials discovered by theoretical physicists from 2023 to 2025 based on BCS superconductivity theory, including the renowned X$_2$YH$_6$ system, perovskite MXH$_3$ system, M$_3$XH$_8$ system, cage-like BCN-doped metal atomic systems derived from LaH$_{10}$ structural evolution, and two-dimensional honeycomb-structured systems evolving from MgB$_2$. The HTSC-2025 benchmark has been open-sourced at https://github.com/xqh19970407/HTSC-2025 and will be continuously updated. This benchmark holds significant importance for accelerating the discovery of superconducting materials using AI-based methods. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:14:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Han",
"Xiao-Qi",
""
],
[
"Gao",
"Ze-Feng",
""
],
[
"Wang",
"Xin-De",
""
],
[
"Ouyang",
"Zhenfeng",
""
],
[
"Guo",
"Peng-Jie",
""
],
[
"Lu",
"Zhong-Yi",
""
]
] |
2506.03838 | Pabitra Barman | Pabitra Barman and Krishnendu Gongopadhyay | On the domination of surface-group representations in $\mathrm{PU}(2,1)$ | 18 pages, 3 figures. Comments welcome! | null | null | null | math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This article explores surface-group representations into the complex hyperbolic group $\mathrm{PU}(2,1)$ and presents domination results for a special class of representations called $T$-bent representations. Let $S_{g,k}$ be a punctured surface of negative Euler characteristic. We prove that for a $T$-bent representation $\rho: \pi_1(S_{g,k}) \rightarrow \mathrm{PU}(2,1)$, there exists a discrete and faithful representation $\rho_0: \pi_1(S_{g,k}) \rightarrow \mathrm{PO}(2,1)$ that dominates $\rho$ in the Bergman translation length spectrum, while preserving the lengths of the peripheral loops. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:15:44 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Barman",
"Pabitra",
""
],
[
"Gongopadhyay",
"Krishnendu",
""
]
] |
2506.03839 | Tobias Pielok | Tobias Pielok, Bernd Bischl, David R\"ugamer | Revisiting Unbiased Implicit Variational Inference | Accepted to ICML 2025 | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | Recent years have witnessed growing interest in semi-implicit variational inference (SIVI) methods due to their ability to rapidly generate samples from complex distributions. However, since the likelihood of these samples is non-trivial to estimate in high dimensions, current research focuses on finding effective SIVI training routines. Although unbiased implicit variational inference (UIVI) has largely been dismissed as imprecise and computationally prohibitive because of its inner MCMC loop, we revisit this method and show that UIVI's MCMC loop can be effectively replaced via importance sampling and the optimal proposal distribution can be learned stably by minimizing an expected forward Kullback-Leibler divergence without bias. Our refined approach demonstrates superior performance or parity with state-of-the-art methods on established SIVI benchmarks. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:16:58 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Pielok",
"Tobias",
""
],
[
"Bischl",
"Bernd",
""
],
[
"Rügamer",
"David",
""
]
] |
2506.03840 | Grischa Liebel | Pragya Verma and Marcos Vinicius Cruz and Grischa Liebel | Differences between Neurodivergent and Neurotypical Software Engineers: Analyzing the 2022 Stack Overflow Survey | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | Neurodiversity describes variation in brain function among people, including common conditions such as Autism spectrum disorder (ASD), Attention deficit hyperactivity disorder (ADHD), and dyslexia. While Software Engineering (SE) literature has started to explore the experiences of neurodivergent software engineers, there is a lack of research that compares their challenges to those of neurotypical software engineers. To address this gap, we analyze existing data from the 2022 Stack Overflow Developer survey that collected data on neurodiversity. We quantitatively compare the answers of professional engineers with ASD (n=374), ADHD (n=1305), and dyslexia (n=363) with neurotypical engineers. Our findings indicate that neurodivergent engineers face more difficulties than neurotypical engineers. Specifically, engineers with ADHD report that they face more interruptions caused by waiting for answers, and that they less frequently interact with individuals outside their team. This study provides a baseline for future research comparing neurodivergent engineers with neurotypical ones. Several factors in the Stack Overflow survey and in our analysis are likely to lead to conservative estimates of the actual effects between neurodivergent and neurotypical engineers, e.g., the effects of the COVID-19 pandemic and our focus on employed professionals. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:17:03 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Verma",
"Pragya",
""
],
[
"Cruz",
"Marcos Vinicius",
""
],
[
"Liebel",
"Grischa",
""
]
] |
2506.03841 | Orlando Luongo | Alessio Belfiglio, Orlando Luongo, Stefano Mancini | Quantum entanglement in cosmology | Invited review on relativistic quantum information, 100 pages, 15 figures | null | null | null | gr-qc astro-ph.CO hep-ph hep-th quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We discuss recent progress in the study of entanglement within cosmological frameworks, focusing on both momentum and position-space approaches and also reviewing the possibility to directly extract entanglement from quantum fields. Entanglement generation in expanding spacetimes can be traced back to the phenomenon of gravitational particle production, according to which the background gravitational field may transfer energy and momentum to quantum fields. The corresponding entanglement amount and its mode dependence are both sensitive to the field statistics and to the details of spacetime expansion, thus encoding information about the background. Gravitational production processes also play a key role in addressing the quantum-to-classical transition of cosmological perturbations. In order to directly extract entanglement from quantum fields, local interactions with additional quantum systems, working as detectors, have been suggested, leading to the formulation of the entanglement harvesting protocol. Despite harvesting procedures are currently unfeasible from an experimental point of view, various proposals for implementation exist and a proper modeling of detectors and local interactions is crucial to address entanglement extraction via realistic setups. In the final part of the work, we address entanglement characterization in position space, primarily focusing on black hole spacetimes. We first investigate a possible interpretation of Bekenstein-Hawking black hole entropy in terms of the entanglement entropy arising in discrete quantum field theories, on account of the area law. Then, we discuss the resolution of the black hole information paradox via the gravitational fine-grained entropy formula, which provides a new way to compute the entropy of Hawking radiation and allows to preserve unitarity in black hole evaporation processes. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:18:41 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Belfiglio",
"Alessio",
""
],
[
"Luongo",
"Orlando",
""
],
[
"Mancini",
"Stefano",
""
]
] |
2506.03842 | ChangJae Lee | ChangJae Lee, Heecheol Yang, Byeonggwon Kim | Improving Post-Processing for Quantitative Precipitation Forecasting Using Deep Learning: Learning Precipitation Physics from High-Resolution Observations | null | null | null | null | physics.ao-ph | http://creativecommons.org/licenses/by/4.0/ | Accurate quantitative precipitation forecasting (QPF) remains one of the main challenges in numerical weather prediction (NWP), primarily due to the difficulty of representing the full complexity of atmospheric microphysics through parameterization schemes. This study introduces a deep learning-based post-processing model, DL-QPF, which diagnoses precipitation fields from meteorological forecasts by learning directly from high-resolution radar estimates precipitation. The DL-QPF model is constructed using a Patch-conditional Generative Adversarial Network (Patch-cGAN) architecture combined with a U-Net generator and a discriminator. The generator learns meteorological features relevant to precipitation, while the adversarial loss from the discriminator encourages the generation of realistic rainfall patterns and distributions. Training is performed on three years of warm-season data over the Korean Peninsula, with input variables derived from ECMWF's Integrated Forecasting System High-Resolution forecast (IFS-HRES). Model verification is conducted against multiple reference models, including global (IFS-HRES, KIM), regional (KIM-Regional, KIM-LENS), and AI-based (GraphCast) forecasts. Verification across multiple rainfall thresholds shows that DL-QPF achieves a frequency bias near one and superior success ratios. Particularly for heavy and intense rainfall events, DL-QPF outperforms both conventional NWP and an AI model, demonstrating improved skill in capturing high-intensity precipitation. This study highlights the potential of observational data-driven deep learning approaches in post-processing QPF. By directly learning from observations, DL-QPF reduces systematic biases and enhances the realism of forecasted rainfall distributions. These results demonstrate the model's potential to enhance QPF realism. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:20:05 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lee",
"ChangJae",
""
],
[
"Yang",
"Heecheol",
""
],
[
"Kim",
"Byeonggwon",
""
]
] |
2506.03843 | Clifford M. Will | Fatemeh Taherasghari and Clifford M. Will | Compact binary systems in Einstein-{\AE}ther gravity. II. Radiation reaction to 2.5 post-Newtonian order | 15 pages, submitted to Phys. Rev. D | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | We obtain the equations of motion for compact binary systems (black holes or neutron stars) in an alternative theory of gravity known as Einstein-AEther theory, which supplements the standard spacetime metric with a timelike four-vector (the AEther field) that is constrained to have unit norm. The equations make use of solutions obtained in Paper I for the gravitational and AEther field potentials within the near zone of the system, evaluated to 2.5 post-Newtonian (PN) order ($O(v/c)^5$ beyond Newtonian gravity), sufficient to obtain the effects of gravitational radiation reaction to the same order as the quadrupole approximation of general relativity. Those potentials were derived by applying the post-Minkowskian method to the field equations of the theory. Using a modified geodesic equation that is a consequence of the effects of the interaction between the AEther field and the internal strong-gravity fields of the compact bodies, we obtain explicit equations of motion in terms of the positions and velocities of the bodies, focussing on the radiation-reaction terms that contribute at 1.5PN and 2.5PN orders. We obtain the rate of energy loss by the system, including the effects of dipole gravitational radiation (conventionally denoted $-1$PN order) and the analogue of quadrupole radiation (denoted $0$PN order). We find significant disagreements with published results, based on calculating the energy flux in the far zone using a ``Noether current'' construction. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:20:24 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Taherasghari",
"Fatemeh",
""
],
[
"Will",
"Clifford M.",
""
]
] |
2506.03844 | Mohammed Mahshook | Mohammed Mahshook and Rudra Banerjee | Beyond Diamond: Interpretable Machine Learning Discovery of Coherent Quantum Defect Hosts in Semiconductors | 12 pages including references, 7 figures | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by-sa/4.0/ | Quantum point defects in wide bandgap semiconductors-such as the nitrogen-vacancy (NV) center in diamond are leading candidates for solid-state spin qubits due to their long coherence times and optically addressable spin states. Yet, identifying host materials capable of supporting such deep-level defects remains a significant challenge, owing to the complex interplay between chemical composition, crystal structure, and electronic environment. Here, we present a scalable, interpretable machine learning framework that combines density functional theory (DFT)-informed descriptors with a structure-agnostic ensemble model to predict quantum-compatible defect-host materials. Trained on a curated dataset from the Materials Project and ICSD, our model achieves a high Matthews correlation coefficient (MCC > 0.95) and assigns confidence scores to guide prioritization of candidates. First-principles calculations of coherence-relevant properties, including static dielectric constants and defect formation energetics, validate key predictions. While high dielectric response is a necessary but not sufficient condition for spin coherence, our model successfully recovers known hosts such as diamond and SiC, and reveals previously overlooked candidates such as WS2 , MgO, CaS and TiO2 . This approach establishes a robust path for discovering next-generation quantum materials, bridging data-driven screening with physically interpretable design principles. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:21:17 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mahshook",
"Mohammed",
""
],
[
"Banerjee",
"Rudra",
""
]
] |
2506.03845 | Rounak Biswas | Rounak Biswas, Falguni Roy | The generalized and pseudo $n$-strong Drazin inverse of the sum of elements in Banach algebras | null | null | null | null | math.FA | http://creativecommons.org/licenses/by/4.0/ | In this paper, we begin by introducing some necessary and sufficient conditions for generalized $n$-strong Drazin invertibility (g$n$s-invertibility) and pseudo $n$-strong Drazin invertibility (p$n$s-invertibility) of an element in a Banach algebra for $n\in\mathbb{N}$. Subsequently, these results are utilized to prove some additive properties of g$n$s (p$n$s)-Drazin inverse in a Banach algebra. This process produces a generalization of some recent results of H Chen, M Sheibani (Linear and Multilinear Algebra \textbf{70.1} (2022): 53-65) for g$n$s and p$n$s-Drazin inverse. Furthermore, we define and characterize weighted g$n$s and weighted p$n$s-Drazin inverse in a Banach algebra. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:22:49 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Biswas",
"Rounak",
""
],
[
"Roy",
"Falguni",
""
]
] |
2506.03846 | Aniello Esposito | Aniello Esposito and Utz-Uwe Haus | SLURM Heterogeneous Jobs for Hybrid Classical-Quantum Workflows | null | null | null | null | cs.DC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | A method for efficient scheduling of hybrid classical-quantum workflows is presented, based on standard tools available on common supercomputer systems. Moderate interventions by the user are required, such as splitting a monolithic workflow in to basic building blocks and ensuring the data flow. This bares the potential to significantly reduce idle time of the quantum resource as well as overall wall time of co-scheduled workflows. Relevant pseudo-code samples and scripts are provided to demonstrate the simplicity and working principles of the method. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:24:10 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Esposito",
"Aniello",
""
],
[
"Haus",
"Utz-Uwe",
""
]
] |
2506.03847 | Michail-Antisthenis Tsompanas | Hugo Alcaraz-Herrera, Michail-Antisthenis Tsompanas, Igor Balaz, Andrew Adamatzky | Designing morphologies of soft medical devices using cooperative neuro coevolution | null | null | 10.1145/3712255.3726671 | null | cs.NE | http://creativecommons.org/licenses/by-sa/4.0/ | Soft robots have proven to outperform traditional robots in applications related to propagation in geometrically constrained environments. Designing these robots and their controllers is an intricate task, since their building materials exhibit non-linear properties. Human designs may be biased; hence, alternative designing processes should be considered. We present a cooperative neuro coevolution approach to designing the morphologies of soft actuators and their controllers for applications in drug delivery apparatus. Morphologies and controllers are encoded as compositional pattern-producing networks evolved by Neuroevolution of Augmented Topologies (NEAT) and in cooperative coevolution methodology, taking into account different collaboration methods. Four collaboration methods are studied: n best individuals, n worst individuals, n best and worst individuals, and n random individuals. As a performance baseline, the results from the implementation of Age-Fitness Pareto Optimisation (AFPO) are considered. The metrics used are the maximum displacement in upward bending and the robustness of the devices in terms of applying to the same evolved morphology a diverse set of controllers. Results suggest that the cooperative neuro coevolution approach can produce more suitable morphologies for the intended devices than AFPO. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:26:55 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Alcaraz-Herrera",
"Hugo",
""
],
[
"Tsompanas",
"Michail-Antisthenis",
""
],
[
"Balaz",
"Igor",
""
],
[
"Adamatzky",
"Andrew",
""
]
] |
2506.03848 | Angsuman Das | Angsuman Das and Arnab Mandal | On Independence Number of Comaximal Subgroup Graph | 13 pages | null | null | null | math.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we establish sharp thresholds on the independence number of the comaximal subgroup graph $\Gamma(G)$ that guarantee solvability, supersolvability, and nilpotency of the underlying group $G$. Specifically: \begin{itemize}
\item For solvability, we prove that any group $G$ with independence number $\alpha(\Gamma(G))\leq 51$ must be solvable, and show that the alternating group $A_5$ is uniquely determined by its graph.
\item For supersolvability, we show that $\alpha(\Gamma(G))\leq 14$ implies $G$ is supersolvable, except for three explicit exceptions.
\item For nilpotency, we prove that $\alpha(\Gamma(G))\leq 6$ ensures nilpotency, except for five groups. \end{itemize} Finally, we conclude with some open issues involving domination parameters. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:28:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Das",
"Angsuman",
""
],
[
"Mandal",
"Arnab",
""
]
] |
2506.03849 | Benjamin Dupuis | Benjamin Dupuis, Dario Shariatian, Maxime Haddouche, Alain Durmus, Umut Simsekli | Algorithm- and Data-Dependent Generalization Bounds for Score-Based Generative Models | null | null | null | null | stat.ML cs.LG | http://creativecommons.org/licenses/by/4.0/ | Score-based generative models (SGMs) have emerged as one of the most popular classes of generative models. A substantial body of work now exists on the analysis of SGMs, focusing either on discretization aspects or on their statistical performance. In the latter case, bounds have been derived, under various metrics, between the true data distribution and the distribution induced by the SGM, often demonstrating polynomial convergence rates with respect to the number of training samples. However, these approaches adopt a largely approximation theory viewpoint, which tends to be overly pessimistic and relatively coarse. In particular, they fail to fully explain the empirical success of SGMs or capture the role of the optimization algorithm used in practice to train the score network. To support this observation, we first present simple experiments illustrating the concrete impact of optimization hyperparameters on the generalization ability of the generated distribution. Then, this paper aims to bridge this theoretical gap by providing the first algorithmic- and data-dependent generalization analysis for SGMs. In particular, we establish bounds that explicitly account for the optimization dynamics of the learning algorithm, offering new insights into the generalization behavior of SGMs. Our theoretical findings are supported by empirical results on several datasets. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:33:04 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Dupuis",
"Benjamin",
""
],
[
"Shariatian",
"Dario",
""
],
[
"Haddouche",
"Maxime",
""
],
[
"Durmus",
"Alain",
""
],
[
"Simsekli",
"Umut",
""
]
] |
2506.03850 | Liang Chen | Liang Chen, Xueting Han, Li Shen, Jing Bai, Kam-Fai Wong | Vulnerability-Aware Alignment: Mitigating Uneven Forgetting in Harmful Fine-Tuning | ICML 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Harmful fine-tuning (HFT), performed directly on open-source LLMs or through Fine-tuning-as-a-Service, breaks safety alignment and poses significant threats. Existing methods aim to mitigate HFT risks by learning robust representation on alignment data or making harmful data unlearnable, but they treat each data sample equally, leaving data vulnerability patterns understudied. In this work, we reveal that certain subsets of alignment data are consistently more prone to forgetting during HFT across different fine-tuning tasks. Inspired by these findings, we propose Vulnerability-Aware Alignment (VAA), which estimates data vulnerability, partitions data into "vulnerable" and "invulnerable" groups, and encourages balanced learning using a group distributionally robust optimization (Group DRO) framework. Specifically, VAA learns an adversarial sampler that samples examples from the currently underperforming group and then applies group-dependent adversarial perturbations to the data during training, aiming to encourage a balanced learning process across groups. Experiments across four fine-tuning tasks demonstrate that VAA significantly reduces harmful scores while preserving downstream task performance, outperforming state-of-the-art baselines. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:33:36 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Chen",
"Liang",
""
],
[
"Han",
"Xueting",
""
],
[
"Shen",
"Li",
""
],
[
"Bai",
"Jing",
""
],
[
"Wong",
"Kam-Fai",
""
]
] |
2506.03851 | Henri Lombardi | Claude Quitt\'e, Henri Lombardi | Algebraic identities to prove that a neat finite free algebra is tracically \'etale | in French language. Keywords: Commutative Algebra, \'Etale algebra, Unramified algebra, Tracically \'etale algebra, Elementary methods, Algebraic identities, Constructive mathematics | null | null | null | math.AC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The central objective of this article is to provide an elementary proof of the following theorem, of which we are unaware of any trace in the existing literature. If $B$ is a net finite free algebra over a commutative ring $A$, then it is tracically \'etale (its trace form is nondegenerate) and a fortiori \'etale over A. As indicated in the title, our proof is based on algebraic identities. This confirms the implicit adage that much of the most abstract commutative algebra is concentrated in algebraic identities concerning matrices of polynomials over an arbitrary commutative ring. -- -- -- L'objectif central de cet article est de donner une d\'emonstration \'el\'ementaire du th\'eor\`eme suivant, dont nous ne connaissons pas de trace dans la litt\'erature existante. Si $B$ est une alg\`ebre libre finie nette sur $A$, alors elle est traciquement \'etale (sa forme trace est non d\'eg\'en\'er\'ee) et \`a fortiori \'etale sur $A$. Comme indiqu\'e dans le titre, notre d\'emonstration est bas\'ee sur des identit\'es alg\'ebriques. Cela confirme l'adage implicite selon lequel une grande partie de l'alg\`ebre commutative la plus abstraite se concentre dans des identit\'es alg\'ebriques concernant les matrices de polyn\^omes sur un anneau commutatif arbitraire. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:33:50 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Quitté",
"Claude",
""
],
[
"Lombardi",
"Henri",
""
]
] |
2506.03852 | Marius Caus | Marius Caus and Musbah Shaat | An OTFS-based Random Access Scheme for GNSS Independent Operation in NTN | 13 pages | null | null | null | eess.SP | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper investigates the random access procedure for non-terrestrial networks operating without global navigation satellite system (GNSS) support. In such scenarios, positioning uncertainties can reach several kilometers, which directly impacts the open-loop compensation mechanisms employed by the user equipment. To ensure that the resulting time and carrier frequency offsets can be handled by the network, the robustness of the standardized random access signal design and detection scheme must be enhanced. To extend radio access capabilities, identical Zadoff Chu (ZC) sequences are concatenated and then modulated into the orthogonal time frequency space (OTFS) modulation. Thanks to the specific characteristics of the OTFS-based random access signal, the received sequences are coherently combined, thereby maximizing the desired signal strength. Additionally, the proposed preamble minimizes the overhead associated with the cyclic prefix (CP) transmission. Numerical evaluations in a regenerative low-Earth-orbit (LEO) satellite scenario show that, despite significant positioning errors, the proposed OTFS random access design attains comparable peak-to-average power ratio (PAPR) and missed detection probability (MDP) to OFDM-based solutions, while improving spectral confinement and reducing overhead. These results demonstrate that the proposed OTFS-based random access design offers a robust and spectrally efficient alternative to OFDM for GNSS-independent NTN access. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:33:53 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Caus",
"Marius",
""
],
[
"Shaat",
"Musbah",
""
]
] |
2506.03853 | Olga Rovenska | Oleksiy Dovgoshey and Olga Rovenska | Labeled Trees Generating Separable and Locally Finite Ultrametrics | null | null | null | null | math.GN | http://creativecommons.org/licenses/by/4.0/ | We analyze the interplay between labeled trees and the ultrametric spaces they present. We provide characterizations of labeled trees that generate separable ultrametric spaces and those that generate locally finite ultrametric spaces. In particular, we establish an analog of K\"onig's Infinity Lemma for locally finite ultrametric spaces generated by labeled trees. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:36:45 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Dovgoshey",
"Oleksiy",
""
],
[
"Rovenska",
"Olga",
""
]
] |
2506.03854 | Polyvios Pratikakis | Emmanouil Anagnostakis, Polyvios Pratikakis | Analysis of Server Throughput For Managed Big Data Analytics Frameworks | null | null | null | null | cs.DC | http://creativecommons.org/licenses/by/4.0/ | Managed big data frameworks, such as Apache Spark and Giraph demand a large amount of memory per core to process massive volume datasets effectively. The memory pressure that arises from the big data processing leads to high garbage collection (GC) overhead. Big data analytics frameworks attempt to remove this overhead by offloading objects to storage devices. At the same time, infrastructure providers, trying to address the same problem, attribute more memory to increase memory per instance leaving cores underutilized. For frameworks, trying to avoid GC through offloading to storage devices leads to high Serialization/Deserialization (S/D) overhead. For infrastructure, the result is that resource usage is decreased. These limitations prevent managed big data frameworks from effectively utilizing the CPU thus leading to low server throughput.
We conduct a methodological analysis of server throughput for managed big data analytics frameworks. More specifically, we examine, whether reducing GC and S/D can help increase the effective CPU utilization of the server. We use a system called TeraHeap that moves objects from the Java managed heap (H1) to a secondary heap over a fast storage device (H2) to reduce the GC overhead and eliminate S/D over data. We focus on analyzing the system's performance under the co-location of multiple memory-bound instances to utilize all available DRAM and study server throughput. Our detailed methodology includes choosing the DRAM budget for each instance and how to distribute this budget among H1 and Page Cache (PC). We try two different distributions for the DRAM budget, one with more H1 and one with more PC to study the needs of both approaches. We evaluate both techniques under 3 different memory-per-core scenarios using Spark and Giraph with native JVM or JVM with TeraHeap. We do this to check throughput changes when memory capacity increases. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:37:51 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Anagnostakis",
"Emmanouil",
""
],
[
"Pratikakis",
"Polyvios",
""
]
] |
2506.03855 | Xiaolong Wang | Xiaolong Wang and Xuerong Yang and Xiaoli Wang and Bo Song | Data-driven balanced truncation for second-order systems via the approximate Gramians | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper studies the data-driven balanced truncation (BT) method for second-order systems based on the measurements in the frequency domain. The basic idea is to approximate Gramians used the numerical quadrature rules, and establish the relationship between the main quantities in the procedure of BT with the sample data, which paves the way for the execution of BT in a nonintrusive manner. We construct the structure-preserving reduced models approximately based on the samples of second-order systems with proportional damping, and provide the detailed execution of the data-driven counterpart of BT in real-value arithmetic. The low-rank approximation to the solution of Sylvester equations is also introduced to speed up the process of the proposed approach when a large amount of samples involved in the modeling. The performance of our approach is illustrated in detail via two numerical examples. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:38:08 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Wang",
"Xiaolong",
""
],
[
"Yang",
"Xuerong",
""
],
[
"Wang",
"Xiaoli",
""
],
[
"Song",
"Bo",
""
]
] |
2506.03856 | Kwanwoo Lee | Kwanwoo Lee, Gyeongjae Park, Jaeheung Park | Phase-based Nonlinear Model Predictive Control for Humanoid Walking Stabilization with Single and Double Support Time Adjustments | 8 pages, 4 figures | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | Balance control for humanoid robots has been extensively studied to enable robots to navigate in real-world environments. However, balance controllers that explicitly optimize the durations of both the single support phase, also known as step timing, and the Double Support Phase (DSP) have not been widely explored due to the inherent nonlinearity of the associated optimization problem. Consequently, many recent approaches either ignore the DSP or adjust its duration based on heuristics or on linearization techniques that rely on sequential coordination of balance strategies. This study proposes a novel phase-based nonlinear Model Predictive Control (MPC) framework that simultaneously optimizes Zero Moment Point~(ZMP) modulation, step location, step timing, and DSP duration to maintain balance under external disturbances. In simulation, the proposed controller was compared with two state-of-the-art frameworks that rely on heuristics or sequential coordination of balance strategies under two scenarios: forward walking on terrain emulating compliant ground and external push recovery while walking in place. Overall, the findings suggest that the proposed method offers more flexible coordination of balance strategies than the sequential approach, and consistently outperforms the heuristic approach. The robustness and effectiveness of the proposed controller were also validated through experiments with a real humanoid robot. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:41:28 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lee",
"Kwanwoo",
""
],
[
"Park",
"Gyeongjae",
""
],
[
"Park",
"Jaeheung",
""
]
] |
2506.03857 | Mingxuan Xia | Mingxuan Xia, Haobo Wang, Yixuan Li, Zewei Yu, Jindong Wang, Junbo Zhao, Runze Wu | Prompt Candidates, then Distill: A Teacher-Student Framework for LLM-driven Data Annotation | Accepted to ACL 2025 (Main conference) | null | null | null | cs.LG cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, Large Language Models (LLMs) have demonstrated significant potential for data annotation, markedly reducing the labor costs associated with downstream applications. However, existing methods mostly adopt an aggressive strategy by prompting LLM to determine a single gold label for each unlabeled sample. Due to the inherent uncertainty within LLMs, they often produce incorrect labels for difficult samples, severely compromising the data quality for downstream applications. Motivated by ambiguity aversion in human behaviors, we propose a novel candidate annotation paradigm wherein large language models are encouraged to output all possible labels when incurring uncertainty. To ensure unique labels are provided for downstream tasks, we develop a teacher-student framework CanDist that distills candidate annotations with a Small Language Model (SLM). We further provide a rigorous justification demonstrating that distilling candidate annotations from the teacher LLM offers superior theoretical guarantees compared to directly using single annotations. Extensive experiments across six text classification tasks validate the effectiveness of our proposed method. The source code is available at https://github.com/MingxuanXia/CanDist. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:42:37 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Xia",
"Mingxuan",
""
],
[
"Wang",
"Haobo",
""
],
[
"Li",
"Yixuan",
""
],
[
"Yu",
"Zewei",
""
],
[
"Wang",
"Jindong",
""
],
[
"Zhao",
"Junbo",
""
],
[
"Wu",
"Runze",
""
]
] |
2506.03858 | Mickael Latocca | Rafik Imekraz (MIA, ULR), Micka\"el Latocca (LaMME) | Almost Sure Uniform Convergence Of Random Hermite Series | null | null | null | null | math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We continue the analysis of random series associated to the multidimensional harmonic oscillator $-\Delta + |x|^2$ on $\mathbb{R}^d$ with d \geq 2$$. More precisely we obtain a necessary and sufficient condition to get the almost sure uniform convergence on the whole space $\mathbb{R}^d$ . It turns out that the same condition gives the almost sure uniform convergence on the sphere $\mathbb{S}^{d-1}$ (despite $\mathbb{S}^{d-1}$ is a zero Lebesgue measure of $\mathbb{R}^d$). From a probabilistic point of view, our proof adapts a strategy used by the first author for boundaryless Riemannian compact manifolds. However, our proof requires sharp off-diagonal estimates of the spectral function of $-\Delta + |x|^2$ . Such estimates are obtained using elementary tools. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:43:05 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Imekraz",
"Rafik",
"",
"MIA, ULR"
],
[
"Latocca",
"Mickaël",
"",
"LaMME"
]
] |
2506.03859 | Bo Fu | Dandan Jiang, Bo Fu, Weiwei Xu | Accelerating Randomized Algorithms for Low-Rank Matrix Approximation | 31 pages, 3 figures | null | null | null | stat.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Randomized algorithms are overwhelming methods for low-rank approximation that can alleviate the computational expenditure with great reliability compared to deterministic algorithms. A crucial thought is generating a standard Gaussian matrix $\mathbf{G}$ and subsequently obtaining the orthonormal basis of the range of $\mathbf{AG}$ for a given matrix $\mathbf{A}$. Recently, the \texttt{farPCA} algorithm offers a framework for randomized algorithms, but the dense Gaussian matrix remains computationally expensive. Motivated by this, we introduce the standardized Bernoulli, sparse sign, and sparse Gaussian matrices to replace the standard Gaussian matrix in \texttt{farPCA} for accelerating computation. These three matrices possess a low computational expenditure in matrix-matrix multiplication and converge in distribution to a standard Gaussian matrix when multiplied by an orthogonal matrix under a mild condition. Therefore, the three corresponding proposed algorithms can serve as a superior alternative to fast adaptive randomized PCA (\texttt{farPCA}). Finally, we leverage random matrix theory (RMT) to derive a tighter error bound for \texttt{farPCA} without shifted techniques. Additionally, we extend this improved error bound to the error analysis of our three fast algorithms, ensuring that the proposed methods deliver more accurate approximations for large-scale matrices. Numerical experiments validate that the three algorithms achieve asymptotically the same performance as \texttt{farPCA} but with lower costs, offering a more efficient approach to low-rank matrix approximation. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:46:45 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Jiang",
"Dandan",
""
],
[
"Fu",
"Bo",
""
],
[
"Xu",
"Weiwei",
""
]
] |
2506.03860 | Gianmassimo Tasinato | Gianmassimo Tasinato | Stochastic inflation as a superfluid | 23 pages | null | null | null | gr-qc astro-ph.CO hep-th | http://creativecommons.org/licenses/by/4.0/ | We point out that inflationary superhorizon fluctuations can be effectively described by a set of equations analogous to those governing a superfluid. This is achieved through a functional Schr\"odinger approach to the evolution of the inflationary wavefunction, combined with a suitable coarse-graining procedure to capture large-scale dynamics. The irrotational fluid velocity is proportional to the gradient of the wavefunction phase. Marginalizing over short superhorizon modes introduces an external force acting on the fluid velocity. The quantum pressure characteristic of the superfluid plays a role in scenarios involving an ultra-slow-roll phase of inflation. Our superfluid framework is consistent with the standard Starobinsky approach to stochastic inflation while offering complementary insights, particularly by providing more precise information on the phase of the inflationary wavefunction. We also discuss a heuristic approach to include dissipative effects in this description. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:46:58 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tasinato",
"Gianmassimo",
""
]
] |
2506.03861 | Qiuhan Han | Qiuhan Han, Qian Wang, Atsushi Yoshikawa, Masayuki Yamamura | PulseReddit: A Novel Reddit Dataset for Benchmarking MAS in High-Frequency Cryptocurrency Trading | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | High-Frequency Trading (HFT) is pivotal in cryptocurrency markets, demanding rapid decision-making. Social media platforms like Reddit offer valuable, yet underexplored, information for such high-frequency, short-term trading. This paper introduces \textbf{PulseReddit}, a novel dataset that is the first to align large-scale Reddit discussion data with high-frequency cryptocurrency market statistics for short-term trading analysis. We conduct an extensive empirical study using Large Language Model (LLM)-based Multi-Agent Systems (MAS) to investigate the impact of social sentiment from PulseReddit on trading performance. Our experiments conclude that MAS augmented with PulseReddit data achieve superior trading outcomes compared to traditional baselines, particularly in bull markets, and demonstrate robust adaptability across different market regimes. Furthermore, our research provides conclusive insights into the performance-efficiency trade-offs of different LLMs, detailing significant considerations for practical model selection in HFT applications. PulseReddit and our findings establish a foundation for advanced MAS research in HFT, demonstrating the tangible benefits of integrating social media. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:48:51 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Han",
"Qiuhan",
""
],
[
"Wang",
"Qian",
""
],
[
"Yoshikawa",
"Atsushi",
""
],
[
"Yamamura",
"Masayuki",
""
]
] |
2506.03862 | Faizan Bhat | Faizan Bhat, Arnab Priya Saha and Aninda Sinha | A stringy dispersion relation for field theory | 29 pages, 9 figures | null | null | null | hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We present a simple, self-contained derivation of the local, parametric crossing symmetric dispersion relation for 2-2 scattering, which is motivated by string theory. In various limits of the parameter, this stringy dispersion relation goes over to various known dispersion relations like fixed-$t$, fixed-$s$, local crossing symmetric etc. We present formulas applicable for any number of subtractions. Several examples are discussed for illustration. In particular, the Veneziano and the Virasoro-Shapiro amplitudes are shown to admit series representations that manifest poles in all channels and converge everywhere. We then discuss applications to weakly-coupled gravitational EFTs, and explain how the parameter can be used to derive bounds on the Wilson coefficients by working in the forward limit ($t=0$), even in the presence of the graviton pole. Finally, our approach also enables us to find series representations for multi-variable, totally symmetric generalisations of the Venziano and Virasoro-Shapiro amplitudes that manifest poles in all the variables. In the future, this may be useful for finding $n$-particle dispersion relations, and we take the first steps in this direction. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:54:06 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Bhat",
"Faizan",
""
],
[
"Saha",
"Arnab Priya",
""
],
[
"Sinha",
"Aninda",
""
]
] |
2506.03863 | Hao Li | Hao Li, Qi Lv, Rui Shao, Xiang Deng, Yinchuan Li, Jianye Hao, Liqiang Nie | STAR: Learning Diverse Robot Skill Abstractions through Rotation-Augmented Vector Quantization | Accepted by ICML 2025 Spotlight | Proceedings of the 42st International Conference on Machine Learning, PMLR 267, 2025 | null | null | cs.RO cs.LG | http://creativecommons.org/licenses/by/4.0/ | Transforming complex actions into discrete skill abstractions has demonstrated strong potential for robotic manipulation. Existing approaches mainly leverage latent variable models, e.g., VQ-VAE, to learn skill abstractions through learned vectors (codebooks), while they suffer from codebook collapse and modeling the causal relationship between learned skills. To address these limitations, we present \textbf{S}kill \textbf{T}raining with \textbf{A}ugmented \textbf{R}otation (\textbf{STAR}), a framework that advances both skill learning and composition to complete complex behaviors. Specifically, to prevent codebook collapse, we devise rotation-augmented residual skill quantization (RaRSQ). It encodes relative angles between encoder outputs into the gradient flow by rotation-based gradient mechanism. Points within the same skill code are forced to be either pushed apart or pulled closer together depending on gradient directions. Further, to capture the causal relationship between skills, we present causal skill transformer (CST) which explicitly models dependencies between skill representations through an autoregressive mechanism for coherent action generation. Extensive experiments demonstrate the superiority of STAR on both LIBERO benchmark and realworld tasks, with around 12\% improvement over the baselines. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:54:42 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Li",
"Hao",
""
],
[
"Lv",
"Qi",
""
],
[
"Shao",
"Rui",
""
],
[
"Deng",
"Xiang",
""
],
[
"Li",
"Yinchuan",
""
],
[
"Hao",
"Jianye",
""
],
[
"Nie",
"Liqiang",
""
]
] |
2506.03864 | Pep Mulet | Juan Barajas-Calonge, Raimund B\"urger, Pep Mulet, Luis-Miguel Villada | Invariant-region-preserving WENO schemes for one-dimensional multispecies kinematic flow models | null | Journal of Computational Physics, Volume 537, 15 September 2025, 114081 | 10.1016/j.jcp.2025.114081 | null | math.NA cs.NA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Multispecies kinematic flow models are defined by systems of N strongly coupled, nonlinear first-order conservation laws, where the solution is a vector of N partial volume fractions or densities. These models arise in various applications including multiclass vehicular traffic and sedimentation of polydisperse suspensions. The solution vector should take values in a set of physically relevant values (i.e., the components are nonnegative and sum up at most to a given maximum value). It is demonstrated that this set, the so-called invariant region, is preserved by numerical solutions produced by a new family of high-order finite volume numerical schemes adapted to this class of models. To achieve this property, and motivated by [X. Zhang, C.-W. Shu, On maximum-principle-satisfying high order schemes for scalar conservation laws, J. Comput. Phys. 229 (2010) 3091--3120], a pair of linear scaling limiters is applied to a high-order weighted essentially non-oscillatory (WENO) polynomial reconstruction to obtain invariant-region-preserving (IRP) high-order polynomial reconstructions. These reconstructions are combined with a local Lax-Friedrichs (LLF) or Harten-Lax-van Leer (HLL) numerical flux to obtain a high-order numerical scheme for the system of conservation laws. It is proved that this scheme satisfies an IRP property under a suitable Courant-Friedrichs-Lewy (CFL) condition. The theoretical analysis is corroborated with numerical simulations for models of multiclass traffic flow and polydisperse sedimentation. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:56:20 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Barajas-Calonge",
"Juan",
""
],
[
"Bürger",
"Raimund",
""
],
[
"Mulet",
"Pep",
""
],
[
"Villada",
"Luis-Miguel",
""
]
] |
2506.03865 | Benjamin Jung | Benjamin Jung, Johannes Bosse, Malte G\"ottsche | Reconstructing North Korea's Plutonium Production History with Bayesian Inference-Based Reprocessing Waste Analysis | null | null | null | null | physics.soc-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Although North Korea's nuclear program has been the subject of extensive scrutiny, estimates of its fissile material stockpiles remain fraught with uncertainty. In potential future disarmament agreements, inspectors may need to use nuclear archaeology methods to verify or gain confidence in a North Korean fissile material declaration. This study explores the potential utility of a Bayesian inference-based analysis of the isotopic composition of reprocessing waste to reconstruct the operating history of the 5 MWe reactor and estimate its plutonium production history. We simulate several scenarios that reflect different assumptions and varying levels of prior knowledge about the reactor. The results show that correct prior assumptions can be confirmed and incorrect prior information (or a false declaration) can be detected. Model comparison techniques can distinguish between scenarios with different numbers of core discharges, a capability that could provide important insights into the early stages of operation of the 5 MWe reactor. Using these techniques, a weighted plutonium estimate can be calculated, even in cases where the number of core discharges is not known with certainty. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:56:24 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Jung",
"Benjamin",
""
],
[
"Bosse",
"Johannes",
""
],
[
"Göttsche",
"Malte",
""
]
] |
2506.03866 | Sumiran Pujari | Sumiran Pujari and Harsh Nigam | {\guillemotleft}Anticommuting{\guillemotright} $\mathbb{Z}_2$ quantum spin liquids | 23 pages, 15 figures, comments welcome | null | null | null | cond-mat.str-el cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We discuss a class of lattice $S=\frac{1}{2}$ quantum Hamiltonians with bond-dependent Ising couplings with a mutually {\guillemotleft}anticommuting{\guillemotright} algebra of extensively many local $\mathbb{Z}_2$ conserved charges that was introduced in [arXiv:2407.06236] including the nomenclature. This mutual algebra is reminiscent of the spin-$\frac{1}{2}$ Pauli matrix algebra but encoded in the structure of \emph{local conserved charges}. These models have finite residual entropy density in the ground state with a simple but non-trivial degeneracy counting and concomitant quantum spin liquidity as proved in [arXiv:2407.06236]. The spin liquidity relies on a geometrically site-interlinked character of the local conserved $\mathbb{Z}_2$ charges that is rather natural in presence of an {\guillemotleft}anticommuting{\guillemotright} structure. One may contrast this with say the bond-interlinked character of the local conserved $\mathbb{Z}_2$ charges on the hexagonal plaquettes of the Kitaev honeycomb spin-$\frac{1}{2}$ model which leads to a mutually commuting local algebra. In this work, we elucidate the differences of this kind of quantum spin liquidity in relation to many-body topological order found in some gapped quantum spin liquids whose canonical example is the Kitaev toric code. The toric code belongs to the more general class of Levin-Wen or string net constructions that possess mutually commuting algebras for the local conserved charges. We will make several exact statements on the kinds of many-body order that can be present within the class of {\guillemotleft}anticommuting{\guillemotright} $\mathbb{Z}_2$ quantum spin liquids co-existing with extensive residual ground state entropy. We will also point out a mutually commuting algebra with local support that are naturally expressed as multi-linear Majorana forms in the Kitaev representation of these quantum spin liquids. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:57:44 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Pujari",
"Sumiran",
""
],
[
"Nigam",
"Harsh",
""
]
] |
2506.03868 | Zhuoyang Pan | Zhuoyang Pan, Boxiao Pan, Guandao Yang, Adam W. Harley, Leonidas Guibas | Animal Pose Labeling Using General-Purpose Point Trackers | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatically estimating animal poses from videos is important for studying animal behaviors. Existing methods do not perform reliably since they are trained on datasets that are not comprehensive enough to capture all necessary animal behaviors. However, it is very challenging to collect such datasets due to the large variations in animal morphology. In this paper, we propose an animal pose labeling pipeline that follows a different strategy, i.e. test time optimization. Given a video, we fine-tune a lightweight appearance embedding inside a pre-trained general-purpose point tracker on a sparse set of annotated frames. These annotations can be obtained from human labelers or off-the-shelf pose detectors. The fine-tuned model is then applied to the rest of the frames for automatic labeling. Our method achieves state-of-the-art performance at a reasonable annotation cost. We believe our pipeline offers a valuable tool for the automatic quantification of animal behavior. Visit our project webpage at https://zhuoyang-pan.github.io/animal-labeling. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 11:59:22 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Pan",
"Zhuoyang",
""
],
[
"Pan",
"Boxiao",
""
],
[
"Yang",
"Guandao",
""
],
[
"Harley",
"Adam W.",
""
],
[
"Guibas",
"Leonidas",
""
]
] |
2506.03869 | Michele Bucelli | Michele Bucelli, Luca Dede' | Coupling models of resistive valves to muscle mechanics in cardiac fluid-structure interaction simulations | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | To accurately simulate all phases of the cardiac cycle, computational models of hemodynamics in heart chambers need to include a sufficiently faithful model of cardiac valves. This can be achieved efficiently through resistive methods, and the resistive immersed implicit surface (RIIS) model in particular [Fedele et al., BMMB, 2017]. However, the conventional RIIS model is not suited to fluid-structure interaction (FSI) simulations, since it neglects the reaction forces by which valves are attached to the cardiac walls, leading to models that are not consistent with Newton's laws. In this paper, we propose an improvement to RIIS to overcome this limitation, by adding distributed forces acting on the structure to model the attachment of valves to the cardiac walls. The modification has a minimal computational overhead thanks to an explicit numerical discretization scheme. Numerical experiments in both idealized and realistic settings demonstrate the effectiveness of the proposed modification in ensuring the physical consistency of the model, thus allowing to apply RIIS and other resistive valve models in the context of FSI simulations. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:00:50 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Bucelli",
"Michele",
""
],
[
"Dede'",
"Luca",
""
]
] |
2506.03870 | Syed Mhamudul Hasan | Mohd. Farhan Israk Soumik, Syed Mhamudul Hasan, Abdur R. Shahid | Evaluating Apple Intelligence's Writing Tools for Privacy Against Large Language Model-Based Inference Attacks: Insights from Early Datasets | null | null | null | null | cs.LG cs.CR | http://creativecommons.org/licenses/by/4.0/ | The misuse of Large Language Models (LLMs) to infer emotions from text for malicious purposes, known as emotion inference attacks, poses a significant threat to user privacy. In this paper, we investigate the potential of Apple Intelligence's writing tools, integrated across iPhone, iPad, and MacBook, to mitigate these risks through text modifications such as rewriting and tone adjustment. By developing early novel datasets specifically for this purpose, we empirically assess how different text modifications influence LLM-based detection. This capability suggests strong potential for Apple Intelligence's writing tools as privacy-preserving mechanisms. Our findings lay the groundwork for future adaptive rewriting systems capable of dynamically neutralizing sensitive emotional content to enhance user privacy. To the best of our knowledge, this research provides the first empirical analysis of Apple Intelligence's text-modification tools within a privacy-preservation context with the broader goal of developing on-device, user-centric privacy-preserving mechanisms to protect against LLMs-based advanced inference attacks on deployed systems. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:01:17 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Soumik",
"Mohd. Farhan Israk",
""
],
[
"Hasan",
"Syed Mhamudul",
""
],
[
"Shahid",
"Abdur R.",
""
]
] |
2506.03871 | Wencheng Wang | Wencheng Wang, Minxue Yang, Wei Chen, Xiangang Wan, Feng Tang | Large Berry curvature effects induced by extended nodal structures: Rational design strategy and high-throughput materials predictions | 15 pages, 7 figures | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Berry curvature can drastically modify the electron dynamics, thereby offering an effective pathway for electron manipulation and novel device applications. Compared to zero-dimensional nodal points in Weyl/Dirac semimetals, higher-dimensional extended nodal structures, such as nodal lines and nodal surfaces, are more likely to intersect the Fermi surface, leading to large Berry curvature effects without fine-tuning the chemical potential. In this work, we propose a strategy that utilizes straight nodal lines (SNLs) and flat nodal surfaces (FNSs) to design large Berry curvature effects, and we exhaustively tabulate SNLs and FNSs within the 1651 magnetic space groups (MSGs). We demonstrate that SNLs and FNSs can generate large Berry curvature widely distributed in the Brillouin zone. As an application, we identify 158 MSGs that host FNSs, SNLs, or both and allow for nonvanishing anomalous Hall conductivity (AHC). Based on these 158 MSGs, we screen materials from the MAGNDATA magnetic material database for high-throughput calculations, identifying 60 materials with AHC values exceeding $500\,\Omega^{-1}{\rm cm}^{-1}$. We select the candidate materials $\rm SrRuO_3$ and $\rm Ca_2NiOsO_6$ to demonstrate the contributions of FNSs and SNLs to one and two nonvanishing AHC components, respectively. We also investigate the tuning of AHC through symmetry breaking, outlining all possible symmetry-breaking pathways, and select the candidate material HoNi to demonstrate this approach by applying an external magnetic field. Additionally, we identify Berry curvature quadrupoles in the candidate materials, indicating that our strategy can be generalized to Berry curvature multipole effects. Our work will guide both the theoretical and experimental design of materials with large Berry curvature effects, with significant implications for a wide range of device applications. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:03:29 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Wang",
"Wencheng",
""
],
[
"Yang",
"Minxue",
""
],
[
"Chen",
"Wei",
""
],
[
"Wan",
"Xiangang",
""
],
[
"Tang",
"Feng",
""
]
] |
2506.03872 | Yang Xiao | Yang Xiao, Guoan Xu, Qiang Wu, Wenjing Jia | JointSplat: Probabilistic Joint Flow-Depth Optimization for Sparse-View Gaussian Splatting | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reconstructing 3D scenes from sparse viewpoints is a long-standing challenge with wide applications. Recent advances in feed-forward 3D Gaussian sparse-view reconstruction methods provide an efficient solution for real-time novel view synthesis by leveraging geometric priors learned from large-scale multi-view datasets and computing 3D Gaussian centers via back-projection. Despite offering strong geometric cues, both feed-forward multi-view depth estimation and flow-depth joint estimation face key limitations: the former suffers from mislocation and artifact issues in low-texture or repetitive regions, while the latter is prone to local noise and global inconsistency due to unreliable matches when ground-truth flow supervision is unavailable. To overcome this, we propose JointSplat, a unified framework that leverages the complementarity between optical flow and depth via a novel probabilistic optimization mechanism. Specifically, this pixel-level mechanism scales the information fusion between depth and flow based on the matching probability of optical flow during training. Building upon the above mechanism, we further propose a novel multi-view depth-consistency loss to leverage the reliability of supervision while suppressing misleading gradients in uncertain areas. Evaluated on RealEstate10K and ACID, JointSplat consistently outperforms state-of-the-art (SOTA) methods, demonstrating the effectiveness and robustness of our proposed probabilistic joint flow-depth optimization approach for high-fidelity sparse-view 3D reconstruction. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:04:40 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Xiao",
"Yang",
""
],
[
"Xu",
"Guoan",
""
],
[
"Wu",
"Qiang",
""
],
[
"Jia",
"Wenjing",
""
]
] |
2506.03873 | Alexander Mrokon | Alexander Mrokon, Heike Kraft, Dongsung Shin, Hiroki Tanaka, Simon J. Herr, Karsten Buse, and Ingo Breunig | Electrically-induced resonance shifts of whispering gallery resonators made of barium magnesium fluoride | null | null | null | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | Barium magnesium fluoride (BMF) is a ferroelectric crystal with a transparency range far beyond the one of other optical materials. In particular, its low loss in the deep ultraviolet makes this material an unique candidate for frequency conversion in this spectral range. Due to its relatively weak second-order nonlinearity, a resonant configuration such as an optical whispering gallery would be beneficial. We show that femtosecond-laser based material processing enables the reliable fabrication of BMF whispering gallery resonators with quality factors beyond $10^7$. Their resonance frequencies can be shifted linearly by applying electric fields between the $+c$ and $-c$ faces of the crystal. The slope of the shift is $-0.8$~MHz/(V/mm). It seems that the origin of this shift is piezoelectricity, while the electro-optic effect is negligible. Our results pave the way for millimeter-sized frequency converters in the deep ultraviolet. Furthermore, they indicate that a careful determination of fundamental material properties is still necessary. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:08:22 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mrokon",
"Alexander",
""
],
[
"Kraft",
"Heike",
""
],
[
"Shin",
"Dongsung",
""
],
[
"Tanaka",
"Hiroki",
""
],
[
"Herr",
"Simon J.",
""
],
[
"Buse",
"Karsten",
""
],
[
"Breunig",
"Ingo",
""
]
] |
2506.03874 | Zhonghao Liang | Zhonghao Liang and Qunying Liao | The equivalent condition for GRL codes to be MDS, AMDS or self-dual | 18 pages | null | null | null | cs.IT math.IT | http://creativecommons.org/licenses/by/4.0/ | It is well-known that MDS, AMDS or self-dual codes have good algebraic properties, and are applied in communication systems, data storage, quantum codes, and so on. In this paper, we focus on a class of generalized Roth-Lempel linear codes, and give an equivalent condition for them or their dual to be non-RS MDS, AMDS or non-RS self-dual and some corresponding examples. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:08:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Liang",
"Zhonghao",
""
],
[
"Liao",
"Qunying",
""
]
] |
2506.03875 | Nathan Szwarcberg | Nathan Szwarcberg (LMA), Tom Colinot, Christophe Vergez (LMA), Michael Jousserand, L\'eonie Maignan (LMA), Hrant Arzumanyan (LMA), Giordano Gatti (LMA), Anthia Patsinakidou (LMA), Pedro Faria Oliveira Morais (LMA) | Register jumps on the clarinet: numerical and in-vitro investigation into basins of attraction and phase-tipping | null | null | null | null | physics.class-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When playing the clarinet, opening the register hole allows for a transition from the first to the second register, producing a twelfth interval. On an artificial player system, the blowing pressure range where the second register remains stable can be determined by gradually varying the blowing pressure while keeping the register hole open. However, when the register hole is opened while the instrument is already producing the first register, the range of blowing pressures that lead to a stable second register is narrower than the full stability zone of the second register. This phenomenon is investigated numerically by performing multiple hole openings at different times, for various values of the blowing pressure and the embouchure parameter. In some narrow regions of the control parameters space, the success of a register transition depends on the phase at which the hole is opened. This illustrates an instance of phase-tipping, where the limit cycle of the closed-hole regime may intersect multiple basins of attraction associated with the open-hole regimes. Furthermore, to assess the robustness of the basins of attraction, random noise is introduced to the control parameters before the register hole is opened. Results indicate that the equilibrium regime is more robust to noise than the other oscillating regimes. Finally, long-lasting transient quasiperiodics are investigated. The phase at which the hole is opened influences both the transient duration and the resulting stable regime. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:09:03 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Szwarcberg",
"Nathan",
"",
"LMA"
],
[
"Colinot",
"Tom",
"",
"LMA"
],
[
"Vergez",
"Christophe",
"",
"LMA"
],
[
"Jousserand",
"Michael",
"",
"LMA"
],
[
"Maignan",
"Léonie",
"",
"LMA"
],
[
"Arzumanyan",
"Hrant",
"",
"LMA"
],
[
"Gatti",
"Giordano",
"",
"LMA"
],
[
"Patsinakidou",
"Anthia",
"",
"LMA"
],
[
"Morais",
"Pedro Faria Oliveira",
"",
"LMA"
]
] |
2506.03876 | Hongliang Tian | Yuke Peng, Hongliang Tian, Zhang Junyang, Ruihan Li, Chengjun Chen, Jianfeng Jiang, Jinyi Xian, Xiaolin Wang, Chenren Xu, Diyu Zhou, Yingwei Luo, Shoumeng Yan, Yinqian Zhang | Asterinas: A Linux ABI-Compatible, Rust-Based Framekernel OS with a Small and Sound TCB | null | USENIX ATC 2025 | null | null | cs.OS | http://creativecommons.org/licenses/by/4.0/ | How can one build a feature-rich, general-purpose, Rust-based operating system (OS) with a minimal and sound Trusted Computing Base (TCB) for memory safety? Existing Rust-based OSes fall short due to their improper use of unsafe Rust in kernel development. To address this challenge, we propose a novel OS architecture called framekernel that realizes Rust's full potential to achieve intra-kernel privilege separation, ensuring TCB minimality and soundness. We present OSTD, a streamlined framework for safe Rust OS development, and Asterinas, a Linux ABI-compatible framekernel OS implemented entirely in safe Rust using OSTD. Supporting over 210 Linux system calls, Asterinas delivers performance on par with Linux, while maintaining a minimized, memory-safety TCB of only about 14.0% of the codebase. These results underscore the practicality and benefits of the framekernel architecture in building safe and efficient OSes. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:10:26 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Peng",
"Yuke",
""
],
[
"Tian",
"Hongliang",
""
],
[
"Junyang",
"Zhang",
""
],
[
"Li",
"Ruihan",
""
],
[
"Chen",
"Chengjun",
""
],
[
"Jiang",
"Jianfeng",
""
],
[
"Xian",
"Jinyi",
""
],
[
"Wang",
"Xiaolin",
""
],
[
"Xu",
"Chenren",
""
],
[
"Zhou",
"Diyu",
""
],
[
"Luo",
"Yingwei",
""
],
[
"Yan",
"Shoumeng",
""
],
[
"Zhang",
"Yinqian",
""
]
] |
2506.03877 | Christian Gang Liu | Christian Gang Liu, Peter Bodorik, Dawn Jutla | Automated Mechanism to Support Trade Transactions in Smart Contracts with Upgrade and Repair | null | Elsevier Journal of Blockchain: Research and Applications, 2025 | 10.1016/j.bcra.2025.100285 | null | cs.SE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In our previous research, we addressed the problem of automated transformation of models, represented using the business process model and notation (BPMN) standard, into the methods of a smart contract. The transformation supports BPMN models that contain complex multi-step activities that are supported using our concept of multi-step nested trade transactions, wherein the transactional properties are enforced by a mechanism generated automatically by the transformation process from a BPMN model to a smart contract. In this paper, we present a methodology for repairing a smart contract that cannot be completed due to events that were not anticipated by the developer and thus prevent the completion of the smart contract. The repair process starts with the original BPMN model fragment causing the issue, providing the modeler with the innermost transaction fragment containing the failed activity. The modeler amends the BPMN pattern on the basis of successful completion of previous activities. If repairs exceed the inner transaction's scope, they are addressed using the parent transaction's BPMN model. The amended BPMN model is then transformed into a new smart contract, ensuring consistent data and logic transitions. We previously developed a tool, called TABS+, as a proof of concept (PoC) to transform BPMN models into smart contracts for nested transactions. This paper describes the tool TABS+R, developed by extending the TABS+ tool, to allow the repair of smart contracts. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:13:07 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Liu",
"Christian Gang",
""
],
[
"Bodorik",
"Peter",
""
],
[
"Jutla",
"Dawn",
""
]
] |
2506.03878 | Thomas Borsoni | Thomas Borsoni (LJLL (UMR\_7598)), Laurent Boudin (REO, LJLL), Julien Mathiaud (IRMAR, UR), Francesco Salvarani (CEREMADE) | A kinetic model for polyatomic gas with quasi-resonant collisions leading to bi-temperature relaxation processes | null | null | null | null | math.AP math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this article, we extend the Boltzmann framework for polyatomic gases by introducing quasi-resonant kernels, which relax resonant interactions, for which kinetic and internal energies are separately conserved and lead to equilibrium states with two temperatures. We establish an H-theorem and analyze the quasi-resonant model's asymptotic behaviour, demonstrating a two-phase relaxation process: an initial convergence towards a two-temperature Maxwellian state followed by gradual relaxation of the two temperatures towards each other. Numerical simulations validate our theoretical predictions. The notion of quasi-resonance provides the first rigorous framework of a Boltzmann dynamics for which the distribution is at all times close to a multi-temperature Maxwellian, relaxing towards a one-temperature Maxwellian. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:14:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Borsoni",
"Thomas",
"",
"LJLL"
],
[
"Boudin",
"Laurent",
"",
"REO, LJLL"
],
[
"Mathiaud",
"Julien",
"",
"IRMAR, UR"
],
[
"Salvarani",
"Francesco",
"",
"CEREMADE"
]
] |
2506.03879 | Xuexiang Xu | Huan Liu, Zu-wu Chen, Xue-feng Zhan, Hong-chun Yuan and Xue-xiang Xu | Exploring entanglement, Wigner negativity and Bell nonlocality for anisotropic two-qutrit states | 13 pages, 11 figures, Comments are welcome | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We introduce a family of anisotropic two-qutrit states (AITTSs). These AITTSs are expressed as $\rho _{aiso}=p\left\vert \psi _{\left( \theta,\phi \right) }\right\rangle \left\langle \psi _{\left( \theta ,\phi \right)}\right\vert +(1-p)\frac{1_{9}}{9}$ with $\left\vert \psi _{\left( \theta,\phi \right) }\right\rangle =\sin \theta \cos \phi \left\vert00\right\rangle +\sin \theta \sin \phi \left\vert 11\right\rangle +\cos\theta \left\vert 22\right\rangle $ and $1_{9}=\sum_{j,k=0}^{2}\left\vert jk\right\rangle \left\langle jk\right\vert $. For a given $p\in \lbrack 0,1]$, these states are adjustable in different ($\theta ,\phi $) directions. In the case of ($\theta ,\phi $) = ($\arccos (1/\sqrt{3}),\pi /4$), the AITTS will reduce to the isotropic two-qutrit state $\rho _{iso}$. In addition, the AITTSs are severely affected by the white noise ($\rho _{noise}=1_{9}/9$). Three properties of the AITTSs, including entanglement, Wigner negativity and Bell nonlocality, are explored detailedly in the analytical and numerical ways. Each property is witnessed by an appropriate existing criterion. Some of our results are summarized as follows: (i) Large entanglement does not necessarily mean high Wigner negativity and strong Bell nonlocality. (ii) A pure state with a large Schmidt number does not necessarily have a greater Wigner negativity. (iii) Only when $\left\vert\psi _{\left( \theta ,\phi \right) }\right\rangle $ has the Schmidt number 3, the AITTS has the possibility of exhibiting Bell nonlocality in proper parameter range. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:15:52 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Liu",
"Huan",
""
],
[
"Chen",
"Zu-wu",
""
],
[
"Zhan",
"Xue-feng",
""
],
[
"Yuan",
"Hong-chun",
""
],
[
"Xu",
"Xue-xiang",
""
]
] |
2506.03880 | Ruihan Jin | Ruihan Jin, Pengpeng Shao, Zhengqi Wen, Jinyang Wu, Mingkuan Feng, Shuai Zhang, Jianhua Tao | RadialRouter: Structured Representation for Efficient and Robust Large Language Models Routing | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | The rapid advancements in large language models (LLMs) have led to the emergence of routing techniques, which aim to efficiently select the optimal LLM from diverse candidates to tackle specific tasks, optimizing performance while reducing costs. Current LLM routing methods are limited in effectiveness due to insufficient exploration of the intrinsic connection between user queries and the characteristics of LLMs. To address this issue, in this paper, we present RadialRouter, a novel framework for LLM routing which employs a lightweight Transformer-based backbone with a radial structure named RadialFormer to articulate the query-LLMs relationship. The optimal LLM selection is performed based on the final states of RadialFormer. The pipeline is further refined by an objective function that combines Kullback-Leibler divergence with the query-query contrastive loss to enhance robustness. Experimental results on RouterBench show that RadialRouter significantly outperforms existing routing methods by 9.2\% and 5.8\% in the Balance and Cost First scenarios, respectively. Additionally, its adaptability toward different performance-cost trade-offs and the dynamic LLM pool demonstrates practical application potential. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:16:41 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Jin",
"Ruihan",
""
],
[
"Shao",
"Pengpeng",
""
],
[
"Wen",
"Zhengqi",
""
],
[
"Wu",
"Jinyang",
""
],
[
"Feng",
"Mingkuan",
""
],
[
"Zhang",
"Shuai",
""
],
[
"Tao",
"Jianhua",
""
]
] |
2506.03881 | Benedikt Tiedemann | Benedikt Tiedemann, Moritz Kreuseler, Jochen Fr\"ohlich | An integrated multi-size collision model for flotation | null | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | The accuracy obtained with CFD and process simulations of flotation critically depends on the quality and robustness of the underlying models for the non-resolved sub-processes. An important issue in flotation is the collision between particles and air bubbles. Many models have been developed, but their accuracy for applications in flotation is limited. In particular, the significant size difference between particles and bubbles and their intricate coupling to the turbulent flow field pose severe challenges. The present paper first reviews presently employed collision models, highlighting their advantages and disadvantages when applied to flotation. On this basis, the "Integrated Multi-Size Collision model" (IMSC) is proposed. After a detailed evaluation, it combines existing approaches from various sources and introduces new developments designed to address present shortcomings. The model is validated by own DNS data as well as data from the literature. It is shown that, overall, the IMSC provides better predictions for the collision rate in typical flotation conditions than presently employed collision models and covers the entire parameter range of the flotation process very well. Using the available data, some of the underlying modelling assumptions are validated. Finally, a comprehensive overview of the model is provided for further use in Euler-Euler frameworks or process simulations. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:17:17 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tiedemann",
"Benedikt",
""
],
[
"Kreuseler",
"Moritz",
""
],
[
"Fröhlich",
"Jochen",
""
]
] |
2506.03882 | Anthony Hastir | Anthony Hastir, Birgit Jacob | LQ optimal control for infinite-dimensional passive systems | null | null | null | null | math.OC math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the Linear-Quadratic optimal control problem for a general class of infinite-dimensional passive systems, allowing for unbounded input and output operators. We show that under mild assumptions, the finite cost condition is always satisfied. Moreover, we show that the optimal cost operator is a contraction. In the case where the system is energy preserving, the optimal cost operator is shown to be the identity, which allows to deduce easily the unique stabilizing optimal control. In this case, we derive an explicit solution to an adapted operator Riccati equation. We apply our results to boundary control systems, first-order port-Hamiltonian systems and an Euler-Bernoulli beam with shear force control. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:19:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Hastir",
"Anthony",
""
],
[
"Jacob",
"Birgit",
""
]
] |
2506.03883 | Gabriele Alberti | G. Alberti, E. Tonello, C. Tuccari, F. Mombelli, S. Brezinsek, T. Dittmar, A. Hakola, A. Kirschner, K. Krieger, M. Rasinski, J. Romazanov, A. Uccello, A. Widdowson, M. Passoni, the ASDEX Upgrade Team and the EUROfusion Tokamak Exploitation Team | Tungsten erosion and scrape-off layer transport modelling in L-mode helium plasma discharges in ASDEX Upgrade | null | null | null | null | physics.plasm-ph | http://creativecommons.org/licenses/by/4.0/ | Due to its unavoidable presence in thermonuclear DT plasmas and to its peculiar effects on materials, investigating the role of helium (He) in plasma-wall interaction (PWI) in current tokamaks is fundamental. In this work, PWI in L-mode He plasma discharges in ASDEX Upgrade (AUG) is modelled by exploiting simplified analytical approaches and two state-of-the-art codes. SOLPS-ITER is employed both to provide a suitable background plasma for erosion simulations and to interpret diagnostics measurements in terms of He+/2+ fraction. In particular, a 50-50% concentration of the two He ions is found in the proximity of the strike-points, while He2+ represents the dominant population farther in the scrape-off layer (SOL). The role of He ion fraction on AUG tungsten divertor erosion is first estimated by means of a simple analytical model and, afterwards, by exploiting ERO2.0, showing the major impact of He2+ in common AUG plasma temperatures. ERO2.0 findings are also compared with experimental erosion data in the strike-point region, showing the possible impact of extrinsic impurities on divertor erosion. Finally, the multi-fluid and kinetic approaches employed in this work to simulate W erosion and migration are compared, including W also in SOLPS-ITER modelling. The impact of target boundary conditions on the W source in SOLPS-ITER is investigated, in order to find a good agreement with ERO2.0 estimation. Then, W migration in the two codes is compared, showing a stronger W transport towards the X-point in ERO2.0 compared to present SOLPS-ITER simulations with no drifts. Similar W influx in core could be achieved by reducing anomalous diffusivity in ERO2.0. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:20:36 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Alberti",
"G.",
""
],
[
"Tonello",
"E.",
""
],
[
"Tuccari",
"C.",
""
],
[
"Mombelli",
"F.",
""
],
[
"Brezinsek",
"S.",
""
],
[
"Dittmar",
"T.",
""
],
[
"Hakola",
"A.",
""
],
[
"Kirschner",
"A.",
""
],
[
"Krieger",
"K.",
""
],
[
"Rasinski",
"M.",
""
],
[
"Romazanov",
"J.",
""
],
[
"Uccello",
"A.",
""
],
[
"Widdowson",
"A.",
""
],
[
"Passoni",
"M.",
""
],
[
"Team",
"the ASDEX Upgrade",
""
],
[
"Team",
"the EUROfusion Tokamak Exploitation",
""
]
] |
2506.03884 | Utkarsh Pathak | Utkarsh Pathak, Chandra Sai Krishna Gunda, Anusha Prakash, Keshav Agarwal, Hema A. Murthy | Kinship in Speech: Leveraging Linguistic Relatedness for Zero-Shot TTS in Indian Languages | Accepted at INTERSPEECH 2025 | null | null | null | cs.CL cs.CV | http://creativecommons.org/licenses/by/4.0/ | Text-to-speech (TTS) systems typically require high-quality studio data and accurate transcriptions for training. India has 1369 languages, with 22 official using 13 scripts. Training a TTS system for all these languages, most of which have no digital resources, seems a Herculean task. Our work focuses on zero-shot synthesis, particularly for languages whose scripts and phonotactics come from different families. The novelty of our work is in the augmentation of a shared phone representation and modifying the text parsing rules to match the phonotactics of the target language, thus reducing the synthesiser overhead and enabling rapid adaptation. Intelligible and natural speech was generated for Sanskrit, Maharashtrian and Canara Konkani, Maithili and Kurukh by leveraging linguistic connections across languages with suitable synthesisers. Evaluations confirm the effectiveness of this approach, highlighting its potential to expand speech technology access for under-represented languages. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:22:24 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Pathak",
"Utkarsh",
""
],
[
"Gunda",
"Chandra Sai Krishna",
""
],
[
"Prakash",
"Anusha",
""
],
[
"Agarwal",
"Keshav",
""
],
[
"Murthy",
"Hema A.",
""
]
] |
2506.03885 | Sam Pollard | Sam Pollard and Michael Wray | Video, How Do Your Tokens Merge? | Accepted at eLVM workshop at CVPR 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Video transformer models require huge amounts of compute resources due to the spatio-temporal scaling of the input. Tackling this, recent methods have proposed to drop or merge tokens for image models, whether randomly or via learned methods. Merging tokens has many benefits: it can be plugged into any vision transformer, does not require model re-training, and it propagates information that would otherwise be dropped through the model. Before now, video token merging has not been evaluated on temporally complex datasets for video understanding. In this work, we explore training-free token merging for video to provide comprehensive experiments and find best practices across four video transformers on three datasets that exhibit coarse and fine-grained action recognition. Our results showcase the benefits of video token merging with a speedup of around $2.5$X while maintaining accuracy (avg. $-0.55\%$ for ViViT). Code available at https://github.com/sjpollard/video-how-do-your-tokens-merge. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:28:23 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Pollard",
"Sam",
""
],
[
"Wray",
"Michael",
""
]
] |
2506.03886 | Anna Lena Sch\"one | Anna Lena Sch\"one, Steffen Sch\"uttler, Talisha Mro{\ss}, Niklas Eichstaedt, Judith Golda, Lars Sch\"ucke, and Andrew R. Gibson | Concentration profiles of OH and H$_2$O$_2$ in plasma-treated water: influence of power, gas mixture and treatment distance | null | null | null | null | physics.plasm-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Plasma liquid interactions are important for a range of applications. For these, H$_2$O$_2$ and OH represent two key reactive species, whose concentrations in liquids need to be controlled for effective application outcomes. Here, a combination of gas and liquid simulations is used to study the concentration profiles of H$_2$O$_2$ and OH in water treated by a radio-frequency-driven plasma jet, with a glass capillary between the electrodes, operated in He with admixtures of water vapour. Simulations are compared with measured H$_2$O$_2$ concentrations and found to be in good qualitative agreement as plasma power and water admixture are varied. Simulation results show that the concentration profiles of H$_2$O$_2$ in the liquid are mainly determined by transport, while those of OH are limited by reactions with H$_2$O$_2$, which consumes OH. For a given plasma operating condition, the concentration and penetration depth of H$_2$O$_2$ increase with plasma treatment time, while those of OH tend to decrease because of the increasing H$_2$O$_2$ concentration. Plasma power, water vapour admixture, and the distance between the jet and the liquid surface all allow for the concentrations of H$_2$O$_2$ and OH to be controlled. The OH delivered from the gas phase to the liquid, and its concentration within the liquid are strongly dependent on the reaction pathways occurring in the effluent region, such that the trends in OH density at the end of the plasma region differ from those in the liquid. While the concentration of OH in the liquid is always much lower than that of H$_2$O$_2$, the ratio of the two species can be controlled over orders of magnitude by varying water admixture and power. The highest selectivity to OH is at low water admixtures, low powers and short treatment times, while the highest selectivity to H$_2$O$_2$ is at high water admixtures, high powers and long treatment times. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:29:57 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Schöne",
"Anna Lena",
""
],
[
"Schüttler",
"Steffen",
""
],
[
"Mroß",
"Talisha",
""
],
[
"Eichstaedt",
"Niklas",
""
],
[
"Golda",
"Judith",
""
],
[
"Schücke",
"Lars",
""
],
[
"Gibson",
"Andrew R.",
""
]
] |
2506.03887 | Junyi Chen | Junyi Chen, Shihao Bai, Zaijun Wang, Siyu Wu, Chuheng Du, Hailong Yang, Ruihao Gong, Shengzhong Liu, Fan Wu, Guihai Chen | Pre$^3$: Enabling Deterministic Pushdown Automata for Faster Structured LLM Generation | Published as a conference paper at ACL 2025 | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Extensive LLM applications demand efficient structured generations, particularly for LR(1) grammars, to produce outputs in specified formats (e.g., JSON). Existing methods primarily parse LR(1) grammars into a pushdown automaton (PDA), leading to runtime execution overhead for context-dependent token processing, especially inefficient under large inference batches. To address these issues, we propose Pre$^3$ that exploits deterministic pushdown automata (DPDA) to optimize the constrained LLM decoding efficiency. First, by precomputing prefix-conditioned edges during the preprocessing, Pre$^3$ enables ahead-of-time edge analysis and thus makes parallel transition processing possible. Second, by leveraging the prefix-conditioned edges, Pre$^3$ introduces a novel approach that transforms LR(1) transition graphs into DPDA, eliminating the need for runtime path exploration and achieving edge transitions with minimal overhead. Pre$^3$ can be seamlessly integrated into standard LLM inference frameworks, reducing time per output token (TPOT) by up to 40% and increasing throughput by up to 36% in our experiments. Our code is available at https://github.com/ModelTC/lightllm. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:30:30 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Chen",
"Junyi",
""
],
[
"Bai",
"Shihao",
""
],
[
"Wang",
"Zaijun",
""
],
[
"Wu",
"Siyu",
""
],
[
"Du",
"Chuheng",
""
],
[
"Yang",
"Hailong",
""
],
[
"Gong",
"Ruihao",
""
],
[
"Liu",
"Shengzhong",
""
],
[
"Wu",
"Fan",
""
],
[
"Chen",
"Guihai",
""
]
] |
2506.03888 | Dipanjan Mukherjee Dr. | Dipanjan Mukherjee | Jet-Feedback on kpc scales: a review | Submitted to Galaxies, MPDI. Minor updates expected. Comments and suggestions are very welcome | null | null | null | astro-ph.GA astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Relativistic jets from AGN are an important driver of feedback in galaxies. They interact with their environments over a wide range of physical scales during their lifetime, and an understanding of these interactions is crucial for unraveling the role of supermassive blackholes in shaping galaxy evolution. The impact of such jets have been traditionally considered in the context of heating the large-scale environments. However, in the last few decades there has been additional focus on the immediate impact of jet feedback on the host galaxy itself. In this review we outline the development of various numerical simulations since the onset of studies of jets to the present day, where sophisticated numerical techniques have been employed to study jet feedback including a range of physical processes. The jets can act as an important agent of injecting energy in the host's ISM, as confirmed both in observations of multi-phase gas, as well as in simulations. Such interactions have the potential to impact the kinematics of the gas as well as its star formation. We summarize the recent results from simulations of jet feedback on kpc scales, and outline the broader implications for observations and galaxy evolution. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:30:30 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mukherjee",
"Dipanjan",
""
]
] |
2506.03889 | Pau Vilimelis Aceituno | Pau Vilimelis Aceituno, Jack William Miller, Noah Marti, Youssef Farag, Victor Boussange | Temporal horizons in forecasting: a performance-learnability trade-off | 33 pages, 12 figures | null | null | null | cs.LG nlin.CD | http://creativecommons.org/licenses/by/4.0/ | When training autoregressive models for dynamical systems, a critical question arises: how far into the future should the model be trained to predict? Too short a horizon may miss long-term trends, while too long a horizon can impede convergence due to accumulating prediction errors. In this work, we formalize this trade-off by analyzing how the geometry of the loss landscape depends on the training horizon. We prove that for chaotic systems, the loss landscape's roughness grows exponentially with the training horizon, while for limit cycles, it grows linearly, making long-horizon training inherently challenging. However, we also show that models trained on long horizons generalize well to short-term forecasts, whereas those trained on short horizons suffer exponentially (resp. linearly) worse long-term predictions in chaotic (resp. periodic) systems. We validate our theory through numerical experiments and discuss practical implications for selecting training horizons. Our results provide a principled foundation for hyperparameter optimization in autoregressive forecasting models. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:34:22 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Aceituno",
"Pau Vilimelis",
""
],
[
"Miller",
"Jack William",
""
],
[
"Marti",
"Noah",
""
],
[
"Farag",
"Youssef",
""
],
[
"Boussange",
"Victor",
""
]
] |
2506.03890 | Christian Tinauer | Christian Tinauer and Maximilian Sackl and Stefan Ropele and Christian Langkammer | Identifying Alzheimer's Disease Prediction Strategies of Convolutional Neural Network Classifiers using R2* Maps and Spectral Clustering | Accepted for the conference EUSIPCO2025 (https://eusipco2025.org/) | null | null | null | eess.IV cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep learning models have shown strong performance in classifying Alzheimer's disease (AD) from R2* maps, but their decision-making remains opaque, raising concerns about interpretability. Previous studies suggest biases in model decisions, necessitating further analysis. This study uses Layer-wise Relevance Propagation (LRP) and spectral clustering to explore classifier decision strategies across preprocessing and training configurations using R2* maps. We trained a 3D convolutional neural network on R2* maps, generating relevance heatmaps via LRP and applied spectral clustering to identify dominant patterns. t-Stochastic Neighbor Embedding (t-SNE) visualization was used to assess clustering structure. Spectral clustering revealed distinct decision patterns, with the relevance-guided model showing the clearest separation between AD and normal control (NC) cases. The t-SNE visualization confirmed that this model aligned heatmap groupings with the underlying subject groups. Our findings highlight the significant impact of preprocessing and training choices on deep learning models trained on R2* maps, even with similar performance metrics. Spectral clustering offers a structured method to identify classification strategy differences, emphasizing the importance of explainability in medical AI. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:35:24 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tinauer",
"Christian",
""
],
[
"Sackl",
"Maximilian",
""
],
[
"Ropele",
"Stefan",
""
],
[
"Langkammer",
"Christian",
""
]
] |
2506.03891 | Hanna Myleiko | Hanna Myleiko, Sergei Solodky | On recovering the Radon-Nikodym derivative under the big data assumption | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The present paper is focused on the problem of recovering the Radon-Nikodym derivative under the big data assumption. To address the above problem, we design an algorithm that is a combination of the Nystr\"om subsampling and the standard Tikhonov regularization. The convergence rate of the corresponding algorithm is established both in the case when the Radon-Nikodym derivative belongs to RKHS and in the case when it does not. We prove that the proposed approach not only ensures the order of accuracy as algorithms based on the whole sample size, but also allows to achieve subquadratic computational costs in the number of observations. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:37:07 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Myleiko",
"Hanna",
""
],
[
"Solodky",
"Sergei",
""
]
] |
2506.03892 | Giyong Choi | Giyong Choi and HyunWook Park | Joint Video Enhancement with Deblurring, Super-Resolution, and Frame Interpolation Network | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Video quality is often severely degraded by multiple factors rather than a single factor. These low-quality videos can be restored to high-quality videos by sequentially performing appropriate video enhancement techniques. However, the sequential approach was inefficient and sub-optimal because most video enhancement approaches were designed without taking into account that multiple factors together degrade video quality. In this paper, we propose a new joint video enhancement method that mitigates multiple degradation factors simultaneously by resolving an integrated enhancement problem. Our proposed network, named DSFN, directly produces a high-resolution, high-frame-rate, and clear video from a low-resolution, low-frame-rate, and blurry video. In the DSFN, low-resolution and blurry input frames are enhanced by a joint deblurring and super-resolution (JDSR) module. Meanwhile, intermediate frames between input adjacent frames are interpolated by a triple-frame-based frame interpolation (TFBFI) module. The proper combination of the proposed modules of DSFN can achieve superior performance on the joint video enhancement task. Experimental results show that the proposed method outperforms other sequential state-of-the-art techniques on public datasets with a smaller network size and faster processing time. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:38:51 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Choi",
"Giyong",
""
],
[
"Park",
"HyunWook",
""
]
] |
2506.03893 | Yixuan Cao | Yuhong Feng, Fangcao Jian, Yixuan Cao, Xiaobin Jian, Jia Wang, Haiyue Feng, Chunyan Miao | An Efficient Candidate-Free R-S Set Similarity Join Algorithm with the Filter-and-Verification Tree and MapReduce | null | null | null | null | cs.DC cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given two different collections of sets, the exact set similarity R-S Join finds all set pairs with similarity no less than a given threshold, which has widespread applications. While existing algorithms accelerate large-scale R-S Joins using a two-stage filter-and-verification framework along with the parallel and distributed MapReduce framework, they suffer from excessive candidate set pairs, leading to significant I/O, data transfer, and verification overhead, and ultimately degrading the performance. This paper proposes novel candidate-free R-S Join (CF-RS-Join) algorithms that integrate filtering and verification into a single stage through filter-and-verification trees (FVTs) and their linear variants (LFVTs). First, CF-RS-Join with FVT (CF-RS-Join/FVT) is proposed to leverage an innovative FVT structure that compresses elements and associated sets in memory, enabling single-stage processing that eliminates the candidate set generation, fast lookups, and reduced database scans. Correctness proofs are provided. Second, CF-RS-Join with LFVT (CF-RS-Join/LFVT) is proposed to exploit a more compact Linear FVT, which compresses non-branching paths into single nodes and stores them in linear arrays for optimized traversal. Third, MR-CF-RS-Join/FVT and MR-CF-RS-Join/LFVT have been proposed to extend our approaches using MapReduce for parallel processing. Empirical studies on 7 real-world datasets have been conducted to evaluate the performance of the proposed algorithms against selected existing algorithms in terms of execution time, scalability, memory usage, and disk usage. Experimental results demonstrate that our algorithm using MapReduce, i.e., MR-CF-RS-Join/LFVT, achieves the best performance. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:42:36 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Feng",
"Yuhong",
""
],
[
"Jian",
"Fangcao",
""
],
[
"Cao",
"Yixuan",
""
],
[
"Jian",
"Xiaobin",
""
],
[
"Wang",
"Jia",
""
],
[
"Feng",
"Haiyue",
""
],
[
"Miao",
"Chunyan",
""
]
] |
2506.03894 | Jan Seyfried | Jan Seyfried, Sayantan Sen, Marco Tomamichel | Testing (Conditional) Mutual Information | 79 pages, accepted for presentation at the Conference on Learning Theory (COLT) 2025 | null | null | null | cs.DS cs.IT math.IT | http://creativecommons.org/licenses/by/4.0/ | We investigate the sample complexity of mutual information and conditional mutual information testing. For conditional mutual information testing, given access to independent samples of a triple of random variables $(A, B, C)$ with unknown distribution, we want to distinguish between two cases: (i) $A$ and $C$ are conditionally independent, i.e., $I(A\!:\!C|B) = 0$, and (ii) $A$ and $C$ are conditionally dependent, i.e., $I(A\!:\!C|B) \geq \varepsilon$ for some threshold $\varepsilon$. We establish an upper bound on the number of samples required to distinguish between the two cases with high confidence, as a function of $\varepsilon$ and the three alphabet sizes. We conjecture that our bound is tight and show that this is indeed the case in several parameter regimes. For the special case of mutual information testing (when $B$ is trivial), we establish the necessary and sufficient number of samples required up to polylogarithmic terms.
Our technical contributions include a novel method to efficiently simulate weakly correlated samples from the conditionally independent distribution $P_{A|B} P_{C|B} P_B$ given access to samples from an unknown distribution $P_{ABC}$, and a new estimator for equivalence testing that can handle such correlated samples, which might be of independent interest. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:43:12 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Seyfried",
"Jan",
""
],
[
"Sen",
"Sayantan",
""
],
[
"Tomamichel",
"Marco",
""
]
] |
2506.03895 | Faegheh Hasibi | Emma J. Gerritse, Faegheh Hasibi, Arjen P. de Vries | Graph-Embedding Empowered Entity Retrieval | arXiv admin note: text overlap with arXiv:2005.02843 | null | null | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | In this research, we investigate methods for entity retrieval using graph embeddings. While various methods have been proposed over the years, most utilize a single graph embedding and entity linking approach. This hinders our understanding of how different graph embedding and entity linking methods impact entity retrieval. To address this gap, we investigate the effects of three different categories of graph embedding techniques and five different entity linking methods. We perform a reranking of entities using the distance between the embeddings of annotated entities and the entities we wish to rerank. We conclude that the selection of both graph embeddings and entity linkers significantly impacts the effectiveness of entity retrieval. For graph embeddings, methods that incorporate both graph structure and textual descriptions of entities are the most effective. For entity linking, both precision and recall concerning concepts are important for optimal retrieval performance. Additionally, it is essential for the graph to encompass as many entities as possible. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:43:17 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Gerritse",
"Emma J.",
""
],
[
"Hasibi",
"Faegheh",
""
],
[
"de Vries",
"Arjen P.",
""
]
] |
2506.03897 | Martina Rama | Martina Rama, Gabriele Santin, Giulia Cencetti, Michele Tizzoni, Bruno Lepri | Forecasting Seasonal Influenza Epidemics with Physics-Informed Neural Networks | null | null | null | null | physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate epidemic forecasting is critical for informing public health decisions and timely interventions. While Physics-Informed Neural Networks (PINNs) have shown promise in various scientific domains, their application to real-time epidemic forecasting remains limited. The reasons are mainly due to the intrinsic difficulty of the task and the tendency to fully leveraging their learning and inference potential, which, however, often results in non-optimal forecasting frameworks. Here, we present SIR-INN, a hybrid forecasting framework that integrates the mechanistic structure of the classical Susceptible-Infectious-Recovered (SIR) model into a neural network architecture. Trained once on synthetic epidemic scenarios, the model is able to generalize across epidemic conditions without retraining. From limited and noisy observations, SIR-INN infers key transmission parameters via Markov chain Monte Carlo (MCMC) generating probabilistic short- and long-term forecasts. We validate SIR-INN using national influenza data from the Italian National Institute of Health, in the 2023-2024 and 2024-2025 seasons. The model performs competitively with current state-of-the-art approaches, particularly in terms of Mean Absolute Error (MAE) and Weighted Interval Score (WIS). It shows accurate predictive performance in nearly all phases of the outbreak, with improved accuracy observed for the 2024-2025 influenza season. Credible uncertainty intervals are consistently maintained, despite occasional shortcomings in coverage. SIR-INN offers a computationally efficient, interpretable, and generalizable solution for epidemic forecasting, appropriately leveraging the framework's hybrid design. Its ability to provide real-time predictions of epidemic dynamics, together with uncertainty quantification, makes it a promising tool for real-world epidemic forecasting. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:53:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Rama",
"Martina",
""
],
[
"Santin",
"Gabriele",
""
],
[
"Cencetti",
"Giulia",
""
],
[
"Tizzoni",
"Michele",
""
],
[
"Lepri",
"Bruno",
""
]
] |
2506.03898 | Pierre-Fran\c{c}ois Massiani | Pierre-Fran\c{c}ois Massiani, Christian Fiedler, Lukas Haverbeck, Friedrich Solowjow, Sebastian Trimpe | A kernel conditional two-sample test | 40 pages, 8 figures, 8 tables. Under review | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a framework for hypothesis testing on conditional probability distributions, which we then use to construct conditional two-sample statistical tests. These tests identify the inputs -- called covariates in this context -- where two conditional expectations differ with high probability. Our key idea is to transform confidence bounds of a learning method into a conditional two-sample test, and we instantiate this principle for kernel ridge regression (KRR) and conditional kernel mean embeddings. We generalize existing pointwise-in-time or time-uniform confidence bounds for KRR to previously-inaccessible yet essential cases such as infinite-dimensional outputs with non-trace-class kernels. These bounds enable circumventing the need for independent data in our statistical tests, since they allow online sampling. We also introduce bootstrapping schemes leveraging the parametric form of testing thresholds identified in theory to avoid tuning inaccessible parameters, making our method readily applicable in practice. Such conditional two-sample tests are especially relevant in applications where data arrive sequentially or non-independently, or when output distributions vary with operational parameters. We demonstrate their utility through examples in process monitoring and comparison of dynamical systems. Overall, our results establish a comprehensive foundation for conditional two-sample testing, from theoretical guarantees to practical implementation, and advance the state-of-the-art on the concentration of vector-valued least squares estimation. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:53:13 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Massiani",
"Pierre-François",
""
],
[
"Fiedler",
"Christian",
""
],
[
"Haverbeck",
"Lukas",
""
],
[
"Solowjow",
"Friedrich",
""
],
[
"Trimpe",
"Sebastian",
""
]
] |
2506.03899 | Wenjing Liao | Jiahui Cheng, Sung Ha Kang, Haomin Zhou, Wenjing Liao | Identification of Differential Equations by Dynamics-Guided Weighted Weak Form with Voting | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | In the identification of differential equations from data, significant progresses have been made with the weak/integral formulation. In this paper, we explore the direction of finding more efficient and robust test functions adaptively given the observed data. While this is a difficult task, we propose weighting a collection of localized test functions for better identification of differential equations from a single trajectory of noisy observations on the differential equation. We find that using high dynamic regions is effective in finding the equation as well as the coefficients, and propose a dynamics indicator per differential term and weight the weak form accordingly. For stable identification against noise, we further introduce a voting strategy to identify the active features from an ensemble of recovered results by selecting the features that frequently occur in different weighting of test functions. Systematic numerical experiments are provided to demonstrate the robustness of our method. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:53:55 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Cheng",
"Jiahui",
""
],
[
"Kang",
"Sung Ha",
""
],
[
"Zhou",
"Haomin",
""
],
[
"Liao",
"Wenjing",
""
]
] |
2506.03900 | Phillip J Black | D. P. Zerafa, P. N. Wilkinson, C. J. Radcliffe, J. P. Leahy, I. W. A. Browne, P. J. Black | L-BASS: A project to produce an absolutely calibrated 1.4 GHz sky map. II -- Technical Description of the System | Accepted for publication by RASTI, 21/05/2025 | null | null | null | astro-ph.IM astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | L-BASS is an instrument designed to make radiometric temperature measurements of the sky with an absolute accuracy of better than 0.1 K at 1.4 GHz. This will be achieved in two steps: first by measuring the sky temperature relative to that of the North Celestial Pole, using two horn-based antennas, and second with the sky antenna replaced with a calibrated cryogenic load to measure the absolute brightness temperature of the North Celesial Pole. Here we describe the design of the L-BASS two-antenna system and report on laboratory measurements to establish its performance at component and sub-system level. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:54:59 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Zerafa",
"D. P.",
""
],
[
"Wilkinson",
"P. N.",
""
],
[
"Radcliffe",
"C. J.",
""
],
[
"Leahy",
"J. P.",
""
],
[
"Browne",
"I. W. A.",
""
],
[
"Black",
"P. J.",
""
]
] |
2506.03902 | Eleftheria Tsipidi | Eleftheria Tsipidi, Samuel Kiegeland, Franz Nowak, Tianyang Xu, Ethan Wilcox, Alex Warstadt, Ryan Cotterell, Mario Giulianelli | The Harmonic Structure of Information Contours | ACL 2025 (main conference) | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | The uniform information density (UID) hypothesis proposes that speakers aim to distribute information evenly throughout a text, balancing production effort and listener comprehension difficulty. However, language typically does not maintain a strictly uniform information rate; instead, it fluctuates around a global average. These fluctuations are often explained by factors such as syntactic constraints, stylistic choices, or audience design. In this work, we explore an alternative perspective: that these fluctuations may be influenced by an implicit linguistic pressure towards periodicity, where the information rate oscillates at regular intervals, potentially across multiple frequencies simultaneously. We apply harmonic regression and introduce a novel extension called time scaling to detect and test for such periodicity in information contours. Analyzing texts in English, Spanish, German, Dutch, Basque, and Brazilian Portuguese, we find consistent evidence of periodic patterns in information rate. Many dominant frequencies align with discourse structure, suggesting these oscillations reflect meaningful linguistic organization. Beyond highlighting the connection between information rate and discourse structure, our approach offers a general framework for uncovering structural pressures at various levels of linguistic granularity. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:56:30 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tsipidi",
"Eleftheria",
""
],
[
"Kiegeland",
"Samuel",
""
],
[
"Nowak",
"Franz",
""
],
[
"Xu",
"Tianyang",
""
],
[
"Wilcox",
"Ethan",
""
],
[
"Warstadt",
"Alex",
""
],
[
"Cotterell",
"Ryan",
""
],
[
"Giulianelli",
"Mario",
""
]
] |
2506.03904 | Jorge Penarrubia | Jorge Pe\~narrubia, Rapha\"el Errani, Eduardo Vitral, Matthew G. Walker | Gravothermal expansion of dwarf spheroidal galaxies heated by dark subhaloes | 19 pages, submitted | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We use analytical and $N$-body methods to study the evolution of dwarf spheroidal galaxies (dSphs) embedded in dark matter (DM) haloes that host a sizeable subhalo population. Dark subhaloes generate a fluctuating gravitational field that injects energy into stellar orbits, driving a gradual expansion of dSphs. Despite the overall expansion, the stellar density profile preserves its initial shape, suggesting that the evolution proceeds in a self-similar manner. Meanwhile, the velocity dispersion profile, initially flat, evolves as the galaxy expands: the inner regions heat up, while the outer regions cool down. Kinematically, this resembles gravothermal collapse but with an inverted evolution, instead of collapsing the stellar system expands within a fluctuating halo potential. As the half-light radius $r_{\rm half}$ approaches the halo peak velocity radius $r_{\rm max}$, the expansion slows, and the velocity dispersion peaks at $\sigma_{\rm max} \simeq 0.27 v_{\rm max}$. The stellar heat capacity remains positive for deeply embedded stars but diverges near $r_{\rm max}$, turning negative beyond this threshold, which indicates a phase transition in the dynamical response to energy injection. The relaxation time scales as $t_{\rm rel} \sim r_{\rm half}^{3/2}$, showing that orbital diffusion slows as the galaxy expands. Ultra-faint dSphs, having the smallest sizes and shortest relaxation times, are particularly sensitive to the presence of dark subhaloes. Some of our dSph models expand beyond the detection of current photometric surveys, becoming `stealth' galaxies with half-light radii and velocity dispersions similar to ultra-diffuse galaxies (UDGs), but orders of magnitude lower luminosities, representing a distinct, yet currently undetected, population of DM-dominated satellites. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:58:17 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Peñarrubia",
"Jorge",
""
],
[
"Errani",
"Raphaël",
""
],
[
"Vitral",
"Eduardo",
""
],
[
"Walker",
"Matthew G.",
""
]
] |
2506.03905 | Carlos Lozano | Jorge Ponsin and Carlos Lozano | Lattice Boltzmann wall boundary conditions for RANS-based simulations with wall functions | 43 pages | null | null | null | physics.flu-dyn physics.comp-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this study, we investigate wall boundary condition schemes for Lattice Boltzmann simulations of turbulent flows modeled using RANS equations with wall functions. Two alternative schemes are formulated and assessed: a hybrid regularized boundary condition and a slip-velocity bounce-back scheme. Their performance is evaluated using two canonical turbulent flow cases, a fully developed channel flow and a zero-pressure-gradient flat plate boundary layer (BL), selected specifically to isolate and analyze the impact of wall boundary condition treatments on turbulence modeling. The comparative analysis reveals that the slip-velocity bounce-back approach, which has received relatively little attention within the context of LBM-RANS with wall functions, consistently outperforms the regularized-based method in terms of both accuracy and sensitivity to mesh resolution. Moreover, the regularized-based approach is shown to be highly sensitive to the reconstruction of the wall-normal velocity gradient, even in simple geometries such as flat walls where no interpolation is required. This dependency necessitates the use of specialized, ad-hoc gradient reconstruction techniques, requirements that are not present in the slip-velocity bounce-back method. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 12:59:41 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Ponsin",
"Jorge",
""
],
[
"Lozano",
"Carlos",
""
]
] |
2506.03906 | Andr\'e Guerra | Andr\'e Guerra, Riccardo Tione | Constancy of the index for gradient mappings | 16 pages | null | null | null | math.AP math.CV math.DG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that if the Hessian of a $C^{1,1}$ function has uniformly positive determinant almost everywhere then its index is locally constant, as conjectured by \v{S}ver\'ak in 1992. We deduce this result as a consequence of a more general theorem valid for quasiregular gradient mappings. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:01:49 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Guerra",
"André",
""
],
[
"Tione",
"Riccardo",
""
]
] |
2506.03907 | Adriano Chialastri | Adriano Chialastri and Ko Sanders | Quasi-equivalence of Gaussian states and energy estimates for functions of modular Hamiltonians | 16 pages | null | null | null | math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | To compare two Gaussian states of the Weyl-CCR algebra of a free scalar QFT we study three closely related perspectives: (i) quasi-equivalence of the GNS-representations, (ii) differences of the total energy (on some Cauchy surface), and (iii) differences between functions of the modular Hamiltonians. (For perspective (ii) we will only consider real linear free scalar quantum fields on ultrastatic spacetimes.) These three perspectives are known to be related qualitatively, due to work of Araki and Yamagami, Verch and Longo. Our aim is to investigate quantitative relations, including in particular estimates of differences between functions of modular Hamiltonians in terms of energy differences. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:01:51 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Chialastri",
"Adriano",
""
],
[
"Sanders",
"Ko",
""
]
] |
2506.03908 | Andreas Katsanikakis | Andreas Katsanikakis and Nikolaos Bekiaris-Liberis | Stabilization of Linear Switched Systems with Long Input Delay via Average or Averaging Predictor Feedbacks | 13 pages, 8 figures, preprint submitted to Systems & Control Letters | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | We develop delay-compensating feedback laws for linear switched systems with time-dependent switching. Because the future values of the switching signal, which are needed for constructing an exact predictor-feedback law, may be unavailable at current time, the key design challenge is how to construct a proper predictor state. We resolve this challenge constructing two alternative, average predictor-based feedback laws. The first is viewed as a predictor-feedback law for a particular average system, properly modified to provide exact state predictions over a horizon that depends on a minimum dwell time of the switching signal (when it is available). The second is, essentially, a modification of an average of predictor feedbacks, each one corresponding to the fixed-mode predictor-feedback law. We establish that under the control laws introduced, the closed-loop systems are (uniformly) exponentially stable, provided that the differences among system's matrices and among (nominal stabilizing) controller's gains are sufficiently small, with a size that is inversely proportional to the delay length. Since no restriction is imposed on the delay, such a limitation is inherent to the problem considered (in which the future switching signal values are unavailable), and thus, it cannot be removed. The stability proof relies on multiple Lyapunov functionals constructed via backstepping and derivation of solutions' estimates for quantifying the difference between average and exact predictor states. We present consistent numerical simulation results, which illustrate the necessity of employing the average predictor-based laws and demonstrate the performance improvement when the knowledge of a minimum dwell time is properly utilized for improving state prediction accuracy. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:01:57 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Katsanikakis",
"Andreas",
""
],
[
"Bekiaris-Liberis",
"Nikolaos",
""
]
] |
2506.03909 | Zhongxing Yu | Lantian Li, Zhihao Liu, and Zhongxing Yu | Solsmith: Solidity Random Program Generator for Compiler Testing | 11 pages, 12 figures | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Smart contracts are computer programs that run on blockchain platforms, with Solidity being the most widely used language for their development. As blockchain technology advances, smart contracts have become increasingly important across various fields. In order for smart contracts to operate correctly, the correctness of the compiler is particularly crucial. Although some research efforts have been devoted to testing Solidity compilers, they primarily focus on testing methods and do not address the core issue of generating test programs. To fill this gap, this paper designs and implements Solsmith, a test program generator specifically aimed at uncovering defects in Solidity compilers. It tests the compiler correctness by generating valid and diverse Solidity programs. We have designed a series of unique program generation strategies tailored to Solidity, including enabling optimizations more frequently, avoiding undefined behaviour, and mitigating behavioural differences caused by intermediate representations. To validate the effectiveness of Solsmith, we assess the effectiveness of the test programs generated by Solsmith using the approach of differential testing. The preliminary results show that Solsmith can generate the expected test programs and uncover four confirmed defects in Solidity compilers, demonstrating the effectiveness and potential of Solsmith. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:04:17 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Li",
"Lantian",
""
],
[
"Liu",
"Zhihao",
""
],
[
"Yu",
"Zhongxing",
""
]
] |
2506.03910 | Shyam Sunder Prabhu | Shyam Prabhu, P Akshay Kumar, Antov Selwinston, Pavan Taduvai, Shreya Bairi, Rohit Batra | Enhancing Experimental Efficiency in Materials Design: A Comparative Study of Taguchi and Machine Learning Methods | 7 pages, 3 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Materials design problems often require optimizing multiple variables, rendering full factorial exploration impractical. Design of experiment (DOE) methods, such as Taguchi technique, are commonly used to efficiently sample the design space but they inherently lack the ability to capture non-linear dependency of process variables. In this work, we demonstrate how machine learning (ML) methods can be used to overcome these limitations. We compare the performance of Taguchi method against an active learning based Gaussian process regression (GPR) model in a wire arc additive manufacturing (WAAM) process to accurately predict aspects of bead geometry, including penetration depth, bead width, and height. While Taguchi method utilized a three-factor, five-level L25 orthogonal array to suggest weld parameters, the GPR model used an uncertainty-based exploration acquisition function coupled with latin hypercube sampling for initial training data. Accuracy and efficiency of both models was evaluated on 15 test cases, with GPR outperforming Taguchi in both metrics. This work applies to broader materials processing domain requiring efficient exploration of complex parameters. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:04:29 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Prabhu",
"Shyam",
""
],
[
"Kumar",
"P Akshay",
""
],
[
"Selwinston",
"Antov",
""
],
[
"Taduvai",
"Pavan",
""
],
[
"Bairi",
"Shreya",
""
],
[
"Batra",
"Rohit",
""
]
] |
2506.03911 | Yichun Hu | Chamsi Hssaine, Yichun Hu, Ciara Pike-Burke | Learning Fair And Effective Points-Based Rewards Programs | null | null | null | null | cs.LG math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Points-based rewards programs are a prevalent way to incentivize customer loyalty; in these programs, customers who make repeated purchases from a seller accumulate points, working toward eventual redemption of a free reward. These programs have recently come under scrutiny due to accusations of unfair practices in their implementation. Motivated by these concerns, we study the problem of fairly designing points-based rewards programs, with a focus on two obstacles that put fairness at odds with their effectiveness. First, due to customer heterogeneity, the seller should set different redemption thresholds for different customers to generate high revenue. Second, the relationship between customer behavior and the number of accumulated points is typically unknown; this requires experimentation which may unfairly devalue customers' previously earned points. We first show that an individually fair rewards program that uses the same redemption threshold for all customers suffers a loss in revenue of at most a factor of $1+\ln 2$, compared to the optimal personalized strategy that differentiates between customers. We then tackle the problem of designing temporally fair learning algorithms in the presence of demand uncertainty. Toward this goal, we design a learning algorithm that limits the risk of point devaluation due to experimentation by only changing the redemption threshold $O(\log T)$ times, over a horizon of length $T$. This algorithm achieves the optimal (up to polylogarithmic factors) $\widetilde{O}(\sqrt{T})$ regret in expectation. We then modify this algorithm to only ever decrease redemption thresholds, leading to improved fairness at a cost of only a constant factor in regret. Extensive numerical experiments show the limited value of personalization in average-case settings, in addition to demonstrating the strong practical performance of our proposed learning algorithms. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:05:16 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Hssaine",
"Chamsi",
""
],
[
"Hu",
"Yichun",
""
],
[
"Pike-Burke",
"Ciara",
""
]
] |
2506.03912 | Aleksandra Marinkovi\'c | Aleksandra Marinkovi\'c | Concave symplectic toric fillings | 11 pages, 9 figures | null | null | null | math.SG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As shown by Etnyre and Honda in [EH], every contact 3-manifold admits infinitely many concave symplectic fillings that are mutually not symplectomorphic and not related by blow ups. In this note we refine this result in the toric setting by showing that every contact toric 3-manifold admits infinitely many concave symplectic toric fillings that are mutually not equivariantly symplectomorphic and not related by blow ups. The concave symplectic toric structure is constructed on certain linear and cyclic plumbings over spheres. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:05:18 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Marinković",
"Aleksandra",
""
]
] |
2506.03913 | Claire Barale | Claire Barale, Michael Rovatsos, Nehal Bhuta | When Fairness Isn't Statistical: The Limits of Machine Learning in Evaluating Legal Reasoning | Preprint | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Legal decisions are increasingly evaluated for fairness, consistency, and bias using machine learning (ML) techniques. In high-stakes domains like refugee adjudication, such methods are often applied to detect disparities in outcomes. Yet it remains unclear whether statistical methods can meaningfully assess fairness in legal contexts shaped by discretion, normative complexity, and limited ground truth.
In this paper, we empirically evaluate three common ML approaches (feature-based analysis, semantic clustering, and predictive modeling) on a large, real-world dataset of 59,000+ Canadian refugee decisions (AsyLex). Our experiments show that these methods produce divergent and sometimes contradictory signals, that predictive modeling often depends on contextual and procedural features rather than legal features, and that semantic clustering fails to capture substantive legal reasoning.
We show limitations of statistical fairness evaluation, challenge the assumption that statistical regularity equates to fairness, and argue that current computational approaches fall short of evaluating fairness in legally discretionary domains. We argue that evaluating fairness in law requires methods grounded not only in data, but in legal reasoning and institutional context. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:05:37 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Barale",
"Claire",
""
],
[
"Rovatsos",
"Michael",
""
],
[
"Bhuta",
"Nehal",
""
]
] |
2506.03914 | Eduardo Santos Escriche | Eduardo Santos Escriche, Stefanie Jegelka | Learning equivariant models by discovering symmetries with learnable augmentations | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Recently, a trend has emerged that favors learning relevant symmetries from data in geometric domains instead of designing constrained architectures. To do so, two popular options are (1) to modify the training protocol, e.g., with a specific loss and data augmentations (soft equivariance), or (2) to ignore equivariance and infer it only implicitly. However, both options have limitations: soft equivariance requires a priori knowledge about relevant symmetries, while inferring symmetries merely via the task and larger data lacks interpretability. To address both limitations, we propose SEMoLA, an end-to-end approach that jointly (1) discovers a priori unknown symmetries in the data via learnable data augmentations, and (2) softly encodes the respective approximate equivariance into an arbitrary unconstrained model. Hence, it does not need prior knowledge about symmetries, it offers interpretability, and it maintains robustness to distribution shifts. Empirically, we demonstrate the ability of SEMoLA to robustly discover relevant symmetries while achieving high prediction accuracy across various datasets, encompassing multiple data modalities and underlying symmetry groups. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:06:05 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Escriche",
"Eduardo Santos",
""
],
[
"Jegelka",
"Stefanie",
""
]
] |
2506.03915 | Matej Ze\v{c}evi\'c | Sebastian R\"odling, Matej Ze\v{c}evi\'c, Devendra Singh Dhami, Kristian Kersting | Causal Explanations Over Time: Articulated Reasoning for Interactive Environments | Main paper: 9 pages, References: 2 pages, Supplementary: 9 pages. Number of figures: 10, number of tables: 3 | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Structural Causal Explanations (SCEs) can be used to automatically generate explanations in natural language to questions about given data that are grounded in a (possibly learned) causal model. Unfortunately they work for small data only. In turn they are not attractive to offer reasons for events, e.g., tracking causal changes over multiple time steps, or a behavioral component that involves feedback loops through actions of an agent. To this end, we generalize SCEs to a (recursive) formulation of explanation trees to capture the temporal interactions between reasons. We show the benefits of this more general SCE algorithm on synthetic time-series data and a 2D grid game, and further compare it to the base SCE and other existing methods for causal explanations. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:07:16 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Rödling",
"Sebastian",
""
],
[
"Zečević",
"Matej",
""
],
[
"Dhami",
"Devendra Singh",
""
],
[
"Kersting",
"Kristian",
""
]
] |
2506.03916 | Agostina Calabrese | Agostina Calabrese, Tom Sherborne, Bj\"orn Ross, Mirella Lapata | Compositional Generalisation for Explainable Hate Speech Detection | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Hate speech detection is key to online content moderation, but current models struggle to generalise beyond their training data. This has been linked to dataset biases and the use of sentence-level labels, which fail to teach models the underlying structure of hate speech. In this work, we show that even when models are trained with more fine-grained, span-level annotations (e.g., "artists" is labeled as target and "are parasites" as dehumanising comparison), they struggle to disentangle the meaning of these labels from the surrounding context. As a result, combinations of expressions that deviate from those seen during training remain particularly difficult for models to detect. We investigate whether training on a dataset where expressions occur with equal frequency across all contexts can improve generalisation. To this end, we create U-PLEAD, a dataset of ~364,000 synthetic posts, along with a novel compositional generalisation benchmark of ~8,000 manually validated posts. Training on a combination of U-PLEAD and real data improves compositional generalisation while achieving state-of-the-art performance on the human-sourced PLEAD. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:07:36 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Calabrese",
"Agostina",
""
],
[
"Sherborne",
"Tom",
""
],
[
"Ross",
"Björn",
""
],
[
"Lapata",
"Mirella",
""
]
] |
2506.03917 | Stefano Damiano | Stefano Damiano and Toon van Waterschoot | Sound Field Reconstruction Using Physics-Informed Boundary Integral Networks | Accepted for publication at EUSIPCO 2025 | null | null | null | eess.AS | http://creativecommons.org/licenses/by/4.0/ | Sound field reconstruction refers to the problem of estimating the acoustic pressure field over an arbitrary region of space, using only a limited set of measurements. Physics-informed neural networks have been adopted to solve the problem by incorporating in the training loss function the governing partial differential equation, either the Helmholtz or the wave equation. In this work, we introduce a boundary integral network for sound field reconstruction. Relying on the Kirchhoff-Helmholtz boundary integral equation to model the sound field in a given region of space, we employ a shallow neural network to retrieve the pressure distribution on the boundary of the considered domain, enabling to accurately retrieve the acoustic pressure inside of it. Assuming the positions of measurement microphones are known, we train the model by minimizing the mean squared error between the estimated and measured pressure at those locations. Experimental results indicate that the proposed model outperforms existing physics-informed data-driven techniques. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:07:54 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Damiano",
"Stefano",
""
],
[
"van Waterschoot",
"Toon",
""
]
] |
2506.03918 | Tomasz Kryjak | Marcin Kowalczyk, Kamil Jeziorek, Tomasz Kryjak | Learning from Noise: Enhancing DNNs for Event-Based Vision through Controlled Noise Injection | null | IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, Nashville, 2025 | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Event-based sensors offer significant advantages over traditional frame-based cameras, especially in scenarios involving rapid motion or challenging lighting conditions. However, event data frequently suffers from considerable noise, negatively impacting the performance and robustness of deep learning models. Traditionally, this problem has been addressed by applying filtering algorithms to the event stream, but this may also remove some of relevant data. In this paper, we propose a novel noise-injection training methodology designed to enhance the neural networks robustness against varying levels of event noise. Our approach introduces controlled noise directly into the training data, enabling models to learn noise-resilient representations. We have conducted extensive evaluations of the proposed method using multiple benchmark datasets (N-Caltech101, N-Cars, and Mini N-ImageNet) and various network architectures, including Convolutional Neural Networks, Vision Transformers, Spiking Neural Networks, and Graph Convolutional Networks. Experimental results show that our noise-injection training strategy achieves stable performance over a range of noise intensities, consistently outperforms event-filtering techniques, and achieves the highest average classification accuracy, making it a viable alternative to traditional event-data filtering methods in an object classification system. Code: https://github.com/vision-agh/DVS_Filtering | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:10:26 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Kowalczyk",
"Marcin",
""
],
[
"Jeziorek",
"Kamil",
""
],
[
"Kryjak",
"Tomasz",
""
]
] |
2506.03919 | Lorenz Kummer BSc MSc | Lorenz Kummer, Samir Moustafa, Anatol Ehrlich, Franka Bause, Nikolaus Suess, Wilfried N. Gansterer, Nils M. Kriege | Weisfeiler and Leman Go Gambling: Why Expressive Lottery Tickets Win | Accepted at ICML 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | The lottery ticket hypothesis (LTH) is well-studied for convolutional neural networks but has been validated only empirically for graph neural networks (GNNs), for which theoretical findings are largely lacking. In this paper, we identify the expressivity of sparse subnetworks, i.e. their ability to distinguish non-isomorphic graphs, as crucial for finding winning tickets that preserve the predictive performance. We establish conditions under which the expressivity of a sparsely initialized GNN matches that of the full network, particularly when compared to the Weisfeiler-Leman test, and in that context put forward and prove a Strong Expressive Lottery Ticket Hypothesis. We subsequently show that an increased expressivity in the initialization potentially accelerates model convergence and improves generalization. Our findings establish novel theoretical foundations for both LTH and GNN research, highlighting the importance of maintaining expressivity in sparsely initialized GNNs. We illustrate our results using examples from drug discovery. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:10:59 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Kummer",
"Lorenz",
""
],
[
"Moustafa",
"Samir",
""
],
[
"Ehrlich",
"Anatol",
""
],
[
"Bause",
"Franka",
""
],
[
"Suess",
"Nikolaus",
""
],
[
"Gansterer",
"Wilfried N.",
""
],
[
"Kriege",
"Nils M.",
""
]
] |
2506.03920 | Melody Lee | Melody Lee | Personalized Improvement of Standard Readout Error Mitigation using Low-Depth Circuits and Machine Learning | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum computers have shown promise in improving algorithms in a variety of fields. The realization of these advancements is limited by the presence of noise and high error rates, which become prominent especially with increasing system size. Mitigation techniques using matrix inversions, unfolding, and deep learning, among others, have been leveraged to reduce this error. However, these methods are not reflective of the entire gate set of the quantum device and may need further tuning depending on the distance from the most recent calibration time. This paper proposes a method of improvement to numerical readout error techniques, where the readout error model is further refined using measured probability distributions from a collection of low-depth circuits. We use machine learning to improve the readout error model for the quantum system, testing the circuits on the simulated IBM Perth backend using Qiskit. We demonstrate a median 6.6% improvement in fidelity, 29.9% improvement for mean-squared error, and 10.3% improvement in Hellinger distance over the standard error mitigation approach for a seven-qubit system with a circuit depth of four. With further focus directed towards such improvement of these error mitigation techniques, we are one step closer to the fault-tolerant quantum computing era. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:13:04 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lee",
"Melody",
""
]
] |
2506.03921 | Xunzhu Tang | Xunzhu Tang and Jacques Klein and Tegawend\'e F. Bissyand\'e | Boosting Open-Source LLMs for Program Repair via Reasoning Transfer and LLM-Guided Reinforcement Learning | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Several closed-source LLMs have consistently outperformed open-source alternatives in program repair tasks, primarily due to their superior reasoning capabilities and extensive pre-training. This paper introduces Repairity, a novel three-stage methodology that significantly narrows this performance gap through reasoning extraction and reinforcement learning. Our approach: (1) systematically filters high-quality reasoning traces from closed-source models using correctness verification, (2) transfers this reasoning knowledge to open-source models via supervised fine-tuning, and (3) develops reinforcement learning with LLM-based feedback to further optimize performance. Empirical evaluation across multiple program repair benchmarks demonstrates that Repairity improves the performance of Qwen2.5-Coder-32B-Instruct, a base open source LLM, by 8.68\% on average, reducing the capability gap with Claude-Sonnet3.7, a state-of-the-art closed-source model, from 10.05% to 1.35%. Ablation studies confirm that both reasoning extraction and LLM-guided reinforcement learning contribute significantly to these improvements. Our methodology generalizes effectively to additional code-related tasks, enabling organizations to leverage high-quality program repair capabilities while maintaining the customizability, transparency, and deployment flexibility inherent to open-source models. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:13:58 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tang",
"Xunzhu",
""
],
[
"Klein",
"Jacques",
""
],
[
"Bissyandé",
"Tegawendé F.",
""
]
] |
2506.03923 | Mohammadamin Shafiei | Mohammadamin Shafiei, Hamidreza Saffari, Nafise Sadat Moosavi | More or Less Wrong: A Benchmark for Directional Bias in LLM Comparative Reasoning | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Large language models (LLMs) are known to be sensitive to input phrasing, but the mechanisms by which semantic cues shape reasoning remain poorly understood. We investigate this phenomenon in the context of comparative math problems with objective ground truth, revealing a consistent and directional framing bias: logically equivalent questions containing the words ``more'', ``less'', or ``equal'' systematically steer predictions in the direction of the framing term. To study this effect, we introduce MathComp, a controlled benchmark of 300 comparison scenarios, each evaluated under 14 prompt variants across three LLM families. We find that model errors frequently reflect linguistic steering, systematic shifts toward the comparative term present in the prompt. Chain-of-thought prompting reduces these biases, but its effectiveness varies: free-form reasoning is more robust, while structured formats may preserve or reintroduce directional drift. Finally, we show that including demographic identity terms (e.g., ``a woman'', ``a Black person'') in input scenarios amplifies directional drift, despite identical underlying quantities, highlighting the interplay between semantic framing and social referents. These findings expose critical blind spots in standard evaluation and motivate framing-aware benchmarks for diagnosing reasoning robustness and fairness in LLMs. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:15:01 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Shafiei",
"Mohammadamin",
""
],
[
"Saffari",
"Hamidreza",
""
],
[
"Moosavi",
"Nafise Sadat",
""
]
] |
2506.03924 | Linjie Zhao | Xiaofeng Xue and Linjie Zhao | Moderate deviation principles for the current and the tagged particle in the WASEP | null | null | null | null | math.PR | http://creativecommons.org/licenses/by/4.0/ | We study the weakly asymmetric simple exclusion process in one dimension. We prove sample path moderate deviation principles for the current and the tagged particle when the process starts from one of its stationary measures. We simplify the proof in our previous works [Xue and Zhao, Electronic Journal of Probability, 2024] and [Xue and Zhao, Stochastic Processes and their Applications, 2023], where the same problem was investigated in the symmetric simple exclusion process. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:15:24 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Xue",
"Xiaofeng",
""
],
[
"Zhao",
"Linjie",
""
]
] |
2506.03925 | Haiye Huo | Lu Li, Haiye Huo | SVD-Based Graph Fractional Fourier Transform on Directed Graphs and Its Application | 30 pages,14 figures | null | null | null | eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Graph fractional Fourier transform (GFRFT) is an extension of graph Fourier transform (GFT) that provides an additional fractional analysis tool for graph signal processing (GSP) by generalizing temporal-vertex domain Fourier analysis to fractional orders. In recent years, a large number of studies on GFRFT based on undirected graphs have emerged, but there are very few studies on directed graphs. Therefore, in this paper, one of our main contributions is to introduce two novel GFRFTs defined on Cartesian product graph of two directed graphs, by performing singular value decomposition on graph fractional Laplacian matrices. We prove that two proposed GFRFTs can effectively express spatial-temporal data sets on directed graphs with strong correlation. Moreover, we extend the theoretical results to a generalized Cartesian product graph, which is constructed by $m$ directed graphs. Finally, the denoising performance of our proposed two GFRFTs are testified through simulation by processing hourly temperature data sets collected from 32 weather stations in the Brest region of France. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:15:40 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Li",
"Lu",
""
],
[
"Huo",
"Haiye",
""
]
] |
2506.03926 | Debarshi Brahma | Debarshi Brahma and Soma Biswas | Multiple Stochastic Prompt Tuning for Practical Cross-Domain Few Shot Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we propose a practical cross-domain few-shot learning (pCDFSL) task, where a large-scale pre-trained model like CLIP can be easily deployed on a target dataset. The goal is to simultaneously classify all unseen classes under extreme domain shifts, by utilizing only a few labeled samples per class. The pCDFSL paradigm is source-free and moves beyond artificially created episodic training and testing regimes followed by existing CDFSL frameworks, making it more challenging and relevant to real-world applications. Towards that goal, we propose a novel framework, termed MIST (MultIple STochastic Prompt tuning), where multiple stochastic prompts are utilized to handle significant domain and semantic shifts. Specifically, multiple prompts are learnt for each class, effectively capturing multiple peaks in the input data. Furthermore, instead of representing the weights of the multiple prompts as point-estimates, we model them as learnable Gaussian distributions with two different strategies, encouraging an efficient exploration of the prompt parameter space, which mitigate overfitting due to the few labeled training samples. Extensive experiments and comparison with the state-of-the-art methods on four CDFSL benchmarks adapted to this setting, show the effectiveness of the proposed framework. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:18:04 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Brahma",
"Debarshi",
""
],
[
"Biswas",
"Soma",
""
]
] |
2506.03927 | Marianne Simonsen | Marianne Simonsen, Lars Skipper and Jeffrey A. Smith | Welfare Reform: Consequences for the Children | 68 pages, 6 main figures, 7 main tables, 2 appendix tables, 13 appendix tables | null | null | null | econ.GN q-fin.EC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper uses register-based data to analyze the consequences of a recent major Danish welfare reform for children's academic performance and well-being. In addition to work requirements, the reform brought about considerable reductions in welfare transfers. We implement a comparative event study that contrasts outcomes for individuals on welfare at the time of reform announcement before and after the implementation of the reform with the parallel development in outcomes for an uncontaminated comparison group, namely those on welfare exactly one year prior. Our analysis documents that mothers' propensity to receive welfare decreased somewhat as a consequence of the reform, just as we observe a small increase in hours worked. At the same time, we do not detect negative effects on short-run child academic performance. We do find small negative effects on children's self-reported school well-being and document substantial upticks in reports to child protective services for children exposed to the reform. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:19:06 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Simonsen",
"Marianne",
""
],
[
"Skipper",
"Lars",
""
],
[
"Smith",
"Jeffrey A.",
""
]
] |
2506.03928 | Ze Feng | Ze Feng, Jiang-Jiang Liu, Sen Yang, Lingyu Xiao, Xiaofan Li, Wankou Yang, Jingdong Wang | Vision Remember: Alleviating Visual Forgetting in Efficient MLLM with Vision Feature Resample | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we study the Efficient Multimodal Large Language Model. Redundant vision tokens consume a significant amount of computational memory and resources. Therefore, many previous works compress them in the Vision Projector to reduce the number of vision tokens. However, simply compressing in the Vision Projector can lead to the loss of visual information, especially for tasks that rely on fine-grained spatial relationships, such as OCR and Chart \& Table Understanding. To address this problem, we propose Vision Remember, which is inserted between the LLM decoder layers to allow vision tokens to re-memorize vision features. Specifically, we retain multi-level vision features and resample them with the vision tokens that have interacted with the text token. During the resampling process, each vision token only attends to a local region in vision features, which is referred to as saliency-enhancing local attention. Saliency-enhancing local attention not only improves computational efficiency but also captures more fine-grained contextual information and spatial relationships within the region. Comprehensive experiments on multiple visual understanding benchmarks validate the effectiveness of our method when combined with various Efficient Vision Projectors, showing performance gains without sacrificing efficiency. Based on Vision Remember, LLaVA-VR with only 2B parameters is also superior to previous representative MLLMs such as Tokenpacker-HD-7B and DeepSeek-VL-7B. | [
{
"version": "v1",
"created": "Wed, 04 Jun 2025 13:22:35 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Feng",
"Ze",
""
],
[
"Liu",
"Jiang-Jiang",
""
],
[
"Yang",
"Sen",
""
],
[
"Xiao",
"Lingyu",
""
],
[
"Li",
"Xiaofan",
""
],
[
"Yang",
"Wankou",
""
],
[
"Wang",
"Jingdong",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.