October 2025 arXiv papers — page 11

J. Marijan, H. Strobel, M. K. Oberthaler, J. Berges

We investigate entropy transport for universal scaling phenomena in closed quantum many-body systems far from equilibrium. From spatially resolved experimental data of a spinor Bose gas, we demonstrate that entropy decreases on long-distance scales while it increases at short distances. A dynamical separation of scales leads to macrophysics with long-range o

Gilberto Colangelo, Martina Cottini, Martin Hoferichter, Simon Holz

A reliable calculation of radiative corrections to $\tau\to\pi\pi\nu_\tau$ decays is an important prerequisite for using hadronic $\tau$ decays for a data-driven evaluation of the hadronic-vacuum-polarization contribution to the anomalous magnetic moment of the muon, $a_\mu^\text{HVP, LO}[\pi\pi,\tau]$. In this Letter, we present an improved model-independen

Zakary Schofield, Vanderli Laurindo, Ori Ezrah Mor, Patrick M. Ledingham

We have demonstrated the coherent storage and retrieval of single-photon-level light using the atomic frequency comb protocol in a room temperature rubidium vapour. Velocity-selective optical pumping is used to prepare the comb within the $F=2$ hyperfine ground state of rubidium, with the spacing between peaks coinciding with half the $F = 2 - F =3$ hyperfin

Rotem Ezra, Hedi Zisling, Nimrod Berman, Ilan Naiman

Diffusion models have become state-of-the-art generative models for images, audio, and video, yet enabling fine-grained controllable generation, i.e., continuously steering specific concepts without disturbing unrelated content, remains challenging. Concept Sliders (CS) offer a promising direction by discovering semantic directions through textual contrasts,

Ziyu Guo, Xinyan Chen, Renrui Zhang, Ruichuan An

Recent video generation models can produce high-fidelity, temporally coherent videos, indicating that they may encode substantial world knowledge. Beyond realistic synthesis, they also exhibit emerging behaviors indicative of visual perception, modeling, and manipulation. Yet, an important question still remains: Are video models ready to serve as zero-shot

Jing-Yu Zhao, Ya-Hui Zhang

Mechanism of superconductivity in twisted bilayer graphene (TBG) remains one of the central problems in correlated moir\'e materials. The most intriguing question is about the nature of the normal state: is the Cooper pair formed from small Fermi surface or large Fermi surface? In this work we point out the possibility of a symmetric pseudogap metal with sma

Chao Feng, Zihao Wei, Andrew Owens

We learn visual features by captioning images with an image-conditioned masked diffusion language model, a formulation we call masked diffusion captioning (MDC). During training, text tokens in each image-caption pair are masked at a randomly chosen ratio, and a decoder conditioned on visual features is trained to reconstruct the original text. After trainin

Saranesh Prembabu, Rahul Sahay, Stefan Divic, Ashvin Vishwanath

Spin polarons are bound states of electrons and spin-flips that form above spin polarized electronic insulators.These bound states conventionally form in one of two settings: in frustrated lattices with dispersive bands -- where the motion of an electron preferences binding a nearby spin-flip -- or in topological flat bands -- where the Chern number enforces

Yu-En Wong, Songtao Chen

High-fidelity spin readout is a crucial component for quantum information processing with optically interfaced solid-state spins. Here, we propose and investigate two theoretical protocols for fast single-shot readout of cavity-coupled single T center electronic spins. For fluorescence-based readout, we selectively couple one of the T center spin-conserving

Dongyue Lu, Ao Liang, Tianxin Huang, Xiao Fu

Immersive applications call for synthesizing spatiotemporal 4D content from casual videos without costly 3D supervision. Existing video-to-4D methods typically rely on manually annotated camera poses, which are labor-intensive and brittle for in-the-wild footage. Recent warp-then-inpaint approaches mitigate the need for pose labels by warping input frames al

Philipp Lindenberger, Paul-Edouard Sarlin, Jan Hosang, Matteo Balice

Determining the precise geographic location of an image at a global scale remains an unsolved challenge. Standard image retrieval techniques are inefficient due to the sheer volume of images (>100M) and fail when coverage is insufficient. Scalable solutions, however, involve a trade-off: global classification typically yields coarse results (10+ kilometers),

Usha Bhalla, Alex Oesterling, Claudio Mayrink Verdun, Himabindu Lakkaraju

Translating the internal representations and computations of models into concepts that humans can understand is a key goal of interpretability. While recent dictionary learning methods such as Sparse Autoencoders (SAEs) provide a promising route to discover human-interpretable features, they often only recover token-specific, noisy, or highly local concepts.

Jing Lin, Ruisi Wang, Junzhe Lu, Ziqi Huang

Despite recent advances in 3D human motion generation (MoGen) on standard benchmarks, existing text-to-motion models still face a fundamental bottleneck in their generalization capability. In contrast, adjacent generative fields, most notably video generation (ViGen), have demonstrated remarkable generalization in modeling human behaviors, highlighting trans

Nafid Enan, Gias Uddin

Logs provide valuable insights into system runtime and assist in software development and maintenance. Log parsing, which converts semi-structured log data into structured log data, is often the first step in automated log analysis. Given the wide range of log parsers utilizing diverse techniques, it is essential to evaluate them to understand their characte

Dominic Agius, Tracy Robyn Slatyer

The 21-cm signal from the epoch of cosmic dawn ($z \sim 10-30$) offers a powerful probe of new physics. One standard mechanism for constraining decaying dark matter from 21-cm observations relies on heating of the intergalactic medium by the decay products, an effect whose observability is entangled with the uncertain Lyman-$\alpha$ fluxes and X-ray heating

Hyunji Lee, Minseon Kim, Chinmay Singh, Matheus Pereira

As coding agents are increasingly deployed in large codebases, the need to automatically design challenging, codebase-level evaluation is central. We propose Gistify, a task where a coding LLM must create a single, minimal, self-contained file that can reproduce a specific functionality of a codebase. The coding LLM is given full access to a codebase along w

Shao-Hua Hu, Po-Sung Liu, Jun-Yi Wu

Distributed quantum computing (DQC) provides a promising route toward scalable quantum computation, where entanglement-assisted LOCC and circuit knitting represent two complementary approaches. The former deterministically realizes nonlocal operations but demands extensive entanglement resources, whereas the latter requires no entanglement yet suffers from e

Penghui Qi, Zichen Liu, Xiangxin Zhou, Tianyu Pang

Reinforcement learning (RL) fine-tuning of large language models (LLMs) often suffers from instability due to the numerical mismatch between the training and inference policies. While prior work has attempted to mitigate this issue through algorithmic corrections or engineering alignments, we show that its root cause lies in the floating point precision itse

Mantas Mazeika, Alice Gatti, Cristina Menghini, Udari Madhushani Sehwag

AIs have made rapid progress on research-oriented benchmarks of knowledge and reasoning, but it remains unclear how these gains translate into economic value and automation. To measure this, we introduce the Remote Labor Index (RLI), a broadly multi-sector benchmark comprising real-world, economically valuable projects designed to evaluate end-to-end agent p

Cheng Zheng, William Koch, Baiang Li, Felix Heide

Hierarchical structures of motion exist across research fields, including computer vision, graphics, and robotics, where complex dynamics typically arise from coordinated interactions among simpler motion components. Existing methods to model such dynamics typically rely on manually-defined or heuristic hierarchies with fixed motion primitives, limiting thei

Joshua D. Lothringer, Nataliea Lowson, Guangwei Fu

The present-day bulk elemental composition of an exoplanet can provide insight into a planet's formation and evolutionary history. Such information is now being measured for dozens of planets with state-of-the-art facilities using Bayesian atmosphere retrievals. We collect measurements of exoplanet composition of gas giants into a Library of Exoplanet Atmosp

Arnab Sen Sharma, Giordano Rogers, Natalie Shapira, David Bau

We investigate the mechanisms underlying a range of list-processing tasks in LLMs, and we find that LLMs have learned to encode a compact, causal representation of a general filtering operation that mirrors the generic "filter" function of functional programming. Using causal mediation analysis on a diverse set of list-processing tasks, we find that a small

Masahiro Kato

This note introduces a unified theory for causal inference that integrates Riesz regression, covariate balancing, density-ratio estimation (DRE), targeted maximum likelihood estimation (TMLE), and the matching estimator in average treatment effect (ATE) estimation. In ATE estimation, the balancing weights and the regression functions of the outcome play impo

Zaishuo Xia, Yukuan Lu, Xinyi Li, Yifan Xu

A world model is an internal model that simulates how the world evolves. Given past observations and actions, it predicts the future physical state of both the embodied agent and its environment. Accurate world models are essential for enabling agents to think, plan, and reason effectively in complex, dynamic settings. However, existing world models often fo

Aniruddh Bansal, Davit Soselia, Dang Nguyen, Tianyi Zhou

Charts play an important role in visualization, reasoning, data analysis, and the exchange of ideas among humans. However, existing vision-language models (VLMs) still lack accurate perception of details and struggle to extract fine-grained structures from charts. Such limitations in chart grounding also hinder their ability to compare multiple charts and re

Meisen Gao, Zhong-Bo Kang, Jani Penttala, Ding Yu Shao

In the high-energy limit of QCD, scattering off nucleons and nuclei can be described in terms of Wilson-line correlators whose energy dependence is perturbative. The energy dependence of the two-point correlator, called the dipole amplitude, is governed by the Balitsky-Kovchegov (BK) equation. The initial condition for the BK equation can be fitted to the ex

Valentyna Starodub, Mantas Lukoševičius

Age-related macular degeneration (AMD) is one of the leading causes of irreversible vision impairment in people over the age of 60. This research focuses on semantic segmentation for AMD lesion detection in RGB fundus images, a non-invasive and cost-effective imaging technique. The results of the ADAM challenge - the most comprehensive AMD detection from RGB

Andreas Auer, Daniel Klotz, Sebastinan Böck, Sepp Hochreiter

Recent research on time series foundation models has primarily focused on forecasting, leaving it unclear how generalizable their learned representations are. In this study, we examine whether frozen pre-trained forecasting models can provide effective representations for classification. To this end, we compare different representation extraction strategies

Jungyeon Koh, Hyeonsu Lyu, Jonggyu Jang, Hyun Jong Yang

How can we explain the influence of training data on black-box models? Influence functions (IFs) offer a post-hoc solution by utilizing gradients and Hessians. However, computing the Hessian for an entire dataset is resource-intensive, necessitating a feasible alternative. A common approach involves randomly sampling a small subset of the training data, but

Yin Tang, Yanyuan Ma, Bing Li

We conduct a KL-divergence based procedure for testing elliptical distributions. The procedure simultaneously takes into account the two defining properties of an elliptically distributed random vector: independence between length and direction, and uniform distribution of the direction. The test statistic is constructed based on the $k$ nearest neighbors ($

M. Guolo, A. Mummery, S. van Velzen, S. Gezari

We present a multi-wavelength analysis of 14 tidal disruption events (TDEs)-including an off-nuclear event associated with an ultra-compact dwarf galaxy-selected for having available thermal X-ray spectra during their late-time UV/optical plateau phase. We show that at these stages, the full spectral energy distribution - X-ray spectra and UV/optical photome

Alice Colpani Serri, Chris A. Flett, Jean-Philippe Lansberg, Olivier Mattelaer

We present an extension of the MadGraph5_aMC@NLO framework that enables the automated calculation of leading-order cross sections for S-wave quarkonium and leptonium production within the non-relativistic QCD (NRQCD) and non-relativistic QED (NRQED) factorisation formalisms. The framework has been validated against a variety of benchmark processes, demonstra

S. D Campos

The thermodynamic properties of black holes have been extensively studied through analogies with classical systems, revealing fundamental connections between gravitation, entropy, and quantum mechanics. In this work, we extend the thermodynamic framework of black holes by incorporating charge and analyzing its role in entropy production. Using an analogy wit

Marco Federici, Riccardo Del Chiaro, Boris van Breugel, Paul Whatmough

Quantization is the key method for reducing inference latency, power and memory footprint of generative AI models. However, accuracy often degrades sharply when activations are quantized below eight bits. Recent work suggests that invertible linear transformations (e.g. rotations) can aid quantization, by reparameterizing feature channels and weights. In thi

Mark Gross, Fatemeh Rezaee

We give a detailed analysis of the stability scattering diagram for $\mathbb{P}^2$ introduced by Bousseau. This scattering diagram lives in a subset of $\mathbb{R}^2$, and we decompose this subset into three regions, $R_{\Delta},R_{\Diamond}$ and $R_{\mathrm{unbdd}}$. The region $R_{\Delta}$ has a chamber structure whose chambers are in one-to-one correspond

Anushka Sivakumar, Andrew Zhang, Zaber Hakim, Chris Thomas

This work introduces SteerVLM, a lightweight steering module designed to guide Vision-Language Models (VLMs) towards outputs that better adhere to desired instructions. Our approach learns from the latent embeddings of paired prompts encoding target and converse behaviors to dynamically adjust activations connecting the language modality with image context.

Shengnan An, Xunliang Cai, Xuezhi Cao, Xiaoyu Li

We present AMO-Bench, an Advanced Mathematical reasoning benchmark with Olympiad level or even higher difficulty, comprising 50 human-crafted problems. Existing benchmarks have widely leveraged high school math competitions for evaluating mathematical reasoning capabilities of large language models (LLMs). However, many existing math competitions are becomin

Alexander Steier, Shamik Ghosh, Jacques Delabrouille

The detection of primordial gravitational waves in Cosmic Microwave Background B-mode polarization observations requires accurate and robust subtraction of astrophysical contamination. We show, using a blind Spectral Matching Independent Component Analysis, that it is possible to infer unbiased estimates of the primordial B-mode signal from ground-based obse

CMS Collaboration

For the first time, correlations among mixed-order moments of two or three flow harmonics$-$($v_{n}^{k},v_{m}^{l}$) and ($v_{n}^{k},v_{m}^{l}, v_{p}^{q}$), with $k$, $l$, and $q$ denoting the respective orders$-$are measured in xenon-xenon (XeXe) collisions and compared with lead-lead (PbPb) results, providing a novel probe of collective behavior in heavy io

Jiawei Gu, Yunzhuo Hao, Huichen Will Wang, Linjie Li

Multimodal reasoning requires iterative coordination between language and vision, yet it remains unclear what constitutes a meaningful interleaved chain of thought. We posit that text and image thoughts should function as complementary rather than isomorphic modalities that mutually advance reasoning. Guided by this principle, we build ThinkMorph, a unified

Mingu Kang, Pavel E. Dolgirev, Chao C. Zhang, Hoyoung Jang

Phase fluctuations are a key factor distinguishing nonthermal (ultrafast) and thermal phase transitions. Charge order in cuprates is characterized by short-range coherence while competing with superconductivity, and as such, it provides a representative case to study the role of phase fluctuation in coupled order parameter dynamics. In this work, we investig

Jiajie Guo, Feng-Xiao Sun, Matteo Fadel, Qiongyi He

Additional state evolutions performed before measurement, also called measurement-after-interactions (MAI) protocols, have shown a great potential for increasing the sensitivity of metrological scenarios. Here, we go beyond this result and show that MAI techniques can significantly enhance the detection capability of witnesses for quantum correlations. In pa

Shaokai Wu, Yapan Guo, Yanbiao Ji, Jing Tong

CT reconstruction provides radiologists with images for diagnosis and treatment, yet current deep learning methods are typically limited to specific anatomies and datasets, hindering generalization ability to unseen anatomies and lesions. To address this, we introduce the Multi-Organ medical image REconstruction (MORE) dataset, comprising CT scans across 9 d

Bertrand Teguia Tabuguia

Given finitely many consecutive terms of an infinite sequence, we discuss the construction of a polynomial difference equation that the sequence may satisfy. We also present a method to seek a candidate polynomial differential equation for its generating function. It appears that these methods often lead to effective D-algebraic operations.

Shozab Qasim, Jason Pollack

Quantum error correction, thermalization, and quantum chaos are fundamental aspects of quantum many-body physics that have each developed largely independently, despite their deep conceptual overlap. In this work, we establish a precise link between all three in systems that satisfy the eigenstate thermalization hypothesis (ETH) and exhibit a well-defined hi

Javier González-Anaya, Brett Nasserden, Sasha Zotine

We study surjective endomorphisms of projective bundles over toric varieties, achieving three main results. First, we provide a structural theorem describing endomorphisms of projectivized split bundles over arbitrary base varieties, which we use to classify all surjective endomorphisms of Hirzebruch surfaces and construct novel families of examples. Second,

Soujanya Hazra, Sanjay Ghosh

Electroencephalography (EEG) acquisition is often affected by large-amplitude variations across subjects and sessions. This makes conventional front ends vulnerable to saturation and clipping. Modulo sampling addresses this limitation by folding the signal into a bounded range rather than truncating it. The recovery task is then to reconstruct the original w

Benjamin Freiman, Xinyuan You, Andy C. Y. Li, Raphael Cervantes

We present a quantum-enhanced protocol for detecting wave-like dark matter using an array of $N$ entangled superconducting cavities initialized in an $m$-photon Fock state. By distributing and recollecting the quantum state with an entanglement-distribution operation, the scan rate scales as $N^2(m+1)$ while thermal excitation is the dominant background, sig

Niccolò Preti, Nicolò Antolini, Charles Drevon, Pietro Lombardi

The famous two-fluid model of finite-temperature superfluids has been recently extended to describe the mixed classical-superfluid dynamics of the newly discovered supersolid phase of matter. We show that for rigidly rotating supersolids one can derive a more appropriate single-fluid model, in which the seemingly classical and superfluid contributions to the

William Overman, Mohsen Bayati

As increasingly capable agents are deployed, a central safety challenge is how to retain meaningful human control without modifying the underlying system. We study a minimal control interface in which an agent chooses whether to act autonomously (play) or defer (ask), while a human simultaneously chooses whether to be permissive (trust) or engage in oversigh

Gianluca Levi, Max Kroesbergen, Louis Thirion, Yorick L. A. Schmerwitz

Rydberg excited states of molecules pose a challenge for electronic structure calculations because of their highly diffuse electron distribution. Even large and elaborate atomic basis sets tend to underrepresent the long-range tail, overly confining the Rydberg state. An approach is presented here where the molecular orbitals are variationally optimized for

Reham Faqehi, Haya Alhuraib, Hamad Saiari, Zyad Bamigdad

In industrial and IoT environments, massive amounts of real-time and historical process data are continuously generated and archived. With sensors and devices capturing every operational detail, the volume of time-series data has become a critical challenge for storage and processing systems. Efficient data management is essential to ensure scalability, cost

Martim Afonso, Nuno Saavedra, Bruno Lourenço, Alexandra Mendes

Systematic reviews and mapping studies are critical to synthesize research, identify gaps, and guide future work, but are often labor-intensive and time-consuming. Existing tools provide partial support for specific steps, leaving much of the process manual and error-prone. We present ProfOlaf, a semi-automated tool designed to streamline systematic reviews

Roman Berens, Lam Hui, Daniel McLoughlin, Adam R. Solomon

We compute the ladder operators for static tidal perturbations to higher-dimensional black holes. These operators map between solutions of the relevant equation of motion at different multipole orders. We focus on spin 0, 1 and 2 perturbations to the Schwarzschild-Tangherlini black hole and on spin 0 perturbations to the 5D Myers-Perry black hole. The ladder

Michelle Zuo

This paper quantitatively analyzes county-level voting patterns in Wisconsin's presidential elections from 2000 to 2024. As a pivotal swing state, Wisconsin has alternated between Democratic and Republican candidates since 2012. Using data from the Wisconsin Elections Commission, we examine vote totals across 72 counties and seven election cycles. Pearson co

Thomas Gawne, Alina Kononov, Andrew Baczewski, Hannah Bellenbaum

X-ray Thomson scattering (XRTS) probes the dynamic structure factor of the system, but the measured spectrum is broadened by the combined source-and-instrument function (SIF) of the setup. In order to extract properties such as temperature from an XRTS spectrum, the broadening by the SIF needs to be removed. Recent work [Dornheim et al. Nature Commun. 13, 79

Rudradip Biswas, Dimitra-Dionysia Stergiopoulou

In this short note, we characterise some Gorenstein versions of the concept of a group being of type $\Phi$ as introduced by Olympia Talelli. And, we also generalize a different Talelli result regarding the coincidence of the classical and the Gorenstein cohomological dimension of torsion-free groups in Kropholler's $\LH\mathscr{F}$ class.

Yang Hu, Donald Stanley

It is a classical problem in algebraic topology to decide whether a given graded $\mathbb{Z}$-algebra can be realized as the cohomology ring of a space. In this paper, we introduce families of Stanley-Reisner algebras depending on graphs, and relate their realizability to the span coloring of the graph.

Bernd C. Kellner

Extending previous work of the author, we compute the Wilson quotient modulo $p^5$ and $p^6$, and equivalently $(p-1)!$ modulo $p^6$ and $p^7$, respectively. Further, we determine some power sums of the Fermat quotients up to modulo $p^6$. Subsequently, we discuss some patterns that occur in the $p$-adic coefficients of the Wilson quotient as well as of $(p-

Yunchao Ma, Yizhuang Zhou, Yunhuan Yang, Tiancai Wang

In this paper, we show how to run pi0-level multi-view VLA at 30Hz frame rate and at most 480Hz trajectory frequency using a single consumer GPU. This enables dynamic and real-time tasks that were previously believed to be unattainable by large VLA models. To achieve it, we introduce a bag of strategies to eliminate the overheads in model inference. The real

Rafael Ballabriga, Eric Buschmann, Michael Campbell, Raimon Casanova Mohr

The H2M (Hybrid-to-Monolithic) is a monolithic pixel sensor manufactured in a modified \SI{65}{\nano\meter}~CMOS imaging process with a small collection electrode. Its design addresses the challenges of porting an existing hybrid pixel detector architecture into a monolithic chip, using a digital-on-top design methodology, and developing a compact digital ce

Ashwin Kumar, William Yeoh

We introduce the General Incentives-based Framework for Fairness (GIFF), a novel approach for fair multi-agent resource allocation that infers fair decision-making from standard value functions. In resource-constrained settings, agents optimizing for efficiency often create inequitable outcomes. Our approach leverages the action-value (Q-)function to balance

Noam Mouelle, Jeremiah Mitchell, Valerie Gibson, Ulrich Schneider

Time-resolved atom interferometry, as employed in applications such as gravitational wave detection and searches for ultra-light dark matter, requires precise control over systematic effects. In this work, we investigate phase noise arising from shot-to-shot fluctuations in the atoms' transverse motion in the presence of the wavefront curvature of the interf

Anthony Francis, Patrick Fritzsch, Rohith Karur, Jangho Kim

We present a nonperturbative determination of the pion valence parton distribution function (PDF) moment ratios $\left\langle x^{n-1} \right\rangle / \left\langle x \right\rangle$ up to $n=6$, using the gradient flow in lattice QCD. As a testing ground, we employ SU($3$) isosymmetric gauge configurations generated by the OpenLat initiative with a pseudoscala

James Broda, Alanna Haslam-Hyde, Mary Lou Zeeman

Even if a linear system of ordinary differential equations has a globally attracting equilibrium at the origin, small disturbances from the equilibrium may lead to large transient excursions before the system stabilizes. This counter-intuitive phenomenon of transient amplification is called reactivity and is often associated with systems that are non-normal.

Narendra N. Hegade, Nachiket L. Kortikar, Balaganchi A. Bhargava, Juan F. R. Hernández

We propose digitized counterdiabatic quantum sampling (DCQS), a hybrid quantum-classical algorithm for efficient sampling from energy-based models, such as low-temperature Boltzmann distributions. The method utilizes counterdiabatic protocols, which suppress non-adiabatic transitions, with an iterative bias-field procedure that progressively steers the sampl

Vivan Doshi

The discovery of conservation laws is a cornerstone of scientific progress. However, identifying these invariants from observational data remains a significant challenge. We propose a hybrid framework to automate the discovery of conserved quantities from noisy trajectory data. Our approach integrates three components: (1) a Neural Ordinary Differential Equa

Daniel Lännström, Patrik Lundström, Johan Öinert, Stefan Wagner

Let $G$ be a group with identity element $e$, and suppose that $S$ is an associative $G$-graded ring that is not necessarily unital. In the case where $G$ is an ordered group, we show that a graded ideal is prime if and only if it is graded prime. Consequently, in that setting, a graded ring is prime if and only if it is graded prime. For any group $G$, if $

Luca Basta, Francesca Picca, Pegah Darvehi, Vincenzo Pagliara

The interaction between hydrogen atoms and carbon nanoparticles is a fundamental process governing the properties of carbonaceous materials in environments ranging from combustion systems to the interstellar medium. This study investigates the effects of controlled atomic hydrogen exposure on young and mature soot nanoparticles, generated in premixed ethylen

J. de Curtò, I. de Zarzà, Pablo García, Jordi Cabot

This paper presents a comprehensive cross-platform evaluation of reasoning capabilities in contemporary foundation models, establishing an infrastructure-agnostic benchmark across three computational paradigms: HPC supercomputing (MareNostrum 5), cloud platforms (Nebius AI Studio), and university clusters (a node with eight H200 GPUs). We evaluate 15 foundat

Maria Acevedo, Daniel Scolnic, Bastien Carreres, Erik R. Peterson

We present an overview of the Weighing Halos Accurately, Locally, and Efficiently with Supernovae (WHALES) survey, the first to discover and measure Type Ia supernovae (SNe Ia) in and around galaxy superclusters. By building a sample of SNe~Ia around these massive environments, we aim to provide new constraints on bulk-flow models while laying the groundwork

Zixu Shen, Kexin Chu, Yifan Zhang, Dawei Xiang

The expansion of large language models is increasingly limited by the constrained memory capacity of modern GPUs. To mitigate this, Mixture-of-Experts (MoE) architectures activate only a small portion of parameters during inference, significantly lowering both memory demand and computational overhead. However, conventional MoE inference approaches, which sel

Alberto Accardi, Matteo Cerutti

We present recent updates from the CTEQ-JLab (CJ) global PDF analysis, focusing on the interplay and implementation systematics of the HT and offshell correction (CJ22ht). We also discuss preliminary results of the CJ25 global analysis, showing the impact of the full JLab 6 GeV datasets, that we recently collected in a comprehensive DIS database, and having

Ivo Siekmann

Aggregated Markov models provide a flexible framework for stochastic dynamics that develops on multiple timescales. For example, Markov models for ion channels often consist of multiple open and closed state to account for "slow" and "fast" openings and closings of the channel. The approach is a popular tool in the construction of mechanistic models of ion c

Haoyi Zhang, Tianyi Zhu

Regulators currently govern the AI data economy based on intuition rather than evidence, struggling to choose between inconsistent regimes of informed consent, immunity, and liability. To fill this policy vacuum, this paper develops a novel computational policy laboratory: a spatially explicit Agent-Based Model (ABM) of the data market. To solve the problem

Daniel O'Hanlon

In Bayesian hierarchical models, group-level parameter arrays must be mapped to the observation axis, often using explicit indexing. In complex models with numerous incompatible data and parameter sets, this introduces the potential for bugs, as indexing with the incorrect indices typically fails silently. Here we present typegeist, a type system for Python

Zsolt Bartha, Brett Kolesnik, Gal Kronenberg, Yuval Peled

We study graph bootstrap percolation on the Erd\H{o}s-R\'enyi random graph ${\mathcal G}_{n,p}$. For all $r \ge 5$, we locate the sharp $K_r$-percolation threshold $p_c \sim (\gamma n)^{-1/\lambda}$, solving a problem of Balogh, Bollob\'as and Morris. The case $r=3$ is the classical graph connectivity threshold, and the threshold for $r=4$ was found using st

Masahiro Kato

The goal of policy learning is to train a policy function that recommends a treatment given covariates to maximize population welfare. There are two major approaches in policy learning: the empirical welfare maximization (EWM) approach and the plug-in approach. The EWM approach is analogous to a classification problem, where one first builds an estimator of

Muhammad Faraz Ul Abrar, Nicolò Michelusi

Over-the-air (OTA) federated learning (FL) has been well recognized as a scalable paradigm that exploits the waveform superposition of the wireless multiple-access channel to aggregate model updates in a single use. Existing OTA-FL designs largely enforce zero-bias model updates by either assuming \emph{homogeneous} wireless conditions (equal path loss acros

Xinhan Zheng, Huyu Wu, Xueting Wang, Duo Su

Multimodal large language models (MLLMs) exhibit a pronounced preference for textual inputs when processing vision-language data, limiting their ability to reason effectively from visual evidence. Unlike prior studies that attribute this text bias to external factors such as data imbalance or instruction tuning, we propose that the bias originates from the m

Li Hu, Rong-Gen Cai, Shao-Jiang Wang

Recently, Ruffini et al. [Phys. Rev. Lett. 134 (2025) 8, 081403] pointed out that the repetitive Penrose process cannot drain the entire extractable energy of a Kerr black hole. In this Letter, we alternatively point out the charge of a Reissner-Nordstr\"{o}m black hole cannot drop down to exactly zero via the repetitive electric Penrose process that is term

Xuan Zou, Zhou-Quan Wan, Hong Yao

Charge-$4e$ superconductivity is an exotic state of matter that may emerge as a vestigial order from a charge-$2e$ superconductor with multicomponent superconducting order parameters. Showing its emergence in a lattice phase model from numerically exact large-scale computations has remained rare. Here, we propose a kagome lattice model with a nematic superco

Fenfen Lin, Yesheng Liu, Haiyu Xu, Chen Yue

Reading measurement instruments is effortless for humans and requires relatively little domain expertise, yet it remains surprisingly challenging for current vision-language models (VLMs) as we find in preliminary evaluation. In this work, we introduce MeasureBench, a benchmark on visual measurement reading covering both real-world and synthesized images of

Isaac Robledo, Juan Alfaro, Víctor Duro, Alberto Solera-Rico

Aerodynamic drag on flat-backed vehicles like vans and trucks is dominated by a low-pressure wake, whose control is critical for reducing fuel consumption. This paper presents an experimental study at $Re_W\approx 78,300$ on active flow control using four pulsed jets at the rear edges of a bluff body model. A hybrid genetic algorithm, combining a global sear

Tommaso d'Orsi, Gleb Novikov

We present a simple perturbation mechanism for the release of $d$-dimensional covariance matrices $\Sigma$ under pure differential privacy. For large datasets with at least $n\geq d^2/\varepsilon$ elements, our mechanism recovers the provably optimal Frobenius norm error guarantees of \cite{nikolov2023private}, while simultaneously achieving best known error

Joongho Kim, Xirui Huang, Zarreen Reza, Gabriel Grand

Tree-of-Thought (ToT) reasoning boosts the problem-solving abilities of Large Language Models (LLMs) but is computationally expensive due to semantic redundancy, where distinct branches explore equivalent reasoning paths. We introduce Semantic Similarity-Based Dynamic Pruning (SSDP), a lightweight method that, to the best of our knowledge, is the first frame

Erik J. Wasleske, Vivienne F. Baldassare, Christopher M. Carroll

We investigate the connection between accretion signatures and host galaxy properties in the context of how active dwarf galaxies are identified. We use the database constructed in Wasleske & Baldassare (2024) which contains dwarf galaxies that were selected as active galaxies by optical spectroscopy, infrared colors, X-ray brightness, and photometric variab

Yuchen Zhang, Hanyue Du, Chun Cao, Jingwei Xu

Low-Rank Adaptation (LoRA) has become a widely adopted parameter-efficient fine-tuning (PEFT) technique for adapting large language models (LLMs) to downstream tasks. While prior work has explored strategies for integrating LLM training and serving, there still remains a gap in unifying fine-tuning and inference for LoRA-based models. We present Loquetier, a

Gabriel Asher, Devesh Shah, Amy A. Caudy, Luke Ferro

A vast majority of mass spectrometry data remains uncharacterized, leaving much of its biological and chemical information untapped. Recent advances in machine learning have begun to address this gap, particularly for tasks such as spectral identification in tandem mass spectrometry data. Here, we present the latest generation of LSM-MS2, a large-scale deep

Aminul Hussain, Nisa Ara, Rudranil Basu, Sudeshna Sen

The interplay between non-trivial band topology and strong electronic correlations is a central challenge in modern condensed matter physics. We investigate this competition on a two-leg ladder model with a p-wave-like hybridisation between the legs. This model hosts a symmetry-protected topological phase in its non-interacting limit. Using the density-matri

Renato Quartullo, Andrea Garulli, Mirko Leomanni

This paper presents a time-optimal Model Predictive Control (MPC) scheme for linear discrete-time systems subject to multiplicative uncertainties represented by interval matrices. To render the uncertainty propagation computationally tractable, the set-valued error system dynamics are approximated using a matrix-zonotope-based bounding operator. Recursive fe

Nguyen Nang Thieu, Nguyen Dong Yen

This paper establishes three minimax theorems for possibly nonconvex functions on Euclidean spaces or on infinite-dimensional Hilbert spaces. The theorems also guarantee the existence of saddle points. As a by-product, a complete solution to an interesting open problem related to continuously differentiable functions is obtained. The obtained results are ana

Chuyan Chen, Chenyang Ma, Zhangxin Li, Yutong He

Communication remains a central bottleneck in large-scale distributed machine learning, and gradient sparsification has emerged as a promising strategy to alleviate this challenge. However, existing gradient compressors face notable limitations: Rand-$K$ discards structural information and performs poorly in practice, while Top-$K$ preserves informative entr

Bowen Li, Jiping Luo, Themistoklis Charalambous, Nikolaos Pappas

Guaranteeing stringent data freshness for low-altitude unmanned aerial vehicles (UAVs) in shared spectrum forces a critical trade-off between two operational costs: the UAV's own energy consumption and the occupation of terrestrial channel resources. The core challenge is to satisfy the aerial data freshness while finding a Pareto-optimal balance between the

Giulia DeSalvo, Clara Mohri, Mehryar Mohri, Yutao Zhong

Learning to defer uncertain predictions to costly experts offers a powerful strategy for improving the accuracy and efficiency of machine learning systems. However, standard training procedures for deferral algorithms typically require querying all experts for every training instance, an approach that becomes prohibitively expensive when expert queries incur

Qiwei Chen, C. Ricardo Constante-Amores

Rayleigh-Benard convection (RBC) is a canonical system for buoyancy-driven turbulence and heat transport, central to geophysical and industrial flows. Developing efficient control strategies remains challenging at high Rayleigh numbers, where fully resolved simulations are computationally expensive. We use a control framework that couples data-driven manifol

Nick Heilenkötter

Can regularization terms in the training of invertible neural networks lead to known Bayesian point estimators in reconstruction? Invertible networks are attractive for inverse problems due to their inherent stability and interpretability. Recently, optimization strategies for invertible neural networks that approximate either a reconstruction map or the for

Paul F. R. Wilson, Mohamed Harmanani, Minh Nguyen Nhat To, Amoon Jamzad

Purpose: Medical foundation models (FMs) offer a path to build high-performance diagnostic systems. However, their application to prostate cancer (PCa) detection from micro-ultrasound ({\mu}US) remains untested in clinical settings. We present ProstNFound+, an adaptation of FMs for PCa detection from {\mu}US, along with its first prospective validation. Meth

Majed El Helou, Chiara Troiani, Benjamin Ryder, Jean Diaconu

Authorizing Large Language Model driven agents to dynamically invoke tools and access protected resources introduces significant risks, since current methods for delegating authorization grant overly broad permissions and give access to tools allowing agents to operate beyond the intended task scope. We introduce and assess a delegated authorization model en

Akhila Kandivalasa, Marcos Netto

The proposed approach yields a numerical method that provably executes in linear time with respect to the number of nodes and edges in a graph. The graph, constructed from the power system model, requires only knowledge of the dependencies between state-to-state and output-to-state variables within a state-space framework. While graph-based observability ana