April 2026 arXiv papers — page 93

Muxin Pu, Xiao-Ming Wu, Mei Kuan Lim, Chun Yong Chong

We present SignDPO, a novel multi-level Direct Preference Optimisation (DPO) framework designed to enhance the alignment of skeleton-based Sign Language Translation. While current skeleton-based models have made significant progress using Maximum Likelihood Estimation, they are primarily constrained by an imitation-based paradigm that lacks discriminative se

A. M. Morgen, S. S. Balling, M. T. Strøe, T. G. Skov

Impurities embedded in a Bose-Einstein Condensate (BEC) of 39K atoms are investigated with a pump-probe ejection spectroscopy sequence. The spectroscopic signal exhibits a strong feature corresponding to a Bose polaron in agreement with prior injection spectroscopy and theory. In addition, significant spectral weight at energies well below the energy of the

Pan Wang, Yihao Hu, Xiujin Liu, Hang Wang

Traditional shadow removal networks often treat image restoration as an unconstrained mapping, lacking the physical interpretability required to balance localized texture recovery with global illumination consistency. To address this, we propose CFSR, a multi-modal prior-driven framework that reframes shadow removal as a physics-constrained restoration proce

Wen Tao, Yiwei Wang, Peng Zhou, Bryan Hooi

Molecule generation requires satisfying multiple chemical and biological constraints while searching a large and structured chemical space. This makes it a non-binary problem, where effective models must identify non-obvious solutions under constraints while maintaining exploration to improve success by escaping local optima. From this perspective, creativit

Sourav Bhattacharya, Moutushi Dutta Choudhury

We have considered the Yukawa theory at two loop in the inflationary de Sitter spacetime, for a massless minimally coupled scalar and a massless fermion. The one loop computation for the same has been investigated in detail in the earlier literatures. The chief motivation behind this study is the fact that at one loop, the scalar self energy contains only fe

Abhijit Sadhukhan, Adri Bhattacharya, Anisur Rahaman Molla

We study the message complexity of leader election in synchronous networks of diameter two. Our main contribution is a refined analysis of the randomized algorithm proposed by Chatterjee et al. [DC, 2020]. In their work, the authors established a lower bound of $\Omega(n)$ messages ($n$ is the number of nodes in the network) and presented a randomized algori

Yu-Hang Guo, Han-Tao Jing, Ming-Yi Dong, Zhi-Ping Li

China's first proton test beam facility, named the High-energy Proton-beam Experiment Station (HPES), is currently under construction in campus of CSNS, as part of the CSNS-II project. Utilizing protons slowly extracted from the Rapid Cycling Synchrotron of CSNS, HPES will deliver 1.6 GeV proton beam with an adjustable flux ranging from 1E3 to 1E8 protons pe

Shangyu Li, Juyong Jiang, Meibo Ren, Sizhe Zhong

Transpilation, or code translation, aims to convert source code from one programming language (PL) to another. It is beneficial for many downstream applications, from modernizing large legacy codebases to augmenting data for low-resource PLs. Recent large language model (LLM)-based approaches have demonstrated immense potential for code translation. Among th

Enze Pan

Many deployed systems expose black-box objectives whose minimizing configuration shifts with an externally observed context. When contexts revisit a small set of latent regimes, an optimizer that discards history pays repeated adaptation cost; when each step must remain inexpensive, full Gaussian-process (GP) refits at high observation counts are difficult t

Nils Lid Hjort, M. C. Jones

This paper develops a nonparametric density estimator with parametric overtones. Suppose $f(x,\theta)$ is some family of densities, indexed by a vector of parameters $\theta$. We define a local kernel smoothed likelihood function which for each $x$ can be used to estimate the best local parametric approximant to the true density. This leads to a new density

S. Dittmaier

We review the salient features of next-to-leading-order QCD and electroweak corrections to the scattering of two and the production of three weak gauge bosons at the Large Hadron Collider. Results for the tower of $O(\alpha_s^m\alpha^n)$ corrections are shown for the exemplary processes of like-sign WW scattering and triple-W production, emphasizing the larg

Mudi Jiang, Jiahui Zhou, Xinying Liu, Zengyou He

In multi-view clustering, the quality of different views may vary substantially, and low-quality or degraded views can impair overall clustering performance. However, existing studies mainly address this issue within the clustering process through view weighting or noise-robust optimization, while paying limited attention to data-level assessment before clus

L. Feher, H. R. Dullin

We investigate certain Liouville integrable systems constructed earlier via reduction of the quasi-Hamiltonian double of $\mathrm{SU}(n)$. These systems live on compact connected symplectic manifolds of dimension $2(n-1)$ and can be interpreted as compactified trigonometric Ruijsenaars--Schneider systems. Depending on the value of a parameter $0<y< \pi$, the

Sanzo Miyazawa

The inverse Potts problem for estimating evolutionary single-site fields and pairwise couplings in homologous protein sequences from their single-site and pairwise amino acid frequencies observed in their multiple sequence alignment would be still one of useful methods in the studies of protein structure and evolution. Since the reproducibility of fields and

Yichen Cai, Yuelong Qiu, Jianhua Zhang, Li Yu

As 6G advances, ubiquitous connectivity and higher capacity requirements of the air interface pose substantial challenges for accurate and real-time wireless channel acquisition in diverse environments. Conventional statistical channel modeling relies on offline measurement data from limited environments, struggling to support online applications facing dive

Shaoliang Yang, Jun Wang, Yunsheng Wang

The matrix-free gather-batched-GEMM-scatter pattern eliminates global stiffness assembly for three-dimensional SIMP topology optimization, but the conventional three-stage implementation forces avoidable DRAM traffic between stages. We present a single fused CUDA kernel, implemented through CuPy's runtime compilation interface, that performs gather, per-elem

Hang Cheng, Muyan He, Mingyu Fan, Chengfeng Xie

Sketch-based 3D shape retrieval (SBSR) aims to retrieve 3D shapes that are consistent with the category of the input hand-drawn sketch. The core challenge of this task lies in two aspects: existing methods typically employ simplified aggregation strategies for independently encoded 3D multi-view features, which ignore the geometric relationships between view

Nina Kastendiek, Jakob Niehues, Frank Hellmann

We study the properties and stability of networks with arbitrary Laplacian coupling. Classic approaches to studying networked systems require unrealistic assumptions, including homogeneous node dynamics, one-dimensional and undirected edges, or constant edge weights. We develop a unified formulation of Laplacian-style couplings that drops these assumptions,

Yinhao Xu, Georg A. Gottwald, Zdenka Kuncic

Self-organizing memristive networks are physical circuits that dynamically reconfigure their circuitry in response to external input signals. Their adaptive behavior arises from intrinsic neuro-synaptic dynamics combined with a heterogeneous network topology. In this work, we demonstrate that such networks naturally generate neuronal population spiking dynam

Siv Marie Cartland Hansen, Rowan Hoogervorst, Otto Anker Nielsen, Richard Martin Lusby

Line planning in public transport is the strategic problem of selecting lines and their operating frequencies. This problem is important as it defines the passenger service, based on available connections and expected travel times, and drives operational cost in terms of the number of vehicles required. This paper presents a line planning model that minimize

Helmut Harbrecht, Christoph Schwab

We prove error bounds for operator surrogates of solution operators for partial differential and boundary integral equations on families of domains which are diffeomorphic to one common reference (or latent) domain $D_{ref}$. The pullback of the PDE to $D_{ref}$ via affine-parametric shape encoding produces a collection of holomorphic parametric PDEs on $D_{

Coleridge Faraday, Ben Bert, Jack Brand, Werner Vogelsang

We present perturbative quantum chromodynamics (pQCD) predictions for high-momentum particle yield modification in very light ion collisions - ${}^{10}\mathrm{B}+{}^{10}\mathrm{B}$, ${}^{6}\mathrm{Li}+{}^{6}\mathrm{Li}$, ${}^{4}\mathrm{He}+{}^{4}\mathrm{He}$, and ${}^{3}\mathrm{He}+{}^{3}\mathrm{He}$ - with and without medium-induced energy loss. We find non

Ryo Noguchi, Jongkeun Jung, Younsik Kim, Sungsoo Hahn

Magnetic materials hosting topological band structures have attracted intense interest due to the interplay between magnetism and spin-orbit coupling (SOC). Here, using temperature- and polarization-dependent angle-resolved photoemission spectroscopy, we investigate the surface ferromagnet GdAg$_2$/Ag(111), a two-dimensional system with Weyl-nodal-line-like

Alexandre Huchet, Tom Laclavère, Leonora Kardum

QUBIC, the Q & U Bolometric Interferometer for Cosmology, is a telescope that observes the polarisation of the sky in the millimetre-wavelength range. Its goal is to detect the primordial B-modes of polarisation in the cosmic microwave background by combining the sensitivity of bolometers with the good understanding of interferometry systematics. This dual a

Nuo Chen, Yicheng Tong, Yuzhe Yang, Yufei He

Multi-agent systems (MAS) are increasingly used for open-ended idea generation, driven by the expectation that collective interaction will broaden the exploration diversity. However, when and why such collaboration truly expands the solution space remains unclear. We present a systematic empirical study of diversity in MAS-based ideation across three bottom-

Mariano Cadoni, Lorenzo Herres, Leonardo Modesto, Lorenzo Orlando

We investigate the leading gravitational eikonal in nonlocal $D$ dimensional theories of gravity. We analyze the simplest cases of $2\rightarrow2$ massless and massive scalar scattering at tree level, studying the effects of nonlocal form factors in the gravitational sector. We give an interpretation of our results in terms of geodesic motion in effective ge

Shaowei Zhang, Faqiang Qian, Yan Chen, Ziliang Wang

Emotion Recognition in Conversation (ERC) has become a fundamental capability for large language models (LLMs) in human-centric interaction. Beyond accurate recognition, coherent emotional expression is also crucial, yet both are limited by the scarcity and static nature of high-quality annotated data. In this work, we propose SELF-EMO, a self-evolution fram

Michael Y. Li, Jubayer Ibn Hamid, Emily B. Fox, Noah D. Goodman

Chain-of-thought reasoning has driven striking advances in language model capability, yet every reasoning step grows the KV cache, creating a bottleneck to scaling this paradigm further. Current approaches manage these constraints on the model's behalf using hand-designed criteria. A more scalable approach would let end-to-end learning subsume this design ch

Julio Silva-Rodríguez, Ender Konukoglu

Super-resolution (SR) models are attracting growing interest for enhancing minimally invasive surgery and diagnostic videos under hardware constraints. However, valid concerns remain regarding the introduction of hallucinated structures and amplified noise, limiting their reliability in safety-critical settings. We propose a direct and practical framework to

Haiweng Xu, Sipeng Zheng, Hao Luo, Wanpeng Zhang

Recent Vision-Language-Action (VLA) models report impressive success rates on standard robotic benchmarks, fueling optimism about general-purpose physical intelligence. However, recent evidence suggests a systematic misalignment between standard benchmark success and true embodied reasoning, raising the question of whether these high scores reflect genuine c

Alexander Sauter, Riccardo Schiavone, Lucía Balsa Picado, Gianluigi Liva

This paper proposes the design of polar and convolutional coset codes for the unequal message protection (UMP) in the short blocklength regime, to overcome the rate loss introduced by preamble-based solutions. After providing conditions to ensure message class disjointness, a two-step decoding architecture is proposed: it first identifies the message class v

Pooyan Khosravinia, João Gama, Bruno Veloso

Anomaly detection in multivariate time series is a central challenge in industrial monitoring, as failures frequently arise from complex temporal dynamics and cross-sensor interactions. While recent deep learning models, including graph neural networks and Transformers, have demonstrated strong empirical performance, most approaches remain primarily correlat

Ralf Hielscher, Rüdiger Kilian, Erik Wünsche, Katharina Tinka Marquardt

Grain boundary plane distributions are widely used to infer the mechanisms governing grain boundary formation in polycrystalline materials. We show that such interpretations are inherently ambiguous. Using a unified eight-parameter boundary distribution framework, we derive both the grain boundary character distribution (GBCD) and the grain boundary normal d

Miltiadis Karakikes, Panagiotis Kostas

We investigate the homological behaviour of compactly generated triangulated categories under separable extensions. We show that homological invariants (finiteness of global dimension, gorensteinness and regularity) are preserved under such extensions. We also establish a relation between singularity categories in this setting, proving that the singularity c

Gautam Kumar, Amit Shivam, Ashwini Ratnoo

This paper presents a decentralized, collision-free framework for path following guidance of multiple uncrewed aerial vehicles (UAVs), while maintaining uniform spacing along a reference path. A vector field-based guidance law is employed to drive each UAV toward the reference path. A rotational repulsion mechanism, utilizing relative distance and bearing be

Yuyao Huang, Wei Nong, Shuya Yamazaki, Martin Hoffmann Petersen

Novelty in materials discovery requires candidates to be distinct, non-redundant, and thermodynamically plausible. While crystallographic databases continue to expand in both size and complexity, making efficient and reliable novelty assessment has become increasingly difficult. This becomes particularly acute when crystallographic disorder is involved, as p

Davide Bigoni, Diego Misseroni, Andrea Piccolroaz

Two equal and opposite distributed dead loads are applied orthogonally to the axis of an elastic rod in its rectilinear reference configuration, one at the extrados and the other at the intrados, such that the resultant applied force per unit length is uniformly zero. In this configuration, the rod is subjected to a transverse (tensile or compressive) stress

Kai-Tao Huang, X. S. Wang

Magnetic solitons are nonlinear, local excitations in magnetic systems. In this study, we theoretically and numerically investigate the properties and generation of one-dimensional (1D) topologically trivial magnetic solitons in ferromagnetic nanowires. An approximate analytical soliton solution described by two free parameters is validated by comparing with

Nimrod Busany

The maintained artifact in an AI-enabled system is not code plus settings, but a versioned governed program space: domains, structural constraints, eligibility, evaluation assets, and a statistical release gate. AI-enabled systems operate under changing world conditions: provider models and APIs change, input distributions drift, evaluation sets age, and obj

Ruben Hefele, Timo Oksanen

Tillage operations account for a large share of on-farm diesel consumption, yet the fuel efficiency of the combined tractor-implement system is not optimised in current practice. Modern continuously variable transmission (CVT) tractors minimise engine fuel consumption internally, but they treat the implement as an unknown load and do not account for the effe

Sylvain Brisson, Brice Lecampion

We investigate two hydraulic stimulation stages performed in April 2022 at the Utah FORGE enhanced geothermal system test site using analytical and numerical models for tensile hydraulic fractures and fluid-induced dilatant shear fractures. The two injection stages differ primarily by the viscosity of the fracturing fluid. Despite similar injection rate sche

Jiaqi Li, Lvyang Zhang, Yang Zhao, Wen Lu

What does it mean to give an AI agent a complete education? Current agent development produces specialists systems optimized for a single capability dimension, whether tool use, code generation, or security awareness that exhibit predictable deficits wherever they were not trained. We argue this pattern reflects a structural absence: there is no curriculum t

Xinyao Zhang, Nicole Sonne Heckmann, Manuela Del Castillo Suero, Francesco Paolo Speca

Background: The potential of large language models (LLMs) to automate and support pharmacoepidemiologic study design is an emerging area of interest, yet their reliability remains insufficiently characterized. General-purpose LLMs often display inaccuracies, while the comparative performance of specialized biomedical LLMs in this domain remains unknown. Meth

Shuyue Guan, Weian Guo, Pengyu Zheng, Xinxuan Lin

The anomalous Nernst effect (ANE), generating a voltage perpendicular to a temperature gradient due to magnetization, is closely linked to the Berry curvature (BC) near the Fermi energy in topological magnets. We report an enhanced spontaneous ANE in the ferromagnetic Kondo lattice CeCo2As2, which features Kondo-screened cerium-based 4f moments embedded in a

Mason Wang, Cheng-Zhi Anna Huang

We introduce the Latent Fourier Transform (LatentFT), a framework that provides novel frequency-domain controls for generative music models. LatentFT combines a diffusion autoencoder with a latent-space Fourier transform to separate musical patterns by timescale. By masking latents in the frequency domain during training, our method yields representations th

Tereza Zemánková, Martin Šarbort, Petr Jákl, Jan Ježek

Non-Hermitian dynamics in open systems can give rise to a variety of fascinating non-equilibrium phenomena, ranging from symmetry-breaking transitions to directional energy flow. Parity-time (PT) symmetry breaking determines the occurrence of dynamical instabilities, while non-reciprocal interactions enable asymmetric energy transfer between modes. Here, we

Junyoung Yang, Kyungmin Kim, Sangdon Park

Uncertainty quantification is crucial in safety-critical systems, where decisions must be made under uncertainty. In particular, we consider the problem of online uncertainty quantification, where data points arrive sequentially. Online conformal prediction is a principled online uncertainty quantification method that dynamically constructs a prediction set

Omrit Filtser, Tzalik Maimon, Ofir Yomtovyan

The minimum convex cover problem seeks to cover a polygon $P$ with the fewest convex polygons that lie within $P$. This problem is $\exists\mathbb R$-complete, and the best previously known algorithm, due to Eidenbenz and Widmayer (2001), achieves an $O(\log n)$-approximation in $O(n^{29} \log n)$ time, where $n$ is the complexity of $P$. In this work we pre

Yu Zhang, Chuyang Sun, Kehai Chen, Xuefeng Bai

Large Vision-Language Models (LVLMs) still struggle with vision hallucination, where generated responses are inconsistent with the visual input. Existing methods either rely on large-scale annotated data for fine-tuning, which incurs massive computational overhead, or employ static post-hoc strategies that overlook the dynamic nature of hallucination emergen

Andrew Fowlie

Prompted by misconceptions in the recent literature, we review the justifications for naturalness arguments and Occam's razor found in Bayesian statistics. We discuss the automatic Occam's razor that emerges in Bayesian formalism, bringing together points of view from diverse fields, including statistics, social sciences, physics and machine learning. In ped

Christopher C. Lovell, Max E. Lee, William J. Roper, Daniel Anglés-Alcázar

We present a new method for emulating the halo mass function (HMF) and other distribution functions in large effective volumes, down to low halo masses, whilst simultaneously modifying large ranges of parameters, for a fraction of the cost of traditional periodic cosmological simulations. We demonstrate the method by selecting small regions, $V \sim (50 \,h^

Aziz M. Embarek, Dmitry V. Shatilovich

We study nonlinear stationary Kolmogorov equations with degenerate diffusion matrices and discontinuous coefficients. The existence of a solution is proved. We propose a new approach based on an integral condition with Lyapunov functions and a regularity of projections of solutions in the partially degenerate case. Examples are given to illustrate the result

Jianan Liu, Jing Yang, Xianyou Li, Weiran Yan

The rapid adoption of artificial intelligence (AI) and large language models (LLMs) is transforming financial analytics by enabling natural language interfaces for reporting, decision support, and automated reasoning. However, limited empirical understanding exists regarding how different LLM-based reasoning architectures perform across realistic financial w

Roberto Capecelatro, Marco Marciani, Claudio Guarcello, Gabriele Campagnano

We study the transport properties of non-Hermitian magnetic Josephson junctions, considering a superconductor-quantum dot-superconductor device coupled to a ferromagnetic metallic reservoir in the presence of an external magnetic field. We focus on the $0-\pi$ transitions that occur when the equilibrium phase difference between the superconductors shifts fro

Xingyu Liu, Zengqin Huang, Xiang Gao, Hailong Sun

Fuzz testing of software libraries relies on fuzz drivers to invoke library APIs. Traditionally, these drivers are written manually by developers - a process that is time-consuming and often inadequate for exercising complex program behaviors. While recent studies have explored the use of Large Language Models (LLMs) to automate fuzz driver generation, the r

Alistair Plum, Felicia Körner, Anne-Marie Lutgen, Laura Bernardy

This paper presents ltzGLUE, the first Natural Language Understanding (NLU) benchmark for Luxembourgish (LTZ) based on the popular GLUE benchmark for English. Although NLU tasks are available for many European languages nowadays, LTZ is one of the official national languages that is often overlooked. We construct new tasks and reuse existing ones to introduc

Nicolas Strangmann

We present the first results on the $\pi^0$ nuclear modification factor $R_{OO}$ in OO collisions at LHC energies by the ALICE experiment. The measurement of the modification of hadron production in nuclear collisions compared to a vacuum baseline in pp collisions is a valuable probe for parton energy loss in the hot medium. The ALICE $R_{OO}$ results show s

Jie Zhu, Huaixia Dou, Junhui Li, Lifan Guo

Emotional Support Conversation (ESC) aims to assist individuals experiencing distress by generating empathetic and supportive dialogue. While prior work typically assumes that each supporter turn corresponds to a single strategy, real-world supportive communication often involves multiple strategies within a single utterance. In this paper, we revisit the ES

Ana Baltaretu, Pascal Benschop, Jan van Gemert

Human Action Recognition (HAR) models are increasingly deployed in high-stakes environments, yet their fairness across different human appearances has not been analyzed. We introduce a framework for auditing bias in HAR models using synthetic video data, generated with full control over visual identity attributes such as skin color. Unlike prior work that fo

Atif Ansar, Bent Flyvbjerg, Alexander Budzier

Do projects learn across space and time? The Olympics, among the largest publicly funded programmes in the world, offer a unique empirical setting. Theoretically, the Games seem ideal for generating "positive learning curves," driving down costs from one iteration to the next. In practice, they do not. Drawing on the concept of "myopia of learning," we argue

Hasan Amin, Harry Yizhou Tian, Xiaoni Duan, Chien-Ju Ho

Although large language models (LLMs) are increasingly used as annotators at scale, they are typically treated as a pragmatic fallback rather than a faithful estimator of human perspectives. This work challenges that presumption. By framing perspective-taking as the estimation of a latent group-level judgment, we characterize the conditions under which moder

Ismaïl Baaj, Henri Prade

In this article, we establish a precise connection between binarized neural networks (BNNs) and Sugeno integrals. The advantage of the Sugeno integral is that it provides a framework for representing the importance of inputs and their interactions, while being equivalent to a set of if-then rules. For a hidden BNN neuron at inference time, we show that the a

Jinglai Zheng, Chuhan Qiao, Haiming Huang

Deploying LLMs as reasoning assistants in safety-critical aerospace engineering requires stricter evaluation criteria than general scientific benchmarks. In hypersonic thermal protection system (TPS) design, inaccurate stagnation-point heat flux or boundary-layer calculations may cause catastrophic design margin violations. Models with numerically reasonable

Wenjie Mu, Zhan Li, Chuanzhou Su, Xuanyi Shen

Generalizable Neural Radiance Fields (GeNeRFs) enable high-quality scene reconstruction from sparse views and can generalize to unseen scenes. However, in real-world settings, transient distractors break cross-view structural consistency, corrupting supervision and degrading reconstruction quality. Existing distractor-free NeRF methods rely on per-scene opti

Francesc Molina, Albert Guillen i Fabregas

This manuscript investigates channel capacity under mismatched stochastic likelihood decoding. We derive Feinstein- and Verd\'u-Han-style bounds on the error probability coded communication. These are used to obtain a general information-spectrum formula for the channel capacity under mismatched stochastic decoding. The mismatch capacity formula is expressed

Sota Asai

For a finite dimensional algebra $A$, the TF equivalence on the real Grothendieck group $K_0(\operatorname{\mathsf{proj}} A)_\mathbb{R}$ can be regarded as a completion of the $g$-fan. For example, the silting cones $C^\circ(U)$ of 2-term presilting complexes $U$ give the most fundamental family of TF equivalence classes. The next step is studying the TF equ

Victoria Bosch, Rowan Sommers, Adrien Doerig, Tim C Kietzmann

Recent studies reveal striking representational alignment between artificial neural networks (ANNs) and biological brains, leading to proposals that all sufficiently capable systems converge on universal representations of reality. Here, we argue that this claim of Universality is premature. We introduce the Umwelt Representation Hypothesis (URH), proposing

Yuxiang Zhao, Wei Huang, Yujie Song, Liu Wang

Expressive Human Pose and Shape Estimation (EHPS) plays a crucial role in various AR/VR applications and has witnessed significant progress in recent years. However, current state-of-the-art methods still struggle with accurate parameter estimation for facial and hand regions and exhibit limited generalization to wild images. To address these challenges, we

Huakang Chen, Jingbin Hu, Liumeng Xue, Qirui Zhan

Instruction-following text-to-speech (TTS) has emerged as an important capability for controllable and expressive speech generation, yet its evaluation remains underdeveloped due to limited benchmark coverage, weak diagnostic granularity, and insufficient multilingual support. We present \textbf{MINT-Bench}, a comprehensive multilingual benchmark for instruc

Raffaele Pisano, Roberto Navigli

Process Reward Models (PRMs) have emerged as a powerful tool for providing step-level feedback when evaluating the reasoning of Large Language Models (LLMs), which frequently produce chains of thought (CoTs) containing errors even when the final answer is correct. However, existing PRM datasets remain expensive to construct, prone to annotation errors, and p

Ke Wan, Kensuke Tanioka, Toshio Shimokawa

Machine learning has become integral to medical research and is increasingly applied in clinical settings to support diagnosis and decision-making; however, its effectiveness depends on access to large, diverse datasets, which are limited within single institutions. Although integrating data across institutions can address this limitation, privacy regulation

Kotaro Kobayashi, Naomichi Yutani, Takayuki R. Saitoh, Junichi Baba

Gas delivery to galactic centers powers nuclear starbursts and active galactic nuclei (AGNs), yet bar-driven inflow is generally expected to stall in a nuclear ring a few hundred parsecs across. Using three-dimensional Lagrangian hydrodynamic simulations in a fixed barred potential, we identify a bypass channel in which a fraction of the inflowing gas acquir

Maximilian Kasy, Elizabeth Linos, Sanaz Mobasseri

This paper develops a framework for identification, estimation, and inference on the causal mechanisms driving endogenous social network formation. Identification is challenging because of unobserved confounders and reverse causality; inference is complicated by questions of equilibrium and sampling. We leverage repeated observations of a network over time a

Emilio Xosé Rodríguez Fernández

The LHCb detector has demonstrated a proven competitiveness across a wide range of physics analyses thanks to its forward coverage. These proceedings describe: i) complementary measurements using heavy flavour jets, ii) Electroweak (EW) measurements with the top and W boson, and iii) searches for New Physics states such as axion-like particles (ALPs), heavy-

Chuhan Qiao

We revisit multi-agent delegation under a stronger and more realistic assumption: an agent's capability is not fixed at the skill level, but depends on task context. A coding agent may excel at short standalone edits yet fail on long-horizon debugging; a planner may perform well on shallow tasks yet degrade on chained dependencies. Static skill-level capabil

Qiuhui Chen, Jiaxiang Song, Shuai Tan, Weimin Zhong

Deep learning-based industrial anomaly detectors often behave as black boxes, making it hard to justify decisions with physically meaningful defect evidence. We propose ZSG-IAD, a multimodal vision-language framework for zero-shot grounded industrial anomaly detection. Given RGB images, sensor images, and 3D point clouds, ZSG-IAD generates structured anomaly

Thomas Führer, Paula Hilbert, Ani Miraçi, Dirk Praetorius

We analyze optimal complexity of adaptive finite element methods (AFEMs) for general second-order linear elliptic partial differential equations (PDEs) in the Lax-Milgram setting. To this end, we formulate an adaptive algorithm which steers the local mesh-refinement as well as the termination of a generalized minimal residual solver (GMRES) with optimal prec

M. Levent Kurnaz

International commerce has long been seen as a key way to keep the global food system stable, allowing agricultural surpluses in some areas to compensate for shortages in others. This strategy has led to the rise of highly specialised processing hubs that combine significant industrial capacity with agricultural inputs sourced from throughout the world. T\"u

Maximilian von Aspern, Felix Buld, Michael Pinedo

We study flow shop scheduling with stochastic reentry, where jobs must complete multiple passes through the entire shop, and the number of passes that a job requires for completion is drawn from a discrete probability distribution. The goal is to find policies that minimize performance measures in expectation. Our main contribution is a reduction to a stocha

Yindong Zhang, Wenmian Yang, Yiquan Zhang, Weijia Jia

Developing agents capable of navigating fragmented, multi-source information remains challenging, primarily due to the scarcity of benchmarks reflecting hybrid workflows combining database querying with external APIs. To bridge this gap, we introduce ReCoQA, a large-scale benchmark of 29,270 real-estate instances featuring machine-verifiable supervision for

Paul Brunet

The recently introduced model of representations has been defined and motivated somewhat ex-nihilo. In this document, I will show that representations are related to a more ''classical'' model through a 2-adjunction. The target model is that of preorder morphisms, i.e. maps between sets equipped with reflexive and transitive relation that satisfy some natura

Qidong Wang, Junjie Hu, Ming Jiang

Recent work has increasingly explored neuron-level interpretation in vision-language models (VLMs) to identify neurons critical to final predictions. However, existing neuron analyses generally focus on single tasks, limiting the comparability of neuron importance across tasks. Moreover, ranking strategies tend to score neurons in isolation, overlooking how

Peerachai Banyongrakkul, Mansooreh Zahedi, Christoph Treude, Haoyu Gao

Modern software systems have transitioned from purely code-based architectures to AI-integrated systems where pre-trained models (PTMs) serve as permanent dependencies. However, while the evolution of traditional software libraries is well-documented, we lack a clear understanding of how these "PTM dependencies" change over time. Unlike libraries, PTMs are c

Rishav Rishav, Pushpak Pujari, Pushpendre Rastogi

Prompt optimization methods either analyze individual failures in isolation or compare prompt variants across examples, operating on single execution traces with no access to the reasoning process distinguishing success from failure on the same input. We introduce ContraPrompt, built on the observation that when a model fails but succeeds on a retry with fee

Thomas Tulinski, Jorge Fernandez-De-Cossio-Diaz, Simona Cocco, Rémi Monasson

Training in machine learning generally consists in finding one model, whose parameters minimize a data-dependent loss. Yet, empirical work shows that ensemble learning, an approach in which multiple models are sampled, can improve performance. Here, we provide an analytical framework to understand these observations in the case of Boltzmann machines, exploit

Xiao Wang

The key-value (KV) cache is the dominant memory bottleneck during Transformer inference, yet little is known theoretically about how aggressively it can be compressed before multi-step reasoning degrades. We study this through $k$-hop pointer chasing on $n$ tokens under a shared KV cache of size $s$, attention dimension $m$, $H$ heads, $p$-bit precision, and

Takumi Namba

In this paper, we study robust distributed sub-optimal coordination of linear agents subject to input nonlinearities. Inspired by the robust agreement literature, we formulate a bounded distributed sub-optimal coordination problem, in which each agent converges to a neighborhood of the optimizer of a global optimization problem defined over a communication n

Nikita S Shcheblanov, Anaël Lemaître

Our understanding of vibrations in solids currently rests on concepts and techniques designed for crystals and explicitly relying on periodicity, hence inapplicable to amorphous materials. As a consequence, no established framework enables a systematic decomposition of the vibrational spectrum of amorphous solids into contributions associated with well-defin

Marios Kokmotos, Dimitri M. Gangardt, Giovanni Barontini

We show numerically that a repulsive Bose-Einstein condensate can be driven into implosive dynamics by a direct topological quench. We first realize giant vortices by quasi-adiabatic phase imprinting, and then perform a sudden anti-imprint that cancels the accumulated winding in a single step, abruptly switching the condensate from a highly charged vortex st

H S V N S Kowndinya Renduchintala, Sumit Bhatia

Large Language Models (LLMs) exhibit a puzzling disparity in their formal linguistic competence: while they learn some linguistic phenomena with near-perfect mastery, they often perform below chance on others, even after training on trillions of tokens. In this work, we investigate whether these failures stem from inherent architectural limitations or simply

Ömer Lütfü Karakelle, Sefa Kayraklık, İbrahim Hökelek, Ali Görçin

Determining the optimal phase configurations of reconfigurable intelligent surface (RIS) elements typically requires complex channel estimation procedures with high pilot overhead, creating a bottleneck for real-time deployment in time-varying wireless environments. In this paper, we propose a digital twin (DT)-driven framework for RIS phase shift optimizati

Zhanyu Liu, Qingguo Hu, Ante Wang, Chenqing Liu

Reinforcement Learning with Verifiable Reward (RLVR) has proven effective for training reasoning-oriented large language models, but existing methods largely assume high-resource settings with abundant training data. In low-resource scenarios, RLVR is prone to more severe entropy collapse, which substantially limits exploration and degrades reasoning perform

Feixue Shao, Guangze Shi, Xueyu Liu, Yongfei Wu

Visual decoding of neurophysiological signals is a critical challenge for brain-computer interfaces (BCIs) and computational neuroscience. However, current approaches are often constrained by the systematic and stochastic gaps between neural and visual modalities, largely neglecting the intrinsic computational mechanisms of the Human Visual System (HVS). To

Tobias Dieselhorst, Tanja Stadler

The parameters of many classes of birth-death processes cannot be inferred uniquely from phylogenetic trees: infinitely many parameter combinations yield the same distribution of phylogenetic trees. Here, we show that parameter identifiability can be recovered even for the most general cases of time-dependent rates when additional information on hidden birth

Ernst Dennis Lægteskov Binau Larsson, Erik Kjellgren, Peter Reinholdt, Jacob Kongsted

Accurate modeling of surface catalytic processes often requires methods capable of describing strong correlation, charge transfer, and multiple closely lying electronic states. While density functional theory remains widely used, its limitations for localized electronic states motivate the use of wavefunction-based approaches and, more recently, quantum comp

Dazhong Wang, Ruqu Wang, Xinyi Xu

This paper studies optimal auction design when valuations depend endogenously on post-auction collaboration between the seller and the winning bidder. Both parties exert non-contractible efforts after the auction, generating a double moral hazard problem alongside adverse selection. We analyze two role structures -- winner-pivotal and seller-pivotal collabor

Soumyodeep Mukhopadhyay, Didier Rullière, Rodolphe Le Riche, David Gaudrie

Approximation of functions satisfying partial differential equations (PDEs) is paramount for simulation of physical fluid flows and other problems in physics. Recently, physics-informed machine learning approaches have proven useful as a data-driven complement to numerical models for partial differential equations, bringing faster responses and allowing us t

Alcides Buss, Julian Kranz

We construct the first explicit examples of locally compact Hausdorff \'etale groupoids that are not inner amenable and that do not arise as transformation groupoids associated to partial actions of discrete groups. This answers questions of Anantharaman--Delaroche and Exel. Our examples include all Higson--Lafforgue--Skandalis groupoids associated to non-am

Islam Mansour, Francescopaolo Sica, Michael Schmitt

Synthetic Aperture Radar (SAR) plays a critical role in maritime surveillance, yet deep learning for SAR analysis is limited by the lack of pixel-level annotations. This paper explores how general-purpose vision foundation models can enable zero-shot ship instance segmentation in SAR imagery, eliminating the need for pixel-level supervision. A YOLOv11-based

Xiaoyuan Cheng, Haoyu Wang, Wenxuan Yuan, Ziyan Wang

Recent advances in flow-based offline reinforcement learning (RL) have achieved strong performance by parameterizing policies via flow matching. However, they still face critical trade-offs among expressiveness, optimality, and efficiency. In particular, existing flow policies interpret the $L_2$ regularization as an upper bound of the 2-Wasserstein distance