November 2025 arXiv papers — page 10

Pirzada Suhail, Rehna Afroz, Amit Sethi

Deep neural networks often struggle to recognize when an input lies outside their training experience, leading to unreliable and overconfident predictions. Building dependable machine learning systems therefore requires methods that can both estimate predictive \textit{uncertainty} and detect \textit{out-of-distribution (OOD)} samples in a unified manner. In

Valerio Buttinelli, Angelo Felice Lopez, Roberto Vacca

We study connectedness of degeneracy loci $D_{r-k}(\varphi)$ of morphisms $\varphi : {\mathcal O}_X^{\oplus (r+1-k)} \to \mathcal E$, where $\mathcal E$ is a rank $r$ globally generated bundle on a smooth $n$-dimensional variety $X$ and $k \le 3$. For $k \le 2$ we give a characterization of connectedness in terms of vanishing of Chern classes. Moreover we pr

Tamara Bottazzi, Cristian Conde

In this paper, we investigate the Schatten $p$-class ideals for $p >1$ as semi-inner product spaces in the sense of Giles and Lumer. Within this framework, we explore several geometric and analytic notions such as Birkhoff-James orthogonality, $p$-parallelism, and related properties that naturally arise when these structures are interpreted through the lens

Zirui Wang, Tao Zhang

3D understanding is a key capability for real-world AI assistance. High-quality data plays an important role in driving the development of the 3D understanding community. Current 3D scene understanding datasets often provide geometric and instance-level information, yet they lack the rich semantic annotations necessary for nuanced visual-language tasks.In th

Jakub Czuchnowski, Chuan Li, Hongli Ni, Brandon Weissbourd

Deconvolution is the most widely used aberration correction technique in microscopy, however most techniques assume that the aberrations are the same for each point in the image, which is rarely true. Methods for tracking spatially varying aberrations require burdensome calibration or computation, or require symmetries in the aberration patterns. Here, we ex

Jananan Arulseelan

Expanding on previous work of the author, we initiate the model theoretic study of W$^*$-dynamical systems. We axiomatize continuous weight-preserving group actions of $G$ on von Neumann algebras for $G$ a given locally compact Hausdorff group. Since our axiomatization is of continuous actions, the ultraproduct is defined so that the ultraproduct action of $

Gabriel Capelo, Eric C. Andrade

Inspired by recent observations of the $1/3$ magnetization plateau in kagome-based magnets, we investigate the $J_1-J_2$ Heisenberg model on the kagome lattice under the influence of an external magnetic field. Although the classical ground state at zero field depends on the sign of $J_2$, we find a robust $1/3$ semiclassical magnetization plateau in both ca

Wei Fan, Kevin Tan, Yuting Wei

Adaptively collected data has become ubiquitous within modern practice. However, even seemingly benign adaptive sampling schemes can introduce severe biases, rendering traditional statistical inference tools inapplicable. This can be mitigated by a property called stability, which states that if the rate at which an algorithm takes actions converges to a det

Stefano Scanzio, Pietro Chiavassa, Gianluca Cena

Executable QR codes, also known as sQRy, are a technology aimed at inserting executable programs in a QR code. Through a concrete example, in this paper, we demonstrate their usage in the context of industrial networks in order to assess the operation of a TSN switch by analyzing its status LEDs even in the absence of an internet connection. The entire gener

Pietari Laitinen, Matti Vihola

Iterated sampling importance resampling (i-SIR) is a Markov chain Monte Carlo (MCMC) algorithm which is based on $N$ independent proposals. As $N$ grows, its samples become nearly independent, but with an increased computational cost. We discuss a method which finds an approximately optimal number of proposals $N$ in terms of the asymptotic efficiency. The o

Akhil Rajeev P

Grammatical error correction for Indic languages faces limited supervision, diverse scripts, and rich morphology. We propose an augmentation-free setup that uses instruction-tuned large language models and conservative decoding. A 12B GEMMA 3 model is instruction-tuned in bnb 4-bit precision with parameter-efficient fine-tuning (PEFT) and Alpaca-style format

Matt MacDermott, Qiyao Wei, Rada Djoneva, Francis Rhys Ward

AI systems that output their reasoning in natural language offer an opportunity for safety -- we can \emph{monitor} their chain of thought (CoT) for undesirable reasoning, such as the pursuit of harmful objectives. However, the extent to which CoT faithfully reflects the underlying reasoning process, and hence the extent to which it can be usefully monitored

Pedro J. Chocano, Luis Felipe Prieto-Martínez

Given a finite poset $P$, its zeta matrix $\mathbf Z$ encode fundamental incidence-theoretic information about the order structure. In this paper we introduce and study the \emph{order-complement matrix} $\overline{\mathbf Z} = \mathbf J - \mathbf Z$, where $\mathbf J$ is the all-ones matrix. We prove a closed formula for its characteristic polynomial and fo

Chandrasekhar Gokavarapu

We develop the geometric and homological framework for non-commutative $n$-ary $\Gamma$-semirings by constructing a sheaf and derived theory over their non-commutative $\Gamma$-spectrum. Starting with a non-commutative $n$-ary $\Gamma$-semiring $(T,+,\Gamma,\mu)$ and its bi-$\Gamma$-modules, we define the space $\Spec_{\Gamma}^{\mathrm{nc}}(T)$, equip it wit

Mohammad Abdollahi, Khandaker Rifah Tasnia, Soumit Kanti Saha, Jinqiu Yang

Understanding a program's runtime reasoning behavior, meaning how intermediate states and control flows lead to final execution results, is essential for reliable code generation, debugging, and automated reasoning. Although large language models (LLMs) can accurately predict program outputs, most prior work has focused on output accuracy and performance, tr

Matej Klemen, Tjaša Arčon, Luka Terčon, Marko Robnik-Šikonja

Empirical grammar research has become increasingly data-driven, but the systematic analysis of annotated corpora still requires substantial methodological and technical effort. We explore how agentic large language models (LLMs) can streamline this process by reasoning over annotated corpora and producing interpretable, data-grounded answers to linguistic qu

Mikhail Chebunin, Günter Last

We consider a random connection model (RCM) $\xi$ driven by a Poisson process $\eta$. We derive exponential moment bounds for an arbitrary cluster, provided that the intensity $t$ of $\eta$ is below a certain critical intensity $t_T$. The associated subcritical regime is characterized by a finite mean cluster size, uniformly in space. Under an exponential de

Padmanabha Bose, Smarajit Karmakar

Soft matter systems are common in nature and make up nearly all the essential components necessary for life, from cells to the organelles within those cells. The ability of these soft materials to deform is crucial for the proper functioning of various biological processes, such as blood flow in our veins and arteries. It is vital to understand how deformabi

Gabriele Formis, Amanda Ericson, Stefan Forsstrom, Kyi Thar

Ensuring reliable and predictable communications is one of the main goals in modern industrial systems that rely on Wi-Fi networks, especially in scenarios where continuity of operation and low latency are required. In these contexts, the ability to predict changes in wireless channel quality can enable adaptive strategies and significantly improve system ro

Timothy Legge, Aaron Wang, Jacob Ortiz, Victor Limouzi

Transformer-based models have achieved state-of-the-art performance in jet tagging at the CERN Large Hadron Collider (LHC), with the Particle Transformer (ParT) representing a leading example of such models. A striking feature of ParT is its sparse, nearly binary, attention structure, raising questions about the origin of this behavior and whether it encodes

Bart Jacobs, Márk Széles, Dario Stein

Inference is a fundamental reasoning technique in probability theory. When applied to a large joint distribution, it involves updating with evidence (conditioning) in one or more components (variables) and computing the outcome in other components. When the joint distribution is represented by a Bayesian network, the network structure may be exploited to pro

Hajra Anwar Beg, Baptiste Chopin, Hao Tang, Mohamed Daoudi

We present ReactionMamba, a novel framework for generating long 3D human reaction motions. Reaction-Mamba integrates a motion VAE for efficient motion encoding with Mamba-based state-space models to decode temporally consistent reactions. This design enables ReactionMamba to generate both short sequences of simple motions and long sequences of complex motion

Owen Dugan, Roberto Garcia, Ronny Junkins, Jerry Liu

The success of large language models (LLMs) can be attributed in part to their ability to efficiently store factual knowledge as key-value mappings within their MLP parameters. Recent work has proposed explicit weight constructions to build such fact-storing MLPs, providing an improved understanding of LLM fact storage mechanisms. In this paper, we introduce

Xia Chen, Ruiji Sun, Philipp Geyer, André Borrmann

Human-factor analysis typically employs correlation analysis and significance testing to identify relationships between variables. However, these descriptive ('what-is') methods, while effective for identifying associations, are often insufficient for answering causal ('what-if') questions. Their application in such contexts often overlooks confounding and c

Wanchen Zhao, Peter Bubenik

As the size of data increase, persistence diagrams often exhibit structured asymptotic behavior, converging weakly to a Radon measure. However, conventional vector summaries such as persistence landscapes are not well-behaved in this setting, particularly for diagrams with high point multiplicities. We introduce continuous persistence landscapes, a new vecto

Dimitris Kompostiotis, Dimitris Vordonis, Konstantinos D. Katsanos, Florin-Catalin Grec

High-precision localization and environmental sensing are essential for a new wave of applications, ranging from industrial automation and autonomous systems to augmented reality and remote healthcare. Conventional wireless methods, however, often face limitations in accuracy, reliability, and coverage, especially in complex non-line-of-sight (NLoS) environm

Bo Zhang

Classical neural networks are known for their ability to approximate mappings between finite-dimensional spaces, but they fall short in capturing complex operator dynamics across infinite-dimensional function spaces. Neural operators, in contrast, have emerged as powerful tools in scientific machine learning for learning such mappings. However, standard neur

Jason Lunder

In creating sentence embeddings for Natural Language Inference (NLI) tasks, using transformer-based models like BERT leads to high accuracy, but require hundreds of millions of parameters. These models take in sentences as a sequence of tokens, and learn to encode the meaning of the sequence into embeddings such that those embeddings can be used reliably for

M. Adeel Ajaib

We show that nilpotent matrices that yield the Schrodinger equation from its first order form encode the fingerprints of grand unified theories. We perform a rigorous search for all such nilpotent matrices and find that the resulting matrices naturally organize into suggestive group theoretic structures without any other a priori assumptions. The antisymmetr

Jonathan Pipping-Gamón, Tianshu Feng, R. Paul Sabin

Expected goals (xG) models estimate the probability that a shot results in a goal from its context (e.g., location, pressure), but they operate only on observed shots. We propose xG+, a possession-level framework that first estimates the probability that a shot occurs within the next second and its corresponding xG if it were to occur. We also introduce ways

Alexander Smith

Among the nondegenerate C^4 hypersurfaces M in R^n, we characterize the rational quadrics as the hypersurfaces that are the least well approximated by rational points. Given M other than a rational quadric, we prove a heuristically sharp lower bound for the number of rational points very near M, improving the sensitivity of prior results of Beresnevich and H

Alexandre Roy

Let $X$ be an algebraic variety over $\mathbb{C}$ and $G$ be an algebraic group acting on $X$ whose action is closed. J. Poineau defined a compactification $X^\urcorner$ of $X(\mathbb{C})$ by using hybrid Berkovich spaces. We will focus on the extension of the action of $G$ on this compactification by characterising the set $\mathcal{U} \subset X^\urcorner$

Mohammed El Abdioui

This study presents an advanced framework for tropopause detection and analysis using ERA5 reanalysis data, with particular application to extreme meteorological events affecting Morocco and Southern Europe. The research implements and compares multiple detection methodologies, including classical approaches (thermal/WMO and dynamical/1.5 PVU criteria) along

K. G. Fripp, A. V. Shytov, V. V. Kruglyak

We use micromagnetic simulations to demonstrate neuron functionality of two-dimensional (2D) chiral magnonic resonators. Our design exploits nonlinear resonant scattering of spin waves propagating in a YIG medium from an edge mode of a permalloy nano-element. The reduced frequency and volume of the edge mode facilitate matching it to the YIG modes and give r

Shantanu Ghosh, Vedant Parthesh Joshi, Rayan Syed, Param Budhraja

Breast cancer is one of the leading causes of death among women worldwide. We introduce Mammo-FM, the first foundation model specifically for mammography, pretrained on the largest and most diverse dataset to date - 140,677 patients (821,326 mammograms) across four U.S. institutions. Mammo-FM provides a unified foundation for core clinical tasks in breast im

Balthazar Fléchelles

We give a complete characterization of the holonomies of strictly convex cusps and of round cusps in convex projective geometry. We build families of generalized cusps of non-maximal rank associated to each strictly convex or round cusp. We also extend Ballas-Cooper-Leitner's definition of generalized cusp to allow for virtually solvable fundamental group, a

Julian Brandon, Angus Chadwick, Arthur Pellegrino

Many tasks require mapping continuous input data (e.g. images) to discrete task outputs (e.g. class labels). Yet, how neural networks learn to perform such discrete computations on continuous data manifolds remains poorly understood. Here, we show that signatures of such computations emerge in the representational geometry of neural networks as they learn. B

Pedro Duarte, Anton Gorodetski, Victor Kleptsyn

We provide an explicit formula for an increment of the fibered rotation number of a one-parameter family of circle cocycles over any ergodic transformation in terms of invariant measures. As an application, for a family of random dynamical systems on the circle, this gives a formula for an increment of the rotation number in terms of the stationary measures.

Zag ElSayed, Grace Westerkamp, Gavin Gammoh, Yanchen Liu

We introduce EEG Autoclean Vision Language AI (ICVision) a first-of-its-kind system that emulates expert-level EEG ICA component classification through AI-agent vision and natural language reasoning. Unlike conventional classifiers such as ICLabel, which rely on handcrafted features, ICVision directly interprets ICA dashboard visualizations topography, time

Anson Ho, Jean-Stanislas Denain, David Atanasov, Samuel Albanie

Most AI benchmarks saturate within years or even months after they are introduced, making it hard to study long-run trends in AI capabilities. To address this challenge, we build a statistical framework that stitches benchmarks together, putting model capabilities and benchmark difficulties on a single numerical scale. This acts as a "Rosetta Stone", allowin

João P. S. Maurício de Carvalho

Understanding how societies react to epidemic threats requires more than tracking infection curves. Public perception, collective memory and behavioural adaptation interact through feedback loops that can amplify or suppress the spread of fear, vigilance and precaution. In this work we reinterpret the classical Lorenz system in a socioepidemic context, gover

Jose Luis Lima de Jesus Silva, Joao Pedro Gomes, Paulo Roberto de Melo Barros Junior, Vitor Hugo Serravalle Reis Rodrigues

Interpreting seismic horizons is a critical task for characterizing subsurface structures in hydrocarbon exploration. Recent advances in deep learning, particularly U-Net-based architectures, have significantly improved automated horizon tracking. However, challenges remain in accurately segmenting complex geological features and interpolating horizons from

Daniel A. Jaume, Victor N. Schvöllner, Cristian Panelo, Kevin Pereyra

We study the nullspace of the adjacency matrix of split graphs, whose vertex set can be partitioned into a clique and an independent set. We introduce the clique-kernel, a subspace that decides whether clique vertices lie in the support of a kernel eigenvector, and we prove that its dimension is at most one. This yields the formula $null(Sp) = null(R) + \dim

Evgeny M. Gromov, Boris A. Malomed

We introduce an extended nonlinear Lugiato-Lefever equation (LLE) with the pseudo-stimulated-Raman-scattering (pseudo-SRS) cubic term, linear damping/gain, and spatial inhomogeneous (weakly or strongly localized) pump. The LLE is derived, in the extended adiabatic approximation, from the underlying Zakharov system (ZS), which includes a viscosity term, actin

Frank Gaede, Gregor Kasieczka, Lorenzo Valente

Accurate particle shower simulation remains a critical computational bottleneck for high-energy physics. Traditional Monte Carlo methods, such as Geant4, are computationally prohibitive, while existing machine learning surrogates are tied to specific detector geometries and require complete retraining for each design change or alternative detector. We presen

Seyed Hadi Mirfarshbafan, Nicolas Filliol, Oscar Castañeda, Christoph Studer

Fixed-point number representation is commonly employed in digital VLSI designs that have stringent hardware efficiency constraints. However, fixed-point numbers cover a relatively small dynamic range for a given bitwidth. In contrast, floating-point numbers offer a larger dynamic range at the cost of increased hardware complexity. In this paper, we propose a

Shashanka B R, Mohith Charan R, Seema Banu F

Chunking has emerged as a critical technique that enhances generative models by grounding their responses in efficiently segmented knowledge [1]. While initially developed for unimodal (primarily textual) domains, recent advances in multimodal foundation models have extended chunking approaches to incorporate diverse data types, including images, audio, and

Sergey G. Bobkov, Friedrich Götze

We discuss variants of construction of measurable subgradients for multivariate convex functions and the problem of characterization of the $\Delta_2$-condition in terms of their directional derivatives. Furthermore we study related basic properties of Luxemburg and Orlicz pseudo-norms for vector-valued functions.

Niki Z. Petrakos, Erica E. M. Moodie, Nicolas Savy

The current literature regarding generation of complex, realistic synthetic tabular data, particularly for randomized controlled trials (RCTs), often ignores missing data. However, missing data are common in RCT data and often are not Missing Completely At Random. We bridge the gap of determining how best to generate realistic synthetic data while also accou

Mohamed Bouadi, Pratinav Seth, Aditya Tanna, Vinay Kumar Sankarapu

Tabular data drive most real-world machine learning applications, yet building general-purpose models for them remains difficult. Mixed numeric and categorical fields, weak feature structure, and limited labeled data make scaling and generalization challenging. To this end, we introduce Orion-Bix, a tabular foundation model that combines biaxial attention wi

Juan A. Delgado-Notario, Cedric Bray, Elsa Perez-Martin, Ben Benhamou-Bui

Graphene plasmons confine incident terahertz fields far below the diffraction limit and, when hosted by a gate-defined Fabry-Perot cavity, enable electrically tunable, frequency-selective photodetectors. In a magnetic field, these plasmons hybridize with the cyclotron motion to form magnetoplasmons, offering a platform for fundamental studies and for nonreci

Kent Quanrud

We present randomized algorithms that compute $(1+\epsilon)$-approximate minimum global edge and vertex cuts in weighted directed graphs in $O(\log^4(n) / \epsilon)$ and $O(\log^5(n)/\epsilon)$ single-commodity flows, respectively. With the almost-linear time flow algorithm of [CKL+22], this gives almost linear time approximation schemes for edge and vertex

Helen Guo, Elizabeth L. Ogburn, Ilya Shpitser

Identifying causal effects in the presence of unmeasured variables is a fundamental challenge in causal inference, for which proxy variable methods have emerged as a powerful solution. We contrast two major approaches in this framework: (1) bridge equation methods, which leverage solutions to integral equations to recover causal targets, and (2) array decomp

R. Scott Barrows, Julia M. Comerford, James Negus, Francisco Muller-Sanchez

From the Mapping Nearby Galaxies at Apache Point Observatory (MaNGA) survey, we identify 14 off-nuclear broad (FWHM>1000 km/s) Halpha and/or Hbeta emission line sources that indicate spatially offset active galactic nuclei (AGN) candidates. In addition to massive black holes (MBHs) in on-going galaxy mergers, this selection can also find MBHs that have been

Benjamin Kostroun, Tomoya Asaba, Sean M. Thomas, Eric D. Bauer

In magnetic topological materials, time-reversal symmetry breaking gives rise to topological point and line nodes with distinctive signatures in the anomalous Hall and anomalous Nernst conductivity that satisfy the well-known Mott relation. However, this relationship can fail for doping-dependent transport measurements of materials with complex magnetism, to

Cagatay Candan

Savitzky-Golay (SG) smoothers are noise suppressing filters operating on the principle of projecting noisy input onto the subspace of polynomials. A poorly selected polynomial order results in over- or under-smoothing which shows as either bias or excessive noise at the output. In this study, we apply the N-fold cross-validation technique (also known as leav

Colin Doumont, Donney Fan, Natalie Maus, Jacob R. Gardner

Existing high-dimensional Bayesian optimization (BO) methods aim to overcome the curse of dimensionality by carefully encoding structural assumptions, from locality to sparsity to smoothness, into the optimization procedure. Surprisingly, we demonstrate that these approaches are outperformed by arguably the simplest method imaginable: Bayesian linear regress

Badal Bhalla, Aurora Ireland, Hongwan Liu, Huangyu Xiao

The temperature of baryons at the end of the cosmic dark ages can be inferred from observations of the 21-cm hyperfine transition in neutral hydrogen. Any energy injection from the dark sector can therefore be detected through these measurements. Dark compact objects and dark-matter substructures can modify the baryon temperature by transferring heat via dyn

Arsham Ghavasieh

Deep neural networks (DNNs) exhibit crackling-like avalanches whose origin lacks a mechanistic explanation. Here, I derive a stochastic theory of deep information propagation (DIP) by incorporating Central Limit Theorem (CLT)-level fluctuations. Four effective couplings $(r, h, D_1, D_2)$ characterize the dynamics, yielding a Landau description of the static

Isaac Iyinoluwa Olufadewa, Miracle Ayomikun Adesina, Ezekiel Ayodeji Oladejo, Uthman Babatunde Usman

Depression is a major contributor to the mental-health burden in Nigeria, yet screening coverage remains limited due to low access to clinicians, stigma, and language barriers. Traditional tools like the Patient Health Questionnaire-9 (PHQ-9) were validated in high-income countries but may be linguistically or culturally inaccessible for low- and middle-inco

James Tian

This paper investigates an iterative rank-one decomposition scheme for positive operators on a Hilbert space based on a residual-weighted congruence update. At each step the operator is compressed along a chosen unit vector while remaining inside the positive cone, and the resulting map defines a monotone dynamical system on the cone of positive operators. W

Alessandro De Palma, Greta Dolcetti, Caterina Urban

Verified explanations are a principled way to explain the decisions taken by neural networks, which are otherwise black-box in nature. However, these techniques face significant scalability challenges, as they require multiple calls to neural network verifiers, each of them with an exponential worst-case complexity. We present FaVeX, a novel algorithm to com

Saeed AlMarri, Mathieu Ravaut, Kristof Juhasz, Gautier Marti

Large Language Models (LLMs) have attracted significant attention for classification tasks, offering a flexible alternative to trusted classical machine learning models like LightGBM through zero-shot prompting. However, their reliability for structured tabular data remains unclear, particularly in high stakes applications like financial risk assessment. Our

Menelaos Raptis, Gwen C. Rudie, Ryan F. Trainor, Noah S. J. Rogers

A galaxy's metallicity and its relation to stellar mass encode the history of gas accretion, star formation, and outflows within cosmic ecosystems. We present new constraints on the low-mass end of the mass-metallicity relation (MZR) at $z\sim2-3$ from ultra-deep JWST/NIRSpec spectroscopy of seven continuum-faint galaxies in the Chemical Evolution Constraine

Isabella Macias, Sydney Jenkins, Andrew Vanderburg

The search for exomoons, or moons in other star systems, has attracted significant interest in recent years, driven both by advancements in detection sensitivity and by the expanding population of known exoplanets. The $\beta$ Pictoris system is a particularly favorable target, as its proximity and directly imaged planets allow for precise astrometric monito

Christopher M. Owen, Michael J. Lawler

Metal-organic frameworks (MOFs) combine high porosity with structural fragility, raising important questions about their mechanical stability. We develop a rigidity-based framework in which spring networks parameterized by UFF4MOF are used to construct rigidity and dynamical matrices. Large-scale analysis of 5,682 MOFs from the CoRE 2019 database shows that

Miguel Vioque, Richard A. Booth, Enrico Ragusa, Álvaro Ribas

Protoplanetary disks with inner dust cavities (often referred to as "transition disks") are potential signposts of planet formation. We use Gaia astrometry to search for planetary and stellar companions in a sample of 98 transition disks, assessing the occurrence rate of such companions and their potential influence on cavity formation. For the 98 Young Stel

Joshua Davies, Dominik Grau, Kay Schönwald, Matthias Steinhauser

We compute three-loop virtual corrections to the associated production of a Higgs boson with a $Z$ boson in the large-$m_t$ limit. We describe in detail the application of the asymptotic expansion and provide, for all form factors, analytic results for the first three terms in the $1/m_t$ expansion. We also provide numerical routines implemented in the C++ l

Shunke Ai, Irene Tamborra

Shocks in astrophysical transients are key sites of particle acceleration. If the shock upstream is optically thick, radiation smoothens the velocity discontinuity at the shock (radiation-mediated shocks). However, in mildly magnetized outflows, a collisionless subshock can form, enhancing the efficiency of particle acceleration. We solve the hydrodynamic eq

Mridul Ahi, Keerti Choudhary, Shlok Pande, Pushpraj

Given a digraph $G = (V, E)$ with a designated source $s$, sink $t$, and an $(s,t)$-max-flow of value $\lambda$, we present constructions for max-flow and min-cut sensitivity oracles, and introduce the concept of a fault-tolerant flow family, which may be of independent interest. Our main contributions are as follows. 1. Fault-Tolerant Flow Family: For any g

Pavel P. Popov, Edoardo Ballini, Alberto Bottarelli, Michele Burrello

Confinement is one of the hallmarks of quantum chromodynamics (QCD). Yet, its first-principle characterization, even in simpler models, remains elusive. Through a combination of group-theoretical arguments and numerical analysis, we show that the physical consequences of confinement in a class of discrete non-Abelian lattice gauge theories (LGTs), the dihedr

Lorenzo Di Pietro, Stefanos R. Kousvos, Marco Meineri, Alessandro Piazza

Yang-Mills theory in AdS$_{4}$ with Dirichlet boundary conditions is expected to undergo a transition as the AdS radius varies, since the boundary data is incompatible with confinement in flat space. Various mechanisms have been proposed for the disappearance of the Dirichlet boundary condition. From the boundary viewpoint, the associated $3d$ CFT is a defor

Jonah C. Rose, Mariangela Lisanti, Paul Torrey, Francisco Villaescusa-Navarro

We introduce a new suite of 1,024 cosmological and hydrodynamical zoom-in simulations of Milky Way-mass halos, run with Cold Dark Matter, as part of the DREAMS Project. Each simulation in the suite has a unique set of initial conditions and combination of cosmological and astrophysical parameters. The suite is designed to quantify theoretical uncertainties f

Kaho Morii, Patricio Sanhueza, Qizhou Zhang, Giovanni Sabatini

The core mass function (CMF) of prestellar cores is essential for understanding the initial conditions of star and cluster formation. However, the universality of the CMF and its relationship to the initial mass function (IMF) remain unclear. We study the CMF in the earliest stage of high-mass star formation using 461 prestellar core candidates and 254 proto

Eloïc Vallée, Owidiusz Makuta, Patrick Emonts, Rhine Samajdar

Bell nonlocality provides a robust scalable route to the efficient certification of quantum states. Here, we introduce a general framework for constructing Bell inequalities tailored to the $\mathbb{Z}_d$ toric code for odd prime local dimensions. Selecting a suitable subset of stabilizer operators and mapping them to generalized measurement observables, we

Vinh Tran, Daniel Gilman, M. Sten Delos, Xuejian Shen

Prompt cusps (PCs) form from the direct collapse of overdensities in the early Universe, reside at the center of every dark matter halo, and have density profiles steeper than $r^{-1}$ NFW cusps. Using a suite of high-resolution N-body simulations, we study the evolution of isolated halos in self-interacting dark matter (SIDM) with massive PCs embedded at th

Mudit Jain, Elijah Sheridan, David J. E. Marsh, Elli Heyes

We develop tools of Bayesian inference on the moduli space of Calabi--Yau (CY) manifolds. We sample from the invariant Weil--Petersson (WP) measure using Markov Chain Monte Carlo and normalising flows on \Kahler moduli space with dimension up to $h^{1,1}=30$, and present results on the spectrum of the CY volume and properties of divisors when the measure is

Muhammad Maaz, Hanoona Rasheed, Fahad Shahbaz Khan, Salman Khan

Reasoning over dynamic visual content remains a central challenge for multimodal large language models. Recent thinking models generate explicit reasoning traces for interpretability; however, their reasoning often appears convincing while being logically inconsistent or weakly grounded in visual evidence. We identify and formalize these issues through two d

Hanoona Rasheed, Mohammed Zumri, Muhammad Maaz, Ming-Hsuan Yang

Recent multimodal large language models (MLLMs) have advanced video understanding, yet most still "think about videos" ie once a video is encoded, reasoning unfolds entirely in text, treating visual input as a static context. This passive paradigm creates a semantic bottleneck: models cannot rewatch, refocus, or verify evidence, leading to shallow visual rea

Bao Shu, Yan Cai, Jianjian Sun, Chunrui Han

Developing robust world model reasoning is crucial for large language model (LLM) agents to plan and interact in complex environments. While multi-turn interaction offers a superior understanding of environmental dynamics via authentic feedback, current approaches often impose a rigid reasoning process, which constrains the model's active learning, ultimatel

Zhizhou Zhong, Yicheng Ji, Zhe Kong, Yiying Liu

Recently, multi-person video generation has started to gain prominence. While a few preliminary works have explored audio-driven multi-person talking video generation, they often face challenges due to the high costs of diverse multi-person data collection and the difficulty of driving multiple identities with coherent interactivity. To address these challen

Yiping Wang, Shao-Rong Su, Zhiyuan Zeng, Eva Xu

Recent advances in large language models (LLMs) have enabled breakthroughs in mathematical discovery, exemplified by AlphaEvolve, a closed-source system that evolves programs to improve bounds on open problems. However, it relies on ensembles of frontier LLMs to achieve new bounds and is a pure inference system that models cannot internalize the evolving str

Yusuke Matsushita, Kengo Hirata, Ryo Wakizaka, Emanuele D'Osualdo

Quantum Separation Logic (QSL) has been proposed as an effective tool to improve the scalability of deductive reasoning for quantum programs. In QSL, separation is interpreted as disentanglement, and the frame rule brings a notion of entanglement-local specification (one that only talks about the qubits entangled with those acted upon by the program). In thi

Jorge E. Pérez-García, Carlos Colchero, Julio C. Gutiérrez-Vega

Accurate identification of Hamiltonian parameters is essential for modeling and controlling open quantum systems. In this work, we demonstrate that the multichannel Hankel alternative view of Koopman (mHAVOK) algorithm is a robust and reliable spectral data-driven method for retrieving Hamiltonian parameters from the evolution of first-moment observables in

Jiahao Guo, Sinan Du, Jingfeng Yao, Wenyu Liu

Large Vision Language Models (VLMs) effectively bridge the modality gap through extensive pretraining, acquiring sophisticated visual representations aligned with language. However, it remains underexplored whether these representations, optimized for multimodal understanding tasks, harbor an inherent potential for visual generation. In this paper, we propos

David Owen Horace Cutler, Mel Deaton

We introduce two valuation-based deviations on convex bodies. Using a construction that allows us to associate to these deviations "intrinsic" pseudometrics, we establish various results which capture information about the underlying valuation in terms of the geometry of their induced deviations.

Rachel Houtz, Martha Ulloa, Mia West

In this work, we present a self-consistent prediction for the gravitational wave signal arising from confinement-induced phase transitions in hidden non-Abelian SU(N) gauge theories with F light flavors. To do this, we impose perturbativity and unitarity constraints on the thermal effective potential to identify the portion of parameter space that admits a r

Danielle Paulson, Souhardya Sengupta, Lucas Janson

We introduce a new procedure for testing the significance of a set of regression coefficients in a Gaussian linear model with $n \geq d$. Our method, the $L$-test, provides the same statistical validity guarantee as the classical $F$-test, while attaining higher power when the nuisance coefficients are sparse. Although the $L$-test requires Monte Carlo sampl

Xinyi Li, Zaishuo Xia, Weyl Lu, Chenjie Hao

Current world models lack a unified and controlled setting for systematic evaluation, making it difficult to assess whether they truly capture the underlying rules that govern environment dynamics. In this work, we address this open challenge by introducing the SmallWorld Benchmark, a testbed designed to assess world model capability under isolated and preci

Samuel E. Gralla, Morifumi Mizuno

In the presence of a strong electric field, the vacuum is unstable to the production of pairs of charged particles -- the Schwinger effect. The created pairs extract energy from the electric field, resulting in nontrivial backreaction. In this paper, we study 1+1D massive QED subject to strong external electric fields in a self-consistent and fully quantum m

Michael W. Toomey, Ellie Hughes, Mikhail M. Ivanov, James M. Sullivan

Recent results from DESI BAO analyses suggest that dark energy may not be a cosmological constant and is in fact dynamical. Furthermore, the data suggest that the equation of state may have been in the phantom regime in the distant past, recently undergoing a phantom crossing. In this work, we investigate whether this preference can be realized within a kine

Marco Drewes, Juraj Klarić, Yuan-Zhen Li

Several proposed future lepton colliders are capable of producing trillions of Z-bosons, including FCC-ee, CEPC, LEP3 and LEP-Z. Such Tera-Z factories can discover new elementary particles with couplings to the Z-boson that are orders of magnitude smaller than current bounds. For couplings near the currently excluded parameter regions they could produce suff

Nihar Paul, Amala Mahadevan

The interactions between near-inertial waves (NIWs) and submesoscale currents in the surface ocean are challenging to deconvolve due to their overlapping temporal and spatial scales. The frequency of NIW is modulated by the relative vorticity, $\zeta$, of submesoscale currents, which varies between positive and negative $\zeta$ of $O(f)$ on spatial scales of

Andrea Crespi, Will J. Percival, Alex Krolewski, Marco Bonici

Galaxy clustering constrains the baryon fraction Omega_b/Omega_m through the amplitude of baryon acoustic oscillations and the suppression of perturbations entering the horizon before recombination. This produces a different pre-recombination distribution of baryons and dark matter. After recombination, the gravitational potential responds to both components

Charles W. Powell

Tropical cyclone activity was intermittent during the 2025 Atlantic season, with extended quiet periods. Cumulative activity was near-average relative to 1991-present, despite warm sea-surface temperatures and La Nina conditions. We compare drivers of activity in 2025 with climatology, using reanalysis data to examine variability in environmental conditions,

Julien Berestycki, Sarah Penington, Oliver Tough

The free boundary problem\[ \begin{cases} \partial_tu=\frac{1}{2}\Delta u+u,\quad &t>0, \, x>L_t,\\ u(t,x)=0,\quad &t>0,\, x\le L_t,\\ \int_{L_t}^{\infty}u(t,y)dy=1,\quad &t> 0,\\ u(t,x)dx \to u_0(dx)&\text{weakly as }t\to 0, \end{cases}\] has long been conjectured to be in the universality class of the so-called FKPP reaction-diffusion equation. It appears

Marwan Bit, Javier González-Anaya, Dagan Karp, Yuanyuan Luo

Gallardo and Routis constructed compactifications of the moduli space of $n$ labeled points in $\mathbb{P}^d$ by assigning weights to points, generalizing Hassett's weighted compactifications of $M_{0,n}$ to higher-dimensional projective spaces. Among their compactifications, there is a toric compactification that generalizes the standard Losev-Manin compact

Hans Gundlach, Jayson Lynch, Matthias Mertens, Neil Thompson

Language models have seen enormous progress on advanced benchmarks in recent years, but much of this progress has only been possible by using more costly models. Benchmarks may therefore present a warped picture of progress in practical capabilities *per dollar*. To remedy this, we use data from Artificial Analysis and Epoch AI to form the largest dataset of

Alexander Heckett, Vincent Conitzer

We consider settings where an uninformed principal must hear arguments from two better-informed agents, corresponding to two possible courses of action that they argue for. The arguments are verifiable in the sense that the true state of the world restricts the arguments that can be made by the agents. Each agent simply wants to be chosen as the winner and d

Andreas Eberle, Francis Lörler

We study the rate of convergence to equilibrium of the self-repellent random walk and its local time process on the discrete circle $\mathbb{Z}_n$. While the self-repellent random walk alone is non-Markovian since the jump rates depend on its history via its local time, jointly considering the evolution of the local time profile and the position yields a pie