April 2026 arXiv papers — page 107

Niels Lauritzen, Jesper Funch Thomsen

We study when algebra endomorphisms can be lifted to first-order flat lifts. To a first-order flat lift of an algebra and an endomorphism, we associate a canonical class in Hochschild cohomology with coefficients in a naturally twisted bimodule. The cohomology class vanishes exactly when the endomorphism admits a multiplicative lift. For an Azumaya algebra o

Louis H. Kauffman, Daniel S. Silver, Susan G. Williams

For any cubic graph in a closed orientable surface and a perfect matching, the Penrose-Kauffman polynomial is a sum of chromatic polynomials of a collection of associated graphs. A knot-theoretic perspective affords elementary proofs of old and new results about the polynomial. The Four Color Theorem is shown to be equivalent to a statement about 3-coloring

Muhammad Adeel Zahid, Ekram Hossain, Peng Hu

We present an end-to-end performance evaluation of MPEG-DASH video streaming over a Low-Earth Orbit (LEO) satellite-based 5G Integrated Access and Backhaul (IAB) network. Our objective is to investigate how modern transport protocols and congestion control algorithms affect adaptive video delivery in an integrated satellite-terrestrial network (ISTN), where

Sarasija Sudharsan

Finite-dimensional state-space representations of unsteady aerodynamics implicitly assume a system with fading memory. However, the impulse response of the two-dimensional inviscid (Euler) equations is characterized by an asymptotic $t^{-3/2}$ power-law decay due to the persistence of shed vorticity. The present work demonstrates that this decay rate constit

A. B. Fernandes, S. H. Pereira, J. F. Jesus, D. M. Soares Junior

In this work, we present a cosmological model in which the cosmological constant term is replaced by the Kretschmann scalar at the level of the action. In this way, it becomes possible to implement a model of dynamical dark energy. After constraining the free parameters using observational data from supernovae and cosmic chronometers, we show that the model

Claudio F. Paganini

The aim of this paper is to provide a largely self-contained, compact and comprehensible introduction to the basic ideas behind correlation geometry, which underlies the theory of causal fermion system (CFS). A key focus here is on the manner in which the framework deals with gauge transformations, including diffeomorphisms via the principle of unitary equiv

Craig Iaboni, Pramod Abichandani

Reliable UAV object detection requires robustness to illumination changes, motion blur, and scene dynamics that suppress RGB cues. Thermal long-wave infrared (LWIR) sensing preserves contrast in low light, and event cameras retain microsecond-level temporal edges, but integrating all three modalities in a unified detector has not been systematically studied.

Muhammad Saif Ullah Khan, Chen-Yu Wang, Tim Prokosch, Michael Lorenz

Inverse kinematics (IK) is a core operation in animation, robotics, and biomechanics: given Cartesian constraints, recover joint rotations under a known kinematic tree. In many real-time human avatar pipelines, the available signal per frame is a sparse set of tracked 3D joint positions, whereas animation systems require joint orientations to drive skinning.

Edilberto O. Silva, Faizuddin Ahmed

In this work, we investigate the optical, dynamical, and radiative properties of an Euler--Heisenberg black hole immersed in a perfect fluid dark matter (PFDM) background. We analyze the photon sphere and shadow, the scalar quasinormal-mode spectrum in the eikonal regime, the grey-body factor through the eikonal QNM correspondence, the sparsity of Hawking ra

Ekin Sıla Yörük, Özgür E. Müstecaplıoğlu, Zafer Gedik

Nonassociative deformations of phase-space structures arise naturally in the presence of magnetic charge, where the Jacobi identity for momentum components fails and the corresponding Moyal product becomes nonassociative. While such structures are well understood at the level of single-particle kinematics, their implications for open-system quantum dynamics

Weihua Du, Jingming Zhuo, Yixin Dong, Andre Wang He

Recent large language model (LLM) agents have shown promise in using execution feedback for test-time adaptation. However, robust self-improvement remains far from solved: most approaches still treat each problem instance independently, without accumulating reusable knowledge. This limitation is particularly pronounced in domain-specific languages such as Tr

Jonan-Rohi S. Plueger, Bethany R. Wilcox, Steven J. Pollock, Gina Passante

The Controlled-Not (CNOT) gate is essential to algorithms in quantum computing for its ability to entangle qubits. As such, it is important to understand how students learning quantum computing reason around the function and use of this critical quantum gate. To investigate this, we conducted think-aloud interviews in which students solved problems involving

Paul Breiding, Taylor Brysiewicz, David K. Johnson

We introduce a low-memory framework for certifying numerical solutions to polynomial systems which uses solution iterators and spatial partitioning trees to reduce memory requirements. We provide a prototypical algorithm, analyze its complexity, and demonstrate the memory reduction on a large example.

Livia Qian, Gabriel Skantze

Backchannels (e.g., `yeah', `mhm', and `right') are short, non-interruptive feedback signals whose lexical form and prosody jointly convey pragmatic meaning. While prior computational research has largely focused on predicting backchannel timing, the relationship between lexico-prosodic form and meaning remains underexplored. We propose a two-stage framework

Arda Fazla, Ege C. Kaya, Antesh Upadhyay, Abolfazl Hashemi

Analysis of Stochastic Gradient Descent (SGD) and its variants typically relies on the assumption of uniformly bounded variance, a condition that frequently fails in practical non-convex settings, such as neural network training, as well as in several elementary optimization settings. While several relaxations are explored in the literature, the Blum-Gladysh

Richard T. Pomeroy, Juan P. Madrid, Conor R. O'Neill, Alexander T. Gagliano

The stability and longevity of globular clusters (GCs) make them effective tracers of the dynamical histories of galaxies in cluster environments. We construct a catalog of 23,351 GC candidates in the Coma cluster using imaging from the Hubble Space Telescope Advanced Camera for Surveys. We cross-match galaxy data from the SIMBAD, NED, and SDSS archives to c

Gareth Speight, Scott Zimmerman

We construct a Lipschitz curve in the free Carnot group of step 3 with 2 generators that meets every $C^{1}$ horizontal curve in a set of measure zero. This shows that the $C^{1}_{H}$-Lusin property fails in a strong sense in this group, and we deduce that such a curve must be purely $C^1_H$ 1-unrectifiable. Hence 1-rectifiability in Carnot groups is wildly

Edson Araujo, Saurabhchand Bhati, M. Jehanzeb Mirza, Brian Kingsbury

Recent advances in reasoning models have shown remarkable progress in text-based domains, but transferring those capabilities to multimodal settings, e.g., to allow reasoning over audio-visual data, still remains a challenge, in part because of the limited availability of high-quality reasoning data in targeted multimodal combinations. To address this proble

Hassan Nasreddine

We study entropy--coherence relations near rank-deficient support boundaries in finite-dimensional quantum systems. For block-diagonal reference states, we establish support-sensitive coercivity estimates showing that the entropy cost of cross-boundary coherence acquires a logarithmic enhancement as the population scale approaches the support boundary. Combi

Habibeh Naderi, Behrouz Haji Soleimani, Stan Matwin

We introduce CoCo-LoRA, a multimodal, uncertainty-aware parameter-efficient fine-tuning method for text prediction tasks accompanied by audio context. Existing PEFT approaches such as LoRA are efficient but typically deterministic, while recent Bayesian low-rank adapters model uncertainty in a lightweight way yet remain largely unimodal and condition uncerta

Tingwei Cao, Yan Xu

Microgrid operation is highly vulnerable to short-term load uncertainty, while conventional predict-then-optimize pipelines cannot fully align probabilistic forecasting quality with downstream robust scheduling performance. This paper proposes a CVaR-guided decision-focused learning and risk-triggered re-optimization framework for two-stage robust microgrid

Seerat Kaur, Sukhjit Singh Sehra, Dariush Ebrahimi

Traffic prediction plays a central role in intelligent transportation systems (ITS) by supporting real-time decision-making, congestion management, and long-term planning. However, many existing approaches face practical limitations. Most spatio-temporal models are trained on centralized data, rely on numerical representations, and offer limited explainabili

Zhi Xiang Chong, Joong-Mok Park, Shuyuan Huyan, Avinash Khatri

Ultrafast spectroscopy under simultaneous high pressure and magnetic field provides a versatile approach for investigating pressure-driven electronic instabilities and correlated phases, and for probing potential bulk superconducting behavior under extreme conditions. However, such an experimental platform has yet to be implemented, standing as a roadblock t

Tejeswar Pokuri, Shivarth Rai

Dehazing is a technique in computer vision for enhancing the visual quality of images captured in cloudy or foggy conditions. Dehazing helps to recover clear, high-quality images from haze-affected remote sensing data. In this study, we introduce IncepDeHazeGAN, a novel Generative Adversarial Network (GAN) involving Inception block and multi-layer feature fu

Christine Breiner, Ben K. Dees

We prove a discreteness result for the possible orders of harmonic maps from surfaces to Euclidean buildings; in particular for a building of type $W$ the order is of the form $\frac mk$ where $k$ divides $|W|$. This generalizes, in the case where the domain has dimension $2$, the "order gap" of Gromov and Schoen. This result follows by directly analyzing th

Kevin Stowe, Kailash Patil

With the rise of generative language models, machine-generated text detection has become a critical challenge. A wide variety of models is available, but inconsistent datasets, evaluation metrics, and assessment strategies obscure comparisons of model effectiveness. To address this, we evaluate 15 different detection models from six distinct systems, as well

Noor Islam S. Mohammad, Uluğ Bayazıt

Large language models (LLMs) are increasingly deployed in high-stakes domains, yet a unified treatment of their overlapping safety challenges remains lacking. We present SafeLM, a framework that jointly addresses four pillars of LLM safety: privacy, security, misinformation, and adversarial robustness. SafeLM combines federated training with gradient smartif

Srishty Aggarwal, Rohan Raha, Mayank Pathak, Banibrata Mukhopadhyay

The general relativistic magnetohydrodynamic (GRMHD) simulations are widely used to study accretion disk and jet dynamics around a black hole. Despite strong observational evidences for intrinsically nonlinear behavior, the interpretations of GRMHD simulation results, more precisely the underlying timeseries, have not been well-explored by nonlinear timeseri

Nerea Gurrutxaga, Joanna Drazkowska, Vignesh Vaikundaraman, Thorsten Kleine

Carbonaceous chondrites are samples from planetesimals that formed 2-4 million years after solar system formation began. They consist of distinct dust components formed at different times and locations in the accretion disk and whose abundances in carbonaceous chondrites vary over planetesimal formation time. The mechanism that led to this time-varied accret

Timothy Launders, Cara Giovanetti, Hongwan Liu

We predict the primordial deuterium abundance using a novel, fully data-driven approach, where we use Gaussian process regression to fit experimental nuclear reaction data for $d$,($d$,$n$)$^3$He, $d$,($d$,$p$)$t$, and $d$($p$,$\gamma$)$^3$He, three reactions to which the primordial deuterium abundance is most sensitive. Using the Planck determination of the

Luca Mantani, Pablo Olgoso, Alejo N. Rossia

The search for New Physics requires ever increasing precision from experimental and theoretical efforts. Within the Standard Model Effective Field Theory (SMEFT) framework, the latest achievement in this quest has been the complete computation of the two-loop Renormalisation Group Equations (RGEs) for the Wilson Coefficients of dimension-six operators. In th

Joël Gogniat, Martin Hoferichter, Gabriele Levati

Asymmetry measurements in $e^+e^-\toτ^+τ^-$ constitute a promising avenue to obtain competitive constraints on the $τ$ dipole moments, the anomalous magnetic moment $a_τ$ and the electric dipole moment $d_τ$, especially, once a polarized electron beam becomes available, as possible at a future polarization upgrade of the SuperKEKB collider. While the main ch

Yukei Murakami, Jack Tweddle, Phil Wiseman, Saurabh Jha

Correlations between standardized Type Ia supernova (SN Ia) luminosities and host-galaxy properties are routinely modeled to avoid bias in cosmological parameter inference. A recent hypothesis attributes these correlations to progenitor-age variations and, combined with a strong ($\sim$5-6 Gyr) age evolution between low- and high-redshift samples, could alte

Tommaso Armadillo, Eugenia Celada, Jaco ter Hoeve, Fabio Maltoni

We present projections for the sensitivity of future high-energy colliders to new physics through precision measurements of the Standard Model (SM) interactions, focusing on near-term electron-positron facilities: FCC-ee, LEP3, and the Linear Collider Facility. We interpret these projections in three complementary frameworks: Higgs coupling modifiers, effect

Milena Crnogorčević, Tim Linden, Ariel Goobar, Brian D. Metzger

Superluminous supernovae (SLSNe) are among the most energetic stellar explosions, yet their central power source remains uncertain. Models invoking magnetar spin-down or circumstellar interaction predict GeV gamma-ray emission once the ejecta becomes transparent to high-energy photons. We search for such emission from 223 hydrogen-poor SLSNe using 17 years o

Haoran Feng, Yifan Niu, Zehuan Huang, Yang-Tian Sun

We introduce LaviGen, a framework that repurposes 3D generative models for 3D layout generation. Unlike previous methods that infer object layouts from textual descriptions, LaviGen operates directly in the native 3D space, formulating layout generation as an autoregressive process that explicitly models geometric relations and physical constraints among obj

Dian Shao, Zhengzheng Xu, Peiyang Wang, Like Liu

UAV vision-language navigation (VLN) requires an agent to navigate complex 3D environments from an egocentric perspective while following ambiguous multi-step instructions over long horizons. Existing zero-shot methods remain limited, as they often rely on large base models, generic prompts, and loosely coordinated modules. In this work, we propose FineCog-N

Caleb Lammers, Joshua N. Winn

Nontransiting planets can reveal themselves through transit timing variations (TTVs), but inferring the properties of the perturbing planet is a highly degenerate inverse problem. We present a systematic reassessment of all 12 published cases in which a nontransiting planet was claimed to have been uniquely characterized using TTVs. Two systems (KOI-142 and

Jakob Hultgren, Sohaib Khalid

Let $\pi:(X,L)\rightarrow \mathbb D^*$ be the Fermat family of cubic curves in $\mathbb P^2$. For each $k\geq 1$, we construct a valuatively independent basis for $H^0(X,L^k)$. The construction uses a canonical cost function determined by a Hessian structure on the essential skeleton $\op{Sk}(X,\pi)$.

Shih-Yu Chang

We introduce an operadic notion of spectrum for algebras over colored operads in a symmetric monoidal category. The construction is defined via a canonical Hochschild-type object together with an operadic residue, which together encode spectral information in a manner compatible with operadic composition. A central result of this work is that classical spect

Kerem Ekec, Uğur Teğin

Lensless cameras based on thin diffusers offer a compact alternative to conventional refractive imaging but rely on computational reconstruction, since the diffuser's point spread function (PSF) globally multiplexes every scene point across the sensor. Here, we report a Fourier Neural Operator (FNO) framework for this reconstruction task. Because a linear sh

Nicolas Ledos, Sebastiano Cantalupo, Titouan Lazeyras, Gabriele Pezzulli

Recent simulations increasingly resolve the small-scale structure of the circumgalactic medium (CGM), but the dynamical impact of ionising radiation on its cold $10^4$ K component remains poorly understood. We investigate the evolution of cold gas structures exposed to quasars' EUV radiation. We develop an analytical framework to describe the evolution of su

Qiaoyi Li, Yang Qi, Wei Li

The Hubbard model and its extensions are canonical theoretical frameworks for understanding correlated electronic states, including those in high-$T_c$ cuprates. Here, we use state-of-the-art thermal tensor network method to map out the temperature-doping phase diagram of the $t$-$t^\prime$ Hubbard model. On the electron-doped side, we find a $d$-wave superc

Jimmy Shih-Chun Hung, Arbel Haim, Mouktik Raha, Gihwan Kim

Erasure qubits are a promising platform for implementing hardware-efficient quantum error correction. Realizing the error-correction advantages of this encoding requires frequent mid-circuit erasure checks that are fast, high-fidelity, and scalable. Here, we realize erasure detection with a hardware-efficient circuit consisting of a single readout resonator

Teodoro Mayayo, Josep Sardanyés, Joan Torregrosa

Modelling how populations respond to habitat loss is crucial for understanding ecosystem stability, especially when positive interactions among resource species, such as plant-plant facilitation, play a key role. Habitat loss not only reduces available organic nutrients and space for primary producers but also disrupts the positive feedbacks that sustain res

Michael H. Heinrich, Alexander Wowchik, Jürgen Berges

We compute thermodynamic properties of dilute Bose gases using non-perturbative approximations of the two-particle irreducible (2PI) effective action. It is shown how to systematically renormalise the self-consistent descriptions beyond conventional Gaussian approximations such as Hartree-Fock-Bogoliubov theory. This allows us to determine the condensate dep

Gian Maria Dall'Ara, Roberto Frigerio, Ervin Hadziosmanovic

Let $M$ be a complete hyperbolic $n$-manifold, $n\geq 2$. Via integration over geodesic simplices, any closed bounded differential 2-form on $M$ defines a bounded cohomology class in $H^2_b(M)$. It was proved by Barge and Ghys (for $n=2$) and by Battista et al. (for $n>2$) that, if $M$ is closed, then this procedure defines an injective embedding of the (inf

Kyunghoo Mun, Matthew Rosenzweig

We study phase transitions for repulsive-attractive mean-field free energies on the circle. For a $\frac{1}{n+1}$-periodic interaction whose Fourier coefficients satisfy a certain decay condition, we prove that the critical coupling strength $K_c$ coincides with the linear stability threshold $K_\#$ of the uniform distribution and that the phase transition i

Eric Gan, Aryan Bhatt, Buck Shlegeris, Julian Stastny

As AI systems are increasingly used to conduct research autonomously, misaligned systems could introduce subtle flaws that produce misleading results while evading detection. We introduce Auditing Sabotage Bench, a benchmark for evaluating the ability of auditors to detect and fix sabotage in ML research codebases. Our benchmark consists of 9 ML research cod

Peter T. J. Bradshaw, Marcus Gouveia, Jonte R. Hance

It is well-known that any two pure quantum states (in the same Hilbert space) can be mapped to any other using unitary transformations. However, previous approaches to this problem required two explicit bases for the Hilbert space, one each for the initial and target states, and thus their complexity necessarily scales with the dimension of the Hilbert space

Shivarth Rai, Tejeswar Pokuri

Atmospheric haze significantly degrades wildlife imagery, impeding computer vision applications critical for conservation, such as animal detection, tracking, and behavior analysis. To address this challenge, we introduce AnimalHaze3k a synthetic dataset comprising of 3,477 hazy images generated from 1,159 clear wildlife photographs through a physics-based p

Daniel E. Salazar, Fabrice P. Laussy

We show that boson correlations from quantum states with a Glauber-Sudarshan representation of their density matrix which provides a well-behaved probability distribution -- including coherent states, thermal states, and all states that can be deemed classical -- are a manifestation of the Simpson paradox: they are spurious correlations from statistical (ens

Sean Hill, Felix X. -F. Ye

Stochastic dynamical systems with slow or metastable behavior evolve, on long time scales, on an unknown low-dimensional manifold in high-dimensional ambient space. Building a reduced simulator from short-burst ambient ensembles is a long-standing problem: local-chart methods like ATLAS suffer from exponential landmark scaling and per-step reprojection, whil

Ziwen Liu, Huawei Lin, Yide Ran, Denghui Zhang

Large language models (LLMs) sometimes memorize undesirable knowledge, which must be removed after deployment. Prior work on machine unlearning has focused largely on optimization methods that adjust parameters to enforce forgetting while preserving retention. However, these approaches assume that the forget and retain sets are readily available, which rarel

Michele Cicoli

We review recent progress in string model building in both early and late time cosmology. We describe the main theoretical and phenomenological features of an entire class of inflationary models where inflation is driven by a type IIB Kaehler modulus which enjoys an effective and approximate shift symmetry. We illustrate how reheating can occur via the pertu

Thomas Bayer, Alexander Lohr, Sarah Weiß, Bernd Michelberger

Explaining Machine Learning (ML) results in a transparent and user-friendly manner remains a challenging task of Explainable Artificial Intelligence (XAI). In this paper, we present a method to enhance the interpretability of ML models by using a Knowledge Graph (KG). We store domain-specific data along with ML results and their corresponding explanations, e

Shriram Chennakesavalu, Kirill Shmilovich, Hayley Weir, Colin Grambow

Large Language Models (LLMs) have the potential to accelerate small molecule drug design due to their ability to reason about information from diverse sources and formats. However, their practical utility remains unclear due to the lack of benchmarks that reflect real-world scenarios. In this work, we introduce a suite of chemically-grounded tasks spanning m

Ari Krishna

We study admissible subcategories of the bounded derived category of a smooth projective surface that are supported on the exceptional locus of a birational morphism. We prove that if $f:X\to Y$ is a birational morphism of smooth projective surfaces, then every admissible subcategory of $D^b(X)$ supported on $\operatorname{Exc}(f)$ is generated by a finite e

Hanyu Xue, Ziqian Tang, Chen Yang, Zizhao Han

Aziz and Howl argued that a classical gravitational field can generate quantum entanglement through a quantum-field-theoretic channel mediated by virtual matter propagation. However, their claimed channel is more naturally and accurately understood as semiclassical wavepacket motion in an external gravitational field, rather than as a distinctively quantum-f

Hitesh Mehta, Arjit Saxena, Garima Chhikara, Rohit Kumar

This paper explores the response of Large Language Models (LLMs) to user prompts with different degrees of politeness and impoliteness. The Politeness Theory by Brown and Levinson and the Impoliteness Framework by Culpeper form the basis of experiments conducted across three languages (English, Hindi, Spanish), five models (Gemini-Pro, GPT-4o Mini, Claude 3.

Christopher N. Gilbreth, Dmytro Filin, Marianna S. Safronova, Guanming Lao

Engineering large-scale quantum computers which simultaneously provide high-fidelity quantum operations, low memory errors, low crosstalk, and reasonable resource usage remains an outstanding challenge across quantum computing platforms. In trapped ions, progress has largely focused on alkaline-earth and ytterbium ions, whose simple electronic structures fac

F. F. Rosales-Ortega, J. E. Méndez-Delgado, J. U. Guerrero-González, C. Esteban

We present the most comprehensive set of empirical optical strong-line metallicity calibrations to date, based on the DEep Spectra of Ionised REgions Database (DESIRED), the largest compilation of HII regions and galaxies with direct electron-temperature determinations assembled to date. We construct a high-quality calibration sample of 2392 spectra$-$1029 e

Xiangbo Gao, Sicong Jiang, Bangya Liu, Xinghao Chen

As AI-assisted video creation becomes increasingly practical, instruction-guided video editing has become essential for refining generated or captured footage to meet professional requirements. Yet the field still lacks both a large-scale human-annotated dataset with complete editing examples and a standardized evaluator for comparing editing systems. Existi

Dursun Bulutoglu

We provide a method for parallelizing the branch-and-bound with isomorphism pruning algorithm developed by Margot [Symmetric ILP: Coloring and small integers, Discrete Optimization (4) (2007), 40-62]. We apply our method to classify orthogonal arrays. For classifying all non-OD- equivalent OA(128, 9, 2, 4) and OA(144, 9, 2, 4) our method results in linear sp

Van-Truong Le

The complexity of Vietnam's legal texts presents a significant barrier to public access to justice. While Large Language Models offer a promising solution for legal text simplification, evaluating their true capabilities requires a multifaceted approach that goes beyond surface-level metrics. This paper introduces a comprehensive dual-aspect evaluation frame

J. Escorza, G. Pellicer, T. de Ara, J. Hurtado-Gallego

Accurate electrical amplification is essential in molecular electronics for measuring conductance through atomic and molecular junctions, where currents often span several orders of magnitude. In this work, we present a systematic design and comparative analysis of four current-to-voltage ($I\text{--}V$) amplifier architectures: single-stage linear, series-l

Mamta Yadav, Clifford E. Woodward, Jan Forsman

Freshwater scarcity demands desalination technologies that are efficient, scalable, and sustainable. Capacitive deionisation (CDI) is promising but remains limited by inefficient ion adsorption and poor charge utilisation. Here, we show that suitably chosen polyampholytic block copolymer grafting can substantially enhance CDI performance, via a combination o

Tejeswar Pokuri, Shivarth Rai

Underwater images often suffer from severe degradation, such as color distortion, low contrast, and blurred details, due to light absorption and scattering in water. While learning-based methods like CNNs and Transformers have shown promise, they face critical limitations: CNNs struggle to model the long-range dependencies needed for non-uniform degradation,

Aswathi Mundayatt, Jaya Sreevalsan-Nair

Existing multi-hazard susceptibility mapping (MHSM) studies often rely on spatially uniform models, treat hazards independently, and provide limited representation of cross-hazard dependence and uncertainty. To address these limitations, this study proposes a deep learning (DL) workflow for joint flood-landslide multi-hazard susceptibility mapping (FL-MHSM)

Seulgi Kim, Mohit Prabhushankar, Ghassan AlRegib

Vision Language models (VLMs) have demonstrated strong performance across a wide range of benchmarks, yet they often suffer from modality dominance, where predictions rely disproportionately on a single modality. Prior approaches primarily address this issue by steering model's attention allocation, implicitly assuming that all modalities provide sufficient

Xiao Wang, Zezhong Zhang, Isaac Lyngaas, Hong-Jun Yoon

Accurate Earth system prediction requires state inference from incomplete observations, but conventional two-stage data assimilation (DA) is computationally prohibitive because repeated PDE-based ensemble forecasts, observation updates, and intermediate data movement limit ensemble size at high resolution. We introduce STORM, a one-stage generative AI framew

Ruiyang Wang, Hao-Lun Hsu, Jiwoo Kim, Miroslav Pajic

Coordinating multi-robot systems (MRS) to search in unknown environments is particularly challenging for tasks that require semantic reasoning beyond geometric exploration. Classical coordination strategies rely on frontier coverage or information gain and cannot incorporate high-level task intent, such as searching for objects associated with specific room

Deshan Sumanathilaka, Nicholas Micallef, Julian Hough, Saman Jayasinghe

Recent advances in language models have substantially improved Natural Language Understanding (NLU). Although widely used benchmarks suggest that Large Language Models (LLMs) can effectively disambiguate, their practical applicability in real-world narrative contexts remains underexplored. SemEval-2026 Task 5 addresses this gap by introducing a task that pre

Ketan Kamat, Ryota Nakano, Masatoshi Hirabayashi

The cold-classical Kuiper Belt Object 486958 Arrokoth is a contact binary composed of two flattened lobes, Weeyo and Wenu, closely aligned along their principal axes, despite each lobe having a highly irregular shape. The object's smooth and relatively undamaged structure suggests the observed bilobate shape results from a gentle, low-velocity merger between

Sarthak Mittal, Leo Gagnon, Guillaume Lajoie

Frontier models have demonstrated exceptional capabilities following the integration of task-reward-based reinforcement learning (RL) into their training pipelines, enabling systems to evolve from pure reasoning models into sophisticated agents. However, debate persists regarding whether RL genuinely instills new skills within a base model or merely sharpens

Reham Alharbi, Valentina Tamma, Terry R. Payne, Jacopo de Berardinis

Competency Questions (CQs) are a cornerstone of requirement elicitation in ontology engineering. CQs represent requirements as a set of natural language questions that an ontology should satisfy; they are traditionally modelled by ontology engineers together with domain experts as part of a human-centred, manual elicitation process. The use of Generative AI

Yige Xu, Yongjie Wang, Zizhuo Wu, Kaisong Song

Reasoning in vision-language models (VLMs) has recently attracted significant attention due to its broad applicability across diverse downstream tasks. However, it remains unclear whether the superior performance of VLMs stems from genuine vision-grounded reasoning or relies predominantly on the reasoning capabilities of their textual backbones. To systemati

Radford Green, Cornell Holmes, Mee Seong Im

The study of spanning trees and related structures is central in graph theory, closely connected to understanding functions between finite sets. This paper generalizes the established relationship between rooted trees and eventually constant endomorphisms to a wider context including $k$-tuples of functions among $k$ disjoint vertex sets. We derive a weighte

Heewon Oh

We present ArtifactNet, a lightweight framework that detects AI-generated music by reframing the problem as forensic physics -- extracting and analyzing the physical artifacts that neural audio codecs inevitably imprint on generated audio. A bounded-mask UNet (ArtifactUNet, 3.6M parameters) extracts codec residuals from magnitude spectrograms, which are then

Jinyuan Wu, Michael Wang, Datao Gong

A clustering functional block implemented in field-programable-gate-array (FPGA) for time projection chambers (TPC) operating with predictable time regardless the complexity of the event is described in this paper. The clustering functional block reorganizes input data and the hits data belonging to the same clusters are output together for further process i

Thibaut Lemoine

We identify a universal finite-$N$ structure underlying Wilson loop expectations in lattice Yang--Mills, in any dimension $d\geq 2$, for gauge group $\mathrm{U}(N)$, and for arbitrary smooth central plaquette actions. The starting point is a state-sum expansion in plaquette labels by irreducible representations, in which each term factorizes into an action-d

Thomas Dave, Jérémy Paltrinieri, Pau Petit Rosàs, William J. Torres Bobadilla

We present a first study of the scattering process $e^+ e^-\to\pi^+\pi^-\gamma$ beyond next-to-leading order, aimed at providing preliminary insights required for future NNLO predictions for radiative return processes. A complete four-dimensional tensor decomposition of the amplitude is developed, and the associated one-loop polarised amplitudes are evaluate

Tianqi Gao

Two-photon absorption (TPA) enables three-dimensional characterisation of silicon detectors by generating charge carriers within a confined volume around a focused laser spot. In combination with pixelated readout systems, TPA measurements provide access to spatially resolved timing observables relevant for electric field reconstruction. However, the interpr

Ishaan Madan, Ben K. D. Pearce

Saturn's moon Titan is a prime destination for investigating prebiotic chemistry beyond Earth, particularly at impact crater sites where transient liquid water may have enabled aqueous reactions between organic molecules. Selk crater represents one such environment and is a primary target of NASA's Dragonfly mission. Here, we present a thermodynamic assessme

Siddhant Bharadwaj, Ashish Vashist, Fahimul Aleem, Shruti Vyas

Image geolocalization has traditionally been addressed through retrieval-based place recognition or geometry-based visual localization pipelines. Recent advances in Vision-Language Models (VLMs) have demonstrated strong zero-shot reasoning capabilities across multimodal tasks, yet their performance in geographic inference remains underexplored. In this work,

Habibeh Naderi, Behrouz Haji Soleimani, Stan Matwin

We propose HILBERT (HIerarchical Long-sequence Balanced Embedding with Reciprocal contrastive Training), a cross-attentive multimodal framework for learning document-level audio-text representations from long, segmented sequences in low-resource data settings. HILBERT leverages frozen pre-trained speech and language encoders to extract segment-level features

Joseph Duque Lopez, Sergei Dudarev, James Kermode, Thomas Hudson

In order to predict the long-term effects of irradiation on the material properties of tungsten, a continuum approach to simulating the interactions of dislocation loops, which arise from radiation damage, is proposed. Continuum models of the displacement, strain and stress fields produced by dislocation loops exhibit unphysical singularities near the defect

Linnuo Zhang, Chihao Li, Jiajin Ge, Tatiana Azaryan

We present results from two test beam campaigns that investigate the performance of straw tube detectors as potential candidates for an FCC-ee straw tracker. These studies were carried out at CERN using 150 GeV muon beams. Dedicated algorithms were developed to determine both single tube spatial resolution for the primary coordinate in the $r-\phi$ plane and

Shunliang Zhang, Zhen Hu

We present the latest physics results from the FASER experiment at the LHC. Using $pp$ collision data at $\sqrt{s} = 13.6$ TeV collected during LHC Run 3, FASER reports new results on four fronts: a search for dark photons with an improved analysis strategy using $177~fb^{-1}$ of data, yielding world-leading exclusion limits; neutrino cross section measureme

Haojian Huang, Chuanyu Qin, Yinchuan Li, Yingcong Chen

Reinforcement learning has advanced video reasoning in large multi-modal models, yet dominant pipelines either rely on on-policy self-exploration, which plateaus at the model's knowledge boundary, or hybrid replay that mixes policies and demands careful regularization. Dynamic context methods zoom into focused evidence but often require curated pretraining a

Songtao Wang, Quang Hieu Pham, Fangcong Yin, Xinpeng Wang

Reinforcement learning with verifiable rewards (RLVR) typically optimizes for outcome rewards without imposing constraints on intermediate reasoning. This leaves training susceptible to reward hacking, where models exploit loopholes (e.g., spurious patterns in training data) in the reward function to achieve high scores without solving the intended task. The

Jiacheng Shen, Masato Hagiwara, Milad Alizadeh, Ellen Gilsenan-McMahon

Large language models have shown strong performance on broad-domain knowledge and reasoning benchmarks, but it remains unclear how well language models handle specialized animal-related knowledge under a unified closed-book evaluation protocol. We introduce BAGEL, a benchmark for evaluating animal knowledge expertise in language models. BAGEL is constructed

Nishq Poorav Desai, Ali Etemad, Michael Greenspan

Time-to-Collision (TTC) forecasting is a critical task in collision prevention, requiring precise temporal prediction and comprehending both local and global patterns encapsulated in a video, both spatially and temporally. To address the multi-scale nature of video, we introduce a novel spatiotemporal hierarchical transformer-based architecture called Collid

Come Fiegel, Victor Gabillon, Michal Valko

In multi-fidelity optimization, biased approximations of varying costs of the target function are available. This paper studies the problem of optimizing a locally smooth function with a limited budget, where the learner has to make a tradeoff between the cost and the bias of these approximations. We first prove lower bounds for the simple regret under diffe

Anzhou Wen, Praneeth Chakravarthula

Holographic displays are widely regarded as the "ultimate" display technology, promising immersive 3D visuals with natural depth cues, continuous parallax, and perceptual realism. Realizing this potential, however, has remained elusive due to persistent image quality limitations -- most notably speckle noise, a byproduct of the random interference inherent t

Hoyeol Kim, Hyeonwoo Kim

This paper presents a systematic benchmark of state-of-the-art multilingual large language models (LLMs) adapted via token pruning - a compression technique that eliminates tokens and embedding parameters corresponding to languages irrelevant to the target application. Focusing on Korean-centric natural language processing (NLP) tasks, we evaluate architectu

Van-Truong Le, Le-Khanh Nguyen, Trong-Doanh Nguyen

Academic integrity continues to face the persistent challenge of examination cheating. Traditional invigilation relies on human observation, which is inefficient, costly, and prone to errors at scale. Although some existing AI-powered monitoring systems have been deployed and trusted, many lack transparency or require multi-layered architectures to achieve t

Manuel Quaschner, Wijnand Steneker

We show that jets of initial data can be approximated up to arbitrary order by finite-gap solutions for classes of so-called BKM systems of PDEs introduced by Bolsinov--Konyaev--Matveev, which include classical PDEs such as KdV, Kaup--Boussinesq and Camassa--Holm. Finite-gap solutions are obtained via a finite-reduction map, defined algebraically, which send

Khawaja Azfar Asif, Rafaqat Alam Khan

Panoramic radiography is a fundamental diagnostic tool in dentistry, offering a comprehensive view of the entire dentition with minimal radiation exposure. However, manual interpretation is time-consuming and prone to errors, especially in high-volume clinical settings. This creates a pressing need for efficient automated solutions. This study presents the f

Noah Rhodes, Sachin Shivakumar, Luke S. Baker, Kaarthik Sundar

We examine market outcomes in energy transport networks with a focus on gas-fired generators, which are producers in a wholesale electricity market and consumers in the natural gas market. Market administrators monitor bids to determine whether a participant wields market power to manipulate the price of energy, reserves, or financial transmission rights. If