April 2026 arXiv papers — page 57

Isabella Illari, Andrei Afanasev, William J. Briscoe, Victor L. Kashevarov

A formalism for radiative correction calculations in exclusive $η$ electroproduction on the proton is presented, extending the treatment developed for the pion channel. The EXCLURAD code is used in the radiative correction procedure with EtaMAID-2023 multipole amplitudes. The cross-section correction factor $δ$ varies by up to ${\sim}\,30\%$ across the reson

Diana Scognamiglio, Jake H. Lee, Eric Huff, Sergi R. Hildebrandt

Future weak lensing analyses with the Nancy Grace Roman Space Telescope will require highly realistic image simulations to control shear systematics at unprecedented precision. A key limitation of existing approaches is their reliance on analytic light-profile models, which cannot fully capture the complex, non-parametric morphologies revealed by high-resolu

Apostolos Mastichiadis

Black-hole X-ray binaries (BHXRBs) in the hard and hard-intermediate spectral states commonly exhibit prominent type-C quasi-periodic oscillations (QPOs) in their X-ray power spectra. Despite extensive observational and theoretical efforts, the physical mechanism responsible for these oscillations has not yet been firmly established. The disk-corona system i

K. Kaur, I. Rammala-Zitha, A. Basu, G. Witzel

The supermassive black hole Sagittarius A* (Sgr A*) exhibits temporal and spectral variability across the electromagnetic spectrum. However, variability at radio frequencies below ~ 5 GHz for timescales shorter than a day remains largely unexplored. We investigate the variability of Sgr A* at 2.79 GHz on short timescales (1 min), to probe an under-explored r

B. I. Ciocan, N. F. Bouché, J. Fensch, D. Krajnović

The radial acceleration relation (RAR) is a tight empirical correlation between the observed radial acceleration (a_tot) and the baryonic radial acceleration (a_bar) measured across galaxy radii: these two accelerations start to deviate significantly from each other below a characteristic acceleration scale, a0. So far, observational studies of the RAR have

Kuldeep Ray, Jérémie Vigier, Sylvain Martin, Chloé Bouard

Combining spin-orbit (SOT) and spin-transfer torques (STT) provides a practical approach for field-free switching in spin-orbit torque magnetic random-access memory (SOT-MRAM), a prerequisite for industrial deployment, but can compromise reliability through phenomena such as backhopping, especially in top-pinned stacks commonly used for SOT-MRAM. We investig

Benjamin Lombardi, Ian Grooms, William Kleiber

Ocean mesoscale eddies can be thought of as the "weather" of the ocean and strongly influence the ocean's physics, chemistry, and biology; they influence other components of the Earth system via air-sea and sea-ice interactions, and are crucial drivers of marine heat waves. Thus, proper modeling of eddies in both historical and future climates is

Andrea Addazi, Konstantin Belotsky, Vitaly Beylin, Timur Bikbaev

The multi-messenger exploration of dark matter and physics beyond the Standard Model has emerged as a central direction in modern astro-particle physics, particularly following the discovery of gravitational waves. In this work, we present a comprehensive review and forward-looking perspective on machine-learning-enhanced multi-messenger approaches, combinin

Daniil Kozhemiachenko, Igor Sedlár

We explore a fuzzy modal logic that can formalise probabilistic reasoning about actions and knowledge. In particular, we deal with contexts involving statements about events expressed via modal formulas, e.g., "after doing $a$, the probability of $A$ knowing that $p$ holds increases / decreases / is equal to $0.25$", "according to $A$, $p$ is equ

H. Y. Yuan, Yizheng Wu, Olena Gomonay

Heat waste is a bottleneck in the development of green information technologies and much effort has been devoted to suppress the heating effect in both electronic and spintronic devices. Here we take an alternative approach and show that controllable heating at the nanoscale can actually benefit information processing. In particular, we study a hybrid nanost

O. Verhamme, J. O. Sundqvist, A. de Koter, H. Sana

Context. For stellar evolution models we rely on mass-loss rate prescriptions that show features that lack empirical confirmation, such as the bi-stability jump. This jump is an increase in mass loss in the decreasing temperature regime Teff 28-21 kK. Although papers compared empirical results to prescriptions,a large observational sample of B stars for whic

Talvikki Hovatta, Elina Lindfors, Heidi Korhonen, Preeti Kharb

Parsec-scale jets of blazars have so far been spatially resolved only in millimeter and submillimeter wavelengths, where very long baseline interferometry can be used to obtain milliarcsecond-scale images of the jets. We have attempted to spatially resolve the near-infrared emission in jet-dominated blazars for the first time. We used the VLTI-GRAVITY instru

Yue Liu

Current LLM coding agents are predominantly trained on composite benchmarks (e.g., bug fixing), which often leads to task-specific overfitting and limited generalization. To address this, we propose a novel scaling paradigm that shifts the focus from task-level optimization to atomic skill mastery. We first formalize five fundamental atomic skills, code loca

Sarafa A. Iyaniwura, Zhiwei Peng

Active particles exhibit self-propulsion, leading to transport behavior that differs fundamentally from passive Brownian motion. In confined or structured domains, activity strongly influence escape probabilities and first-passage behavior. Understanding these effects is essential for describing transport in biological microenvironments, microfluidic devices

Maximillan Sölch, Alexandra Neagu, Marcus Messer, Peter Johnson

Learning at scale often requires domain-specific automation such as assessment and feedback. An organization locked in to a general learning platform without these specialist automations limits its pedagogical offering. An ecosystem of interoperable, platform-agnostic microservices for domain-specific automation would solve this problem. To develop an effect

Wenyang Jia, Jingjing Wang, Xianneng Zou, Kai Lei

The stability of Internet services is persistently challenged by large volumetric TCP SYN floods, for which conventional defenses such as SYN Cookies preserve server state but still amplify bandwidth pressure. This paper presents SDN-SYN PoW, an ingress aware defense architecture that integrates non interactive Proof of Work with an SDN control plane for man

Baris Karacan, Barbara Di Eugenio, Patrick Thornton

Clinical free-text notes contain vital patient information. They are structured into labelled sections; recognizing these sections has been shown to support clinical decision-making and downstream NLP tasks. In this paper, we advance clinical section segmentation through three key contributions. First, we curate a new de-identified, section-labeled obstetric

Asmaa Cherkaoui, Ramon Flores, Delaram Kahrobaei, Richard Wilson

We propose Eidolon, a post-quantum signature scheme grounded on the NP-complete k-colorability problem. Our construction generalizes the Goldreich-Micali-Wigderson zero-knowledge protocol to arbitrary k >= 3, applies the Fiat-Shamir transform, and uses Merkle-tree commitments to compress signatures from O(tn) to O(t log n). We generate hard instances by plan

Filippo Ruffini, Camillo Maria Caruso, Claudia Tacconi, Lorenzo Nibid

Accurate survival prediction in Non-Small Cell Lung Cancer (NSCLC) requires integrating clinical, radiological, and histopathological data. Multimodal Deep Learning (MDL) can improve precision prognosis, but small cohorts and missing modalities limit its clinical applicability, as conventional approaches enforce complete case filtering or imputation. We pres

Anish Ghoshal, Pratyay Pal

We study a minimal type-I seesaw framework in which a first-order phase transition (FOPT), driven by a singlet scalar, produces right-handed neutrinos (RHNs) through bubble collisions, realizing a cosmic-scale collider that probes ultra-high energy scales. The resulting RHN distribution sources novel low-frequency gravitational-waves (GWs) in addition to the

Jia Bao, Bin Guo, Shu Qu, Fanqin Xu

Multipartite nonlocality has been extensively investigated within one-dimensional quantum lattices. Previous research has primarily focused on the nonlocality measure $S$, which quantifies the violation of Bell-type inequalities. However, the optimal nonlocal operators, which are related to specific experimental settings required to achieve the violation, of

Ya-Tang Yu, Hsin-Lien Lee, Ting Hsu, Guin-Dar Lin

Fast coherent state transport is essential to quantum computation and quantum information processing. While an adiabatic transport of atomic qubits guarantees a high fidelity of the state preparation, it requires a long timescale that defies efficient quantum operations. Here, we propose an adaptable and fast bang-bang-bang (BBB) protocol, utilizing a combin

Ida Svenningsson, Emiliya Yordanova, Yuri V. Khotyaintsev, Mats André

Despite heat flux's role in regulating energy conversion in collisionless plasmas, its properties and evolution in the magnetosheath downstream of the Earth's bow shock are scarcely explored. We use MMS in situ measurements to quantify and characterize the electron heat flux in the magnetosheath. We find that the heat flux is shaped by the magnetoshe

Yulin Ou, Yu Wang, Yang Xu, Hendrik Buschmeier

Recent theoretical advancement of information density in natural language has brought the following question on desk: To what degree does natural language exhibit periodicity pattern in its encoded information? We address this question by introducing a new method called AutoPeriod of Surprisal (APS). APS adopts a canonical periodicity detection algorithm and

Georg Ahnert, Anna-Carolina Haensch, Barbara Plank, Markus Strohmaier

Many in-silico simulations of human survey responses with large language models (LLMs) focus on generating closed-ended survey responses, whereas LLMs are typically trained to generate open-ended text instead. Previous research has used a diverse range of methods for generating closed-ended survey responses with LLMs, and a standard practice remains to be id

Luca Cotti, Idilio Drago, Anisa Rula, Devis Bianchini

System logs represent a valuable source of Cyber Threat Intelligence (CTI), capturing attacker behaviors, exploited vulnerabilities, and traces of malicious activity. Yet their utility is often limited by lack of structure, semantic inconsistency, and fragmentation across devices and sessions. Extracting actionable CTI from logs therefore requires approaches

Gayoung An, Donghyun Lee

Regularity of the Boltzmann equation, particularly in the presence of physical boundary conditions, heavily relies on the geometry of the boundaries. In the case of non-convex domains with specular reflection boundary conditions, the problem remained outstanding until recently due to the severe singularity of billiard trajectories near the grazing set, where

Luigi C. Bresciani, Gabriele Levati, Paride Paradisi

We develop a formalism, based on spinor-helicity techniques, to generalize the formulation of partial wave unitarity bounds. We discuss unitarity bounds for $N \to M$ (with $N,M \geq 2$) scattering processes -- relevant for high-energy future colliders -- and spin-2 or higher-spin theories -- relevant for effective field theories of gravity -- that are not a

Ellen M. Considine, Rachel C. Nethery

Open burning of plastic waste may pose a significant threat to global health by degrading air quality, but quantitative research on this problem -- crucial for policy making -- has been stunted by lack of data. Many low- and middle-income countries, where open burning is most concerning, have little to no air quality monitoring. Here, we leverage remotely se

Rutwig Campoamor-Stursberg, Francisco J. Herranz, Danilo Latini, Ian Marquette

We consider the quantum analog of the generalized Zernike systems given by the Hamiltonian: $$\hat{\mathcal{H}}_N =\hat{p}_1^2+\hat{p}_2^2+\sum_{k=1}^N γ_k (\hat{q}_1 \hat{p}_1+\hat{q}_2 \hat{p}_2)^k ,$$ with canonical operators $\hat{q}_i,\, \hat{p}_i$ and arbitrary coefficients $γ_k$. This two-dimensional quantum model, besides the conservation of the angu

Alex Colling, Maciej Dunajski, Hari Kunduri, James Lucietti

We construct all axi-symmetric non-gradient $m$-quasi-Einstein structures on a two-sphere. This includes the spatial cross-section of the extreme Kerr black hole horizon corresponding to $m=2$, as well as a family of new regular metrics with $m\neq 2$ given in terms of hypergeometric functions. We also show that in the case $m=-1$ with vanishing cosmological

David Pinto-Ramos, Marcel Gabriel Clerc, Abdelkader Makhoute, Mustapha Tlidi

Due to climatic changes, excessive grazing, and deforestation, semi-arid and arid ecosystems are vulnerable to desertification and land degradation. As aridity increases, vegetation cover often self-organizes into spatial patterns before collapsing to bare soil. While recent theoretical work has established that spatially heterogeneous yet isotropic environm

Aaron Weber, Daksh Dhingra, Sawyer B. Fuller

Sub-gram flapping-wing flying insect robots (FIRs) are challenging to model because of mechanical complexity in their wings, unsteady aerodynamic flow, and the difficulty of making precise measurements at a small scale. Coupling effects between roll and pitch torque actuation have not previously been measured because a two-axis sensor that is sensitive enoug

Mohamed Ouf, Mariam Guizani

Open source projects depend on newcomers who stay, yet most leave after a single contribution. Contribution events such as Google Summer of Code, LFX Mentorship, Hacktoberfest, and 24 Pull Requests attract thousands of newcomers each year, but whether they produce lasting contributors remains unclear. We conduct the first matched-cohort study comparing 2,001

Jean Odilon Andrianirina, Philippe Manjakasoa Randriantsoa, Georgette Ramanantsizehena, Domohina Raharinirina

Access to safe drinking water remains a major challenge in rural areas of developing countries. This study investigates the feasibility of a simple, low-cost solar photocatalytic reactor coated with commercial titanium dioxide (TiO$_2$) for the disinfection of well water contaminated with fecal coliforms. A TiO$_2$ film was deposited on a glass plate using a

Tianyi Liu, Christopher Twigg, Patrick Grady, Kevin Harris

Optical motion capture (mocap) systems are widely used for ground-truth capture in AR/VR, SLAM and robotics datasets. These datasets require extrinsic calibration to align mocap coordinates to external camera frames -- a step that is subject to multiple sources of error in practice, and failures often go undetected until they corrupt downstream data. These i

Harsh Kumar, Rahul Maity, Tanmay Joshi, Aman Chadha

Aligned large language models (LLMs) remain vulnerable to adversarial manipulation, and their reliance on web-scale pretraining creates a subtle but consequential attack surface. We study Stealth Pretraining Seeding (SPS), a threat model in which adversaries distribute small amounts of poisoned content across stealth websites, increasing the likelihood that

Adyant Balaji, Abhinav Uppal, Min Suk Lee, Yuchen Xu

Tinnitus is a prevalent auditory condition lacking objective biomarkers, motivating the search for reliable neural signatures. EEG, being a noninvasive method of brain imaging with a high temporal resolution provides a way to investigate the neural dynamics that may be associated with tinnitus. The generalizability of EEG-based tinnitus biomarkers across dif

Vladislav Kargin

We investigate the Brown measures of compressions of $R$-diagonal random variables, extending previous results to include unbounded cases. For random variables with finite variance, we demonstrate that the Brown measures of their compressions converge to the uniform distribution on the unit disc. In the case of infinite variance, we characterize the Brown me

Oleh E. Omel'chenko, Carlo R. Laing

We study networks of theta neurons arranged on a ring with delayed interactions. In the continuum limit the systems are described by next generation neural field models with delays. We consider distributed delays with both finite and infinite support, and conduction delays. The stability of spatially uniform and localized bump states is determined, and we fi

Oscar P. Bruno, Allen Yang

This paper presents a fast "two-dimensional Fourier Continuation" (2D-FC) method for the construction of biperiodic extensions of smooth, non-periodic functions defined over general two-dimensional (2D) domains, including domains with corners. The algorithm operates with an O(N log N) computational cost, for an N-point discretization grid, and it achieves a

Mahdi Kallel, Johannes Tölle, Ahmed Hendawy, Carlo D'Eramo

Standard supervised classification trains models to imitate the exact labels provided by a perfect oracle. This imitation happens in a single pass, restricting the model to a fixed compute budget even when inputs vary in complexity. Moreover, the rigid training objective forces the model to express absolute certainty on its training data, resulting in overco

Nalin Poungpeth, Nicholas Clark, Tanu Mitra

Large language models (LLMs) possess strong persuasive capabilities that outperform humans in head-to-head comparisons. Users report consulting LLMs to inform major life decisions in relationships, medical settings, and when seeking professional advice. Prior work measures persuasion as intentional attempts at producing the most effective argument or convinc

Razvan Gabriel Iagar, Ariel Sánchez

The large time behavior of solutions to the following generalized Burgers-Fisher-KPP equation $$ \partial_tu=u_{xx}+k(u^n)_x+u^p-u^q, \quad (x,t)\in\mathbb{R}\times(0,\infty), $$ with $n\geq2$, $p>q\geq1$ and $k\in\mathbb{R}$, is considered in this work. Denoting by $H(x,t)$, respectively $\widetilde{H}(x,t)$ the solutions having as initial condition the Hea

Bernard T. Agyeman, Zhe Li, Ilias Mitrai, Prodromos Daoutidis

We propose a hybrid reinforcement and self-supervised learning framework for accelerating generalized Benders decomposition (GBD). In this framework, a graph based reinforcement learning agent operates on a bipartite representation of the master problem and, together with a verification mechanism, determines the integer variable assignments that solve the ma

Douglas Stryker

We prove that the stable 2-systole is uniformly bounded on the space of Riemannian metrics with scalar curvature at least one for closed spin 2-essential manifolds, which includes $S^2 \times S^2$, $S^2 \times T^n$, and $\mathbb{C}\mathbb{P}^{2n+1}$.

Brandon Collins, Logan Bolton, Hung Huy Nguyen, Mohammad Reza Taesiri

When answering questions about images, humans naturally point, label, and draw to explain their reasoning. In contrast, modern vision-language models (VLMs) such as Gemini-3-Pro and GPT-5 only respond with text, which can be difficult for users to verify. We present SketchVLM, a training-free, model-agnostic framework that enables VLMs to produce non-destruc

Kritti Sharma, Elisabeth Krause, Vikram Ravi, Dhayaa Anbajagane

Fast Radio Bursts (FRBs) probe baryons permeating the cosmic web through their dispersion measures (DMs), which encode the integrated electron density along cosmological sightlines. Using 3,455 unique FRB sources from CHIME/FRB with $\sim 15$ arcmin localizations, we present an anthology of DM correlations with tracers of large-scale structure and baryonic m

Hamidreza Moradi, Scott David Kelly

An asymmetric two-link robot supported atop a flat platform by wheels that roll and pivot freely, but do not slip laterally, will develop forward momentum if the joint between the links is actuated internally. In particular, oscillations in the joint angle will generate undulatory locomotion suggesting fishlike swimming. If two such robots surmount a common

Jason Tang, Stephen Law

Street-view perception models predict subjective attributes such as safety at scale, but remain correlational: they do not identify which localized visual changes would plausibly shift human judgement for a specific scene. We propose a lever-based interventional counterfactual framework that recasts scene-level explainability as a bounded search over structu

Arthur Jakobsson, Abhinav Mahajan, Karthik Pullalarevu, Krishna Suresh

Many robotic tasks are unforgiving; a single mistake in a dynamic throw can lead to unacceptable delays or unrecoverable failure. To mitigate this, we present a novel approach that leverages learned simulation priors to inform goal-conditioned dynamic manipulation of ropes for efficient and accurate task execution. Related methods for dynamic rope manipulati

Gabriel Le Guay, Matteo Castellani, Reed Foster, Francesca Incalza

Scaling photonic quantum-information platforms requires arrays of superconducting nanowire single-photon detectors (SNSPDs) for feedforward control, in which optical operations are conditioned on Bell-state measurements relying on photon-coincidence detections. On-chip superconducting cryotron electronics, performing logic on detector outputs and driving opt

Mohamed Ragab, Faria Ferooz, Mohammad Bahrani, Helen Oliver

In decentralized personal data ecosystems grounded in architectures such as Solid, users retain sovereignty over their data via personal online data stores (pods), hosted on Solid-compliant server infrastructures. In such environments, data remains under the control of pod owners, which complicates search due to distribution across numerous pods and user-spe

Ousmane Assani-Amate, Mohammadreza Bakhtyari, Émilie Roy, Vladimir Makarenkov

Dimensionality reduction is a critical preprocessing step for clustering high-dimensional data, yet comprehensive evaluation of its impact across diverse methods and data types remains limited. In this study, we systematically assess the influence of five dimensionality reduction techniques - Principal Component Analysis (PCA), Kernel Principal Component Ana

Weisi Liu, Guangzeng Han, Xiaolei Huang

Time introduces fundamental challenges in model development and deployment: models are usually trained on historical data while deployed on future data where semantic distributions and domain knowledge may evolve. Unfortunately, existing studies either overlook temporal shifts or hardly capture rich shifting patterns of both semantic and knowledge. We develo

Zhaohui Wang

In enterprise fraud detection, model accuracy alone is insufficient when insiders can tamper with audit logs or bypass approval workflows. Real-world incidents show that fraud often persists not because detection algorithms fail, but because the audit trail itself is controllable by privileged operators. This exposes a fundamental trust gap: *who audits the

Mykola Trokhymovych, Yana Oliinyk, Nazarii Nyzhnyk

This paper presents a highly efficient Retrieval-Augmented Generation (RAG) system built specifically for Ukrainian document question answering, which achieved 2nd place in the UNLP 2026 Shared Task. Our solution features a custom two-stage search pipeline that retrieves relevant document pages, paired with a specialized Ukrainian language model fine-tuned o

Janak Bhandari, Robert Catuto, Zhumin Zhang, Bradley D. Smith

The coupling between propagating surface plasmon polaritons (SPPs) in Au nanostructures and the exciton transitions of cyanine dye J-aggregates has been examined using leakage radiation microscopy. Real space images of the nanostructures give the propagation lengths of the leaky SPP modes, and Fourier space images yield their dispersion curves. The dispersio

Nathan Shankar, Pawel Ladosz, Hujun Yin

Reliable visual perception under low illumination remains a core challenge for autonomous robotic systems, where degraded image quality directly compromises navigation, inspection, and various operations. A recent training free approach showed that Bayesian optimisation with Gaussian Processes can adaptively select brightness, contrast, and denoising paramet

Heman Shakeri, Behnaz Moradi-Jamei, Aram Vajdi, Ehsan Ardjmand

Non-Markovian (renewal) epidemic simulation on multi-million-node contact networks is essential for realistic forecasting under general age-dependent holding-time distributions (log-normal, Weibull, Erlang, and similar), but the age-dependent hazard forces dense per-step updates that render the sparse event-queue strategies of standard CPU methods ineffectiv

Caleb Dobias, Miguel Römer, Swati Bhargava, Tara Crowe

A fiber-based photonic lantern can couple an array of single-mode optical fibers to the guided modes of a multimode fiber, with the mapping between the single-mode fibers and guided modes fully described by a complex-valued transfer matrix. Recent experimental studies have reported strong wavelength-dependent evolution of this matrix in non-mode-selective ph

Yaping Mao

A finite Euclidean set is diameter-Ramsey if, for every number of colors, some finite same-diameter witness has the property that every coloring of the witness contains a monochromatic congruent copy of the set. Frankl, Pach, Reiher and R\"odl asked whether any obtuse triangle is diameter-Ramsey. We prove the stronger statement that every non-degenerate tria

Shin Hwei Tan, Haibo Wang, Heng Li

Generative Artificial Intelligence (GAI) systems that can automatically generate content in the form of source code or other contents (e.g., images) has seen increasing popularity due to the emergence of tools such as ChatGPT which rely on Large Language Models (LLMs). Misuse of the automatically generated content can incur serious consequences due to potent

Alberto Cattaneo, M Keith Ballard, Robert M. Kirby, Varun Shankar

The rapid rise of scientific machine learning (SciML) has expanded the role of differentiable modeling, surrogate modeling, and data-driven constitutive laws in large-scale simulation. The JAX framework provides an attractive environment for these workflows through automatically differentiable programs, vectorization, GPU acceleration, and while enabling sea

Ekta Bhatia, Yingge Du, Krishna P Koirala, Chung Kow

Surface oxides contribute to losses in superconducting transmon devices resulting in degraded performance. We explore the use of the damascene process to replace the sidewall native oxide of a device with a metal/substrate interface. We simulate sidewall oxidation by burying an oxide layer during fabrication. We observe a modest improvement between the two t

Seyed Moein Abtahi, Rasa Rahnema, Hetkumar Patel, Neel Patel

The transition from stateless language model inference to persistent, multi session autonomous agents has revealed memory to be a primary architectural bottleneck in the deployment of production grade agentic systems. Existing methodologies largely depend on hybrid semantic graph architectures, which impose substantial computational overhead during both inge

Tiffany Bao, Kylie Trousil, Quang Duy Tran, Fabio Di Troia

Malware attacks have a significant negative impact on organizations of varied scales in the field of cybersecurity. Recently, malware researchers have increasingly turned to machine learning techniques to combat sophisticated obfuscation methods used in malware. However, collecting a diverse set of malware samples with various obfuscation techniques is chall

Baobiao Yue, Karl-Heinz Kampert, Julian Rautenberg

Ultra-high-energy (UHE) neutrinos are unique cosmic messengers that can traverse cosmological distances unattenuated, providing direct insight into the most energetic processes in the universe. Radio detection offers significant advantages for detecting highly inclined air showers induced by UHE neutrinos. This is due to a larger exposure range compared to p

Emil Ryd, Henning Bartsch, Julian Stastny, Joe Benton

As AI systems begin to automate complex tasks, supervision increasingly relies on weaker models or limited human oversight that cannot fully verify output quality. A model more capable than its supervisors could exploit this gap through sandbagging, producing work that appears acceptable but falls short of its true abilities. Can training elicit a model's be

Anne E. Staples

Most reinforcement-learning (RL) controllers used in continuous control are architecturally centralized: observations are compressed into a single latent state from which both value estimates and actions are produced. Biological control systems are often organized differently. Insects, in particular, coordinate navigation, heading stabilization, memory, and

Meihong Liu

The Belle and Belle~II experiments have accumulated a combined data set of $1.2~\mathrm{ab}^{-1}$ of $e^+e^- \to B\bar{B}$ collisions at the $\Upsilon(4S)$ resonance. Owing to the clean event environment and the well-constrained initial-state kinematics, these data are particularly well suited for studying channels involving missing energy from neutrinos. Th

Mohamad Maassarani

For $G$ a finite group, one way to construct irreducible quandle representations over $\mathbb{C}$ of the conjugacy quandle $Conj(G)$ is by taking the product of an irreducible linear group representation of $G$ by what we call a quandle character of $Conj(G)$ (a quandle morphism into $\mathbb{C}^\times$ ). We show that these are all the irreducible quandle

Muhy Eddin Za'ter, Bri-Mathias Hodge

Accurate forecasting of electric load and renewable generation is essential for reliable and cost effective power system operations. Recent advances in transformer based and foundation machine learning models, driven by large scale pretraining, increased available data and computation, in addition to architectural innovations, have shown promise in time seri

Xiaoyi Chen, Haoyuan Wang, Siyuan Tang, Sijia Liu

Large language models (LLMs) often memorize private information during training, raising serious privacy concerns. While machine unlearning has emerged as a promising solution, its true effectiveness against privacy attacks remains unclear. To address this, we propose PrivUn, a new evaluation framework that systematically assesses unlearning robustness throu

Qinan Yu, Alexa Tartaglini, Peter Hase, Carlos Guestrin

Reinforcement Learning from Verifiable Rewards (RLVR) on chain-of-thought reasoning has become a standard part of language model post-training recipes. A common assumption is that the reasoning chains trained through RLVR reliably represent how a model gets to its answer. In this paper, we develop two metrics for critically examining this assumption: Causal

Helge Kristian Jenssen

We consider the compressible Euler system for ideal gas flow in the absence of any forces except the internal thermodynamic pressure. In this setting, and in dimensions higher 1, it is known that wave-focusing can drive Euler solutions to amplitude blowup in finite time from bounded initial data. In the known cases (self-similar, radial flows \cites{gud,hun_

Amine Barrak

Federated learning (FL) aggregation on serverless platforms faces a hard scalability ceiling: existing architectures (lambda-FL, LIFL) partition clients across aggregators, but every aggregator must hold the complete model gradient in memory. When gradients exceed the per-function memory limit (e.g., 10 GB on AWS Lambda), aggregation becomes infeasible regar

Hiranya V. Peiris

If a Large Language Model (LLM) can replicate your scientific contribution, the problem is not the LLM. What does it say about our field that so much of the anxiety about AI comes down to the fear that a machine could do what we do? Perhaps it says we should be doing something better.

Jackson L. Jewett, Josephine V. Carstensen

The production of concrete generates roughly 8% of anthropogenic CO2 globally, largely because of the massive quantities that are manufactured. New design methods must be developed and deployed to improve the material efficiency of reinforced concrete structures, and reduce concrete's carbon impact. This research uses topology optimization, a free-form struc

Andrey Urusov, Rostislav Berezovskiy, Anatoly Krestenko, Andrei Kornilov

The emergence of Concentrated Liquidity Market Makers (CLMMs) has made liquidity provision on decentralized exchanges an active and risk-sensitive task. However, the standalone profitability of liquidity provision remains unclear for liquidity providers (LPs) who neither hedge their inventory risk nor receive off-pool profits. This paper studies the actual o

Nahian Salsabil, Sebastian Elbaum

Validating Autonomous Vehicles (AVs) requires exposure to rare, safety-critical scenarios, infrequent in routine driving data. Existing benchmarks address this by generating synthetic conflicts or mapping accident descriptions to abstract road geometries, failing to capture the topological complexity of real-world crashes. We introduce TRACE , a pipeline tha

Guan Gui, Peter Zandi, Jacob Taylor, Ananya Joshi

Psychiatric intake is a sequential, high-stakes information-gathering process in which clinicians must decide what to ask, in what order, and how to interpret incomplete or ambiguous responses under limited time. Despite growing interest in conversational AI for healthcare, there is still limited infrastructure for conversational AI in this application. Acco

Mehmet Kemal Gümüş

We develop a minisuperspace formulation of the classical double copy for anisotropic Lifshitz spacetimes in arbitrary dimension. By imposing static symmetries at the level of the action, the gravitational system reduces to an effective one-dimensional radial problem with a universal structure, in which all theory dependence is captured by a single potential.

Anushka Kulkarni, Sarthak Dubey

We present Selective Non-Gaussian Refinement (SNGR), a SLAM framework that augments iSAM2 with targeted nested sampling on windows where Gaussian approximations are likely to fail. We detect such regions using the condition number of joint marginal covariances and selectively refine them using the full nonlinear factor graph likelihood, with a gating mechani

Shevya Panda, Shinjini Bose, Ananya Joshi

Large language models (LLMs) are increasingly utilized in clinical reasoning and risk assessment. However, their interpretive reliability in critical and indeterminate domains such as psychiatry remains unclear. Prior work has identified algorithmic biases and prompt sensitivity in these systems, raising concerns about how contextual information may influenc

Karthic Palaniappan

There are 7,407 languages in the world. But, what about the languages that are not there in the world? Are humans so narrow minded that we don't care about the languages aliens communicate in? Aliens are humans too! In the 2016 movie Arrival, Amy Adams plays a linguist, Dr. Louise Banks who, by learning to think in an alien language (Heptapod) formed of non-

Xiaodi Li, Yang Xiao, Munhwan Lee, Konstantinos Leventakos

Patient-trial matching requires reasoning over long, heterogeneous electronic health records (EHRs) and complex eligibility criteria, posing significant challenges for scalability, generalization, and computational efficiency. Existing approaches either rely on full-document processing with large language models (LLMs), which is computationally expensive, or

Jaeyoung Jung, Marco Giometto

This study derives conservative and skew-symmetric formulations of the incompressible flow equations in a terrain-following sigma-coordinate system that preserve key structural properties of the Cartesian formulation. Unlike conventional formulations based on the direct application of the sigma-transformation to Cartesian equations, in which metric-induced t

Sikander Azam, Asif Zaman, Qaiser Rafiq, Amin Ur Rahman

All inorganic halide perovskites offer promising optoelectronic properties at low cost, but their structural softness and thermal instability limit applications. Density functional theory using the FP-LAPW method (WIEN2k) was used to study Ni2+/Pr3+ co-doping in CsPbCl3. Results show Ni2+ substitutes for Pb2+ at the B-site and Pr3+ for Cs at the A-site, keep

Jasmine Parsons, Anatoly Spitkovsky, Alexander Philippov, Hayk Hakobyan

We present the first 3D global kinetic simulations of the interacting magnetospheres of pre-merger binary neutron stars. The stars, whose magnetic moments are anti-aligned, twist the field lines connecting them, leading to periodic eruptions. Each eruption consists of an expanding magnetic flux tube with a reconnecting current sheet trailing behind it, topol

Andreas Weingartner

We establish a relationship between error terms appearing in estimates for the counting functions of smooth and rough numbers. We then apply this link to obtain an explicit upper bound for the error term in de Bruijn's approximation $\Lambda$ for the count of smooth numbers, from an explicit upper bound, due to Fan, for the error term in a variant of de Brui

Elias Hossain, Mohammad Jahid Ibna Basher, Ivan Garibay, Ozlem Garibay

Offline reinforcement learning (RL) can learn effective policies from fixed datasets, but deployment objectives may change after training, and in many applications the trained actor cannot be retrained because of data, cost, or governance constraints. We study deployment-time adaptation for frozen offline actors using Product-of-Experts (PoE) composition wit

Md Tanjemul Islam, Md Rafiul Kabir

Autonomous vehicles (AVs) rely on real-time perception systems to understand road environments and ensure safe navigation. However, implementing reliable perception algorithms on resource-constrained embedded platforms remains challenging due to limited computational resources. This paper presents a lightweight vision-based framework that integrates lane det

Xuyuan Wang, Donglin Han

Markov chain Monte Carlo (MCMC) methods provide powerful framework for sampling unknown probability measures across a wide range of scientific applications. In some settings, the target distribution is supported on a lower-dimensional submanifold of Euclidean space defined by nonlinear constraints, motivating the development of constrained Hamiltonian Monte

Henglin Pu, Husheng Li

Integrated sensing and communication (ISAC) is poised to be a defining feature of 6G networks, promising to transform cellular base stations (BSs) into ubiquitous radar sensors. However, a significant gap exists between the theoretical promise of ISAC and the commercial reality of legacy cellular communication infrastructure. Existing communication networks

Cheng Jiang, Yinchen Liu, Huacheng Yu

A fundamental question in streaming complexity is whether every space-efficient turnstile algorithm is implicitly a linear sketch. The landmark work of Li, Nguyen, and Woodruff [LNW14] established an equivalence between the two, but their reduction requires a stream length that is at least doubly exponential in the dimension $n$. In the opposite direction, r

Robert W. Krause

Interaction effects are ubiquitous in applied statistical modelling, yet their meaningful interpretation remains challenging. The classic Johnson-Neyman (JN) technique (Johnson and Neyman 1936) addresses this challenge for two-way interactions by identifying the regions of a moderator's range over which a focal effect is and is not statistically significant.

Martin Charron, Vincent Tabard-Cossa

An exact solution is presented for the resistance of an orifice in a 2D membrane separating two infinitely large conductive reservoirs and obstructed by an infinitely long cylinder. The solution is obtained by constructing a curvilinear coordinate system that captures the symmetry of the obstructed system with constant-coordinate surfaces mapping the system

Ismael Mendoza, Axel Guinot, Matthew R. Becker, Camille Avestruz

Forthcoming Stage-IV dark energy optical surveys, such as LSST, have the ambitious goal of measuring cosmological parameters at sub-percent precision. Realizing their full scientific potential requires very precise measurement of the cosmic shear signal and control of corresponding systematics. In this work, we present a modern implementation of the Bayesian

Ahmed Mohammed Cherif

We examine biharmonic submanifolds within warped product structures. For a submanifold $(M,g)\subset (N,h)$ and a positive smooth function $f:I\to\mathbb{R}^+$, we study the inclusion $\varphi:(I\times M,\widetilde{g})\to (I\times N,\overline{h})$, where $\widetilde{g}=dt^2+f^2g$ and $\overline{h}=dt^2+f^2h$. We relate the tension and bitension fields of $\v