March 2026 arXiv papers — page 90

Nitesh Kumar, Jianwei Lai, Casey S. Mezerkor, Jiaqi Wang

Machine learning interatomic potentials (MLIPs) trained on large, chemically diverse datasets are revolutionizing computational chemistry, enabling molecular dynamics simulations of battery electrolytes with near-DFT accuracy over 10,000 times faster than DFT. While previous MLIP training datasets with suitable elemental coverage for electrolytes have been b

Jianan Huang, Rodolfo V. Valentim, Luca Vassio, Matteo Boffa

The use of ML in cybersecurity has long been impaired by generalization issues: Models that work well in controlled scenarios fail to maintain performance in production. The root cause often lies in ML algorithms learning superficial patterns (shortcuts) rather than underlying cybersecurity concepts. We investigate contrastive multi-modal learning as a first

Yuning Huang, Xiaoyu Ji, Joseph Huang, Yichi Zhang

Large vision--language models (VLMs) are increasingly applied to long-video question answering, yet inference is often bottlenecked by the number of input frames and resulting visual tokens. Naive sparse sampling can miss decisive moments, while purely relevance-driven selection frequently collapses onto near-duplicate frames and sacrifices coverage of tempo

Juan Andrés Urrea-Niño, Francesco Knechtli, Tomasz Korzec, Michael Peardon

Construction of creation operators which can properly sample the underlying energy eigenstates remains a fundamental first step in lattice QCD spectroscopy calculations, particularly when the spectrum includes states with different composition such as mesons, glueballs, multi-particle states, etc. We tackle this issue in the study of the scalar glueball and

Burak Öz, Akaki Mamageishvili, Christoph Schlegel, Ali Taslimi

We study Arbitrum's Timeboost auction, an ahead-of-time mechanism that sells a 200ms ordering advantage in an otherwise first-come, first-served transaction ordering policy. The market naturally divides into two phases: a competition phase, in which the dominant searchers compete directly in the primary auction, and a coordination phase, in which they so

Luigi Capogrosso, Michele Magno

Earth observation (EO) missions traditionally rely on transmitting raw or minimally processed imagery from satellites to ground stations for computationally intensive analysis. This paradigm is infeasible for CubeSat systems due to stringent constraints on the onboard embedded processors, energy availability, and communication bandwidth. To overcome these li

Surasak Phetmanee

We develop a formal theory of throughput in finite serial pipeline systems subject to stage multiplicative capacity perturbations, motivated by the deployment of AI tools in cybersecurity operations. A pipeline is a finite totally ordered set of stages each with a positive capacity throughput is the minimum stage capacity. An admissible multiplier assigns to

Lars Becker, Polona Durcik

We prove $L^p$ estimates for the shifted bilinear Hilbert transform, with a polylogarithmic bound in the size of the shift. As applications, we obtain $r$-variation estimates for bilinear ergodic averages in the sharp range $r > 2$, a sharp bilinear H\"ormander multiplier theorem, and a $\log$-Dini theorem for bilinear singular integrals.

Richard J. Young

Recent work on chain-of-thought (CoT) faithfulness reports single aggregate numbers (e.g., DeepSeek-R1 acknowledges hints 39% of the time), implying that faithfulness is an objective, measurable property of a model. This paper provides evidence that it is not. Three classifiers (a regex-only detector, a regex-plus-LLM pipeline, and a Claude Sonnet 4 judge) a

R. Rodríguez-Cardoso, S. Roca-Fàbrega, Oscar Agertz, Jesus Gallego

Satellite galaxies in the Local Group tend to be distributed in thin, planar configurations, with many sharing coherent orbital motion. Galaxy formation simulations in $\Lambda$CDM have historically struggled to produce similar structures, leading to the so-called "planes of satellites problem". In this work, we investigate whether the emergence of such stru

Xuanwang Zhang, Yuteng Han, Jinnan Qi, Mulong Xie

Despite significant advances in autonomous web navigation, current methods remain far from human-level performance in complex web environments. We argue that this limitation stems from Topological Blindness, where agents are forced to explore via trial-and-error without access to the global topological structure of the environment. To overcome this limitatio

Ruxiao Chen, Xilei Zhao, Thomas J. Cova, Frank A. Drews

Theory of Mind (ToM) reasoning with Large Language Models (LLMs) requires inferring how people's implicit, evolving beliefs shape what they seek and how they act under uncertainty -- especially in high-stakes settings such as disaster response, emergency medicine, and human-in-the-loop autonomy. Prior approaches either prompt LLMs directly or use latent-stat

Yifan Shen, Jiateng Liu, Xinzhuo Li, Yuanzhe Liu

Generative world models have shown promise for simulating dynamic environments, yet egocentric video remains challenging due to rapid viewpoint changes, frequent hand-object interactions, and goal-directed procedures whose evolution depends on latent human intent. Existing approaches either focus on hand-centric instructional synthesis with limited scene evo

Nahid Binandeh Dehaghani, Rafal Wisniewski, A. Pedro Aguiar

High-dimensional Schr\"odinger systems arising from tensor-product discretizations suffer from exponential state growth, making direct controller synthesis and real-time closed-loop simulation computationally challenging. Hierarchical Tucker (HT) tensor representations offer scalable low-rank surrogates, but the impact of fixed-rank truncation on closed-loop

Maria Eduarda Veras, Eduardo Freitas, Assis T. de Oliveira Filho, Djamel Sadok

The demand for ultra-low latency in modern applications, such as cloud gaming and augmented reality, has exposed the limitations of traditional congestion control algorithms regarding bufferbloat. The Low Latency, Low Loss, and Scalable Throughput (L4S) architecture addresses this challenge by combining scalable congestion controls, such as TCP Prague, low-l

Ali Darijani, Jürgen Beyerer, Zahra Sadat Hajseyed Nasrollah, Luisa Hoffmann

Probability theory has become the predominant framework for quantifying uncertainty across scientific and engineering disciplines, with a particular focus on measurement and control systems. However, the widespread reliance on simple Gaussian assumptions--particularly in control theory, manufacturing, and measurement systems--can result in incomplete represe

Candice R. Gerstner

With the advancements in AI speech synthesis, it is easier than ever before to generate realistic audio in a target voice. One only needs a few seconds of reference audio from the target, quite literally putting words in the target person's mouth. This imposes a new set of forensics-related challenges on speech-based authentication systems, videoconferencing

Jiyu Lim, Youngwoo Yoon, Kwanghyun Park

Conventional robot social behavior generation has been limited in flexibility and autonomy, relying on predefined motions or human feedback. This study proposes CRISP (Critique-and-Replan for Interactive Social Presence), an autonomous framework where a robot critiques and replans its own actions by leveraging a Vision-Language Model (VLM) as a `human-like s

W. Quan, E. Camphuis, C. Daley, N. Huang

Maps of the sky in millimeter wavelengths contain rich information on cosmology through anisotropies of the cosmic microwave background (CMB). Creating multifrequency sky maps of anisotropies in the $I$, $Q$, and $U$ Stokes parameters is one of the first steps of CMB cosmology analyses. In this work, we describe the production and validation of a set of sky

Sai Koneru, Elphin Joe, Christine Kirchhoff, Jian Wu

In contested domains, instruction-tuned language models must balance user-alignment pressures against faithfulness to the in-context evidence. To evaluate this tension, we introduce a controlled epistemic-conflict framework grounded in the U.S. National Climate Assessment. We conduct fine-grained ablations over evidence composition and uncertainty cues acros

Qi Cao, Andrew Gambardella, Takeshi Kojima, Yutaka Matsuo

Large language models (LLMs) have demonstrated remarkable capabilities across diverse tasks. However, the truthfulness of their outputs is not guaranteed, and their tendency toward overconfidence further limits reliability. Uncertainty quantification offers a promising way to identify potentially unreliable outputs, but most existing methods rely on repeated

Rodrigo F. Pinheiro, André A. Costa, Yu Sang

We compute the full-sky angular power spectrum and bispectrum, along with their Fisher matrices, to forecast constraints on cosmological parameters for the BINGO and SKA1-MID Band 2 radio telescopes. This represents the first forecast analysis using the full-sky relativistic bispectrum in redshift space for these surveys. Our results show that the second-ord

Lara Kim Linke, Yvonne Tomm, Xinyun Liu, Galina Gurieva

Atomic-scale disorder can create hidden optical anisotropy even in crystals that are structurally cubic on average. Here, we show that CuInSnS$_4$ single crystals host locally symmetry-broken environments arising from intrinsic In/Sn cation disorder, which affect vibrational and excitonic properties in markedly different ways. Combining polarization- and tem

Gandalf Lechner

The problem of classifying all unitary R-matrices of arbitrary finite dimension that have precisely two distinct eigenvalues is described, working up to a natural equivalence relation given by the characters of their braid group representations. Up to one class that might or might not exist in even dimension larger than two, a full classification theorem is

Júnio Luan Pereira

This text highlights issues present in the proof of Lemma 6.10 of the Baumgartner (1943 -- 2011) article "Almost disjoint sets, the dense set problem and the partition calculus" of 1976, and intends to present a correction at the same time it proves a stronger result mentioned in the article to have similar proof.

Ahmed Abuali, David A. Clarke, Morten Hjorth-Jensen, Ioannis Konstantinidis

We develop a one-class, deep-learning framework to detect the phase transition and recover critical behavior of the 3D Ising model. A 3D convolutional neural network autoencoder (CAE) is trained on ground-state configurations only, without prior knowledge of the critical temperature, the Hamiltonian, or the order parameter. After training, the model is appli

Jiaxuan Cai, Xinmiao Zhang

Hamming Quasi-Cyclic (HQC) was chosen for the latest post-quantum cryptography standardization. A concatenated Reed-Muller (RM) and Reed-Solomon (RS) code is decoded during the HQC decryption. Soft-decision RS decoders achieve better error-correcting performance than hard-decision decoders and accordingly shorten the required codeword and key lengths. Howeve

Emiel Hoogeboom, David Ruhe, Jonathan Heek, Thomas Mensink

It is currently difficult to distill discrete diffusion models. In contrast, continuous diffusion literature has many distillation approaches methods that can reduce sampling steps to a handful. Our method, Discrete Moment Matching Distillation (D-MMD), leverages ideas that have been highly successful in the continuous domain. Whereas previous discrete disti

Aashish Joshi, Prisha, Neetu Raj Singh Chundawat, Jitendra Kumar

The $B$-factory experiments operate at electron-positron colliders with beam energies precisely tuned for optimal $B^{0}\text{-}\bar{B}^{0}$ meson pair production. These $B^{0}\text{-}\bar{B}^{0}$ meson pairs are produced entangled and offer a unique opportunity to explore quantum correlations and examine the foundational aspects of quantum mechanics. In thi

Jakub Skrzeczkowski

We consider cross-diffusion systems describing evolution of two species $u$ and $v$ moving according to Darcy's law with the pressure law $p(s) = \frac{1}{\alpha-1} s^{\alpha-1}$ where $s=u+v$. One of the most challenging questions in the field is the construction of solutions to the problem in the presence of additional advection fields, without imposing an

Mandana Mohammadi Looey, Amrita Basak, Satadru Dey

Extrusion-based 3D printing of cementitious materials enables fabrication of complex structures, however it is highly sensitive to disturbances, material property variations, and process uncertainties that decrease flow stability and dimensional fidelity. To address these challenges, this study proposes a robust linear quadratic optimal control framework for

H. Sinan Bank, Daniel R. Herber, Thomas H. Bradley

Engineering system design -- whether mechatronic, control, or embedded -- often proceeds in an ad hoc manner, with requirements left implicit and traceability from intent to parameters largely absent. Existing specification-driven and systematic design methods mostly target software, and AI-assisted tools tend to enter the workflow at solution generation rat

Shuoyuan Xu, Zhipeng Zhong, Tiago Barros, Matthew Coombes

Agricultural robotics is gaining increasing relevance in both research and real-world deployment. As these systems are expected to operate autonomously in more complex tasks, the availability of representative real-world datasets becomes essential. While domains such as urban and forestry robotics benefit from large and established benchmarks, horticultural

Ali Sakour, Zoalfekar Sakour

The Hyperspace Analogue to Language (HAL) model relies on global word co-occurrence matrices to construct distributional semantic representations. While these representations capture lexical relationships effectively, aggregating them into sentence-level embeddings via standard mean pooling often results in information loss. Mean pooling assigns equal weight

Qianqi Yan, Yichen Guo, Ching-Chen Kuo, Shan Jiang

Modern multimodal large language models (MLLMs) generate fluent responses from interleaved text, image, audio, and video inputs. However, identifying which input sources support each generated statement remains an open challenge. Existing attribution methods are primarily designed for classification settings, fixed prediction targets, or single-modality arch

Hui Zhong, Yichun Gao, Luyan Liu, Hai Yang

Automated building facade inspection is a critical component of urban resilience and smart city maintenance. Traditionally, this field has relied on specialized discriminative models (e.g., YOLO, Mask R-CNN) that excel at pixel-level localization but are constrained to passive perception and worse generization without the visual understandng to interpret str

Huihua Zhao, Rafael Cathomen, Lionel Gulich, Wei Liu

Recent advances in reinforcement learning (RL) have enabled impressive humanoid behaviors in simulation, yet transferring these results to new robots remains challenging. In many real deployments, the primary bottleneck is no longer simulation throughput or algorithm design, but the absence of systematic infrastructure that links environment verification, tr

Ruixuan Zhao, Guitao Yang, James Fleming, Boli Chen

With the advancement of IoT technologies and the rapid expansion of cyber-physical systems, there is increasing interest in distributed state estimation, where multiple sensors collaboratively monitor large-scale dynamic systems. Compared with its continuous-time counterpart, a discrete-time distributed observer faces greater challenges, as it cannot exploit

Hui Zhong, Yichun Gao, Luyan Liu, Xusen Guo

Building facade defect inspection is fundamental to structural health monitoring and sustainable urban maintenance, yet it remains a formidable challenge due to extreme geometric variability, low contrast against complex backgrounds, and the inherent complexity of composite defects (e.g., cracks co-occurring with spalling). Such characteristics lead to sever

C. Antonio, I. Chifu, R. Gafeira, J. J. G. Lima

The Potential Field Source Surface (PFSS) model is the most used approach for extrapolating the global coronal magnetic field, offering efficiency and strong performance at large scales. However, PFSS assumes a potential coronal field, so it cannot account for distortions from electric currents. More advanced methods, such as nonlinear force-free field (NLFF

Haoran Ma, Yuchen Zheng, Leining Zhang, Xiaofei Chen

Strain engineering provides a powerful route for tuning the electronic properties of two-dimensional (2D) materials, but exploring the full multidimensional strain space with density functional theory (DFT) is computationally prohibitive due to the nonlinear coupling between normal and shear components. In this work, we introduce a Transformer-based, multi-t

Atabey Kaygun

We construct a category $\OrdFor$ as an arboreal extension of $\Delta_{\mathrm{epi}}\subseteq\Delta$, whose morphisms are ordered forests composed by grafting. We define a full functor $\pi\colon \OrdFor\to\Delta_{\mathrm{epi}}^{op}$ extracting the semisimplicial shadow. For every complete category $\mathcal C$, this induces a fully faithful functor from sem

S. Bogdanov, S. Sygletos, O. Sidelnikov, G. Gomes

In this letter, we numerically investigate a long-haul coherent data transmission system with a cascade of semiconductor optical amplifiers (SOAs). We exploit low-complexity neural networks that can be implemented in real time to compensate for the accumulated distortions induced by a cascade of SOAs. This equalization provides an order-of-magnitude reductio

Nima Farahmand Bafi, Robert Evans, Anna Maciolek

The phase behavior of a single type of colloid C suspended in near-critical solvents is known to be very rich. Motivated in part by recent experiments we consider a mixture of two colloidal types C1 and C2 in a binary solvent close to its demixing critical point. We extend a mean-field description of a lattice model, previously used to investigate systems wi

Guy Perrin

The effects of the polarization characteristics of beam trains in optical long-baseline interferometers are well known and have led to difficulties in measuring the spatial coherence of astronomical sources in the past. This has been overcome by designing symmetrical optical trains. With the advent of interferometers using large telescopes, observations of f

Chia-Yu Hsu, Shubhanshu Shekhar

We consider the problem of constructing sequential power-one tests where the null and alternative classes are specified indirectly through historical or offline data. More specifically, given an offline dataset consisting of observations from $L+1$ distributions $\{P_0, P_1, \ldots, P_L\}$, and a new unlabeled data stream $\{X_t: t \geq 1\} \overset{i.i.d}{\

Denis Chetverikov, Jesper R. -V. Sørensen, Aleh Tsyvinski

In this paper, we propose a triple (or double-debiased) Lasso estimator for inference on a low-dimensional parameter in high-dimensional linear regression models. The estimator is based on a moment function that satisfies not only first- but also second-order Neyman orthogonality conditions, thereby eliminating both the leading bias and the second-order bias

Elka Rusta, Stefania Salvadori, Roberto Maiolino, Viola Gelli

The properties of the first metal-free stars remain largely unknown, and so far, the only data-driven constraints on their mass distribution (IMF) come from near-field cosmology. Here, we interpret new observations of the C1 and C2 components of Hebe, the HeII emitter near the galaxy GN-z11. Using a locally calibrated model, we robustly confirm the pristine

Roberto Maiolino, Hannah Übler, Michele Perna, Joris Witstok

We report the confirmation of a HeII$\lambda$1640 emitter located at 3 pkpc from the galaxy GN-z11, at z=10.6. The detection, based on JWST NIRSpec-IFU high-resolution spectroscopy, confirms a previous claim based on medium-resolution spectroscopy. The HeII$\lambda$1640 identification is further supported by the independent detection of H$\gamma$ obtained by

Shiliang Zhang, Sabita Maharjan

The efficient management and planning of urban energy systems require integrated three-dimensional (3D) models that accurately represent both consumption nodes and distribution networks. This paper introduces our developed approach and openly released software that automate the generation of digital 3D urban energy model from open data. We synthesize data fr

Ivan Kartáč, Mateusz Lango, Ondřej Dušek

Large Language Models (LLMs) achieve strong performance on many reasoning benchmarks, yet these evaluations typically focus on isolated tasks that differ from real-world usage in task-oriented dialogue (TOD). In this setting, LLMs must perform reasoning inherently while generating text and adhering to instructions on role, format, and style. This mismatch ra

Cen Wan, Alex A. Freitas

Large language models have achieved great success in multiple challenging tasks, and their capacity can be further boosted by the emerging agentic AI techniques. This new computing paradigm has already started revolutionising the traditional scientific discovery pipelines. In this work, we propose a novel agentic AI-based knowledge discovery-oriented virtual

Ravish Gupta, Saket Kumar, Shreeya Sharma, Maulik Dang

Getting a real cybersecurity risk assessment for a small organization is expensive -- a NIST CSF-aligned engagement runs $15,000 on the low end, takes weeks, and depends on practitioners who are genuinely scarce. Most small companies skip it entirely. We built a six-agent AI system where each agent handles one analytical stage: profiling the organization, ma

Seungwon Kim, Gheehyun Nahm, Alison Tatsuoka

In this paper, we construct infinitely many non-isotopic 3-knots in the 5-sphere, each of which has four critical points with respect to the standard height function of the 5-sphere. This contrasts with a theorem of Scharlemann which says that any 2-knot in the 4-sphere with four critical points is unknotted, and also provides infinitely many knotted solid t

Yifan Xu, Qizhang Shen, Vineet Kamat, Carol Menassa

This paper presents KUKAloha, a general, low-cost, and shared-control teleoperation framework designed for construction robot arms. The proposed system employs a leader-follower paradigm in which a lightweight leading arm enables intuitive human guidance for coarse robot motion, while an autonomous perception module based on AprilTag detection performs preci

Wanqi Yuan, Omkar Sharad Mayekar, Connor Pennington, Nianyi Li

Neural Radiance Fields (NeRF) achieve photorealistic novel view synthesis but become costly when high-resolution (HR) rendering is required, as HR outputs demand dense sampling and higher-capacity models. Moreover, naively super-resolving per-view renderings in 2D often breaks multi-view consistency. We propose Generalizable NGP-SR, a 3D-aware super-resoluti

Abtin Molavi, Feras Saad, Aws Albarghouthi

Quantum error correction (QEC) enables reliable computation on noisy hardware by encoding logical information across many physical qubits and periodically measuring parities to detect errors. A decoder is the classical algorithm that uses these measurements to infer which error most likely occurred, so that the system can correct it. The decoder's accuracy-h

T. M. Kamsma, Y. Gu, D. Shi, C. Spitoni

We present an integrated iontronic memristor circuit that reproduces biologically inspired Spike Rate-Dependent Plasticity (SRDP) and functions as a physical nonlinear frequency kernel, which we demonstrate can be used to classify natural auditory data. The fluidic circuit integrates two parallel memristive membranes containing short and long conical memrist

J. L. Velasco, I. Calvo, J. M. García-Regaña

Quasi-axisymmetric stellarators are the stellarator analogue of the axisymmetric tokamak, retaining many of its favorable confinement properties, its compacity and its relative coil simplicity, while avoiding its principal limitation, the need for an inductively driven plasma current. Despite these attractive physics properties, the development of quasi-axis

Max M. Briel, Jeff J. Andrews

The orbital and eccentricity evolution for compact object binaries through gravitational wave emission first derived by Peters and Mathews are used extensively throughout the gravitational wave community for calculating the orbital evolution and merger time of compact binaries. While improved calculations of the binary merger time have been the focus of seve

Wenjing Hong, Zhonghua Rong, Li Wang, Feng Chang

Large Language Models (LLMs) have been widely deployed, especially through free Web-based applications that expose them to diverse user-generated inputs, including those from long-tail distributions such as low-resource languages and encrypted private data. This open-ended exposure increases the risk of jailbreak attacks that undermine model safety alignment

Ruiping Liu, Jingqi Zhang, Junwei Zheng, Yufan Chen

Guide dogs offer independence to Blind and Low-Vision (BLV) individuals, yet their limited availability leaves the vast majority of BLV users without access. Quadruped robotic guide dogs present a promising alternative, but existing systems rely solely on the robot's ground-level sensors for navigation, overlooking a critical class of hazards: obstacles that

N. Plungė, P. Brommer, R. S. Edwards, E. G. Kakouris

This work presents a variational physics-informed deep learning framework for phase-field modelling of brittle crack propagation in anisotropic media. Previous Deep Ritz Method (DRM) approaches have focused on second-order, isotropic phase-field fracture formulations. In contrast, the present work introduces, for the first time within a variational deep lear

Hajar Ajiyel, Anthony J. Genot, Soo Hyeon Kim, Nicolas Schabanel

DNA self-assembly is a well-understood nanotechnology to obtain extremely ordered structures from the nanometer to up to the hundred of microns scale. By contrast, DNA hydrogels rely on the disordered assembly of DNA building blocks to reach macroscopic volumes. However, in order to hold the promise of DNA bulk materials, the sequence designer needs a system

Yuanbo Hou, Vanja Zdravkovic, Marianne Sinka, Yunpeng Li

Mosquito-borne diseases affect more than one billion people each year and cause close to one million deaths. Traditional surveillance methods rely on traps and manual identification that are slow, labor-intensive, and difficult to scale. Audio-based mosquito monitoring offers a non-destructive, lower-cost, and more scalable complement to trap-based surveilla

Jiajie Li, Chenhui Xu, Meihuan Liu, Jinjun Xiong

Conventional fine-tuning on domain-specific datasets can inadvertently alter a model's pretrained multimodal priors, leading to reduced generalization. To address this, we propose Chain-of-Adaptation (CoA), an adaptation framework designed to integrate domain knowledge while maintaining the model's inherent reasoning and perceptual capabilities. CoA introduc

Mohammed Q. Shormani, Yehia A. AlSohbani

We aim to examine the extent to which Large Language Models (LLMs) can 'talk much' about grammar modules, providing evidence from syntax core properties translated by ChatGPT into Arabic. We collected 44 terms from generative syntax previous works, including books and journal articles, as well as from our experience in the field. These terms were translated

Ansgar Lowack

In solid-state batteries, ceramic solid electrolytes are penetrated by dendrites when plating above a critical current density $J_\mathrm{crit}$. A dendrite will propagate by metal deposition at a pre-existing dendrite tip if the mechanical energy required to crack the ceramic open is less than the electrical energy (Joule heating) wasted by forcing the curr

Niclas Pokel, Yiming Zhao, Pehuén Moure, Yingqiang Gao

Personalizing Automatic Speech Recognition (ASR) for non-normative speech remains challenging because data collection is labor-intensive and model training is technically complex. To address these limitations, we propose Adapt4Me, a web-based decentralized environment that operationalizes Bayesian active learning to enable end-to-end personalization without

Moritz Gögl, Christopher Yau

The Joint-Embedding Predictive Architecture (JEPA) is often seen as a non-generative alternative to likelihood-based self-supervised learning, emphasizing prediction in representation space rather than reconstruction in observation space. We argue that the resulting separation from probabilistic generative modeling is largely rhetorical rather than structura

Maaninee Gupta, Kyle J. DeMars

The complex cislunar dynamical environment poses challenges for spacecraft navigation and Space Domain Awareness (SDA) operations, where the knowledge of current and future spacecraft states is essential. Conventional Gaussian-based approaches for SDA degrade under the nonlinearities that manifest in this regime. To accurately model the underlying dynamics a

Pietro Talli, Qi Liao, Alessandro Lieto, Parijat Bhattacharjee

Current network data telemetry pipelines consist of massive streams of fine-grained Key Performance Indicators (KPIs) from multiple distributed sources towards central aggregators, making data storage, transmission, and real-time analysis increasingly unsustainable. This work presents a generative AI (GenAI)-driven sampling and hybrid compression framework t

Krzysztof Kotowski, Ramez Shendy, Jakub Nalepa, Agata Kaczmarek

Forecasting plays a crucial role in modern safety-critical applications, such as space operations. However, the increasing use of deep forecasting models introduces a new security risk of trojan horse attacks, carried out by hiding a backdoor in the training data or directly in the model weights. Once implanted, the backdoor is activated by a specific trigge

Mahyar Karimi, K. S. Thejaswini, Roderick Bloem, Thomas A. Henzinger

In traditional runtime verification, a system is typically observed by a monolithic monitor. Enforcing privacy in such settings is computationally expensive, as it necessitates heavy cryptographic primitives. Therefore, privacy-preserving monitoring remains impractical for real-time applications. In this work, we address this scalability challenge by distrib

Amartya Roy, Rasul Tutunov, Xiaotong Ji, Matthieu Zimmer

LLMs are increasingly used as general-purpose reasoners, but long inputs remain bottlenecked by a fixed context window. Recursive Language Models (RLMs) address this by externalising the prompt and recursively solving subproblems. Yet existing RLMs depend on an open-ended read-eval-print loop (REPL) in which the model generates arbitrary control code, making

David Anderson, Greta Panova, Leonid Petrov

We present computational results on principal specializations $\mathfrak{S}_w(1^n)$ of Schubert polynomials, which count reduced pipe dreams and reduced bumpless pipe dreams (RBPD). We find the first counterexample, at $n=17$, to the Merzon-Smirnov conjecture (arXiv:1410.6857) that the maximum of $\mathfrak{S}_w(1^n)$ over $S_n$ is attained at a layered perm

Seyed Mahdi B. Azad, Jasper Hoffmann, Iman Nematollahi, Hao Zhu

Forward-backward (FB) representations provide a powerful framework for learning the successor representation (SR) in continuous spaces by enforcing a low-rank factorization. However, a fundamental spectral mismatch often exists between the high-rank transition dynamics of continuous environments and the low-rank bottleneck of the FB architecture, making accu

Dimitrios Giannakis, Michael Montgomery

The study of mathematical connections between operator-theoretic formulations of classical dynamics and quantum mechanics began at least as early as the 1930s in work of Koopman and von Neumann and was developed in later decades by many authors, often independently, into a framework now broadly known as Koopman-von Neumann representation of classical dynamic

Tal Haklay, Nikhil Prakash, Sana Pandey, Antonio Torralba

Automated interpretability systems aim to reduce the need for human labor and scale analysis to increasingly large models and diverse tasks. Recent efforts toward this goal leverage large language models (LLMs) at increasing levels of autonomy, ranging from fixed one-shot workflows to fully autonomous interpretability agents. This shift creates a correspondi

Yuming Feng, Christy Yang

Direct Preference Optimization (DPO) is widely used after supervised fine-tuning (SFT) to align language models, yet empirical behavior under small backbones and modest data is under-specified. We systematically compare SFT-only, DPO-only, and staged SFT-to-DPO training alongside full fine-tuning (FFT) versus LoRA on a GPT-2-scale decoder, evaluating paraphr

Jan Philipp Klinger, Reinhold Kaiser, Owe Philipsen, Jonas Schaible

When the number of massless fermions exceeds a critical value $N_f^*$, QCD enters the conformal window and becomes chirally symmetric already in the vacuum. Determining $N_f^*$ from lattice simulations is challenging, since calculations are performed at finite lattice spacing, quark mass, and temporal lattice size, where both a thermal transition and an unph

Ruijie Gao, Jun Yang, Yang Gao, Jingdong Zhang

RS Canum Venaticorum (RS CVn) close binaries, characterized by tidal locking, rapid rotations, and strong magnetic fields, are ideal laboratories for high-resolution radio observations to probe emission processes, magnetic field configurations, and interaction activity. Despite their importance, only a few RS CVn sources have been explored by polarimetric ob

Bret Benesh

Many recent proposals for reducing tanking in draft lotteries share a common structure: losses improve draft position early in the season while wins improve draft position later. While such systems improve late-season incentives, they retain a predictable pivot point that tanking teams can exploit strategically. This paper proposes a simple modification that

Gabriel Luz Almeida, Alan Müller, Stefano Foffa, Riccardo Sturani

We derive the effective action governing the dynamics of a compact binary system when gravitational radiation is emitted by any mass or current multipole, scattered by the quasi-static field associated with the binary's angular momentum, and then reabsorbed. Among such angular momentum failed-tail processes, the ones involving multipole moments up to mass an

Julian Fernandez Bonder, Martin Guzman, Juan F. Spedaletti

In this paper we analyze an eigenvalue problem associated to fractional operators of the form \[ L_a^s u(x)=2 \text{p.v.}\int_{\mathbb{R}^n}a(x,y,D^su(x,y))\,\frac{dy}{|x-y|^{n+s}},\] which represents a generalization model for nonlocal, nonstandard growth diffusion problems. We study this problem in the context of the fractional Orlicz Sobolev spaces withou

Antonio De Santis, Marco Balduini, Matteo Belcao, Andrea Proia

Large manufacturing companies face challenges in information retrieval due to data silos maintained by different departments, leading to inconsistencies and misalignment across databases. This paper presents an experience in integrating and retrieving qualification data for electronic components used in satellite board design. Due to data silos, designers ca

Alexandre Bailleul, Mounir Hayani, Théo Untrau

We study generalized Skewes' numbers, which are the locations of the first sign change between two comparable prime counting functions. In the context of the race between quadratic residues and quadratic nonresidues, we construct sequences of highly composite moduli $q$ such that those Skewes' numbers grow very rapidly in some sense. This disproves unconditi

Luca Ambrogioni

Diffusion models generate structure by progressively transforming noise into data, yet the mechanisms underlying this transition remain poorly understood. In this work, we show that pattern formation in trained diffusion models can be explained as an out-of-equilibrium phase transition driven by instabilities in the denoising dynamics. We develop a theoretic

Baptiste Debecker, Eduardo Serrano-Ensástiga, Thierry Bastin, François Damanet

We prove a no-go theorem for symmetry-based dissipative engineering of collective-spin steady states: in spin-only Lindblad dynamics with jump operators linear in the collective-spin operators, any unique steady state exhibiting at least $\mathbb{Z}_2 \times \mathbb{Z}_2$ symmetry is necessarily the maximally mixed state. We then show that bath memory lifts

M. L. Bissell, M. Jankowski, A. Antušek, N. Azaryan

To date, the magnetic structure of nuclei has been poorly constrained, with limited information on its spatial distribution. In this work, we address the composition and distribution of nuclear magnetization in a precision study of short-lived $^{47}$K. We measure the Larmor frequency with part-per-million precision using liquid-state $β$-detected nuclear ma

Thierno Mamadou Baldé, Vuk Milisic, Steffen Plunder

Comparison principles for Volterra equations play a role analogous to maximum principles in PDEs: they provide positivity and stability information on the solution and allow one to control the output of bounded inputs. In the continuous setting, such results often rely on Laplace-transform or spectral methods (see Gripenberg, Londen, and Staffans, Volterra I

ATLAS Collaboration

This article details two studies of Higgs boson properties using the vector-boson fusion production mode and the $\gamma\gamma jj$ final state. Both efforts are based on a data sample corresponding to 164 fb$^{-1}$ of $\sqrt{s}=13.6$ TeV proton--proton collisions recorded by the ATLAS experiment at the Large Hadron Collider. The first study employs matrix el

Shiqi Gao, Kang Fu, Zitong Xu, Huiyu Duan

Current no-reference image quality assessment (NR-IQA) models for enhanced images often struggle to generalize, as they tend to overfit to the distinct patterns of specific enhancement algorithms rather than evaluating genuine perceptual quality. To address this issue, we propose a preference-guided debiasing framework for no-reference enhancement image qual

Wen-Zhe Yan, Lan-Tian Feng, Zhibo Hou, Yuan-Yuan Zhao

Programmable photonic quantum processors face a critical challenge: despite significant advances in quantum state preparation and manipulation, measurements remain limited to projective techniques. Here, we demonstrate a programmable measurement processor that overcomes this limitation by enabling arbitrary quantum measurements within a scalable circuit fram

Edoardo Calvello, Elizabeth Carlson, Nikola Kovachki, Michael N. Manta

Machine learning has opened new frontiers in purely data-driven algorithms for data assimilation in, and for forecasting of, dynamical systems; the resulting methods are showing some promise. However, in contrast to model-driven algorithms, analysis of these data-driven methods is poorly developed. In this paper we address this issue, developing a theory to

Augustus Brown, Daniele Dorigoni, Congkao Wen

We study the giant graviton integrated correlator in SU$(N)$ $\mathcal{N}=4$ super Yang-Mills at finite complexified coupling $\tau$. Despite the formidable complexity arising from the heavy nature of the operators considered, the large-$N$ expansion simplifies dramatically and exhibits manifest modular invariance. At each order in $1/N$, the expansion coeff

Josh Miles, Sohom Bhattacharya

We investigate the problem of statistical inference for logistic regression with high-dimensional covariates in settings where dependence among individuals is induced by an underlying Markov random field. Going beyond the pairwise interaction models such as the Ising model, we consider a framework to accommodate more general tensor structures that capture hi

Levin Maier

In this paper, we introduce \emph{$\ell^p$-information geometry}, an infinite-dimensional framework that shares key features with the geometry of the space of probability densities \( \mathrm{Dens}(M) \) on a closed manifold, while also incorporating aspects of measure-valued information geometry. We define the \emph{$\ell^2$-probability simplex} with a nonc

S. Mazzolani, I. Mattei, L. Servoli

The Microstrip Silicon Detector (MSD) is one of the subsystems of the FragmentatiOn Of Target (FOOT) experiment whose goal is to measure double differential nuclear fragmentation cross sections for applications in particle therapy and radioprotection in space. The MSD, composed of six 150 $\mu$m-thick silicon sensors arranged in three X-Y measuring planes, i

Yu Wang, Olcay Türk, Angela Grimminger, Hendrik Buschmeier

We investigate how verbal and nonverbal linguistic features, exhibited by speakers and listeners in dialogue, can contribute to predicting the listener's state of understanding in explanatory interactions on a moment-by-moment basis. Specifically, we examine three linguistic cues related to cognitive load and hypothesised to correlate with listener understan