October 2025 arXiv papers — page 21

Jonas Spenger, Kolya Krafeld, Ruben van Gemeren, Philipp Haller

Scaling global aggregations is a challenge for exactly-once stream processing systems. Current systems implement these either by computing the aggregation in a single task instance, or by static aggregation trees, which limits scalability and may become a bottleneck. Moreover, the end-to-end latency is determined by the slowest path in the tree, and failures

David S Maxwell, Michael Darkoh, Sidharth R Samudrala, Caroline Chung

With the recent growth of Deep Learning for AI, there is a need for tools to meet the demand of data flowing into those models. In some cases, source data may exist in multiple formats, and therefore the source data must be investigated and properly engineered for a Machine Learning model or graph database. Overhead and lack of scalability with existing work

Bohan Wu, Paul de La Sayette, Li Fei-Fei, Roberto Martín-Martín

The ability to use random objects as tools in a generalizable manner is a missing piece in robots' intelligence today to boost their versatility and problem-solving capabilities. State-of-the-art robotic tool usage methods focused on procedurally generating or crowd-sourcing datasets of tools for a task to learn how to grasp and manipulate them for that task

Samet Demir, Zafer Dogan

Pretrained Transformers demonstrate remarkable in-context learning (ICL) capabilities, enabling them to adapt to new tasks from demonstrations without parameter updates. However, theoretical studies often rely on simplified architectures (e.g., omitting MLPs), plain data models (e.g., linear regression with isotropic inputs), and single-source training, limi

James V. Roggeveen, Michael P. Brenner

Solving inverse and optimization problems over solutions of nonlinear partial differential equations (PDEs) on complex spatial domains is a long-standing challenge. Here we introduce a method that parameterizes the solution using spectral bases on arbitrary spatiotemporal domains, whereby the basis is defined on a hyperrectangle containing the true domain. W

Diego Cuevas, Mikel Gutiérrez, Jesús Ibáñez, Ignacio Santamaria

This paper proposes a noncoherent low probability of detection (LPD) communication system based on direct sequence spread spectrum (DSSS) and Grassmannian signaling. Grassmannian constellations enhance covertness because they tend to follow a noise-like distribution. Simulations showed that Grassmannian signaling provides competitive bit error rates (BER) at

Kean Chen

In this paper, we present an inverse-free pure quantum state estimation protocol that achieves Heisenberg scaling. Specifically, let $\mathcal{H}\cong \mathbb{C}^d$ be a $d$-dimensional Hilbert space with an orthonormal basis $\{|1\rangle,\ldots,|d\rangle\}$ and $U$ be an unknown unitary on $\mathcal{H}$. Our protocol estimates $U|d\rangle$ to within trace d

Boris Y. Rubinstein

Recently we introduced a new class of relations for Bernoulli symmetric polynomials. This manuscript shows that these relations are valid for arbitrary homogeneous symmetric polynomial. Analysis of these relations leads to the discovery of a new type of nonlinear relations for the Bernoulli numbers.

Ayush Subedi, Anthony Torres, Jeff Ganley, Ujjwal Dhakal

ZBLAN glass is a promising material for infrared optical fibers due to its wide transmission window and low theoretical attenuation. However, its strong tendency to crystallize during processing limits optical performance. While microgravity environments have been shown to suppress crystallization, the role of mechanical vibration under normal gravity condit

Hesam Arabzadeh, Brad Lee Holian

Recent laboratory experiments suggest an intrinsic asymmetry between heating and cooling, with heating occurring more efficiently. Two decades earlier, molecular dynamics (MD) simulations had examined a related setup - heating one side of a computational cell while cooling the other via distinct thermostats. We revisit those calculations, recapitulating the

Charlie Harrison, Pasin Manurangsi

Zero-concentrated differential privacy (zCDP) is a variant of differential privacy (DP) that is widely used partly thanks to its nice composition property. While a tight conversion from $\epsilon$-DP to zCDP exists for the worst-case mechanism, many common algorithms satisfy stronger guarantees. In this work, we derive tight zCDP characterizations for severa

Christodoulos Benetatos, Yongyi Zang, Randal Leistikow

State-of-the-art vocal separation models like Mel-Band-Roformer rely on full temporal self-attention mechanisms, where each temporal frame interacts with every other frames. This incurs heavy computational costs that scales quadratically with input audio length, motivating chunking and windowing approaches. Through analysis of a pre-trained vocal separation

Shannon Zejiang Shen, Valerie Chen, Ken Gu, Alexis Ross

Current evaluations of agents remain centered around one-shot task completion, failing to account for the inherently iterative and collaborative nature of many real-world problems, where human goals are often underspecified and evolve. We argue for a shift from building and assessing task completion agents to developing collaborative agents, assessed not onl

Fabio Centofanti

In modern industrial settings, advanced acquisition systems allow for the collection of data in the form of profiles, that is, as functional relationships linking responses to explanatory variables. In this context, statistical process monitoring (SPM) aims to assess the stability of profiles over time in order to detect unexpected behavior. This review focu

Zhi-Kai Chen, Jun-Peng Jiang, Han-Jia Ye, De-Chuan Zhan

Autoregressive (AR) image generation models are capable of producing high-fidelity images but often suffer from slow inference due to their inherently sequential, token-by-token decoding process. Speculative decoding, which employs a lightweight draft model to approximate the output of a larger AR model, has shown promise in accelerating text generation with

Joseph O. Indekeu, Kenichiro Koga

Fluid three-phase equilibria, with phases $\alpha, \beta, \gamma$, are studied close to a tricritical point, analytically and numerically, in a mean-field density-functional theory with two densities. Employing Griffiths' scaling for the densities, the interfacial tensions of the wet and nonwet interfaces are analysed. The mean-field critical exponent is obt

Tobin South, Subramanya Nagabhushanaradhya, Ayesha Dissanayaka, Sarah Cecchetti

The rapid rise of AI agents presents urgent challenges in authentication, authorization, and identity management. Current agent-centric protocols (like MCP) highlight the demand for clarified best practices in authentication and authorization. Looking ahead, ambitions for highly autonomous agents raise complex long-term questions regarding scalable access co

Shreya Meel, Sennur Ulukus

We revisit the problem of symmetric private information retrieval (SPIR) in settings where the database replication is modeled by a simple graph. Here, each vertex corresponds to a server, and a message is replicated on two servers if and only if there is an edge between them. To satisfy the requirement of database privacy, we let all the servers share some

Maxime Dixmerias, Giuseppe Del Vecchio Del Vecchio, Cyprien Daix, Joris Verstraten

The pursuit of universal governing principles is a foundational endeavor in physics, driving breakthroughs from thermodynamics to general relativity and quantum mechanics. In 1951, Wigner introduced the concept of a statistical description of energy levels of heavy atoms, which led to the rise of Random Matrix Theory (RMT) in physics. The theory successfully

Emily Heath, Grace McCourt, Alex Parker, Coy Schwieder

We define the $r\textit{-Kneser Ramsey number}$ $R^{\textrm{KG}}_{r}(s, t)$ as the minimum integer $n$ such that every red/blue edge-coloring of the Kneser graph $\textrm{KG}(n,r)$ contains a red $s$-clique or a blue $t$-clique. We obtain general bounds on the numbers $R^{\textrm{KG}}_{r}(s, t)$, and make progress on two related Ramsey-type problems, one rai

Livia Ferro, Ross Glew, Tomasz Lukowski, Jonah Stalknecht

In this paper we investigate the loop-level geometry of ABJM theory from the perspective of lightcone geometries in dual space. This geometry admits a natural fibration, where one of the loop variables can be naturally interpreted as living in a fiber for each fixed point of a lower-loop geometry. When varying the latter, this leads us to the definition of $

Aakriti Shah, Yifan Hu, Thai Le

Unlearning in large language models (LLMs) is usually evaluated as whether an "unlearned" fact can be recovered. We instead ask whether a fact's structural entanglement with the rest of a model's knowledge predicts whether it leaks after unlearning, whether this relationship changes systematically, and whether it is causal. Across varied unle

Felipe A. Barros, Italo Salas, Enkeleida Lushi, Francisca Guzman-Lastra

Slicks are thin viscous films that can be found at the air--water interface of water bodies such as lakes, rivers and oceans. These micro-layers are enriched in surfactants, organic matter, and microorganisms, and exhibit steep physical and chemical gradients across only tens to hundreds of micrometers. In such geometrically confined environments, the hydrod

Shirin Chehelgami, Joe LoVetri, Vahab Khoshdel

A conditional latent-diffusion based framework for solving the electromagnetic inverse scattering problem associated with microwave imaging is introduced. This generative machine-learning model explicitly mirrors the non-uniqueness of the ill-posed inverse problem. Unlike existing inverse solvers utilizing deterministic machine learning techniques that produ

Ciera McFarland, Margaret McGuinness

Soft, vine-inspired growing robots that move by eversion are highly mobile in confined environments, but, when faced with gaps in the environment, they may collapse under their own weight while navigating a desired path. In this work, we present a comprehensive collapse model that can predict the collapse length of steered robots in any shape using true shap

Junlong Li, Wenshuo Zhao, Jian Zhao, Weihao Zeng

Real-world language agents must handle complex, multi-step workflows across diverse Apps. For instance, an agent may manage emails by coordinating with calendars and file systems, or monitor a production database to detect anomalies and generate reports following an operating manual. However, existing language agent benchmarks often focus on narrow domains o

Chen-Wei Chang, Yu-Chieh Cheng, Yun-En Tsai, Fanglan Chen

Access to metro systems plays a critical role in shaping urban housing markets by enhancing neighborhood accessibility and driving property demand. We present RailEstate, a novel web based system that integrates spatial analytics, natural language interfaces, and interactive forecasting to analyze how proximity to metro stations influences residential proper

Vanya Arikutharam, Arkadiy Ukolov

Retrieval-Augmented Generation allows LLMs to access external knowledge, reducing hallucinations and ageing-data issues. However, it treats retrieved chunks independently and struggles with multi-hop or relational reasoning, especially across documents. Knowledge graphs enhance this by capturing the relationships between entities using triplets, enabling str

Tobias König, Jonas W. Peteranderl

In the conformal class of the standard metric on the $3$-sphere, we prove a quantitative refinement of the Andrews-De Lellis-Topping inequality in terms of a two-term distance to the set of minimizing conformal factors. This inequality is itself a stability result for the well-known Schur lemma and is therefore referred to as almost-Schur lemma. Hence, our s

Stewart Slocum, Asher Parker-Sartori, Dylan Hadfield-Menell

The ability of LLMs to represent diverse perspectives is critical as they increasingly impact society. However, recent studies reveal that alignment algorithms such as RLHF and DPO significantly reduce the diversity of LLM outputs. Not only do aligned LLMs generate text with repetitive structure and word choice, they also approach problems in more uniform wa

Louis A. Angeloni, Ir-Jene Shan, J. H. Leach, W. Andreas Schroeder

Measurements of the spectral emission properties of an iron-doped a beta-Ga2O3(010) photocathode at 300 K reveal the presence of an ultracold contribution to the total electron beam emission with a 6 meV mean transverse energy (MTE) in the 3.5-4.4 eV photon energy range (282-354 nm). This extreme sub-thermal photoemission signal is consistent with direct emi

Yi W. Ren, Akio K. Inoue, Javier Álvarez-Márquez, Takuya Hashimoto

We present analysis of JWST NIRCam and NIRSpec observations of the galaxy SXDF-NB1006-2 at $z = 7.212$, as part of the Reionization and the ISM/Stellar Origins with JWST and ALMA (RIOJA) project. We derive the physical properties by conducting spectral energy distribution (SED) fitting, revealing that our target is a young (age $\sim2$ Myr) starburst galaxy

D. Q. Adams, C. Alduino, K. Alfonso, A. Armatol

The Cryogenic Underground Observatory for Rare Events (CUORE) experiment set the most stringent limit on the neutrinoless double-beta ($0\nu\beta\beta$) decay half-life of $^{130}$Te with 2 ton yr TeO$_2$ analyzed exposure. In addition to $0\nu\beta\beta$ decay, the CUORE detector -- a ton-scale array of nearly 1000 cryogenic calorimeters operating at $\sim$

Nikolaos Koukoulekidis, Iman Marvian

We study the interplay of symmetries and Gaussianity in bosonic systems, under closed and open dynamics, and develop a resource theory of Gaussian asymmetry. Specifically, we focus on Gaussian symmetry-respecting (covariant) operations, which serve as the free operations in this framework. We prove that any such operation can be realized via Gaussian Hamilto

Jamie Mahowald, Benjamin Charles Germain Lee

Multimodal approaches have shown great promise for searching and navigating digital collections held by libraries, archives, and museums. In this paper, we introduce map-RAS: a retrieval-augmented search system for historic maps. In addition to introducing our framework, we detail our publicly-hosted demo for searching 101,233 map images held by the Library

Jarmo Mäkelä

We formulate an argument, based on the use of the action-angle variables and the Bohr-Sommerfeld quantization rule, to the effect that if there exists the smallest possible non-zero area, there also exists, for massive particles, the largest possible observable speed, which is a bit less than the speed of light.

Abha Jha, Abel Salinas, Fred Morstatter

The rise of Large Language Models (LLMs) offers transformative potential for interpreting complex legal frameworks, such as Title 18 Section 175 of the US Code, which governs biological weapons. These systems hold promise for advancing legal analysis and compliance monitoring in sensitive domains. However, this capability comes with a troubling contradiction

David Bate, Pietro Wald

We construct a family of purely PI unrectifiable Lipschitz differentiability spaces and investigate the possible of Banach spaces targets for which Lipschitz differentiability holds. We provide a general investigation into the geometry of \emph{shortcut} metric spaces and characterise when such spaces are PI rectifiable, and when they are $Y$-LDS, for a give

Dan Barry, Davoud Shariat Panah, Alessandro Ragano, Jan Skoglund

We present Binaspect, an open-source Python library for binaural audio analysis, visualization, and feature generation. Binaspect generates interpretable "azimuth maps" by calculating modified interaural time and level difference spectrograms, and clustering those time-frequency (TF) bins into stable time-azimuth histogram representations. This allows multip

Boshi An, Chenyu Yang, Robert Katzschmann

We adapt a pre-trained Vision-Language-Action (VLA) model (Open-VLA) for dexterous human-robot collaboration with minimal language prompting. Our approach adds (i) FiLM conditioning to visual backbones for task-aware perception, (ii) an auxiliary intent head that predicts collaborator hand pose and target cues, and (iii) action-space post-processing that pre

Jack Storror Carter, Cesare Molinari

The partial correlation graphical LASSO (PCGLASSO) is a penalised likelihood method for Gaussian graphical models which provides scale invariant sparse estimation of the precision matrix and improves upon the popular graphical LASSO method. However, the PCGLASSO suffers from computational challenges due to the non-convexity of its associated optimisation pro

Priyavrat Deshpande, Amit Roy, Rutuja Sawant

We introduce a new family of pure simplicial complexes, called the $r$-co-connected complex of $G$ with respect to $A$, $\Sigma_r(A,G)$, where $r\geq 1$ is a natural number, $G$ is a simple graph, and $A$ is a subset of vertices. Interestingly, when $A$ is empty, this complex is precisely the Alexander dual of the $r$-independence complex of $G$. We focus on

Dario Sauro

The unique off-shell fermionic gauge invariance of a vector-spinor field theory is found, and the invariant action is derived. The latter is Weyl invariant in any dimension in the massless limit, and it coincides with the singular point of the one-parameter family of Rarita-Schwinger Lagrangians, in agreement with previous findings in flat space. Pure gauge

Hans Jockers, Pyry Kuusela, Maik Sarve

Under certain assumptions, we show that unitary rational $\mathcal{N}=(2,2)$ conformal field theories together with a certain generating set of Cardy boundary states in the associated boundary conformal field theories give rise to rational Hodge structures of complex multiplication type. We argue that these rational Hodge structures for such rational conform

Loan Renaud, Tomasz Poreba, Simone Di Cataldo, Alasdair Nicholls

Hydrogen hydrates (HH) are a unique class of materials composed of hydrogen molecules confined within crystalline water frameworks. Among their multiple phases, the filled ice structures, particularly the cubic C2 phase, exhibit exceptionally strong host-guest interactions due to ultra-short H2-H2O distances and a 1:1 stoichiometry leading to two interpenetr

Sungho Koh, SeungJu Cha, Hyunwoo Oh, Kwanyoung Lee

Text-to-image diffusion models often exhibit degraded performance when generating images beyond their training resolution. Recent training-free methods can mitigate this limitation, but they often require substantial computation or are incompatible with recent Diffusion Transformer models. In this paper, we propose ScaleDiff, a model-agnostic and highly effi

Andrés N. Salcedo, Eduardo Rozo, Hao-Yi Wu, David H. Weinberg

We present a simulation-based forward-modeling framework for cosmological inference from optical galaxy-cluster samples, and apply it to the abundance and weak-lensing signals of DES-Y1 redMaPPer clusters. The model embeds cosmology-dependent optical selection using a counts-in-cylinders approach, while also accounting for cluster miscentering and baryonic f

Elif Ebru Ohri, Qi Liao, Anastasios Giovanidis, Francesca Fossati

As augmented and virtual reality evolve, achieving seamless synchronization between physical and digital realms remains a critical challenge, especially for real-time applications where delays affect the user experience. This paper presents MetaLore, a Deep Reinforcement Learning (DRL) based framework for joint communication and computational resource alloca

Claudio Bonanno, Andrea Bulgarelli, Elia Cellini, Alessandro Nada

We develop a methodology based on out-of-equilibrium simulations to mitigate topological freezing when approaching the continuum limit of lattice gauge theories. We reduce the autocorrelation of the topological charge employing open boundary conditions, while removing exactly their unphysical effects using a non-equilibrium Monte Carlo approach in which peri

Chenxi Zhao, Dong Wu, Weiyi Kong, Oskar J. Haidn

Accurate simulation of blood flow in deformable vessels is critical in cardiovascular research for understanding disease progression and informing clinical decision-making. However, due to the thin-walled nature of arteries, traditional smoothed particle hydrodynamics (SPH) approaches based on full-dimensional volume modeling often require extremely fine par

Feng Wang, Xiangyu Hu

To effectively handle flows characterized by strong backflow and multiple open boundaries within particle-based frameworks, this study introduces three enhancements to improve the consistency, independence, and accuracy of the buffer-based open boundary condition in SPHinXsys. First, to improve the buffer consistency, the continuum hypothesis is introduced t

Saeed AlMarri, Kristof Juhasz, Mathieu Ravaut, Gautier Marti

Large Language Models (LLMs) are increasingly explored as flexible alternatives to classical machine learning models for classification tasks through zero-shot prompting. However, their suitability for structured tabular data remains underexplored, especially in high-stakes financial applications such as financial risk assessment. This study conducts a syste

Charles Hyde, Mitch Kerver, Christos Tsolakis, Polykarpos Thomadakis

We describe an iterative algorithm to construct an unstructured tessellation of simplices (irregular tetrahedra in 3-dimensions) to approximate an arbitrary function to a desired precision by interpolation. The method is applied to the generation of Compton Form Factors for simulation and analysis of nuclear femtography, as enabled by high energy exclusive p

Georg C. Hofstätter, Jonas Knoerr

We provide a new proof of Alesker's Irreducibility Theorem. We first introduce a new localization technique for polynomial valuations on convex bodies, which we use to independently prove that smooth and translation invariant valuations are representable by integration with respect to the normal cycle. This allows us to reduce the statement to a correspondin

Abha Jha, J. Hunter Priniski, Carolyn Steinle, Fred Morstatter

Networked environments shape how information embedded in narratives influences individual and group beliefs and behavior. This raises key questions about how group communication around narrative media impacts belief formation and how such mechanisms contribute to the emergence of consensus or polarization. Language data from generative agents offer insight i

Junyu Luo, Bohan Wu, Xiao Luo, Zhiping Xiao

Post-training of Large Language Models (LLMs) is crucial for unlocking their task generalization potential and domain-specific capabilities. However, the current LLM post-training paradigm faces significant data challenges, including the high costs of manual annotation and diminishing marginal returns on data scales. Therefore, achieving data-efficient post-

Edgard Moreira Minete, Mathis Immertreu, Fabian Teichmann, Sebastian Müller

We formulate mold filling in metal casting as a 2D neural operator learning problem that maps geometry and boundary data on an unstructured mesh to time resolved flow quantities, replacing expensive transient CFD. In the proposed method, a graph based encoder aggregates local neighborhood information on the input mesh and encodes geometry and boundary data,

Azza Bouleimen, Giordano De Marzo, Taehee Kim, Nicol`o Pagan

Large Language Models (LLMs) offer new avenues to simulate online communities and social media. Potential applications range from testing the design of content recommendation algorithms to estimating the effects of content policies and interventions. However, the validity of using LLMs to simulate conversations between various users remains largely untested.

Yesmine Abdennadher, Eleonora Cicciarella, Michele Rossi

Spike-based temporal messaging enables SNNs to efficiently process both purely temporal and spatio-temporal time-series or event-driven data. Combining SNNs with Gated Recurrent Units (GRUs), a variant of recurrent neural networks, gives rise to a robust framework for sequential data processing; however, traditional RNNs often lose local details when handlin

Emerson J. Hollar, Esmat Farzana

We report over 3 kV breakdown voltage and ultra-low leakage (011) \b{eta}-Ga2O3 power devices utilizing Schottky barrier engineering and high-permittivity (\k{appa}) dielectric (ZrO2) field plate. The (011) orientation of \b{eta}-Ga2O3 enabled low background doping and thick drift layers which are promising to support kV-class vertical \b{eta}-Ga2O3 power sw

Jiayi Kuang, Yinghui Li, Xin Zhang, Yangning Li

Large language model-based agents show promise for software engineering, but environment configuration remains a bottleneck due to heavy manual effort and scarce large-scale, high-quality datasets. Existing benchmarks assess only end-to-end build/test success, obscuring where and why agents succeed or fail. We introduce the Environment Configuration Diagnosi

Tim Strnad, Blaž Bertalanič, Carolina Fortuna

As machine learning underpins more critical applications, the value of a reported result depends on whether it can be compared and repeated. In practice, this remains difficult: research groups often assemble their own tools for configuration, execution, versioning, and evaluation, while also repeating the domain-specific work such as dataset preparation and

Pierre Béchaz, Giuseppe Fanizza, Giovanni Marozzi, Matheus R. Medeiros Silva

We develop a second-order cosmological perturbation theory on a background geometry expressed in terms of light-cone coordinates, extending the first-order analyses available in the literature. In particular, we investigate the gauge transformations of second-order perturbative quantities on the light-cone and establish their connection with standard perturb

Tibor Jordán, Xuemei Liu, Soma Villányi

We study sufficient conditions for the generic rigidity of a graph $G$ expressed in terms of (i) its minimum degree $\delta(G)$, or (ii) the parameter $\eta(G)=\min_{uv\notin E}(\deg(u)+\deg(v))$. For each case, we seek the smallest integers $f(n,d)$ (resp.\ $g(n,d)$) such that every $n$-vertex graph $G$ with $\delta(G)\geq f(n,d)$ (resp.\ $\eta(G)\geq g(n,d

Subarsha Banerjee

The Szeged index of a graph is an invariant with several applications in chemistry. The power graph of a finite group $G$ is a graph having vertex set as $G$ in which two vertices $u$ and $v$ are adjacent if $v=u^m$ or $u=v^n$ for some $m,n\in \mathbb{N}$. In this paper, we first obtain a formula for the Szeged index of the generalized join of graphs. As an

Boris Bukh, Jun Gao, Xizhi Liu, Oleg Pikhurko

Determining the minimum density of a covering of $\mathbb{R}^{n}$ by Euclidean unit balls as $n\to\infty$ is a major open problem, with the best known results being the lower bound of $\left(\mathrm{e}^{-3/2}+o(1)\right)n$ by Coxeter, Few and Rogers [Mathematika 6, 1959] and the upper bound of $\left(1/2+o(1) \right)n \ln n$ by Dumer [Discrete Comput. Geom.

Yujun He, Hangdong Zhao, Simon Frisk, Yifei Yang

Minimizing intermediate results is critical for efficient multi-join query processing. Although the seminal Yannakakis algorithm offers strong guarantees for acyclic queries, cyclic queries remain an open challenge. In this paper, we propose SplitJoin, a framework that introduces split as a first-class query operator. By partitioning input tables into heavy

Alessandro Lucchetti, Francesco Cadini, Marco Giglio, Luca Lomazzi

Graph Neural Networks (GNNs) have recently been explored as surrogate models for numerical simulations. While their applications in computational fluid dynamics have been investigated, little attention has been given to structural problems, especially for dynamic cases. To address this gap, we introduce the Graph Network-based Structural Simulator (GNSS), a

Jiani Zheng, Zhiyang Teng, Kunpeng Qiu, Xiangtai Li

Unified Vision-Language Models (UVLMs) perform both understanding and generation within a single architecture. Since these models rely on heterogeneous data and supervision, balancing both generation and understanding in reinforcement learning (RL) is challenging. To address this challenge, we propose PairUni, a unified framework that reorganizes data into u

Enrica Barrilli, Bernard Mourrain, Daniele Taufer

This article addresses the Generalized Additive Decomposition (GAD) of symmetric tensors, that is, degree-$d$ forms $f \in \mathcal{S}_d$. From a geometric perspective, a GAD corresponds to representing a point on a secant of osculating varieties to the Veronese variety, providing a compact and structured description of a tensor that captures its intrinsic a

Takato Tokuno

A recent asteroseismic analysis suggests that Kepler-56 -- a planet-hosting red giant -- exhibits a unique spin structure: (1) the spin axes of the core and envelope are misaligned; and (2) the envelope rotates approximately an order of magnitude faster than typical red giants. In this paper, we investigate a feasible scenario to reproduce this spin structur

Federica Tonti, Ricardo Vinuesa

Unmanned Aerial Vehicles (UAVs) are increasingly populating urban areas for delivery and surveillance purposes. In this work, we develop an optimal navigation strategy based on Deep Reinforcement Learning. The environment is represented by a three-dimensional high-fidelity simulation of an urban flow, characterized by turbulence and recirculation zones. The

S. Umayal, Vaidehi S. Paliya, D. J. Saikia, C. S. Stalin

Powerful, large-scale relativistic jets are usually associated with massive, old elliptical galaxies. This paradigm has recently been challenged by the identification of narrow-line Seyfert 1 (NLSy1) galaxies, thought to be young active galactic nuclei with low-mass black holes, harboring relativistic jets. Among them, sources hosting $>$100 kpc radio jets a

Hasan Akgul, Mari Eplik, Javier Rojas, Aina Binti Abdullah

ZK-SenseLM is a secure and auditable wireless sensing framework that pairs a large-model encoder for Wi-Fi channel state information (and optionally mmWave radar or RFID) with a policy-grounded decision layer and end-to-end zero-knowledge proofs of inference. The encoder uses masked spectral pretraining with phase-consistency regularization, plus a light cro

Teena tom Dieck, Lukas Brand, Sebastian Lotter, Kathrin Castiglione

Molecular communication (MC) enables information exchange through the transmission of signaling molecules (SMs) and holds promise for many innovative applications. However, most existing works in MC rely on simplified transmitter (TX) models that do not account for the physical and biochemical limitations of realistic biological hardware and environments. Th

Kesha Sorathia, Cono Di Paola, Gabriel Greene-Diniz, Carlo A. Gaggioli

Due to the wide range of technical applications of actinide elements, a thorough understanding of their electronic structure could complement technological improvements in many different areas. Quantum computing could greatly aid in this understanding, as it can potentially provide exponential speedups over classical approaches, thereby offering insights int

Elia Torre, Michele Viscione, Lucas Pompe, Benjamin F Grewe

Recurrent neural networks (RNNs) provide a powerful approach in neuroscience to infer latent dynamics in neural populations and to generate hypotheses about the neural computations underlying behavior. However, past work has focused on relatively simple, input-driven, and largely deterministic behaviors - little is known about the mechanisms that would allow

Kyle G Miller, Tomas E Gutierrez, Archis S Joglekar, Amanda Elliott

Optical techniques for spatiotemporal control can produce laser pulses with custom amplitude, phase, or polarization structure. In nonlinear optics and plasma physics, the use of structured pulses typically follows a forward design approach, in which the efficacy of a known structure is analyzed for a particular application. Inverse approaches, in contrast,

Tarun Kumar Chawdhury, Jon D. Duke

Electronic Health Records (EHR) store clinical documentation as base64 encoded attachments in FHIR DocumentReference resources, which makes semantic question answering difficult. Traditional vector database methods often miss nuanced clinical relationships. The Clinical Entity Augmented Retrieval (CLEAR) method, introduced by Lopez et al. 2025, uses entity a

Swagat S. Mishra, Athul K. Soman

The expansion history of the Universe prior to Big Bang Nucleosynthesis (BBN) remains largely unconstrained. The high-energy post-inflationary era may involve multiple distinct epochs, each characterized by a different equation of state (EoS). A key prediction of inflation is the generation of tensor perturbations that later manifest as a stochastic backgrou

Hongjin Du, Rahul Rane, Weijie Xia, Pedro P. Vergara

The increasing integration of renewable energy, particularly offshore wind, introduces significant uncertainty into hybrid AC-HVDC systems due to forecast errors and power fluctuations. Conventional control strategies typically rely on fixed setpoints and neglect frequency deviations, which can compromise system stability under rapid renewable variations. To

Phuc Tran, Nisheeth K. Vishnoi, Van H. Vu

A central challenge in machine learning is to understand how noise or measurement errors affect low-rank approximations, particularly in the spectral norm. This question is especially important in differentially private low-rank approximation, where one aims to preserve the top-$p$ structure of a data-derived matrix while ensuring privacy. Prior work often a

Ethan Obie Romero-Severson, Tara Harvey, Nick Generous, Phillip M. Mach

Understanding how AI systems are used by people in real situations that mirror aspects of both legitimate and illegitimate use is key to predicting the risks and benefits of AI systems. This is especially true in biological applications, where skill rather than knowledge is often the primary barrier for an untrained person. The challenge is that these studie

Tianyu Yang, Terry Ruas, Yijun Tian, Jan Philip Wahle

Vision-language models (VLMs) excel at interpreting text-rich images but struggle with long, visually complex documents that demand analysis and integration of information spread across multiple pages. Existing approaches typically rely on fixed reasoning templates or rigid pipelines, which force VLMs into a passive role and hinder both efficiency and genera

Tribikram Choudhury, Yacine Haddad, Michael Doser

Infrared quantum dots, operating in the near-infrared (NIR, 700-1400 nm), short-wavelength infrared (SWIR, 1400-3000 nm), mid-infrared (MIR, 3000-8000 nm) and long-wavelength infrared (LWIR, 8000-15000 nm) regions, have promising potential in optoelectronics, nanotechnology and military surveillance applications. The properties of infrared quantum dots exhib

Dinesh Kumar Keshari, Avijit Pal, Bhaskar Paul

A $7$-tuple of commuting bounded operators $\textbf{T} = (T_1, \dots, T_7)$ on a Hilbert space $\mathcal{H}$ is called a \textit{$\Gamma_{E(3; 3; 1, 1, 1)} $-contraction} if $\Gamma_{E(3; 3; 1, 1, 1)}$ is a spectral set for $\textbf{T}. $ Let $(S_1, S_2, S_3)$ and $(\tilde{S}_1, \tilde{S}_2)$ be tuples of commuting bounded operators defined on a Hilbert spac

Ayse Irmak Ercevik, Aidan Dakhama, Melane Navaratnarajah, Yazhuo Cao

Fuzzing has become a key search-based technique for software testing, but continuous fuzzing campaigns consume substantial computational resources and generate significant carbon footprints. Existing grey-box fuzzing approaches like AFL++ focus primarily on coverage maximisation, without considering the energy costs of exploring different execution paths. Th

Kristóf Bérczi, Karthekeyan Chandrasekaran, Tamás Király, Daniel P. Szabo

Narayanan showed the existence of the principal partition sequence of a submodular function, a structure with numerous applications in areas such as clustering, fast algorithms, and approximation algorithms. In this work, motivated by two applications, we develop a theory of $\{s,t\}$-separating principal partition sequence of a submodular function. We defin

Corrado Lattanzio, Ramón G. Plaza, José Manuel Valdovinos

This paper studies the diffusion approximation, non-equilibrium model of radiation hydrodynamics derived by Buet and Despr\'es (J. Quant. Spectrosc. Radiat. Transf. 85 (2004), no. 3-4, 385-418). The latter describes a non-relativistic inviscid fluid subject to a radiative field under the non-equilibrium hypothesis, that is, when the temperature of the fluid

Gabrielle O'Brien, Antonio Pedro Santos Alves, Sebastian Baltes, Grischa Liebel

Programming assistants powered by large language models (LLMs) have become widely available, with conversational assistants like ChatGPT particularly accessible to novice programmers. However, varied tool capabilities and inconsistent availability of extensions (web search, code execution, retrieval-augmented generation) create opportunities for user misconc

Chaithanya Naik Mude, Swamit Tannu

Quantum Error Correction (QEC) protects qubits against bit- and phase-flip errors in the |0> or |1> subspace, but physical qubits can also leak into higher energy levels (e.g., |2>). Leakage is especially harmful, as it corrupts all subsequent syndrome measurements and can spread to neighboring qubits. Detecting leakage on data qubits is particularly challen

Diego Royo, Jorge Garcia-Pueyo, Miguel Crespo, Guillermo Enguita

mitransient is a light transport simulation tool that extends Mitsuba 3 with support for time-resolved simulations. In essence, mitransient extends conventional rendering by adding a temporal dimension which accounts for the time of flight of light. This allows rapid prototyping of novel transient imaging systems without the need of costly or difficult-to-op

N. Roldan-Levchenko, D. J. P. de Sousa, C. O. Ascencio, J. D. S Forte

Optical gain is a critical process in today's semiconductor technology and it is most often achieved via stimulated emission. In this theoretical study, we find a resonant TE mode in biased low-symmetry two-dimensional metallic systems which may lead to optical gain in the absence of stimulated emission. We do so by first modeling the optical conductivity us

Javad Aliakbari, Johan Östman, Ashkan Panahi, Alexandre Graell i Amat

We consider the problem of federated learning (FL) with graph-structured data distributed across multiple clients. In particular, we address the prevalent scenario of interconnected subgraphs, where interconnections between clients significantly influence the learning process. Existing approaches suffer from critical limitations, either requiring the exchang

Van-Vi Vo, Tien-Dung Nguyen, Duc-Tai Le, Hyunseung Choo

Efficient routing in IoT sensor networks is critical for minimizing energy consumption and latency. Traditional centralized algorithms, such as Dijkstra's, are computationally intensive and ill-suited for dynamic, distributed IoT environments. We propose a novel distributed Q-learning framework for constructing shortest-path trees (SPTs), enabling sensor nod

Cynthia A. Huang, Mitchell O'Hara-Wild, Rob J. Hyndman, Matthew Kay

Visualizing changes over time is fundamental to learning from the past and anticipating the future. However, temporal semantics can be complicated, and existing visualization tools often struggle to accurately represent these complexities. It is common to use bespoke plot helper functions designed to produce specific graphics, due to the absence of flexible

J. Rodrigues, F. Turoboś, M. Lenartowicz, Z. Puchała

This paper presents AIRS, a day-of-operations disruption-recovery system. AIRS.ACR models integrated aircraft-crew recovery on a Time-Space Network (TSN) and solves a mixed-integer linear program (MILP) that enforces rotation continuity, crew legality, maintenance windows, slot capacities, and multi-leg integrity via flow-balance constraints; disruption-awar

Giulia Venditti, Francesco Macheda, Paolo Barone, José Lorenzana

KTaO$_3$ (KTO) is an incipient ferroelectric, characterized by a softening of the lowest transverse optical (TO) mode with decreasing temperature. Cooper pairing in the recently discovered KTO-based heterostructures has been proposed to be mediated by the soft TO mode. Here we study the electron coupling to the zone-center odd-parity modes of bulk KTO by mea

Steffen Schultze, Helmut Grubmüller

MINFLUX microscopy allows for localization of fluorophores with nanometer precision using targeted scanning with an illumination profile with a minimum. However, current scanning patterns and the overall procedure are based on heuristics, and may therefore be suboptimal. Here we present a rigorous Bayesian that offers maximal resolutions from either minimal

Michelle Chen Huebscher, Katharine Mach, Aleksandar Stanić, Markus Leippold

Evaluating how Large Language Models (LLMs) handle complex, specialized knowledge remains a critical challenge. We address this through the lens of climate change by introducing CLINB, a benchmark that assesses models on open-ended, grounded, multimodal question answering tasks with clear requirements for knowledge quality and evidential support. CLINB relie