Peer Revenue Play credits. Real papers.

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Tuesday 29 September

Cryptography and Security

Security-Enhanced Seed-Based Weight Quantization for Large Language Models

Qiuyu Ren, Sudipta Paria, Aritra Dasgupta et al. (+1 more)

Large language models (LLMs) incur substantial storage, memory-bandwidth and energy costs, motivating compact weight representations. Existing seed-based compression methods reconstruct weights from compact pseudo-random representations but do not explicitly account for the non-uniform sensitivity of model weights. We introduce Seed-Q, a security-enhanced sensitivity-aware seed-based weight compression framework that uses lightweight Linear Feedback Shift Register (LFSR)-based weight generation with non-uniform bit allocation. Our approach assigns larger representation budgets to sensitive weights while aggressively compressing less sensitive regions. Importantly, this non-uniform allocation requires no side-information: the decoder deterministically reconstructs the bit-allocation schedule, with no rung depending on the decoded weights, eliminating the need to store per-block metadata or use calibration data while preserving the baseline coding rate. Experiments across diverse LLMs show that Seed-Q matches 4-bit perplexity of SeedLM with fewer bits, while at the same 4 bits/weight it reduces both perplexity degradation and zero-shot accuracy loss relative to SeedLM. We also show that Seed-Q simultaneously achieves high security against bit-flip attacks on model parameters, as bit corruption affects multiple reconstructed weights, greatly amplifying its impact and making it easier to detect. We further implement Seed-Q in an ASIC-based accelerator and demonstrate modest hardware overhead compared to prior seed-based approaches.
Read the full abstract

All authors Qiuyu Ren, Sudipta Paria, Aritra Dasgupta, Swarup Bhunia

Computer Vision

Curating Synthetic Data for Task-Specific Visual Perception

Saptarshi Neil Sinha, Paul Julius Kühn and Michael Weinmann

Synthetic data are most valuable where general-purpose datasets cannot provide the domain-specific priors a task requires, and where manual annotation is expensive, imprecise, or infeasible. In this article we argue that the central question for specialized vision systems is not how to generate more data, but which data to generate. We therefore discuss curated synthetic data, whose scene content, appearance variations, sensing characteristics, and annotations are deliberately designed around a given task. We examine three complementary curation paradigms. Procedural rendering offers explicit control over scene parameters and the annotations follow by construction. Physically-based simulation encodes the mechanism behind an observed effect and yields exactly aligned supervision pairs. Generative AI learns sensor-specific appearance from small real seed sets and attains plausible realism, though it remains prone to hallucination and to inaccurate annotation. These paradigms are illustrated with examples from industrial surface defect detection, restoration of degraded digitized autochrome plates, and 6DoF pose estimation from RGB and event data. Using these examples, we analyze different data regimes and training strategies that combine synthetic and real data across these paradigms. We conclude that curated synthetic data are best understood as a complement to real observations, and that hybrid pipelines combining controllable supervision with learned appearance are the most promising direction for reliable sim-to-real transfer.
Read the full abstract

Analysis of PDEs

Derivation of the Enskog–Enskog System from a Particle System

Zhe Chen

We formally derive a coupled Enskog system for a gas–particle mixture from the hard-sphere BBGKY hierarchy. The derivation retains the finite separation between colliding centres, includes collisions within and between species and explicitly states the direction of spatial shifts and the mass-dependent prefactors of the four collision operators. The resulting system provides a kinetic starting point for two-phase flows for which the volume effect cannot be ignored. For unit contact factors, we derive the corresponding gas and particle balance laws, and establish an entropy inequality under suitable regularity assumptions.
Read the full abstract

Systems and Control

Condition-Based Maintenance of Degrading Assets underIntermittent Accessibility

Muhammet Ceyhan Sahin, Murat Yildirim and Onur A. Kilic

Many maintenance models implicitly assume that maintenance can be performed whenever intervention is warranted. In practice, however, environmental uncertainty, such as weather and sea conditions, can make maintenance opportunities intermittent and dynamically evolving. We consider a degrading asset for which preventive maintenance may be performed before failure, but intervention is possible only when the asset is accessible. Accessibility evolves stochastically over time and therefore affects not only whether maintenance can be performed now, but also the value of waiting for future opportunities. We formulate this setting as a finite-state Markov decision process under a long-run average cost criterion and investigate the structure of the optimal maintenance policy. Under monotone degradation and cost conditions, we show that an optimal policy retains a threshold form in asset condition, but unlike a single threshold, the optimal threshold varies with the accessibility state. Thus, intervention depends jointly on the asset condition and the expected evolution of future maintenance opportunities. Through a numerical study motivated by offshore wind turbine maintenance, we examine how accessibility dynamics, degradation characteristics, and economic parameters shape the optimal thresholds and compare the optimal policy with constant-threshold and age-based maintenance policies. In a representative setting, adapting the condition threshold to accessibility reduces long-run average cost by 4.31% relative to an optimized condition-based maintenance policy. Relative to an optimized age-based policy, the optimal policy reduces long-run average cost by 33.95%. The results show that maintenance thresholds are not universal: they should adapt to the degradation and accessibility characteristics of the system, and failing to account for these conditions can lead to substantial cost increases.
Read the full abstract

Information Retrieval

Re-ranking and Late Interaction Drive Retrieval Quality: A Controlled Comparison of RAG Strategies for Scientific Question Answering

Bhagyesh Rathi, Eshan Chawla and William B. Andreopoulos

Retrieval-Augmented Generation (RAG) is now the standard way to ground Large Language Models (LLMs) in external knowledge, yet the design space of retrieval pipelines is large and the trade-offs between variants are not well understood, especially on domain-specific corpora at realistic scale. In this work, we present a controlled comparison of six retrieval strategies for scientific question answering: (i) classic top-k dense retrieval, (ii) LLM-based query rephrasing, (iii) query rephrasing followed by LLM-based reranking, (iv) multi-query fusion via Reciprocal Rank Fusion (RRF), (v) an agentic tool-call pipeline in which the generator decides for itself whether to retrieve, and (vi) late-interaction retrieval with ColBERTv2. All six pipelines share the same generator (Meta-Llama/Llama-3.1-8B-Instruct), prompt, and evaluation protocol; the five single-vector pipelines additionally share SPECTER2 embeddings and a Chroma vector store; and all six retrieve from the full corpus of 463,971 arXiv papers dated 2024-2025. To support reproducible, large-scale evaluation, we also release a synthetic question dataset of 19,484 problem-statement and methodology questions generated by Llama-3.1-8B-Instruct from a random sample of 10,000 papers across academic domains (query generation succeeded for 9,742 of them), and every strategy is evaluated on this same query set. We describe the architecture and implementation of each pipeline, release the code and the synthetic question dataset, and evaluate each strategy with an LLM-as-a-judge protocol along multiple quality dimensions, together with direct gold-paper retrieval metrics. The result is an open testbed for studying the cost and quality trade-offs of RAG design choices on a research-literature corpus, and a basis for future work on faithfulness, retrieval robustness, and agentic retrieval.
Read the full abstract

Robotics

Diffusion-2BC: Hybrid Diffusion and Regression Training for Offline Behavior Cloning in Autonomous Driving

Bruno Maciel Machado and Eric Aislan Antonelo

Behavior cloning provides an offline route to autonomous-driving policy learning, but mean-squared-error regression is poorly matched to demonstrations in which one observation admits several valid actions. Diffusion policies can represent conditional multimodal action distributions, yet their closed-loop performance may be unstable when visual features and control are learned from limited data. This paper presents Diffusion-2BC, which combines a diffusion denoising objective with an auxiliary deterministic behavior-cloning loss over a shared visual encoder. The auxiliary branch is used only during training; inference remains diffusion-based. The proposed method is evaluated in the controlled Claw environment and in bird's-eye-view CARLA navigation, including route-conditioned driving, route-free navigation through multiple intersections, and cross-map evaluation from Town01 to Town02. In the Claw task, Diffusion-2BC reduced the mean mask-distance error by approximately 10% relative to a diffusion-based behavior-cloning baseline and by 85% relative to standard deterministic behavior cloning. In route-free CARLA, Diffusion-2BC traveled substantially farther before termination under the evaluation protocol than both baselines in Town01 and Town02. Additional qualitative rollouts revealed distinct route choices, showing the multimodal behavior of the proposed diffusion-based agent. The results indicate that an auxiliary regression signal can improve the closed-loop reliability of diffusion behavior cloning while preserving multimodal prediction in the controlled benchmark.
Read the full abstract

Information Theory

Derandomizing Dense Binary Hypervector Codebooks for Quantized Scalars

Dmitri Rachkovskij, Evgeny Osipov, Olexander Volkov et al. (+2 more)

Hyperdimensional computing and vector symbolic architectures often represent quantized scalar levels by dense binary codebooks whose level-to-level similarity is intended to follow a prescribed function of scalar separation. At finite dimensionality, randomized scalar codebook constructions deviate from this target because of sampling noise, random-start imbalance, update-count fluctuations, component dependence, and finite-capacity effects. We develop a transition-based derandomization framework for dense binary scalar codebooks across two target-similarity families, with similarity decaying exponentially or linearly with level separation. The framework separates the target similarity law, the derandomization variant, and the concrete generator construction, making explicit how initialization, selection, update, and capacity-handling mechanisms shape the induced similarity profile. We formalize derandomization variants that separately constrain initial Hamming weight, update-count variability, and update balance, thereby controlling distinct sources of finite-dimensional error. For each family and variant, we derive the induced mean similarity, identify realization-wise and mean target-matching regimes, and derive exact finite-dimensional expressions for bias, variance, and root-mean-square error. Simulations across dimensions, quantization ranges, reference scalar levels, and generator constructions validate the theory and show how each constraint removes or reduces a specific source of similarity mismatch. The results provide practical guidance for choosing scalar codebook generators that more closely match a desired similarity law under finite-dimensional and hardware-relevant constraints.
Read the full abstract

All authors Dmitri Rachkovskij, Evgeny Osipov, Olexander Volkov, Denis Kleyko, Vaclav Snasel

Fluid Dynamics

Stable and Interpretable Multi-Mode Rheological Universal Differential Equations (mmRUDEs) for Data-Driven Constitutive Modeling

Mohua Das, Nicholas King, Navid Azizan et al. (+1 more)

The Rheological Universal Differential Equation (RUDE) framework embeds neural networks within a frame-indifferent tensorial constitutive backbone and so enables data-driven discovery of complex material rheological behavior. However, the flexibility that makes the RUDE framework attractive also makes it difficult to deploy: the learned neural network correction can make the constitutive equation numerically stiff during training, distribute corrections across non-unique combinations of tensor-basis terms, and extrapolate poorly with increasing nonlinearity, especially for industrially relevant materials with broad relaxation spectra. We develop a stability- and reliability-focused multi-mode RUDE framework that addresses three key challenges: thermodynamic admissibility, out-of-distribution prediction, and interpretability. Logarithmic compression of the input invariants and a split-network architecture improve numerical conditioning during training, while a differentiable projection layer enforces the Clausius–Duhem inequality at every time step. We illustrate the framework with both synthetic (a multi-mode Giesekus model) and experimental data on an entangled silicone polymer melt. Both models were trained on LAOS; they remain stable and accurate under unseen flow conditions, including steady shear, transient stress growth, and extensional flows, for which the unconstrained models may diverge numerically. To aid in interpretation of the contributions of each learned correction to the rheological behavior, we introduce the rheome, a compact way of representing the dominant weighted tensor-basis contributions to the overall rheological behavior. These improvements to the RUDE framework lead to stable and interpretable trained models, paving the way for implementation in computational fluid dynamics simulations and for guiding the rational design of soft material processing operations.
Read the full abstract

All authors Mohua Das, Nicholas King, Navid Azizan, Gareth H. McKinley

Computation and Language

The Backdrop Exposes What the World Around an Agent Costs It

Nusrat Jahan Lia and Shubhashis Roy Dipta

Agent benchmarks test agents in worlds that stay still. Deployed agents work in worlds that other people also change. Someone texts the agent to send the money elsewhere or an order confirmation asks it to reply with a door code. We present BACKDROP, which asks how much of an agent's capability in a clean world survives in such a world. BACKDROP takes a task along with the agents execution environment, and plants four everyday hazards in its world, one at a time and all together. The instruction and the correct end state stay the same. Each hazard asks one question. Authority: does a message from another person override the user? Injection: does text planted in a record redirect the agent? Boundary: does a request pull it into an app it was not given? Fault: after a write fails without saying whether it landed, does the agent check before it retries? Across 3,678 variants and 16 models, , the average pass rate falls from 69.5% to 31.3% once all four hazards are present; the strongest models fall furthest (Claude Fable 5.1 from 96.6% to 56.0%). Agents have learned to resist injected text but often follow other unauthorized requests of other people. With all four hazards present, and counting only runs where the planted text reached the agent, agents followed another person's message in 46.4% of runs and injected text in 20.3%. The gap is consistent throughout all 16 models. BACKDROP formalizes these gaps and shows how an agent's score in a task's world is a ceiling on real-world performance.
Read the full abstract

Earth and Planetary Astrophysics

Migration and Evolution of giant ExoPlanets (MEEP). III. Twenty-Nine Giant Planets from the TESS Mission

Jack Schulte, Joseph E. Rodriguez, David W. Latham et al. (+97 more)

We present the discovery and characterization of 29 hot and warm Jupiter systems transiting bright ($G < 12.8$) FGK stars using TESS data and ground-based photometry and spectroscopy. Through the use of high angular resolution imaging, we discovered three bound stellar companions, named TOI-3988 B, TOI-6171 B, and TOI-7266 B. The planets that we confirmed span an orbital period range of $1.28 - 16.9$ days, a mass range of $0.35 - 8.7$ $\mathrm{M}_{\mathrm{J}}$, and an orbital eccentricity range of $0 - 0.6$, and add to the growing self-consistent sample of hot and warm Jupiter systems detected using TESS and analyzed using the Markov-Chain Monte Carlo code EXOFASTv2. By performing two-sample Kolmogorov-Smirnov tests on a set of host star and planetary parameters, we find that this growing self-consistent sample of giant planets has a different distribution of host star masses, surface gravities, and metallicities, compared to the hot Jupiter and warm Jupiter systems in the literature, likely due to the additional constraint on the host star density from the planetary transits included in our global fits. We find that seven of the 29 newly confirmed planets have significant orbital eccentricity, and that one of these, TOI-3365 b, is in a system which likely hosts a massive outer companion. The continued confirmation of hot Jupiter and warm Jupiter systems and the follow-up RV monitoring on these systems to identify outer companions will pave the way for important investigations of giant planet migration.
Read the full abstract

All authors Jack Schulte, Joseph E. Rodriguez, David W. Latham, Allyson Bieryla, Karen A. Collins, Samuel N. Quinn, Xian-Yu Wang, Rafael Brahm, Abdullah Al-Haji, Robert Aloisi, Caitlyn Altermatt, Nikola Antonov, David Baker, Pavel Balanutsa, Khalid Barkaoui, Özgür Baştürk, Slawomir Bednarz, Paul Benni, Krzysztof Bernacki, Mario Billiani, Ventsislav Bodakov, Pau Bosch-Cabot, Valerio Bozza, Ahmet Cem Kutluay, David R. Ciardi, Catherine A. Clark, Jerome de Leon, Jason D. Eastman, İpek Aleyna Ertürk, Thomas M. Esposito, Phil Evans, Miranda Felsmann, Raquel Forés-Toribio, Isaac Fournier, Akihiko Fukui, Tianjun Gan, Cristilyn Gardner-Watkins, Purvi Garg, Eric Girardin, Tateki Goto, Ferran Grau Horta, Thomas Henning, Eric G. Hintz, Keith Horne, Kai Ikuta, Jon M. Jenkins, Andrés Jordán, Jenson Keen, Adam Kraus, Valentina Kuzmina, Artem Kuznetsov, Adam Lark, Rodrigo Leiva, Vladimir Lipunov, Michael B. Lund, Franck Marchis, Joshua Margherone, Robert Massey, Kim McLeod, Edward Michaels, Mayuko Mori, J. A. Muñoz, Felipe Murgas, Norio Narita, Takaya Okada, Shane Painter, Enric Palle, Ivan Panchenko, Hannu Parviainen, Anastasiia Pechko, Adam Popowicz, Don Radford, Fabian Rodriguez Frustaglia, Sylvain Roux, Subhroja Roy, Matthew Ryno, Gülben Şanc{\i}, Richard P. Schwarz, Victor Senik, Lauren A. Sgro, Avi Shporer, Gregor Srdoc, Denise Stephens, Chris Stockdale, Ashley Stone, Ivan Strakhov, Marcelo Tala Pinto, Thiam-Guan Tan, Suzanne Taylor, Neil Thomas, Nataly Tiurina, Julian C. van Eyken, Songhu Wang, Noriharu Watanabe, Levi R. Webb, Francis P. Wilkin, Joe Williams, Selçuk Yalç{\i}nkaya, Mesut Y{\i}lmaz, Carl Ziegler

Robotics

L1-MPPI: L1 Adaptive Model Predictive Path Integral for Agile UAV Control

Lukáš Kotek, Ondřej Procházka, Vojtěch Vonásek et al. (+2 more)

This work proposes the L1 Adaptive Model Predictive Path Integral (L1-MPPI). It cascades L1 adaptive control with the Model Predictive Path Integral (MPPI) to improve tracking of high-speed UAV trajectories. Thanks to the L1augmentation, the tracking remains accurate even under model uncertainties and external disturbances, such as an additional payload or a mismatch in the modeled aerodynamic drag. In contrast to existing MPPI approaches for UAV control that do not explicitly model aerodynamic effects, varying payloads, and typically neglect the dynamics of low-level motor controllers, our L1-MPPI approach enhances the dynamic model used in the MPPI by incorporating the low-level flight controller and motor dynamics, as well as an iterative mixing scheme that reflects the approach of the low-level controller. The proposed method demonstrates improved tracking performance in both simulation and the real world, even when the UAV is subjected to an unknown payload. In flight with 35% mass increase, our approach lowers the RMSE by 58.61% with respect to plain MPPI. Compared to the same MPPI using an online mass estimator in place of the L1 augmentation, the RMSE is lower by 38.59%. During the real-world experiments the UAV reaches speeds up to 13.50 m/s and accelerations up to 2.5 g.
Read the full abstract

All authors Lukáš Kotek, Ondřej Procházka, Vojtěch Vonásek, Martin Saska, Robert Pěnička

Computer Vision

PAMI: Part Anchored Motion for Text to Human-Object Interaction Generation

Chuqiao Li, Xianghui Xie, Yong Cao et al. (+2 more)

Text-conditioned full-body human-object interaction (HOI) generation requires synthesizing human motion and object trajectories that match the input text while remaining precisely coordinated over time. Most methods represent the human and object as separate trajectories and predict the global human-object couplings. Learning this complex, dynamically changing relationship implicitly, however, often yields object drift, missed contact, and penetration. We introduce PAMI, a Part-Anchored Motion framework for Interaction generation. Inspired by the classic Hough Transform, our key idea is to localize object motion by letting body-part anchors vote for it: we express object motion relative to multiple body-part anchors and use PamiVAE to learn an interaction latent space, decoding frame-wise weights that aggregate these part-specific votes. Building on this representation, PAMI generates interactions in a coarse-to-fine hierarchy. PamiGen first generates a coarse human-object interaction from text in this structured latent space, and PamiRefiner then recursively resolves fine-grained contact geometry using a hybrid surface-sensing representation, combining long-range probes that capture overall body-part influence with short-range sensors that resolve detailed contacts near the object surface. Experiments on InterAct show that PAMI generates more faithful interactions and more accurate human-relative object motion than previous methods, achieving 14.5% higher contact recall than the previous state of the art. Extensive ablations validate the contributions of both the part-anchored voting representation and hybrid surface-sensing refinement.
Read the full abstract

All authors Chuqiao Li, Xianghui Xie, Yong Cao, Andreas Geiger, Gerard Pons-Moll

Machine Learning

Does Gradient Conflict Predict the Understanding–Generation Trade-off? A Controlled Audit of Conflict-Metric Validity in Unified Multimodal Models

Shuyang Jiang, Fucheng Deng, Yuchuan Luo et al. (+1 more)

Unified multimodal models (UMMs) are increasingly designed around gradient conflict between understanding and generation objectives. The premise that reducing these metrics improves the downstream understanding-generation trade-off has never been tested directly. We audit it in a controlled testbed, GRIDUMM, which mirrors key structural ingredients of UMM training while making the ground-truth trade-off exactly computable. Across 63 configurations and 372 measured checkpoints, no directional conflict metric reaches an absolute Spearman correlation of 0.3 with a confidence interval excluding zero for conflict measured during training against the eventual trade-off. A dose-response intervention that monotonically suppresses conflict leaves the trade-off flat, separating correlation from causation. The norm ratio is a generation-failure detector and becomes null among configurations that master generation. Functional interference measures outperform directional conflict metrics, while training loss tracks the trade-off strongly. Our results do not show that conflict is useless; they show that its validity as a diagnostic target must be established, not assumed, and we release the audit protocol as a reusable standard.
Read the full abstract

All authors Shuyang Jiang, Fucheng Deng, Yuchuan Luo, Zhenyu Wu

Soft Condensed Matter

Morphodynamics of lipid vesicles under osmotic forcing

Nicholas Broussinos and David Saintillan

We investigate the dynamics of spherical lipid vesicles subjected to an osmotic shock using continuum simulations of a permeable, area-incompressible fluid membrane with bending elasticity and surface viscosity. The dynamics are governed by a single dimensionless parameter $\varGamma$, which measures the relative strength of osmotic pressure and bending forces. Osmotic forcing drives a three-stage evolution comprising rapid buckling and collapse, smoothing of the resulting wrinkles and ridges, and slow folding into an invaginated morphology resembling the precursor to experimentally observed vesicle-in-vesicle structures. Linear stability analysis predicts the wavelength and growth rate of the initial buckling instability and shows that the dimensionless buckling time scales as $\varGamma^{-2}$, in quantitative agreement with simulations. At longer times, the energy budget reveals a reversal in elastic energy transfer: bending energy accumulated during buckling is released during relaxation and dissipated by membrane viscosity. A scaling analysis of the folding dynamics predicts that the dimensionless folding time scales as $\varGamma^{-2/3}$, again in excellent agreement with simulations. These results provide a continuum description of the mechanisms and widely separated timescales governing osmotically driven membrane collapse and invagination.
Read the full abstract

Robotics

TrafficSignBench: Rule-Centric Closed-Loop Evaluation of Traffic-Sign Compliance in Autonomous Driving

Victoria Smirnova, Viktoriia Zinkovich, Gregorii Bukhtuev et al. (+4 more)

Autonomous driving planners are typically evaluated using aggregate metrics such as driving score, destination rate, and collision rate, which do not explicitly measure compliance with traffic rules. As a result, planners can achieve high benchmark scores while still exhibiting unsafe or illegal behaviors, limiting their applicability to real-world deployment. To address this gap, we introduce TrafficSignBench, a large-scale, traffic sign-centric benchmark for systematic and interpretable evaluation of traffic-rule compliance in autonomous driving. Our framework combines real-map-based simulation for realistic road layouts with rule-targeted procedural scenario generation for scalable and balanced coverage of underrepresented rules. We implement traffic rules corresponding to 34 traffic signs, each equipped with an automatic rule checker for detecting violations during closed-loop execution. This design yields 29,000 diverse road scenes and 29 distinct testing scenario types, enabling controlled evaluation of rule-specific planner behavior. We construct 5,800 testing scenes and demonstrate that current autonomous driving planners can exhibit poor traffic-rule compliance despite strong performance on standard evaluation metrics. To address this limitation, we transform existing planners into rule-compliant trajectory experts via explicit traffic-sign constraints, enabling scalable generation of high-quality oracle trajectories for fine-tuning.
Read the full abstract

All authors Victoria Smirnova, Viktoriia Zinkovich, Gregorii Bukhtuev, Artem Belyaev, Andrey Kuznetsov, Denis Shepelev, Vlad Shakhuro

Robotics

Observability Analysis and Online Calibration of Visual-Inertial-Wheel Odometry for 4WIS4WID Mobile Robots

Branimir Ćaran, Vladimir Milić, Bojan Šekoranja et al. (+1 more)

In this paper, we present a visual-inertial-wheel odometry (VIWO) framework with online calibration for four-wheel independently steered and driven (4WIS4WID) mobile robots. We derive a 2D odometry model directly from the four driving velocities and steering angles, using both the longitudinal rolling constraints and the lateral no-slip constraints of all wheels. A preintegration model and analytical Jacobians are developed for efficient filtering and calibration. An observability analysis of the linearized VIWO system shows that a drive-only model makes all steering offsets unobservable, whereas the proposed redundant model restores their observability. The analysis also identifies four standard VINS unobservable directions and three additional directions associated with the arbitrary placement of the odometry reference frame. Furthermore, we characterize several degenerate motions, including zero yaw rate, constant steering, and a non-rolling wheel, and derive the corresponding excitation conditions for the thirteen wheel intrinsics that remain after fixing the odometry frame reference. The performance of the proposed system has been demonstrated in both simulation and real-world experiments on a 4WIS4WID mobile robot.
Read the full abstract

All authors Branimir Ćaran, Vladimir Milić, Bojan Šekoranja, Bojan Jerbić

General Relativity and Quantum Cosmology

Explicit Analytical Representations for the Schwarzschild Radial Equation via Hypergeometric and Frobenius Expansions

J. G. R. Valangelis, Ailton C. Nascimento and Helder A. S. Costa

The massive Klein-Gordon equation on the Schwarzschild exterior reduces to a radial differential equation of confluent-Heun type. We present its exact reduction by rigorously retaining a subleading centrifugal term that preserves the horizon indicial exponents but critically shifts the accessory parameters. Within the Svartholm–Schmidt hypergeometric-expansion framework, we derive the three-term recurrence relation and establish the corrected continued-fraction compatibility condition matching the minimal tail to the lower-end recurrence. Crucially, we demonstrate that the physical compactification coordinate $z=(r-1)/r$ shifts the irregular singular point to $z=1$, inherently transforming the hypergeometric expansion into a five-term recurrence that cannot degenerate to three terms. To circumvent this obstruction, we construct the horizon-normalized physical branch via a direct Frobenius series at $z=0$. This representation exhibits high-order convergence and near-precision-floor residuals against the exact radial ODE, while clarifying why quasinormal-mode spectral selection remains a connection problem at the irregular singular endpoint.
Read the full abstract

Artificial Intelligence

NAQD Env: A benchmark for selective withdrawal in language agents

Mohamed Abouzahra

Language agents must revise planned actions when evidence changes, permission is revoked, or a stop instruction arrives. A useful response is selective: suspend affected actions, preserve unaffected work, and resume only after sufficient repair. We introduce NAQD-Env, a synthetic environment that evaluates these decisions against a deterministic reference policy over explicit evidence, authorization, and constraint dependencies. Eleven dependency families support evaluation on development structures, held-out families, and held-out combinations of structures. Metrics distinguish attempted violations from violations permitted by a simulated execution gate and jointly report policy agreement, task value, withdrawal, resumption, and event reporting. We evaluate three open-weight instruction-tuned models from two families under three prompt conditions on 350 frozen scenarios, yielding 3,150 model-prompt episodes before gate replay. Across the reported conditions, withdrawal recall is at most 0.06, no valid resumption is observed at eligible opportunities, and only one episode matches the complete reference policy. Under the NAQD prompt, Qwen2.5-7B has fewer unsafe-attempt episodes than Qwen2.5-3B and Llama-3.1-8B, but also completes less useful work and preserves unaffected actions less accurately. Exploratory supervised fine-tuning probes increase Qwen2.5-3B decision accuracy from 0.45-0.54 to 0.83-0.92; separate diagnostics reveal inappropriate withdrawal after curriculum omissions and a loss of event reporting. These results motivate evaluating selective withdrawal as a distinct component of agent reliability. The setting measures policy application with trusted structured inputs and does not establish real-world containment or source-verification ability.
Read the full abstract

General Economics

Maintaining Human Verification Capacity under Automation

Li Gan and Eric Gan

Human verification depends on expertise that must be maintained before it is needed. This paper links reliance on automated checks, investment in human checking ability, and performance during an interruption. Better checking lowers the error reduction gained from an extra unit of human skill while the checker works. It can therefore reduce the incentive to preserve independent expertise, even when it lowers the best achievable expected cost of maintenance and errors. In an illustration, a more informative checker raises detection while it works from \hoPowWorkA{} to \hoPowWorkC{} percent, but the organization then keeps no routine practice and detects \hoPowOnsetC{} percent of errors when the checker first fails, against \hoPowOnsetA{} percent with a less informative checker. A detection requirement therefore concerns both current capability and its survival until new training becomes effective. Evidence from colonoscopy and aviation documents weaker unaided performance under routine automation, without isolating the mechanism. The framework connects a detection target to an explicit reserve of expertise and a training pipeline. Standard detection tests estimate each quantity and reveal automation bias and silent checker failures. The framework distinguishes the requirement from minimizing expected loss and proposes a longitudinal test.
Read the full abstract

Artificial Intelligence

PrivMeSA: Privacy-Aware Self-Evolving Multi-Agent System for Medicine via Local-Remote LLM Collaboration

Dannong Wang, Yuran Zhang, Bian Sun et al. (+4 more)

Clinical large language model (LLM) agents deployed locally can consult more capable remote models, but doing so risks exposing patient information. Privacy-conscious delegation places disclosure decisions with a local agent, yet removing explicit identifiers is insufficient: quasi-identifiers can accumulate across multi-turn consultations and repeated patient visits to enable re-identification. We introduce PrivMeSA, a privacy-aware self-evolving multi-agent system that learns to control disclosure and retains remote expertise for local reuse. A local agent manages each encounter and consults remote specialists that may request additional information. Reinforcement learning balances task accuracy against direct disclosure and registry-based re-identification risk, with privacy evaluated over the complete outbound transcript of each encounter. A local lesson memory distills completed consultations into generalized clinical guidance and retrieves relevant lessons before transmission, allowing subsequent cases to reuse expertise without another remote exchange. Memory grows without additional outcome labels or parameter updates. On an emergency-department benchmark built from MIMIC-IV-ED records, PrivMeSA improves mean task accuracy over delegation by up to 15.8 percentage points. In the same setting, PrivMeSA reduces the disclosure of personal details from 98.0% to 0.2% of cases and the share of cases in which the patient can be narrowed to ten or fewer registry patients from 74% to 0%.
Read the full abstract

All authors Dannong Wang, Yuran Zhang, Bian Sun, Alex Stinard, Yuzhang Shang, Song Wang, Yu Tian

Cosmology and Nongalactic Astrophysics

Toward Detecting the Moving Lens Effect with Optical Spectroscopy

David Wittman, Rodrigo Stancioli and Christopher Hopp

Relativity predicts that sky-plane motion of a gravitational lens changes the wavelengths of lensed photons. This moving lens effect is orders of magnitude smaller than cosmological and Doppler effects on the same photons, but can be isolated by comparing different strongly lensed images of a given source. With a massive, fast-moving lens such as a galaxy cluster falling into another cluster, lensed spectra of a given source could differ in wavelength by up to 1 part in $10^6$. We examine the feasibility of detecting this effect with optical spectroscopy, using Keck/Deimos data on the merging cluster RM J001938.7+033557.3 as a case study. We also quantify the predicted sky-plane motions of systems likely to be identified as merging clusters in a $\Lambda$CDM cosmology, and investigate how information on sky-plane motions could help constrain merger models. We find that measurement of the moving lens effect would disambiguate between models in which the subclusters are outbound after first pericenter and those in which the subclusters have returned to the same separation after first apocenter.
Read the full abstract

Combinatorics

A combinatorial interpolation between the hypercube and the associahedron

Lara Bossinger and Ömer Gürdoğan

We study a family of polytopes interpolating between the hypercube and the associahedron. We give an explicit description of their normal fans and describe their combinatorics in terms of chord diagrams associated to their vertices. This yields interesting number sequences generalizing Pell numbers and asymptotic to Catalan numbers. We describe the transition between the normal fans in terms of star subdivisions and relate them to polytopes associated to Dyck paths as in [Veronica Calvo Cortes and Hadleigh Frost. Dyck paths, Configuration Spaces and Polytopes for Linear Nakayama algebras. https://arxiv.org/abs/2602.04571] and hypergraphic polytopes defined in [Carolina Benedetti, Nantel Bergeron, and John Machacek. Hypergraphic polytopes: combinatorial properties and antipode. Journal of Combinatorics. 2019]. Furthermore, all polytopes in our families yield binary geometries.
Read the full abstract

Information Retrieval

AdaM-Rec: Adaptive Modality Routing for Multimodal Recommendation

Honghao Fu, Jiacheng Chen, Manxi Lin et al. (+7 more)

While recent multimodal recommender systems have demonstrated the effectiveness of incorporating visual and textual information to improve downstream performance, most existing methods rely on static modality fusion, assuming that the relative importance of textual and visual signals remains stable across recommendation scenarios. This design may not fully account for an important variation across recommendation requests: some queries require fine-grained visual cues, whereas others are better served by textual or functional semantics, in which case indiscriminate modality fusion brings in uninformative cues and impairs recommendation quality. To address this, we propose AdaM-Rec, an LLM-based framework for adaptive modality routing in multimodal recommendation, which enables dynamic calibration of reliance on textual and multimodal evidence for user-specific queries. Built on structured natural-language representations of items and user preferences, it estimates modality reliability using proxy recall tasks. Specifically, it generates pseudo-queries that match the granularity of the actual query while pointing to the user's positively interacted items as verifiable proxy targets, evaluating which modality yields better recall performance in analogous scenarios and optimizing the routing strategy in an agentic manner. It then performs routed recall with optimized strategy, enriches results with collaborative items, and ranks candidates by their relevance to both the query and user preferences. Experiments demonstrate that AdaM-Rec delivers strong performance against state-of-the-art baselines, highlighting the effectiveness and broader potential of adaptive control over modality reliance in multimodal recommendation.
Read the full abstract

All authors Honghao Fu, Jiacheng Chen, Manxi Lin, Junjun Zheng, Xiangheng Kong, Yiwei Wang, Xin Yu, Miao Xu, Yuning Jiang, Yujun Cai

Hardware Architecture

NDS: Programmer-Free Offload of High-Performance Near Data Strands

Shreyas Singh, Pratyush Nandi, Lin Jia et al. (+2 more)

Near Data Processing (NDP) has the potential to significantly improve system performance and energy by alleviating data movement bottlenecks. However, most NDP proposals pose heavy requirements for the software stack, data layout, and/or the underlying hardware. To broaden NDP adoption, this work focuses on a modular hardware-centric approach that automates the above steps and can target unmodified binaries. This paper presents Near Data Strands (NDS), a framework that orchestrates tasks and data automatically without involving the programmer and the software stack. As a first step towards this ambitious goal, this work focuses on regular loops that meet specific criteria. The framework identifies potential offloadable loops, decomposes them into parallel execution strands, creates a high performance instruction schedule, performs the required data marshaling and orchestration, and initiates near-data execution. We demonstrate that NDS achieves a 3.14x geomean speedup over a single-core host baseline and a 1.82x geomean speedup over 8-core host-parallel variants, all without any modifications to the software stack. These gains grow as repeated loop invocations amortize offload costs, showing that legacy applications can leverage a transparent hardware approach to extract benefits from NDP.
Read the full abstract

All authors Shreyas Singh, Pratyush Nandi, Lin Jia, Shankar Balachandran, Rajeev Balasubramonian

Machine Learning

Grokking through the Lens of Minimum-Norm Interpolation

Gil Kur, Ileana Rugina, Clémentine Carla Juliette Dominé et al. (+1 more)

Grokking shows that fitting the training data and learning the underlying signal can occur at very different stages. However, existing theories offer limited quantitative insight into how this delayed generalization depends on inductive bias and signal structure. Our work addresses the gap by developing a statistical theory that characterizes how regularization geometry and signal sparsity govern generalization near interpolation. In particular, we focus on the prototypical setting of high-dimensional regression and identify regimes in which sparsity-promoting regularization makes exact interpolation much more accurate than approximate fitting. In strongly overparameterized noiseless problems, we prove a zero–one generalization law and construct a family of convex norms whose interpolators transition from the trivial risk of the all-zero predictor to exact recovery, while keeping the training error equal to $0$. Furthermore, when feature dimension and sample size are proportional, we provide a precise characterization of training and generalization errors along $\ell_r$-regularization paths. This in turn allows us to quantify the generalization gain that remains near interpolation: we show that this gain increases as the norm becomes more sparsity-promoting and as the target becomes sparser, with a sharp drop in generalization reached for noiseless data and $\ell_1$ regularization. Experiments on diagonal linear networks and transformers trained on modular arithmetic demonstrate the generality of our theoretical predictions. Finally, beyond grokking, our work reveals a statistical instability in minimum-norm interpolation: small perturbations in the regularization strength can lead to drastically different generalization, while preserving small training error.
Read the full abstract

All authors Gil Kur, Ileana Rugina, Clémentine Carla Juliette Dominé, Marco Mondelli

Quantum Physics

Quantum Secure Non-Interactive Reductions

Maxwell Gold, Sarah Hagen, Daniel Alabi et al. (+1 more)

Efficient models for secure computation often rely on offline preprocessed correlations, which subsequently enable private online computation from minimal assumptions. Within these models, the study of secure non-interactive reductions (SNIR) formalizes when one classical correlation can be non-interactively transformed into another, while guaranteeing information-theoretic simulation-based privacy. In this work, we introduce quantum secure non-interactive reductions (QSNIR), a natural extension of SNIR, in which the source and target resource are each a bipartite quantum state. Specifically, when the target resource is classical, our framework enables us to analyze the problem of disseminating private correlations from entanglement, in addition to classical resources. Our primary technical result is that QSNIR privacy in this context can be computed exactly via a semidefinite program (SDP) pair that quantifies an error operationally rooted in a simulation-based cheating advantage. We prove that QSNIR privacy is lower-bounded by a decision problem related to minimum-error state discrimination (MED), and use both definitions to compute exact one-shot privacy errors for universal two-party computation (2PC) preprocessing correlations.
Read the full abstract

All authors Maxwell Gold, Sarah Hagen, Daniel Alabi, Eric Chitambar

Materials Science

Ab initio calculated diamagnetic and paramagnetic susceptibility of carbonate minerals

Jaroslav Hamrle, Matěj Machek, Vladimír K. Kusbach et al. (+1 more)

The anisotropy of magnetic susceptibility (AMS) is a widely used tool to infer rock fabrics, yet quantitative interpretation is limited by sparse single-crystal magnetic properties for rock-forming minerals and by the difficulty of separating intrinsic diamagnetism from impurity-related magnetism. Here we use density-functional theory (DFT) combined with perturbation theory to compute the diamagnetic susceptibility and AMS of calcite-group carbonates (calcite, magnesite, and dolomite) and to quantify the additional paramagnetic contribution from transition-metal doping. For pure calcite, the calculated susceptibility and its anisotropy are in good agreement with published single-crystal measurements, validating the ab initio approach for diamagnetic phases. We provide improved intrinsic diamagnetic reference values for magnesite and dolomite, for which experimental susceptibilities are commonly affected by magnetic impurities. To address impurity effects explicitly, we model Fe and Mn substitution in calcite using supercells. The computed spin moments reproduce expected high-spin states (Fe$^{2+}$, $S=2$; Mn$^{2+}$, $S=5/2$) and yield a susceptibility anisotropy per Fe concentration that matches the experimental slope. The strong anisotropy is primarily governed by an orbital contribution tied to the crystallographic $c$-axis rather than by spin-orbit coupling, highlighting orbital magnetisation as a key but numerically challenging ingredient for modelling of paramagnetic AMS in carbonates.
Read the full abstract

All authors Jaroslav Hamrle, Matěj Machek, Vladimír K. Kusbach, Zuzana Roxerová

Dynamical Systems

Transition Dynamics in General Two-Component Reaction-Diffusion Systems

Taylan Şengül, Burhan Tiryakioglu and Esmanur Y{\i}ld{\i}z Ak{\i}l

This work provides a comprehensive characterization of the first dynamic transition of a general two-component reaction-diffusion system on a bounded interval in one spatial dimension. By rigorously addressing the difficulties arising from the non-linear interactions between the two components, this study significantly extends the scope of previous studies on dynamic transitions of reaction-diffusion systems, which focused on single-equation cases. The analysis is conducted by employing center manifold reduction within the standard classification of dynamic transition theory. A key contribution of this study is that the nonlinear terms are treated to involve arbitrary powers of the unknowns $u$ and $v$. Furthermore, the theoretical framework is complemented by the analysis of a representative and simplified system. As a result of this generalization, the dynamics of a broad class of fundamental reaction-diffusion systems can be systematically classified.
Read the full abstract

Earth and Planetary Astrophysics

Space Journey Beyond Earth: Exploring the properties of exoplanets and multi-planetary systems

Alexandra Muresan

In contrast to the observed exoplanet diversity, planets within many multi-planetary systems exhibit intra-system similarities in their radii, masses, and orbital spacings. This thesis investigates the planetary properties and system architectures, focusing on the intra-system similarities, the inferred planetary interior compositions, and the differences between single- and multi-planetary systems. Paper A shows that planets within the same system tend to have similar orbital spacings even when their sizes or masses differ. The spacing similarity is more pronounced in orbital period ratios than in mutual Hill separations, while the period ratios are more strongly correlated with the masses than with the radii of adjacent planets. These results indicate that uniform orbital spacings represent a common architectural property and are not necessarily associated with similarities in planetary size or mass. Paper B compares transiting planets in observed single- and multi-planetary systems. After excluding hot Jupiters, the distributions of planet types and radii of singles and multis orbiting FGK stars show no significant differences, suggesting a common underlying planet population. Around FGK host stars, however, an overabundance of multis is identified within a specific planet radius range. Paper C finds that non-gas-giant planets in systems hosting detected gas giants are, on average, larger, more massive, and more irregularly spaced than planets in systems without detected gas giants. Adjacent planets exhibit significant similarities in radius, mass, and bulk density, whereas their inferred core and water mass fractions show substantial scatter and no strong correlations. These results indicate that similarities in bulk properties do not necessarily translate into similarities in compositions. These studies place new constraints on the processes governing planet formation and evolution.
Read the full abstract

Artificial Intelligence

Reach Into The CHOIR: Free-List Elicitation Uncovers Distinct Model Voices in LLM Ensembles

Ben Wigler and Maria Tsfasman

Open-ended LLM homogeneity can create false plurality when several systems appear to offer independent perspectives while returning the same familiar default. Single-pass answers obscure the distinction between agreement produced by a tightly constrained answer space, prompt-vocabulary echo, and broader answer spaces with stable alternatives beneath the surface. We introduce CHOIR (Collective Hierarchically-Ordered Inquiry Responses), a framework that adapts free-list elicitation from cognitive anthropology to LLM ensembles. CHOIR repeatedly elicits ranked lists, clusters items into prompt-level concepts, and measures concept salience across models, prompt variants, and persona conditions. We evaluate CHOIR on Infinity-Chat 100, an external prompt bank from recent work on open-ended model homogeneity, and on a 27-question targeted diagnostic bank designed to isolate mechanism-level contrasts. On Infinity-Chat 100, CHOIR reproduces high surface agreement (93/100 prompts above chance) while separating narrow prompts from broad prompts with recoverable depth. Across targeted probes and the external prompt bank, base-model identity remains the strongest recoverable signature, and persona prompts shift surfaced concepts within base-model signatures. A source-blind ranking module prioritises rare-but-stable candidates for later inspection. CHOIR turns open-ended homogeneity into a diagnostic measurement problem by asking where models converge, why they converge, and what remains reachable under structured depth probing.
Read the full abstract