Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 4 September

Probability

An Exact Tail Condition for the Largest Error in Estimating a Rank-One Direction

We consider a rectangular random matrix formed by adding a rank-one term to a matrix with independent entries. The direction on the left side of that term is estimated by the leading left singular vector, and the error is multiplied by the square of the size of the added term. For entries with mean zero and variance one, we identify the exact tail condition for the following statement to hold for every deterministic sequence of directions: the largest error over all sizes of the added term converges to the same fixed value determined by the limiting ratio of rows to columns. This condition (weaker than the existence of fourth moments) requires that the tail probability, multiplied by the fourth power of the threshold, to tend to zero. If it fails, convergence to this value already fails when both directions are coordinate vectors, even at a size fixed before the matrix is drawn. For regularly varying tails of order between two and four, the largest error tends to infinity.
Read the full abstract

Fluid Dynamics

An Aeroelastic Solver Integrating Reformulated-Vortex-Particle and Finite-Element Methods across Non-Conforming Interfaces

We present $\texttt{VarFlExI}$ (Variable Fidelity Unsteady Flow-FEniCS Exchange Interface), a modular aeroelastic solver for modeling two-way fluid-structure interaction (FSI) of flexible lifting surfaces. The framework employs a reformulated Vortex Particle Method (rVPM) to solve the incompressible Navier-Stokes equations without the need for computationally expensive volumetric meshing, while supporting variable-fidelity aerodynamic modeling using the $\texttt{FLOWUnsteady}$ framework. On the structural side, a Reissner-Mindlin plate formulation is discretized using the finite element method and integrated in time through the generalized-$\alpha$ method, with the nonlinear equilibrium equations solved using a Gauss-Newton procedure within the $\texttt{FEniCS}$ framework. Fluid and structural solvers are coupled through an explicit staggered partitioned scheme, ensuring conservation of virtual work for load and displacement transfer across the non-matching interface. To accommodate the multi-representative nature of the aerodynamic loads and geometry, the interface coupling employs separate work-conservative force and reverse-geometry transfer operators via a common intermediate interface. The framework is validated against water-tunnel experiments, demonstrating accurate prediction of the coupled aeroelastic response rather than independent validation of the constituent solvers. The computational efficiency of the meshless aerodynamic solver enables simulations at significantly lower computational cost while maintaining accuracy. The solver is further evaluated through sensitivity analyses and parameter studies spanning different flow conditions, structural properties, and coupling parameters.
Read the full abstract

Computer Vision

Few-Shot Video Recognition via Hierarchical Metric Learning

Few-shot action recognition (FSAR) aims to recognize unseen action categories with only a small number of annotated video samples. Recent works typically apply single-prototype supervision at the network output and fail to sufficiently exploit rich cross-frame global spatial information in videos. Even existing multi-level metric schemes only impose parallel prototype constraints on intermediate layers, without progressive supervision along the full feature pipeline, which results in limited generalization ability of the learned class prototypes. Inspired by this, we present a novel method, hierarchical metric learning for few-shot action recognition (HML-FSAR). First, a spatial-enhanced module is developed to capture cross-frame global spatial representations. Combined with temporal MHA, heterogeneous alignment, spatial-temporal feature fusion and dictionary learning modules, it constructs the complete feature processing pipeline. Second, a hierarchical metric learning (HML) strategy is embedded into HML-FSAR. Composed of center metric, alignment metric, contrastive metric, dictionary metric and prototype metric, HML imposes progressive multi-stage complementary constraints from frame-level representations to final class prototypes, so as to jointly optimize feature compactness, heterogeneous spatial-temporal alignment, inter-class discriminability and anti-noise robustness. The proposed HML-FSAR method is validated on five widely-used FSAR datasets, and experimental results fully demonstrate its effectiveness.
Read the full abstract

All authors Jiaxin Zhang, Haoran Gao, Xizhan Gao, Zihao Dong, Tingwei Wang, Sijie Niu

Artificial Intelligence

Uncensored Open-weight Models: Redistribution as the Persistence Layer

A rapidly expanding ecosystem of actors is removing built-in safety guardrails from open-weight AI models. We profile this ecosystem by identifying key producers, downstream reproductions, and emerging applications. Between January 2024 and March 2026, we identified 3,471 original uncensored models on HuggingFace, each repackaged an average of 2.4 times; three actors account for 52% of all 8,164 compressed redistributions. Once quantized and mirrored across separate accounts, formats, and registries such as Ollama, these models persist regardless of upstream removal and become easier to deploy downstream. Of the 1,643 identified GitHub applications integrating uncensored large language models (ULLMs), 25% were classified as explicitly malicious.
Read the full abstract

All authors 10a Labs: Juliette Garcia, Hailey May, Bobby McKenzie, David Pham, Matthew Swain, Joshua Valdez, Corie Wieland, Zachary Yahn

Instrumentation and Methods for Astrophysics

What is new in NuRadioMC: Multilayer Analytic Raytracer

NuRadioMC is a framework for the simulation of ultra-high-energy neutrino detectors that measure the radio signal emitted in neutrino-induced particle cascades used in different radio neutrino experiments. We present an extension to the analytic raytracing method for multilayered exponential refractive index models, where the full refractive index profile is described by a set of different single-exponential layers, supporting realistic medium descriptions at various experiment sites while maintaining computational efficiency. This approach also enables a realistic exponential refractive index description of the atmosphere and allows us to model signal propagation over non-smooth changes in the refractive index. This report outlines the fundamentals of the multilayered analytic raytracing method and shows some applied examples.
Read the full abstract

All authors Hannes Warnhofer, Sjoerd Bouma, Vital De Henau, Bryan Hendricks, Jakob Henrichs, Nils Heyer, Enrique Huesca Santiago, Annanay Jaitly, Philipp Laub, Anna Nelles, Martin Ravn, Felix Schlüter, Zeynep Su Selcuk, Karen Terveer, Christian Glaser, Christoph Welling, Philipp Windischhofer

Computer Vision

Cross-Domain Tracker Adaptation Without Target-Domain Labels via Vision-Language Agents

We present a system that uses a Vision-Language Model (VLM) as a diagnostic agent for adapting a detect-to-track pipeline to a new target domain without access to target-domain labels. Rather than optimizing against annotated metrics, the VLM directly inspects rendered tracking outputs, identifies visual failure modes, and recommends parameter updates through an iterative tuning loop. We first demonstrate that ground-truth-supervised hyperparameter transfer can be brittle. On MOT17->MOT20, applying a source-derived oracle configuration reduces mean HOTA by 0.090, from a target-domain ceiling of 0.357, to 0.267. Without using any target-domain labels, our VLM-based tuner recovers 67.8% of this lost headroom, finishing within 0.029 HOTA of the target ceiling; on the highest-density target sequence, it recovers up to 86.7%. We further show that label-free Bayesian optimization with handcrafted proxy objectives struggles under large domain shifts and can degrade configurations that are already strong. In contrast, the VLM tuner acts selectively: when its visual diagnosis reveals no clear failure mode, it declines to modify the configuration, preserving performance on easy transfers while improving hard ones. Finally, we characterize the conditions under which this approach succeeds, namely, when domain shift manifests through exposed detection-level parameters, versus where it is less effective, such as MOT17->DanceTrack, where the source oracle is already near-optimal.
Read the full abstract

Quantum Physics

Quantum Optimisation for Protein-Protein Interaction Network Alignment

Protein-protein interaction (PPI) network alignment combines topological and sequence information to identify conserved modules across species, but global alignment remains challenging: heuristics sacrifice optimality, while exact methods lack scalability. We model the alignment as a weighted maximum common induced subgraph problem and reformulate it through the modular product graph to a minimum-weight vertex cover on the complement, with node weights carrying sequence similarity. To solve this problem, we develop a hybrid framework combining kernelisation, branch-and-bound, and seven Quantum Approximate Optimisation Algorithm (QAOA) formulations. These formulations differ in how the cover constraints are enforced, from penalty terms in the cost Hamiltonian to mixers confined to the feasible subspace. For single round QAOA, we derive closed-form expressions for the expected cost of four circulant mixer variants, enabling performance characterisation without circuit simulation. Applied to synthetic and real-world networks reduced to KEGG pathways, the QAOA formulations achieve high topological conservation on the aligned core while at least maintaining biological conservation comparable to leading classical aligners, at the cost of reduced node coverage. Across selected KEGG pathways, the aligned subnetworks retain disease-associated proteins, preserving biologically relevant information. Cheaper formulations leave more edges uncovered, while enforcing feasibility in the mixer raises circuit depth by one to two orders of magnitude. Together, these results highlight the potential of quantum optimisation for PPI network alignment and the resource trade-offs that will shape its scalability as quantum hardware matures.
Read the full abstract

All authors Merle Stahl, Robert J. Banks, Matthias Traube, Josua Unger, Wolfgang Lechner, Jan Baumbach, Mhaned Oubounyt

Mathematical Physics

Representation Theory of Canonical Commutation Relations Arising from the Quantization of the Klein-Gordon Equation with an External Potential

We investigate the Weyl representation of the canonical commutation relations for a model describing a quantized massive scalar field under the influence of an external potential. The main problem is to determine whether the Weyl representation remains equivalent to or becomes inequivalent to the original one when the mass and/or potential are changed. This problem is reduced to the study of Schrödinger operators. It turns out that the Weyl representations are inequivalent when the masses differ. Moreover, when the masses are the same, the transition between equivalence and inequivalence occurs at the decay rate $-3/2$ of the difference between the potentials. This contrasts with the decay rate $-1$ that defines the short-range condition in scattering theory.
Read the full abstract

Cryptography and Security

Governing Bring Your Own AI: A Parameterized Maturity Model

Employees are increasingly using personally owned generative AI tools such as ChatGPT, Gemini, and Claude for their daily work. This practice is known as Bring Your Own AI (BYOAI), which is a distinct form of Shadow AI in which employee-authenticated personal accounts are used outside of enterprise identity and security controls. Existing frameworks were designed for AI tools managed by organizations, and their coverage does not extend to unmanaged AI tools used with a personal account. In addressing these issues, we developed a governance model through a systematic review of the literature that produces a risk taxonomy and a framework-engagement profile. We also developed a parameterized governance model that measures how much a level of governance maturity reduces residual risk. A five-level maturity ladder is coupled to a technical control architecture through a chain in which the coverage of the control layer influences the security outcomes. Our study of a curated corpus of 30 records (24 research studies and 6 framework documents) indicated that the most prominent categories identified were data exposure and compliance, and framework engagement was inconsistent. Three mutually supporting pillars (technical, governance, and human) were established to support safeguards. Additionally, the results of the model demonstrated that prohibition-based solutions will result in residual risk levels close to those achieved through baseline solutions. Under the specified parameterization, layered control-based solutions substantially reduce modeled exfiltration risk and increase enforceable coverage.
Read the full abstract

Machine Learning

PRICE: A Systematic Study of LLM Adaptation Choices for Bitcoin Price Forecasting

Cryptocurrency markets exhibit extreme volatility and non-stationary dynamics that challenge conventional forecasting methods. Although Large Language Models (LLMs) have shown promise for time series forecasting, the combined effects of adaptation choices remain largely unexplored in financial settings. This study introduces PRICE, a structured approach for adapting LLMs to short-term Bitcoin price forecasting. Built on a 4-bit quantized LLaMA-3 8B model, PRICE investigates how fine-tuning, numerical representation, prompting, inference, and decoding jointly influence forecasting performance. PRICE integrates Parameter-efficient fine-tuning with Low-Rank Adaptation (LoRA), Recursive multi-step inference, Integer-rounded numerical representation, Context-Task-Format (CTF) prompting, and Exact zero-temperature decoding. Ablation studies show that each component contributes to forecasting accuracy and reliability. LoRA enables efficient training on limited hardware, recursive inference improves accuracy, integer-rounded values reduce errors, CTF prompting outperforms Chain-of-Thought, Implicit Chain-of-Thought (iCoT), and few-shot prompting, and zero-temperature decoding improves stability during recursive forecasting. Comparative evaluation against eight transformer-based and time-series foundation models shows that PRICE achieves the lowest forecasting errors on both validation and test sets while maintaining robust performance across evaluation periods. Despite being based on a model primarily pretrained on text rather than time-series data, PRICE achieves competitive or superior performance relative to specialized foundation models. These findings demonstrate that adaptation choices critically determine the accuracy and robustness of LLMs for numerical time-series forecasting.
Read the full abstract

Computer Vision

Measured Sliders: Learning Continuous Controls from Differentiable Image Measurements

Continuous sliders are useful only when coefficient changes produce predictable image changes. Yet most diffusion sliders derive their axes from text or learned representations, leaving their scales disconnected from observable image properties. Consequently, we cannot tell in advance which attributes are learnable, compare control strengths directly, or anticipate interference when multiple controls are combined. We propose Measured Sliders, a framework that defines continuous controls through closed-form differentiable image measurements. A common measurement space unifies the pipeline. Before training, an observability test identifies usable supervision. During training, a measurement-guided objective learns target movement while suppressing non-target changes. After training, decoded calibration expresses controls in comparable units of realized image change. Multiple LoRA branches are stored in one checkpoint and composed without training on joint activations. Across SDXL and FLUX.1-dev, the resulting controls are ordered, selective, and composable. On 553 prompts, lighting direction reaches rho = 0.995 and 98.9% monotone sweeps. A five-attribute checkpoint achieves average selectivity 2.59, compared with 1.50 for the strongest baseline, and preserves every requested direction in 96.7% of pair and 86.1% of triple compositions. The observability test also separates every subsequently successful measurement from the failed candidate. Overall, image-space measurement provides a common basis for learning, diagnosing, calibrating, and composing continuous generative controls.
Read the full abstract

Machine Learning

Hessian-Informed Conformation Augmentation for Potential Energy Surface Curvature in Machine Learning Interatomic Potentials

While machine-learning interatomic potentials (MLIPs) have successfully learned potential energy surfaces (PES) and atomic forces, many practical applications, such as vibrational analysis and transition state search, rely heavily on the PES Hessian. Yet standard MLIPs are trained on energy and forces only, and using the Hessian as an additional training objective requires backpropagation whose memory cost grows quadratically with the number of atoms. To address these limitations, we propose two Hessian-derived data augmentation schemes: isotropic Gaussian displacement (UniAug) and normal mode-weighted displacement (ModeAug). Both methods utilize simple Taylor expansions, generating physically grounded molecular structures with highly accurate energy and force labels. By preserving the underlying MLIP architecture and training procedure, this approach contributes to improving energy and force predictions in various benchmarks. Comprehensive evaluations across non-equilibrium and equilibrium datasets demonstrate that our approach enhances model accuracy where reference forces are large and training data are limited.
Read the full abstract

Artificial Intelligence

Substrate-Aware AI Agents: Execution Context as a First-Class Input

Autonomous AI agents increasingly select actions in environments whose memory, execution-time, runtime, compute, and operational constraints determine what counts as a suitable plan. We call the absence of this execution context from an agent's planning state substrate blindness. We test this general proposition through numerical code generation, where selected implementation choices and operational consequences are directly observable. Three frontier model configurations–Anthropic Claude Opus 5, OpenAI GPT-5.6-Sol, and Google Gemini 3.7 Flash–generate code for a high-dimensional pairwise Euclidean-distance task either from the task alone or with a 128 MB RAM and 10.0 s wall-time contract. Contract disclosure reduced measured peak process memory in 13 of 14 executable index-aligned task-only versus contract-disclosed comparisons and reduced mean wall time in all three cohorts, making execution up to 3.1x faster. Across the audited corpus, disclosure produced structural code changes including bounded blocking, float32 retention, upper-triangle traversal, and in-place or memory-mapped buffers. At a tighter 96 MB contract, independently sampled contract-disclosed cohorts achieved correct-and-within-budget outcomes of 4/5 for Claude Opus 5, 5/5 for GPT-5.6-Sol, and 3/5 for Gemini 3.7 Flash, compared with task-only outcomes of 0/5, 1/5, and 0/5; cohort mean MaxRSS and wall time were 49-74% and 35-64% lower than their task-only references. These results establish a controlled proof of concept for substrate-aware agent planning: a minimal execution contract induces proactive structural adaptation in generated programs, shifting computation away from unconstrained allocations and substantially improving observed resource-time profiles before execution.
Read the full abstract

High Energy Physics (Phenomenology)

New measurements in leptonic kaon and pion decays: experimental and theoretical perspectives

We study the future prospects for precision measurements of purely leptonic charged-current kaon and pion decays. Together with the ratio of leptonic widths $R_K^\nu=\Gamma(K\to e\nu)/\Gamma(K\to\mu\nu)$, we consider the single ratios $R_{K\pi}^{\ell=\mu,e} = \Gamma(K \to \ell\nu) / \Gamma(\pi \to \ell\nu)$, for which many experimental uncertainties cancel, as well as the double ratio $R_{K\pi}^e / R_{K\pi}^\mu$, which provides a particularly sensitive probe of new physics. This study is timely in light of two converging developments. On the experimental side, a new conceptual design for a dedicated setup based on a slow-extracted proton beam and a multi-year data-taking program, is expected to achieve a precision at the few $10^{-4}$ level. On the theory side, recent lattice-QCD calculations have reached a comparable level of accuracy. Finally, we assess the impact of these next-generation measurements on searches for physics beyond the Standard Model and show that they can provide constraints that are either competitive with or complementary to existing bounds.
Read the full abstract

All authors Lukas Allwicher, Stefan Alexandru Ghinescu, Federico Mescia, Tommaso Spadaro, Lucine Tabatt

Earth and Planetary Astrophysics

Planetary gas gaps and kinematic signatures in the planet forming disk around WISPIT 2

The giant planet formation process in disks of gas and dust surrounding young stars is observationally still poorly constrained. Direct detection of protoplanets within the disk remains limited with current facilities. Indirect observational evidence of protoplanets through substructures or deviations from Keplerian rotation in the gas remains ambiguous in absence of detected planets. The lack of the combined detection of substructures and their corresponding planets makes it challenging to connect the planet formation process to the host environment. Observations with the Atacama Large Millimeter Array (ALMA) reveal clear detections of gas gaps and kinematic signatures in the WISPIT 2 protoplanetary disk that are cospatial with the locations of previously detected giant protoplanets. The newly identified gas gaps, observed in both the 12CO integrated intensity map and its rotation curve, correspond with the previously identified gaps in scattered light. Their morphologies are shown to be consistent with eccentric gaps, as expected from planet-disk interaction models of massive planets. The lack of eccentricity in the outer disk can be explained if a third planet is present in the system. The observations of WISPIT 2 provide a long-sought empirical bridge between kinematic signatures, gas gaps and giant planet formation in disks.
Read the full abstract

All authors Nienke van der Marel, Jochen Stadler, Andres F. Izquierdo, Ruobing Dong, Bertram Bitsch, Jake Byrne, Richelle van Capelleveen, Chloe Lawlor, Christian Ginski, Matthew Kenworthy.

Information Theory

Latency-Optimal Geo-Distributed Storage over Structured Networks

We study latency-optimal file assignment in geo-distributed storage systems modeled as weighted graphs, where edge weights represent communication delays and each node stores one (possibly coded) file. Our goal is to minimize the average time required to retrieve an original file, taken uniformly over all nodes and files. We show that for every fixed number of files $k \geq 3$, computing a latency-minimizing assignment is NP-hard via a reduction from the domatic number problem. On the positive side, we identify natural network topologies that admit uncoded, structured optimal assignments in which, for every node, one can choose its $k$ closest nodes, including itself, so that they store distinct original files. We prove that every weighted tree, certain weighted cycles, and unit-weight graphs with sufficiently large minimum degree admit such assignments. For these graph classes, we provide efficient algorithms to construct latency-optimal file assignments.
Read the full abstract

Artificial Intelligence

ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs

Mixture-of-Experts (MoE) architectures provide an efficient paradigm for scaling large language models (LLMs), yet fixed top-k routing activates the same number of expert slots for every token, causing substantial redundant computation. Existing expert-skipping methods often rely on router confidence, calibration data, or additional training, and therefore cannot reliably estimate the actual contribution of routed experts. To this end, we propose ACE, a training-free, calibration-free, and checkpoint-preserving framework for token-adaptive expert skipping in MoE-based LLMs. ACE contains two complementary components: 1) Global Spectral Proxy (GSP), which estimates global transformation capacity from the coupled gate, up, and down projections together with RMSNorm scaling; and 2) Router-Conditioned Refinement (RCR), which constructs expert-specific direction prototypes from centered router weights and evaluates expert responses along routing-preferred directions. During inference, ACE combines both estimates with runtime router gates and skips an expert slot only when both views identify it as low-contribution, while always retaining the top-1 expert. All expert statistics are computed offline, leaving only table lookups and lightweight scalar operations online. Extensive experiments across three MoE-based LLMs and eight benchmarks demonstrate that ACE consistently outperforms existing static and dynamic baselines, with increasingly pronounced advantages under aggressive expert skipping. For instance, at a 50% skipping ratio on Qwen3.6-35B-A3B, ACE reduces WikiText-2 perplexity by 7.96% and improves average downstream accuracy by 4.15 percentage points over the strongest competing method.
Read the full abstract

All authors Zukang Xu, Zhixiong Zhao, Xing Hu, Jiangyong Yu, Houji Wen, Jun Li, Zhe Jiang, Dawei Yang

Artificial Intelligence

CABAL: Multi-Agent Simulacra for Tracing the Effects of Collusive Bidding in Peer Review

Recent reports during the AAAI-27 review cycle highlight the risk of reviewers coordinating bids for reciprocal assignment advantage. Prior work treats bidding, reviewer assignment, and review manipulation as separate stages, leaving the lifecycle effects of collusive bidding unclear. Real-world analysis is further constrained by typically unobservable collusive intent and the lack of counterfactuals for the same conference. Motivated by this gap, we introduce \alg, an end-to-end multi-agent simulacra framework for studying reviewer assignment integrity by holding the conference environment fixed and configuring LLM-driven reviewer agents with honest or collusive policies. We further develop an affinity-guided collusive bidding strategy that uses mutual reviewer-paper affinities to construct collusion rings and select target papers, producing expertise-consistent rather than arbitrarily targeted attacks. Controlled experiments show that collusive bidding more than doubles target-paper capture and that assigned colluders score target papers about two points higher than honest co-reviewers, while conference-wide effects remain comparatively modest. Evaluated bid-phase detectors provide only limited evidence of collusion: in a fixed-triplet detector stress test, native positive-bid graphs are confounded by benign affinity, while a Very-High-only diagnostic view enables precise but low-coverage local recovery.
Read the full abstract

All authors Jicheng Zhou, Kemou Li, Kahim Wong, Zheyuan Li, Zhuan Shi, Fengpeng Li, Haiwei Wu, Jiantao Zhou

Quantum Physics

TETRIS-Q: Tiling-based Effective Transient-fault Reduction on Interleaved Superconducting Qubits

The struggle of the hour in quantum computing research is achieving effective suppression of the error mechanisms induced by the interaction of external radiation with superconducting quantum devices. Despite the rapid advancements in quantum error correction (QEC) of recent years, radiation-induced faults are yet to be fully addressed. These events are known to be the cause of simultaneous correlated defects in qubits that lie onto a single substrate, ultimately jeopardising QEC code effectiveness. In this paper, we propose to selectively combine substrate-level phonon barriers and QEC interleaving via a planar-mesh tiling algorithm, TETRIS-Q, reaching efficient and effective suppression of radiation events. Our cross-layer solution comes at no extra cost in terms of QEC code execution or decoding time. We model and simulate radiation-induced transient faults over a plethora of barrier and QEC interleaving configurations. Through more than 51 million quantum circuit simulations, we show peak logical error reductions of more than $99.8 \%$, together with an $80\%$ reduction of the observable transient duration with permeable barriers. We find that sparser tiling can reach comparable performance to single qubit tiling, prompting cost reductions of upwards of $87 \%$ in barrier tracing. By leveraging independent QEC code interleaving, we measure up to one order of magnitude average logical error rate reductions without the use of permeable barriers, and up to three orders of magnitude with the joint usage of barriers.
Read the full abstract

All authors Marzio Vallero, Gioele Casagranda, Flavio Vella, Paolo Rech

Quantum Physics

The marginal is pretty good

One-shot information theory measures often require an optimization over states, but the form of these optimizers can be complicated or depend on the initial problem in nonlinear ways. In this note, we show that in many instances using the marginal instead of the optimal state is sufficiently good and only changes the result by a small factor. We prove that for the Petz–Rényi and sandwiched–Rényi divergences of order $\alpha\in(0,1)$, replacing the optimizing state on $B$ by the marginal $\rho_B$ results in a multiplicative overhead of at most $1/\alpha$.
Read the full abstract

All authors Lukas Schmitt, Gui-Long Jiang, Shi-Bing Li, Joseph M. Renes

2609.05225 · 4 Sep · 12 pages, Gui-Long Jiang and Shi-Bing Li added as authors, Theorem 3.1 now holds for $\alpha \in(0,1)$ (instead of $[1/2,1)$), Theorem 4.1 and 4.2 show the same statement for the sandwiched Rényi divergence. This generalizes the previous result that only worked for $\alpha=1/2$ and special classes of states

Computer Vision

First Things First: Teaching LLM-Based Agents to Prioritize Must-Haves before Nice-to-Haves

Recent progress in multimodal large language models (MLLMs) has fueled significant enthusiasm in their potential to act as autonomous agents for real-world tasks. However, scenarios requiring agents to fulfill users' complex, structured requirements remain largely underexplored. In this work, we examine reasoning tasks under three distinct requirement scenarios: (i) Must-have requirements uniquely determine a unique feasible solution; (ii) Multiple answers satisfy the must-have requirements and are prioritized via the nice-to-have requirements; and (iii) No candidate solution satisfies the must-have requirements, in which case the agent should abstain from generating a response. We evaluate state-of-the-art MLLMs on 3,649 carefully constructed problems that reflect realistic service scenarios, including e-commerce, booking, and map-based or ride-hailing. Our evaluation reveals that existing MLLMs exhibit catastrophic failures in all scenarios. They frequently misinterpret task requirements, violate must-have requirements, and produce invalid solutions. To address this critical gap, we propose First Things First Reinforcement Learning FTF-rl that explicitly optimizes reasoning over multi-priority user requirements. Experimental results show that our method substantially improves the task success rate compared to strong baselines. Moreover, FTF-rl yields general effectiveness on popular logical and mathematical reasoning tasks, including LogicVista, MathVision, and InfoQA. Our findings suggest that enhancing requirement-aware reasoning capability provides a simple yet effective pathway to improve generalization of MLLM agents. Code and dataset are available at https://github.com/claire62/FTF-RL.
Read the full abstract

All authors Tianjie Ju, Xinyue Xu, Wanxuan Sun, Lingxiao Diao, Gongshen Liu, Zhuosheng Zhang, Cheng Yang

Machine Learning

FedDRAW: Federated Dual Reputation Annealing Weighting for Heterogeneous Multi-Institutional Chest Radiograph Classification

Artificial intelligence models are promising for medical diagnosis, but they require large numbers of unbiased data, which in medicine are distributed across hospitals and cannot be centralized to protect patient privacy. Federated Learning (FL) addresses this, since hospitals train one shared diagnostic model while patient data remain local. Training proceeds in communication rounds, in which each hospital trains the shared model locally and returns it to the server for merging by weighted average. This aggregation weight determines whose institutional knowledge shapes the result. Federated averaging (FedAvg) sets it in proportion to local sample count, so a small but informative hospital is permanently assigned a small influence, andl argest clients could dominate the global model even when they are less informative. We propose Federated Dual Reputation Annealing Weighting (FedDRAW), a server-side aggregation method that combines a data-size prior with the cosine similarity between client and global parameters under two coupled annealing schedules. An inner schedule shifts client reputation from the size prior towards similarity. An outer, deferred annealing schedule on the softmax inverse temperature keeps the weighting selective in the early and middle rounds and relaxes it to uniformity at convergence. We evaluate FedDRAW on 12 simulated client-partition scenarios of two chest radiograph datasets (CheXpert and ChestMNIST), against seven federated baselines under identical local training settings. FedDRAW achieved the highest average rank among all eight methods under both AUC and the geometric mean (GM) of sensitivity and specificity, which a Friedman test with Nemenyi post-hoc analysis confirmed to be a statistically significant difference between the methods. Scheduling two signals, rather than fixing the weights by sample count alone, could enable less biased diagnostic models.
Read the full abstract

Quantum Physics

A double-resonator coupler for high-fidelity two-qubit gates between superconducting qubits

Tunable couplers have enabled two-qubit gate fidelities in superconducting quantum processors to approach $99.9\%$, yet simultaneously suppressing residual interactions and maintaining flexible qubit-frequency allocation remain central challenges for scaling. Here, we propose a double-resonator coupler (DRC) consisting of two resonators interconnected by a single Josephson junction and a capacitor. The hybridized resonator modes provide two mediated exchange paths whose interference controls the qubit-qubit interaction. The DRC enables complete cancellation of residual $ZZ$ interaction for qubit-qubit detunings well outside the straddling regime, even without direct inter-qubit capacitance, thereby relaxing constraints on frequency allocation and qubit placement. Away from the idle point, the same circuit provides a strong $ZZ$ interaction of approximately $70\,\mathrm{MHz}$, enabling a $20\,\mathrm{ns}$ controlled-Z gate with simulated coherent infidelity below $10^{-5}$. These results establish the DRC as a flexible single-junction coupler architecture for high-fidelity superconducting quantum processors.
Read the full abstract

All authors Seunghyeon Jin, Seungha Woo, Shinyoung Hwang, June-Young M. Lee, Jeongmin Shim, Sunje Kim, Dae Seok Han, Jaeho Shin, Eunjong Kim

Computation and Language

A Verifier-Guided Explainable Reasoning Framework with Gold-Anchored QLoRA, Task-Aware Mixture-of-Experts, and Group-Relative RLVR

Large language models (LLMs) show strong reasoning ability, but their explanations can remain inconsistent, weakly grounded, or difficult to verify. We propose a verifier-guided explainable reasoning framework for transparent educational question answering that combines gold-anchored QLoRA, task-aware symbolic routing, and group-relative RLVR. Qwen2.5-3B-Instruct is first adapted with field-weighted QLoRA supervision anchored to authoritative answers. A lightweight router then assigns logic problems to a FOL/Z3 verifier and physics problems to a formula- and unit aware symbolic solver. Verifier feedback is further used to support candidate evaluation, self-revision, and reward construction during RLVR. Candidate responses are evaluated along three complementary dimensions: P1 for answer correctness, P2 for evidence or unit consistency, and P3 for reasoning depth and explainability. At inference, gold-free self-consistency aggregates multiple candidate responses before an optional question-only physics verifier performs conservative system-level correction. On 438 held-out examples, RLVR increases P3 from 50.68% to 72.20%, while hybrid P1 remains approximately stable at 55.94%. Self-consistency improves model only P1 from 48.86% to 50.23%, with symbolic verification providing the remaining hybrid gain. These results indicate that RLVR primarily strengthens explicit reasoning structure, while symbolic verification complements the neural policy by improving answer reliability at the system level.
Read the full abstract

All authors Thi Kim Trang Vo, Nam Tien Le, Thi Kim Nguyet Vo, Minh Khang Tran, Duy Phuong Tran

Number Theory

Local Jacquet-Shalika integrals and modifying factors

Over any local field of characteristic zero, we develop a local theory of Jacquet-Shalika integrals that were originally introduced by Jacquet and Shalika to study exterior square local $L$-factors for $GL_m$. In particular, at least in the archimedean case, we establish a complete functional equation matching Artin local $\varepsilon$-factors. Central to the argument is a novel comparison strategy between Jacquet-Shalika integrals and open-orbit integrals associated to the Shalika subgroup.
Read the full abstract

All authors Dihua Jiang, Dongwen Liu, Binyong Sun, Fangyang Tian

Mesoscale and Nanoscale Physics

Robustness of RKKY interactions across a Weyl node-annihilation transition

Weyl semimetals (WSMs), with their unique topological properties and distinct electronic structure, exhibit intriguing properties when either time-reversal or inversion symmetries are broken. In this work, we consider the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction between magnetic impurities in time-reversal symmetry-breaking WSMs. We derive analytical expressions for the full RKKY exchange tensor in arbitrary two-band spinful lattice systems. Our approach reveals both Heisenberg, anisotropic Ising and Dzyaloshinsky-Moriya terms, which can be calculated by energy-integrating real-space Green's functions across the entire Brillouin zone, with the band edge acting as a natural energy cutoff. We apply this framework to study a two-band tight-binding model for a time-reversal symmetry-breaking WSM that interpolates between a Weyl phase with well-separated chiral nodes and a quadratic band-touching semimetal phase. Remarkably, the spatial profile, magnitude, and anisotropic tensor structure of the exchange couplings remain persistent across the node-annihilation transition. This topological robustness reveals that short- and intermediate-range RKKY interactions are mediated by the global, Brillouin-zone-integrated quantum metric of the full valence band rather than being strictly dictated by local low-energy Berry curvature monopoles. These findings demonstrate the necessity of full-band tight-binding formulations when predicting real-space magnetic interactions, providing key insights for electric-field tuning of magnetic anisotropy and constructing realistic models of heavy-fermion and Weyl-Kondo semimetals.
Read the full abstract

Optics

Screening bolt loosening in a four-bolt plate with global FRF correlation and local FRAC maps from full-field laser Doppler vibrometry

Full-field laser Doppler vibrometry (LDV) can reveal how bolt torque loss redistributes a frequency response function (FRF) over an entire structure rather than only at a few sensor positions. This work presents a screening procedure for a four-bolt aluminum plate using pointwise amplitude and phase exports of scanned H1 FRFs. Candidate resonances were selected from the all-tight spatial RMS spectrum and tracked in four single-bolt 0 Nm cases. Seven baseline responses remained trackable within prescribed group-specific search windows and were retained as experimental resonance groups RG1-RG7; a dimensionless plate-frequency coefficient was reported alongside each measured frequency. Global changes were quantified using amplitude-only and complex, phase-retaining modal-assurance dissimilarities, and local changes were evaluated using matched-window FRAC-deficit maps and a normalized hotspot-area fraction. The retained groups separate into low-, intermediate-, and high-distortion responses, while the local maps distinguish compact joint-centered changes from distributed FRF redistribution. The workflow therefore links global FRF distortion with spatially resolved interpretation without relying on a trained classifier or specimen-specific node labels.
Read the full abstract

All authors Berkay Kullukcu, Robin Pianowski, Mehmet Sait Özer, Ercan Altinsoy, Dina Hannebauer

Soft Condensed Matter

Coarse-graining far-from-equilibrium dynamics using oblique projectors

We generalize the Mori-Zwanzig projection operator formalism by introducing an oblique projector using biorthogonal variables to coarse-grain the dynamics of systems far from equilibrium, such as active or living matter. Improving on previously used orthogonal projectors, the oblique projector correctly decomposes the emergent interactions of the coarse-grained particles into systematic and stochastic contributions. We show that the formalism directly connects to established results in non-equilibrium linear response theory and adiabatic perturbation theory. Based on this formalism, we develop a coarse-graining algorithm which we apply to extract the emergent non-equilibrium chiral dynamics of a passive tracer in a chiral active bath. Our theory correctly predicts the odd transport properties and a confinement-induced shift of the chiral frequency of the tracer.
Read the full abstract

Astrophysics of Galaxies

Stellar Population and Dynamical Modeling of Galaxies with Kinematically Misaligned Components: Age and Metallicity of Structural Components

We present stellar population and orbit modeling of two galaxies with kinematically misaligned components, PGC 35706 and LEDA 2220522, extending previous Schwarzschild dynamical models. Using spatially resolved age and metallicity maps from NBursts full spectral fitting of MaNGA data with the E-MILES grid, we solve a regularized bounded linear inverse problem that assigns light-weighted SSP-equivalent age, metallicity, and mass-to-light ratio to orbital cells in the $(R,\lambda_z)$ plane. Component means and radial profiles use projected mass within the actual MaNGA field of view; profile segments with local component fractions below $1\%$ are treated as weakly constrained. In LEDA 2220522, the counter-rotating disk is younger than the co-rotating disk, $1.32\pm0.04$ versus $1.61\pm0.06$ Gyr, and more metal-rich, $[\mathrm{M/H}]=-0.450\pm0.043$ versus $-0.613\pm0.024$. Its rotation agrees with that of the H$\alpha$ ionized gas, supporting late supply of chemically enriched retrograde gas without requiring multiple accretion episodes with alternating angular momentum. In PGC 35706, the counter-rotating disk contains $8.8\%$ of the mass in the observed field and is formally older than the co-rotating disk by $1.36\pm0.51$ Gyr and more metal-poor by $0.130\pm0.046$ dex. These differences are preliminary because the projected data provide limited support. Fifty Monte Carlo realizations confirm stability of the projected maps against statistical noise, but not uniqueness in individual phase-space cells. The stellar population and orbit approach is therefore applicable to MaNGA data when local projected component fractions and systematics from regularization and the dynamical model are explicitly considered.
Read the full abstract

All authors V. S. Goradzhanov, K. O. Parfenov, D. D. Mishurin, O. K. Sil'chenko

Algebraic Geometry

Filtrations on D-modules and multiplicities of roots of Bernstein-Sato polynomials

In this paper, we relate multiplicities of Bernstein–Sato-type polynomials with respect to a function f to several singularity invariants. First, we introduce certain “mod” b-functions and show that they characterize the weight filtration on the localization of a simple algebraic holonomic D-module along f, and we provide an algorithm to compute them. Second, we show that they can be approximated by the multiplicities of roots of power b-functions (i.e., the b-functions with respect to powers of f). Furthermore, we give a sharp upper bound for the Hodge level of elements given by a certain sum of such multiplicities. Next, we give an effective asymptotic solution to the Gelfand problem by determining an explicit threshold after which every integer shift of a root of b_f(s) is a pole of the Archimedean zeta function of f. We also show that the order of these poles is equal to the nilpotency index of the logarithmic monodromy operator, which we further express as the limit of the multiplicities of roots of power b-functions. We define several filtrations, relating them to the weight and Hodge filtrations, based upon which we leave some open questions that we address in the affirmative in the case when f has a homogeneous isolated singularity, or it is a hyperplane arrangement, or a semi-invariant on a spherical variety. We give several immediate applications to our results, including a solution to a question of Torrelli on whether 1/f lies in the intersection complex of the hypersurface of f if and only if -1 is a simple root of b_f(s).
Read the full abstract