Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 4 September

Quantum Physics

Nitrogen Vacancy Centers in Diamond for Quantum Biosensing: Magnetometry Techniques, Platforms and Applications

Quantum sensing using nitrogen-vacancy (NV) centers in diamond has emerged as a powerful platform for detecting ultra-low magnetic fields under ambient conditions. Owing to their long spin coherence times, optical addressability, and compatibility with aqueous environments, it has found widespread applications in biosensing and bio-imaging. This review presents the fundamental principles and recent advances in NV-based quantum magnetometry for biosensing applications, with a particular focus on measurements in aqueous medium and at cellular and molecular length scales. We discuss the underlying spin physics of NV centers and highlight two primary detection modalities: optically detected magnetic resonance (ODMR) and T1 relaxometry-based sensing and how these approaches aids in the detection of both static magnetic fields and dynamic magnetic noise arising from biological processes. The review explores key application areas, including nanoscale nuclear magnetic resonance (NMR), monitoring of neural activity, detection of abnormal or rogue cells using NV-based platforms etc. In addition, strategies for enhancing sensitivity, such as surface functionalization of nanodiamonds, femtosecond (fs) laser-written photonic structures, and integration with microfluidic and lab-on-chip systems have also been discussed in depth. We also address the critical challenges, including surface-induced decoherence, charge-state instability, and signal-to-noise limitations in biofluids associated with NV-based biosensing applications. Finally, we outline future prospects, highlighting how NV-based magnetic biosensing provides a promising pathway for translating quantum sensing technologies into practical biomedical applications.
Read the full abstract

All authors Sirsendu Ghosal, Umakant Prajapati, Sachin Negi, Saifian Farooq Bhat, Thejas B, Vibhav Bharadwaj

Statistical Mechanics

Thermodynamic learning

We discuss the possibility to train a thermodynamic system, whose micro-variables are fully determined through the Hamiltonian by the usual statistical mechanical rules, to perform tasks such as memorization and generalization. Training is achieved by the application of suitable external fields, playing the role of data. At variance with conventional machine learning, no other logical or algorithmic rules are introduced. The system is amenable, in principle, to exact analytical calculations. We specialize this general approach to a prototypical Ising system with annealed dichotomous couplings and study its learning ability. Results indicate excellent memorization and good generalization capacity already for small systems, and a tendency to improve with system size.
Read the full abstract

All authors F. Corberi, S. dello Russo, G. Messuti, S. Scarpetta, L. Smaldone

High Energy Astrophysical Phenomena

Evidence of Orbital Evolution of Gigahertz-Peaked Spectra in Binary Millisecond Pulsar

We present the first systematic investigation of the continuous orbital evolution of the Gigahertz-Peaked Spectum turnover frequency ($\nu_{peak}$) in the binary millisecond pulsar, PSR J2144$-$5237. Using approximately 50 hours of densely sampled observations spanning the $\sim10$-day orbit with the Parkes Ultra-Wideband Low frequency receiver, we obtained continuous frequency coverage from 704 MHz to 4 GHz, enabling robust measurements of the spectral evolution throughout the binary orbit. We detect a clear and systematic modulation of the turnover frequency with orbital phase for both the $1^{st}$ pulse and $2^{nd}$ pulse components, with both emission components exhibiting remarkably similar evolutionary trends despite originating from distinct emission regions within the pulsar magnetosphere. The correlated evolution of the two pulse components strongly suggests that the observed spectral variability is governed by propagation through the intra-binary environment. The large orbital modulation of turnover frequency can be explained by cyclotron resonance absorption in magnetized plasma associated with the companion, while alternative absorption mechanisms are unable to reproduce the observed behaviour. Modelling the orbital evolution of the turnover frequency within the cyclotron absorption framework, can constrain the electron density distribution in the pulsar wind, estimate the magnetic field strength of the companion, and place limits on the geometry of the absorbing plasma and the binary system. This study establish orbital evolution of the Gigahertz Peaked Spectum as a powerful new diagnostic of the magnetized intra-binary medium and provide a hitherto unexplored probe of pulsar wind-companion interactions in eclipsing millisecond pulsar binaries.
Read the full abstract

Quantum Physics

One Bit of Collective Information Is Worth N ln 2 Bits of Local Information in a Many-Body Quantum Battery

Charging a quantum battery through a collective non-adiabatic stroke stores energy in a shared bosonic mode, but part remains locked in correlations with the collective spin and is inaccessible to cyclic unitaries acting on the mode alone. A demon holding one bit can unlock this energy, quantified by the daemonic ergotropy. We investigate the value of one bit and its dependence on where the information is obtained. Two protocols are compared at matched stored energy and matched information, using balanced two-outcome measurements carrying exactly one bit. We find that one bit about the collective coordinate unlocks N ln(2) times as much work as one bit about a single ion. For three stored-energy settings and N = 4-24, the measured scaling exponent is 0.990 +/- 0.043, while double extrapolation gives a prefactor of 0.69298 +/- 0.00044, within 0.02% of ln(2). To leading order, the daemonic gain equals nu mu^2 times the between-outcome variance of Jx, verified numerically to 1.3%. A balanced single-ion measurement resolves 1/4 of this variance, whereas a balanced collective split resolves (ln(2)/4)N. The microscopic origin of the prefactor remains open; a Gaussian median-split estimate of 1/(2pi) is excluded by 9%. One bit recovers a constant fraction of the locked energy independent of N and yields roughly twice as much work per bit as a complete readout, indicating strong diminishing returns. We also show that unnormalized gain comparisons can reverse the conclusion and that the break-even ion number does not collapse onto the Dicke superradiant threshold when the mode frequency is varied.
Read the full abstract

Mesoscale and Nanoscale Physics

Interlayer Exciton Condensate Stiffness Is Non-Monotonic in Quantum Metric

Identifying the origin of the superfluid stiffness of electron-electron and electron-hole pair condensates is an important issue in flat-band physics. Here, we study the stiffness of bilayer exciton condensates using exact diagonalization and realistic Coulomb interactions across a wide variety of flat Chern band systems, including Landau levels, mixed Landau levels, and moiré bands. We find that stiffness is non-monotonic in the trace of the quantum metric and that it develops peaks when the band wavefunctions are engineered to be similar to those of Landau levels. The stiffness predicted by mean-field theory agrees quantitatively with exact diagonalization in these optimal cases, but systematically overestimates it otherwise. Flat bands with identical quantum geometry tensors can exhibit substantial differences in stiffness. The stiffness of condensates formed between moiré flat bands, which typically have strong variations in Berry curvature and quantum metric across their Brillouin zones, tends to be larger when the bands have non-zero Chern numbers and can be larger than that of Landau levels. Our results reveal a behavior that is richer than that suggested by simple geometric bounds and provide new guiding principles for the design of robust flat-band condensates with large superfluid stiffness.
Read the full abstract

All authors Bo Zou, Yitong Zhang, Siddhartha Sarkar, Shi-Zeng Lin, Allan H. MacDonald, Kai Sun

Instrumentation and Methods for Astrophysics

Validating the SCALES DRP: A Performance Study of the Medium-Resolution IFS Mode

We present a performance verification of the medium-resolution integral field spectrograph (IFS) mode of the SCALES Data Reduction Pipeline (SCALES-DRP) using data obtained during recent cryogenic cooldown testing. SCALES (Slicer Combined with Array of Lenslets for Exoplanet Spectroscopy), scheduled for deployment on the Keck II Telescope in late 2026, will be the first facility-class high-contrast thermal-infrared IFS, operating over the 2-5 micrometer wavelength range. The SCALES-DRP, developed within the Keck Data Reduction Pipeline Framework and implemented in Python, converts raw detector reads into calibrated three-dimensional datacubes using rectification matrices derived from position-dependent monochromatic lenslet point-spread functions, and supports both optimal and chi-sqaure-based spectral extraction to maximize signal-to-noise and extraction fidelity. We present laboratory verification results for the medium-resolution IFS mode, which combines a 0.34x0.36 arcsec lenslet subarray with an image slicer to achieve spectral resolutions of R ~ 2,000 - 5,000, focusing on the performance of the rectification matrices and the robustness of the extraction algorithms. These results demonstrate the readiness of the SCALES-DRP for instrument commissioning and early science operations. The pipeline is publicly available on GitHub.
Read the full abstract

All authors Athira Unni, Steph Sallum, Max Brodheim, Rosalie McGurk, Andy Skemer, Matthew Brown, Peyton Benac, Michael P. Fitzgerald, Deno Stelter, Dusty Reichwein, Aaron Hunter, William T. S. Deich, Isabel Kain, Mackenzie Lach

Algebraic Geometry

Moduli of Conics on General Plucker Linear Sections of Grassmannians

Let $G=\operatorname{Gr}(k,V)$ be a complex Grassmannian in its Pl$\"u$cker embedding, and let $Y_E$ be a general codimension-$r$ linear section. We study the open Hilbert scheme $R_2(Y_E)$ of smooth conics using the kernel and span of a conic. These define a flag variety $B=\operatorname{Fl}(k-2,k+2;V)$ and a relative Grassmannian of three-planes in a rank-six bundle. For $0\le r\le3$, we prove that $R_2(Y_E)$ is nonempty, smooth, irreducible and rational of dimension $e=2\dim V+k(\dim V-k)-3-3r$, with birational model $B\times\operatorname{Gr}(3,6-r)$. For $r\ge4$, a general-position argument gives a componentwise birational correspondence with a rank-three degeneracy locus and two smooth projective incidence resolutions. Every component meets the locus of smooth minimal-envelope conics; no additional density assumption is needed for general $E$. If $0\le e\le r$, the degeneracy locus itself is smooth. In dimension zero we obtain a top Chern class formula, recovering the known counts $1225$ and $1176$ by exact localization.
Read the full abstract

Computer Vision

Bridging Modalities and Tasks: A Unified Hierarchical ViT for SAR-to-Optical Translation and Semantic Segmentation

Synthetic Aperture Radar (SAR) remains available through cloud and darkness, yet speckle and non-intuitive scattering make it hard to inspect and interpret. SAR-to-optical (S2O) translation renders radar observations as optical-like images for visual access, while semantic segmentation on SAR remains necessary since land-cover decisions cannot rely on synthesised pixels. We propose BMT, a hierarchical Vision Transformer coupling both tasks on a shared encoder via multi-scale one-hot conditioning, a gated LocalViTBlock, a colour-calibrating generation head, ControlNet-style zero-initialized injection, and uncertainty-based task weighting. The joint framework's value is not label-free training: it uses paired optical images and the same pixel annotations as a dedicated SAR segmentor, supervising the SAR decoder and teacher-forcing the generator with ground-truth one-hot maps. At test time, the model takes only SAR and returns an optical-like readout and a radar-decided land-cover map, with the translation objective supplying dense appearance supervision that speckle lacks. On WHU-OPT-SAR, the predicted-condition model achieves the best SSIM and PSNR among compared translators; its land-cover map surpasses the same backbone trained for labelling alone and is within about one point of DeepLabV3+.
Read the full abstract

All authors Siyuan Liu, Xuze Zhang, Yongshun Wang, Licong Pan, Hang Liu, Huihui Li

Number Theory

Numerical experiments on the Hardy conjecture for the Gauss circle problem

The classical unsolved Gauss circle problem concerns estimating the error between the number of lattice points inside a circle and the area of the circle as its radius tends to infinity. About a century ago, Hardy proposed a conjecture concerning this problem. In this paper, we attempt to provide numerical evidence in support of the Hardy conjecture through large-scale numerical computations.
Read the full abstract

All authors Satoshi Yamaguchi, Shoichi Fujima, Shigehiko Kuratsubo, Eiichi Nakai, Tsuyoshi Yoneda

Hardware Architecture

FlexPosit: Tunable Fractional Precision for LLM Inference Accelerators

Large language models (LLMs) offer remarkable capabilities but impose prohibitive compute and energy costs. Quantization governs the trade-offs between accuracy and hardware efficiency across granularity and bit-width. Finer granularity (e.g., group-wise) provides high accuracy but incurs scaling and control overhead, while coarser granularity (e.g., channel-wise) has lower overhead but loses accuracy at low precision. Meanwhile, mixed-precision quantization exposes rich accuracy-efficiency trade-offs algorithmically, but existing LLM accelerators remain limited to discrete precision modes, leaving the fractional design space between them unexplored. FlexPosit bridges these gaps through co-design of Posit-based quantization and a precision-tunable bit-serial architecture. Algorithmically, FlexPosit employs distribution-aware quantization with hardware-aligned, sensitivity-guided mixed-precision allocation, leveraging the Posit format's tapered precision to achieve group-wise-like accuracy with channel-wise-like regularity. Architecturally, FlexPosit is a unified bit-serial systolic array with lightweight per-column decoders, unified Processing Elements (PEs), and a global precision controller, enabling tunable fractional precision while preserving fully regular systolic dataflow. Across diverse LLMs, FlexPosit achieves near-FP16 accuracy with sub-5-bit fractional weights. It achieves 1.8x higher throughput and 1.2x lower energy than BitMoD (group-wise quantization), and 1.5x higher throughput and 2.0x lower energy than OliVe (channel-wise quantization), establishing a new Pareto frontier for precision-tunable LLM acceleration.
Read the full abstract

All authors Yimin Gao, Liangtao Dai, Jun Yin, Xinfei Guo, Mircea Stan

General Relativity and Quantum Cosmology

Naked Singularities beyond Spherical Symmetry: Instability of $\kappa$-Self-Similar Solutions via an Iteration Scheme

This paper provides the instability counterpart to our recent construction of nonspherically symmetric approximating $\kappa$-self-similar naked-singularity solutions for the Einstein–scalar field system. These singular solutions contain pervasive nonspherical borderline terms, and to prove instability the delicate renormalization procedure developed in [2] does not extend to the more singular setting considered here. To overcome these difficulties, we introduce a new iteration scheme adapted to singular backgrounds whose leading-order geometry depends on the angular variables. At each step, the nonlinear coefficients are frozen using the preceding double-null geometry, and the resulting equations are solved in a triangular order. In this way, the nonspherical borderline terms are incorporated into the approximate geometry rather than treated as perturbative errors, yielding successively sharper estimates. After sufficiently many iterations, the scheme controls the singular angular structure and produces an approximate spacetime whose difference from the exact solution satisfies the required bounds. In particular, these bounds provide an existence region large enough to carry out the instability argument. We further prove that anisotropic perturbations of the outgoing data, arbitrarily small in a scale-critical norm, lead to the formation of a trapped surface. We also formulate and verify a matter-focusing condition under which a parabolic flow argument guarantees the existence of a corresponding marginally outer trapped surface (MOTS). Together, these results establish the nonlinear instability of $\kappa$-self-similar naked singularities beyond spherical symmetry in the Einstein–scalar field system and introduce a framework for applications across Einstein systems.
Read the full abstract

Probability

Annealed Survival Probability of Random Walk in an Inhomogeneous Poisson Environment of Mobile Traps

We study the annealed survival probability of a random walk in an inhomogeneous Poisson environment of mobile traps on $\mathbb{Z}^d$. In dimensions $d=1,2$, we determine the asymptotics for the annealed survival probability of the walker under suitable assumptions on the average trap intensity and study how inhomogeneity in the initial trap configuration affects these asymptotics. Our results extend the asymptotics proved for the homogeneous setting in \cite{DGRS2012}. We also establish a law of large numbers, central limit theorem and large deviation principle for the inhomogeneous Poisson trap environment in $d\ge 1$, generalising the corresponding results proved for the homogeneous case considered in \cite{CG1984}. We present a class of examples of inhomogeneous trap environments where the decay rate of survival probability can be identified and also instances where it does not decay with time.
Read the full abstract

Machine Learning

Locating and Steering Refusal Beyond Attention

Where inside a language model does refusal live, and does that place change when the architecture does? In a transformer, refusal is governed by a single direction in the residual stream, a finding that safety and interpretability tooling now depend on. State-space models (SSMs) route information through a recurrent update instead of attention, sharing no token-mixing mechanism with a transformer. Does the same safety representation survive this shift, or must it be rediscovered per architecture? It survives. A single rigid rotation, which can only reorient a space and not reshape it, aligns one model's representation space with another's, so the two genuinely share the representation. A harm probe trained on a transformer then flags an SSM's harmful inputs, and removing the aligned direction makes a model answer attacks it would otherwise refuse, while a random direction of the same size does far less. What is architecture-specific is not where the direction is steered but where it must be read. Each layer computes a fresh output that is then added into the residual stream, and harm is cleanly readable at this output, the write site, before the addition. A control that holds the intervention's strength fixed shows that what matters is where the direction is estimated, not where it is applied. Applied through a detector-triggered gate, this direction lowers jailbreak success in all four architecture families we test (SSM, transformer, recurrent, hybrid), and on the SSM it holds against an attacker that tunes its prompt against the defense. The gate only matches a trivial rule that returns a fixed refusal whenever the same detector fires, so what transfers across architectures is the direction itself, not defense strength. Safety tooling built on refusal therefore ports to a new architecture by re-estimating the direction at that architecture's write site, not by rebuilding it.
Read the full abstract

Computation and Language

Knowing What Not to Answer: Selective Non-Compliance in Vision-Language Models

Vision-language models (VLMs) are expected to respond helpfully to appropriate requests while withholding compliance with requests that are incorrect, unsafe, infeasible, or unanswerable. However, existing benchmarks predominantly evaluate non-compliance at the level of the query as a whole, assuming that each request either warrants compliance or requires withholding compliance. In practice, real-world queries can contain a mixture of answerable content and components for which compliance should be withheld. In this paper, we introduce KoNA, a benchmark for evaluating selective non-compliance in VLMs across five categories: False Premise, Visual Inaccessibility, Universal Unknown, Task Feasibility, and Safety. Each task evaluates two capabilities: query-level non-compliance and component-level non-compliance under paired single and compound queries. Our evaluation across diverse VLMs shows that models often fail to refuse, correct, or abstain appropriately, and these failures become more pronounced when queries require selective non-compliance. To address this challenge, we fine-tune VLMs using KoNA examples that require selective non-compliance, together with a fully answerable set that should receive direct answers. Our fine-tuned models achieve substantial improvements in non-compliance accuracy while largely maintaining performance on fully answerable tasks. These results suggest that the fine-tuned models can distinguish between answerable components and those requiring non-compliance and respond in a task-appropriate manner.
Read the full abstract

Statistical Mechanics

Scaling laws and energy dissipation in the dynamics of SLE curve as a 1D turbulence model

In this study, we demonstrate the characteristics of the stochastic Loewner evolution (SLE) as a one-dimensional (1D) turbulence model. First, we show that the diffusion process $x(t)$ obtained by the time coordinate change in the SLE becomes the time dependent diffusion process that satisfies Richardson's law of turbulence.Subsequently, we derive the condition on the diffusivity parameter $\kappa$ such that the diffusion process $x(t)$ satisfies Kolmogorov's 4/5th law, which determines the relation between the mean energy dissipation and velocity in the particles of turbulence. Further mathematical analysis showed that this condition is seen as the condition on Loewner entropy $S_{\mathrm{Loew}}$, which includes the white Gaussian noise term $W_{s}$ in the Loewner driving function of the present model. In addition, the numerical simulations were performed to verify the scaling relations of this 1D turbulence model. These results suggest that the standard SLE becomes a beneficial model of 1D turbulence by introducing an appropriate time coordinate change and imposing the conditions on $\kappa$ and $S_{\mathrm{Loew}}$.
Read the full abstract

Computer Vision

HiSfM: Disambiguating Structure-from-Motion via Scaffold-Anchored Hierarchical Reconstruction

Structure-from-Motion (SfM) is a fundamental tool for sparse 3D reconstruction with broad impact in robotics and vision, supporting mapping, localization, and large-scale scene modeling. However, conventional pipelines often fail under hard visual ambiguity caused by repeated or symmetric structures, and incur heavy computational cost due to redundant cameras and constraints. We present HiSfM, a hierarchical coarse-to-fine SfM framework that improves robustness and efficiency through scaffold construction. HiSfM first forms strong local communities using geometrical induced heuristics, then connects communities with a compact yet strong skeleton by packing edge-disjoint spanning trees (EDST) while verifying skeletal edges with a two-view disambiguator. We reconstruct a stable scaffold on this verified skeleton, serving as an anchor to capture the essence of the scene, and subsequently absorb remaining images via efficient registration and triangulation for further refinements. Experiments on ambiguity-focused benchmarks and general datasets show that HiSfM prevents ambiguity-induced failures while substantially reducing runtime compared to previous methods, and improves completeness over aggressive sparsification methods. Code is available at https://github.com/3dv-casia/HiSfM.
Read the full abstract

All authors Ziding Zhao, Hainan Cui, Peilin Tao, Shuhan Shen

Statistical Mechanics

Mean-field theory for quantum spin chains

Mean-field theory is a widely used approximation for describing phase transitions, particularly effective above the upper critical dimension. Its origins can be traced back to the van der Waals theory of the liquid-gas transition and Weiss's molecular field theory of the paramagnetic-ferromagnetic transition. However, it was Lev D. Landau who provided a unifying and general framework applicable to a broad class of physical systems. The mean-field approach typically involves neglecting thermal fluctuations, which is a reasonable assumption in many classical contexts. However, its application to quantum phase transitions at zero temperature is less common. The aim of this short pedagogical paper is to explore and clarify the use of the mean-field approach in the less familiar domain of quantum phase transitions. We specifically consider the Ising model and the Blume-Capel model.
Read the full abstract

All authors Mylène Martirosyan, Astrid Monin-Baroille, Leïla Moueddene, Mohammed M. Shabat, Bertrand Berche

Computer Vision

Counting Beyond Instances: A Benchmark for Group-Individual Object Counting

Visual counting is commonly formulated at the instance level, aiming to estimate how many objects of a queried category appear in an image. However, real-world counting often involves higher-level semantic units formed by multiple instances, such as a bunch of grapes, a stack of plates, or a pair of shoes. This exposes a key limitation of existing counting formulations, which mainly focus on what to count, while largely overlooking at which semantic unit to count. We introduce Group-Individual Object Counting (GIC), a new setting that requires models to count both individual objects and semantic groups within a unified framework. To support this new task, we present BunchCount, a real-world benchmark with 1,330 images, 89,254 individual annotations, and 11,065 group annotations. BunchCount provides paired individual-group annotations within the same image and explicitly records containment relations between each group and its constituent individuals. Experiments on BunchCount show that current advanced counting models perform well on individual instances but fail to count semantic groups more accurately. To mitigate semantic granularity conflict, we propose a counting-unit guided relational counting framework, which exploits group-individual containment relations to regularize cross-granularity representations during training. Our method substantially improves group-level counting while better preserving individual-level counting ability, establishing a strong baseline for counting beyond instances.
Read the full abstract

All authors Rui Wang, Junyi Huang, Jiahui Li, Qiao Yu, Yixue Hao, Long Hu, Baoru Huang

Artificial Intelligence

PLUME: Parameter-Efficient Personalization of Large Language Models via Low-Rank User Modulation in Shared Subspaces

Personalizing large language models (LLMs) is essential for delivering AI assistance that aligns with individual users' styles, intents, and preferences. While per-user fine-tuning can substantially enhance personalization quality, it introduces significant parameter and storage overhead, limiting scalability to large user populations. We propose PLUME (Personalized Low-Rank Adaptation through User Modulation and Shared Subspace), a lightweight framework that achieves efficient and expressive per-user adaptation by leveraging a shared task-specific subspace. Specifically, PLUME first learns a global task subspace from aggregated user data. Personalization is then achieved by training only a lightweight small square matrix within this subspace, enabling each user to obtain a tailored model while keeping shared components fixed. Cross-layer shared parameters and rank-1 residual terms are further introduced to significantly reduce redundancy while maintaining expressiveness. Experiments on multiple personalized text generation benchmarks demonstrate that PLUME achieves comparable or superior performance to strong baselines, while reducing per-user parameters by over 95%. These results establish shared-subspace modulation with minimal residuals as a scalable and semantically grounded approach to LLM personalization.
Read the full abstract

All authors Xinyu Li, Hao Zhou, Jianfeng Zhu, Julina Maharjan, Ruixin Guo, Feodor Dragan, Ruoming Jin

Computation and Language

Refuse without Refusal: A Structural Analysis of Safety-Tuning Responses for Reducing False Refusals in Language Models

Striking a balance between helpfulness and safety remains a fundamental challenge in aligning large language models. To achieve this balance, models should refuse harmful queries (e.g., "How do I shoot someone?") while remaining responsive to benign inputs, even those superficially resembling harmful queries (e.g., "Where can I shoot a good photo?"). However, models often struggle to distinguish genuinely harmful queries from benign queries that contain superficially risky language, resulting in false refusals. In this paper, we address the issue by decomposing a response in the safety-tuning dataset into two distinct components: (i) a boilerplate refusal statement and (ii) a rationale explaining the refusal. Our experiments and analyses show that refusal statements impede accurate discrimination between harmful and benign queries by inducing reliance on superficial cues. In contrast, training solely on rationales reduces false refusals while maintaining a comparable level of safety performance. Rationale-Only benefits also appear in our ICL configuration and remain compatible with the evaluated inference-time mitigation methods. The results emphasize the necessity of precisely curated, fine-grained safety supervision datasets and outline directions for constructing aligned agents that better reconcile helpfulness with safety.
Read the full abstract

Combinatorics

The size-Ramsey numbers of $k$-uniform relaxed $\ell$-trees

The size-Ramsey number $\widehat{R}_k(\mathcal G)$ of a $k$-uniform hypergraph $\mathcal G$ is the minimum number of edges in a $k$-uniform hypergraph $\mathcal H$ such that every $2$-edge-coloring of $\mathcal H$ contains a monochromatic copy of $\mathcal G$. Fox asked whether, for fixed $2\le \ell<k$, the size-Ramsey number of every $k$-uniform relaxed $\ell$-tree on $n$ vertices is bounded by a polynomial in $n$, as recorded by Dudek, La Fleur, Mubayi and Rödl. We give a negative answer when $\ell+1\le k\le 2\ell-2$. More precisely, for every sufficiently large $n$, we construct a $k$-uniform relaxed $\ell$-tree $\bar{\mathcal T}_{n,\ell}^{(k)}$ on exactly $n$ vertices such that \[ \widehat{R}_k\bigl(\bar{\mathcal T}_{n,\ell}^{(k)}\bigr) \ge 2^{c_{k,\ell}n^{\,1-1/(2\ell-k)}}, \] where $c_{k,\ell}>0$ depends only on $k$ and $\ell$. Moreover, any two distinct edges of the construction intersect in at most $2\ell-k-1$ vertices. In contrast, for $k\ge 2\ell$, we prove that every $k$-uniform relaxed $\ell$-tree $\bar{\mathcal T}_{n,\ell}^{(k)}$ on $n$ vertices satisfies \[ \widehat{R}_k\bigl(\bar{\mathcal T}_{n,\ell}^{(k)}\bigr) =O_{k,\ell}(n^{2k}). \]
Read the full abstract

Optimization and Control

Convex Modeling of Price Cross-Impact over Time

Transaction costs can make or break a trading strategy, particularly in relative-value trading of commodity and macro markets, where edges are a few basis points. Price impact is a central component of transaction cost. Price impact models usually include self-impact (a trade in a contract moves that contract's price) but omit two well-documented effects: cross-impact (a trade in one contract also moves the prices of related contracts) and transient impact (price impact decays over time, so an unwind recovers part of the entry cost). A model without these effects overprices the impact of relative-value trades, whose correlated legs are built and unwound over days, and so forgoes potentially profitable trades. This paper models both effects with a convex quadratic cost. In each period, a positive semidefinite matrix built from volatility, volume, and correlation forecasts couples trades across contracts. A power-law decay kernel then couples trades across periods. The resulting cost admits no price manipulation even when liquidity varies over the planning horizon. The model is demonstrated empirically on calendar spread trading of crude oil futures around the commodity index roll.
Read the full abstract

Software Engineering

Building a research-software catalog with a coding agent: from hackathon prototype to public deployment

Generative AI and coding agents can accelerate research software development, but they also increase the need for efficient software discovery and maintenance. We developed a repository catalog during a three-day hackathon and subsequently examined the engineering required to make it suitable for public deployment, including adversarial review, data-quality checks, browser-level validation, and publication safeguards. We then explored whether the lessons learned from this prototype could be transferred to a much larger, human-curated portal, through a retrieval agent under development for MateriApps that combines curated portal metadata, external documentation, vector search, and local language-model generation. Implementation with coding agents was rapid, but achieving reliable operation required substantial additional engineering: the most consequential problems were not crashes but silent failures that produced plausible yet incomplete or incorrect outputs, arising from incomplete data acquisition, misleading assessments, and retrieval or preprocessing failures. These observations suggest that AI-assisted software portals require explicit validation, monitoring, and repeated review, and that curated metadata and maintained documentation remain essential. The MateriApps work is exploratory and remains under active development, so the observations reported for it are preliminary; a comparable combination of curated metadata, automatically collected documentation, and retrieval-based assistance may nevertheless be useful for extending other research-software portals.
Read the full abstract

Machine Learning

Simulation-free Unbalanced Dynamic Optimal Transport with General Growth Penalty

Inferring cellular dynamics from unpaired single-cell snapshots requires modeling both state transitions and population growth or death. Unbalanced dynamic optimal transport (UDOT) addresses this by penalizing growth along transport paths, making the choice of growth penalty a key way to encode biological priors on proliferation and apoptosis. However, existing UDOT solvers either rely on computationally expensive NeuralODE simulations or depend on analytical solutions of conditional paths, restricting their efficiency solely to quadratic penalties, i.e. Wasserstein-Fisher-Rao (WFR) geodesics. To enable an efficient UDOT solver for general growth penalties, we first show that concave growth penalties lead to degenerate solutions where growth and transport are separated. We then introduce Simulation-free Unbalanced Dynamic Optimal transport (SUDO), a simulation-free framework for UDOT with general non-quadratic convex growth penalties. SUDO learns the conditional paths and transport costs, solves the induced semi-coupling problem, and subsequently leverages unbalanced flow matching to achieve a simulation-free solution. On WFR benchmarks, SUDO matches the accuracy of efficient, analytical solution-driven algorithms while outperforming simulation-based methods in computational speed. Beyond WFR, SUDO supports asymmetric penalties that encode proliferation-dominant priors and produce more plausible trajectories and growth estimates on synthetic and single-cell datasets.
Read the full abstract

All authors Junda Ying, Yuxuan Wang, Bowen Yang, Peijie Zhou, Lei Zhang

Computer Vision

AngelFingerprint: A Traceable, Explainable, and White-Box Stealthy Watermark for Text-Guided Image Editing

Text-guided diffusion editing raises disinformation concerns, making reliable image provenance essential. While watermarks are commonly used for this purpose, most methods carry a fixed ID that cannot explain what was changed and which prompt produced it. Furthermore, under open-source white-box access, attackers can easily locate and remove watermarks added as separate modules. Targeting this setting, we propose AngelFingerprint, a novel watermarking framework ensuring edit traceability, explainability, and white-box stealthiness. It integrates a LoRA into the diffusion model to embed the editing prompt's CLIP text embedding directly into the model's weights. An extractor then recovers this embedding from the image pixels alone. This semantic payload explains the edit, while the weight-integrated design makes it hard to detect and isolate even under full white-box access. Two techniques make this possible: a velocity-alignment anchor that preserves edit quality, and a specially designed frequency filter that keeps the watermark imperceptible yet recoverable and robust. On the MagicBrush dataset, our extractor achieves $86\%$ top-1 accuracy in a 200-way prompt retrieval, versus $20\%$ for prompt inversion.
Read the full abstract

All authors Bo-Han Kung, Futa Waseda, Ching-Chun Chang, Isao Echizen, Shang-Tse Chen

Computation and Language

How Do Language Models Represent and Use Phonological Information for Allomorph Selection?

Language models are trained on tokenized text that obscures the sound structure of words, yet they reliably produce morphemes whose form is phonologically conditioned. It remains unclear whether they rely on item-specific memorization or rule-like generalization and, if the latter, how that generalization is implemented. We therefore ask whether this phonological condition is represented within language models and how it is causally used for allomorph selection. For the English indefinite article a/an, we show that the phonological condition is encoded along a single linear direction in trigger-token embeddings, that this direction causally drives article selection in token-level wug tests, and that, at the article-prediction position, the model forecasts the upcoming trigger token and uses the forecasted trigger's phonological feature to choose the article. We then ask whether this rule-like generalization extends beyond English article selection, both to allomorph selection in other languages and to explicit phonological judgment. Together, these results provide a mechanistic account of phonologically conditioned allomorph selection in language models, and dissociate this generation-time ability from explicit metalinguistic judgments.
Read the full abstract

Signal Processing

Wireless Foundation Models: State-of-the-Art and Open Challenges

Wireless foundation models (WFMs) have emerged as a promising approach for learning reusable representations from large-scale wireless data and adapting them to downstream tasks. However, the rapidly growing literature remains fragmented across modalities, pretraining objectives, architectures, adaptation strategies, and evaluation protocols, making it difficult to assess progress toward broadly transferable models. This survey provides a systematic analysis of WFMs for physical-layer applications. We first introduce the main WFM design components, including pretraining, backbone architectures, and downstream adaptation. We then organize the literature into five physical-layer task families: signal recognition and demodulation, channel representation learning, RF sensing and localization, beam management, and spectrum sensing and monitoring, while separately examining multi-task PHY models. Across these categories, we analyze how existing models are pretrained, adapted, and evaluated, with particular attention to downstream task diversity and the distinction between in-distribution, partial-shift, and out-of-distribution transfer. Our analysis shows that current WFMs provide increasing evidence of reusable wireless representations, but this evidence varies considerably across task families and evaluation settings. Differences in datasets, modalities, architectures, pretraining objectives, adaptation protocols, and distribution shifts make it difficult to determine which design choices drive transfer and generalization. We conclude by identifying open directions for improving data availability, evaluation rigor, generalization, efficient adaptation, and real-world deployment, providing a unified framework for understanding the current WFM landscape and the requirements for developing more reusable foundation models for future physical-layer wireless systems.
Read the full abstract

All authors Alonso M. Pacheco Huachaca, Juan J. Rodriguez Rodriguez, Ahmed Aboulfotouh, Nelson L. S. da Fonseca, Carlos A. Astudillo, Hatem Abou-Zeid

Artificial Intelligence

FinalityBench: An Effect-Level Benchmark for Agent Decisions Under Delayed and Conflicting Financial Finality

A merchant's payment processor, ledger, ERP and bank feed are updated by messages that get delayed, duplicated, dropped and reordered, so for minutes at a time the four hold contradictory beliefs about the same order. An agent resolving the exception must decide whether to ship goods, re-submit a capture, refund or wait, knowing some of those cannot be undone. We present FinalityBench, an executable benchmark for that decision. It keeps a hidden canonical event log and derives each system's view from a separately faulted delivery stream, so disagreement follows from specified fault semantics rather than being authored. Grading is on executed monetary effects: an episode is scored by the merchant's terminal economic position, relative to a privileged reference told when the pending capture resolves. The corpus of 321 tasks includes 45 twin pairs (90 tasks): tasks whose four system views are identical at the decision instant, whose authoritative probes both return unknown, and whose eventual correct dispositions differ. That snapshot indistinguishability is checked under every evaluation seed rather than assumed; equivalence over all interaction traces is not claimed. Over 14,445 graded episodes from nine programmatic policies, ranking by single-task accuracy and by paired loss disagree in 7 places: a ship-on-first-sign policy is second-best by accuracy at 65.7% and worst in the suite by paired loss, because it cannot tell the two members apart. A runtime gating irreversible actions on an authoritative finality probe reaches 85.4% and, unlike every polling policy, loses nothing to pass^5; its residual loss is almost entirely one archetype, which prices finality information directly. Language models reach the same exact rate as the hand-written gate on a stratified subset, lose about twice as much money, and discover the finality-gating strategy without being told it.
Read the full abstract

Hardware Architecture

Sustainable Edge Vision via Empirically Calibrated DVFS: Eliminating Thermal Throttling on Passively Cooled Hardware

Passive cooling eliminates the energy overhead and mechanical failure modes of fans, making it attractive for edge deployment, yet sustained Deep Neural Network (DNN) inference on passively cooled edge Systems-on-Chip (SoCs) is bottlenecked by thermal throttling. To address this, we propose an empirically calibrated, state-aware Dynamic Voltage and Frequency Scaling (DVFS) scheduler. Unlike heuristic-driven controllers, our methodology utilizes time-domain guards and absolute temperature bounds, with derivative triggers acting as safeguards against sharp thermal spikes. Evaluated on a passively cooled Raspberry Pi 5 running YOLOv8n, our scheduler eliminates all observed thermal throttling events during sustained 30-minute workloads. It outperforms a temperature-only reactive baseline by achieving a 6.8% higher frame rate (Cohen's d = 8.73) while consuming 1.9% less energy per frame. Furthermore, our optimized passive scheduling surpasses an actively cooled reference system in energy efficiency (Joules/frame), though active cooling remains superior for raw throughput. Through isolated ablations, we show that the dwell guard is necessary for run-to-run reproducibility. Finally, exploratory boundary probes indicate that the passive operating envelope closes at ambient temperatures ($\ge 27^\circ$C) where nonlinear leakage defeats DVFS-based control. These results indicate that, within the mapped envelope, correct scheduling can make mechanical cooling unnecessary for sustained edge inference on this platform.
Read the full abstract

Quantum Physics

Strong-Drive Limits in Josephson Circuits: From Chaos to an Unbound-Resonance Threshold

Strong microwave drives enable fast measurement and parametric control in superconducting circuits but can induce transitions out of the intended low-energy manifold. We develop a unified description of strong-drive limits in flux- and charge-driven Josephson circuits across drive frequency and dc flux bias. Using classical phase-space analysis and Floquet–Markov simulations, we identify distinct low- and high-frequency mechanisms. At low frequency, we characterize bound-state resonances and separatrix chaos and find that the flux-drive chaos threshold depends strongly on dc flux bias. At high frequency, these mechanisms are suppressed, and the dissipative steady state transfers from the central bound-state sector to outer resonances formed by above-barrier running trajectories. The resulting unbound-resonance threshold is nearly independent of drive frequency and circuit parameters over the regime studied and is controlled primarily by dc flux bias. Coherent simulations show that parametric operation persists beyond this threshold, but at a reduced rate, setting an effective upper bound on the achievable operation speed. We derive analytical criteria for both thresholds, validate them numerically, and experimentally confirm the predicted dc-bias dependence of the low-frequency threshold in a flux-driven SQUID. We also determine the timescales for transfer into the unbound-resonance regime and relaxation back to the bound-state manifold after the drive is removed. Finally, we relate the stability limits to a complementary picture based on the junction critical current and extend the framework to multitone drives and inductively shunted circuits. Together, these results identify the mechanisms limiting strong driving and suggest routes to extend the stable operating range of Josephson circuits.
Read the full abstract