Peer Revenue Returns subject to peer review

Papers

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

All of arXiv

Tuesday 1 September

Analysis of PDEs

Asymptotic Expansions of the Fundamental Solution for a Mixed Local-Nonlocal Operator

In this paper, we study the fundamental solution of the mixed local-nonlocal operator $ -\Delta+(-\Delta)^s,$ $ 0<s<1. $ We derive precise asymptotic expansions of the fundamental solution and its gradient, up to the first nontrivial correction term, both near the origin and at infinity. The expansions describe the transition between the local and nonlocal diffusion scales and include the critical regimes in which logarithmic terms occur. As applications of these asymptotics, we establish a distributional Bôcher-type theorem for nonnegative supersolutions with an isolated singularity. We further obtain quantitative positive and antisymmetric maximum principles on punctured balls.
Read the full abstract

Algebraic Geometry

Parabolic Lie algebroid connections on parabolic principal bundles over curves

Let $X$ be a compact connected Riemann surface and $S\,\subset\, X$ a finite subset. We consider parabolic principal $G$–bundles $\mathcal{E}_{G}$ on $X$ with parabolic structure on $S$, where $G$ is a connected complex reductive affine algebraic group. Let $P\, \subset\, G$ be a parabolic subgroup and $\mathcal{E}_{P}\, \subset\, \mathcal{E}_{G}$ a reduction of structure group of $\mathcal{E}_{G}$ to $P$. We give a criterion for the existence of a parabolic Lie algebroid connection on $\mathcal{E}_{P}$ for any given parabolic Lie algebroid on $(X,\,S)$ whose anchor map is not surjective. More precisely, $\mathcal{E}_{P}$ admits a parabolic Lie algebroid connection if the reduction $\mathcal{E}_{P}\, \subset\, \mathcal{E}_{G}$ is parabolically infinitesimally rigid. In particular, the Harder–Narasimhan reduction of $\mathcal{E}_{G}$ admits a parabolic Lie algebroid connection.
Read the full abstract

Quantum Algebra

On Braided Differential Calculi and Quantum G-structures

We develop a theory of first order differential calculi in braided monoidal categories and classify braided covariant and bicovariant calculi on braided Hopf algebras. We show that, under certain conditions, bicovariant calculi can be transmuted to braided bicovariant calculi. For Radford–Majid biproducts, we combine bicovariant calculi on a Hopf algebra and braided bicovariant calculi on the corresponding braided Hopf algebra to covariant smash product calculi. The associated Maurer–Cartan form is shown to decompose into a direct sum of the Maurer–Cartan forms of the structure Hopf algebra of the quantum principal bundle and the braided Hopf algebra on the base. Geometrically, this construction realises the quantum affine extension of a given Hopf algebra, and we prove that the resulting quantum principal bundle is equipped with a frame resolution induced by the quantum Maurer–Cartan form. Building on this correspondence, we introduce and develop the notion of quantum $\textrm{G}$-structure, proving that quantum $\textrm{G}$-structures are quantum frame resolutions on the reduction. The theory is illustrated by examples based on transmutations of higher analogues of Sweedler's Hopf algebra and on the braided quantum plane, seen as a Yetter–Drinfeld module of $O_q(\mathrm{GL}_2)$.
Read the full abstract

All authors Antonio Del Donno, Giovanni Gava, Emanuele Latini, Thomas Weber

Astrophysics of Galaxies

Semiresolved Stellar Populations as Distance Indicators

Galaxy distances are central to our understanding of the Universe. Despite the success of existing approaches, independent and complementary methods remain valuable for testing systematic effects and extending the applicability range of different distance metrics. Here we demonstrate that the spectrum of an individual semi-resolved stellar population encodes direct information about its distance and that valuable distance constraints can, in principle, be obtained by combining multiple independent measurements. When applied to optical spectra of bulge of the Andromeda galaxy (M\,31), we derive a stellar population-based distance of 756$\pm$57 kpc ($\mu_0 = 24.39 \pm 0.15$), in agreement with state-of-the-art measurements. Crucially, semi-resolved stellar population distances does not require secondary calibrations, although its absolute scale remains conditional on the adopted stellar population models. Our findings provide a first assessment of the feasibility of deriving stellar population-based distance estimates from simple stellar population models, motivating further tests of the broader applicability and precision of the method.
Read the full abstract

All authors Ignacio Martín-Navarro, Patricia Iglesias Navarro, Francesco La Barbera, Alexandre Vazdekis, Michael A. Beasley, Marc Huertas-Company, ChangHoon Hahn, Luis Peralta de Arriba, Isaac Alonso Asensio, Eirini Angeloudi, Miguel Cerviño, Katja Fahrion, Tereza Jerabkova, Jesús Falcón-Barroso, Sebastián F. Sánchez, Prashin Jethwa

Dynamical Systems

The Quadratic Easy Coefficients Conjecture via Finite-Type Shifts and Zeta Functions

We prove the Quadratic Easy Coefficients Conjecture stated as Conjecture 1 in T. W. Cusick, Recursions for quadratic rotation symmetric functions weights, Discrete Applied Mathematics 378 (2026), 93–101. For an arbitrary finite sum of quadratic monomial rotation symmetric Boolean functions, we identify the recurrent part of the rules matrix with a signed binary de Bruijn transfer matrix $B$. We then give a one-step presentation of the finite-type shift associated with the Boolean function in the symbolic-dynamics construction of Chirvasitu and Cusick. Fourier transformation in an auxiliary $\F_2$ coordinate decomposes the adjacency matrix of this shift into an unsigned de Bruijn block and the signed block $B$. Consequently the dynamical zeta function is \[ \zeta_{X_f}(z)=\frac{1}{\det(I-z\cR(f))}, \] where $\cR(f)$ is the rules matrix. This equality identifies, with their algebraic multiplicities, the characteristic values, supplied by symbolic dynamics, with the roots of the characteristic polynomial of the rules matrix. The desired easy coefficients formula follows from the trace of $B^n$. We also prove nonsingularity and justify the unique backward extension of the weight recurrence.
Read the full abstract

Optimization and Control

A Scalable Framework for Simultaneously Optimizing Damper Positions and Viscosities

We consider damping optimization for vibrating systems described by a second-order differential equation. The goal is to determine the damping positions and viscosity values of the dampers in the system so that they produce the lowest average total energy. We propose a new framework for determining the number and positions of the dampers, using re-weighted $l_1$ minimization and pruning techniques. The efficiency and performance of our new approach are verified and illustrated in several numerical examples.
Read the full abstract

Machine Learning (Statistics)

Measuring consistency via ensemble margin and local prediction variability: Auditing decision systems in the presence of predictive multiplicity

The Rashomon effect is a machine learning phenomenon where equally accurate models produce different predictions for the same inputs (predictive multiplicity). Existing work primarily focuses on multiplicity within individual models, but in more complex decision systems, the impact of the Rashomon effect is less well understood. In this work, we study multiplicity from the perspective of auditing incorrect ensemble predictions, where the decision to divert an instance for human review is based on a consistency criterion that combines the ensemble margin with a measure of local prediction variability for each constituent model. With mild assumptions about stability and smoothness, we show that the consistency scores of finite ensembles converge to the corresponding consistency score of the expected model from the Rashomon set as the ensemble size and the number of samples used to measure local prediction variability increase. To demonstrate the efficacy of the proposed criterion, we evaluate the framework with respect to transformer models applied to natural language understanding tasks and parameter-efficient fine-tuning of large language models used for tabular data classification tasks. Our experiments show that ensembling models from the Rashomon set substantially reduces the risk of incorrect predictions going unchecked compared with auditing a single model, while incurring only a moderate increase in the number of diversions. Moreover, the auditing behavior of the full Rashomon set can be closely approximated by finite ensembles of relatively modest size, with the risk approaching zero for some datasets. We further demonstrate that the proposed measure exhibits stronger agreement with established predictive multiplicity metrics than existing consistency measures, providing a more reliable way to capture multiplicity in the Rashomon set.
Read the full abstract

Systems and Control

Equivalent DQ Sequence-Domain Model of Unbalanced Three-Phase Passive Elements for Power Converter Controllers

Equivalent dq0 models of three-phase passive impedances are widely used for the design of three-phase inverter and rectifier controllers in the synchronous frame. However, these equivalent models assume that the impedance in each phase are the same, which is not always applicable. The fact that the impedance in each phase is different, generates second order harmonic content in the dq0 components, which hinders the advantages of the dq0 transformation. To avoid the second harmonic in these scenarios, it is possible to use the Fortescue's theorem to represent any asymmetrical set of three-phase signals as a linear combination of three symmetrical sequences. In the literature, only a few works have attempted to derive the equivalent model of unbalanced three-phase passive elements in the sequence domain, and none is able to account for the interaction between the zero sequence and the positive and negative sequences. This work presents the derivation of the equivalent dq sequence-domain model of unbalanced three-phase passive elements both analytically and with an intuitive graphical approach. Experimental results are shown to validate the proposed models.
Read the full abstract

All authors Air\án Franc\és, Luis Saz, Rafael Castillo, Dionisio Ramirez, Javier Uceda

Applied Physics

Quasi-normal-mode signatures of periodicity and hyperuniformity

Hyperuniform materials occupy an intermediate regime between periodic order and conventional disorder, characterized by a suppression of long-wavelength density fluctuations, commonly quantified by the structure factor. This characterization, however, relies on the statistical properties of an extended configuration or supercell representation and does not directly describe the resonant response of a finite structure, as typically encountered in scattering experiments. Here, we introduce the quasi-normal-mode spectrum as a complementary framework for characterizing the wave properties of finite stealthy hyperuniform materials. We analyze the complex eigenfrequencies of both periodic and stealthy hyperuniform structures over a range of the stealthiness parameter ($\chi$), and investigate how the spatial organization of scatterers is encoded in both frequency and spatial profiles of their quasi-normal modes. We find systematic changes in the quasi-normal modes spectrum as the degree of hyperuniformity is varied, revealing signatures that are not captured solely by the structure factor. In particular, the evolution of the resonant states provides a direct connection between structural correlations and the scattering response of finite systems. Our results establish quasi-normal modes as a real-space and open-system characterization of hyperuniformity and provide a unified framework for comparing discrete heterogeneous materials from the perspective of their scattering and resonant dynamics.
Read the full abstract

All authors V. Romero-García, M. Martí-Sabaté, V. F. Dal Poggetto, L.M. García-Raffi, M. Lázaro

Robotics

Autonomous robotic bridging using distributed swarm control without inter-agent communication

We describe SCARAB–Swarm-Capable Autonomous Robotic Aquatic Bridging. Using distributed swarm control and multi-model sensing of agents and docking targets, agents can localize themselves, join into formations and proceed to desired target locations. Our technologies would eventually allow the Army to perform unpredictable, dispersed river crossings, enhance crew survivability, and minimize the logistics footprint compared to the current Improved Ribbon Bridge. Our methods operate without GPS or RF communications, though these can be used in non-contested environments (e.g., civilian disaster relief for flooding, etc.). We demonstrate our system via physics-engine-based simulation of several agents using unmanned surface vehicles (USVs) in the presence of currents and wind.
Read the full abstract

All authors Vishwaak C. Thamaraiselvan, Cody L. Lundberg, Michail Theofandis, Suhas Chelian, Nicholas R. Gans

Soft Condensed Matter

Cell size and confinement drive asymmetric cell division through a cortical instability

Asymmetric cell division – in which a mother cell divides into two daughter cells of unequal size – is a fundamental problem in biology. It is believed that the asymmetry originates from the prior polarization of the mother cell. Here we show that division asymmetry can occur spontaneously even in unpolarized mother cells. Specifically, curvature-dependent active stresses in the cell cortex can lead to this symmetry breaking without any molecular polarity cue if the mother cell is confined within a restricted space. Either reducing the cell size or tightening mechanical confinement triggers the same spontaneous symmetry-breaking instability, in which the contractile ring slips off the equator to yield daughters of unequal volume. In the presence of a polarity cue, this instability cooperates with the cue to program the division asymmetry. The model prediction is compared with the imaging data of C. elegans embryogenesis, in which successive cell divisions in a confined eggshell lead to smaller and smaller cell sizes. The measured division asymmetry indeed increases as the cells shrink, and is further amplified when the embryo is mechanically compressed, both in agreement with the model prediction.
Read the full abstract

All authors Da Gao, Guoye Guan, Chao Tang, Rui Ma

Computer Vision

Scale-based Approach for Active Wildfire Segmentation on Satellite Imagery

Active wildfire mapping from satellite imagery is challenging due to the sparse and highly imbalanced nature of fire pixels, especially in early-stage or low-density fire observations. This work investigates the use of multispectral Landsat-8 imagery for active-fire segmentation under multi-scale wildfire size conditions. We propose a data-driven protocol to characterize fire-region size distributions through connected-component analysis and an interquartile range criterion, enabling the evaluation of model robustness across different local fire-region densities. Three segmentation architectures, U-Net, DeepLabV3+, and SegFormer, are evaluated under different SWIR-based spectral configurations. Results show that U-Net achieves the strongest robustness across the evaluated conditions, SegFormer provides competitive performance, and DeepLabV3+ tends to produce conservative predictions with reduced recall. Across architectures, SWIR2 consistently achieves the strongest or near-best results, highlighting its importance for active-fire segmentation in Landsat-8 imagery. These findings suggest that both spectral band selection and architectural design are critical for robust satellite-based active wildfire mapping trained on low active fire-pixel density images.
Read the full abstract

General Relativity and Quantum Cosmology

Scalar wave scattering by black holes embedded in dark matter halos

We study the scattering of massless scalar waves by static black holes embedded in dark matter halos. The halo is modelled by means of a Hernquist profile, for which the Einstein equations admit an exact solution. We analyse the scattering using null geodesics, the glory approximation, the partial wave expansion, and the complex angular momentum (CAM) method. The geometric analysis shows that the presence of the halo changes the light ring radius weakly but shifts the critical and glory impact parameters. The wave analysis shows that the characteristic black hole interference pattern is preserved, with halo compactness providing the leading correction and the halo scale radius playing a secondary role. For sufficiently compact halos, the environment also induces a sizeable gravitational redshift. Most of the displacement of the interference extrema can be reproduced by a vacuum Schwarzschild solution with an appropriately chosen $\omega M_{\rm BH}$. A non-vanishing residual, however, reveals a genuine halo contribution that cannot be absorbed into such an effective frequency shift. Finally, we perform the CAM reconstruction of the scattering cross section and find that only a few Regge poles are required to achieve high accuracy at large angles.
Read the full abstract

Computer Vision

Multimodal RGB-Infrared Combination for UAV-Based Wildfire Segmentation: A Comparative Study on FLAME3

Unmanned Aerial Vehicles (UAVs) have emerged as a promising platform for firefighting operations due to their flexibility, low operational cost, and ability to acquire high-resolution imagery in locations that may be difficult or dangerous to access using conventional methods. Recent advances in deep learning have significantly improved the capabilities of UAV-based wildfire monitoring systems. The present work investigates RGB-infrared fusion for binary wildfire segmentation on the FLAME3 dataset. In this Study, RGB and Infrared baselines are compared with three representative fusion strategies across three segmentation architectures, including U-Net, DeepLabV3+, and SegFormer. The key motivation of this work is to analyze the contribution of each modality, evaluate the impact of fusion timing, and examine how different network architectures exploit multimodal information for UAV wildfire delineation. The findings indicate that thermal information plays a dominant role in UAV segmentation and that feature-level multimodal fusion combined with transformer-based architectures offers the most promising direction for future research.
Read the full abstract

All authors Matheus F. Kovaleski, Luís Garrote, Cristiano Premebida, Jérôme Mendes, João Ruivo Paulo

Solar and Stellar Astrophysics

The quiescent states of V745 Sco and V3890 Sgr: VLT/X-shooter and Swift/XRT+UVOT observations

Symbiotic recurrent novae are symbiotic binaries in which multiple thermonuclear outbursts have been observed. Only four such systems are currently known in the Galaxy: T CrB, RS Oph, V745 Sco, and V3890 Sgr. They provide convenient laboratories for studying accretion and binary interaction in symbiotic novae during quiescence. We characterise the quiescent states of the less-studied systems V745 Sco and V3890 Sgr and compare their properties with those of T CrB and RS Oph. We analyse VLT/X-shooter spectra together with Swift/XRT and UVOT observations, supplemented by ground-based optical photometry and archival data. We determine the spectral types of the red-giant donors and compare the spectral energy distributions of the systems with those of single bulge giants of the same spectral type to investigate the properties of the accretion. We classify the donor stars in V745 Sco and V3890 Sgr as M7III and M6.5III, respectively. We derive reddenings and distances consistent with both systems belonging to the Galactic bulge. Both objects exhibit a significant ultraviolet excess relative to normal late-type giants, indicating ongoing accretion during quiescence. From this excess we estimate mass transfer rates of the order of $\sim 5 \times 10^{-8}\ M_{\odot}\ \mathrm{yr}^{-1}$, broadly consistent with the observed recurrence timescales. These results suggest that recurrent nova eruptions in V745 Sco and V3890 Sgr can be sustained by long-term accretion at approximately the present rate, without requiring an episode of enhanced accretion similar to that observed in T CrB.
Read the full abstract

All authors Bor Jamnik, Ulisse Munari, Nicola Masetti, Gregor Traven, Manica Perko

Statistics Theory

Weighted persistence intensity regression

Persistence diagrams summarize the multiscale topological structure of data, and in applications they often arrive paired with covariates. We develop nonparametric methodology and theory for estimating the expected weighted persistence diagram conditional on a Euclidean covariate. Representing each weighted diagram as a finite random measure on a compact window, we take the density of its conditional expectation as the regression target, the conditional weighted persistence intensity. For a conditional double-kernel estimator we establish finite-sample sup-norm rates with a matching minimax lower bound, uniform rates in partial optimal transport, and an unbiased-risk cross-validation criterion for bandwidth selection. Simulations with analytically known intensities corroborate the theory and show that cross-validation selects the oracle candidate bandwidth in the exact-intensity design. The method is illustrated by studying how radial geometry in cerebral artery trees varies with age.
Read the full abstract

Computation and Language

When Tokenization is Secretly Output Supervision

Tokenization in language models is treated by default as an input preprocessing decision. We argue that this framing is incomplete: in autoregressive models, tokenizer granularity determines what the model must resolve in a single forward pass, and therefore the supervision signal it receives. This affects both the difficulty of the learning problem and the representations that emerge inside the model. We test this in a controlled experiment on numeric reasoning with a novel decoupling of input and output tokenization. As the output supervision view predicts, differences in task performance, training dynamics, and model internals are induced by output tokenization and largely invariant to input tokenization. This may matter in practice, because models with different tokenization strategies differ not only in input representation but in the task they were trained on. Comparisons between models may thus partly reflect task definition rather than ability. A survey of 120 recent *CL papers on numeric reasoning confirms that this is rarely acknowledged: only about 10% report the numeric tokenization of the models they evaluate, while 69% compare across tokenization, and thus supervision, regimes without reporting it. While prior work documents that tokenization consistently affects model performance, there is no principled account of why. We argue that framing tokenization as output supervision provides that account.
Read the full abstract

High Energy Physics (Experiment)

Constraint on the Higgs boson width in the diphoton final state using signal-background interference in proton-proton collisions at $\sqrt{s}$ = 13 TeV

The standard model Higgs boson, with a mass of 125 GeV, has a predicted decay width, $\Gamma_\mathrm{H}$, of 4.1 MeV. This Letter presents the first constraint on $\Gamma_\mathrm{H}$ obtained through the interference between the resonant H $\to$ $\gamma\gamma$ process and the continuum $\gamma\gamma$ production, both produced via gluon-gluon fusion. The analysis is performed using proton-proton collision data at $\sqrt{s}$ = 13 TeV, collected by the CMS experiment between 2016 and 2018, corresponding to an integrated luminosity of 138 fb$^{-1}$. The observed (expected) limit on the Higgs boson width is $\Gamma_\mathrm{H}$ $\lt$ 92 (138) MeV at 95% confidence level. This is the most stringent constraint on $\Gamma_\mathrm{H}$ from on-shell Higgs boson measurements to date.
Read the full abstract

Robotics

Obstacle-Aware Autonomous Coverage and Navigation for Outdoor Robots

Long-duration outdoor coverage with autonomous platforms remains challenging beyond classical planning: deployments face localization drift in open spaces, obstacles in cluttered sites, controller feasibility in turn-heavy maneuvers, and persistent autonomy with energy management. We propose a unified ROS 2 architecture for single-robot outdoor coverage: a dual-antenna RTK-GNSS fused in an EKF keeps both position and heading accurate across long missions; three controller-aware refinements extend a mature coverage planner; a Nav2-based behavior-tree mission executive coordinates multi-goal execution, layered recovery, cost-aware goal management, and autonomous docking for return-to-charge. In real-world trials across five outdoor areas with varying geometries and obstacle densities, the robot completed every coverage route, sweeping 93.1% to 96.1% of the planned coverage area.
Read the full abstract

Computation and Language

InSight: A Benchmark for Agentic Claim Verification in Interactive Visualizations

Vision Language Models have demonstrated remarkable proficiency in interpreting static visual artifacts, but modern data analysis is inherently dynamic, requiring the active interrogation of interactive environments. Existing benchmarks are predominantly constrained to static imagery and one-shot question answering and fail to capture the epistemic demands of this domain, where evidence is frequently occluded, distributed across linked views, or conditionally revealed through user agency. In this paper, we introduce InSight, a benchmark for agentic claim verification over interactive visualizations. The dataset consists of 21,349 claims derived from human-authored analytical narratives and grounded in fully interactive web-based environments. Agents must navigate these environments to determine whether a natural language claim is supported, refuted or not verifiable given the available evidence. Unlike traditional evaluations, InSight treats interaction traces as intrinsic proxies for reasoning, enabling a rigorous audit of how models seek and synthesize visual evidence. We evaluate state-of-the-art models, revealing that interactive verification remains a non-trivial challenge. We release InSight at https://github.com/maevehutch/insight.
Read the full abstract

All authors Maeve Hutchinson, Syed Mahbubul Huq, Mohammad Albinhassan, Radu Jianu, Aidan Slingsby, Pranava Madhyastha

Symplectic Geometry

Floor-crossing for Legendrian Clasp Move

We investigate the effect of the clasp move of Legendrian submanifolds of dimension at least two on the moduli spaces of pseudo-holomorphic curves that contribute to the differential of the Chekanov-Eliashberg algebra. As an application, we compute the differential of Chekanov-Eliashberg algebra of twist spheres $\Lambda_k$ of dimension at least two. Moreover, we construct an infinite family of Legendrians $\mathcal{T}(\Lambda,k)$ from any horizontally displaceable Legendrian $\Lambda$ in $P\times\mathbb{R}$, such that the Chekanov-Eliashberg algebra of $\mathcal{T}(\Lambda,k)$ admits an augmentation, while $\mathcal{T}(\Lambda,k)$ has no exact embedded Lagrangian filling of vanishing Maslov class in the symplectization of $P\times \mathbb{R}$ for sufficiently large $k$.
Read the full abstract

Signal Processing

Satellite Swarms for Direct-to-Cell Networks: A Distribution-Performance Trade-off Analysis

This work investigates 2D distributed satellite swarm configurations for direct-to-cell (D2C) applications. Unlike previous studies, which primarily considered swarms as deployment alternatives to monolithic arrays, this paper focuses on exploiting the increased spatial resolution enabled by large distributed apertures. Simulation results show that, for the considered scenarios, increasing the level of antenna distribution across satellite platforms enlarges the effective aperture and improves the system sum rate under ideal operating conditions. While gains are moderate for uniformly distributed users, they become particularly pronounced in scenarios including hotspot with high user densities. The results further show that user scheduling strategies do not invalidate the superiority of highly distributed configurations, as these architectures provide a more balanced rate distribution across the coverage area, including hotspot regions. In addition, the paper analyzes several key implementation challenges associated with large distributed swarms, including errors in inter-satellite relative positioning, synchronization impairments, and limited beamforming and user-position update rates. Although within the range of swarm configurations and scenarios considered in this work, performance gains continue to increase with aperture size, these practical constraints may restrict swarm sizes in the medium term. Nevertheless, the observed performance gains strongly motivate further research into scalable synchronization, positioning and data distribution techniques for future large-scale satellite swarms.
Read the full abstract

All authors Xavier Artiga, Marius Caus, Ana I. Pérez-Neira, Yerassyl Akhmetkaziyev, Malte Schellmann

Computation and Language

Polish ModernBERT: The Long and Short of Polish Language Understanding

Encoder-only Transformers remain effective for discriminative and representation-learning tasks, yet Polish encoders still largely rely on BERT/RoBERTa-style architectures. We introduce Polish ModernBERT, a family of four Polish encoders available at Base and Large scales, each with 512-token and 8K context variants. We adapt the ModernBERT pretraining recipe through staged selection experiments and release a long-context benchmark covering legal topic classification, ideological decision-direction prediction, factual-consistency assessment over literary plot summaries, and human-rights violation assessment. Across 30 tasks, Polish ModernBERT achieves the best overall performance among the evaluated Polish encoders, reaching 83.99 and 85.11 for the Base-8K and Large-8K models, respectively. On long-context tasks, the 8K variants improve over matched Polish RoBERTa-8K baselines from 67.47 to 77.15 and from 75.88 to 78.49 at the Base and Large scales, respectively. The Base-8K model achieves this gain with 22% fewer parameters (149M vs.\ 190M). Efficiency measurements in representative inference setups show lower peak memory usage and latency than matched Polish RoBERTa baselines in both 512-token and 8K settings. Polish ModernBERT-8K-Base additionally achieves the best result on a Polish retrieval benchmark among the evaluated encoders below 300M parameters.
Read the full abstract

All authors Michał Perełkiewicz, Sławomir Dadas, Rafał Poświata, Małgorzata Grębowiec

Materials Science

Simulating Continuous-Rotation 3D Electron Diffraction: A Multislice and Bloch Wave Framework

To improve the agreement between measured and calculated intensities of three-dimensional electron diffraction (3D ED) experiments, a simulation pipeline is needed to assess and quantify the influence of various structural and experimental parameters. We present a computational pipeline, built upon the abTEM Python package, to simulate continuous-rotation 3D electron diffraction data based on either the Bloch wave or the multislice formalism. Multislice calculations in arbitrary orientations are achieved through large supercells and windowing. We investigate the convergence of key simulation parameters and test their consistency, establishing suitable parameters for accurate and efficient simulations. The pipeline's applicability and robustness are demonstrated through case studies on cubic silicon as well as stretched and sheared variants, analyzing the influence of electron kinetic energy, sample thickness, orientation, and symmetry on simulated diffraction intensities. Finally, we investigate a representative selection of seven compounds, including cubic $\mathrm{SrTiO_3}$, monoclinic $\alpha$-glycine $\mathrm{C_2H_5NO_2}$, and triclinic kyanite $\mathrm{Al_2SiO_5}$ to validate the method. This framework for the simulation of 3D electron diffraction data establishes an approach to investigate the dependence of diffracted intensities on experimentally relevant parameters which are difficult to systematically investigate in experiments.
Read the full abstract

All authors Małgorzata K. Cabaj, Jacob Madsen, Toma Susi, Lukáš Palatinus, Paul B. Klar

Earth and Planetary Astrophysics

Transiting Planetary Systems with Distant Giant Companions Remain Moderately Coplanar

The mutual inclination between inner planets and distant giant companions provides an important probe of planetary system formation and dynamical evolution, yet direct measurements of this quantity remain scarce. We combine radial velocity (RV) observations with Hipparcos–Gaia astrometry to constrain the orbital architecture of 19 planetary systems hosting at least one transiting inner planet and one outer giant companion. Using a hierarchical Bayesian framework, we infer the population-level distribution of the minimum mutual inclination, $\Delta I$, between the inner and outer planetary orbits. We find that the $\Delta I$ distribution is well described by a Rayleigh model with a scale parameter of $\sigma = 15.8^{+2.8}_{-2.6}\deg$, which is strongly preferred over an isotropic distribution ($\Delta\log Z=5.45$). This result suggests that transiting systems hosting distant giant companions remain substantially more coplanar than expected for an isotropic population, consistent with the partial preservation of primordial coplanarity. A division by the mass ($0.3\,M_{\rm Jup}$) of the inner transiting planet suggests that giant-inner-planet systems may have lower $\Delta I$ than small-inner-planet systems, with $P(\sigma_{\rm giant}<\sigma_{\rm small})=0.952$; however, the current data do not significantly favor a model allowing different $\sigma$ values for the two subsamples over one in which they share a common $\sigma$. Future Gaia DR4 astrometry will enable more robust population-level studies of the three-dimensional architectures of systems with distant giant companions.
Read the full abstract

All authors Guang-Yao Xiao, Huan-Yu Teng, Xiumin Huang, Fabo Feng, Dong Lai, Fei Dai, Yu-Juan Liu

High Energy Physics (Phenomenology)

Phase structure of magnetized quark matter at imaginary chemical potential

We investigate the phase structure of magnetized quark matter within a nonlocal SU(2) Polyakov-Nambu-Jona-Lasinio model. Our work focuses on the interplay between temperature and imaginary chemical potential in the presence of an external uniform magnetic field, with emphasis on the deconfinement and Roberge-Weiss transitions. We analyze the dependence of the relevant critical temperatures on the external field, and we explore the sensitivity of the phase structure to different Polyakov loop effective potentials. Our results show qualitative agreement with lattice QCD findings, reproducing inverse magnetic catalysis in the chiral sector and a decreasing behavior of the Roberge-Weiss transition temperature with increasing magnetic field.
Read the full abstract

Computer Vision

IntroConformal: Conformal Factuality Guarantees for Large Vision-Language Models via Introspective Signals

Large Vision-Language Models (LVLMs) have achieved strong multimodal performance, yet ensuring the factual correctness of generated content remains challenging. Existing methods that provide statistical guarantees on factuality typically rely on external verifiers or generation-time confidence signals, which introduce auxiliary dependencies or often fail for confident but incorrect outputs. We argue that reliable factuality control can instead be achieved through introspective signals derived from the model itself. We introduce IntroConformal, a training-free Conformal Risk Control (CRC) framework that provides finite-sample, distribution-free factuality guarantees. We first instantiate it with layer-wise semantic stability, a conformity score derived from hidden-state representations, and then propose verification probability, a stronger score capturing the model's self-administered judgment on claim factuality. Across multiple LVLM architectures, IntroConformal satisfies the conformal risk guarantee while substantially reducing abstention and achieving competitive or superior claim-level discrimination relative to external verifier-based baselines.
Read the full abstract

Computation and Language

Behaviorally Effective LoRA Writes Are Sparse and Structured

Low-rank adaptation fixes the rank of the update, but it does not identify which parts of a trained write actually carry behavior. We study that question directly and show that behaviorally effective LoRA writes are sparse, structured, and far more concentrated than the raw low-rank parameterization suggests. We use Learned-Basis LoRA, a learned-basis continuation recipe, to expose that structure. The recipe warms up an unconstrained adapter, converts its learned write columns into a module-wise orthonormal basis, freezes that basis, and continues training inside the constrained parameterization. Across 14 exact switches from unconstrained to constrained form, held-out accuracy is unchanged at the conversion step and reconstructed write matrices differ by at most 0.25% relative Frobenius error. Same-state continuation then shows that the same trained checkpoint develops differently under different write subspaces, establishing write geometry as a causal state variable. A no-retraining projection test shows that useful write signal stays inside the learned write space and largely disappears from random or frozen-activation PCA controls. The concentration pattern is strong at both local and global scales. Across GSM8K, MathQA, and AQuA, per-module top-k continuation reaches its optimum at k in {2, 4} in all twelve seed-level cases we test. A stricter global ranking test shows that learned top-16 and top-32 subsets outperform matched random subsets, especially on GSM8K/Qwen and MathQA/Qwen. Single-direction ablations further reveal a sparse set of late q_proj, o_proj, and down_proj components with outsized behavioral impact.
Read the full abstract

All authors Haruto Sato, Yuki Tanaka, Ren Nakamura, Aoi Kobayashi, Mei Ito