Peer Revenue Fully funded

Papers

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

All of arXiv

Monday 5 October

High Energy Astrophysical Phenomena

Adaptive proposals for nonparametric equation-of-state inference with RIFT: iterative Gaussian-process refinement in a compressed sequence representation

R. O'Shaughnessy and Z. Zhu

Nonparametric equation-of-state (EOS) inference evaluates the likelihood on a finite ensemble drawn from a prescribed functional prior. Reweighting that ensemble suffices for the binary neutron stars detected so far, and loses effective sample size only once the likelihood concentrates — which needs third-generation catalogues, or extra electromagnetic and nuclear-theory information folded into the same target. We develop and validate an adaptive-proposal procedure that leaves the base prior fixed: data-informed generations are combined by multiple importance sampling and, where that degenerates at large catalogue size, by a calibrated Gaussian projection. We implement it in two representations — a compressed principal-component parametrization of the neutron-star sequence, whose draws need no stellar-structure solve, and a sound-speed parametrization with a differentiable Tolman–Oppenheimer–Volkoff solver — both on a constrained low-density-to-pQCD bridge. Because that anchor, and not the bridge, fixes the low-density prior, we rebuild the ensemble with a crust-plus-chiral-EFT anchor in place of the wide nuclear metamodel: the maximum-mass prior is unmoved, the radius gains a hard prior ceiling at the measured NICER value. GW170817 with the specified NICER and radio factors gives the calibration-conditional $R(1.4\,M_\odot)=11.7\pm0.4$km; in a synthetic $1000$-event catalogue it reproduces the mean fixed-prior $\Lambda(1.4\,M_\odot)$ to $0.4$%, $1.5$ times wider, as the fixed archive's effective sample size falls from $\sim\!470$ to 22.
Read the full abstract

Artificial Intelligence

2d-fet-bench: from spatial reasoning to fet design on flakes

Dunzhi Zhou, Chengyu Zhu, Gang Qiu et al. (+1 more)

Field-effect transistor (FET) layouts on exfoliated two-dimensional flakes are typically drawn by hand for each flake, placing contacts and gates to match its position and outline in optical micrographs. To our knowledge, no executable benchmark tests whether language-model agents can perform this flake-specific construction reliably. We introduce 2D-FET-Bench V2, a benchmark of 128 layout tasks built from microscopy-derived flake contours, including hole-containing flakes and multi-flake tasks. Each task supplies a textual device specification and contour coordinates. An agent generates typed polygon and path operations rendered to GDSII. A deterministic verifier checks geometric and structural requirements, and a separate integrity check verifies that the supplied contours remain unchanged. Scripted reference layouts pass all 128 tasks, showing that every task is solvable. We evaluate six models and seven workflow and scaffold variants of GPT5.6-Luna, with five attempts per task. The best-performing configuration in the six-model panel, GPT5.6-Luna with ReAct-3, passes 62.3% of attempts and solves 80.5% of tasks at least once (coverage) and 43.8% in all five attempts (consistency). ReAct-3 exceeds the one-pass Plan-and-Execute by 27.0 pass@1 points at 2.46 times the tokens. An expert audit of one sampled verifier-passing layout per covered task, across five ReAct-3 configurations, accepts 56.4% to 63.5% of them. The benchmark evaluates geometric and structural FET layout construction.
Read the full abstract

All authors Dunzhi Zhou, Chengyu Zhu, Gang Qiu, Caiwen Ding

Quantum Physics

White paper: 1-10 Hz matter-wave interferometer to test the spin entanglement witness for quantum gravity

Sougato Bose, Anupam Mazumdar, Marko Toroš et al. (+118 more)

In this white paper, we highlight the importance of the ($1-10~{\rm Hz}$) frequency range for laboratory tests of the quantum nature of gravity using the quantum gravity-induced entanglement of masses (QGEM) protocol. QGEM requires matter-wave interferometers with masses ($m\sim10^{-15}-10^{-14}~{\rm kg}$), brought within separations ($d\sim30-50~\mu{\rm m}$), while maintaining spatial superpositions of ($1-20~\mu{\rm m}$) and coherence for ($\tau\sim 0.1 - 1~{\rm s}$). These requirements make low-frequency environmental noise a central experimental challenge and place QGEM in a regime closely related to the low-frequency goals of the Einstein Telescope (ET) and the Cosmic Explorer (CE). In particular, QGEM is sensitive to relative acceleration noise (RAN) and to gravity-gradient noise (GGN) generated by seismic and other environmental mass-density fluctuations. For representative parameters $m=10^{-14}~{\rm kg}$, $\Delta x=10~\mu{\rm m}$, and $\tau=1~{\rm s}$, the differential acceleration-noise amplitude spectral density must be suppressed well below the $10^{-15}~{\rm m\,s^{-2}/\sqrt{Hz}}$ level to keep acceleration-induced dephasing below the relevant experimental scale. Achieving this level of low-frequency noise suppression is therefore a key requirement for QGEM and closely parallels the seismic and gravity-gradient noise challenges that ET and CE address.
Read the full abstract

All authors Sougato Bose, Anupam Mazumdar, Marko Toroš, Tian Zhou, Tadeusz Adach, Niayesh Afshordi, Agya Sewara Alam, Alexandre Arbey, Navdeep Arya, Simon Baier, Peter F. Barker, Angelo Bassi, Ettore Bernardi, Lorenzo Braccini, Robert Brandenberger, Daniel Braun, Guri K. Buza, Luigi Cacciapuoti, Carlo Cepollaro, Lin-Qing Chen, Yanbei Chen, Ralph Jason Costales, Marion Cromb, Álvaro de la Cruz-Dombriz, Catalina Curceanu, Shubhang Dadhich, Debarshi Das, Saurya Das, Pratika Dayal, Subhadeep De, Ema Dimastrogiovanni, Lajos Diósi, Or Dobkowski, Kemal Döner, Brian D'Urso, Gurudev Dutt, Shafaq Gulzar Elahi, Samira Elghaayda, Matteo Fadel, Samuel Fedida, Omer Feldman, Fabiano Feleppa, Ron Folman, Joshua Foo, Paolo Fragolino, Laurent Freidel, Giulio Gasbarri, Marco Genovese, Andrew Geraci, Menachem Givon, Cisco Gooding, Jonathan M. H. Gosling, Piotr T. Grochowski, David Groswasser, Mustafa Gündoğan, Ekim Taylan Han{\i}meli, Bas Hensen, Dipankar Home, Richard Howl, Yonathan Japha, Maciej T. Jarema, Rainer Kaltenbaek, Adrian Kent, Eva Kilian-Rademacher, M. S. Kim, Jarosław K. Korbicz, Timothy Kovachy, Samuel Kováčik, Ohkyung Kwon, Gaetano Lambiase, Naor Levi, Iarley P. Lobo, Leon Loveridge, Adrian Lupascu, Paolo Luppi, Marta Maria Marchese, Antonino Marcianò, Aaron G. Markowitz, Chiara Marletto, Ryan J Marshman, J. D. D. Martin, Florian Millo, Gavin W. Morley, Maria Muretova, Sebastian Murk, Robin Oberfrank, Daniel K. L. Oi, Jerzy Paczos, Papadopoulos Stylianos, Matteo G. A. Paris, Mauro Paternostro, Alessandro Pesci, Luciano Petruzziello, Fabrizio Piacentini, Tanmay Kumar Poddar, Sofia Qvarfort, Markus Rademacher, Dennis Rätzel, Anna Chiara Rescigno, Ryan Rizaldy, Albert Roura, Carlos Sabín, Barry C. Sanders, Martine Schut, Helen M Sheehy, Suprit Singh, Aninda Sinha, Urbasi Sinha, Peter Skakunenko, Michael E Tobar, Géza Tóth, Hendrik Ulbricht, Gislaine Varão, Vlatko Vedral, Vincent Vennin, Francesca Vidotto, Giuseppe Filiberto Vitale, Marko Vojinović, Chenan Wei, Qian Xiang, Magdalena Zych

Solar and Stellar Astrophysics

Axial-Dipole and Equatorial-Dipole Coronal Holes: Congruent Magnetic Geometry, Divergent Evolution

Olga A. Panasenco, Yi-Ming Wang, Samuel T. Badman et al. (+2 more)

While the Sun's polar fields reverse, the axial dipole is weak and low-latitude coronal holes are common. Equatorial-dipole coronal holes (EDCHs) appear when the equatorial dipole is strong, forming at its magnetic poles, similar to the axial-dipole coronal holes (ADCHs) of minimum. Not every low-latitude hole qualifies: the 2024 maximum was quadrupole-led. Parker Solar Probe found fast wind from an EDCH deeply sub-Alfvénic at $9.9\,R_\odot$, at only $60\%$ of an asymptotic speed matching polar wind, which is usually thought to accelerate faster. We ask whether coronal structure predicts this, using potential-field models of synthetic and observed fields at three cycle phases. We show that pure axial and equatorial dipoles are exactly congruent, with the same open areas and core expansion factors, so any difference must arise from orientation relative to the differential-rotation pattern or from the surrounding multipoles. We find that ADCHs are stationary, sustained by converging meridional flow, whereas EDCHs depend on an equatorial dipole that active-region emergence builds and flux transport dismantles: over $18$-day sequences the axial-dipole strength at minimum is steady to $1\%$, while the equatorial dipole varies by $21\%$ at maximum and $8\%$ in the declining phase. Most open flux in EDCHs expands less than in ADCHs, but their boundaries are pseudostreamer-dominated, with non-monotonic expansion. Despite these differences, magnetic geometry alone does not predict slower acceleration in EDCHs; we argue that any real difference points to extended heating by reflection-driven Alfvén-wave turbulence, perhaps fed differently at those boundaries, which potential-field models cannot capture.
Read the full abstract

All authors Olga A. Panasenco, Yi-Ming Wang, Samuel T. Badman, Adam J. Finley, Yeimy J. Rivera

Machine Learning

Benchmarking Label-Revealed Online Updates for EEG BCI Decoding

Bogdan Kozyrskiy, Artem Grachev and Abraham I. Camelo Guerrero

Electroencephalography (EEG) signals drift over time, which can cause static brain-computer interface (BCI) models to degrade in practice. We present a benchmark for online adaptation and compare two widely used pipeline families, Common Spatial Patterns (CSP) and Riemannian covariance-based methods, under time-ordered prequential (test-then-train) evaluation. We examine (i) which pipelines benefit most from label-revealed updates, (ii) whether controlled forgetting of older data improves robustness, and (iii) how a minimal-calibration cold start compares with starting from a pretrained model. Across four datasets (three motor-imagery datasets and one movement-decoding dataset), label-revealed online updates improve 13 of 14 model/dataset pairs on the two largest streams, with relative accuracy gains of up to about 18% over a frozen model. A Shapley-based data-valuation analysis over temporal blocks assigns the largest mean value to the most recent block in each of the three analyzed datasets, while older blocks retain positive value.
Read the full abstract

Machine Learning

Learnable Spectral Activations

Tamir Shor, Or Litany and Alex Bronstein

Implicit neural representations (INRs) are shaped by the spectral structure induced by their input encodings and activation functions. Existing methods improve fitting primarily by modifying which frequencies are available to the network, through coordinate encodings or periodic nonlinearities. However, frequency access is not the only bottleneck: signals with localized or spatially varying structure require the network to efficiently compose frequencies into multi-harmonic internal responses. We introduce learnable spectral activations (LSA), which replace fixed neuron-level nonlinearities with a residual truncated Fourier series whose harmonic amplitudes are learned during training. LSA does not expand the asymptotic function class. Instead, it changes the factorization of the representation: linear weights select features while activation coefficients control spectral shaping, and the two are updated by separate gradients. Because the activation output is affine in the coefficients given fixed pre-activations, spectral tuning becomes a more direct subproblem compared to architectures where it is entangled with feature selection. Empirically, this factorization concentrates more target-signal energy in the leading eigenmodes of the neural tangent kernel, consistent with improved optimization behavior. Across audio, image, neural radiance field, and neural acoustic field tasks, LSA also improves reconstruction quality.
Read the full abstract

Statistical Methodology

Large-scale linear hypothesis testing for high-dimensional online M-estimation

Daeyoung Ham, Myeonghun Yu and Tate Jacobson

With the growing prevalence of streaming data, renewable estimation has become an important tool for updating statistical analyses without retaining historical raw observations. We study estimation and linear hypothesis testing in high-dimensional online M-estimation. Existing renewable procedures based on quadratic approximations can incur first-order approximation errors because the gradient of the unpenalized loss generally does not vanish at a preceding penalized estimator, posing a particular challenge for valid inference. To address this issue, we propose a corrected renewable loss that preserves historical first-order information and combine it with partial folded concave regularization and an iterative local linear approximation. We establish finite sample $\ell_1$- and $\ell_2$-error bounds, together with contraction bounds and strong oracle properties for both the constrained and unconstrained iterative estimators. Building on these results, we derive oracle Bahadur representations and establish asymptotically valid Wald and score tests for general linear hypotheses, allowing both the testing-oracle dimension and the number of restrictions to diverge. We further characterize their asymptotic power under local alternatives. The proposed procedure requires only recursively updated summaries while recovering the first-order inferential behavior of its pooled-data oracle counterpart. Simulation studies and real-data applications demonstrate the finite-sample performance of the proposed estimation and testing procedures.
Read the full abstract

Machine Learning (Statistics)

Bayesian Optimization on Function Spaces via Sparse RKHS Manifolds

Davide Sartor, Meghan E. Huber, Donghyun Kim et al. (+1 more)

Bayesian Optimization (BO) has become an established methodology for minimizing black-box functions of a vector input. Often, however, this parameter vector arises from the discretization of an inherently functional relationship. Several recent articles have considered the Functional Bayesian Optimization (FBO) setting, in which the variable to be optimized is not a member of a finite dimensional vector space, but rather an infinite dimensional function space. In this work, we propose $L^0$ Manifold Optimization (L0MO), a simple approach to FBO which searches the subset of a Reproducing Kernel Hilbert Space (RKHS) consisting of functions with a sparse representation in the kernel functions, optimizing both the kernel locations and their coefficients. We discuss in detail the relationship between our method and existing ones, providing a unifying lens through which to view prior works. To assess our method against the state of the art, we conduct an extensive computational study, and along the way develop a novel set of benchmark test functions which port standard finite-dimensional ones to the infinite dimensional domain. Our experiments demonstrate that, on balance, the proposed method achieves superior performance across a wide range of test benchmarks.
Read the full abstract

All authors Davide Sartor, Meghan E. Huber, Donghyun Kim, Nathan Wycoff

Human-Computer Interaction

Knit-Structure Effects on Electromechanical Metrics and Their Correlation with Joint-Angle Estimation Error in Knitted Strain Sensors

Annika Eloranta, Zhuchenyang Liu, Iiro Naulapaa et al. (+5 more)

Knitted resistive strain sensors show strong promise for joint motion sensing in sports and rehabilitation, but the linkage between sensor design and in situ performance remains unclear. We investigate how knit structure and machine settings (e.g., stitch size) shape electromechanical properties and which metrics predict sensing performance during bending. Sensors spanning seven common knit structures at two stitch sizes were fabricated, characterized under uniaxial cyclic tension, and evaluated on a joint emulating bending rig. Joint-angle estimation was assessed with machine learning models, and correlations with electromechanical metrics were analyzed. Experimental results show that, among six common metrics, gauge factor and baseline resistance are largely set by knit structure, while working range, linear range, hysteresis, and cyclic stability vary only modestly across designs. Gauge factor correlates negatively and baseline resistance positively with joint-angle estimation error, mainly in lower-sensitivity designs, whereas the other metrics have weak or no predictive value. These results support using uniaxial tensile tests to screen out weak designs, while underscoring the need for joint-relevant evaluation and application-specific metrics to identify top performers.
Read the full abstract

All authors Annika Eloranta, Zhuchenyang Liu, Iiro Naulapaa, Iida Arvola, Yao Zhang, Anna-Mari Leppisaari, Lulu Xu, Yu Xiao

Networking

Understanding Conflict and Compatibility Conditions for Agents in AI-RAN

Arshia Zolghadr, Joao F. Santos, Imtiaz Nasim et al. (+3 more)

Disaggregated mobile networks expose open interfaces for independent agents to control the Radio Access Network (RAN). The interactions between agents with contrasting objectives can result in conflicts. However, existing conflict mitigation methods treat agents that control the same parameter as being in a conflict that requires arbitration of their actions, without assessing whether their objectives can be satisfied simultaneously. In this paper, we introduce a control-theoretic approach to determine when agents can coexist and share control of the same parameters. We model threshold-based agents often encountered in the literature as sampled deadband feedback controllers and represent their objectives through Key Performance Indicator (KPI) targets and tolerances that define acceptable deadband bounds. We derive compatibility conditions from these bounds and show that compatibility for any number of agents and monitored KPIs can be established through pairwise tests. We propose a conflict mitigation policy that allows compatible agents to share control of the same network parameters while restricting arbitration to cases where their objectives cannot be satisfied simultaneously. We evaluate our policy through simulations using NIST's ns3-oran extension of ns-3, and our results show that it enables agents to share control with near-zero Service Level Agreement (SLA) drift when their objectives are compatible and the shared acceptable range is attainable, while matching Priority Mitigation when their objectives are incompatible.
Read the full abstract

All authors Arshia Zolghadr, Joao F. Santos, Imtiaz Nasim, Deniz Aytemiz, Nicholas J. Kaminski, Jacek Kibilda

Fluid Dynamics

(Why) Jets quantise trapped acoustic modes

Brandon Yeung, Michael N. Stavropoulos, Vincent Jaunet et al. (+2 more)

High-subsonic turbulent jets trap acoustic modes in their potential core. The modes lie in narrow frequency bands that are $O(0.1)$-wide in Strouhal number, $St$. By studying a Mach 0.9, Reynolds number $10^6$ jet, this work confirms an earlier hypothesis: each band is made up of $O(10)$ discrete frequencies, $\Delta St \sim O(0.01)$ apart. To resolve them, we collect $O(100\,000)$ snapshots from dedicated schlieren experiments and large-eddy simulations (LES), and apply spectral proper orthogonal decomposition (SPOD). Our SPOD algorithm uses sine tapers and enforces the jet's rotational and reflectional symmetries. For both the schlieren and the LES, the trapped SPOD modes are bounded in the $St$-$x$ and $St$-$k$ planes, where $x$ and $k$ are respectively the axial coordinate and wavenumber. The bounds quantise the modes in $St$, $x$, and $k$, such that the $n_x$th mode in $St$ has $n_x$ antinodes along $x$ and $k$. We then predict trapped modes with local, global, modal, and non-modal stability theories. We find evidence of a competition between modal and non-modal stability. From only the theoretical bounds, we build a simple in-phase standing-wave resonance model that predicts the quantised global-mode frequencies.
Read the full abstract

All authors Brandon Yeung, Michael N. Stavropoulos, Vincent Jaunet, Peter Jordan, Oliver T. Schmidt

Combinatorics

A refinement of the edge theorem for order and chain polytopes

Jon Lee

For a finite poset, we partition the edges of the order polytope and of the chain polytope into classes indexed by the connected convex subsets of the poset, and we establish that corresponding classes have the same cardinality. This yields a closed formula for the common number of edges, and it identifies the bijection of Hibi, Li, Sahara and Shikama as a disjoint union of simple bijections between corresponding classes, explaining why it is a bijection and how it acts on edge directions and lengths. As consequences, the two polytopes have equally many edge directions, with matching multiplicities, and, for every m, equally many pairs of vertices whose symmetric difference has m components; the chain polytope has at least as many edges parallel to each coordinate subspace as the order polytope; and the edge lengths of the chain polytope are dominated by those of the order polytope, with strict inequality in the last two comparisons exactly when the poset contains a three-element chain. We discuss implications for linear and convex combinatorial optimization over ideals and antichains. We also express the number of edges in terms of the comparability graph, extending to stable-set polytopes of all graphs; give a recursion for series-parallel posets and closed formulas for layered, zigzag and crown posets; and characterize the distributive lattices for which the two polytopes are unimodularly equivalent. Finally, either diameter can exceed the other arbitrarily; both are bounded by the width, and they coincide for series-parallel, zigzag and crown posets.
Read the full abstract

Neural and Evolutionary Computing

Simplified Swarm Optimization for Surrogate-Assisted Reliability Design of Insulated-Gate Bipolar Transistor Power Modules Using an Open-Source Process Finite-Element Model

Wei-Chang Yeh

Process-induced warpage, ceramic stress and solder strain limit the reliability of insulated-gate bipolar transistor (IGBT) modules on direct-bonded copper (DBC) substrates. Surrogate-assisted design studies train regression models on finite-element analysis (FEA) databases, but rarely check the optimized designs against new FEA or report how surrogate error interacts with the optimizer. This paper builds and evaluates an open pipeline: an open-source process finite-element model, surrogates tuned by Simplified Swarm Optimization (SSO), multi-objective design search, FEA confirmation of selected designs and confirmation-driven infill. The model starts at the second reflow and reproduces measured warpage within 18.2%, 33.6% and 15.3% at the reflow, housing and molding stages without fitted parameters. On a balanced 60-design database, all stochastic tuners reach the same test accuracy, outperform the published grid on cross-validated performance for every output but generalize better only for warpage; the cross-validated ranking of tuners does not transfer to the test set. With equal result reporting, multi-objective SSO and the non-dominated sorting genetic algorithm II give comparable Pareto fronts; a corrected multi-objective particle swarm optimizer trails both. FEA confirmation shows that warpage predictions hold (mean absolute error 0.5%), whereas at the design-space bounds reached by the optimizers the ceramic-stress surrogate is optimistic by up to 26%. Two confirmation-driven infill rounds reduce this error to 1-10% and halve the out-of-sample error; no confirmed design improves on the database in ceramic stress. Ceramic-stress results are indicative, as the metric is mesh-sensitive at production resolution. The model, database, scripts and pre-registered and post-registration results are released.
Read the full abstract

Analysis of PDEs

Minimax solutions for FitzHugh-Nagumo systems with critical exponential growth

João Marcos do Ó, José Carlos de Albuquerque and Hugo Henryque Silva

We study a class of FitzHugh-Nagumo type systems with critical exponential growth in a smooth bounded domain of $\mathbb{R}^{2}$. The classical reduction method turns the strongly indefinite system into a scalar nonlocal problem whose quadratic part induces a norm equivalent to, but different from the standard Sobolev norm, so that the Trudinger-Moser threshold must be handled in this framework. We prove the existence of a ground state solution via a mountain pass argument combined with a Nehari level comparison. We also prove the existence, without imposing the monotonicity condition required in the ground state argument, of a nontrivial solution obtained through a saddle-point linking argument based on the spectral decomposition of $\operatorname{T}=-\Delta+B$.
Read the full abstract

Astrophysics of Galaxies

BRAHMa: Bar Recognition And Hatching using Machine learning – a framework for oriented bounding box detection of galactic bars

Rajit Shrivastava, Narendra Nath Patra and Keerthi K

Galactic bars are fundamental components in studies of dark matter, galaxy evolution, and secular dynamics. However, current methods for measuring bar properties suffer from significant limitations. Manual and crowdsourced catalogs are not scalable to next-generation surveys, while recent machine learning (ML) approaches based on pixel-wise segmentation are computationally intensive and require complex post-processing to extract physical parameters. In this paper, we introduce BRAHMa (Bar Recognition And Hatching using Machine Learning), a publicly available tool based on an oriented-bounding-box YOLO11x model, designed for the rapid and objective detection of galactic bars. Our model was initially trained on synthetic data (Shrivastava 2025), validating its efficacy before training with real data. BRAHMa was trained on bar masks from the Galaxy Zoo 3D (GZ-3D) citizen science project, utilizing images from the DESI Legacy Survey. We address the inherent challenge of 'ground truth' in astronomical data by comparing GZ-3D labels with the Hoyle dataset for 586 common galaxies. This comparison yields a baseline mean absolute error of 0.99 kpc between the GZ3D-derived and Hoyle measurements for 586 common galaxies, providing an empirical reference for catalogue-to-catalogue disagreement. Our BRAHMa tool achieves an L1 error of 1.1 kpc on a large set of 3,150 galaxies from the Hoyle dataset, demonstrating performance that approaches this fundamental limit. We further validate the model's robustness by applying it to distinct datasets, proving its generalizability. BRAHMa provides a scalable and reliable method for bar extraction, enabling large-scale morphological studies. The tool is publicly available through the BRAHMa web interface. (https://brahma-bar-detector.netlify.app/)
Read the full abstract

Robotics

RACER: Residual-Adaptive Closed-Loop Estimation for Sampling-Based Planning in Wheeled-Quadruped Racing

Yuxiang Liu, Marla Eisman, Lizhi Yang et al. (+2 more)

We present RACER, a hierarchical control framework for wheel-based quadruped racing that combines an MPPI planner with a learned residual dynamics model and a low-level RL velocity tracker. The planner augments a nominal unicycle kinematic model with a neural residual term to capture the closed-loop tracking behavior of the RL policy. To train this residual model under limited real-world data, we propose Low-Rank Residual Adaptation (LoRRA), a two-stage approach that pre-trains on large-scale simulation data for broad coverage and then fine-tunes on a small real-world dataset with a low-rank constraint. In simulation, we empirically validate our engineering choices by showing (A) Residual dynamics improve the overall performance of our pipeline by capturing the tracking error of RL velocity tracker at high-speed cornering. (B) Residual dynamics trained with both source-domain and target-domain data gives racing performance significantly better than the residual dynamics trained with only target-domain data. (C) Low-rank constraint at target-domain adaptation gives higher success rates and higher performance than full-tune and from-scratch when domain gap in ground coefficient or joint gain increases.
Read the full abstract

All authors Yuxiang Liu, Marla Eisman, Lizhi Yang, Aaron Ames, Francesco Borrelli

Optics

Compact Circuit Models for Nanoantenna-Based Lightwave Electronics

Adina Bechhofer, Felix Ritzkowsky, Karl K. Berggren et al. (+1 more)

As the field of nanoscale lightwave electronics matures from centering around fundamental science to focusing on device engineering, there is a growing need for compact models that can provide rapid and scalable quantitative analysis of the performance of petahertz-frequency optical-field-driven electronics. In this work, we developed a circuit model framework for describing nano-scale lightwave electronics. Our framework captures the physics governing nanoscale optical devices, such as electrically-connected nanoantennas, using compact circuit models that predict their electromagnetic response, their electron emission, and their charge transfer dynamics. This approach allows scalable simulation of large integrated systems containing networks of nanoscale lightwave electronic components, which is not possible using full-wave electromagnetic and particle-in-cell simulation methods. We implemented the compact model in an free commercial circuit solver (LTspice) and validated its electromagnetic response against a full-wave electromagnetic solver (MEEP). We propose an experiment to fully benchmark the model's ability to capture non-linear charge-transfer-based coupling between devices. The circuit model implementation speeds up the electromagnetic analysis from hours to seconds and accounts for charge-driven coupling, enabling rapid quantitative studies of device operation which provide new insights into how femtosecond signals propagate through nanoscale lightwave electronic structures. We anticipate that the methods we introduce here will become essential to the development and performance analysis of nanoscale lightwave electronics for communication, computation, and signal processing at optical frequencies.
Read the full abstract

All authors Adina Bechhofer, Felix Ritzkowsky, Karl K. Berggren, Phillip D. Keathley

Social and Information Networks

Simulating Strategies for Defense Against Brand-Targeted Online Disinformation

Bryan C. Boots

Brand-targeted disinformation can damage firms not only by spreading false beliefs, but by eroding trust, amplifying reputational uncertainty, and persisting across networked digital environments. This paper models brand-targeted disinformation as a trust-weighted contagion process using a multi-agent simulation across six network structures, three amplification levels, and six intervention strategies. Results show that scale-free and influencer-heavy networks are most vulnerable to rapid spread, while clustered and polarized networks provide partial containment until amplification bridges communities. Hybrid intervention combining prebunking, trusted-node activation, and rapid correction performs best overall. The paper contributes a reusable simulation framework for studying active brand disinformation defense strategies.
Read the full abstract

Machine Learning

Evaluation of Active Feature Acquisition Policies with Tabular Foundation Models

Yuta Kobayashi, Divyam Madaan and Shalmali Joshi

Active feature acquisition learns policies that sequentially acquire features to maximize information about a target variable. We study how to learn and evaluate such policies from finite offline data using prior-data fitted networks (PFNs), which are off-the-shelf models that output posterior predictive distributions without task-specific training. We show that under the imbalanced coverage of offline data, using total predictive entropy as a reward creates an epistemic bias that penalizes acquiring sparsely observed features. Specifically, this reward conflates epistemic uncertainty (arising from lack of offline data) with aleatoric uncertainty (arising from uninformative features). To address this, we target the posterior expected (aleatoric) entropy instead of the total predictive entropy output by a PFN for evaluating feature acquisitions. Empirical evaluations on synthetic and real-world datasets demonstrate that our approach consistently reduces value estimation bias and yields credible intervals with strong empirical coverage, which can translate to improved downstream policy selection.
Read the full abstract

Machine Learning

What pass@k Cannot Measure: Evaluating Diversity and Capability Retention after Post-Training

Subham Rath, Raj Dandekar, Rajat Dandekar et al. (+1 more)

pass@$k$, the fraction of problems a model solves within $k$ sampled attempts, is the field's default protocol for deciding whether reinforcement-learning (RL) post-training on verifiable rewards improved a model. At the population level, pass@$k$ depends only on a problem's probability of a correct sample, with no term for how it is distributed across outputs. We show this gap is not academic. Training Qwen2.5-1.5B-Instruct on grade-school math with Group Relative Policy Optimization (GRPO) and with rejection-sampling fine-tuning (RFT, training on the model's own shortest verifier-passed rollout) moves three complementary diversity measures (token-level entropy, answer-level entropy, unique answers per prompt) in opposite directions, with zero overlap across three seeds per arm. The gap survives restricting to verifier-correct completions only (lexical diversity among correct solutions is 15% lower for GRPO, after controlling for length) and a count-controlled check isolating diversity among incorrect answers alone, ruling out that GRPO's higher accuracy alone explains it. Yet pass@8 and pass@32 show no consistent winner on GSM8K, and a hard MATH-500 subset shows the same pattern: separation only at low $k$. Compared against the starting checkpoint, no trained arm significantly improves hard-problem coverage: RFT is significantly worse, while GRPO is statistically indistinguishable from it - so GRPO's pass@1 edge over RFT reflects a smaller loss relative to Base, not a capability gain, a missing-control issue, not a failure of pass@$k$. On GSM8K, only pass@1, with no role in detecting diversity by construction, separates the arms cleanly, rewarding the arm whose correct solutions are least diverse. We argue this is a concrete instance of a standard evaluation protocol missing a property it is routinely used to certify.
Read the full abstract

All authors Subham Rath, Raj Dandekar, Rajat Dandekar, Sreedath Panat

Statistics Theory

Convex Order Beyond Dimension One: Projection Tests, Counterexamples and Gaussian Mixtures

Olivier Guéant

In actuarial science and quantitative finance, convex order provides a natural way to compare risks with the same mean. In dimension one, convex order is well understood through several characterisations. In higher dimensions, a natural approach is to compare all one-dimensional projections, but, although necessary, the resulting condition is in general not sufficient for multivariate convex order. A simple counterexample due to Pinelis exploits Popoviciu's inequality. We revisit this example using gauge functions, and show that gauge functions themselves do not suffice to characterise multivariate convex order for symmetric distributions, even in dimension two. We then show that the radial case makes the source of the failure transparent and construct radial counterexamples in every dimension at least two. Nevertheless, projection tests are sufficient for some important families of distributions. In particular, this is the case for linear transformations of a common radially distributed random vector. This framework includes elliptical distributions generated from a common radial distribution and, in particular, centred Gaussian distributions. Beyond comparisons between Gaussian distributions, Jourdain and Pagès recently studied the comparison of a Gaussian distribution with a Gaussian mixture, obtaining necessary conditions and sufficient conditions whose equivalence was left open in general. In this paper, we settle the comparison of a centred Gaussian distribution with a mixture of two centred Gaussian distributions: in every finite dimension, the Gaussian is dominated by the mixture in multivariate convex order if and only if all one-dimensional projections are ordered in convex order.
Read the full abstract

Artificial Intelligence

Defense-in-Depth for LLMs: Evaluating Memory Gates Against Activation-Induced and Memory-Induced Sycophancy

Ritvij Sharma, Russell Dlugosz, Ryan Zhou et al. (+1 more)

Long-term memory allows Large Language Models (LLMs) to maintain personalized context across interactions, but retrieved user history can induce memory-induced sycophancy, causing models to favor stored user beliefs over objective evidence. Existing defenses primarily operate on retrieved context and are rarely evaluated jointly with internal behavioral bias. We introduce a $2 \times 2$ defense-in-depth framework separating internal activation steering from external memory handling. We extract sycophancy steering directions from 100 paired prompts and evaluate four open-weight models across 10 steering coefficients and five memory-defense configurations on MemSyco-Bench (answers for all 1,550 items; defense conditions judged on a fixed 250-item subsample), with three LLM judges. Three of the five configurations are new (rewriting every memory, a Router Gate that keeps, rewrites, or drops each memory, and dropping all memory); the other two are MemSyco's baselines. Selective Router Gate filtering preserves substantially more of MemSyco's average accuracy than complete memory removal, and this separation persists when the models are steered toward sycophancy. On Llama 3.1 8B with Router Gate, mild inverse steering ($\alpha = -1.5$) lowers judge-averaged sycophancy from 35.80% to 31.32% while average accuracy moves from 43.99% to 43.31%; this reduction has the same direction under all three judges but is not statistically significant (paired $p = 0.08$ to $0.63$ on 149 items). External memory filtering is the part of the design that holds up; our data do not show that inverse steering adds to it.
Read the full abstract

All authors Ritvij Sharma, Russell Dlugosz, Ryan Zhou, Maheep Chaudhary

Information Retrieval

Rethinking Semantic ID Construction for Generative Recommendation: SimHash with Parallel Decoding and Semantic Alignment

Yuqing Liu, Huiyuan Chen, Yibo Wang et al. (+2 more)

Semantic ID-based generative recommendation represents each item as a sequence of discrete tokens, enabling structured modeling of item semantics. A critical challenge is constructing semantic IDs that are both semantically expressive and computationally efficient. While recent approaches favor complex learned quantization, simple hashing-based methods such as SimHash are widely regarded as fundamentally inferior. In this work, we challenge this consensus by showing that the apparent performance gap does not stem from inherent limitations of hashing, but rather from a structural mismatch with autoregressive decoding, coupled with the inevitable information loss during rigid discretization. Based on this insight, we propose FLASH, a two-stage framework that revitalizes training-free SimHash tokenization through parallel decoding and explicit semantic alignment. Despite its simplicity, FLASH achieves state-of-the-art performance across multiple datasets without requiring any tokenizer training, while exhibiting stronger generalization in cold-start scenarios. Notably, we demonstrate that semantic alignment acts as a universally effective mechanism across diverse paradigms. Our findings suggest that, with compatible decoding and semantic grounding, simple and efficient tokenizers can achieve performance comparable to complex learned counterparts in generative recommendation. Our code is available at https://github.com/KevinC2015/Flash.
Read the full abstract

All authors Yuqing Liu, Huiyuan Chen, Yibo Wang, Wooseong Yang, Philip S. Yu

General Topology

$\varkappa$-Fréchet–Urysohn and Baire subgroups of free topological groups

Arkady Leiderman, Evgenii Reznichenko and Ol'ga Sipacheva

A topological space is $\varkappa$-Fréchet–Urysohn if every point in the closure of an open set $U$ is the limit of a sequence of points of $U$. We prove that every $\varkappa$-Fréchet–Urysohn subgroup and every Baire subgroup of the free topological group on a Tychonoff space $X$ in any full variety of topological groups (in particular, of the free topological group $F(X)$, the free Abelian topological group $A(X)$, and the free Boolean topological group $B(X)$) is discrete. In particular, each of these groups is $\varkappa$-Fréchet–Urysohn (or Baire) only if $X$ is discrete, and all their metrizable and even Fréchet–Urysohn subgroups are discrete. Along the way we characterize strongly $\varkappa$-Fréchet–Urysohn spaces in the spirit of Michael's theorem and study countable families of closed sets absorbing convergent sequences in $\varkappa$-Fréchet–Urysohn spaces and groups.
Read the full abstract

High Energy Physics (Phenomenology)

Born-Oppenheimer Effective Field Theory as a unified framework for all the XYZ exotics

Tommaso Scirpa

We discuss some recent applications of the Born-Oppenheimer effective field theory (BOEFT) formalism, directly derived from QCD and grounded in scale separation and symmetries, to hadronic systems containing two heavy quarks. This formalism provides a unified description of both ordinary and exotic quarkonium states (i.e., hybrids, tetraquarks, pentaquarks, ...) dynamics via coupled Schrödinger equations, whose potentials are computable on the lattice. The aim is to unveil the elusive pattern of the XYZ exotic hadrons. Here, we apply the BOEFT formalism to the $\chi_{c1}(3872)$, $T_{cc}(3875)^+$, and their bottom counterparts. Moreover, we assess the threshold effects on the quarkonium spectrum and discuss the inclusive production cross section for $\chi_{c1}(3872)$.
Read the full abstract

Machine Learning

Fed-BRDECS: Privacy-Preserving and Heterogeneity-Aware Federated Deep Embedded Clustering

Haemin Park, Diego Klabjan, Martin W. Braun et al. (+2 more)

Federated deep clustering seeks to learn clustering-friendly representations from decentralized unlabeled data while preserving client privacy. However, Deep Embedded Clustering (DEC)-style objectives depend on global soft-assignment statistics that require clients to reveal their sensitive information. We propose Fed-BRDECS, a privacy-preserving and heterogeneity-aware federated deep embedded clustering framework. Fed-BRDECS replaces the globally normalized clustering objective with a locally computable sample-stability loss, avoiding the transmission of local soft-assignment distributions. To tackle non-IID client distributions, we introduce prediction-balanced sampling, which oversamples locally rare predicted clusters without requiring ground-truth labels, and centroid-level restarting, which periodically refreshes biased or inactive centroids. Experiments on image and text clustering benchmarks show that Fed-BRDECS consistently outperforms representative federated clustering and deep clustering baselines under both IID and non-IID partitions. We further demonstrate its applicability to federated time-series anomaly detection, where it improves reconstruction-based detectors without adding inference-time cost.
Read the full abstract

All authors Haemin Park, Diego Klabjan, Martin W. Braun, Xiuqi Li, Balakrishnan Ananthanarayanan

Robotics

An Autonomous, 3D Printed, Waterjet-Powered, Open-Source Robotic Trimaran for Environmental Inspection and Monitoring

Reuben O'Brien, Martin Lambrechtse-Reid and Minas Liarokapis

Versatile, autonomous robotic boats can offer excellent environmental inspection and monitoring solutions for remote, dangerous, hard to reach, or access protected water bodies. This paper introduces such a platform in the form of an autonomous, cost-effective, waterjet-powered robotic trimaran. Motivated by the need for an efficient aquatic monitoring, particularly in Aotearoa - New Zealand's diverse environments, the trimaran provides an efficient, low-cost, and easy to replicate alternative to resource-intensive research vessels. The proposed platform, costs $600-1,500 USD to develop (depending on the sensing system configuration), weighs under 5 kg, and excels in bathymetry and water quality testing. The trimaran can reach speeds of up to 2 m/s offering obstacle avoidance of natural features, such as rocks. Utilizing off-the-shelf components and 3D printing technology, the proposed platform offers excellent reproducibility and robustness while operating in shallow waters with its jet propulsion system. The paper presents in detail the design characteristics, the sensing system employed, testing results focusing on bathymetry, and highlights the ability of vessel and the potential for future research and data collection.
Read the full abstract

Digital Libraries

Citations Are Late: Reading epistemic instability from what papers believe, years before the citation graph catches up

Andrey Ustyuzhanin, Ekaterina Trofimova and Denis Zuenko

Paradigm shifts in science are visible in what researchers assert and contest before they are visible in the citation graph. We ask whether a cheap, content-level signal of epistemic instability, derived from the changing distribution of stated modelling beliefs in paper abstracts, can anticipate a paradigm shift earlier than the dominant citation-based disruption index (CD5). On the displacement of recurrent networks by Transformers in NLP, a pre-registered content signal crosses its detection threshold in 2016-Q1, whereas a real-time CD5 monitor cannot even observe the 2017 breakthrough until 2022-Q2, since CD5 needs a five-year forward-citation window: a lead of about 25 quarters. The flat citation baseline is not an artifact of one index, as our CD5, a reference-normalised variant, and an authoritative precomputed index all sit near zero across the shift. The lead is also not a faster proxy: at equal latency the signal beats the content competitors tested, including a learned CD-from-text model and an embedding disruption measure, and the decomposition sees what a keyword cannot (keyword-blind AUC of about 0.9, reproduced on human labels). The lead over CD5 is an observability lead. Which signal carries it depends on the shift: belief adoption in NLP, contestation and applicability stress in vision. Measured against the breakthroughs themselves, the signal leads by five quarters in NLP and is contemporaneous in computer vision.
Read the full abstract

Robotics

What the Elevation Map Cannot See: Semantic-Aware Locomotion and Execution-Aware Navigation for Humanoid Robot

Shunyu Yao, Songyang Liu, Dinghao Chen et al. (+2 more)

Navigation for humanoid robots is critical, yet large-scale evaluation on physical hardware is often impractical due to cost and safety concerns, making simulation benchmarks essential. Existing VLN benchmarks achieve physically executable navigation, but still assume (1) all hazards are observable from elevation maps; (2) realized motions closely match desired motions. In real environments, however, fallen bottles may be ambiguous in elevation maps, while phones and water spills may be difficult to differentiate; hazard avoidance by the locomotion policy can cause the robot's actual trajectory to deviate from the path intended by the VLN policy. Such command-execution mismatch can accumulate and lead the robot toward unintended locations. To expose these failure modes, we introduce a benchmark that models both elevation-subtle hazards and execution deviations, together with a closed-loop VLN + locomotion control framework that continuously realigns high-level navigation with the robot's actual state. We evaluate navigation in simulation and further validate the locomotion policy on a physical Unitree G1 humanoid robot. Results show that semantic input reduces contact with hazards poorly represented in elevation maps, while anti-deviation improves navigation success. These findings highlight the need to evaluate humanoid navigation jointly in terms of route completion and hazard avoidance.
Read the full abstract

All authors Shunyu Yao, Songyang Liu, Dinghao Chen, Yuanyuan Lei, Shuai Li

Earth and Planetary Astrophysics

Tracing the main belt origins of S-complex near-Earth objects: From space mission targets to meteorite sources

P. N. Simon, P. Vernazza, M. Marsset et al. (+5 more)

Context. Space missions to near-Earth objects (NEOs) provide unique opportunities to study and characterise their physical and compositional properties. Interpreting these results requires knowledge of their provenance, i.e. their source family in the main asteroid belt. Aims. We aim to identify the family sources of six S-complex NEOs visited or soon to be visited by space missions, and to investigate the size-dependent orbital behaviour of the broader S-complex NEO population using the MITHNEOS dataset. Methods. We combine near-infrared spectral fitting with ordinary chondrite (OC) analogues reddened by a space-weathering model, using reduced $\chi^2$ minimisation, and cross-validation with probabilistic dynamical maps of the main NEO source families. Results. We identify LL6-7 chondrites and the Flora family for Eros, Itokawa, Apophis, and Torifune; LL3 chondrites and the Nysa family for Toutatis; L chondrites and the Massalia family for Didymos-Dimorphos. Where in situ observations exist (Itokawa and Eros), the classifications are consistent. Applied to the broader S-complex NEO population, the spectral classification and orbital properties distributions are broadly consistent with dynamical models. Accounting for observational biases, we find that at small diameters ($\leq$100 m), the fractions of H- and L-type objects increase at the expense of LL6-7 types, with low-inclination orbits consistent with origins in the Massalia (L) and Koronis (H) families. Conclusions. Combining spectral fitting with OC analogues and dynamical probability maps robustly identifies the source families of individual NEOs. Below $D\sim$100 m, the compositional distribution of S-complex NEOs approaches that of meteorite falls, providing direct observational evidence, based on spectroscopic measurements of S-complex NEOs, of the emergence of the source regions of OC meteorites.
Read the full abstract

All authors P. N. Simon, P. Vernazza, M. Marsset, M. Brož, R. P. Binzel, F. DeMeo, C. A. Thomas, L. Jorda