Plasma Physics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Wednesday 2 September
Expensive fusion simulations are commonly preserved in code-specific formats that limit discovery, comparison, and reuse. We present the Multiscale GyroKinetic DataBase (MGKDB), an open-source software framework and curated archive that converts heterogeneous simulation campaigns into traceable scientific records. Each record links code-native inputs and outputs to provenance and quality metadata, an IMAS-aligned physics representation, and derived diagnostics, preserving model-specific evidence while enabling common-field queries. Production pathways support linear and nonlinear GENE and CGYRO calculations and reduced quasilinear TGLF evaluations. At the September 1, 2026 snapshot, MGKDB contained 1,068,089 records, nearly all of which included a populated gyrokinetics IMAS branch. The software is openly available, while access to the NERSC-hosted production records is managed. Three demonstrations show how these linked representations support scientific reuse. Standardized quantities stored in the Diagnostics branch enable population-scale analysis of archived linear modes; common input coordinates reveal coverage, redundancy, and campaign-driven sampling structure across a multicode collection; and record-level retrieval of native CGYRO inputs drives matched TGLF calculations and produces a traceable dataset for exploratory surrogate modeling. Together, these examples demonstrate how MGKDB supports archive characterization, candidate cross-code and cross-fidelity comparisons, campaign planning, and reproducible data-driven modeling without treating different models as automatically equivalent.
Commutative Algebra
Symbolic Powers and Asymptotic Invariants of GL-Invariant Ideals
This work concerns ideals invariant under the action of the group of linear base changes on a generic matrix; we call these GL-invariant ideals. We determine the ordinary powers, their saturations with respect to determinantal ideals, and the symbolic powers of GL-invariant ideals. We give explicit formulas for asymptotic invariants known as (skew) Waldschmidt constants and asymptotic resurgence, which measure the growth of these families and compare the ordinary and symbolic topologies. We also prove that the generalized symbolic Rees algebras associated with these ideals are Noetherian.
Read the full abstract
Medical Physics
Beyond Dose in Boron Neutron Capture Therapy: Cellular $^{10}$B-Capture Statistics and Microdosimetric Context in Effect Prediction
Boron neutron capture therapy (BNCT) produces high-linear-energy-transfer, short-range particles through the $^{10}$B(n,$\alpha$)$^{7}$Li reaction. Conventional descriptors, including absorbed dose, compound biological effectiveness (CBE)-weighted dose, and photon-isoeffective dose ($D_{\mathrm{isoE}}$), remain clinically useful but compress cellular heterogeneity, compartmental boron localization, and stochastic reaction occurrence into macroscopic summaries. We propose a hierarchical framework that places cell-associated $^{10}$B-capture reaction statistics upstream of microdosimetric effect modeling and links clinical observations to cellular-scale determinants of BNCT response. We synthesize published work on BNCT microdosimetry, CBE and $D_{\mathrm{isoE}}$, stochastic microdosimetric kinetic modeling, PHITS-based cellular simulations, boron imaging, tissue microstructure, and boron-agent characterization, reorganizing these elements into observational, latent-cellular, reaction-count, and biological-effect layers. Imaging, pathology, agent-specific data, and neutron-field information constrain latent variables such as cellular boron burden, compartmental localization, and geometry; these variables determine expected reaction burdens and distributions of realized reaction counts, which are then translated into cell survival, tumor control, or normal-tissue injury by microdosimetric models. Cell-associated reaction-count statistics thus provide an interpretable inferential bridge between macroscopic measurements and established BNCT effect models, supporting spatially resolved and uncertainty-aware effect prediction without replacing existing approaches based on CBE, $D_{\mathrm{isoE}}$, SMK, or PHITS.
Read the full abstract
Machine Learning (Statistics)
A Closed-Form Formula for Consistent Lipschitz Regression on Metric Spaces with Sparse Neural Network Realizations
Several classical machine-learning methods, such as KRRs and SVRs, are both computationally and analytically tractable since their estimators either admit closed-form expressions or are obtained by minimizing convex training objectives; neither feature is generally available for deep neural networks. We address this by introducing a simple closed-form “two-stage” compositional formula $\hat{f}$ for reconstructing an unknown Lipschitz function $f:\mathcal{X}\to \mathbb{R}$ on a metric space $(\mathcal X,\rho)$ from $N$ i.i.d. noisy observations. Our main result is a high-probability uniform ($L^{\infty}$) recovery guarantee that jointly controls approximation and statistical errors while enjoying an optimization error of zero; in particular, we do not assume oracle access to an approximate ERM. Our secondary main results establish the optimality of our formula in three complementary senses. 1) Function space: On Ahlfors-regular metric spaces, the hypothesis class parameterized by our formula attains the optimal fat-shattering dimension. 2) Parameter space: Its dependence on the parameters is maximally numerically stable, in the sense that a smaller approximation error cannot be achieved with a smaller Lipschitz dependence on the model parameters. 3) Forward pass: Its dependence on the input is maximally regular, matching the Lipschitz constant of the target function $f$. When $\mathcal X=[0,1]^d$ is equipped with the $\ell^\infty$ norm, $\hat{f}$ admits algorithmic ReLU-MLP and exact ReLU-multi-head transformer realizations of depth $\mathcal{O}(\log(N))$ with $\mathcal{O}(N)$ nonzero parameters.
Read the full abstract
Analysis of PDEs
Global existence and uniqueness of solutions in a chemotaxis system with density-suppressed motility, a $\theta-$logistic source, and indirect signal production under a strong Allee effect
We consider the following chemotaxis-growth system, featuring density-suppressed motility and a $\theta$-logistic growth term that incorporates a strong Allee effect: \begin{equation*} \left\{ \begin{aligned} u_t &= \Delta(\gamma(v)u) + \mu u(1-u^\theta)(u-A), & x \in \Omega, \quad t > 0, v_t &= \Delta v - v+ w^\beta, & x \in \Omega,\quad t > 0, w_t &= -\delta w +u, & x \in \Omega, \quad t > 0, \end{aligned} \right. \end{equation*} subject to homogeneous Neumann boundary conditions in a bounded domain $\Omega\subset \mathbb{R}^d (d\geq 2)$ with smooth boundary. Here $\mu\in \mathbb{R} , \delta,\beta>0, \theta\geq1$, and the positive motility function $\gamma(v) \in C^3([0,\infty))$ fulfills $\gamma'(v) \leq 0$ for all $v\geq 0$. The main objective of this paper is to establish the global existence and boundedness of classical solutions to the proposed problem. Our analysis combines the Schauder fixed point theorem for proving local-in-time existence with the extensibility criterion, fundamental energy estimates, $L^p-$bounds, the Gagliardo-Nirenberg interpolation inequality, Young's inequality, semigroup estimates, and a Moser-type iteration scheme to derive uniform-in-time $L^{\infty}-$bounds, thereby ensuring global bounded classical solutions. More precisely, we establish that the chemotaxis-growth system admits a unique globally bounded classical solution whenever $\beta < \frac{2(\theta+2)}{d}$. Furthermore, if the logarithmic derivative of the motility function is assumed to be uniformly bounded on $[0,\infty)$, that is, $\frac{\gamma'(v)}{\gamma(v)} \in L^{\infty}([0,\infty))$, then the above restriction on $\beta$ can be weakened to $\beta < \frac{d(\theta+1)+2(\theta+2)}{2d}$, provided that $\theta > \frac{4-d}{d-2}$.
Read the full abstract
Number Theory
Skolem-Mahler-Lech in rings of positive characteristic: a shorter proof and a multi-dimensional generalization
Let $R$ be a commutative ring and $f(a_1, \ldots, a_n) = \sum_{i=1}^k r_{i1}^{a_1} \cdots r_{in}^{a_n} m_i$ be a linear-exponential map over an $R$-module $M$. Dong and Shafrir (2026) showed that, when $\ell M = 0$ for some $\ell \in \mathbb{N}_{>0}$, the zero set of $f$ is the intersection of effectively computable $p$-normal sets, where $p$ ranges over the prime divisors of $\ell$. This generalizes an earlier theorem of Derksen and Masser (2012) on the solution set of $S$-unit equations over fields of positive characteristic. The purpose of this paper is twofold. First, we give a shorter proof of Dong and Shafrir's result, using the theorem of Derksen-Masser as a blackbox. Our proof also yields a decomposition of the zero set as a positive Boolean combination of affine transformations of zero sets of linear-exponential equations over fields. Second, we prove a multi-dimensional generalization of the Skolem-Mahler-Lech theorem over rings of finite characteristic. Specifically, we show that the zero set of every $n$-dimensional linear recurrence sequence over an $R$-module $M$ satisfying $\ell M = 0$ is the intersection of effectively computable $p$-normal sets (in $\mathbb{N}^n$), where $p$ ranges over the prime divisors of $\ell$. For example, this gives a decision procedure for whether two classical linear recurrence sequences have a common value over a ring of characteristic $p^a$ or $p^a q^b$, where $p$ and $q$ are primes.
Read the full abstract
High Energy Physics (Theory)
Boundary-Value Problem in Type IIB Supergravity and Holography
In precision holography, the Euclidean on-shell action of type IIB supergravity often fails to reproduce the leading large-$N$ free energy of the dual field theory: the type IIB pseudo action, for instance, vanishes identically on the $\mathrm{EAdS}_5\times S^5$ background. We address this issue by revisiting the boundary-value problem of the Euclidean type IIB pseudo action from first principles. Classifying the admissible boundary conditions through the variational principle, we construct a generalized pseudo action whose boundary terms implement the choice of ensemble — fixed potentials versus fixed quantized Page charges — in which the holographic comparison is performed. For the fixed five-form-charge ensemble, the resulting boundary term reproduces the recently proposed topological correction to the Pasti–Sorokin–Tonin formulation of type IIB supergravity for holographic backgrounds of interest. We then test the generalized pseudo action on two complementary backgrounds: the $\mathrm{EAdS}_3\times S^3\times M_4$ near-horizon geometry of the D1-D5 system and the warped $\mathrm{EAdS}_6\times S^2\times\Sigma$ solutions dual to five-dimensional SCFTs. In both cases the ten-dimensional on-shell action agrees exactly with that of the corresponding lower-dimensional supergravity, and thereby with the leading large-$N$ free energy of the dual SCFT. Our results establish the choice of boundary conditions — and hence of ensemble — as an essential ingredient of precision holography directly in type IIB supergravity, extending recent analyses in eleven-dimensional supergravity.
Read the full abstract
Hardware Architecture
A Time-Encoded Analog Photonic Interposer for Energy-EfficientIntegration of Analog Vision Sensors and Analog Accelerators
This work introduces a time-encoded analog photonic interposer that enables long-distance, high-fidelity transport of analog signals between spatially separated chiplets. Unlike prior silicon-photonic links limited to digital data, the interposer preserves analog information by converting amplitudes into timing intervals using an analog-to-time converter (ATC), transmitting them over a wavelength-division-multiplexed photonic link, and reconstructing values at the receiver without an explicit high-precision ADC/DAC data-converter pipeline. The link instead embeds an implicit 6-bit time-domain quantization and uses a single wavelength per processing element independent of bit precision. Evaluated in a fully analog vision pipeline with an in-pixel computing (IPC) sensor, it achieves a 2.04x energy–delay product (EDP) improvement over an 8-bit digital electrical baseline on the 560x560 Visual Wake Words dataset, with the advantage widening with link length even against a precision-matched 6-bit baseline. The pipeline holds 89.87% and 86.15% accuracy on ResNet18 and MobileNetV2 for VWW and generalizes across CIFAR-10 and ModelNet40 within 2% of the digital baseline.
Read the full abstract
General Relativity and Quantum Cosmology
Relativistic transformation of temperature for exotic cosmological fluids
We study the relativistic transformation of temperature for effective cosmological fluids with negative pressure within a covariant thermodynamic framework. The Lorentz transformation of temperature is derived for barotropic fluids with $p=w\rho$ and for the (generalized) Chaplygin gas with $p=-A/\rho^\alpha$. For barotropic fluids with constant negative equation-of-state parameter, we show that negative pressure suppresses the growth of temperature under relativistic boosts. In contrast, the Chaplygin gas exhibits a qualitatively different behavior due to its density-dependent pressure, leading to a monotonic increase of the transformed temperature with velocity. The dependence of the relativistic temperature on the relative velocity and background energy density is illustrated through representative examples.
Read the full abstract
Statistical Mechanics
Death by mutants: unusual multicritical dynamics in a two-species model for absorbing state transitions
We explore the phases and active-to-absorbing state phase transitions (AAPT) in a two-species model, where the species A and its mutant B are asymmetrically or nonreciprocally coupled. We identify a multicritical point that connects the global absorbing state of both the species and its mutant, with a uniform active state. The critical dynamics at this multicritical point is studied within the lowest order perturbation theory. This asymmetric coupling between species A and mutant B leads to unequal and distinct upper critical dimensions $d_c^A$ and $d_c^B$ respectively for A and B dynamics. We show that the multicritical point in this model is characterised by an unusual breakdown of scale-invariance by fluctuation-induced logarithmic modulations with power law behaviour of the order parameter and correlation lengths of the species A at all dimensions $d<d_c^A$. Above $d_c^A$, conventional scale-invariance is restored. The dynamics of the mutant species B belongs to the DP universality class with an upper critical dimension of $d_c^B=4$
Read the full abstract
Game Theory
Turn-Based Combat Arena: A New Framework for Multiagent Training and Game Balancing
This paper is the first in a series on Turn-Based Combat Arena, a configurable framework for turn-based strategy games designed to support the efficient training and evaluation of machine learning agents. The proposed framework enables flexible modification of game rules and parameters, allowing rapid experimentation across diverse scenarios. Its architecture is optimized for high-throughput simulation, supporting tens of thousands of games per second and enabling the storage and processing of billions of gameplay records on a single machine. The problem of balancing the game, and particularly the parameters of game units, is investigated in detail. We evaluate several optimization approaches and show that multiple methods converge to comparable solutions, suggesting robustness in identifying balanced game configurations. These results indicate that the framework can serve as a practical platform for both game design analysis and agent training.
Read the full abstract
Metric Geometry
The stealthiest particle configurations in 8 and 24 dimensions
We show that the $E_8$ and Leech lattices generate the stealthiest isometry-invariant, locally square-integrable point processes of intensity $1$ in $\mathbb{R}^8$ and $\mathbb{R}^{24}$, respectively, and that they are the unique stealthiest processes. We also apply the proof technique to sphere packing and energy minimization, and we construct a $20$-dimensional periodic configuration that is stealthier than any known $20$-dimensional lattice with the same particle density. This periodic configuration is a formal dual of Vardy's $20$-dimensional sphere packing.
Read the full abstract
Systems and Control
On a Gradation for Asymptotic Stability
Classical asymptotic stability guarantees convergence but does not quantify the rate at which convergence occurs. This paper introduces a gradation of asymptotic stability where degree zero corresponds to exponential stability and degree $m>0$ corresponds to algebraic decay of order $t^{-1/m}$. We provide direct and converse Lyapunov characterizations for admissible degrees and conditions for certifying the exact stability degree. Hopf, Bautin, fractional-degree, and time-varying examples demonstrate how the degree identifies the leading stabilizing mechanism.
Read the full abstract
Solar and Stellar Astrophysics
Infrared-excess sources in VVV: candidate debris discs, SED characterization, and variability analysis
Debris disks around main-sequence stars provide important insights into planetary formation and evolution; however, identifying these systems in the Galactic plane, where high interstellar extinction and source crowding are significant, remains challenging. Deep, wide-area surveys such as the VISTA Variables in the Via Lactea (VVV) ESO Public Survey provide multi-epoch infrared observations, enabling systematic searches in these environments. We present a method to construct a refined sample of main-sequence candidates by cross-matching data from Gaia DR3, GLIMPSE, DECaPS, and VIRAC2, to identify infrared excesses through spectral energy distribution (SED) analysis, and to analyse time-series data to detect variability potentially associated with circumstellar material or other astrophysical phenomena. Although infrared excess is commonly associated with debris disks, contributions from unresolved binaries or line-of-sight contamination cannot be excluded. Applying this methodology to the VVV tile d077 using the VIRAC2 catalogue, we analyse an initial sample of 879,556 sources and identify 446 sources flagged by the VO SED Analyser (VOSA) excess-detection algorithm as having infrared excess, of which 171 show candidate periodic variability in the VIRAC2 Ks-band light curves. We further perform SED fitting using theoretical models to derive physical parameters and approximate spectral types for these sources. This work provides a catalogue of infrared-excess candidates in a Galactic field affected by high extinction, along with a reproducible method suitable for future large-scale applications.
Read the full abstract
Networking
Efficient Co-simulator Integration with Application to Smart Grids
The evolution of the electric power grid towards a “smart grid” is an example of an emerging large-scale cyber-physical system. To study the potential of cyberattacks or other sources of extreme events, high-fidelity co-simulation is one of the few realistic strategies. In this work, we demonstrate a practical and high-performance co-simulation strategy for smart grids, which provides lessons about co-simulation integration that can be carried over to other cyber-physical systems. Specifically, we present a co-simulation platform based on the Mosaik framework that integrates several federates including OpenDSS for power flow, a refined NS-3 for communication networks, and custom Python-based simulators for on-load tap changer control, distributed state estimation, and data collection. We compare synchronization strategies – exhaustive lock-step time advancement versus an optimized event-driven approach that exploits lower-bound time stamps (LBTS) and next-event prediction – and introduce targeted refinements to NS-3 event handling and relevance filtering of internal events. Performance is evaluated on two standard IEEE benchmark systems: the 13-node test feeder with tap-changer voltage regulation and a large-scale 33-bus system augmented with 32 European low-voltage feeders (1,793 nodes total) performing distributed state estimation with thousands of phasors and smart meters. Experimental results demonstrate that the refined event-based synchronization that uses LBTS-based NS-3 event filtering reduces execution time by up to 50% over non-filtered LBTS, and more than 4$\times$ in the tap control scenario compared to naïve lock-step methods, while strictly preserving temporal correctness and simulation accuracy.
Read the full abstract
Machine Learning
Kernel Reboot: Breaking the Boundaries of Neural Tangent Kernels for Neural Fields
Neural fields (NFs) map continuous coordinates to signals such as color or density, but fast high-quality reconstruction from sparse observations remains difficult. Classical Neural Tangent Kernel (NTK) regression gives closed-form fits, yet it is fundamentally linear and cannot accumulate reusable task priors. We develop three algorithms that address these gaps. NTK-KIP learns a distilled support set of coordinates (and optional labels) so that a finite NTK can inpaint large missing regions from little observed data, yielding a compact non-linear representation instead of a raw kernel solve. MetaQuill meta-learns a shared initialization for an INR so that new scenes can be adapted by updating only a small task-specific weight offset, which provides true feature learning and a reusable prior. Finally, MetaQuill-KIP fuses both ideas: it seeds the task with a KIP-style non-linear warm start, then refines only that small offset around the meta-learned initialization. MetaQuill-KIP achieves high-PSNR reconstructions and semantically plausible inpainting under very sparse observations, while requiring only lightweight per-instance adaptation, whereas diffusion-style baselines typically depend on large pretrained generative priors and costly per-image tuning. This shows that NTK-driven neural fields can be made both non-linear and meta-learnable, narrowing the gap between analytic kernels and practical few-shot reconstruction.
Read the full abstract
Combinatorics
On the reconstruction of $n$-quasigroups of order $4$ and upper bounds on their number
(This is a reprint of the thesis published in the Transactions of the Conference Devoted to the 90th Anniversary Of Alexei A. Lyapunov, Novosibirsk, October 8-12, 2001, pages 323-327). The number of ways for partial $n$-quasigroups of order $4$ to be extended to $n$-quasigroups is investigated. Upper bounds on the number of $n$-quasigroups of order $4$ are obtained. The asymptotic $3^{n+1}2^{2^n+1}$ of that number is established.
Read the full abstract
Trading and Market Microstructure
Mean-field equilibrium of heterogeneous agents under market impact
Although market participants generally have access to a common information set, they make decisions based on forecasts formed over heterogeneous horizons. Because market impact depends on aggregate positions rather than trader identities, these decisions feed back into prices through their collective effect. We introduce a linear mean-field model of this interaction. The observed price is decomposed into a martingale component, a common predictable signal represented by a Volterra process, and the market impact generated by aggregate positions. Agents take positions according to conditional forecasts of future signal increments and a fraction of anticipated aggregate impact over their respective horizons. Within a Gaussian-Volterra framework, we characterize equilibrium through a linear fixed-point equation for aggregate positions and establish existence and uniqueness under explicit conditions. At equilibrium, we identify a balance condition that cancels the direct transmission of the common signal to the observed price. We then study the limit in which agents fully account for market impact. Along a suitably scaled family of equilibria satisfying explicit conditions, the contributions of the predictable signal and the market impact cancel in the limit, and the observed price converges to its martingale component. For fractional-type signals and Gamma-distributed horizons, we further derive local Hölder bounds and identify the horizon distributions for which the observed price has local regularity compatible with that of Brownian motion.
Read the full abstract
Programming Languages
Enhancing the Power of Polyhedral-Based Optimizations with Coordinate-Based Hill Climbing
This paper describes our experience extending the polyhedral compiler Pluto with a lightweight, coordinate-wise hill-climbing tuner that adjusts numeric transformation parameters, such as tile sizes and thread-block dimensions, after Pluto selects the kernel's loop structure. To ensure fast convergence and escape local minima, hill climbing is augmented with two techniques: expanded neighborhood exploration and a shortest-hop refinement phase. On x86 and ARM CPUs, tuned kernels outperform Pluto's default configuration (1.06-1.28x geometric mean speedup across 11 benchmarks) and static optimizers (Clang -O3, Polly, IOOpt), reaching performance competitive with the AutoTVM autotuner at substantially lower search cost. Applying the same technique to GPU thread-block allocation on an NVIDIA A100 yields 5.5-8.5% improvement over default configurations. These results position post-optimization parameter tuning as a practical middle ground between fixed-cost-model polyhedral compilation and full autotuning.
Read the full abstract
Statistical Methodology
Multimarker genetic association tests for panel count data
The existing multimarker survival tests focus on time to event outcomes. However, recurrent events are common in real world clinical and biomedical studies, especially in the research of chronic and recurrent diseases. In this paper, we develop a suite of set based genetic association tests for panel count outcomes under a unified weighted V statistic framework. These tests can effectively account for genetic effect heterogeneity in panel count data. Additionally, we develop small sample corrections to the tests to enhance the accuracy of the tests under small samples. Simulation studies show that the proposed tests perform well in terms of size and power across various scenarios and have a bigger power than the existing set based tests for interval censored outcomes. A dental caries GWAS data set is analyzed to illustrate the utility of the tests.
Read the full abstract
Number Theory
Poissonian pair correlations for the three-particle Sutherland model
We prove a Berry-Tabor theorem for the repulsive three-particle trigonometric Sutherland Hamiltonian. For fixed total momentum, its relative spectrum is an affine translate of the positive-definite $A_2$-form. For a Diophantine coupling, the desymmetrized spectrum has Poissonian pair correlation. The proof reduces the problem to a finite-congruence and Weyl-sector version of Marklof's theorem on inhomogeneous quadratic forms.
Read the full abstract
Statistical Mechanics
Large deviations for linear regressions
Linear regression is one of the simplest and most widely used tools to learn patterns from data: it fits a set of coefficients so that a linear combination of predictors best matches observed responses. The quality of the fit is measured by the residual sum of squares, the total squared mismatch between predictions and data, whose minimum defines the training loss. We consider Gaussian design and noise, with teacher coefficients independently drawn from a general distribution $p(\beta)$, and a general class of separable regularizers, including Ridge and Lasso. Using the zero-temperature replica method, we compute analytically the large-deviation statistics of the minimum training loss for large numbers $P$ of predictors and $N$ of observations, with $r=P/N$ fixed. The rate function we compute governs rare sample-to-sample fluctuations of the optimal loss. Extensive numerical simulations are in excellent agreement with our theory and clearly show a pronounced deviation from the Gaussian regime of typical fluctuations in the tails.
Read the full abstract
Soft Condensed Matter
Critical scaling and soft granular rheology of frictionless non-Brownian suspensions across jamming
We carry out 3D computer simulations to study suspension rheology across the jamming transition for deformable non-Brownian frictionless particles. The simulations are carried out at constant shear rates and suspension densities ($\dot{\gamma} , \phi$), for polydisperse spherical particles. The suspension is probed at densities in the neighbourhood of the jamming point $\phi_c$ , as well as across much wider density ranges. We carry out critical scaling analysis of our close-to-jamming data and confirm that shear-driven jamming is a second-order critical phenomenon requiring strong corrections on shear. We build maps of the local strain rate exponent to classify flow regimes and examine the effect of the corrections on the suspension rheology. Further, using the critical scaling jamming density, we carry out soft granular rheological scaling over the wider density range, demonstrating an excellent data collapse. The soft granular rheology relations are able to predict our data over almost the entire dynamical and density range.
Read the full abstract
Computer Vision
SLIDEFORGE: An LLM Agent for Controllable Editing of Slides as Structured Artifacts
Current AI agents compellingly describe slides. However, AI-assisted slide editing requires more than understanding: the output must retain layout, style, component structure, and native editability. Towards, AI-assisted slide editing, existing agents operate on screenshots or weak document representations and often fragment coherent visual units, rasterize editable content, or break layout. In contrast, for controllable slide editing, we introduce an agentic framework, SLIDEFORGE, which builds a Deck State Graph, an executable slide state that links visual decomposition, native pptx object structure, and perceptual organization. By recovering human-referable components while retaining fine-grained editable structure, SLIDEFORGE supports theme-preserving reconstruction through slide-native operations and rendered-state verification. We further introduce an evaluation paradigm for controllable slide transformation that jointly measures component recovery, preservation, restyling consistency, visual quality, and native editability. Experiments show that SLIDEFORGE outperforms direct prompting, screenshot-based agents, and generic code-agent baselines across these dimensions. Code is available at https://github.com/UIUC-MONET/SLIDEFORGE.
Read the full abstract
Applied Physics
A Langmuir-Like Adsorption Model for MoS2 Gas Sensors: Bridging Theory and Experiment
Gas sensors based on layered TMDCs (transition-metal dichalcogenides) such as MoS2 have attracted considerable attention recently and are considered as a promising alternative to the conventional metal-oxide-based gas sensing devices. Recent studies on TMDC gas sensors have mainly addressed either the theoretical description of adsorption mechanisms or the fabrication of test devices and the investigation of their gas sensing performance. However, a suitable approach for modeling the gas sensing behavior of TMDC devices and for establishing a reliable connection between theoretical results and the response of experimental devices when exposed to gases remains absent. To bridge this gap, in the present work a new Langmuir-based adsorption model, capable of calculating both the sensor sensitivity and selectivity, is developed and applied to MoS2-based gas sensors. It is shown that by using appropriate published data for the adsorption energy and the charge transfer as input data, the reported experimental sensitivities of MoS2 gas sensors can be well reproduced by the model.
Read the full abstract
Sound
CRAW: Codec Robust Audio Watermarking
Recent advances in generative speech models have made it increasingly difficult to distinguish authentic from synthetic audio, enabling new forms of fraud and misinformation. Audio watermarking offers a promising defense by embedding an imperceptible signal into generated speech that can later be detected to verify its provenance. However, recent studies have shown that existing post-hoc watermarking methods fail under neural codecs and denoisers, transformations routinely applied during real-world storage, transmission, and processing, severely limiting their practical utility. Here we introduce CRAW, a codec-robust audio watermarking framework that jointly improves robustness against neural re-synthesis while maintaining high perceptual quality. CRAW combines distortion-aware training with an attention-based pooling mechanism, inference-time perceptual mask- ing, and an error-correcting code to recover the fidelity lost during robust training. Experiments demonstrate that CRAW achieves state-of-the-art robustness against neural codecs, denoisers, and vocoders while maintaining perceptual quality comparable to existing post-hoc watermarking methods. The code is available at https://github.com/DavidC1212/craw.
Read the full abstract
Machine Learning
Scaling Laws, Tabular Data and Actuarial Ratemaking Models
Scaling laws in modern deep learning describe how held-out loss improves as model capacity, training data, and compute increase, often following power-law trends. We investigate whether analogous scaling regularities arise in actuarial ratemaking, where data are tabular, heterogeneous, and noisy, and where classical models such as GLMs remain strong baselines. Using a real-world motor insurance portfolio, we train models from different families across increasing fractions of the training data and multiple random seeds, evaluating out-of-sample Poisson deviance, a likelihood-based loss for Poisson count predictions in which lower values indicate better held-out fit. We find that all model families improve with additional data, but scaling exponents differ substantially: TabM exhibits markedly stronger data scaling than purely supervised tabular Transformers and standard MLP baselines. Transformer variants show weak parameter scaling unless augmented with additional inductive biases (TabM-style adaptation or self-supervision). These results provide quantitative guidance on model selection by data regime and suggest that effective scaling on actuarial tabular tasks depends on architecture and loss function objective design, with simple increases in Transformer size providing limited gains.
Read the full abstract
Digital Libraries
FOXDEN: FAIR Services for AI-Ready Scientific Datasets
Scientific datasets are most compatible with AI workflows when they are described by rich, machine-readable metadata and provenance information, following the FAIR guiding principles. The FAIR Open-Science Extensible Data Exchange Network (FOXDEN) is a set of cyberinfrastructure building blocks developed at the Cornell High Energy Synchrotron Source (CHESS) for annotating raw, reduced, and analyzed datasets with both structured and unstructured metadata and provenance records. It also allows researchers to publish these records with Digital Object Identifiers (DOIs) to create AI-ready datasets. We describe FOXDEN's architecture and its deployment at CHESS and at the National High Magnetic Field Laboratory, where it serves as a foundation for future agentic scientific workflows.
Read the full abstract
Machine Learning (Statistics)
Occupancy-based Quantile Risk Control
Conformal risk control is an emerging framework for the safe deployment of machine learning models with finite-sample guarantees. To accommodate a broader class of risk notions, quantile risk control extends this framework to quantile-based risk measures. However, existing methods either suffer from excessive conservatism or lack rigorous finite-sample guarantees. To address these limitations, we introduce Occupancy-based Quantile Risk Control (OQRC), a novel method that provides tight risk control bounds with finite-sample validity. Our key idea is to formulate risk control as a finite-occupancy problem by partitioning the loss space with the ordered calibration losses. Specifically, we estimate the distribution of test losses across the resulting bins and upper-bound the risk by the maximum loss attained within each bin. We then select the parameter $\lambda$ such that this upper bound does not exceed a predefined threshold $\alpha$ with high probability $1-\delta$. Theoretically, we establish a finite-sample guarantee showing that OQRC yields tight risk control bounds that converge to the optimal bounds at a provable rate of $\mathcal{O}_ p(n^{-1/2})$. Extensive experiments demonstrate the effectiveness of our method, reducing the risk gap by up to 78.64% on common benchmarks.
Read the full abstract
Instrumentation and Methods for Astrophysics
Optimizing Field Geometry for Modular Micropore-Optics Soft X-ray Time-Domain Surveys
Wide-field micropore (lobster eye) optics enable focusing soft X-ray instruments with fields of view of hundreds of square degrees, but square-format modules require careful geometric arrangement to support both wide-area surveys and sensitive pointed observations. We examine this design problem using the X-ray TRansient Array / Micropore Optics X-ray Imager trade study, developed for a NASA Small Explorer mission concept, as a worked example. Single-module performance scaling is combined with geometric survey simulations to compare modular field layouts, overlap regions, and repeated survey tilings. For realistic micropore-optic parameters, increasing module size primarily expands the field of view once the boresight-centered effective area saturates, making modular architecture and field geometry central design variables. A three-module tri-petal configuration, formed by differentially rolling and offsetting three 10 x 10 sq deg square fields, produces a compound field of approximately 240 sq deg with one-module discovery regions, two-module bridge regions, and a compact three-module core. When repeated in a dense hexagonal lattice, the footprint closes gaps, builds a more uniform exposure floor, and provides enhanced-depth survey regions. This geometry offers an efficient compromise among survey coverage, pointed sensitivity, system complexity, and resilience, and can be scaled to larger persistent soft-X-ray monitoring architectures.