Peer Revenue Play credits. Real papers.

Papers

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

All of arXiv

Thursday 10 September

Physics Education

Questioning your brilliance in physics: Differential shifts in fixed mindsets by grade and gender

Students' domain-specific mindsets and their beliefs about their capacity to improve through effort play a crucial role in shaping their experiences and decisions to persist in STEM disciplines. Physics is generally seen as a field requiring innate brilliance, which can reinforce fixed mindsets, particularly after initial setbacks in performance that are common in introductory university courses. In this study, we examine changes in fixed mindsets and potential gender differences in an introductory calculus-based physics course. Our sample consisted of 508 students with an average age of 18, predominantly White, with men comprising the majority. Based upon survey response distributions, three distinct mindset categories were identified: Hesitant, Hopeful, and Confident, describing how strongly students rejected a fixed mindset in physics. The results suggested large gender differences in distributions at the high and low-end groups. We also found an overall decline toward fixed mindsets across the course, and logistic regressions controlling for initial mindsets showed that women were significantly more likely than men to shift away from the Confident category. While the majority of men tended to stay within the Confident category, the majority of women moved away from it. Particularly, this differential shift was seen among students receiving Bs or Cs, the most commonly awarded grades in this course. Furthermore, there were relatively small differences in the probability of change within men as a function of grades received, whereas women showed marked declines toward fixed beliefs with either a B or C. Our findings provide empirical evidence for the dynamic, grade-sensitive nature of students' mindsets in a calculus-based physics course.
Read the full abstract

Analysis of PDEs

The Schrödinger-Klein-Gordon System Revisited

We introduce a microlocal formulation of the Schrödinger-Klein-Gordon system describing the interaction between a non-relativistic quantum particle and a Klein-Gordon field through Yukawa coupling. Instead of working directly with the Schrödinger wave function, we represent the quantum component by its Fourier-Wigner transform, and we rewrite the Klein-Gordon equation in terms of a complex Fourier variable. Eliminating the field variable yields a closed nonlinear Fourier-Moyal equation on phase space whose unknown is the Fourier-Wigner distribution associated with the quantum component. The resulting formulation provides a refined description of the particle-field interaction at the microlocal level. In particular, the original cubic coupling is transformed into a quadratic self-interaction governed by an explicit bilinear operator. This new representation permits the use of techniques from time-frequency analysis. Building on this, we develop a well-posedness theory for the Fourier-Moyal equation in anisotropic spaces involving Wiener-type norms and prescribed moduli of continuity. Local existence and uniqueness are established under general assumptions on the ultraviolet cutoff, together with propagation of microlocal regularity. We then introduce a new construction of weak L 2 -solutions by exploiting compactness properties of Fourier-Wigner transforms and obtain global existence for prepared data. The aim is to provide an alternative analytical framework for the study of Yukawa-type interactions and to establish a bridge between nonlinear dispersive equations and phase-space methods.
Read the full abstract

Probability

Quantitative Asymptotics for Time-Inhomogeneous Lévy-Driven SDEs with Asymptotically Vanishing Drifts

In this work, we are concerned with a class of multi-dimensional time-inhomogeneous stochastic differential equations (SDEs) on $\R^d$ driven by pure-jump Lévy processes, where the drift coefficient $b(t,x)$ satisfies $\lim_{t\to \infty}b(t,x) =0$ for every $x\in \R^d$. On account of three regimes associated with the index $\alpha$ of large jumps corresponding to the driven Lévy noise, we investigate the quantitative asymptotics of the corresponding rescaled processes. More precisely, for $\alpha\in (0,2)$, we prove that the rescaled processes, governed by time-inhomogeneous SDEs subject to additive processes, converge with respect to suitably chosen Wasserstein distances to time-homogeneous SDEs driven by symmetric $\alpha$-stable processes. Notably, the driven noise in the limiting SDEs depends only on large jumps of the underlying additive processes. In case of $\alpha\ge2$, a phase transition occurs and a diffusive phenomenon arises. In particular, in the setting $\alpha>2$, we establish the ergodicity of the rescaled process by means of asymptotic pseudotrajectories. The resulting time-homogeneous limiting SDEs are driven by Brownian motions, even though the transformed time-inhomogeneous SDEs are driven by (discontinuous) additive processes, and the effective noise intensity is determined by the entire Lévy measure of the original pure-jump process. As far as the critical case $\alpha=2 $ is concerned, we demonstrate that the noise intensity of the limiting SDEs driven by Brownian motions relies merely on the large-jump part of the Lévy measure.
Read the full abstract

Machine Learning

Learnware and AI Model Management System

The transition from file storage to database management systems transformed stored data into managed resources. AI now faces an analogous transition from AI model storage to AI model management. Existing model pools essentially serve as AI model storage systems. What is needed instead are AI model management systems that enable models trained by different developers, for different tasks, with different data, and under different objectives to be identified, reused, and even assembled to address future user tasks. Because AI model developers are generally unwilling to share their training data, such systems should operate without accessing the training data of model developers and, ideally, without accessing raw data of future users. This requirement poses a fundamental challenge: the functionality of a modern AI model may not be fully understood even by the developer who trained it. How, then, can a system identify which models are useful for a given user task, let alone assemble models developed independently for different purposes? At first glance, this objective may appear unattainable. It becomes possible, however, by upgrading the basic unit of management from a machine learning model to a learnware. Learnware = Model + Specification. The specification, whose assignment transforms a trained model into a learnware, is generated with the help of a machine learning process without disclosing the training data of the developer and has a theoretically established data-preservation property. The Learnware Dock System (LDS) provides a path toward powerful AI model management systems. Because specifications are generated according to a published reference and are comparable across models, they can also serve as an AI model collaboration protocol through which independently developed models, including intelligent agents, can collaborate.
Read the full abstract

Machine Learning

Musec: MomentUm SpEctral Clipping for Stable Muon-type Training

Muon has emerged as a highly effective optimizer for large language model training, often achieving superior convergence and performance compared with the widely adopted Adam and AdamW optimizers. Nevertheless, Muon is prone to training instability due to its spectral flattening, manifested by loss spikes and unbounded growth of model weights. Existing approaches primarily rely on weight or attention-logit clipping, which require architecture-specific modifications and do not directly address instability across all model components. We propose MomentUm SpEctral Clipping (Musec), which replaces Muon's spectral flattening with spectral clipping: rather than setting all singular values of the momentum matrix to approximately one, Musec clips singular values that exceed a threshold while preserving the underlying spectral structure of the momentum. Our strategy provides an optimizer-level, architecture-agnostic mechanism for stabilizing Muon training. Theoretically, we establish convergence guarantees for Musec in nonconvex nonsmooth stochastic optimization. Practically, we develop Soft Musec, an efficient implementation that uses a smooth spectral saturation function approximated by coupled Newton-Schulz iterations. Empirically, Soft Musec consistently improves training stability over existing Muon variants across a wide range of learning rates and model sizes, remaining stable in settings where existing Muon variants diverge while matching their performance under well-tuned configurations.
Read the full abstract

Astrophysics of Galaxies

The atomic multiphase interstellar medium of galaxies in the COLIBRE simulations

We analyse the neutral ISM of galaxies at redshift z=0 in the COLIBRE hydrodynamical simulations and investigate the conditions under which these two gas phases coexist in a narrow range of thermal pressures. COLIBRE galaxies are selected based on the metallicity of their ISM. The median and mass-weighted distributions of thermal pressures of the multiphase ISM are analysed and compared to thermal equilibrium models and observations. The ISM in galaxies with gas metallicities similar to solar values exhibits a clear multiphase structure with a warm and cold phase coexisting in a certain range of thermal pressures. The pressures at which the ISM is multiphase depend on the gas metallicity. For COLIBRE galaxies with lower metallicities ($Z_{\mathrm{ISM}}\lesssim0.1\,\mathrm{Z}_{\odot}$), this multiphase structure largely disappears, partly due to resolution. The thermal pressures weighted by the HI mass of the neutral phases in COLIBRE galaxies are lower than, but still comparable to, some theoretical works and observational estimates. The thermal pressures show a strong dependence on the weighting scheme. If weighted by the star formation rate or CI mass, the thermal pressures of the cold phase match those derived from observations. The resulting pressures depend on a combination of the assumed radiation field strength, dust abundance, limited resolution and the weighting scheme. The strong dependence of the thermal pressures on the weighting scheme indicates observational tracers used to estimate the thermal pressure are biased towards high-pressure regions.
Read the full abstract

All authors Moa Huppenkothen, Sylvia Ploeckinger, Joop Schaye, Alejandro Benítez-Llambay, Evgenii Chaikin, Filip Huško, Alexander J. Richings, James W. Trayford, Alvaro Hacar, Juan D. Soler

Differential Geometry

A Thorpe Trick for the Bochner Technique

We prove that for $n \geq 6$ there exists $\varepsilon (n)>0$ such that every closed orientable Riemannian manifold with $\left( \lceil \frac{n}{2} \rceil+ \varepsilon\right)$-positive curvature operator is a real homology sphere. For $n=6$ we show that the same result holds for manifolds with $4$-positive curvature operators.
Read the full abstract

Cosmology and Nongalactic Astrophysics

TDCOSMO XXXI: New techniques in line-of-sight studies of time delay lenses

The distribution of matter along the same line of sight but external to the main lens, quantified by $\kappa_\mathrm{ext}$, is a key source of uncertainty in time delay cosmography and other applications of strong lensing, and must be independently estimated to avoid biasing the inferred value of the Hubble constant ($H_0$). We present advancements and standardisations in the weighted number counts techniques used to constrain $\kappa_\mathrm{ext}$, in particular the use of the Euclid Flagship Simulation and a breakdown of the line-of-sight contributions between the observer, lens and source. As part of the TDCOSMO 2026 milestone analysis, we apply this updated method to the sample of 11 time delay lenses used in that study. Our estimates for certain systems are sensitive to these methodological changes, but are nonetheless consistent within $1\sigma$ for all but one of the systems which had been studied previously, with the median $\kappa_\mathrm{ext}$ across those systems changing from $-0.002$ to $-0.006$ following this new analysis. This work represents the first estimate of $\kappa_\mathrm{ext}$ values which includes the contribution of the observer-lens and lens-source terms, and largest standardised analysis of time delay lens environments to date.
Read the full abstract

All authors Daniel Johnson, Devon Williams, Chris Fassnacht, Patrick Wells, Pierre Fleury, Allan Schweinfurth, Martin Millon, Aymeric Galan

Systems and Control

Elementwise Positivity of the Solution to Lyapunov Equation for Hurwitz Companion Matrices

We prove, with the aid of AI, that for every real symmetric forcing matrix $Q\succeq0$, the unique solution of a continuous-time Lyapunov equation is entrywise nonnegative whenever the state matrix is a real Hurwitz companion matrix. This proves an earlier conjecture. The proof makes no assumption on the spectrum of the state matrix, and is made possible by means of using Horner polynomial matrices. We also show that discrete-time counterpart of the conjecture is false: Counterexamples can be readily constructed to show that Schur companion matrices, including one generated by a polynomial with all positive coefficients, yield solutions with negative off-diagonal entries even with identity forcing.
Read the full abstract

Computer Vision

Self-Supervised Cardiac Phase Detection via Single-Parameter Latent Orbits

Accurate identification of end-diastole (ED) and end-systole (ES) in echocardiography underpins the quantification of ventricular function, yet manual selection of these key frames is subjective and introduces clinically significant inter-operator variability. Recent self-supervised methods either prescribe strict periodic trajectories or learn an unconstrained low-dimensional motion subspace from reconstruction or registration objectives. The former offers interpretability but imposes restrictive assumptions on temporal progression, whereas the latter leaves cardiac phase implicit and ED/ES must be recovered through post-hoc geometric processing of the learned trajectory. We translate the physiological observation that cardiac phase is a one-dimensional signal into a prior by constraining the latent motion component to a single-parameter latent orbit, i.e., a global linear trajectory in latent space indexed by a bounded scalar phase variable. Mapping this variable through a sinusoidal nonlinearity yields an oscillatory motion signal with consistent temporal ordering, enabling direct identification of ED and ES from the learned phase signal. This inductive bias allows the model to capture an interpretable representation of the cardiac cycle, while maintaining flexibility to capture irregular heartbeats. Trained on EchoNet-Dynamic without annotations, our minimal single-parameter cardiac phase model learns an effective latent orbit, significantly improves upon the previous state of the art in ED localisation and matches it in ES localisation while using a more constrained representation and fewer training epochs. This demonstrates that a principled physiological inductive bias can match or exceed the performance of more complex representations. Code is available at: https://github.com/BonniciJ/OrbitalEcho/
Read the full abstract

All authors John Bonnici, Matthew Baugh, Aleksandra Kulbaka, Sarah Cechnicka, Bernhard Kainz, Alberto Gomez

Emerging Technologies

From Grid to Chip: Power Architecture, Stability, and Flexibility of AI Data Centers

The rapid growth of artificial intelligence (AI) computing is transforming data centers into large, dynamic electrical loads. Their deployment is primarily constrained by energy availability and grid-connection capacity, which is further aggravated by the ability of power-delivery architectures, control systems, and computing workloads to operate reliably during fast grid disturbances. This article provides a comprehensive overview of forefront challenges and trends in the grid-to-chip framework, and presents a technological perspective on AI data centers as grid-interactive computing systems. First, it reviews grid-integration bottlenecks, evolving connection policies, grid-code requirements, which has fostered new technological trends via spatio-temporal flexibility available through workload orchestration, cooling systems, on-site resources, and energy storage. Second, it maps the evolution of power-delivery architectures from medium-voltage grid interfaces to chip-level, discussing higher-voltage DC distribution, solid-state transformers, wide-bandgap devices, advanced chip-level power delivery, and liquid cooling. Third, it establishes a three-level stability framework spanning rack-level DC-bus dynamics, facility-level converter interactions, and system-level grid-coupled behavior. The framework connects dominant instability mechanisms, including constant power load effects, impedance interactions, forced oscillations, and operating-mode transitions, with suitable modeling, assessment, and mitigation approaches. Synthesizing these topics, this article highlights grid-to-chip co-design as a central requirement for scalable AI infrastructure, linking computing workloads, power-delivery systems, energy buffers, and grid operation.
Read the full abstract

All authors Yubo Song, Rui Kong, Takuro Umihara, Pooya Davari, Frede Blaabjerg, Subham Sahoo

Machine Learning

RDDMPI: Residual Denoising Diffusion Model for Probabilistic Multivariate Time Series Imputation

Multivariate time series imputation (MTSI) aims to recover missing values in temporal data composed of multiple interdependent variables. This problem is central to real-world applications such as healthcare monitoring, traffic networks, and energy systems. Recent diffusion-based approaches have shown strong potential for probabilistic imputation by learning to generate missing values through iterative denoising. However, most existing approaches perform diffusion directly in the original data space, requiring the denoising network to simultaneously capture global structure, temporal dynamics, and stochastic variability. This makes the generative task unnecessarily complex, especially when modern deterministic imputers can already provide accurate initial reconstructions. To address this limitation, we propose RDDMPI, a conditional residual diffusion framework that operates directly in residual space. Instead of modeling the full missing signal directly, we reformulate probabilistic imputation as a baseline-residual decomposition, where a pretrained model captures the dominant signal and a diffusion process models the residual uncertainty. To better exploit deterministic guidance, \model{} conditions the reverse denoising process on both the baseline-completed signal and its latent representation, while a reliability-aware conditioning mechanism adaptively controls the influence of baseline information during residual generation. This formulation simplifies the diffusion learning objective, enabling it to focus on structured correction terms rather than reconstructing the full signal. Experiments on multiple benchmark datasets demonstrate that RDDMPI consistently improves both reconstruction accuracy and uncertainty quantification.
Read the full abstract

Analysis of PDEs

Stable solutions in the abelian Higgs model

We classify entire stable solutions with quadratic energy growth in the $4$-dimensional abelian Higgs model. More precisely, we prove that such solutions are holomorphic with respect to some complex structure, namely they satisfy the first order equations introduced by Bradlow, and consequently their nodal sets are holomorphic curves. Conversely, we construct solutions whose nodal set is any prescribed holomorphic curve with quadratic area growth. As a corollary, we show that all stable solutions with linear energy growth in $\mathbb{R}^3$ are two-dimensional Bogomolnyi vortices.
Read the full abstract

Information Retrieval

Your Retriever Already Knows: Distribution-Shape QPP for RAG Retrieval Sufficiency

Standard Retrieval-Augmented Generation (RAG) pipelines often provide no reliable inference-time signal of whether retrieval succeeded; on ambiguous or out-of-scope queries, generation may then hallucinate. Motivated by a Czech nuclear-regulator deployment where data sensitivity precludes third-party LLM APIs, we compare three Query Performance Prediction (QPP) paradigms for retrieval sufficiency in RAG: score-based features, a content-based LLM judge, and a hybrid. On the eight ViDoRe vision domains (14,514 queries), our 24 non-lexical features (GeneralQPP; 15 distribution-shape, 5 query-surface, 4 global) reach a weighted-average AUROC of 0.856 at 2 ms per query, ahead of a classic-QPP literature pool (Classic Full, 0.835) and well above a local Qwen3.5 LLM judge (0.649, +0.207 gap; $\sim$3000$\times$ faster and cheaper per query). Adding the LLM judgment as one feature (hybrid) matches S1 on ViDoRe (0.863) but gains a statistically significant edge on SÚJB (AUROC 0.911 at Hit@5, adversarial-detection 0.954; 1,510 queries, 500 synthetic adversarial), at LLM latency. Rankings agree across datasets (Spearman $\rho = 0.90$). Under Leave-One-Domain-Out, S1 drops to 0.706; a 13-feature LODO-stepwise subset (S1-Lean) recovers to 0.719 (+0.032 over the literature pool).
Read the full abstract

Probability

Persistence and long-time breakdown of most probable paths under time-dependent fractional noise with applications to KAM tori

We investigate the persistence of most probable paths through the Onsager–Machlup functional for multidimensional stochastic differential equations driven by fractional Brownian motion with time-dependent diffusion coefficients and Hurst parameter $H\in(1/4,1)$. Under suitable structural and variational conditions, deterministic trajectories remain most probable paths for sufficiently small noise in both the fixed-endpoint transition problem and the free-endpoint evolution problem, whereas sufficiently large noise destroys their local minimality. More generally, when exact persistence does not hold, global most probable paths converge to the corresponding trajectories of the noise-free system in both the uniform and Hölder topologies at the rate $O(\epsilon)$. We further analyze the second variation along periodic deterministic trajectories over time intervals of length $NT$. For $H>1/2$, positive definiteness, and hence local minimality, is lost on sufficiently long intervals. For $H\in(1/4,1/2]$, long-time positive definiteness holds for the fixed-endpoint problem, but this conclusion does not directly extend to the free-endpoint setting. We also establish the persistence of KAM tori in nearly integrable Hamiltonian systems in the sense of most probable evolution paths. Finally, a two-dimensional numerical example illustrates the persistence of deterministic trajectories under small noise and their pronounced deviation under large noise.
Read the full abstract

Analysis of PDEs

Gaussians Do Not Always Maximize Mixed-Norm Strichartz Inequalities for the Schrödinger Equation

We investigate the maximization problem for the family of mixed-norm Strichartz inequalities for the Schrödinger equation, $\|e^{-it\Delta/2}f\|_{L_t^qL_{\boldsymbol{x}}^r(\mathbb{R}^{1+d})}\le C_{q,r}\lVert f\rVert_{L^2(\mathbb{R}^d)}$, with $2/q+d/r=d/2$, $q,r\geq 2$, and thus $r\leq 2d/(d-2)$ if $d\geq 3$. We show that, in low dimensions $1\leq d\leq 5$, the thresholds $\rho_1=10$, $\rho_2=6$, $\rho_3=4\sqrt{7}-6\approx 4.583$, $\rho_4=2\sqrt{15}-4\approx 3.746$, and $\rho_5=10/3\approx 3.333$ are such that gaussians are stable local maximizers for $2<r<\rho_d$, and fail to be local maximizers for $\rho_d<r\leq 2d/(d-2)$ (with the conventions there is no upper bound on $r$ when $d\in\{1,2\}$ and that $r=\infty$ is excluded when $d=2$). In the cases $(q,r,d)\in\{(6,6,1),(8,4,1),(4,4,2)\}$, we establish global stability inequalities with effective stability constants. Both proofs hinge on spectral gaps which we compute exactly.
Read the full abstract

High Energy Physics (Theory)

Effective Dynamics of Inflationary End-of-the-World Branes in AdS$_3$

We develop an effective description of a two-dimensional cosmological end-of-the-world brane embedded in AdS$_3$, with a scalar field localized on the brane. Integrating out bulk degrees of freedom, especially in the small-gradient regime, the effective action reduces to a Liouville-like theory coupled to the brane scalar. We then construct brane trajectories whose induced geometry realizes slow-roll inflation and reconstruct the associated scalar profile and potential. We also study regular Euclidean brane geometries that admit a smooth continuation to Lorentzian de Sitter and inflationary universes, and evaluate the semiclassical on-shell action for the de Sitter solution. Finally, we analyze linearized inhomogeneous perturbations and find no exponentially growing mode within the regime of validity of the approximation.
Read the full abstract

All authors Kosei Fujiki, Michitaka Kohara, Jason Kristiano, Yu-ki Suzuki, Tadashi Takayanagi

Sound

ZipCodec: Ultra-Low-Frame-Rate Streaming Speech Coding

Neural audio codecs are a fundamental component of modern speech generation systems. While recent codecs achieve increasingly low bitrates, reducing frame rate remains challenging, as each token must preserve more information while maintaining reconstruction quality. We present ZipCodec, a streaming neural speech codec operating at 6.25 Hz and 0.80 kbps with a theoretical latency of 160 ms. Our approach combines large-scale WavLM distillation with a redesigned transformer-based architecture, a scalar spherical quantizer, and a latency-aware streaming decoder. Experiments show that ZipCodec substantially outperforms existing streaming codecs at comparable bitrates in both reconstruction and downstream tasks, while operating at a significantly lower frame rate. Despite its 842M parameters, ZipCodec achieves real-time single-stream inference on a consumer-grade CPU. Demo samples, code and checkpoints are available at https://lucadellalib.github.io/zipcodec-web/.
Read the full abstract

High Energy Physics (Phenomenology)

Distinguishing a pseudoscalar from a vector $t\bar t$ resonance with top-quark spin correlations at the HL-LHC

Top-quark spin correlations provide a direct probe of the quantum numbers of a resonance decaying to $t\bar t$. We compare a CP-odd type-II two-Higgs-doublet-model pseudoscalar $A$ with a spin-1 leptophobic topcolor $Z'$ in the dilepton channel. The two hypotheses differ in the parton-level helicity correlation by $\Delta c_{\rm hel}\simeq0.43$, nearly independently of mass, and neutrino-weighted reconstruction preserves this separation with an effective dilution factor of about 0.66. At 400 and 800 GeV, the pseudoscalar normalization is anchored to CMS HIG-22-013 coupling limits, while the $Z'$ rate is matched to the pseudoscalar so that the comparison primarily tests spin structure. The leading-order $t\bar t$-only projection gives expected separations of 2.27 and 2.74 standard deviations at $3000\,{\rm fb}^{-1}$. Equalizing the selected signal yields gives pure spin-shape separations of 1.86 and 2.70 standard deviations. Profiling a Gaussian-constrained Standard Model $c_{\rm hel}$-shape nuisance gives 2.07 and 2.63 standard deviations for an optimistic sideband benchmark, or 1.67 and 2.26 for a weaker constraint. A local fixed-template extrapolation corresponds to five-standard-deviation discrimination at couplings about 22% and 16% above the reference values. A 1500 GeV $g=1$ point is shown separately. The reach is limited by the sub-percent signal fraction and control of the mass-dependent Standard Model spin shape, rather than by the intrinsic resonance-spin separation.
Read the full abstract

Materials Science

High-throughput thermodynamic screening of oxide-scale adhesion across the CoCrFeMnNiAl high-entropy alloys

One significant benefit of reactive element (RE) additions is the colossal improvement in oxide-scale retention during high-temperature oxidation. Selecting optimal RE dopants in high-entropy alloys remains empirical because the relevant thermodynamic landscape is inaccessible to first-principles at the required compositional resolution. Here we apply the macroscopic atom model, coupled with McLean isotherm and Guttmann models, to screen adhesion across nine CoCrFeMnNiAl sub-families at \ce{Cr2O3} and \ce{Al2O3} interfaces, ranking five REs (Hf, Y, Zr, La, Ti) for segregation, adhesion enhancement, and sulfur displacement. The screening reveals an oxide-dependent ranking inversion, with Hf dominating at \ce{Cr2O3} and La dominating at \ce{Al2O3}, driven by the interplay between RE–O and RE–matrix interaction enthalpies. Mn-containing alloys exhibit intrinsic sulfur resistance, consistent with their experimentally observed oxidation characteristics. A sulfur immunity phase diagram identifies compositions with Al~+~Mn~$\gtrsim$~25~at% as thermodynamically immune to S-induced adhesion loss. All crossover concentrations collapse onto a universal exponential governed by the segregation enthalpy difference, providing a transferable design rule. Inverse design identifies \ce{Co16Cr16Fe16Ni16Al35} as the optimal S-immune composition with $W_\text{sep} = 5.95$~J/m$^2$ without RE doping.
Read the full abstract

Machine Learning

LoaDiff: Conditional Generation of Electricity Consumption Time Series for Energy Analytics

The energy transition is reshaping residential electricity consumption through the increasing adoption of distributed generation, electrified appliances, and demand-response programs. Understanding these evolving behaviors requires access to granular smart-meter data for applications such as load forecasting, appliance detection, and demand-side flexibility analysis. However, such data are subject to strict access restrictions and data-protection regulations. Thus, realistic synthetic alternatives are necessary. In this paper, we introduce LoaDiff, a diffusion-based generative model for year-long, sub-hourly smart-meter load curves. LoaDiff supports flexible conditioning on static household attributes, such as appliance ownership, and dynamic contextual variables, including calendar information and outdoor temperature. We evaluate the model against multiple generative baselines on three residential electricity-consumption datasets. Our experiments assess four complementary dimensions: fidelity and diversity, training-record memorization risk, downstream utility for load forecasting and appliance detection, and conditional controllability under alternative temperature conditions. The results show that LoaDiff generates realistic and diverse load profiles, achieves a favorable trade-off between generation quality and limited evidence of memorization, preserves information useful for downstream energy applications, and responds coherently to changes in conditioning variables.
Read the full abstract

All authors Mariia Baranova, Adrien Petralia, Etienne Le Naour, Nathan Etourneau, Guillaume Hofmann, Themis Palpanas

Computer Vision

Vidu S2: Real-Time Interactive, Editable, and Spatial Video Generation

We present Vidu S2, which comprises Vidu S2-Avatar, a real-time interactive digital-character model, and Vidu S2-Editing, a real-time video editing model. Moreover, we explore the feasibility of real-time spatial video generation for both Vidu S2-Avatar and Vidu S2-Editing. Compared with Vidu S1, Vidu S2-Avatar supports real-time 720p video generation, generation with dynamic references that can be updated at any moment, and stronger instruction following, such as dancing. Vidu S2-Editing supports editing a video stream in real time, including style rendering, clothing replacement, character replacement, and background replacement. Experiments show that Vidu S2 outperforms all baselines. A playable online demo is available at https://vidu.com/vidu-stream.
Read the full abstract

All authors Jintao Zhang, Kai Jiang, Jintao Chen, Xu Wang, Deyuan Liu, Jungang Li, Dechuang Chen, Ming Lin, Jingjiang Zhou, Haopeng Jin, Qi Jia, Xiaohang Wang, Yaole Wang, Zhanqiang Zhang, Ran Li, Zhengkun Huang, Shuyue Xiong, Yuji Wang, Zikun Dai, Hui He, Yang Luo, Mang Ning, Weiqi Feng, Chengyang Ye, Xinyue Lin, Min Zhao, Hongzhou Zhu, Hengkai Tan, Zeyuan Wang, Chendong Xiang, Kaiwen Zheng, Zhijie Deng, Fan Bao, Jianfei Chen, Jun Zhu

Quantum Physics

Certifying Adversarial Robustness of Quantum Classifiers under Known-Readout Query Access

A quantum classifier assigns labels by evolving an input quantum state and measuring the output, so repeated executions reveal only a distribution over labels. We study certified adversarial robustness for such classifiers under known-readout query access (KRQA), where an evaluator can prepare inputs, knows the quantum measurement, and observes finite-shot outcomes but cannot inspect the internal evolution, parameters, or gradients. We give a measurement-only framework that returns two complementary guarantees for each input: a lower bound ruling out untargeted errors within a radius, and an attack-independent upper bound witnessing an adversarial state within a radius. Both are estimable from the known readout measurement and sampled outcomes, require no tomography or circuit description, and have finite-sample control of probability-estimation error. The upper bound uses gap operators induced by the quantum measurement; the lower bound relaxes state-space search to an efficient optimization over outcome distributions with operator-spectrum constraints, yielding certificates that are never weaker than prior probability-only certificates and can be strictly stronger when the spectral constraints are active. On tractable instances, we compare the lower bound with numerical white-box reference estimates; across multiple classifiers, the upper bound remains informative when standard attacks fail. We further demonstrate real-device feasibility on IBM Quantum hardware: from 40 executions of two 8-qubit quantum neural networks, our method estimates both bounds, with the expected lower-upper ordering on every tested input. Taken together, these results show that robustness claims for quantum classifiers can be audited directly from observable statistics under KRQA.
Read the full abstract

Artificial Intelligence

MAPLE: Memory-Augmented Planning with Language and Evolution

Domain practitioners understand their business constraints but may lack operations-research expertise or dedicated support. LLM-based optimization agents translate natural-language requirements into models or solver programs that established optimization tools can execute. This progress makes optimization more accessible, but real-world operations are dynamic: changing demand, resources, and priorities require updates to data, constraints, and objectives. Methods centered on isolated requests offer limited support for rapid adaptation that preserves earlier decisions and reuses useful search results. We introduce MAPLE (Memory-Augmented Planning with Language and Evolution), an agent for maintaining optimization problems through successive natural-language requests. MAPLE combines language-based problem construction with mathematical programming and evolutionary search. It retains the optimization program, accepted plans, earlier updates, and candidate solutions for subsequent requests. We introduce NLDO, a benchmark of 15 trajectories and 180 updates spanning selection, scheduling, rostering, routing, and cloud-resource placement. In the main evaluation, MAPLE completes all trajectories and achieves online scalar quality of 0.951 and a Pareto hypervolume ratio of 0.875. Controlled comparisons further show that maintaining executable state improves update validity and can preserve useful search information across substantial revisions.
Read the full abstract

Quantum Physics

Quantum-hardware spectral co-design framework for multifrequency Rydberg electrometry

Engineering electromagnetic hardware to satisfy discrete quantum-defined spectral constraints constitutes a general spectral co-design problem for quantum systems. Here we address this challenge in multifrequency Rydberg electrometry by directly coupling a fabrication-constrained simultaneous perturbation stochastic approximation (SPSA)–Adam optimizer to full-wave finite-element eigenmode simulations. Requiring neither analytical nor adjoint gradients, the method operates over a discrete design space containing approximately $10^{300}$-configurations and yields a novel multimode electrometry architecture that simultaneously aligns four high-$Q$ eigenmodes with four selected Cs Rydberg transitions. The optimized design remains highly robust to fabrication imperfections, achieving a relative frequency error as low as $8.42\times10^{-7}$ while reducing the device length by a factor of $1.75\times10^{2}$, thereby overcoming the difficulty of simultaneous multimode spectral matching encountered in conventional design. For a comparable simulation budget, the proposed co-design framework achieves frequency-matching errors approximately 10 and 41 times smaller than those of covariance matrix adaptation evolution strategy and discrete simulated annealing, respectively. The electrometry is predicted to provide an average input-power-sensitivity enhancement of approximately $4.04\times10^{3}$, demonstrating quantum–hardware spectral co-design as a general route toward compact hardware for multichannel quantum sensing.
Read the full abstract

All authors Song Huang, Guibin Tian, Fernando Quijandría, Zhengjiang Li, Zhimin He, Franco Nori, Yong Lu

Atomic Physics

Impact of Raman lasers beam profile inhomogeneities on a spaceborne quantum accelerometer

We report on the study of the impact of laser beam profile inhomogeneities on the acceleration measurements of a quantum accelerometer in space. This sensor uses a low ballistic expansion source based on delta-kick collimated Bose-Einstein condensates, and an interferometer configuration based on a sequence of Raman double diffraction beamsplitters, best suited to the microgravity environment. We characterize both phase and intensity fluctuations in terms of power spectral densities and calculate their impact on the interferometer phase, averaging over all trajectories of the atoms, via numerical simulations based on randomly drawn laser profiles as well as via analytical treatments. We show that high quality optics, such as those realized for gravitational wave detectors, combined with large initial size atomic sources, will allow to reduce interferometer phase bias and fluctuations to the order of 1 mrad corresponding to accelerations in the low 10 -12 m.s -2 range.
Read the full abstract

All authors Benoit Kaczmarczuk, Gabriel Ducasse, Louis Pagot, Quentin Beaufils, Franck Pereira dos Santos

Mesoscale and Nanoscale Physics

Phase-Controlled Majorana Zero Modes in Altermagnetic Topological-Insulator Josephson Junctions

We exploit facet-dependent Andreev phase shifts to control topological superconductivity with a phase bias in a three-dimensional altermagnetic topological-insulator Josephson junction. In the weak link between two conventional $s$-wave superconductors, the $d$-wave altermagnetic order produces anisotropic momentum shifts of the surface Dirac cones. The resulting net momentum of the states involved in Andreev reflection generates additional propagation phases that differ between facets. Consequently, the facet-resolved Andreev spectra exhibit gap closings at distinct phase biases, giving rise to topological superconducting regimes that host Majorana zero modes (MZMs). We further show that the spatial locations of the MZMs can be controlled by the phase bias. Moreover, these topological superconducting transitions are only weakly affected by moderate variations in the chemical potential, obviating the need for fine-tuning to the Dirac point. Our results establish a platform for realizing and spatially controlling MZMs by tuning the superconducting phase bias in altermagnetic topological-insulator Josephson junctions.
Read the full abstract

All authors Hao Dong, Xun-Jiang Luo, Xiao-Hong Pan, Xin Liu

Information Retrieval

FedHUR: Learning Hierarchical Utility-Guided Client Relations for Personalized Federated Recommendation

Federated recommendation enables collaborative model training while keeping user interaction data on local clients. A central problem in federated recommendation is how to aggregate useful information across clients for personalized recommendation. Existing personalized aggregation methods usually construct client relations from predefined parameter-based assumptions, such as parameter similarity or complementarity, and use these relations to determine aggregation weights. However, such methods construct a single global relation, which is insufficient to capture the hierarchical and multi-granularity nature of user relations in recommendation. Moreover, these predefined relations cannot directly reflect whether the related clients can improve prediction performance after aggregation. To address these limitations, we propose FedHUR, a federated recommendation framework for learning hierarchical utility-guided client relations. FedHUR takes item-item filters as the object for relation construction and aggregation. Specifically, it first aggregates and clusters each client's local information to obtain global hierarchical information. Each client computes hierarchical utility signals based on its local information and the global hierarchical information, indicating which collaborative information is useful for improving its prediction. The server uses these utility signals to retrieve clients that are useful to that client for further personalized aggregation. Extensive experiments on five real-world datasets show that FedHUR consistently outperforms existing federated recommendation baselines, demonstrating the effectiveness of hierarchical utility-guided client relation learning. Code is available at https://github.com/Mingzhe-Han/FedHUR.
Read the full abstract

All authors Mingzhe Han, Jiahao Liu, Dongsheng Li, Jiankui Zhou, Hansu Gu, Peng Zhang, Ning Gu, Tun Lu

Fluid Dynamics

Surface gravity wave-mean flow interaction with comparable spatial scales. Part II: two-way coupling and wave-wave interactions

We consider narrow-band deep-water surface gravity waves propagating above a background flow whose spatial scale is comparable to the wavelength. We focus on the regime where the background-flow speed is comparable to the Stokes drift of the waves to derive a two-way coupled model between the fast waves and the slow background flow. Nonlinear wave interactions arise at the same asymptotic order as the two-way coupling and must therefore be included in the derivation. The resulting model consists of a nonlinear version of the reduced wave equation obtained in part I, coupled to a Craik-Leibovich equation governing the evolution of the background flow. The model describes the evolution of the wave field and background flow over the slow turnover frequency of the background flow, thus saving the computational burden of temporally resolving the fast wave frequency. The inviscid model exactly conserves wave action, mechanical energy and horizontal momentum, while the viscous model exactly conserves horizontal momentum. The coupling terms are cast in compact form for ease of physical interpretation and numerical implementation.
Read the full abstract

Probability

Precise Delocalisation and Gumbel Laws for Eigenvectors of Wigner Matrices

We prove a delocalisation bound for eigenvectors of Wigner matrices with the precise relationship between the size of the largest entry and the decay exponent of the probability. We also prove that the largest entry of an individual eigenvector and the largest entry of all eigenvectors are both Gumbel distributed. The proof is based on the inclusion-exclusion principle, the small probability comparison of Erdős–Xu, and partial diagonalisation of Gaussian divisible matrices.
Read the full abstract