Peer Revenue Equity in science

Papers

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

All of arXiv

Thursday 27 August

Quantum Physics

Optimal cloning of mixed states

We consider the problem of approximate cloning of quantum states: given $n$ copies of an unknown state $\rho \in \mathbb{C}^{d \times d}$, prepare an $(n+k)$-copy state with high fidelity to $\rho^{\otimes (n+k)}$. Werner's pure state cloner is the optimal channel for the pure state case, and shows that $n = \Theta(kd/\varepsilon)$ copies are necessary and sufficient to clone $k$ additional copies of an unknown pure state to fidelity $1-\varepsilon$. The random purification channel gives a straightforward extension of Werner's cloner to mixed state inputs: given $n$ copies of a mixed state, randomly purify your input, apply Werner's channel in the larger Hilbert space, and then trace out the auxiliary registers. This gives a mixed state cloner using $n = O(krd/\varepsilon)$ copies to clone rank-$r$ states. Can one do any better? We show that the answer is no: one must use $n = \Omega(krd/\varepsilon)$ copies. We prove our lower bound by studying the special case of projector cloning, in which the input state $\rho$ is promised to be of the form $P/r$, where $P$ is a rank-$r$ orthogonal projector. As a further application of our techniques, we consider the closely related problem of approximate transposition of quantum states, where one seeks to convert $\rho^{\otimes n}$ to a $k$-copy state with high fidelity to $(\rho^T)^{\otimes k}$. Here, we again show $n = \Theta(krd/\varepsilon)$ copies are necessary and sufficient for this task.
Read the full abstract

All authors Marco Fanizza, Dmitry Grinko, Thilo Scharnhorst, Jack Spilecki

General Relativity and Quantum Cosmology

Emergent vacua and stability constraints on black hole solutions in higher-dimensional $f(R)$ gravity

We investigate static spherically symmetric vacuum solutions in higher-dimensional $f(R)$ gravity, beginning with the five-dimensional Starobinsky model governed by the action $f(R) = R + \alpha R^2 - 2\Lambda$. By enforcing the ghost-free stability criterion $f'(R) > 0$ on constant scalar curvature spacetimes, we show that a stable effective cosmological constant cannot be dynamically generated from pure $R^2$ geometric corrections in five dimensions; its existence is inextricably tied to a bare cosmological constant. Generalizing this analysis to arbitrary dimensions $D$ and single-term curvature corrections $f(R) = R + \alpha R^n$ with a vanishing bare cosmological constant, we derive a universal stability bound, $n > D/2$, required for the existence of stable emergent vacua. Finally, we demonstrate that expanding the gravitational action to a multi-term polynomial hierarchy circumvents this strict limitation. By including curvature corrections up to $\mathcal{O}(R^3)$, the extended geometric degrees of freedom simultaneously satisfy the trace constraint and the stability criterion. Furthermore, we establish the exact parameter space boundaries that ensure not only asymptotic vacuum stability but strict global stability ($f'(R) > 0$ for all $R$), allowing for the dynamical generation of exact, globally ghost-free vacuum spacetimes in $D \ge 5$ purely from higher-order geometric terms. The scalaron mass requirement, $m_s^2>0$ is not imposed in full in our analysis. A detailed investigation of its implications on the models considered in this work are left for future study.
Read the full abstract

Artificial Intelligence

LLMs Can Design Near-Optimal OR Algorithms

We ask whether large language models (LLMs) can design effective algorithms for well-specified operations research (OR) problems. We study inventory control, queueing network control, and assortment optimization, using benchmark instances from recent papers in each area. We evaluate two levels of LLM use: at level 1, the model receives one problem instance and returns a solution for that instance; at level 2, it receives only the problem class description and broad parameter ranges, and returns an algorithm that maps instance parameters to solutions. Human input is minimal: we give one untuned prompt that describes the problem, and the model has access to a Python sandbox tool with a fixed compute budget. The strongest model we test, gpt-5.6-sol, matches or outperforms the best method reported in those papers on almost all evaluated instances. This holds even at level 2, where the returned algorithm is fixed before seeing the evaluation instances. Performance also improves sharply across models released less than eight months apart, suggesting that this capability is moving quickly. Thus, for the well-specified operations problems we study, a single untuned LLM query can already produce algorithms competitive with specialized methods. These results suggest that frontier LLMs can be a serious empirical baseline for algorithm design in well-specified OR problems.
Read the full abstract

Computational Finance

A Temporal Multiplex Graph Neural Network for Systemic Risk Transmission in Global Banking

This paper develops a unified framework for assessing systemic risk and identifying contagion channels in the global banking system using a Temporal Heterogeneous Multiplex Graph Neural Network. We construct a harmonised quarterly panel combining bank fundamentals, CDS spreads, and macroeconomic indicators, and represent these data as dynamic multiplex networks linking banks through financial similarity and liquidity co-movement, augmented with country-level macroeconomic relationships. The model integrates graph convolutional layers with recurrent GRU dynamics and incorporates a learnable fusion gate to capture time-varying reliance on alternative contagion channels. Empirical results show that the framework outperforms conventional econometric, machine learning, and graph-based benchmarks for short-term changes in CDS spreads. Beyond forecasting, we provide an interpretable framework to quantify bank-level systemic importance via stress testing, assess country-level spillovers under macroeconomic shocks, and uncover transmission pathways through edge perturbation analysis. Robustness tests confirm the stability of both predictive accuracy and systemic risk rankings.
Read the full abstract

Astrophysics of Galaxies

Exploring the survivability of higher-order multiple protostellar systems – The case of VLA1623

Higher-order protostellar systems ($\geq$3 components) are commonly observed in the early stages of low-mass star formation. A persistent question in star formation and evolution is whether these higher-order protostellar systems survive as gravitationally bound systems or dissolve into binaries over time. We explore this question with a case study of the embedded quadruple protostellar system VLA1623 and its observational constraints, assuming that the components A1, A2, B, and W are gravitationally bound. Using $N$-body simulations, we run a grid of models considering gravity, mass accretion onto protostars, the presence of the protostellar cloud core, and its dispersal. The simulations are integrated over a period of 8 Myr to take into account the evolution from the protostellar phase (Class 0 and I, 1 Myr), through the pre-main sequence (Class II and III, 2 – 3 Myr), and into the main sequence (4 Myr). Our results show that VLA1623 has a probability of 30% to remain as a gravitationally-bound quadruple system up to 8 Myr from the current state. There is also a 25–30% probability of VLA1623 dissolving into a triple system. This suggests that the dissolution of quadruple protostellar systems contributes to the formation of stable triple (proto)stellar systems. Components A1 and A2 are the most likely to be ejected, while W has a lower probability of being ejected from the system. The stability of VLA1623 depends on a combination of component mass ratios wrt the primary, separations, and eccentricities.
Read the full abstract

General Relativity and Quantum Cosmology

Electric and magnetic Penrose processes, charged-particle collisions and superradiance around a Lorentz-violating dyonic black hole

This paper is aimed at investigating electromagnetic energy extraction, charged-particle collisions, charged-field superradiance, and horizon stability in a Lorentz-violating dyonic Kalb-Ramond black hole. The geometry differs from the dyonic Reissner-Nordström solution through a modified asymptotic normalization and an effective charge combining electric and magnetic sectors. We derive the dynamics of electrically and magnetically charged probes, including monopole-induced conical motion, and formulate electric and magnetic Penrose processes as decay or collision mechanisms in which a negative-energy fragment falls into the horizon while another escapes with enhanced energy. The electric and magnetic channels are controlled by \(\Phi_H=Q/[(1-\ell)r_+]\) and \(\Psi_H=p/[(1-2\ell)r_+]\), respectively. Negative-energy states arise from electromagnetic canonical energy rather than from a geometrical ergoregion. We also obtain escape conditions for the outgoing products, test the possibility of overcharging or overmagnetizing the black hole, and analyze the center-of-mass energy of charged-particle collisions. Nonextremal same-direction collisions remain finite, whereas head-on or near-critical configurations can become large. In the extremal limit, electrically or magnetically critical particles may generate the Bañados-Silk-West divergence when the radial reachability condition is fulfilled.
Read the full abstract

Group Theory

Kleisli convolution representations of power monoids

We show that power semigroups of groups, and more generally reduced finitary power monoids, arise naturally as convolution monoids in Kleisli categories of powerset monads: (1) for the non-empty powerset monad, the Kleisli Hom-space $\mathrm{Hom}_{\mathbf{Kl}(\mathscr P_+)}(1,G)$ is isomorphic to the power monoid $\mathcal P_+(G)$; (2) for the reduced finite powerset monad on pointed sets, the Kleisli Hom-space $\mathrm{Hom}_{\mathbf{Kl}(\mathscr P_{\mathrm{fin}})}$ $(\mathbb Z/2\mathbb Z,H)$ is isomorphic to the reduced finitary power monoid $\mathcal P_{\mathrm{fin},1}(H)$. This unifies several constructions in power semigroup theory: Kleisli convolution representations of semigroups, base change along surjective group homomorphisms, and rigidity of automorphism groups. As an application, we prove that for every proper numerical monoid $S$, the Kleisli Hom-monoid $\mathrm{Hom}_{\mathbf{Kl}(\mathscr P_{\mathrm{fin}})}(\mathbb Z/2\mathbb Z,S)$ is rigid. This gives a proof of the Tringali–Yan conjecture in the language of Kleisli categories. It should be mentioned that this conjecture was already proved by Bhowmik and Tringali in a preprint posted on arXiv on July 25, 2026.
Read the full abstract

Machine Learning (Statistics)

Recovering Expert Critic-Sourced Network Adjacency between Musical Artists from Acoustic Distributions: A Construct-Validity Approach

Music recommendation relies primarily on two signals: user-item interactions, which fail in the cold-start regime, and intrinsic musical content, available for any recording. We argue that a third, largely untapped signal is both richer and more principled: critical adjacency, the pairwise relation established when an expert critic explicitly links two artists in long-form prose. It encodes deliberate judgments about which artists belong together. Prior work established its internal validity, showing it recovers coherent, interpretable communities and can match collaborative filtering in user-satisfaction simulations, with no user data. What has been missing is external validation: whether this critic-sourced relation is grounded in the music itself versus sociological context. We test it against acoustic content, reframing the question as one of construct validity. Representing artists as empirical distributions over 80 low-level Essentia acoustic descriptors and modeling pairwise proximity via marginal optimal-transport (Wasserstein) distances, we evaluate how far critical adjacency is sonically recoverable under a cold-start, artist-disjoint split. Our ensemble recovers these edges at out-of-sample AUC of 0.767 (95% CI 0.761-0.775). Recoverability rises monotonically with critical consensus, reaching 0.865 on multi-source attested edges. Stratified evaluations align with sociological models of genre: tightly bounded, scene-based genres show higher recoverability than broad industry umbrella terms. Critical discourse is thus a rich source of information for recommendation, decomposing into a reproducible "sonic core" and a "sociological remainder" driven by narrative positioning, subcultural context, and canonical placement. The work offers both a scalable cold-start discovery mechanism and a sociologically grounded approach to MIR and MRS research.
Read the full abstract

Computer Vision

Detection of Christmas tree plantations from high-resolution aerial imagery. A case study in the French Morvan

Christmas tree plantations are economically relevant, yet a largely unexplored application domain in Remote Sensing (RS). Their delineation is challenging because of high planting density, short rotation cycles, visual confusion with surrounding vegetation, the availability of dense labels for one reference year only, and severe class imbalance at the landscape scale. Although Deep Learning (DL) methods have shown strong potential for vegetation mapping, existing approaches are typically designed for forests, generic plantation systems, or orchards, and do not explicitly address the structural specificity and hard-negative confusion that characterize Christmas tree plantations. In response to these challenges, this work makes three main contributions: (i) it frames Christmas tree plantation mapping as a distinct rare-target semantic segmentation problem; (ii) it introduces a Hard Negative Mining (HNM) strategy to improve discrimination against confusing background patterns; and (iii) it evaluates the proposed framework across complementary levels, including supervised testing, temporal transfer, and large-scale validation. On the 2020 test set held out, the best model, DeepLabV3 with a ResNet-34 encoder, achieves an IoU of 0.733 and an F1-score of 0.846. HNM substantially improves precision-recall behavior, increasing the area under the precision-recall curve from 0.204 to 0.913. Temporal inference further shows meaningful transferability, reaching IoU/F1 values of 0.751/0.858 on 2017/2018 and 0.691/0.817 on 2023. Large-scale validation further highlights the intrinsic difficulty of the task, as Christmas tree plantations occupied only a very small fraction of the extent of the common evaluation, corresponding to 1,498.4 ha (1.72%) in 2017/2018 and 1,782.2 ha (2.04%) in 2023 out of 87,309.4 ha in total.
Read the full abstract

All authors Francesca Razzano, Emanuele Dalsasso, Adrien Baysse-Lainé, Silvia Liberata Ullo, Gilda Schirinzi, Jocelyn Chanussot

Statistical Methodology

Combining covariate adjustment with information from secondary endpoints to improve precision in randomized trials

Background/Aims: Adjustment for prognostic baseline covariates can improve precision in randomized trials. Previous work has shown that jointly modeling primary and secondary endpoints can yield additional precision by borrowing information across endpoints. We investigated whether these approaches can be combined to achieve efficiency gains beyond those obtained through covariate adjustment alone. Methods: We extended a previously proposed one-factor structural equation modeling framework for borrowing information from secondary endpoints to incorporate baseline covariates while retaining the average treatment effect on the primary endpoint as the estimand. To mitigate sensitivity to model misspecification, we combined this estimator with a conventional covariate-adjusted estimator using cross-validated model averaging. We evaluated operating characteristics in simulations and applied the methods to a randomized trial of very low versus normal nicotine content cigarettes. Results: When the structural equation model was correctly specified, endpoint borrowing improved efficiency beyond conventional covariate adjustment across all simulated settings. Model misspecification could induce bias and undercoverage. Model averaging reduced bias and improved coverage relative to the structural equation model estimator, although coverage remained imperfect under severe misspecification. In the trial application, the model-averaged estimate was 21% more precise than the unadjusted estimate and 13% more precise than covariate adjustment alone. Conclusion: Secondary endpoints can contribute meaningful information about the average treatment effect on a primary endpoint even after baseline covariates have been incorporated. These gains require stronger assumptions than conventional covariate adjustment; model averaging provides a practical compromise between efficiency and robustness.
Read the full abstract

Cryptography and Security

Low-ASR Backdoors: Exploiting Attack Success Rate Reduction and Attacker-Defender Asymmetry

Backdoor attacks are among the most effective and stealthy attacks in deep learning. Existing attacks and defenses are largely designed and evaluated under the assumption that successful backdoors exhibit high Attack Success Rates (ASRs). In this paper, we show that this assumption creates a fundamental weakness in existing defense paradigms. ASR is not an intrinsic property of a backdoor; rather, it is an attacker-controlled variable that can be deliberately reduced without eliminating the underlying backdoor behavior. We introduce a reverse-training framework that weakens the trigger-target association, producing low-ASR backdoor models while preserving clean-input performance. Through extensive evaluation across multiple datasets, diverse attack families, and multiple architectures, we show that state-of-the-art defenses fail consistently under low-ASR conditions, exposing a fundamental attacker-defender asymmetry.
Read the full abstract

Information Retrieval

Astar: Learning to Propose Evolution Directions for Self-Evolving Industrial AI Systems

Modern AI systems advance through continuous iteration: a loop of proposing evolution directions, implementing code, training, and evaluation. While the latter three stages are increasingly automated, the starting point — proposing effective evolution directions — remains a critical bottleneck that still relies heavily on senior experts. In this work, we explore whether AI can take over this role. We find that general-purpose LLMs, even the advanced GPT-5.5, offer only generic and misaligned suggestions: the required expertise is accumulated through experience rather than explicitly codified, and thus hard to inject directly. To this end, we propose Astar, a training-based approach that learns a specialized evolution-guiding model from the abundant iteration histories of industrial systems. Realizing this idea, however, raises four challenges: sparse supervision, noisy data, a vast direction space, and prohibitively expensive verification. We address them along two fronts. On the data side, we design a pipeline that turns noisy historical commits into a large, clean evolutionary corpus via pairwise sample expansion and noise filtering. On the model side, we train the model through mid-training, SFT, and RL, guiding evolution direction generation with hierarchical hints and using the reward model in RL as a fast surrogate evaluator. Astar has been deployed in Alibaba's Lazada advertising system for evolution direction proposal. Astar-8B achieves a single-proposal success rate of 0.6786 in real-execution evaluation, far exceeding human experts (0.3229) and the strongest general-purpose LLM (0.3071). More importantly, Astar closes the loop and enables fully automatic iteration: it guided 20 consecutive iterations over two weeks, improving offline Hitrate@200 by 23.6%, while an online A/B test yielded relative lifts of 4.86% in GMV and 1.82% in advertising revenue.
Read the full abstract

All authors Jinxin Hu, Hao Deng, Haibo Xing, Lingyu Mu, Muyu Zou, Weiqin Yang, Sirui Chen, Bohao Wang, Zhezheng Hao, Hao Zhang, Zulong Chen, Shizhun Wang, Yu Zhang, Xiaoyi Zeng, Jiawei Chen

Machine Learning

MM-Spectrum: Multimodal Multi-spectral Molecular Structural Elucidation with a Stable MoE Framework

Inferring molecular structures from multimodal spectroscopic measurements requires integrating complementary yet highly heterogeneous signals. However, the common paradigm of directly concatenating multispectral sequences can exhibit anomalous performance degradation, primarily due to pronounced heterogeneity and the resulting multimodal imbalance across modalities. As a remedy, we propose MM-Spectrum, a sparse Mixture-of-Experts framework tailored for multimodal multispectral spectra-to-structure elucidation. To better match the information characteristics under multispectral imbalance, MM-Spectrum introduces an explicit modality-aware routing mechanism that exposes spectral identity to the router in addition to token content representations. Moreover, it incorporates shared and interaction experts, together with heterogeneous expert capacities, to extract multispectral modality-unique and cross-modal synergistic information while suppressing noise-induced interference. Across full-modality, bimodal, and missing-modality settings on molecular structural elucidation, MM-Spectrum achieves consistent and substantial improvements, supported by ablation studies and interpretability analyses.
Read the full abstract

All authors Hai-tao Yu, Nan Min, Zheng Fang, Hongyu Zhan, Yusen Tan, Yuhan Wang, Jun Xia

Number Theory

Birch's theorem over function fields with quadratically many variables

For smooth hypersurfaces over rational function fields of characteristic greater than the degree and with sufficiently large constant field, we improve the number of variables required in Birch's theorem from an exponential function of the degree to a quadratic one. This agrees, up to constants, with the sharp quadratic threshold for the unconditional existence of rational points on smooth hypersurfaces. A circle method argument reduces the required cancellation to lower bounds for the codimensions of certain singular loci associated with complete exponential sums over finite fields. Our main innovation is a new method for proving these bounds: we introduce the notion of multiplication rank for the linear functionals indexing these exponential sums and combine the resulting rank stratification with a weighted degeneration of the Jacobian equations to obtain a codimension estimate that grows linearly with multiplication rank.
Read the full abstract

Representation Theory

Noncommutative Cluster Varieties and Moduli Spaces of Local Systems

In this article, we construct noncommutative cluster varieties, $\mathcal{A}_{R,S}$, for each reduced root system $R$ and marked surface $S$ simultaneously generalizing the cluster varieties of Fock-Goncharov, Li, Goncharov-Shen, Berenstein-Retakh, Goncharov-Kontsevich, and our previously introduced polygonal cluster algebras. Additionally, we define a large class of algebraic groups, we call Jordan split groups. Given a reduced root system $R$ and a family of Jordan algebras, the Lie algebra for $G$ is constructed by unifying the Tits-Kantor-Koecher construction for a single Jordan algebra with the construction of a split Lie algebra. The notion of Jordan split groups is closely related to a grading of its Lie algebra by the root system $R$. We show that these gradings are usually induced by a choice of standard parabolic subalgebra $\mathfrak{p}_\Theta$ and we classify $R$-graded pairs $(G,\Theta)$ via a condition depending only on the subset $\Theta\subset \Delta$ of the set of simple roots. Next, we define Jordan algebra points of $\mathcal{A}_{R,S}$ which parameterize $G$-local systems on $S$ with boundary decoration related to cosets $G/U_\Theta$ when $G$ is Jordan split of type $R$. When $S$ is a disk, points of $\mathcal{A}_{R,S}$ parameterize configurations of decorated flags. We use this to give noncommutative cluster structures on the double $R$-Bruhat cells of $G$, generalizing the cluster algebras of Berenstein-Fomin-Zelevinsky. When each Jordan algebra is formally real, we say that $G$ has a positive structure with respect to $\Theta$. This defines a positive semigroup in $G$. For real algebraic groups, the pairs $(G,\Theta)$ which have positive structures are exactly those which admit a positive structure as defined by Guichard-Wienhard and we give algebraic proofs of many of the properties of positive configurations of flags and of positive representations.
Read the full abstract

All authors Zachary Greenberg, Dani Kaufman, Merik Niemeyer, Anna Wienhard

Number Theory

The Matrix Pythagorean Equation over $\mathrm{GL}_2(\mathbb{Z})$

In this paper, we study the ordered solutions of the matrix Pythagorean equation $X^2+Y^2=Z^2$ over $\mathrm{GL}_2(\mathbb{Z})$ . Exploiting the independent sign symmetry of the equation, we first reduce the full solution set to the trace-nonnegative subset $\mathcal M$, consisting of those solutions for which all three traces are nonnegative. We then determine a canonical decomposition of $\mathcal M$ into orbits under simultaneous integral conjugacy.
Read the full abstract

Computer Vision

TADP: Task-Aware Deformable Prediction for Single-Stage 3D Object Detection

Most single-stage 3D object detectors complete different tasks with the same extracted features. Nevertheless, it is impossible to project features into a common space that is adaptive for all the tasks. We present a novel task-aware deformable prediction (TADP) method for single-stage 3D object detection to solve this problem. Firstly, a triple feature refinement aggregation module is designed to extract three-level features adaptively. Additionally, we design the multi-scale feature aggregation block to fuse multi-scale features in a scale-aware manner. Finally, the prediction of each task is deformed with the designed plug-and-play task-aware deformation head. It can percept the emphasis and interaction of each task. We also designed three different deformation modules. The experimental results demonstrate that the proposed deformation head shows good results on other detection methods. The experimental results on the KITTI dataset demonstrate that the car mAP is 80.91%, surpassing many state-of-the-art methods on the KITTI benchmark.
Read the full abstract

All authors Su Wang, Yaochen Li, Min Yang, Jiaohao Nie, Chang Liu, Yuehu Liu

Quantum Physics

Efficient formation and identification of single emitters in 4H-SiC following maskless heavy ion implantation

Single photon emitters in silicon carbide (SiC) are a leading platform for scalable quantum technologies. Recent interest has focused on oxygen-vacancy-related emitters, which show exceptionally high optical brightness and strong spin readout contrast. One barrier to scalable quantum devices based on these emitters is the challenge of maskless formation and rapid identification. Here, we demonstrate the formation of isolated bright single emitters in 4H-SiC, using low-energy maskless implantation of heavy ions bismuth and tin. Following annealing, up to 18% of implanted sites host a single emitter, with optimal yields achieved at annealing temperatures of 900-1000 degrees C. Occupancy statistics are modelled to estimate the implantation dose that maximises single-emitter yield. We introduce a tiered characterisation scheme, where a simple intensity threshold isolates single-emitter candidates, confirmed through photon correlation measurements, after which correlations between polarisation, saturation count rate and magnetic resonance frequency assign emitter type. It is shown that time-consuming low-temperature spectroscopy is not necessary to distinguish emitter types. Together, maskless heavy-ion implantation and selective screening offer an efficient route to forming and rapidly identifying near-surface single emitters for room-temperature quantum technologies such as quantum sensing.
Read the full abstract

All authors S. Eserin, L. K. S. Zimmermann, E. Schneider, M. Ludlow, J. Heiler, O. G. Lloyd-Willard, K. Stockbridge, G. Aresta, L. Antwis, S. K. Clowes, M. Hofmann, S. Kucera, P. Berwian, F. Kaiser, B. N. Murdin

Computer Vision

Sidecar: Training-Free Semantic Reuse for Character-Consistent Free-form Visual Storytelling

Visual storytelling requires generating images that follow a narrative while preserving consistent character identities across frames. In free-form story generation, a character is fully described only when first introduced and is later referred to by a type-level mention or pronoun. Although this setting better reflects natural storytelling, later prompts may omit important identity-related semantics, making character consistency more difficult to maintain. We propose Sidecar, a plug-and-play semantic augmentation module that preserves entity-level information from the initial description and injects the missing semantics into later prompt embeddings. Sidecar requires no additional training and does not modify the architecture of the base diffusion model. Experiments on FreeStoryBench show that Sidecar consistently improves prompt-image alignment and character consistency across multiple SDXL- and FLUX-based baselines, with negligible computational overhead.
Read the full abstract

High Energy Physics (Phenomenology)

From the November Revolution toward the Millennium

The 4th International Symposium on the History of Particle Physics spanned two decades, from approximately 1980 to 2000. In this opening keynote lecture, I survey developments in particle physics from the prehistory of the $J/\psi$ discovery to the early nineteen-eighties. I offer context for the contributions that follow in the symposium program.
Read the full abstract

Distributed Computing

Decoupled I/O-Dominant Pipelines for Large-Scale Whole-Slide Image Embedding Extraction

Whole-slide images (WSIs) are central to computational pathology but are prohibitively large, making patch-based processing the practical unit for foundation model inference. At scale, however, generating and handling massive numbers of patches on quickly introduces significant I/O and orchestration overhead, often dominating end-to-end performance. We present a decoupled, I/O-aware pipeline for large-scale WSI embedding extraction that decomposes the workflow into three stages: (1) patch generation and staging, (2) embarrassingly parallel embedding inference, and (3) sharded vector database ingestion. This design isolates data movement from compute, enabling efficient patch delivery, scalable multi-node inference with minimal communication. The resulting system produces a distributed vector database where embeddings are persistently coupled with rich metadata (e.g., patient, slide, and patch attributes), enabling efficient filtering, retrieval, and downstream reuse. This representation database is compact and reusable for tasks such as retrieval, classification, and few-shot learning, particularly benefiting low-resource environments. We show that decoupling I/O, computation, and ingestion enables high-throughput WSI embedding extraction at scale. By characterizing the scaling envelope, we demonstrate that storage dominates beyond moderate concurrency, reframing WSI embedding extraction as a data-centric systems problem rather than a purely compute-bound workload.
Read the full abstract

All authors Mayanka Chandrashekar, Xi Zhang, Ethan Seefried, Tirthankar Ghosal, John Gounley, Heidi Hanson

Materials Science

Benchmarking of Fast and Interpretable UF Machine Learning Potentials

Machine learning interatomic potentials (MLIPs) have emerged as a powerful alternative to density functional theory (DFT) for molecular dynamics simulations, offering near-DFT accuracy at a fraction of the computational cost. However, many state-of-the-art MLIPs remain computationally demanding and act as black boxes, limiting physical interpretability. In this work, we evaluate the ultra-fast force field (UF$^3$) potential, which employs linear regression with cubic B-spline basis to represent effective two- and three-body interactions. We show that UF$^3$ displays accuracy comparable to established models such as GAP, MTP, NNP (Behler Parrinello), and qSNAP MLIPs. We further investigate the transferability of UF$^3$ by computing melting points for six elemental systems with potentials fitted without any solid-liquid interface configurations or explicit thermodynamic information about melting. The model reproduces experimental melting points within $\sim$6% for simple metals (Ni, Cu, Li), but substantially underestimates them for Mo and Si and fails to yield a stable potential for Ge, reflecting the limitations of a fixed expansion truncated at the three-body term for systems with strong angular or covalent bonding. We further illustrate how UF$^3$'s spline-based formulation allows direct visualization of the learned interactions, enabling identification of unphysical behavior that black-box approaches often obscure.
Read the full abstract

Analysis of PDEs

The complete spectrum of the linearized $p$-Laplacian at a Sobolev extremal

Let $1<p<n$ and let $v(x)=(1+|x|^{p/(p-1)})^{-(n-p)/p}$ be the standard radial extremal for the sharp Sobolev inequality. We determine all eigenvalues and eigenspaces of the linearized $p$-Laplacian at $v$, defined by its closed quadratic form in $L^2(\mathbb{R}^n,v^{p^*-2} dx)$. After decomposition into spherical harmonics, an explicit gauge transformation and a change of variables identify each radial operator with a shifted Jacobi operator. This yields a complete eigenbasis indexed by $(\ell,k)\in\mathbb{N}_0^2$, where $\ell$ is the angular degree and $k$ is the radial mode number.
Read the full abstract

Numerical Analysis

A Multilevel Interacting Particle System Method for the estimation of Failure Probabilities

We propose a novel method that combines a multilevel decomposition with the framework of Interacting Particle Systems to compute the probability that given quantities of interest take values above or below a certain pre-specified value. Through the application of a sequential sampling scheme, the algorithm is able to achieve convergence rates faster than those achieved by the classical Multilevel Monte Carlo algorithm, and competes with the current state of the art techniques, while also being dimension independent. Furthermore, the sample points generated by the algorithm form a sequence which concentrates near the critical value of the quantity of interest. Our results are rigorously established and later verified through numerical experiments.
Read the full abstract

Dynamical Systems

On some aspects of discrete groups acting ergodically on the boundary

We show that if $G$ is a real semisimple Lie group and $\Gamma<G$ is a discrete subgroup with slow growth, in the sense that its growth indicator function is smaller than $\rho$, then $\Gamma$ acts totally dissipatively on the Furstenberg boundary of $G$. Moreover, for $G= SO(n,2)$ and $n\geq 3$, we construct infinite-covolume discrete subgroups that act ergodically on the Furstenberg boundary of $G$, providing a counterexample to a conjecture of Margulis for $G = SO(n,2), n > 2$. The examples arise from lattices $\Gamma<H= SO(n,1)$ and their deformations in $G$. Perhaps more importantly, we describe a new approach to studying deformations of such lattices by relating them to deformations of the smooth right-translation action of $SO(n-1,1)$ on $\Gamma\backslash H$. This correspondence allows us to apply recent results of DeWitt and Dolgopyat on smooth group actions and to give examples where the right-translation action of $SO(n-1,1)$ on $\Gamma\backslash H$ can fail to be $C^0$-locally rigid.
Read the full abstract

Materials Science

Grain-Boundary Premelting in High-Entropy Transition Metal Carbides

Grain-boundary segregation and thermally induced interfacial disordering were investigated in four high-entropy transition metal carbides using Monte Carlo (MC) sampling and molecular dynamics (MD) with the universal MACE-OMAT-0 machine-learning interatomic potential. MC sampling segregated the group-VI element (Cr, Mo, or W) and Zr to grain boundaries, where the group-VI content reached approximately 45 at.%, consistent with STEM-EDS observations. During MD heating, the grain-boundary Lindemann index, a normalized measure of interatomic distance fluctuations, reached the liquid-like threshold of $\delta=0.15$ near $1390^{\circ}\mathrm{C}$ for the Cr-containing carbides, $1660^{\circ}\mathrm{C}$ for Mo, and $1890^{\circ}\mathrm{C}$ for W, while the grain interiors remained below the threshold. A chemically random (Cr,Hf,Ta,Ti,Zr)C reference crossed about $60^{\circ}\mathrm{C}$ later and showed less boundary-localized disorder, highlighting the role of interfacial chemistry in premelting. Species-resolved displacements showed enhanced grain-boundary mobility, particularly for carbon. Overall, Cr-rich interfaces showed the earliest and most extensive premelting-like response, followed by Mo- and W-containing boundaries.
Read the full abstract

All authors Marium M. Mou, Caleb Schenck, Samuel E. Daigle, William G. Fahrenholtz, Bharat Gwalani, Stefano Curtarolo, Donald W. Brenner

Fluid Dynamics

Energy transfer and scale organisation in dense canopy turbulence

This study identifies the scale-dependent processes that sustain turbulence in dense submerged canopy flows. Using a scale-resolved energy budget, we determine where in space and at which scales production, pressure-strain redistribution, and inter-scale transfer predominantly occur, and how they link the canopy layer to the overlying shear flow. We show that energy production is localised at the interfacial shear layer, over a narrow range of streamwise and spanwise scales, while fluctuations within the canopy are primarily maintained through inter-scale transfer and pressure-strain redistribution. The dynamically active scales in the canopy are largely imposed by outer-layer structures, with their organisation and coherence mediated by these inherited motions. Energy exchange across the canopy interface is asymmetric but not unidirectional: although the dominant transfer is from the outer layer towards the canopy, intermittent reverse interactions occur at all scales. The most intense cross-interface exchanges are associated with finer-scale motions rather than large-scale structures, indicating that extreme interfacial energy fluxes are governed predominantly by small-scale dynamics. The flexibility of the canopy weakens the coherence of outer-layer structures and reduces the efficiency of inter-layer energy transfer, thereby altering both the organisation and scale of the fluctuations within the canopy. These results clarify how turbulence in dense canopies is organised and sustained, providing a unified energetic interpretation that links coherent structures to scale-dependent mechanisms.
Read the full abstract

All authors Riccardo Bertoncello, Alessandro Chiarini, Giulio Foggi Rota, Maurizio Quadrio, Marco Edoardo Rosti

Optimization and Control

Marine Autonomous Vehicle Fleet Scheduling to Maximise Scientific Impact

The marine science community increasingly relies on Marine Autonomous Vehicles (MAVs) to collect the critical environmental data required to understand global ocean systems. However, as these operations scale, manually routing and planning large autonomous fleets becomes exponentially complex and time-consuming. To address this, we propose a mixed-integer linear programming (MILP) model designed to automate and optimise MAV deployment schedules. The model accounts for strict operational constraints, including battery capacities and time windows for data collection, while aiming to maximise total data collection and minimise both the number of deployed vehicles and their energy consumption. A key novelty of this framework is integrating conventional ship itineraries, allowing MAVs to support vessels with mid-mission battery swapping or accelerated transit between waypoints. Computational experiments demonstrate that the model is highly scalable, solving routing problems for fleets of dozens of MAVs in seconds, and scaling to hundreds of vehicles in only a few minutes. Beyond operational scheduling, the framework serves as a robust simulation tool for evaluating 'what-if' scenarios and analysing the impact of varying parameters on deployment strategies. Finally, the solution generates a suite of visualisations designed to enhance explainability and support strategic decision-making for stakeholders.
Read the full abstract

Astrophysics of Galaxies

Revisiting the enigmatic sixth star cluster in the Fornax dwarf spheroidal galaxy

The Fornax dwarf spheroidal (dSph) galaxy is one of only two Local Group dSphs that host a population of globular star clusters (GCs), the present-day properties of which have often been used to investigate the nature of the dark matter. An additional overdensity of stars called Fornax 6, lying at a projected distance of $\sim$0.3 kpc from the galaxy centre, was recently identified as a sixth GC residing in Fornax. However, this conclusion was based on relatively shallow photometric observations and a limited number of spectra. Here, we reinvestigate the nature of Fornax 6 by analysing MUSE/Wide-Field-Mode observations alongside deep GMOS-S imagery. Using a sample of 131 spectra of red giant and red clump stars, we confirm the Fornax 6 overdensity as a distinct chemo-dynamical component with respect to the surrounding field population. Specifically, we identify 43 likely members associated with the cluster, from which we measure $\rm [Fe/H]=-0.63\pm0.03\,$dex, $\overline{v_{los}} = 50.8^{+0.8}_{-0.7}\, \rm km\,s^{-1}$, and $\sigma_v = 3.3^{+1.4}_{-1.9}\, \rm km\,s^{-1}$. The main-sequence turn-off, observed here for the first time, strongly suggests that these stars are coeval and well described by a $\sim$3 Gyr old isochrone. The cluster is characterised by an irregular morphology, a large half-light radius ($R_h=8.8^{+1.1}_{-1.3}\,$pc), a small flattening ($e<0.26$, 68% upper limit), and low luminosity ($M_V = -5.0 \pm 0.4$). Our improved characterisation of Fornax 6 supports its classification as a genuine low-mass cluster likely undergoing tidal disruption, making it the youngest and most metal-rich member of Fornax's unique GC system.
Read the full abstract

All authors Chiara Crociati, Annette M. N. Ferguson, Raffaele Pascale, Sebastian Kamann, Silvia Martocchia, Jess M. Howell, Pete B. Kuzma, Dougal Mackey, Sakurako Okamoto, Masato Onodera

Algebraic Topology

Persistence Meets Resistance: Doubling Down on Hardness

We present results on the approximate computation of stable invariants for filtrations of finite metric spaces in the context of persistent homology. We establish novel approximation algorithms in the setting of $n$-point metric spaces where the growth of the doubling dimension is in $o(\log n)$ and the diameter is bounded. In the $1$-parameter case, by revisiting known techniques (greedy permutations) in a new way, we derive the first linear-time algorithms for the problem of computing additive $\varepsilon$-approximations of any stable barcode. By deriving bounds on the convergence rate and the approximation quality of uniform samples, we extend the approach to selected multiparameter filtrations. We show that for normalized measure bifiltrations, including the multicover and subdivision-Rips bifiltration, any stable invariant can be probabilistically approximated in time constant in $n$. The constants in the running times of our algorithms depend on the doubling dimension, the diameter and the success probability. We further study the problem through the lens of fine-grained complexity and show that computing the rank of a matrix reduces to that of approximating the barcode of the Vietoris–Rips or Čech filtration. We present two variants of the reduction, one for sufficiently good additive approximations and the other for any constant factor multiplicative approximations.
Read the full abstract