Peer Revenue Play credits. Real papers.

Papers

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

All of arXiv

Thursday 8 October

Computational Finance

Multi-period Mean-Expectile Portfolio Optimization under Wasserstein Ambiguity: Reformulation, Degeneracy and the Role of the Ground Metric

Rupendra Yadav and Aparna Mehra

Expectiles are the only law-invariant risk measures that are both coherent and elicitable. Unlike Conditional Value-at-Risk (CVaR), however, they do not admit a Rockafellar–Uryasev representation that admits tractable Wasserstein reformulations. We address this difficulty by developing an envelope theorem for worst-case expectiles that characterizes the worst-case expectile over a Wasserstein ambiguity set as the unique root of a worst-case expectation with a two-piece affine integrand. This representation permits direct application of standard Wasserstein duality. Using this result, we reformulate a multi-period tri-level mean–expectile portfolio problem as four parametric linear programs with constraints. We establish four structural properties of the proposed model: an endogenously damped price of robustness, a decision-dependent critical radius beyond which the expectile tail component becomes inactive, exact recovery of the nominal model at zero ambiguity, and a characterization of how the Wasserstein ground metric determines whether the limiting portfolio becomes more concentrated or more diversified. Numerical experiments on 90 FTSE constituents over 3,341 out-of-sample trading days show that the expectile model outperforms a CVaR model matched on ambiguity set, radius, ground metric, and trade-off weight in all nine parameter cells—significantly so whenever the radius is non-trivial. The experiments further confirm the predicted degeneracy under the ground metric.
Read the full abstract

Dynamical Systems

Bounds on the maximum number of limit cycles of piecewise linear Lienard systems II. The discontinuous case

Hebai Chen, Jie Jin, Shu Li et al. (+1 more)

This paper concerns the planar Liénard system \(\dot x=y-F(x),\ \dot y=-x\), where \(F(x)\) is a piecewise linear function with exactly \(m\) jump points and no fold points. Tonnelier [SIAM J. Appl. Math. 63 (2002)] conjectured that the maximum number of limit cycles of the system is \(2m\). The conjecture was confirmed for \(m=1\) in [J. Lond. Math. Soc. 113 (2026)], and a lower bound of \(2m\) for arbitrary \(m\) was established by Chen et al. [arXiv:2608.19542]. In this paper, we construct systems with at least \(4m-2\) hyperbolic crossing limit cycles for every positive integer \(m\), thereby disproving Tonnelier's conjecture for \(m\geq2\). We also establish the uniform upper bound \(2^{224(m+1)^2}\) for the total number of crossing, grazing, sliding, and composite limit cycles.
Read the full abstract

All authors Hebai Chen, Jie Jin, Shu Li, Yuhuan Lu

Computation and Language

Not Every Change Is Necessary: Recoverable Drift in Large Language Model Unlearning

Xunlei Chen, Qinghui Gong, Jingkun Xue et al. (+3 more)

Machine unlearning in large language models aims to remove unwanted knowledge while preserving the model's remaining capabilities. Although existing methods use retention objectives or restrict where edits occur, achieving the desired forgetting level can still leave collateral changes that impair non-target behavior. Our recovery comparisons suggest that some of these changes can be reversed while preserving observed forgetting performance. In this work, we present Propose-Then-Project Unlearning (PTP-U), a framework that combines targeted forgetting with the recovery of non-target capabilities. PTP-U first applies local analytic edits to weaken target knowledge associations, then aligns non-target output distributions with those of the original model to recover capabilities while maintaining fixed forgetting constraints. Both stages serve a common goal: satisfying the forgetting requirements while preserving fluent generation and performance on non-target tasks. Across three benchmarks, PTP-U achieves the strongest forgetting-retention trade-off among evaluated methods, reaching 81.22%-91.03% forgetting while preserving 94.20% non-target utility on average. At matched forgetting, PTP-U consistently retains higher non-target utility.
Read the full abstract

All authors Xunlei Chen, Qinghui Gong, Jingkun Xue, Qihe Liu, Shijie Zhou, Fei Ye

Machine Learning

Puffin: Probabilistic Learning of Spatial Detail From Coarse Observations

Chaitanya Jobanputra, Sebastian Vollmer and Gerrit Großmann

High-resolution socioeconomic variables are important for applications such as urban planning, public health, disaster response, and resource allocation. In practice, however, these variables are often observed only at a coarse spatial resolution. We introduce Puffin, a probabilistic framework for statistical disaggregation that raises the resolution of coarse totals using high-resolution satellite embeddings as covariates. Instead of predicting a single value for each fine-resolution subregion, Puffin learns a probability distribution and is trained through an aggregation-aware likelihood. At inference, Puffin conditions these predictions on the observed regional total and splits it among the subregions. The resulting fine-scale estimates are consistent with the observed aggregate and come with calibrated uncertainty, without requiring fine-resolution labels for training. We evaluate Puffin on German and US census, employment, and election data across population, jobs, and other count variables, and study when statistical disaggregation succeeds or fails across regions, countries, and targets.
Read the full abstract

Logic in Computer Science

Formally Certifying the Vertex Set of a Polyhedron Faster than Informal Enumeration

Xavier Allamigeon, Yazid Id-Sahra and Pierre-Yves Strub

The computation of the vertices of a polyhedron described by a system of linear inequalities is a central problem in polyhedral computation. It is a fundamental step in the conversion between H-representations, by linear inequalities, and V-representations, by vertices and extreme rays. This operation plays an important role both in the study of polyhedra and their combinatorics in mathematics and in applications to software and system verification. We present a certificate-based approach for formally verifying the computation of the vertices of a polyhedron. Given an informally computed list of vertices, our method allows to certify in the proof assistant Rocq that the list is complete, or even exact. The cornerstone of the method is a new completeness criterion based on an abstract simplicial complex that generalizes a triangulation of the normal fan of the polyhedron. A significant advantage over previous approaches is that the usually expensive numerical computations are essentially reduced to membership tests to the polyhedron, while the other steps are cheap combinatorial tests. We implement the certification method and prove its correctness in the proof assistant Rocq. We experiment with it on a variety of polyhedra, including Birkhoff polytopes, cross-polytopes, cubes, permutahedra, hypersimplices, and high-dimensional polytopes involved in the disproof of the Hirsch conjecture. Our experiments show that certification with the Rocq-to-OCaml extracted checker is typically 1.5x to over 5x faster than vertex enumeration by the state-of-the-art informal C implementation lrslib of the reverse search method.
Read the full abstract

Numerical Analysis

Acceleration of Newton's method for nonlinear algebraic systems by preflattening and elimination techniques

Ngoc Do Quyen Dang, Ibtihel Ben Gharbia, Clément Cancès et al. (+1 more)

In this paper, we propose a new approach called preflattening to speed up the resolution of nonlinear algebraic systems by Newton-type methods. Preflattening is meant to be for nonlinear systems what preconditioning is for linear ones. The idea is to transform the system into an equivalent one that is more favorable to Newton's method. To this end, we introduce the notion of flatness, which is intended to play for nonlinear systems a role similar to that of the condition number for linear systems. For a physical model of interest in the geosciences, namely the chemical equilibrium problem, we combine this strategy with the nonlinear elimination technique and demonstrate the value of the resulting approach through numerical simulations.
Read the full abstract

All authors Ngoc Do Quyen Dang, Ibtihel Ben Gharbia, Clément Cancès, Quang Huy Tran

Differential Geometry

The Hawking and Penrose singularity theorems for low regularity Finsler spacetimes

Darius Erös, Ettore Minguzzi, Argam Ohanyan et al. (+1 more)

We prove the Hawking and Penrose singularity theorems for Lorentz–Finsler structures of mixed horizontal/vertical regularity $C^{(1,2)}$ in the $C^1$ weighted case and $C^{(1,3)}$ in the unweighted case. At this regularity, geodesics need not be uniquely determined by their initial conditions, and the (weighted) Ricci curvature must be interpreted distributionally. Our approach relies on a two-stage approximation procedure: first, a homogeneity-preserving convolution, which is of independent interest in the broader semi-Riemann–Finsler setting, and second, a causality-adapted approximation à la Chruściel–Grant. In the special case of $C^1$ Lorentzian metrics, our results extend the $C^1$ singularity theorems of Graf to the $C^1$ weighted case.
Read the full abstract

All authors Darius Erös, Ettore Minguzzi, Argam Ohanyan, Shin-ichi Ohta

Materials Science

Comprehensive study of massively overlapping cascades in common elemental metals

Aslak Fellman, Jesper Byggmästar and Fredric Granberg

Massively overlapping cascades simulations were carried out in 21 elemental metals: Be, Al, Ti, V, Cr, Fe, Co, Ni, Cu, Zr, Nb, Mo, Rh, Pd, Ag, Hf, Ta, W, Pt, Au and Pb. These elements have simple FCC, BCC and HCP lattice structures. For each element, 2000 cumulative 5 keV cascades were simulated using molecular dynamics. The overlapping cascades were simulated using various classical analytical interatomic potentials as well as with machine-learning interatomic potentials for many of the elements. The general conclusion is that results from massively overlapping cascades simulations are very sensitive to the choice of interatomic potential across the periodic table. Furthermore, we see that FCC metals all form stacking fault tetrahedra due to irradiation, and also typically include interstitial type Shockley partial and Frank type dislocations. In BCC materials typically form interstitial 1/2$\langle$1 1 1$\rangle$ dislocations, but vacancy type 1/2$\langle$1 1 1$\rangle$ dislocations were also observed in V and Nb. HCP materials tend to form complex dislocation structures mainly consisting of a-type dislocations that can be of both vacancy or interstitial type. Correlations between saturated defect concentrations and fundamental underlying material properties are explored.
Read the full abstract

Statistical Mechanics

The Fiedler dimension of networks of networks: from fractal to small-world architectures

Lorenzo Grimaldi, Andrea Gabrielli and Pablo Villegas

Is the relaxation of a modular network dictated by its modules or by the network that connects them? The answer fixes the time scales of diffusion, consensus, and synchronization, all set by the Fiedler eigenvalue. We show that, for bundled networks, the two levels act one after the other: a random walker must first escape from its module and then spread over the network that connects them. The equilibration time is the sum of these two times, the time needed to reach the base from inside a fiber and the relaxation time of the base, slowed down by the mass of the fibers. The consequences are unexpected. However large the modules are, a small-world core always imposes its own Fiedler dimension; the modules survive only as a logarithmic correction, which slows down equilibration and hides the true exponent up to sizes far beyond any real network. Effective dimensions measured on finite modular systems can therefore be systematically biased. We analytically work out all combinations of finite-dimensional and small-world bases and fibers, and compare them with numerically exact spectra of representatives of each category.
Read the full abstract

Databases

Cost-Aware Mixture-of-Experts Coordination for Model Markets

Yizhou Ma, Wenbo Wu, Xikun Jiang et al. (+2 more)

Existing model marketplaces typically trade and select individual models as indivisible units, limiting their ability to exploit complementarities among heterogeneous experts. This paper proposes an MoE-based model market framework that lifts Mixture-of-Experts from a model-level learning architecture to a market-level coordination mechanism. In this framework, brokers use gating networks to coordinate multiple heterogeneous experts and deliver a composite model service. We formalize the market participants, service workflow, expert cost structure, and a welfare objective that combines predictive utility with heterogeneous execution costs. We then derive a cost-aware gating mechanism and market-aware training objective, and introduce a cost-adjusted revenue allocation rule that distributes residual revenue according to realized expert participation and execution cost. We also establish basic theoretical properties of the allocation rule, including budget balance, participation monotonicity, and cost sensitivity. Experiments over five random seeds on fifteen tabular and image benchmarks use independently trained and frozen neural and tree-based experts together with latency-derived execution costs. MoE Market achieves the highest mean welfare on all fifteen datasets and a lower mean expected cost than Standard MoE in every case, while maintaining competitive predictive performance. The allocation experiments further demonstrate systematic sensitivity to expert participation and cost, together with substantially lower computational overhead than exact Shapley allocation. These results suggest that MoE can serve as a market-level coordination principle for collaborative, cost-aware, and economically grounded model marketplaces.
Read the full abstract

All authors Yizhou Ma, Wenbo Wu, Xikun Jiang, Zhuoqin Yang, Luis-Daniel Ibáñez

Computer Vision

Beyond Visual Enhancement: Adaptive Multi-Context Steering to Mitigate LVLM Hallucinations

Shuran Ma, JiaLe Li, Yuxin Dong et al. (+9 more)

Hallucination remains a significant challenge in Large Vision-Language Models (LVLMs). Existing training-free methods generally mitigate hallucinations through contrastive decoding or visual enhancement, often increasing the relative influence of visual evidence during generation. This raises a fundamental question: Can LVLMs dynamically regulate the contributions of different context sources to suppress hallucinations? In this work, we investigate and quantify how LVLMs coordinate multiple context sources during decoding and examine how this intrinsic behavior can guide hallucination mitigation. We find that LVLMs exhibit an intrinsic vision-attending tendency that can guide adaptive visual steering, while textual contexts can also contribute to hallucination mitigation. Motivated by these findings, we propose AIMS (Adaptive Information Multi-source Steering), a lightweight training-free framework that adaptively coordinates visual, prefilled textual, and generated contexts during decoding. Specifically, AIMS constructs compact prototypes for the three context domains and estimates their affinities with the current query to determine head-wise steering weights. The resulting multi-source steering direction is applied to the query representation, enabling adaptive context integration without additional model training or auxiliary forward passes. Extensive experiments across multiple LVLMs and decoding strategies demonstrate that AIMS effectively mitigates object hallucination while maintaining competitive general-purpose multimodal capabilities.
Read the full abstract

All authors Shuran Ma, JiaLe Li, Yuxin Dong, Shan Zheng, Qingyun Jiang, Xiang Chen, Qi Zhu, Deyi Ji, Yifan Yang, Jianfeng Pan, Yu Tian, Xue Yang

Machine Learning (Statistics)

RobustLDS: Learning linear dynamical systems under adversarial corruptions

Aravinda Kanchana Ruwanpathirana and Hemant Tyagi

We consider the problem of learning linear dynamical systems under adversarial contamination from a single trajectory of length $T$. While identification of linear dynamical systems itself is well-studied, the problem of robust system identification under adversarial contamination is relatively less explored. In this work, we study the setting where a fraction of the $T$ observations are contaminated by adversarial outliers. We propose different estimators based on relaxations of least-trimmed squares along with an alternating minimization algorithm. Furthermore, we also propose two estimators which exploit the group-sparsity (through penalization/hard-constraints) of the outliers. For the estimator with group-sparse penalty, we derive non-asymptotic error bounds which establish its robustness to outliers. We also show empirically that the proposed estimators work well in practice.
Read the full abstract

Number Theory

Uniform Height Gaps in Arbitrary Characteristics

Alice Lin, Frank Lu and Jit Wu Yap

We prove uniform height gaps for subvarieties of abelian varieties in arbitrary characteristic, extending the new gap principle of Gao-Ge-Kühne. Our proof goes through the theory of adelic curves and globally valued fields, and proves the Bogomolov conjecture for an arbitrary globally valued field. This gives a different way to obtain uniform Bogomolov-type results, differing from the approaches of Dimitrov-Gao-Habegger-Kühne and Yuan. First, we prove the Bogomolov conjecture over globally valued fields for divisors. We then reduce the case of general subvarieties to the case of divisors by an induction argument. Specializing to the case of global function fields, we obtain a new proof of the geometric Bogomolov conjecture following the strategy of Gubler and Yamaki.
Read the full abstract

Signal Processing

Large-Scale Partition-Based RIS Beamforming For Uplink RIS-Equipped Multi-User Systems: Asymptotic Analysis

Samira Rahimian and Haris Gacanin

Combining a reconfigurable intelligent surface (RIS) with a receive antenna array is a promising low-complexity architecture for multi-user uplink reception, but its performance analysis for more than two users has remained an open problem: the zero-forcing (ZF) signal-to-interference-plus-noise ratio (SINR) no longer admits an explicit, low-dimensional closed-form expression, and its distribution is analytically intractable for design purposes. This paper addresses this gap for a K-user, K-antenna uplink system in which a large-scale, L-element RIS, partitioned into K user-dedicated sub-surfaces, precedes ZF reception at the base station. Through an asymptotic analysis in which the sub-surface sizes grow without bound, we show that the orthogonal projector underlying the ZF SINR converges to a rank-one matrix aligned with the desired user's channel. We use this convergence to derive a closed-form asymptotic approximation for the average per-user SINR that depends only on deterministic channel parameters. Treating this expression as a tractable design objective, we prove that equal partitioning is approximately sum-rate-optimal at leading order regardless of path-loss asymmetry across users. We also develop a low-complexity greedy pairwise-transfer search that refines the partition beyond this leading-order optimum. Monte Carlo simulations across a range of system and RIS sizes confirm that the closed-form SINR tracks the exact simulated rate closely once the RIS is large relative to the number of users. The greedy search also yields consistent, if modest, sum-rate gains over equal partitioning, validating the theory as both an accurate performance predictor and a practical design tool.
Read the full abstract

Soft Condensed Matter

Fluid deformable surfaces with variable thickness - a Surface Shallow-Water-Helfrich model

Jan Magnus Sischka and Axel Voigt

Epithelial tissues play a fundamental role in morphogenesis. Mechanically they can be viewed as thin soft materials exhibiting a solid-fluid duality. The Surface Navier-Stokes-Helfrich model accounts for these properties by combining bending and surface hydrodynamics. In order to account for varying cell thickness we incorporate a thickness field in the spirit of a shallow-water equation, but defined on the (self-)evolving surface. This replaces the inextensibility constraint of the two-dimensional fluid and with it the conservation of surface area by an incompressibility constraint of the thin film fluid allowing for changes in surface area. We develop a numerical scheme based on surface finite elements, perform convergence tests, demonstrate the impact on shape evolution of closed surfaces with a constant enclosed volume and discuss implications on modeling morphogenesis.
Read the full abstract

Functional Analysis

The differentiation and integration operator on weighted spaces of analytic functions in the weight sequence setting

Gerhard Schindl

We investigate the differentiation and integration operator defined on resp. between weighted spaces of analytic functions in the special situation when the weight is given in terms of a weight sequence. More precisely, in this case the weight is defined via the associated weight function which is a crucial object in the ultradifferentiable setting and the information can be then expressed purely in terms of the underlying weight sequence. Due to the additional structure this setting is exploited in order to construct (counter-)examples, to compare conditions on weights in the different weighted frameworks, and to study and to understand in a more precise qualitative way required and admissible growth restrictions on the weights.
Read the full abstract

Computation and Language

Can Decision Models Understand Stance? Evaluating Jev Against General-Purpose LLMs

Xing Li, Jinzhong Ning, Yijia Zhang et al. (+2 more)

Stance detection requires identifying an author's attitude toward a given target, sometimes based on conversational context. Jev, a specialized decision model designed for structured decision-making, offers an alternative to general-purpose large language models (LLMs). In this work, we evaluate Jev on two stance detection datasets, VAST (English texts) and ZS-CSD (Chinese conversations), comparing it with four general-purpose LLMs and two fine-tuned models. Results show that Jev achieves competitive performance on VAST, matching GPT-5.6 and outperforming the other general-purpose LLMs. However, it falls behind stronger LLMs on ZS-CSD, particularly in distinguishing favor from against. Further analysis suggests that this limitation may be related to understanding reply relationships and stance direction rather than conversation length alone. These findings highlight both the potential and limitations of Jev for stance detection.
Read the full abstract

All authors Xing Li, Jinzhong Ning, Yijia Zhang, Liang Yang, Hongfei Lin

Optimization and Control

Reach-Stabilize Control of Control-Affine Systems with Unknown Affine Parameters

Alexander Dorsey, Kunal Garg and Ankit Goel

This paper considers the reach-stabilize prob- lem for a class of nonlinear control-affine systems with unknown parametric uncertainties, where the system states must remain in a safe set at all times, or enter a safe set in finite time and then remain in it, and converge to a goal point. An estimator is designed that generates a parameter estimate and a computable, nonincreasing bound on the estimation error from a known initial error bound, without requiring persistence of excitation. The bound defines margins that are added to a control barrier and a control Lyapunov function condition, which are then enforced in a quadratic program for efficient control design. It is shown that, for the closed-loop system, the safe set is forward invariant when the system starts within the safe set and is finite-time reachable from outside the safe set with a recovery-time bound that does not depend on the unknown parameter, and that the goal point is exponentially stable. In two numerical examples, with the initial state outside and inside the safe set, the proposed control policy, a baseline oracle control policy that uses the true parameter, both recover or remain in the safe set and converge to the goal point, while another baseline control policy that uses a fixed, incorrect, parameter does not remain
Read the full abstract

Computation and Language

Forms of LLM-Integrated Applications from LLM-Chats to Autonomous AI Agent System

Irene Weber

Large language models (LLMs) are increasingly embedded as components in software systems, marketed under labels such as chatbot, copilot, retrieval-augmented generation, workflow, coding agent and AI agent. Whether these labels denote genuine architectural forms or serve as branding has not been assessed systematically. In the sources surveyed, labels do carry architectural content, most clearly in vendor usage: copilot denotes a router-worker architecture operating a host application under step-by-step user confirmation, while the more recent shift to the label agent coincides with AI-planned multi-step execution of which the user sees only the outcome. The coding agents of four major providers share one architecture, a reason-and-act loop delegating to subagents. This survey describes seven recurring forms—LLM chats, custom agents, retrieval-augmented generation (RAG), AI-enhanced workflows, copilots, coding agents, and, in part, agentic RAG—in a common vocabulary of agents and tools. Each is characterized along four structural dimensions (agentic RAG only partially): the architectural pattern, the control of execution and the point of user intervention, the number of agent calls per task, and tool use. An illustrative corpus of 22 systems from research publications and vendor documentation grounds the descriptions and shows where they reach their limit.
Read the full abstract

Econometrics

Inference in Panel SVARs with Two-Way Dependence

Lennart Empting, Saskia Öztürk, Simone Maxand et al. (+1 more)

We develop inference for heterogeneous panel vector autoregressive (VAR) models and their structural impulse response functions, where the error terms are dependent in the cross-sectional and time dimensions (two-way dependence). For proxy-identified structural VARs, we first adapt mean-group estimation and construct a closed-form pooled identification. Considering the reduced-form VAR dynamics, residual covariances, and structural parameters, we derive a joint central limit theorem under joint limits in the cross-sectional and time dimensions. We then propose a recursive-design panel moving-block bootstrap that resamples the estimated error terms in (i) the temporal, (ii) the cross-sectional, or (iii) both dimensions jointly, and prove consistency of the joint panel-block scheme under two-way dependence. Simulations show coverage close to the nominal level for the joint scheme but severe undercoverage for cross-sectional resampling.
Read the full abstract

All authors Lennart Empting, Saskia Öztürk, Simone Maxand, Konstantin Wagner

Quantum Physics

Emergent SSH physics and localization in a cavity–atom system beyond the rotating-wave approximation

H. Shen, J. Cao, R. Wang et al. (+1 more)

We investigate a cavity–atom system beyond the rotating-wave approximation and recast its dynamics as an effective tight-binding problem in the excitation-number basis. The same experimentally relevant light–matter platform exhibits two distinct regimes controlled by the photon number. In the large-photon-number regime, the hopping amplitudes become nearly uniform and the effective lattice reduces to a dimerized SSH-type chain subject to a linear energy gradient. The spectrum then organizes into one or two Wannier–Stark ladders, giving rise to controllable Bloch and Bloch–Zener oscillations and quantitatively explaining the observed revival patterns. In the small-photon-number regime, the intrinsic $\sqrt{\bar n+(\cdots)}$ dependence produces a pronounced hopping deformation, namely dimerized couplings that increase along the excitation-number lattice. This deformation reshapes the density of states, enhances collective localization, and generates an energy-resolved spatial bias of the eigenstates. The corresponding real-time dynamics displays direction-dependent anomalous diffusion together with Bloch-like revivals and beat phenomena in selected parameter windows. Our results show that the same cavity–atom platform can realize both emergent ladder spectra and hopping-induced localization in excitation-number space.
Read the full abstract

All authors H. Shen, J. Cao, R. Wang, X. Z. Zhang

Machine Learning

Automated Assembly Instruction Generation from CAD Models Using Grounded Large Language Models: A Human-in-the-Loop Framework

Aaron Dsouza, Mohammed Azeez Khan, Ashutosh Mishra et al. (+3 more)

Assembly documentation is a downstream manufacturing artifact that is still usually authored by interpreting CAD models by hand. Structured product data and large language models are both available, yet studies of CAD interpretation, assembly sequence planning, instruction writing, and human oversight have largely proceeded separately. This paper formulates CAD-grounded assembly instruction generation: the production of natural-language assembly procedures constrained by structured engineering information extracted from CAD models. The proposed framework maps a STEP assembly to a typed ProductGraph intermediate representation, derives a precedence order by deterministic topological sorting, realizes each step as language conditioned only on selected graph context, attaches per-step visual documentation, and applies rule-based and model-assisted checks. PDF export remains disabled until a human reviewer resolves every quality flag. The case study establishes endto-end feasibility on a built-in six-part reference assembly: the pipeline preserves a reported assembly order and carries quantity, material, and torque into an exported manual page. Generalization and geometric validation remain open empirical questions. The contribution is an architecture that separates engineering state, deterministic reasoning, grounded language realization, verification, and human release.
Read the full abstract

All authors Aaron Dsouza, Mohammed Azeez Khan, Ashutosh Mishra, Arshaan Khan, Neha K. Nair, Amar Kumar Behera

Probability

Permutations from ranking independent random variables

Luke Turvey

Let $X_1,\ldots,X_n$ be almost surely distinct independent real-valued random variables. Let $\sigma$ be the random permutation of $\{1,\ldots,n\}$ such that $X_{\sigma(1)}<X_{\sigma(2)}<\cdots< X_{\sigma(n)}$. We show that the set of laws of $\sigma$, as the laws of $X_1,\ldots,X_n$ vary, has semialgebraic dimension \[ \sum_{k=2}^n {n \choose k}(k-1)! \] as a subset of the $(n!-1)$-simplex. This establishes a conjecture of Babson, Duchin, Iseli, Poggi-Corradini, Thurston, and Tucker-Foltz who proved that the dimension is upper bounded by the above expression. We prove their conjecture by constructing a family of finitely supported laws for $X_1,\ldots,X_n$ that provides the correct dimension. We also give an alternative proof of the upper bound using a theorem of Radford. Furthermore, we show that the Mallows law on permutations cannot arise as a law of $\sigma$ for $n\geq 4$, answering a question of Ed Crane.
Read the full abstract

Human-Computer Interaction

STcubeOperator: A Framework for Analyzing Spatiotemporal Event Data

Julius Rauscher, Lucas Joos, Mark Rohonyi et al. (+2 more)

The analysis of spatiotemporal event data is essential for informed decision-making in domains such as disaster response, conflict analysis, or intelligence investigations. However, the complexity and interdependence of spatial, temporal, and multiple thematic attributes pose significant challenges for both analysis and visualization. While space-time cubes (STCs) present a powerful integrated visualization technique to analyze this kind of data, existing approaches often lack support for complex exploratory workflows, thus limiting the ability to derive meaningful insights. We address this gap by introducing STcubeOperator, a novel framework that models analysis tasks through space-time cube operations, considering them in context of visualizations, interactions, and computational choices, and implement them in an interactive visual analytics environment. By expressing analysis tasks as a sequence of multiple elementary operations–such as filtering, chopping, and flattening–our approach enables analysts to dynamically explore data from different perspectives. We further provide an open-source prototype implementing the operations in a 3D interactive environment to facilitate task-based exploratory analysis of spatiotemporal event data. We demonstrate the applicability of our framework with a case study based on real-world data on strategic and military operations in the Russia-Ukrainian War, showing its capabilities to reveal spatiotemporal patterns. An expert user study (n=8) shows how specific tasks can be solved with our framework, highlights the versatility of our approach, and provides valuable insights on which operations experienced analysts utilize in practice.
Read the full abstract

All authors Julius Rauscher, Lucas Joos, Mark Rohonyi, Daniel A. Keim, Maximilian T. Fischer

Cryptography and Security

A Security Meta-Model for Retrieval-Augmented Generation Systems

Steve Nouyep, Sébastien Salva and Maxime Puys

Retrieval-Augmented Generation (RAG) systems extend large language models (LLMs) with external knowledge through a multi-stage pipeline. While this architecture can improve the factual grounding of generated answers, it introduces structural attack surfaces that extend beyond those of standalone LLMs. In this paper, we introduce a security meta-model that captures explicit causal relationships between RAG surfaces, attacks, weaknesses, risks, and CIA impact (Confidentiality, Integrity, Availability). Its purpose is to provide security engineers with a structured and user-friendly framework for gathering and assessing the risks, weaknesses, and mitigations relevant to their RAG deployment. We designed the meta-model through an iterative, structured analysis of 43~publications (2023–2026) and instantiated it as a catalog populated with the security threats and remediations reported in the literature. Filtering the catalog according to a deployment configuration produces a risk profile containing the risks applicable to that deployment. An interactive web visualizer lets users navigate the catalog as a graph, follow causal chains, and explore stakeholder-specific views. Analysis of the catalog revealed a persistent imbalance between attack-focused and defense-focused research, a concentration of threats at ingestion, and coverage gaps affecting output integrity. Coverage is assessed against the OWASP LLM Top~10, and operational applicability is illustrated across textual, graph-based, and multimodal RAG configurations.
Read the full abstract

Biological Physics

Femtosecond Three-Dimensional Imaging of Single-Protein with Hard X-ray Laser

Paul Lourdu Xavier, Andrew J. Morgan, Johan Bielecki et al. (+67 more)

The extreme intensity of X-ray free-electron laser (XFEL) pulses has enabled imaging of radiation-sensitive samples, such as macromolecular microcrystals, beyond radiation damage limits. These sources have the potential to enable biomolecular single-particle imaging, similar to cryo-electron microscopy but without the need for cryo-fixation and with temporal resolution from femtoseconds to milliseconds. While this possibility was recognized before XFELs were built, the biological single-particle imaging work flow has previously only been demonstrated on large virus particles. Building on improvements in X-ray beam focusing, particle delivery, diffraction detection, and advanced analysis, here we demonstrate imaging of a single molecular complex, the giant-hemoglobin erythrocruorin (Ery) with X-ray laser pulses. Two-dimensional classes of diffraction patterns could be reconstructed to 15 Angstrom resolution, and 3D images to ~20 to 30 Angstrom, even though the 3D merged intensity in reciprocal space extended beyond 20 Angstrom. The resolution discrepancy is likely due to heterogeneity caused by gas-phase compaction of the complexes. With increased throughput, this approach could be used to reveal in-situ structural details during mass spectrometry studies of biomolecules, while improvements in sample delivery may provide ultrafast snapshot imaging of biological single-particles in their native-state beyond the limitations of radiation damage.
Read the full abstract

All authors Paul Lourdu Xavier, Andrew J. Morgan, Johan Bielecki, Amit K. Samanta, Simon Dold, Chufeng Li, Xinyue Gao, Spencer K. Passmore, Ruslan P. Kurta, Tim B. Berberich, Emiliano De Santis, Wenhui Zhang, Juncheng E, Faisal H. M. Koua, Romain Letrun, Luca Gelisio, Oleksii Turkot, Egor Sobolev, Lukas V. Haas, Jannik Lübke, Jingxuan He, Madeline B. F. Memovich, Kevin Janson, Stefanie Lenzen, Safi Rafie-Zinedine, Mansi Butola, Oleksandr Yefanov, Fabian Trost, Chan Kim, Tomas Popelar, Tokushi Sato, Mohammad Vakili, Yevheniy Ovcharenko, Rebecca Boll, Sergey Usenko, Thomas M. Baumann, Alberto De Fanis, Björn Senftleben, Matteo Porro, Andrea Castoldi, Ibrahym Dourki, Cyril Danilevski, Ekaterina Round, Christina Schmidt, Huijong Han, Roberto Alvarez, Ervin Chia, Cornelia Cazey, Carolin Seuring, Ruojie Sha, Maia Azubel, David A. Bushnell, Sébastien Boutet, Sergey Ovchinnikov, Adrian P. Mancuso, Michael Meyer, Richard Bean, Anton Barty, Filipe R. N. C. Maia, Richard A. Kirian, Thomas D. Grant, N. Duane Loh, Carl Caleman, Erik G. Marklund, Nadia Zatsepin, Andrew V. Martin, Saša Bajt, Roger D. Kornberg, Jochen Küpper, Henry N. Chapman

Physics and Society

Tracking High-Impact Researchers Through Highly Cited Tails

Hirushan Sajindra and Eoin Ó Colgáin

It is well recognised that field weighted citations are approximately lognormally distributed. This skewness reflects competition for attention, whereby most publications are gradually forgotten and only a minority continue to be recalled years later. Exploiting this observation, we construct a researcher-level indicator based on authorship of top 10% cited papers in All Science Journal Code (ASJC) bins and apply it to publications affiliated with Irish institutions over a typical funding cycle. Restricting attention largely to original research papers, we find that researchers authoring strictly more than one ASJC top-decile paper in a given year exhibit persistent participation in the highly cited tail of the literature in subsequent years, thus the simple track record indicator possesses predictive value. In contrast, the median researcher in cohorts of Research Ireland and European Research Council awardees selected through qualitative peer review either authors no top-decile paper or less than the median researcher in our ranking. These results demonstrate that research excellence as judged by qualitative peer review and influence on the research literature identify substantially different cohorts of researchers. In the absence of evidence, funding bodies and universities may conflate these two. While research reputation may be enhanced through many channels, highly cited papers constitute the most widely observable manifestation of international scholarly influence. Our analysis indicates that literature influence may be improved through greater use of data in track record pre-screening prior to qualitative peer review.
Read the full abstract

Plasma Physics

Compressional heating above 1 keV on the LM26 magnetized target fusion machine

S. J. Howard, D. Krotez, P. Carle et al. (+43 more)

The LM26 device has achieved a peak electron temperature of $T_e = 1180 \pm 65$ eV as measured by filtered X-ray diodes, supported by adjacent Thomson scattering measurements of $T_e = 1090 \pm 40$ eV, with a deuterium ion temperature of $T_i = 459 \pm 33$ eV as inferred from neutron yield. LM26 compresses a spherical tokamak plasma inside an initially 1.76 m diameter solid lithium liner imploded by theta-pinch coils. Radial compression by a factor of 2.75 increases $T_e$ over fivefold, while $T_i$ increases as much as twofold. The integrated observation of magnetic flux compression, electron heating, ion heating, and neutron production in a magnetized plasma compressed by a large lithium liner is an important milestone for magnetized target fusion
Read the full abstract

All authors S. J. Howard, D. Krotez, P. Carle, J. Sanchez Rojo, R. Underwood, A. Froese, M. Reynolds, N. Sirmas, K. Conquergood, K. Epp, P. Forysinski, D. Plant, R. Zindler, E. Love, D. P. Brennan, N. Kumar, C. MacDonald, A. Wong, W. Zawalski, B. Rablah, W. Kozicki, C. Preston, A. Lee, X. Feng, M. Schellenberg-Beaver, R. Tingley, J. Gorenstein, V. Suponitsky, Z. Seifollahi Moghadam, D. Froese, E. Cessford, J. Pratt, J. Crofts, J. Sardari, G. Faust, D. Ross, J. Wilkie, S. Bernard, C. Connor, S. Bolanos, C. Gutjahr, E. Chan, C. Eyrich, A. Mahoney, M. Davidson, M. Laberge

Software Engineering

Evaluating Exact Output and Checkpoint-State Prediction in Real Programs

Xiaohong Chen, David Bucur, Chenglong Ma et al. (+4 more)

We present a benchmark for predicting final output and checkpoint state from source and input alone. It extends CRUXEval-style output prediction with paired shorter- and longer-trace inputs and checkpoints inside and after a loop. The benchmark contains 400 cases from 371 Python and C++ programs, evaluated under seven settings from four model families without tools or code execution. Of 11,200 planned predictions, 11,151 produced gradable responses. Reasoning-enabled settings outperform their off counterparts by 33.1 to 55.2 percentage points on completed responses. The strongest setting scores 93.0% on shorter-trace final output, 77.0% on longer-trace final output, and 65.5% and 63.5% on the two state tasks; these scores also hold when missing responses count as wrong. Across 2,397 matched Python comparisons with identical source, changing to the longer-trace input yields 528 correct-to-wrong changes and 147 reversals. Source-clustered analyses preserve this accuracy gap, while adjusted Python models give no evidence of a positive incremental association between cumulative state load and error. Changed inputs and checkpoint tasks alter several factors together, so the gaps do not isolate trace length or an internal state-tracking mechanism. The benchmark exposes errors hidden by short-output scores alone.
Read the full abstract

All authors Xiaohong Chen, David Bucur, Chenglong Ma, Yi Zhang, Lingming Zhang, Sriram Vishwanath, Grigore Rosu

Functional Analysis

Universal weighted sampling of positive and normal operator orbits

Jian Wu

We study common weighted sampling designs for operator systems that are exactly observable at nonnegative integer times. For positive operators with original frame condition number at most $\kappa$, we construct a common integer design with arbitrarily small relative Gramian error and $O(\log R)$ distinct times up to $R$. The optimal logarithmic counting coefficient for preserving the frame property satisfies $C^*(\kappa)\sim\pi^{-2}\log\kappa$ as $\kappa\to\infty$. Real and integer times give the same infimum, and we determine its asymptotics as $\kappa\downarrow1$. A finite bound on the sampled condition number strictly increases this coefficient. For tight frames, we determine the exact optimum under this constraint and construct an integer design attaining it. Without this constraint, the infimum is never attained. For positive operators, a design valid for every finite $\kappa$ requires counting faster than $\log R$, with arbitrarily slow divergence of the ratio. For unrestricted normal operators and fixed $\kappa>1$, at least linear counting is necessary. Under the logarithmic sector condition $|\arg\lambda|\le M(-\log|\lambda|)$ with fixed $M\ge0$, logarithmic sampling is possible with a sharp leading coefficient. For fixed Borel spectral sets satisfying a radial density condition, we characterize logarithmic sampling and determine the minimum upper counting exponent, attained by an integer design. Counterexamples show that the radial condition cannot be omitted.
Read the full abstract