Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 4 September

Sound

Emotion as a Distribution: Joint Valence-Arousal Probability Learning for Speaker-Independent Multimodal Emotion Recognition

Human emotion is graded and frequently mixed, yet most multimodal recognizers collapse it onto a single hard label. We argue the recognizer should instead expose a distribution over affective space. Our text+speech system, alongside its categorical decision, emits a $9\times9$ probability matrix over the Valence-Arousal plane, trained with a two-dimensional Gaussian soft target under a Kullback-Leibler/cross-entropy objective, aimed at counseling support. Evaluation is strict: speaker-independent 5-fold leave-one-session-out IEMOCAP with rotating-session inner validation, headline metrics only on the held-out session. Within one fixed encoder-fusion-head pipeline we compare Transformer and state-space (Mamba-1/2/3) backbones at matched depth and width, at two operating points ($T\approx550$, $T\approx2750$). The featured dual-head system reaches 73.0% $\pm$ 0.3 unweighted accuracy over three seeds (separate rerun: 72.1%), exceeding the Transformer fusion baseline by 3.0 UA points (95% session-bootstrap CI [1.0,4.7]; significant under paired t-test and session-level bootstrap), with no latency or memory advantage at these lengths; swapping the ~1M trainable front-end for frozen WavLM-Large features (learnable layer weights) lifts the same architecture to 76.6% $\pm$ 1.3. Pre-specified controls scope the claims honestly: simpler valence-arousal auxiliaries reproduce the classification lift within noise, and a dedicated regression head tracks the continuous ratings slightly better, so the head's specific value is the normalized affect distribution itself. That distribution recovers the circumplex: its center of mass tracks valence and arousal (CCC 0.66/0.66; predominantly between-class structure, weaker within-class tracking), and its entropy is weakly but consistently linked to categorical rater ambiguity, not dimensional spread.
Read the full abstract

Physics and Society

The collective dynamics of online harassment

Fringe online message boards are often studied in the context of the extreme ideology that they produce. So far, however, not much of this research has focused on direct real-world harm in the all-too-common form of collective harassment. We directly analyze the complex dynamics of KiwiFarms, an online message board dedicated largely to the harassment of individuals from vulnerable communities. We conduct exploratory analyses of the hyperlink structure of the platform and linguistic changes over time, and prospective modeling of thread size. After establishing this broader picture of the complex traits of the system, we observe the temporal evolution of community-specific vocabulary, finding that the community's framing of their harassment targets persistently evokes more danger in the early 2020s than the late 2010s. We lastly find that early thread-growth behavior is predictive of longer-term thread virality. We discuss the implications for broader understanding of toxic online behavior and threat assessment, and make the case for studying fringe platforms as complex systems with significant societal impact.
Read the full abstract

Statistics Applications

Estimating water levels in the High Plains Aquifer by synthesizing satellite data with groundwater well observations

The High Plains Aquifer (HPA) is a critical water resource in the Central United States, yet its depletion remains a major concern. While the Gravity Recovery and Climate Experiment (GRACE) satellite mission provides large-scale estimates of liquid water equivalent thickness (LWET), its coarse spatial resolution (approx. 24 km) limits local inference. In contrast, well observations from the National Ground-Water Monitoring Network (NGWMN) offer valuable but spatially sparse measurements of depth-to-groundwater. In this paper, we develop a downscaling framework for the satellite data that integrates the two sources and covariates using a Bayesian hierarchical framework. Our model uses a latent Gaussian Markov Random Field (GMRF) that describes the groundwater storage at a high spatial resolution. To address computational issues, we use a basis representation approach specified via the Moran basis. We find that fine-scale covariates like irrigation intensity and precipitation help refine spatial predictions. We thus provide, to our knowledge, the first statistically-rigorous approach for downscaling groundwater information based on GRACE satellite data and NGWMN groundwater measurements. Our approach yields high-resolution estimates of groundwater variations across the HPA from 2002–2022. The resulting fine-scale inference provides valuable insights into groundwater dynamics, highlighting the effects of land use and local extraction patterns.
Read the full abstract

Statistics Theory

A complete characterization of sequential testability and change detectability in i.i.d. models

We give a necessary and sufficient condition for the existence of power-one sequential tests in an i.i.d. composite testing problem. A level-\(\alpha\) test with power one against every alternative exists if and only if the alternatives are separated from the null by a countable family of finite-block events. We provide other equivalent conditions using randomized fixed-sample tests, bounded finite-block scores, e-processes, reduced-filtration test supermartingales, and a countable cover whose finite-block weak-$*$ closed convex hulls are positively separated in total variation. As a bonus, the constructive proof yields tests have pointwise expected sample size \(O_Q(\log(1/\alpha))\). Exactly the same conditions also characterize i.i.d.\ change detectability under optional-horizon average-run-length control: for every \(\eta>0\), they are equivalent to an alarm family \((T_\gamma)_{\gamma\ge1}\) satisfying \(\Prob_{P^\infty}(T_\gamma\le\sigma)\le \E_{P^\infty}\sigma/\gamma\) for every null law and every stopping time \(\sigma\). In fact, when these conditions hold, we can construct a single e-detector such that every null-law average run length lies between \(\gamma\) and \((1+\eta)\gamma+1\), and having robust Lorden delay \(O_Q(\log\gamma)\).
Read the full abstract

Group Theory

Every Reflection Group Is Natural

By a reflection group we mean an abstract group generated by nonidentity involutions. Knill proved that every reflection group of cardinality at most the continuum is natural, in the sense that some metric determines its group structure up to isomorphism among group structures whose right translations are isometries, and asked whether the cardinality hypothesis can be removed. We answer this question affirmatively. More precisely, every reflection group G admits a right invariant metric taking at most seven values for which the full isometry group is the right-regular action of G. The proof replaces distinct numerical labels on generators by a finite-valued encoding of a rigid graph on a well-ordered irredundant generating set.
Read the full abstract

Probability

Optimal mixing of the systematic scan dynamics via approximate tensorization of entropy

We study the mixing time of the systematic scan dynamics for high-dimensional discrete distributions. This Markov chain updates coordinates sequentially according to a fixed predetermined order, in contrast to the Glauber dynamics that updates coordinates selected uniformly at random. The systematic scan is often favored in practice because it exhibits strong empirical performance, but its theoretical analysis remains far less developed than that of Glauber dynamics. We take a step toward addressing this imbalance by showing that two standard functional notions of weak dependence between the coordinates of the distribution provide strong convergence guarantees for the systematic scan dynamics. First, we show that approximate tensorization of entropy implies optimal $O(\log n)$ mixing time for every scan order under standard marginal, connectivity, and bounded interaction degree assumptions about the distribution. Second, we show that approximate tensorization of variance yields a constant-factor contraction of the variance functional per scan, which in turn implies an optimal $O(1)$ relaxation time for the natural additive and multiplicative reversibilizations of the systematic scan dynamics. Compared with our entropy result, the variance bound improves the dependence on the maximum interaction degree from exponential to quadratic and requires weaker assumptions on the distribution. As concrete applications of our results, we establish optimal $O(\log n)$ mixing of the systematic scan dynamics for bounded-degree antiferromagnetic two-spin systems in the tree-uniqueness region and for the ferromagnetic $q$-state Potts model on square boxes in $\mathbb{Z}^2$ throughout its subcritical regime.
Read the full abstract

Artificial Intelligence

The Normalization of Deviance in AI Development

Work on the risks of artificial intelligence has focused predominantly on capability risk: the danger that systems become too powerful, too autonomous, or too misaligned with human values. Far less attention has been paid to the organizational level—to whether the institutions building these systems are themselves predisposed to drift toward failure. This paper argues that they are. Regardless of how capable AI systems become, the organizations building them face the same structural dynamics that preceded past major technological disasters. Drawing on case studies of the Space Shuttle Challenger, the Three Mile Island accident, and the Boeing 737 MAX crashes, this paper identifies the common structural mechanisms preceding each failure and maps them onto contemporary AI development. The findings suggest that existing safety infrastructure may provide less protection than it appears, as organizations can complete safety processes in full compliance and still produce catastrophic outcomes. The pre-disaster period of AI development is still underway; the purpose of this paper is to make these dynamics legible while they can still be interrupted.
Read the full abstract

All authors Emilio Barkett, Alexander Kimpton, Daniel Graham, Yusuf Kundgol

Machine Learning

A Multi-Source Ensemble Approach to Candidate Generation for Alternative Vacation Rental Property Recommendations

Alternative property recommendations play a critical role in vacation rental marketplaces, helping users discover relevant options when viewing a specific listing. However, generating high-quality candidate alternatives presents unique challenges: heterogeneous inventory, geographic constraints, rapid availability changes, and long-tail property distributions. We present a comprehensive study of candidate generation (CG) approaches for vacation rental alternatives, comparing collaborative filtering, shallow embeddings, and graph neural network (GNN) methods. Our experiments on a large-scale vacation rental platform (over 2M active properties) show that a hybrid architecture combining item-based collaborative filtering with GNN-based retrieval improves Recall@300 by 14.8% over the strongest baseline, by leveraging the complementary strengths of the two sources: collaborative filtering excels at early recall for properties with rich interaction history, while GNNs discover diverse, non-obvious alternatives and handle cold-start scenarios more effectively. As a component result, GNN-based embeddings alone substantially outperform shallow Hotel2Vec embeddings (48-68% relative recall improvement across K), motivating their inclusion in the ensemble. Crucially, we examine how CG-stage gains carry through to the downstream ranking stage, and find that a stronger candidate pool yields higher downstream ranking quality, though attributing this effect cleanly is complicated by the coupling between candidate generation and ranker training. This recall-conversion gap is an important consideration for practitioners deploying new retrieval methods in two-stage recommendation systems.
Read the full abstract

Strongly Correlated Electrons

Ultrafast electron crystallography reveals the atomic pathway of a light-driven correlated insulator-to-metal transition

Ultrafast phase transitions in correlated materials are often inferred from selected diffraction peak intensities or diffraction peak displacements, leaving the underlying three-dimensional atomic trajectories elusive. Resolving these trajectories is essential for identifying which atomic motions drive changes in electronic properties and how they couple to electronic degrees of freedom. We address this challenge in vanadium dioxide (VO2), a correlated oxide with a near-room-temperature transition between the insulating monoclinic (M1) phase and metallic rutile (R) phase. For this purpose, we introduce ultrafast three-dimensional electron diffraction, which enables refinement of the transient unit cell and internal V and O coordinates, revealing the V-V dimerization and zigzag motion during the phase transition. The refined atomic coordinates follow a linear trajectory in real space during the transition, in contrast to nonlinear or sequential transformation pathways inferred from more indirect observables in earlier work. Quantum many-body calculations treating each V-V pair as a correlated unit show that dimerization creates the level splitting responsible for the electronic gap, which is further enhanced by nonlocal electronic interactions between the paired V atoms. The gap collapses when dimerization is lost. This work turns time-resolved diffraction from order-parameter tracking into transient crystallography, directly connecting atomic trajectories to electronic mechanisms in correlated materials.
Read the full abstract

All authors Arthur Niedermayr, Hongyi Xu, Chin Shen Ong, Patrik Thunström, Michael Yannai, Jianyu Wu, Gaolong Cao, Lior Kornblum, Ido Kaminer, Oscar Grånäs, Xiaodong Zou, Jonas Weissenrieder

Algebraic Geometry

On endomorphisms of affine spaces and the Jacobian problem

Let $p$ be a prime. We provide examples which show that étale endomorphisms of affine planes over an algebraically closed field $k$ of characteristic $p$ can have fibers of arbitrary finite cardinal. Let $(l,m)\in\mathbb N\times\mathbb N^{\ast}$. We provide examples of such étale endomorphisms whose images have complements of cardinality $l$ and whose geometric degrees are $pm$. Several conjectures are disproved, and in particular we provide an analog over $k$ of Kulikov's counterexample to the `Generalized Jacobian Conjecture' of Bass for surfaces. Surjective (resp.\ surjective and non-surjective) counterexamples to Adjamagbo's Jacobian Conjecture over each $k$ are included in dimension $2$ (resp.\ in all dimensions at least $3$); for instance, if $p=2$, then we show for each $m$ there exist surjective étale endomorphisms of the affine spaces over $k$ of dimension at least $3$ of geometric degree $m$. If $e:X\rightarrow X$ is an endomorphism of a variety over an algebraically closed field $K$, then we show that there exists $n\in\mathbb N$ such that $\Imm(e^n)=\Imm(e^{n+1})$ provided either (i) $e$ is quasi-finite or (ii) $\dim(X)\le 2$ and $X\setminus\Imm(e)$ is finite. We provide examples of endomorphisms of affine planes of dimensions at least $3$ over $K$ whose images have complements of cardinality $l$ and for all $n\in\mathbb N$ we have $\Imm(e^n)\neq\Imm(e^{n+1})$. We prove that all affine moduli schemes of étale endomorphisms of affine spaces with Jacobian matrices of determinant 1 over $K$ are connected and classify all those that are smooth. We prove that the Jacobian Conjecture over $\mathbb C$ and Adjamagbo's analog of it over $k$ hold for étale endomorphisms of affine spaces that are composites $g\circ f$, where $f$ is quasi-finite and a locally closed embedding in codimension $1$ outside a specific finite subset and $g$ is a projection that omits one coordinate.
Read the full abstract

High Energy Astrophysical Phenomena

A Possible X-ray and Gamma-ray Quasi-Periodic Oscillation in GRB 241030A

Quasiperiodic oscillations (QPOs) in gamma-ray intensity have been reported from a few gamma-ray bursts. These QPOs could be related to fundamental frequencies in the sources, which would lead to new insights about these systems. Here we report on our analysis of GRB~241030A, a several-minute-long burst with $\sim 10$ gamma-ray flares that have approximate periodicities at f ~ 0.04 Hz and f ~ 0.08 Hz. This quasi-periodicity is seen in X-rays with the Swift X-Ray Telescope (XRT) as well as in gamma-rays with the Swift Burst Alert Telescope (BAT) and the Fermi Gamma-ray Burst Monitor (GBM). Compared with a power spectral model that has only red noise, the Bayes factor in favor of a Lorentzian-shaped QPO is more than $10^{200}$, even when the red noise is described by segments with up to four different slopes. However, the interpretation of this power spectral feature as a QPO rather than as a fluctuation in red noise is complicated by the low frequency of the feature as well as the extremely large number of counts, which amplifies differences from smooth power spectral distributions. After discussing the properties of the apparent QPO, we discuss a few processes that could plausibly modulate the flux at this frequency. Among the mechanisms considered, we find the most plausible explanation to be a GRB jet passing through regularly spaced shells of circumburst material, possibly produced by binary interactions near periastron in an eccentric progenitor system.
Read the full abstract

All authors Noel Klingler, Cecilia Chirenti, M. Coleman Miller, Amy Lien, Simone Dichiara

Computer Vision

Gaussian Linear Functional Manifold Method for Massive Point Cloud Data

Reconstructing continuous terrain manifolds from massive, unstructured airborne LiDAR point clouds remains challenging in complex Wildland-Urban Interface (WUI) environments, where deep neural networks require costly point-wise annotations and nonparametric surface reconstruction methods often lack structural interpretability. This paper introduces the Gaussian Linear Functional Manifold (GLFM), a physics-informed statistical framework that represents continuous surface topography using deterministic linear functional bases while modeling microscale diffuse laser backscatter as an isotropic Gaussian process. To avoid the quadratic computational cost of exact constrained maximum likelihood estimation, we develop an algebraic singular value decomposition (SVD) rank-reduction algorithm that enables linear-time parameter estimation and closed-form quadric classification. Evaluated on 35.2 km^2 of real-world aerial LiDAR data, GLFM automatically filters ground points and extracts morphological features, achieving an adjusted Rand index (ARI) of 0.9933 against field-verified ground truth and outperforming four leading baselines while maintaining an out-of-core memory footprint. The framework provides a rigorous, interpretable, and scalable foundation for large-scale point cloud analytics.
Read the full abstract

All authors Hong Zhao, Tonglin Zhang, Baijian Yang, Jin Wei-Kocsis, Songlin Fei

Computation and Language

Some Tokens Behave like Magnets: Revealing Linguistic Organization in the Layers of Language Models

We identify a special group of token vectors inside large language models (LLMs), which we term magnetic vectors, that organize the surrounding tokens by either attracting or repelling them. Particularly, tokens pointing the same way as an attracting magnet are elongated; tokens pointing the same way as a repelling magnet are compressed. Just as physical magnets pull or push away the iron filings around them, these vectors organize their surroundings through two opposing polarities. Moreover, we identify a statistically significant pattern in linguistic category where function words consistently act as repelling magnets in early layers, and we also find magnets consistently reorganize their polarities in unique ways deeper in the model. In a further case study we find this observation may unveil a deliberate, layer-wise organization in how LLMs process language. This pattern is consistent across different LLM architectures, sizes, and layer configurations. It is also causally relevant. When the LLM is fine-tuned for a downstream task, the task-functional tokens emerge as magnets. E.g., in question answering, the answer-span tokens become uniquely repelling magnets in the final layer, geometrically carving the answer out of the surrounding context. Furthermore, removing early-layer repelling magnets devastates syntactic tasks (POS tagging accuracy drops from 91% to below 10%) while sparing semantic ones, and removing late-layer attracting magnets does the reverse. We believe this phenomenon warrants further investigation, as it opens the first probe-free path to understanding how language models geometrically organize linguistic computation across their layers.
Read the full abstract

Computer Vision

SeRV: Semantic-Aligned Residual Vector Quantization for American Sign Language Generation

American Sign Language (ASL) generation remains challenging due to limited paired text-ASL motion data and the difficulty of learning motion representations both precise for reconstruction and predictable from linguistic input. Existing methods rely on motion tokenizers optimized for reconstruction, without explicit semantic supervision from paired text. As a result, the learned tokens remain limited in supporting semantically consistent and fine-grained ASL motion generation. To address this limitation, we propose SeRV (Semantic-Aligned Residual Vector Quantization), a semantic-aligned RVQ tokenizer for ASL generation. SeRV learns a semantically structured residual token space by combining global-level motion-text alignment with token-level text-conditioned supervision. Building on this tokenizer, a Hierarchical GPT predicts residual motion tokens in a coarse-to-fine manner, generating structurally coherent and semantically aligned 3D ASL motion. We further construct a large-scale reconstructed 3D ASL motion-text benchmark by recovering paired 3D motion from OpenASL videos. Experiments on How2Sign and OpenASL show that SeRV outperforms six state-of-the-art baselines, with user studies assessing reconstruction quality and semantic fidelity. Samples of our reconstructed 3D ASL motion are available on our project page: https://signvideos.github.io/Demo/.
Read the full abstract

All authors Hongyu Wu, Xu Wu, Tianhao Wu, Jiawei Yu, Phuc Nguyen, Jian Liu, Yi Wu

Systems and Control

Practical Zero-Trust for Mission-Critical Robotic Fleets via Hardware Attestation and Packet Timing Watermarking

Autonomous unmanned vehicles are vital to tactical missions, mission-critical public-safety operations like search and rescue and disaster response. However, their reliance on open wireless links and standard Robot Operating System (ROS 2) middleware exposes a broad cyber-physical attack surface. A compromise of these systems can disrupt real-time control loops, leading to mission failure or asset loss in high-stakes environments. This paper presents and empirically evaluates a layered, context-aware cybersecurity framework enforcing Zero-Trust principles for a multi-node robotic fleet over Wi-Fi. The framework integrates an active hardware root of trust (TPM 2.0), centralized in-band and out-of-band SIEM telemetry monitoring (ELK Stack and Kismet), and a non-cryptographic Inter-Packet Delay (IPD) timing watermark. Evaluated on a live ROS 2 mobile testbed under multi-layer exploits (OSI Layers 2-5), results demonstrate that while volume-based filters isolate brute denial-of-service floods, tracking the statistical sample kurtosis (K) of the embedded IPD watermark exposes stealthy Man-in-the-Middle command injections with complete detection accuracy without payload overheads.
Read the full abstract

All authors Ryne Gonzales, Ethan Liesdyanto, Rex Worley, Michael Frederick, Jaewon Kim, Eman Hammad

Multiagent Systems

SimTIO: A Simulation-Grounded Multi-Agent LLM Framework for Compositional Traffic Intervention Optimization

Traffic analysts must translate diagnosed bottlenecks into executable interventions without allowing local improvements to degrade network-wide performance. This study presents SimTIO, a simulation-grounded multi-agent large language model framework for composing and selecting traffic interventions under explicit operational constraints. SimTIO first simulates an unmodified SUMO scenario to identify a baseline-frozen set of ten bottleneck edges. A grounded sampler then initializes signal-control, corridor-speed, and demand-preserving routing actions, while three specialist agents use measured simulation feedback to select one-parameter refinements from validator-confirmed mutation catalogs. Compatible actions are combined and re-simulated so that their interaction effects are measured rather than inferred. Final selection minimizes bottleneck time loss while constraining network-wide delay, neighboring-road spillover, throughput loss, and teleport events, with the unmodified scenario retained as a no-operation guard. Across 15 cases covering five U.S. urban networks, three synthetic-demand seeds, and 2,400 origin-destination trips per scenario, SimTIO reduced Top-10 bottleneck time loss by an average of 9.18 percent and network-wide delay by 2.78 percent. It found a feasible improving plan in 86.7 percent of cases, compared with 73.3 percent for grounded random search and 80.0 percent for a deterministic heuristic under the same seven-simulation budget, although the differences in Top-10 improvement were not statistically significant. These results support using LLMs as constrained, feedback-guided local search operators while reserving final decision authority for executable tools, microscopic simulation, and explicit safety constraints.
Read the full abstract

Systems and Control

Resilient Control Loops in Autonomous Vehicles Under Adversarial Jamming via Spectral Perception and Network-Layer Failover

The operational integrity of autonomous mobile robots relies on the continuous availability of wireless control loops, making them highly attractive targets for adversarial intentional electromagnetic interference. This paper introduces a resilient, cross-layer framework that combines physical-layer spectral perception with network-layer routing optimization to protect middleware stability, such as ROS2, during intentional electromagnetic interference. Utilizing a software-defined radio front-end, the system extracts dynamic spectral descriptors, including spectral entropy and channel occupancy, to inform a Random Forest classifier that establishes adaptive environmental baselines. To ensure uninterrupted data flow, the architecture maintains dual pre-authenticated physical interfaces in a hot-standby configuration, enabling instantaneous failover through automated network routing table updates. Empirical validation on a physical ROS2 mobile robot testbed demonstrates that this adaptive hardware-assisted architecture optimizes communication recovery to an average of 141ms. This sub-second restoration translates directly into a 78.9% reduction in pooled root-mean-square path tracking error compared to software re-association, successfully securing system-level mission integrity.
Read the full abstract

All authors Luis Barajas, Colin Jeardoe, Jaewon Kim, Eman Hammad

Computer Vision

RenderFormer-V2: Neural Rendering with Heterogeneous Scene Primitives

We present 'RenderFormer-V2', a unified learned transformer-based neural rendering model, complementary to modern physics-based rendering systems, that can handle diverse light-transport effects such as caustics, volumetric scattering, environment lighting, textured and displaced surfaces and out-of-distribution materials without per-scene training or specialized code. RenderFormer-V2 models global light transport as a sequence-to-sequence transformation. Following its predecessor, RenderFormer-V2 also employs a two stage process: a view-independent stage that resolves intra-scene primitive to primitive transport, and a view-dependent stage that transforms the internal neural scene representation into image pixels. Different from RenderFormer, our model employs a novel combined windowed-attention and rendering-informed attention sink in the view-independent stage to improve scalability while maintaining render accuracy. To further improve versatility, RenderFormerV2 supports heterogeneous scene primitives, including environment maps and participating media, and it employs a material encoding independent of the underlying surface reflectance model that encodes material appearance via a novel neural embedding. We demonstrate the versatility of RenderFormer-V2 on a variety of scenes and perform an extensive ablation of the improved attention mechanism.
Read the full abstract

All authors Chong Zeng, Yue Dong, Pieter Peers, Lvmin Zhang, Maneesh Agrawala

Quantum Physics

SAQC: A SAT-Aware Compilation Framework for QAOA-Based Quantum Optimization

The Quantum Approximate Optimization Algorithm (QAOA) is a promising variational approach for solving Boolean satisfiability (SAT) problems. Existing SAT-to-QAOA workflows first translate Boolean formulas into penalty Hamiltonians before applying generic quantum compilation, thereby discarding SAT-specific information such as clause structure, logical operators, and variable dependencies. Consequently, subsequent compiler optimizations cannot exploit the underlying Boolean formulation. This paper presents SAQC, a SAT-aware compilation framework that introduces a SAT Intermediate Representation (SIR) to preserve Boolean structure during compilation. The proposed framework supports both CNF and XOR-extended CNF (eCNF) formulations and performs SAT-aware optimizations including clause rewriting, dependency analysis, and clause scheduling prior to quantum lowering. Building on the optimized SIR, SAQC provides a unified framework for both penalty Hamiltonian generation and direct clause-to-ansatz synthesis, together with ansatz optimizations based on relative-phase decomposition, ancilla-assisted synthesis, and dynamic uncomputation. Experimental results demonstrate significant reductions in circuit depth, two-qubit gate count, and compilation time compared with conventional Hamiltonian-based workflows while remaining compatible with existing quantum compilation frameworks.
Read the full abstract

All authors Till Schnittka, Julie Maria Raju, Abhoy Kole, Rolf Drechsler

Artificial Intelligence

Beyond Prompts: Measuring and Optimizing LLM Tool-Agent Harnesses

LLM tool agents can be improved without retraining by modifying the runtime harness around a fixed model: prompts, tool interfaces, middleware, state handling, and recovery logic. We study this setting as resource-bounded harness selection for fixed-model multi-turn tool agents, with the search surface scoped to prompts and tool-boundary middleware: edits are guarded intercepts at the tool boundary, not arbitrary rewriting of agent execution logic. Our optimizer-agnostic protocol reports mean held-out lift, worst-condition lift, repeatability, logged cost diagnostics, and RelLift95(B), a conservative estimate of the held-out gain of the harness selected under budget B. We instantiate the protocol with prompt-only and prompt-plus-middleware optimizers, including PRISM, which clusters failures and routes repairs to prompt, tool-boundary middleware, or joint edit surfaces within a Pareto search. On BFCL multi-round, tau2-Retail, and tau2-Telecom, PRISM obtains mean held-out lifts of 14.2, 14.9, and 10.1 percentage points and positive empirical RelLift95 on all three benchmarks, and a component ablation attributes the margin chiefly to failure-surface routing and the edit-pattern constraint. Across optimizers, the results show that some search procedures can occasionally find large gains but still choose brittle updates, so the reliability of the chosen harness should be reported alongside average held-out lift.
Read the full abstract

All authors Cen Mia Zhao, Haibo Ruan, Wenjie Chen, Pei-fen Tu, Usman Abbasi, Joel Hesch

Functional Analysis

Can one feel the existence of a non-trivial invariant measure?

It is shown that a bounded linear map on a complex separable Hilbert space with non-trivial invariant probability measures doesn't have to possess eigenvalues. This resolves a question indicated by Flytzanis in 1995 and concretely asked by Grivaux–López-Martínez from 2023 resp. Grivaux–Matheron–Menet from 2021. Still, as a positive result, we prove that every bounded linear operator on a separable complex Banach space for which a non-fixing invariant probability measure exists, has to have some approximate point spectrum on the unit circle minus $1$.
Read the full abstract

Optics

Phase-aware Inverse Design for Silicon Photonic Logic Gates

Inverse design has emerged as a powerful strategy for realizing compact photonic devices; however, its application to logic operations remains constrained by challenges in physical interpretability, architectural generality, and experimental validation. This work introduces an experimentally validated, unified, and physics-driven inverse design framework that implements all fundamental Boolean logic gates within a single silicon photonic architecture. Devices are fabricated within a 2 x 2 um^2 design region on a silicon-on-insulator platform, representing one of the smallest areas reported for photonic logic elements. By integrating amplitude, phase, and energy conservation into a composite figure of merit, the proposed approach enables direct control over constructive and destructive interference. Consequently, all logic functions, including XOR and three-input NAND/NOR operations, are achieved using a standardized configuration with two logical inputs and a bias port. Experimental results exhibit good agreement with numerical simulations across the C-band, confirming both the predictive accuracy and fabrication robustness of the method. Performance benchmarking reveals competitive contrast ratios compared to previous implementations, while providing a unified and scalable design strategy. These findings establish a physically interpretable and experimentally validated paradigm for inverse-designed photonic logic, advancing the development of compact, integrated optical computing systems.
Read the full abstract

All authors Ary Portes, Narges Dalvand, Felipe M. F. Teixeira, Omar P Vilela Neto, Julian L. Pita Ruiz, Michael Menard, Jhonattan C. Ramirez

Differential Geometry

The topology of tridiagonal isospectral sets

We describe the topology of the set of tridiagonal symmetric real matrices with fixed non-simple spectrum. The tridiagonal isospectral set is not a submanifold of the corresponding orthogonal conjugacy class, but a collection of submanifolds which intersect cleanly. We show that the tridiagonal isospectral set is an aspherical space and that its Betti numbers are combinatorial numbers. This generalizes to arbitrary spectrum constructions and results known for simple spectrum. We also extend these results to split real semisimple Lie algebras.
Read the full abstract

Metric Geometry

Continuity of the normal cycle with respect to $C^0$ topologies

We show that the generalized curvatures and more specifically the normal cycle of a compact subset of $\mathbb{R}^d$ are continuous under weak notions of convergence, assuming an a priori mass bound. In particular, provided its mass remains bounded, the normal cycle behaves continuously when the subset is perturbed by a homeomorphism that is $C^0$-close to the identity. Our approach relies on a combination of persistent homology with the geometric measure theory framework classically used in the study of curvatures of singular sets. As an application, we prove that every compact definable set in an o-minimal structure admits a normal cycle by showing that any such set is a limit, in the above sense, of a family of smooth sets whose normal cycles have uniformly bounded mass. We also show that WDC sets are limits of nested smooth sets with uniformly bounded normal cycle masses.
Read the full abstract

Computers and Society

From Voice to Leverage in Data Governance. Building Enforceable Social License for Data ReUse

Contemporary data governance has converged on participation. Citizens assemblies, deliberative polls, codesign workshops, and community advisory panels are now routinely suggested to build trust and secure a social license for the reuse of data. Yet this consensus rests on an unexamined assumption: that giving affected communities a voice will, on its own, change what data holders do. This paper argues that the field needs to complement voice with leverage: the capacity of individuals and communities to make their participation consequential by attaching credible legal, institutional, contractual, economic, or reputational costs to being ignored. Drawing on labor relations bargaining theory, the sociology of disputes, and James C. Scott's account of legibility, we develop leverage as an analytic category defined by three properties: it is counterfactual, enforceable, and durable. We further argue that leverage presupposes epistemic legibility affected publics can rarely impose costs on a system they cannot see, document, or contest; and that most existing transparency instruments equip regulators and procurers rather than the governed. We then map six sources from which communities and data subjects can derive leverage (legal and regulatory; institutional decision; contractual; economic; data access; and political and reputational), examine the intermediary structures that aggregate and sustain it, and propose a six dimension evaluative framework: obligation, enforceability, consequence, monitoring equipment, revocability, and institutional durability for auditing whether any social license process actually redistributes power. The paper's central claim is that a social license worth the name must share the properties of a legal license: conditions, a licensor, and enforceable remedies for breach.
Read the full abstract

All authors Stefaan G. Verhulst, Adam Zable, Begona Gonzalez Otero, Jacqueline Lu

Computer Vision

Bigger Text Encoders Can Hurt CLIP Zero-Shot Performance

Contrastive Language-Image Pretraining (CLIP) is a building block of many machine learning applications. Scaling laws have guided resource allocation for large-scale training, yet prior work treats total CLIP model size as a single variable, without exploring how the capacity split between encoders impacts downstream performance. Here, we train multiple CLIP models with different vision and text encoder sizes, revealing that for most vision encoders, there is an optimal text encoder size beyond which zero-shot performance degrades—even as total parameter count increases. Exploiting this behavior yields efficient configurations that match the zero-shot performance of the standard ViT-B/16 architecture with up to 55% fewer parameters. We further show that this degradation stems from overfitting induced by the oversized text encoder, and that using modality-specific weight decay coefficients not only recovers but improves performance across all degraded configurations. A geometric analysis reveals a trade-off in which scaling the text encoder improves embedding uniformity but worsens cross-modal alignment; we further show that these metrics are predictive of zero-shot performance. We hope these findings motivate CLIP architectures and training methods that counteract this degradation, a prerequisite for scaling CLIP reliably and efficiently.
Read the full abstract

Networking

SDN-Orchestrated Dual-Path 5G/SATCOM Maritime Communications for Carrier Strike Groups

Carrier Strike Group (CSG) communications must sustain mission traffic across heterogeneous links whose quality varies with distance, weather, fading, and intermittent outages. We present an integrated Software-Defined Networking (SDN)/Software-Defined Radio (SDR) framework that unifies 5G and satellite communications (SATCOM) under a single control plane for dual-path failover in CSG communications. The framework is evaluated via stochastic simulation across four operational regions: Norfolk, the Norwegian Sea, the Philippine Sea, and the North Pacific storm track. The framework couples a centralized SDN controller with a Rician-faded 5G channel under Adaptive Modulation and Coding (AMC), a Ku-band SATCOM channel with ITU-R P.618/P.838 rain attenuation, and Gilbert-Elliott burst-availability processes on both links. The controller routes each packet by traffic class and current link state. Experiments use 60 independent trials of 50,000 packets per parameter setting with common random numbers, paired confidence intervals, and convergence diagnostics across fault, fading, and region sweeps. For the Norfolk weather baseline, SDN routing reaches 93.09% reliability versus 36.22% (5G-only) and 89.87% (SATCOM-only), with paired gains of +56.87 and +3.22 percentage points. The advantage grows under joint low-Rician-K/high-burst-fault stress, and AMC improves 5G robustness as Signal-to-Interference-plus-Noise Ratio degrades. SDN remains the highest-reliability mode across all four regions, indicating that cross-layer SDN-controlled failover improves maritime resilience without relying on deterministic link abstractions.
Read the full abstract

All authors Avinash Srinivasan, Dalibor Španjević, Kevin H. Nguyen, Bannon Ireton, Christopher B. Landis

Computation and Language

MedWER: A Reproducible, Model-Free Evaluation Protocol for Medical Speech Recognition

Overall word error rate hides clinically critical errors: a transcript can be 95% correct and still swap one drug for another. The usual fix weights errors on medical entities, and almost always depends on an evaluation-time named-entity recognition (NER) model or cloud API, which makes the metric's denominator a versioned black box. We present MedWER, an evaluation protocol and open-source tool for medical ASR whose denominator is a fixed, license-clean term list: 19,373 drug, diagnosis, symptom, and injury-mechanism entries projected from public sources. The protocol couples a pinned text normalizer with a phrase-aware term-restricted WER, the MedWER, so the only versioned component is a normalizer dependency held at an exact release and checked against committed golden fixtures. Coverage is validated against an independent provincial drug-benefit file the list was not built from; the matching heuristic is calibrated against ground-truth entity spans. Baselines for Moonshine~base, Whisper~base.en, and MedASR on two open benchmarks are scored with the released tool and reported with 95% confidence intervals from resampled per-utterance scores.
Read the full abstract

Machine Learning

Newton Matching for Generative Modeling: A Unified Framework for Fine-Tuning and Sampling

We develop Newton Matching, a unified framework for fine-tuning and sampling in generative modeling. The target is $\pi\propto\mu e^{\tau r}$, where $r$ is the reward, $\tau>0$ the inverse temperature, and $\mu$ denotes the pretrained model's terminal density for fine-tuning or the constant $1$ for sampling. We shift the paradigm from isolated losses to iterative optimization over canonical models: population minimizers of standard conditional matching for terminal densities. Under compatible smooth-realization assumptions, canonical velocities form a manifold diffeomorphic to the density manifold. Transporting the Fisher-Rao metric and mixture connection to this manifold, we show that the reverse-KL Hessian equals the metric, so the Newton direction coincides with the negative Fisher-Rao gradient. At terminal density $\rho$, each stage takes a tangential step generated by the regularized reward $r-\frac1\tau\log(\rho/\mu)$, followed by terminal-density-preserving canonicalization. This canonical retraction yields an exact finite-stepsize density characterization. For the ideal iteration, we prove strict reverse-KL descent away from the target for $0 < \eta \le \tau$, global convergence under mild conditions, and local quadratic convergence for full steps ($\eta=\tau$). Covariance and gradient forms, each with forward or reverse regression-pair constructions, yield sample-wise tangential-update losses with the same population minimizer, without importance sampling or full-trajectory backpropagation. We develop approximate updates and define critical-point consistency as vanishing tangential displacement if and only if $\rho=\pi$. We recover representative methods as exact realizations, critical-point-consistent approximations, or objective-altering variants, enabling modular algorithm design. Our work advances the theory and algorithms of reinforcement learning for generative models.
Read the full abstract

All authors Zeyang Li, Yunan Wang, Paolo Giaretta, Navid Azizan

Functional Analysis

A class of port-Hamiltonian systems with general time delays

We define a class of infinite dimensional linear port-Hamiltonian delay systems, give a characterization for such systems, and provide an easily verifiable sufficient condition to belong to this class, thereby extending the study by Breiten, Hinsen and Unger in [IEEE Trans. Autom. Control vol. 69, no. 12 (2024), pp. 8924-8930, doi: 10.1109/TAC.2024.3464332]. The results rely on the operator semigroup theoretic treatment of linear delay PDEs.
Read the full abstract

All authors Rogelio Arancibia Bustos, Bálint Farkas, Birgit Jacob, Merlin Schmitz