Peer Revenue Publish or profit

Papers

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

All of arXiv

Thursday 3 September

Differential Geometry

Rigidity of Minimal Submanifolds in Spheres of Higher Codimension

We investigate Chern's conjecture for minimal submanifolds of higher codimension in the unit sphere. We establish two integral inequalities for closed minimal submanifolds. Then we obtain a new rigidity theorem under a pinching condition on the squared norm of the second fundamental form.
Read the full abstract

Computer Vision

FAVE: Foveated Adaptive Visual Encoding for Efficient Fine-Grained Visual Understanding

Fine-grained visual understanding depends on local detail, yet visual encoders face a trade-off between costly full-image high-resolution processing and compact global encoding that can weaken such evidence. Inspired by human active vision, we separate where to look from what to encode. We focus on the latter and introduce FAVE (Foveated Adaptive Visual Encoding), a lightweight variable-resolution ViT that encodes externally selected regions at high acuity while preserving native geometry. We first isolate this encoding problem using oracle ground-truth crops in a controlled small-object regime. On ImageNet objects with a native maximum side of 96 pixels, FAVE improves Top-1 by 9.4 points over a fixed-resolution ViT on the same crop window with 12.7 times lower FLOPs. Increasing global resolution or backbone capacity does not recover the same operating point. We then integrate FAVE as a complementary local branch in FastVLM. Its local tokens are combined with FastVLM's global visual tokens, while the original global pathway and language model remain frozen. With at most 16 additional local tokens, FAVE improves TextVQA by 1.60 points and achieves a 3.3 times controlled TTFT speedup over SmolVLM2-2.2B. On GQA attribute questions, it improves FastVLM-1.5B by 1.31 points, extending the benefit beyond text while narrowing the gap to FastVLM-7B. Together, these results show that selectively allocating high-acuity local capacity provides an efficient complement to broader global representations and model scaling for fine-grained understanding of small objects, text, and attributes.
Read the full abstract

Computation and Language

Evaluation of Phonetic Encoding Algorithms on Transcription Datasets

In this work, a novel evaluation scheme built on a generalized variant of the Rand Index measure, namely, the Hüllermeier-Rifqi Index, is proposed in order to assess how well phonetic encoding algorithms conform to word-based transcriptions in IPA (International Phonetic Alphabet) notation. For this objective, the discordance score is obtained by calculating the absolute difference between the pairwise similarity values of ground-truth transcriptions and those of corresponding phonetic encodings, which are computed using normalized edit distance as a permutation dependent string metric. The resulting score is subsequently adjusted with respect to that of a random string generator incorporating the same alphabet as the encoder under consideration. A wide range of phonetic encoders were evaluated as such on multi-lingual transcription datasets along with their recall capabilities based on the collision rate. The validity of the proposed scheme is further supported by its applicability in measuring the orthographic transparency of a language when the writing system is viewed as an inherent phonetic representation.
Read the full abstract

Image and Video Processing

Development and Evaluation of Ultrasound Image Learning Pipelines for MASLD Risk Stratification

Metabolic dysfunction-associated steatotic liver disease (MASLD) affects approximately 30% of the general population. Ultrasound-based imaging, including B-mode imaging and shear wave elastography (SWE), is widely used for noninvasive fibrosis assessment; however, the role of deep learning-based ultrasound image learning for MASLD risk stratification remains insufficiently characterized. In this study, we developed and evaluated ultrasound image learning pipelines using B-mode and SWE images for fibrosis staging and identification of patients with at-risk metabolic dysfunction-associated steatohepatitis (MASH). A total of 250 ultrasound examinations, one exam per subject, were included. Model performance was evaluated using 3-fold cross-validation with area under the receiver operating characteristic curve (AUROC). End-to-end SWE image learning achieved performance comparable to operator-guided SWE across fibrosis stages. Overall, SWE-based learning consistently outperformed B-mode image learning in fibrosis staging, with AUROC improvements from 0.64 (95%CI: [0.56, 0.72]) to 0.72 (95% CI: [0.65, 0.79]) for F>=2 (significant fibrosis, p=0.11), from 0.67 (95%CI: [0.58, 0.75]) to 0.78 (95% CI:[0.72, 0.85]) for F>=3 (advanced fibrosis, p=0.02), and from 0.69 (95%CI: [0.56, 0.82]) to 0.80 (95%CI: [0.72, 0.89]) for F4 (cirrhosis, p=0.10). These findings highlight the potential of SWE image learning for MASLD risk stratification.
Read the full abstract

All authors Guangyi Zhang, Xiaohong Wang, Eugene Cheah, Peng Guo, Brian A. Telfer, Theodore T. Pierce, Anthony E. Samir

Data Structures and Algorithms

Multiobjective Hypergraph Min-Cut in Quasi-Polynomial Time

We study the multiobjective hypergraph min-cut problem: Given a hypergraph $H=(V,E)$ and $k$ cost functions $c_1, c_2, \ldots, c_k:E\to\mathbb{Z}_{\ge 0}$, the goal is to find a non-empty proper subset $U\subsetneq V$ of vertices with minimum $\max_{i\in [k]} c_i(\delta(U))$. When $k$ is part of input, the problem is NP-hard (even in graphs). We focus on fixed-constant $k$ setting (e.g., $k=1, 2, 3, \ldots$). Single-objective hypergraph min-cut as well as multiobjective graph min-cut for a constant number of objectives admit polynomial-time algorithms. In contrast to these special cases, the complexity of multiobjective hypergraph min-cut remains open even for $k=2$. Known techniques fail to extend due to structural differences between graphs and hypergraphs. For $k$-objective hypergraph min-cut when $k$ is a fixed constant, we design a randomized PTAS, and two different randomized quasi-polynomial time algorithms. As an application of our $2$-objective hypergraph min-cut results, we obtain a quasi-polynomial time approximation scheme (QPTAS) for hypergraph connectivity interdiction. AI tools were used to iterate and refine the algorithmic ideas underlying this work.
Read the full abstract

Cryptography and Security

Candidate Comparability Before Promotion: Conditional Validation in Adaptive Network Intrusion Detection

Adaptive network intrusion detection systems retrain classifiers after drift alarms, but an alarm detects change; it does not establish that a challenger should replace the deployed incumbent. Promotion is security-relevant because it changes the model responsible for subsequent attack detection, and evaluating it has a methodological problem: promotion conclusions may depend on how the challenger was constructed and on how much evidence supports it. We test that dependence on CICIDS2017, UNSW-NB15 and ToN-IoT with self-contained challenger pipelines, nested candidate-size controls, a common-harness comparison of nine update policies, and a final sensitivity confining every exact feature vector to one evaluation, training or probe role. Incumbent-owned frozen preprocessing amplified apparent promotion harm; with self-contained challenger pipelines the mean full-drift harm did not persist. Raising nominal candidate evidence from 512 to 2,000 samples per class improved promotion under pool-constructed progressive drift by +0.53, +1.67 and +0.38 balanced-accuracy points: positive and statistically resolved in all three benchmarks, but materially benchmark-dependent rather than homogeneous, and driven mainly by fewer false positives. Policy conclusions were partially robust: policy ordering changed with candidate comparability, no policy globally dominated, and earlier compatibility statements for a label-free estimator and a calibrated ensemble narrowed. Validation helped evidence-disadvantaged challengers but added no average benefit at parity. Thirteen replays on real, time-ordered traffic showed no net harm from always deploying. Challenger construction and evidence should be controlled, reported and interpreted explicitly when promotion is evaluated.
Read the full abstract

All authors Roberto Fernández-Barrios, Iker Pastor-López, Amaia Pikatza-Huerga, Pablo García Bringas

Quantum Physics

Toward Resilient Many-Body Formulations under Incomplete Correlation Models: A Dual-Space Variational Formulation

Quantum simulations of correlated many-body systems will inevitably operate with imperfect information: the prepared states may arise from incomplete or approximate ansatze, while their measured Hamiltonian and overlap matrix elements are additionally affected by hardware and statistical noise. Robust quantum algorithms will be those that can improve upon reliable lower-level descriptions in spite of these imperfections. We introduce a dual-space nonorthogonal configuration-interaction (DS-NOCI) framework built around this principle. Rather than replacing a classically accessible NOCI reference manifold with correlated quantum states, DS-NOCI retains both spaces and couples them variationally. The reference sector provides a stable many-body backbone, while the quantum states contribute whatever additional correlation directions they contain. With exact matrix elements, the enlarged dual space gives a ground-state energy no higher than either the reference-only NOCI or correlated-only quantum-NOCI spaces, ensuring that an incomplete correlation model cannot degrade the underlying variational description. The additional reference–correlated matrix elements require only one correlated state preparation and are therefore less demanding than the correlated–correlated block already required by quantum variant of NOCI. Molecular benchmarks further show enhanced resilience to imperfect amplitudes and noisy matrix elements. DS-NOCI thus provides a general strategy for building quantum many-body formulations that remain useful when both correlation models and quantum computations are necessarily incomplete and noisy.
Read the full abstract

All authors Vibin Abraham, Bhumika Jayee, Bo Peng, Nicholas P. Bauman, Karol Kowalski

High Energy Physics (Phenomenology)

Earth-scale searches for displaced vertices: KM3NeT meets the LHC

At high-energy colliders, enormous numbers of feebly interacting particles beyond the Standard Model can be produced with negligible missing-energy signatures in the main detectors. These particles can travel macroscopic distances and decay inside a distant large-volume detector, leaving visible signals. In this sense, we may not need new detectors to search for very long-lived particles. In this paper, we show that long-lived dark particles produced in rare meson decays in proton-proton collisions during LHC Run 3 and at the HL-LHC can leave observable imprints in KM3NeT detectors such as ORCA. If no anomalous signal has been observed in ORCA to date, our projected sensitivity for LHC Run 3 will correspond to one of the strongest limits on kaon decays into light long-lived particles.
Read the full abstract

Astrophysics of Galaxies

Linking orbital history to the quenching of isolated dwarf galaxies

Recent discoveries of isolated dwarf galaxies (${M}_{\star}\sim 10^{7-9}~{\rm M}_{\odot}$) with no signs of ongoing star formation have challenged the prevailing notion that these solitary systems are exclusively star-forming. Investigating the origins of isolated quenched dwarfs provides important clues into the dominant processes driving galaxy quenching as a function of environment and stellar mass. In this study, we identify central dwarf galaxies in the Illustris-TNG50 cosmological simulation with ${M}_{\star} \sim 10^{7-9}~{\rm M}_{\odot}$, evaluate their star formation and orbital histories, and investigate how their dark matter halo properties correlate with quenching and interactions with massive neighbors. We find that $R_{\rm{peri}}$, defined as the closest separation ever attained between a central dwarf and a galaxy more massive than ${M}_{\star}> 10^{10}~{\rm M}_{\odot}$, is a reliable metric for inferring the star formation history of central dwarfs in TNG50. We also show that isolated quenched dwarf galaxies separate into backsplash and non-backsplash subpopulations as a function of $R_{\rm{peri}}$, and that non-backsplash galaxies are consistent with being quenched by cosmic web stripping or internal feedback mechanisms. Furthermore, we find that $R_{\rm{peri}}$ correlates strongly with dark matter halo mass and star formation proxies based on halo maximum circular velocity, which are commonly used in empirical galaxy-halo connection models. These predictions can be tested with data from upcoming deep and wide surveys (e.g., Rubin LSST and Roman) that are anticipated to significantly increase the cosmic census of dwarf galaxies.
Read the full abstract

Computation and Language

You Really Didn't Get That? Benchmarking Social Pragmatic Inference for Indirect and Playful Chinese Online Comments

Chinese online comments often convey social meaning through indirect and playful language that is hard to interpret without context. Existing evaluations largely organize items around predefined phenomena or controlled pragmatic categories, leaving open whether models can distinguish plausible readings of what a naturally occurring comment is doing in a particular exchange. We introduce a benchmark for evaluating whether LLMs can recover such situated pragmatic meanings. From more than 200,000 public Chinese social media interaction records, we construct 4,735 human-validated diagnostic items, each pairing a target comment with reconstructed preceding context and plausible misreadings. We evaluate eight LLMs as both question writers and solvers in a cross-writer setting. The task is challenging: the strongest model achieves 81.42% leave-writer-out accuracy. Across all eight models, the mean leave-writer-out accuracy is 68.70% while human accuracy was 90.8%. Case analysis shows that models often recognize broad irony or playfulness while misidentifying the mechanism or interactional move.
Read the full abstract

All authors Shiwei Hong, Junjie Ma, Emma Jiren Wang, Ethan Z. Rong, Siying Hu, Haichang Li, Ziying Wang, Zhicong Lu

Computer Vision

What Moves? Localized Motion Representations for Compositional Scene Control

Real-world dynamics are inherently compositional: multiple entities move simultaneously within a shared scene, each exhibiting distinct motion patterns. Yet current motion representation models entangle the dynamics of different entities, without explicitly capturing localized motion for each individually. Crucially, motion is defined relative to a global reference frame, including camera motion and scene layout. However, localized embeddings are often computed from cropped images or obtained by masking features after encoding, discarding the context needed to interpret motion. To address this, we introduce a promptable localized motion representation that produces persistent embeddings for user-specified regions defined by spatial masks. Rather than cropping the input or masking features, our model processes the full video and conditions motion encoding directly on the queried region. This yields temporally consistent, region-addressable embeddings that isolate local dynamics while retaining the global context required for disambiguation. We demonstrate object-level motion transfer, enabling controlled composition of dynamic scenes. Beyond generative control, our embeddings support localized action classification in multi-actor videos. Across both tasks, our approach improves controllability and outperforms global representations localized through cropping or post-hoc masking.
Read the full abstract

All authors Frank Fundel, Malek Ben Alaya, Thomas Ressler-Antal, Stefan Andreas Baumann, Björn Ommer

Cryptography and Security

Privacy Failure in Split-LLM Training, The Returned Gradient Nullifies the Decoys

We present a systems-security case study of a two-node split-LLM training system whose privacy evaluation passed while leaving an observable channel untested. The Trusted Local Node (TLN) sends protected activations to the Untrusted Cloud Node (UCN), the UCN returns its output, and TLN, holding the private loss, returns the output gradient. The frame the UCN receives mixes real rows with decoys, and the loss ignores the decoys. Their gradients are exactly zero, so the pattern of zeros reveals which rows were real. We measure it with a protocol fixed in advance: a leak injected at known strength to prove the instrument can see one, a shuffled-label control to prove it does not report absent leaks, and a threshold set before the runs. Across nine seeds, the zeros identified the real rows on every frame, 4,096 of 4,096 per run. An attack on the frame contents recovered about one extra token per hundred over a constant-guess baseline (+0.65 to +1.50 percentage points); the shuffled controls recovered nothing. A second set of runs repeated this on a configuration that keeps model quality within budget, so the finding is not confined to a setting nobody would deploy. On both datasets, every such run passed the forward-channel privacy check and the quality check, yet failed that same check once the returned gradient was included. Clipping and noising each row of the gradient closed the leak for about 0.01 nats of held-out cross-entropy. The system is not thereby safe: five classes of attack, including those accumulating observations across training steps, were never measured.
Read the full abstract

Computer Vision

What Do Scan-Derived Class Prototypes Add? Disentangling Supervision, Prototype Content and Query Protocol in Recognition over Frozen Foundation Features

A scan supplies labeled images and a geometric reference. We separate their contributions in a recognizer whose scan-derived prototype matrix acts as a supervised head's fixed output layer. On T-LESS, HOPE and 18 self-collected industrial parts, we test real, random and exactly permuted prototypes, matched geometry-free classifiers, stronger appearance rules and paired background protocols. Across DINOv2-giant and MetaCLIP-H with real-background queries, the largest fused-accuracy advantage of the real prototypes over either control is one percentage point; larger differences favor controls, by up to 2.8 points in arm means. On HOPE with DINOv2-giant the head alone is 2.8 points above exact permutations (95% interval: 0.8-4.7); this advantage does not reach fusion and is not observed on MetaCLIP-H. On DINOv2-giant, matched logistic regression comes within 0.5 points of fusion on T-LESS and exceeds it on HOPE and the self-collected parts. Against white cutouts, real HOPE query backgrounds lower image-prototype accuracy by 43 points on DINOv2-giant and 13 on MetaCLIP-H. The audit separates prototype content, label supervision and query protocol.
Read the full abstract

Soft Condensed Matter

Speed-quality tradeoff in a physical model of molecular sorting

Eukaryotic cells rely on membrane-mediated processes to compartmentalize biomolecules, counteracting diffusion-driven homogenization. These processes involve the selective sorting and packing of molecules into lipid vesicles, which are then dispatched to appropriate intracellular destinations. Previous works introduced an abstract statistical physics framework for studying this molecular distillation process, where the membrane was treated as static and the focus was on molecular aggregation and extraction. Here, we extend this framework to explicitly incorporate dynamic membrane behavior, including bending, curvature generation, and changes in membrane size due to vesicle fusion and fission events. Sorting domains drive membrane curvature, leading to vesicle formation, detachment, and a molecular distillation process that alters membrane size. Using mesoscopic modeling and numerical tools, we investigate the resulting interplay between membrane mechanics and molecular sorting. We determine a well-defined parameter region where vesicle fission and efficient molecular sorting occur, controlled by membrane rigidity, spontaneous curvature, and pressure difference across the membrane. We further identify a trade-off between speed and quality of molecular distillation: parameters that accelerate vesicle formation tend to reduce the quality of distillation, and vice versa. We propose maximization of the rate of negative entropy production as a natural criterion to optimally balance these competing effects. In this context, optimal parameters emerge naturally from the coupled dynamics of molecular aggregation and membrane mechanics, suggesting a possible strategy for cellular sorting systems to strike an efficient balance between rapid vesicle formation and high sorting quality.
Read the full abstract

All authors Elisa Floris, Damiano Andreghetti, Thibaut Caplet, Luca Dall'Asta, Andrea Gamba

Machine Learning

On the Abundance of Critical Points of the t-SNE Energy

This paper considers the energy landscape of the t-SNE algorithm. While this algorithm has enjoyed broad adoption, the non-convexity of the associated energy has made it difficult to rigorously understand what the algorithm captures in many settings. In particular, a number of well-known numerical examples, several of which are reproduced in this article, suggest a complicated energy landscape with many local minimizers that do not respect the topology or clustering structure of the underlying data. This work seeks to provide first steps towards a rigorous explanation of these phenomena. Specifically, for a general family of energies, which include both the original t-SNE algorithm and recently identified large data limits, and for densities in feature space which obey a continuous symmetry, we construct infinite families of distinct critical points. These critical points are based upon identifying pairs of discrete symmetries, one in the original feature space and the other in the target embedding space, which are preserved under gradient dynamics. These critical configurations exhibit many characteristics, such as topology breaking and spurious clustering, which are often observed empirically. Finally, numerical and analytical examples are given throughout as a means of illustrating the approach.
Read the full abstract

Statistical Mechanics

Fukui-Kawakami chains: spectrum and hidden $\mathfrak{gl}(1|1)$-symmetry

Fukui and Kawakami showed that the trigonometric Haldane–Shastry (HS) spin chain can be deformed by introducing twisted boundary conditions. In this work we revisit the resulting twisted HS chains, which we call Fukui–Kawakami (FK) chains. We analyse their spectra in dependence of the twist parameter, utilising a direct connection with the (untwisted) HS chain. This allows us to explain why part of the spectrum can be described by Fukui and Kawakami's twisted Bethe equations, and can be constructed from Yangian highest weight states. These states can be labelled by combinatorial data called `motifs', as in the HS chain, which cover some of the (deformed) descendants. We furthermore show that there are other descendants which do not follow from these Bethe equations, but whose energy can be described as a sum of two single-particle energies, suggesting additional hidden symmetries. We then focus on the special case of antiperiodic boundary conditions, and show that this chain coincides with the `minimally polarised' long-range model recently introduced by Basu-Mallick, Finkel, and González-López. Remarkably, this connection implies that the antiperiodic chain has a hidden $\mathfrak{gl}(1|1)$-symmetry, which we use to relate the HS motifs to their `supersymmetric' $\mathfrak{gl}(1|1)$ counterparts.
Read the full abstract

Artificial Intelligence

Corporate Language Model (CLM): Transforming Tacit and Fragmented Enterprise Knowledge into a Sovereign, Auditable, and Executable Corporate Intelligence Layer

Enterprise AI deployments fail not from model inadequacy, but because organizations lack a structured substrate encoding how they decide, negotiate, and execute. Generic LLMs carry no firm-specific ontological priors; RAG remains brittle, with no path to executable action; static playbooks encode logic but cannot reason or adapt. This demands an architecture treating tacit-knowledge capture, ontological grounding, sovereign deployment, and auditable actuation as co-designed from the start. This paper introduces the Corporate Language Model (CLM), a framework transforming a firm's structured, unstructured, multimodal, and tacit knowledge into an ontology-grounded enterprise foundation upon which reasoning and governed execution are composed. CLM has five capability planes and four architectural pillars: a Neurosymbolic Mesh coupling generative models with a knowledge graph; a Skill Graph where reusable tactics, personas, objections, and goals are typed and composed; Living Digital Twins modeling functional areas as reasoning surrogates; and a Deep Security Layer enforcing sovereignty, traceability, and human oversight. A Spec-as-Code paradigm bridges grounded intent and executable artifact. CLM is one instantiation of this foundation-centric class. Four contributions follow: CLM is defined as a distinct object of study; the Skill Graph is introduced for compositional explainability by construction; the Wisdom Listener effect is proposed, whereby tacit-capable foundations compound in value with use, connecting to dynamic capabilities and organizational learning; and evidence from a JCI-accredited tertiary hospital in Brazil instantiates three of the six maturity stages under LGPD.
Read the full abstract

Nuclear Experiment

Spin-dependent azimuthal asymmetry of coherent J/$\psi$ photoproduction in hadronic PbPb collisions at $\sqrt{s_\mathrm{NN}}$ = 5.36 TeV

We report a measurement of the azimuthal asymmetry of muons from the decay of coherently photoproduced J/$\psi$ mesons with respect to the second-order event plane in hadronic lead-lead (PbPb) collisions. The analysis is based on data collected by the CMS experiment at the LHC at a nucleon-nucleon center-of-mass energy of 5.36 TeV, corresponding to an integrated luminosity of 1.26 nb$^{-1}$. The asymmetry is quantified using the amplitude $A_{2}$ of the second-order harmonic modulation of the decay-muon azimuthal distribution. A negative $A_{2}$ = $-$0.194 $\pm$ 0.022 (stat.) $\pm$ 0.018 (syst.) is observed with a significance greater than five standard deviations for the first time. The $A_{2}$ values show no significant dependence on the J/$\psi$ meson rapidity or the degree of overlap of the colliding ions. A theoretical model in which the coherently photoproduced J/$\psi$ meson inherits the linear polarization of the incident photon is consistent with the data. When considered together with recent measurements of coherent $\rho^0$ $\to$ $\pi^+\pi^-$ photoproduction in ultraperipheral collisions, these results provide a direct observation of spin dependence in the azimuthal asymmetry of decay products from coherently photoproduced vector mesons. In addition, the observed correlation with the event plane may enable a novel approach to reaction plane determination in the future.
Read the full abstract

General Relativity and Quantum Cosmology

An Attractor-Repeller model of the Local Universe : The $\Lambda$-Szekeres spacetime and Perturbation Theory

The standard linearly perturbed Friedmann-Lemaître-Robertson-Walker (FLRW) framework provides an accurate description of the Universe on large scales but cannot fully capture the non-linear inhomogeneities that characterize the local Universe. In this work, we investigate whether an exact inhomogeneous solution of the Einstein field equations can provide a more realistic description of the nearby cosmic environment. To this end, we employ the quasi-spherical Szekeres class I solution with $\Lambda$ and focus on its axisymmetric subclass, motivated by recent observations indicating an axisymmetric expansion rate of the local Universe. We determine light propagation by numerically integrating the Sachs optical equations, thereby obtaining the luminosity distance and its evolution with redshift. We further compute the multipoles of the covariant cosmographic parameters and apply the formalism to an attractor-repeller configuration representative of the local matter distribution. We find that the resulting covariant cosmographic parameters are consistent with current observational constraints. Finally, we derive the gravitational potential and peculiar velocity field generated by the Szekeres density distribution, establishing the connection between the exact relativistic solution and its Newtonian counterparts.
Read the full abstract

High Energy Physics (Theory)

Higher-Dimensional Integrable Scattering

This paper describes classical soliton scattering in the 2+1d integrable chiral model, related to the Bogomolny monopole equation and self-dual Yang-Mills, as a case study for higher-dimensional integrable scattering. Extended line solitons exhibit nontrivial scattering with lump solitons and with other line solitons, and $N \to N$ interactions can be factorised into a series of $2 \to 2$ interactions. Scattering between parallel line solitons in 3d is related to lump soliton scattering in 2d, allowing for direct comparison with results in 1+1d integrable field theories.
Read the full abstract

Artificial Intelligence

Why Better Models Can Create Riskier Systems: Evidence from LLM Agents in Financial Markets

Large language models (LLMs) are being deployed at scale in consequential real-world systems, from financial markets to content moderation to hiring. We show that improving individual model capability can degrade rather than improve system-level outcomes. We hypothesize that shared training and architectures can lead more capable LLMs to behave more similarly, creating correlated actions that do not diversify away. We develop a general framework showing how this correlation creates a non-diversifiable risk floor and test its predictions in financial markets using an agent-based simulation with LLM traders of varying general-purpose capability. We find that: (1) frontier LLMs exhibit significantly correlated behavior that increases with capability; (2) when their shared reasoning is accurate, increasing agent participation reduces market-level risk; and (3) when agents share a common misinformation environment, the same correlated behavior becomes a liability. Together, these results identify a capability paradox: improving individual models does not necessarily produce better system-level outcomes. Whether the same dynamics arise in other domains is an open empirical question.
Read the full abstract

All authors Jillian Ross, Eric So, Zoe De Simone, Charles Pozniak, Andrew W. Lo

Programming Languages

Corten - Foundational Verification of Rust Programs

We present Corten, a foundational verification framework for Rust programs in the Rocq theorem prover, built on the Iris separation logic framework. Corten provides the first semantics of surface-level Rust mechanized in a proof assistant with an attached program logic, directly grounded in the Rust Reference: it deeply embeds the Typed High-level Intermediate Representation (THIR) into Rocq and formalises Rust's dynamic semantics as a weakest-precondition predicate transformer calculus. By operating at THIR rather than on internal compiler representations, Corten proof goals display the THIR AST, which pretty-prints to surface Rust, keeping verification close to the source code and facilitating maintainability as code evolves. Atop this semantics, Corten develops a program logic and a syntax-directed proof automation layer; the program logic includes defunctionalized continuation stacks that keep proof goals first-order and compact. Soundness is established incrementally, construct by construct, against an interaction-trees denotation. A synthetic test suite demonstrates a two-to-four times reduction in proof size compared to raw semantic proofs. We further showcase Corten on a buddy allocator case study, verifying memory safety of the allocation and deallocation functions, laying the groundwork for end-to-end verification in a shared Rocq semantic foundation spanning hardware-software boundaries.
Read the full abstract

All authors František Farka, Carmine Abate, Sven Linker, Sebastian Ertel

Dynamical Systems

Loss of memory in the periodic Ehrenfest model with small polyhedral scatterers

We determine the Boltzmann–Grad limit of the Ehrenfest “wind-tree” model with a periodic configuration of polyhedral scatterers. The central result is a loss-of-memory effect in dimension $d\geq 3$ for a generic choice of polyhedron. This yields a Markovian limiting transport process on an extended state space and a generalised linear Boltzmann equation for the time evolution of the particle density.
Read the full abstract

Strongly Correlated Electrons

Dual Charge-Density-Waves in Two-dimensional DyTe3 and Their Distinct Impacts on Magneto-Transport Properties

Charge-density-waves (CDWs) are macroscopic quantum states defined by periodic modulations in electronic charge density coupled with lattice distortions. Despite significant research efforts, the evolution of electromagnetic transport properties in the presence of CDWs remains largely unexplored. Here, we report the effects of two orthogonal CDWs (CDW1 and CDW2) coexisting at low-temperature in DyTe3 on its electronic transport properties. Importantly, we find that while no clear magnetotransport anomaly is resolved across CDW1 transition under the present experimental conditions, the emergence of CDW2 drives a drastic enhancement in magnetoresistance and a multiband-governed nonlinear Hall response. These results demonstrate that CDW1 and CDW2 in DyTe3 are fundamentally different and impact its electronic transport in qualitatively distinct ways, establishing DyTe3 as a model system for understanding the transport consequences of multiple CDWs in quasi-two-dimensional materials.
Read the full abstract

All authors Shuvankar Gupta, Yasemin Ozbek, Olajumoke Oluwatobiloba Emmanuel, Maya Bostock, Johnathan Kowalski, Pengpeng Zhang, Xianglin Ke

Computer Vision

AdaptVPR: Route-Aware Hard Positive Generation for Robust Visual Place Recognition

Visual Place Recognition (VPR) localizes a query image by retrieving database images of the same or nearby place, yet its robustness is often degraded by domain shifts arising from illumination, weather, seasonal changes, and dynamic occlusions. One contributing factor is the limited appearance diversity of the same place in existing training data. To address this issue, we propose AdaptVPR, a route-aware generative augmentation framework that constructs same-place hard positives for robust VPR training. AdaptVPR first uses a vision language model to parse scene attributes and estimate editing feasibility, while a rule-based scheduler determines the generation route according to editability scores and risk constraints. The generation process is decomposed into three complementary routes: the Global Appearance Route introduces global scene changes in weather, illumination, and time of day; the Local Occlusion Route inserts plausible dynamic occluders; and the Dual Route combines both types of perturbations to produce more challenging appearance shifts. Each generated candidate is evaluated using a VPR-oriented verification scheme based on geometric consistency and appearance diversity, reducing the risk of structural drift while ensuring sufficient appearance variation. Global candidates are generated once and rejected if verification fails, while Local Occlusion and Dual candidates use verification feedback for limited prompt refinement and regeneration. Using this framework, we construct AdaptCities, containing 160K verified synthetic same-place hard positives. Experiments across multiple VPR baselines and vision foundation backbones show consistent gains on standard benchmarks and substantial improvements under challenging domain shifts, with R@1 gains of up to 9.2%. The source code and data resources are publicly available at https://github.com/chenshunpeng/AdaptVPR.
Read the full abstract

All authors Shunpeng Chen, Jingyi Zhang, Changwei Wang, Shengpeng Xu, Yukun Song, Xingtian Pei, Jinzhou Lin, Li Guo, Shibiao Xu

Quantum Physics

Three-Photon and Hybrid Coherent-Fock Interference in a Two-Phase Six-Port Mach-Zehnder Interferometer

We present a unified theoretical analysis of three-photon quantum interference in a Six-Port Mach-Zehnder Interferometer (6p-MZI) constructed from two cascaded tritters, with two independent phase modulators placed between the tritter arms. We analytically derive the transfer matrix of the 6p-MZI and show how they organize into three symmetry classes, governed by the discrete Fourier transform (DFT) structure of the tritter and the conjugate relations. Furthermore, we analyze two input regimes: First, three indistinguishable single photons are injected into the tritter, and the output probability distributions $P_{[111]}$, $P_{[\{300\}]}$, and $P_{[\{210\}]}$ are derived as functions of the two relative phases $(\phi_1, \phi_2)$. At $\phi_2 = 0$, the single-phase limit is recovered, which confirms 100% visibility of the even-distribution fringe. Second, a hybrid coherent-Fock input $|\alpha\rangle_1|\alpha\rangle_2|1\rangle_3$ is analyzed via the density matrix formalism. The average photon number at each output port exhibits amplitude-dependent phase shifts. Our results establish the 6p-MZI as a programmable platform for tripartite quantum state manipulation and coherent amplitude sensing.
Read the full abstract

All authors P. O. Amadi, P. Pewkhom, N. Sitpathom, R. Endut, N. Ali, S. A. Aljunid, S. Suryadi, P. Kalasuwan

High Energy Physics (Phenomenology)

Quantum Graph Neural Networks for Jet Tagging on Quantum Hardware

Jets are central to the physics programs of both current and future colliders, from precision Standard Model measurements and searches for new physics at the Large Hadron Collider to studies of nucleon structure at the future Electron-Ion Collider. Motivated by these applications, we explore quantum machine learning for jet classification and present a permutation-invariant Quantum Graph Neural Network (QGNN) applied to particle-cloud representations of jets. We apply the model to two such discrimination tasks: quark vs. gluon and up vs. down quark flavor tagging, with the latter being, to our knowledge, the first application of a quantum model to this problem. In the ideal simulation, the QGNN performs competitively against the Particle Flow Network and traditional QCD observables. We further deploy scaled-down models to IBM and IonQ quantum processing units (QPUs), where we train and evaluate them, obtaining promising results. Finally, we perform an interpretability analysis to characterize the observables learned by the quantum model, relating them to generalized angularities for the quark-gluon study and to jet charge for the flavor study.
Read the full abstract

All authors Benjamin Jobilal, Jinghong Yang, Trevor Smith, Vincent Calvo, Zhong-Bo Kang, Shabnam Jabeen

Computation and Language

VERGE: Verification-Enhanced Refinement for Grounded Extraction of Early-Onset Colorectal Cancer Symptoms in Clinical Notes

Early-onset colorectal cancer is increasing among younger adults, yet red-flag symptoms in this age group have no evidence-based guidelines for follow-up testing, and structured encounter data do not capture the detail needed to support early detection and inform follow-up, including symptom duration, context, and fam- ily history, an established colorectal-cancer risk factor. This study aimed to develop and evaluate an automated method for extracting six red-flag symptoms and family-history risk status from free-text clinical notes. We developed VERGE, an agentic workflow in which an initial label and evidence are proposed using retrieval-augmented generation, then passed through a bounded verification- refinement cycle that checks textual grounding and clinical validity, corrects and rechecks a claim until resolved or a limit is reached, and escalates unresolved claims for human review. VERGE was evaluated on 4,033 clinician-labeled note-finding pairs against a single-agent baseline, a rule-based clinical language-processing baseline, and an alternative underlying language model. Compared with the single-agent baseline, VERGE reduced false positive find- ings, improving precision from 0.764 to 0.849 and MCC from 0.681 to 0.730, a balanced gain across the precision-recall trade-off, and resolved most flagged errors autonomously, with human review required for only 1.5 percent of claims. These results indicate that a bounded, verification-based workflow can reduce unnecessary positive findings without sacrificing the ability to detect true ones. This approach offers a path toward more reliable and trustworthy clinical language-processing tools to support colorectal cancer risk assessment in younger patients.
Read the full abstract

All authors Nikkie Hooman, Monarch Nigam, Amy E. Hughes, Rasmi G. Nair, Mehak Gupta

Image and Video Processing

Ultrasound-Based Prediction of Cirrhosis Decompensation Using Large-Scale Computer Vision Models

Decompensation represents a critical transition in the course of cirrhosis, yet clinicians have limited non-invasive tools to reliably predict its onset. In this study, we propose a novel imaging-based approach that leverages large-scale computer vision models to analyze routine abdominal ultrasound images and extract predictive features beyond those captured by traditional laboratory-based risk scores. Ultrasound is widely available, low cost, and suitable for longitudinal surveillance, making it an attractive modality for scalable risk stratification and long-term follow-up. Our framework integrates automated ultrasound data processing with modern deep learning architectures to identify patients at high risk of decompensation prior to the occurrence of clinical deterioration. This non-invasive strategy offers a practical complement to existing clinical scoring systems and may enable earlier, more proactive management of patients with compensated cirrhosis.
Read the full abstract

All authors Guangyi Zhang, Peiyun Ni, Eugene Cheah, Rajat Chandra, Peng Guo, Raymond T. Chung, Anthony E. Samir

Robotics

Scalable Edge-assisted Fusion and Path Prediction for Connected Autonomous Vehicles

The planning algorithms inside an Autonomous Vehicle (AV) rely on information from on-board sensors whose line of sight is limited by emerging traffic conditions and occlusions. Edge-assisted creation of a unified world model fusing information from AVs and Road Side Units (RSUs) in a geographical locale, and the prediction of AVs' future trajectories, can enhance the planning algorithms inside AVs to improve quality metrics, such as better traffic flow and collision prevention. AVs participating in such enhancements are called Connected Autonomous Vehicles (CAVs). However, such information generated by the edge (world model and motion predictions) must reach the planners within a tight Age of Information (AoI) time budget to be useful. The state of the art fuses per-CAV information: each AV fuses inputs from other actors locally, which limits both scalability with actor count and quality of results. We present Conductor, an edge-based solution for creating a unified world model from the perspective of a fixed anchor (e.g., an RSU) in a locale and predicting future trajectories of AVs in that locale. Our solution adheres to the AoI time budget by dynamically limiting the number of AVs that would lead to the best quality of results. Specifically, we introduce an occlusion-aware selector that favors information contribution by AVs that detect objects in the locale not covered by RSUs. We pair this selector with a runtime controller that adapts both the number of AV inputs to fuse and the amount of trajectory predictions in each cycle to stay within the AoI time budget. Evaluation on CAV simulation infrastructure shows our joint selector-controller meets the AoI safety bound across traffic scenarios with up to 31 CAVs, with fusion fidelity close to an Oracle and much better than a random selector under the same AoI constraint.
Read the full abstract

All authors Tyler Landle, Jackson Isenberg, Abhijit Chatterjee, Alexandros Daglis, Umakishore Ramachandran