Peer Revenue Impact factor: volatile

Papers

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

All of arXiv

Thursday 24 September

Mathematical Physics

The para–Krawtchouk oscillator and its continuum limit

Stéphane Z. Beaulac, Nicolas Crampé, Quentin Labriet et al. (+3 more)

The para-Krawtchouk polynomials define, through their recurrence relation, a finite oscillator model that exhibits fractional revival. We determine the continuum limit of this Hamiltonian, both from its recurrence relation and directly on the polynomials. The limit is the singular (isotonic) oscillator $-\partial_\eta^2+\eta^2+\tfrac{\gamma^2-1}{4\eta^2}$. The deformation parameter $\gamma$ survives as a central impurity of the couplings, which decays as the inverse square of the distance. This is the marginal rate, which the continuum scaling turns into a finite $\sfrac{1}{\eta^2}$ term. Using the decomposition of the para-Krawtchouk polynomials into two Hahn families, we prove the confluence, for every degree, of the eigenvectors on the two sub-lattices to generalized Hermite functions. The bispectral pair of the discrete model, which generates the Hahn algebra, becomes in the limit a pair of generators of $\su$, the dynamical algebra of the singular oscillator.
Read the full abstract

All authors Stéphane Z. Beaulac, Nicolas Crampé, Quentin Labriet, Lucia Morey, Marlon Josue Rivera Valladares, Luc Vinet

Medical Physics

MBFormer: Microbubble Transformer for 3D Time-Series Data Processing to Improve Bound Bubble Detection in Nonde-structive Ultrasound Molecular Imaging

Jihye Baek, Jeong Hoon Lee, Hoda Hashemi et al. (+4 more)

Development of nondestructive ultrasound molecular imaging (UMI) is essential for early cancer detection through real-time screening using clinical ultrasound systems. Current techniques face challenges in accurately detecting targeted microbubbles (MBs) bound to specific biomarkers, primarily due to false-positive detections of unbound free-floating MBs. We propose a transformer model for time-series video processing to improve the differentiation of bound MBs. We propose a hierarchical transformer, termed MBFormer (microbubble transformer), featuring a positional-embedding-free encoder and a lightweight decoder. Leveraging attention within 3D spatio-temporal data to effectively capture stationary signals from bound MBs while suppressing nonstationary signals from unbound MBs. Since MBs appears as relatively small textures compared with conventional segmentation targets in medical imaging, such as organs and tumors, we optimized the model with two hierarchical layers, each with an attention block, to process ultrasound video data. The network outputs the molecular signal amplitude to visualize fine MB textures. Performance was evaluated using an in vivo breast cancer model, compared against a prior CNN-based UMI method and SegFormer3D baseline, a representative 3D transformer. MBFormer (AUC = 0.943) outperformed both CNN (AUC = 0.897) and SegFormer3D (AUC = 0.766) in detecting bound MBs. The CNN showed residual molecular signal from free MBs in the cardiac chambers, whereas SegFormer3D failed to detect fine MB textures. Overall, MBFormer demonstrated enhanced detection of bound MBs while suppressing free MBs and achieved a frame rate of 16.7 to 18.1 FPS, demonstrating its potential for real-time application. We anticipate that this transformer-based UMI model can facilitate real-time, free-hand nondestructive UMI in clinical systems.
Read the full abstract

All authors Jihye Baek, Jeong Hoon Lee, Hoda Hashemi, Arutselvan Natarajan, Farbod Tabesh, Ramasamy Paulmurugan, Jeremy J. Dahl

Econometrics

Risk-Averse Welfare Maximization via Marginal Treatment Effects

Jarrod Burgh and Emerson Melo

This paper studies risk-averse treatment allocation when individuals self-select into treatment based on unobserved characteristics. We develop a framework that combines the marginal treatment effect approach to endogenous selection with a general class of coherent risk measures that capture distributional preferences over welfare outcomes. We show that the planner's problem admits equivalent interpretations in terms of uncertainty aversion, distributional robustness, and worst-case welfare. For law-invariant coherent risk measures, we derive a Kusuoka representation that expresses the planner's objective as a weighted evaluation of different regions of the welfare distribution and characterize the resulting optimal allocation rule. We further establish finite-sample regret guarantees for empirical risk-averse policy learning, showing how the statistical difficulty of learning a policy depends on the planner's sensitivity to adverse welfare outcomes. The framework nests the risk-neutral policy learning model of \cite{Kiatagawa_Tetenov_2018} as a special case. An application to the \cite{Card1995} college proximity data demonstrates that incorporating risk aversion can lead to economically meaningful changes in optimal college admission policies.
Read the full abstract

High Energy Astrophysical Phenomena

Tracing the Host-Galaxy Environments of FRBs with a New Sample of Localized Sources Detected with the DSA-110

Frank M. Verdi, Vikram Ravi, Myles B. Sherman et al. (+7 more)

We characterize the source environments of a sample of fast radio bursts (FRBs) detected with the Deep Synoptic Array (DSA-110). We present new analyses of a sample of 24 FRBs which have been localized to host galaxies with spectroscopic redshifts. Four of these bursts have been confidently associated with galaxies near or beyond redshift 1, and several are published here for the first time. Combining these new bursts with previously analyzed DSA-110 sources, we use a sample of 43 DSA-110 FRBs with known redshifts to measure host-galaxy contributions to dispersion measure (DM) and Faraday rotation measure (RM). We find characteristically large and significantly correlated host-galaxy contributions to DM and RM, suggesting that most FRBs reside in overdense, magnetized regions within their host ISM. Additionally, we identify a subset of FRBs with detectable scattering, and show that their scattering timescales and inferred host DM values are broadly consistent with an empirical relation established for Galactic pulsars. Combined with two-screen modeling of some bursts, our results indicate that density fluctuations within the host-galaxy ISM are the dominant source of extragalactic scattering. Comparisons of FRB host-galaxy properties with pulsars and simulations of young and old progenitor populations support a scenario in which FRBs preferentially form and remain associated with HII regions throughout their emission lifetime, favoring magnetars formed in core-collapse supernovae as the sources of FRBs.
Read the full abstract

All authors Frank M. Verdi, Vikram Ravi, Myles B. Sherman, Jakob T. Faber, Stella Koch Ocker, Kritti Sharma, Casey J. Law, Gregg Hallinan, Kaitlyn Shin, Liam Connor

Astrophysics of Galaxies

Evidence for the Typhon Stellar Stream as the Remnant of a Recently-disrupted Dwarf Galaxy

João V. Nogueira-Santos, Guilherme Limberg, Silvia Rossi et al. (+4 more)

In the hierarchical model of galaxy formation, the Milky Way's stellar halo assembles through the successive accretion of dwarf galaxies. While many resulting substructures have been identified, the origin of the recently discovered Typhon stellar stream remains ambiguous, as dynamical data alone are insufficient to distinguish between a globular cluster and a dwarf galaxy progenitor. To resolve this, we combine astrometry from Gaia with spectroscopic data from available surveys to comprehensively analyze the stream's chemical and dynamical properties. We identify 107 candidate members, significantly expanding the known sample. Three of these candidates lie beyond $55\,\mathrm{kpc}$, with the farthest at $67\,\mathrm{kpc}$, lending further support to Typhon's inferred large apocentric distance. Our analysis yields a mean metallicity of $\rm[Fe/H]=-1.42\pm0.07$ and a metallicity dispersion of $\sigma_{\text{[Fe/H]}} = 0.26^{+0.06}_{-0.05}$ dex. This metallicity dispersion supports a dwarf-galaxy origin for the Typhon stream. Typhon's high orbital energy is consistent with a relatively recent accretion of a fully disrupted dwarf galaxy with a stellar mass in the range of $10^6 M_{\odot} < M_\star \lesssim 10^7 M_{\odot}$. Furthermore, Typhon is the only major dwarf-galaxy stellar stream in the Milky Way's halo to exhibit such an extreme orbital eccentricity of $e \gtrsim 0.85$. We conjecture that dwarf-galaxy streams from high-eccentricity mergers are fully dissolved much more rapidly than those on circular orbits, in line with Typhon's proposed recent accretion. Additional dwarf-galaxy stellar streams on high-eccentricity orbits should be discoverable with next-generation photometric surveys as well as future releases from the Gaia space mission.
Read the full abstract

All authors João V. Nogueira-Santos, Guilherme Limberg, Silvia Rossi, Hélio D. Perottoni, Lais Borbolato, Fabrícia O. Barbosa, André R. da Silva

Cryptography and Security

Subjects, Not Authors: The Authorship Hazard in Agentic Dataspaces

Seungho Lee and Changbin Lee

Dataspace connectors decide whether a transfer may occur, not what the transferred value contains, tolerable for contracted applications, not for LLM agents that compose tool calls. Work on agents that generate governance artifacts evaluates output quality, not who may authorize an artifact for use. A published policy is what the decision point enforces, so publication is a governance event, and agents that are both policy subjects and policy authors write the norms that bind them. We name this the authorship hazard and state one principle: an agent is a subject of the governance plane, never an author of it. Its authorization channel to publication is closed by construction; its influence channel, drafting what humans approve, becomes an enforcement problem. Across 90 preregistered edits to the paper's running agreement, each evaluated on 344,512 requests, the six that only reclassify a field all change authorization and narrow a duty without touching policy text, and a policy-diff classifier passes all six. Read as worded, the privilege-delta conditions also pass 33 of 69 effective policy-text edits; read as covering any relaxation, none. Treating classification as authorship routes all six to review; the registry this requires is not yet built. At the execution boundary, protected fields reach the model in 105 of 105 cases under prompt-stated duties and in 0 of 105 under a compiled tool-call constraint, but values outside named fields are exposed in 7 of 7. At the review share measured, a central approval pool needs one approver per 20 to 138 participants.
Read the full abstract

Computer Vision

MedTokenBudget: Lesion-Preserving Token Routing for Dermoscopic Image Classification

Zhexiang Li

Dermoscopy classifiers built on Vision Transformers process all image patches uniformly, although diagnostic evidence is concentrated in the lesion region. Existing token pruning methods reduce tokens using generic saliency or similarity signals, but rarely ask whether the retained subset still contains the lesion. This paper introduces MedTokenBudget, a supervised post-backbone token routing framework that learns to construct compact lesion-enriched representations when auxiliary lesion masks are available. Its Lesion-Aware Token Scoring (LATS) module fuses attention entropy, feature norm, and local feature contrast through a learned scorer, then routes the top-$K$ patches under a target budget. LATS is trained with budget curriculum learning, diversity regularization, attention distillation, and lesion-mask supervision. The trained router is evaluated with a lesion retention rate that directly measures how much ground-truth lesion evidence survives the token budget. On ISIC 2019, mask-supervised LATS consistently outperforms Random and ToMe at headline budgets while retaining substantially more lesion patches. Code is provided for reproducibility, and complete tabulated results are included in the supplementary material.
Read the full abstract

General Relativity and Quantum Cosmology

Exceptional points and near-extremal boundary response in rotating BTZ black holes

Sheng Long, Yu Shi, Zhongwu Xia et al. (+4 more)

We investigate whether the coalescence of long-lived quasinormal modes enhances the boundary response of rotating BTZ black holes. Using exact scalar solutions with mixed Robin boundary conditions and the retarded boundary correlator, we compare frequency-domain resonances and time-domain signals at fixed source and operator normalizations. An exceptional point (EP) can be crossed by varying the spin alone in a suitably selected fixed theory. Along a distinct near-extremal path that tracks the co-rotating EP surface by tuning the scalar mass and boundary coupling, a taller, narrower resonance coexists with a weaker but longer-lasting double-pole impulse response. The frequency-domain enhancement persists away from the EP and is therefore not unique to mode coalescence. Thus, the EP determines the double-pole structure and Jordan relaxation, but does not by itself determine the response strength, which also depends on the physical path and excitation protocol.
Read the full abstract

All authors Sheng Long, Yu Shi, Zhongwu Xia, Huajie Gong, Zhoujian Cao, Qin Tan, Qiyuan Pan

Human-Computer Interaction

Epstein Files Engine: Agentic Search for Investigative Journalism

Duy K. Nguyen, Teresa Mondría Terol, Dylan Freedman et al. (+1 more)

On Jan. 30, 2026, the U.S. Department of Justice released a mixed-media collection concerning Jeffrey Epstein, including about three million pages of PDFs. We describe the Epstein Files Engine, an A.I. agent The New York Times deployed to investigate the files. The Engine translated reporter questions into Google BigQuery SQL queries across three corpora: Epstein-related releases, the Times's archive and external, Epstein-related news headlines. It used an LLM to plan queries and returned citation-rich answers a reporter could verify and trust. More than 100 journalists used the Engine, and it contributed to at least 20 published stories. We report how reporters queried it and describe Diff, our text-and-visual duplicate matching method that amplified novelty signals and allowed the Engine to surface genuinely new information. We argue that newsroom agents serve newsrooms best not as autonomous writers, but as interfaces to source material and institutional knowledge.
Read the full abstract

All authors Duy K. Nguyen, Teresa Mondría Terol, Dylan Freedman, Zach Seward

Statistical Computation

Scientific Data Analysis for Class-Informatics in Computational Taxonomy

Thomas B. Michelon and Hsieh Fushing

In this A.I. era, Computational Taxonomy is proposed to study complex systems by analyzing their databases under taxonomic hierarchies abiding the Principle of Science by providing "good explanations". Comparisons among branches or classes are carried out by Scientific Data Analysis (SDA) paradigm that explores all potential associative patterns, including interacting effects of all high orders, and evaluate finite sample precisions for all information pieces individually by effectively making use of all variables' categorical nature. Under each comparison, all confirmed information pieces are collected and displayed along row-axis of a heatmap with all involved study-subjects on the column-axis. Each comparison's heatmap individually characterizes participating classes and study-subjects and simultaneously provides a scientific basis for outlier detection upon all non-participants. All these heatmaps then collectively constitutes so-called Class-informatics that offers good explanations based on characteristic of all classes and study-subjects. Computational Taxonomy's Class-informatics indeed resolves multiple fundamental issues: Tukey's more than 60 years outlier detection problem, issue of self-correction annotation, and a crucial check on assumption of information-content equality between testing and training data sets in Machine Learning. A showcase of Computational Taxonomy is exclusively illustrated on Iris data.
Read the full abstract

Computer Vision

MVAgent: Multi-Agent Video Generation via Consistent Condition Construction and Shot-Level Policy Optimization

Xiangyu Kong, Wenjie Zhou, Fengping Tian et al. (+5 more)

Multi-shot agentic video generation requires consistent character appearance, stable spatial layout across camera angles, and continuous character state between shots. When every shot is a separate request to a frozen generator, repeated text does not determine appearance, layout or state. We therefore recast the problem as condition construction and present MVAgent, a multi-agent pipeline whose agents collaborate through typed conditioning inputs. Because an environment image shows one viewpoint, a Spatial Grounding agent samples views from generated camera-traversal clips and anchors each shot to the view matching its framing. As generated shots drift from the plan, an Observer records how each shot ends in a continuity memory, from which a Transition agent builds character action and spatial references for the next shot. An Orchestrator composes these inputs into each request. Since a request reveals its effect only after rendering, we train it by agentic reinforcement learning with Trunk-GDPO, which compares rendered candidates at every shot rather than once per video and continues the best as the trunk. With generator and judges frozen, MVAgent attains the highest cross-shot consistency and narrative-planning quality among the compared methods on ViMax-Bench and is preferred over the strongest agentic baseline in human evaluation.
Read the full abstract

All authors Xiangyu Kong, Wenjie Zhou, Fengping Tian, Lihua Fang, Haoqin Sun, Chenyang Lyu, Longyue Wang, Weihua Luo

Robotics

Audit Before You Commit: Locating Belief Failures in Active Identification for One-Shot Manipulation

Mohamed Abouagour and Byung-Cheol Min

A robot that probes a few times before one irreversible action, such as tapping a surface before inserting a peg, must decide when the evidence is enough to commit. We argue that this decision rests on two conditions that existing methods do not separate: the belief must still cover the truth in the coordinate that decides the action, and the failure model that scores actions must track realized failure. We audit both conditions separately, offline and with ground truth, on a deployed probe-then-commit pipeline: a particle belief, a scenario failure score, and one commit. On simulated insertion, more taps sharpen the belief while the truth leaves its support on 16.9% of episodes and the failure score turns optimistic by 0.31. Conformal calibration restores coverage but not the decision: confidently wrong instances still pass a confidence gate. The audit's signatures instead point at the observation model, where a hand scan finds a 2.1 mm error in the tap boundary. Correcting that one number cuts failure from 0.354 to 0.112 on untouched instances and transfers unrefitted to a second engine, while in a third engine the same audit suggests an execution-model mismatch instead. Across seven task families in three engines, a few probes at a fixed executor reduce miss or failure. On a physical arm inserting a tool into a rigid pocket by touch, the gain and the audit's two conditions reproduce, and replaying the recorded taps under an injected model error shows the audit's signature on real data. Additional materials are available at https://sites.google.com/view/auditbeforeyoucommit.
Read the full abstract

Algebraic Geometry

The norm principle for extended Clifford groups

Federico Scavia

We prove the norm principle for the extended Clifford group $\Omega(q)$ of every nondegenerate quadratic form $q$ of even dimension at least $4$ over an arbitrary field $F$ of characteristic different from $2$, for all finite separable field extensions $L/F$. The main new ingredient is a factorization theorem for proper similitudes: for a quadratic extension $L/F$ and $f\in \mathrm{GO}^+(q)(L)$, there exist $u\in \mathrm{SO}(q)(L)$, a maximal $F$-torus $S\subset \mathrm{GO}^+(q)$, and $s\in S(L)$ such that $f=us$.
Read the full abstract

Systems and Control

Resilience Enhancement of Distribution Grids Through a Three-Stage Framework for Scheduling Mobile Energy Storage Systems

Ali Abbasi and Kyri Baker

As high-impact, low-probability disturbances such as extreme weather events increase in frequency, enhancing power grid resilience has become a critical priority. Mobile Energy Storage Systems (MESSs) have strong potential for helping with this purpose due to their high operational flexibility and fast deployability. They can be rapidly relocated to supply critical loads, support islanded operation, and adapt to changing conditions of the grid, making them a promising addition or alternative to conventional resilience-providing methods. This paper proposes a three-stage framework for scheduling MESS units. In the first stage (Normal Operation), charging/discharging scheduling of MESSs at buses is optimized for economic arbitrage. Upon receiving an early warning signal, the second stage (Proactive Positioning) shifts the focus to resilience. Here, the model determines the optimal staging points within the transportation network to minimize the expected time of arrival at critical loads, considering the probability distribution of line outages. The third stage (Dynamic Relocation) addresses the post-event restoration phase, in which MESSs are relocated in response to grid updates, such as subsequent line failures. Furthermore, a graph neural network-based solution approach is proposed to further enhance computational efficiency in determining the optimal placing of MESS units during the pre-allocation and relocation stages. The simulation results demonstrate that the proposed framework reduces expected energy not served and improves the restoration speed.
Read the full abstract

Machine Learning

Probabilistic Robustness-driven Universal Adversarial Perturbations with Explainability against Deep Reinforcement Learning-based Intrusion Detection System

Hongsen Zhang, Lu Zhang, Mingjing Xu et al. (+3 more)

Deep reinforcement learning (DRL) enables adaptive intrusion detection in dynamic network environments but also exposes intrusion detection systems (IDS) to adversarial threats such as universal adversarial perturbations (UAPs), which apply a single input-agnostic perturbation to degrade detection performance across traffic. Probabilistic Robustness (PR), as a post-hoc evaluation metric, provides a principled, population-level measure of adversarial impact that conceptually aligns with the universality objective of UAPs, i.e., PR quantifies the prevalence of misclassification in the input space, making it a natural signal for guiding UAP generation. Hence, we propose PR-based UAP, which represents the first integration of an explicit PR-driven objective into generating UAPs against DRL-based IDS. Building on this formulation, we introduce PX-UAP, which leverages explainable artificial intelligence (XAI) to guide perturbation shaping under realistic domain constraints, and provides a rigorous theoretical analysis of its design. Extensive experiments demonstrate that PX-UAP consistently outperforms state-of-the-art UAP methods in attack effectiveness.
Read the full abstract

All authors Hongsen Zhang, Lu Zhang, Mingjing Xu, Yi Zhang, Gregory Epiphaniou, Carsten Maple

Computation and Language

The Hard Part Comes After Search: Benchmarking Web Agents on Synthesizing, Organizing, and Displaying Knowledge

Alexander Gill, Md Farhan Ishmam, Xuyen Nguyen et al. (+6 more)

Existing computer-use agent benchmarks do not fully evaluate agents acting as assistants. A useful assistant retrieves information across complex, multi-step workflows, synthesizes it into artifacts (documents, presentations, spreadsheets), and navigates program interfaces to produce a coherent final product. Such workflows demand reasoning and synthesis, decomposition of complex tasks, as well as visual and spatial understanding. To study agents on workflows like these, we introduce KNOWS, a benchmark of open-ended, complex, browser-based tasks that jointly evaluate these capabilities, with each task culminating in a produced artifact. To write tasks, we develop a task design rubric and a protocol for ensuring that tasks meet the requirements. Each task is paired with an evaluator, a program that combines deterministic checks with LLM judgments to balance the richness, reliability, and automation tradeoff inherent to agent evaluation. We evaluate and analyze frontier computer-use agents and browser-based harnesses. They achieve moderate scores on partial-success metrics, but the best performer fully succeeds in fewer than 3% of our complex, long-horizon tasks. Failures on visual steps render the resulting artifacts unusable, even when agents complete more than 50% of other evaluation steps. Our results expose limitations of current agents acting as end-to-end assistants, and call for progress on tool use, visual understanding, and long-horizon reasoning.
Read the full abstract

All authors Alexander Gill, Md Farhan Ishmam, Xuyen Nguyen, Neha Bhat, Parker Henry DeYoung, Fateme Hashemi Chaleshtori, Nathan Stringham, Kenneth Marino, Ana Marasović

Geometric Topology

Topological components of surface group representations into the unitary group

Xueyuan Wan

We study representations of compact oriented surface groups into $\mathrm U(p)$ with elliptic-unipotent boundary holonomies, meaning that no boundary holonomy has eigenvalue $1$. We prove that the signature of the associated flat Hermitian bundle completely determines the connected component, and that every component is path-connected. For genus $g\geq1$ and $n\geq1$ boundary components, there are $np-1$ components; for $g=0$ and $n\geq2$, there are $(n-2)p+1$. The disk case is empty, while the closed-surface representation spaces are connected. The same component classification holds after taking the quotient by conjugation. Our proofs use the boundary rho invariant and explicit matrix deformations. For a three-holed sphere, the representation components have the homotopy types of complex Grassmannians, and their conjugation quotients are contractible.
Read the full abstract

Systems and Control

Decentralized Gain Learning for Voltage Control via Local Trajectory Convolution

Yiwei Dong and Wenqi Cui

Fast and spatially heterogeneous fluctuations from distributed energy resources call for voltage regulation that is both fast and responsive to changing operating conditions. Local linear Volt/VAR control offers the advantage of promptly regulating voltages without real-time communication, but its performance depends critically on the choice of control gains. Optimizing these gains for a system-level objective, however, typically requires an accurate network model or centralized communication. In this work, we uncover a structural property that enables such optimization using only local measurements. Specifically, by exploiting the self-adjoint structure of the closed-loop voltage dynamics, we show that each component of the system-level objective gradient can be exactly recovered from a self-convolution of the corresponding bus's voltage deviation trajectory. Building on this result, we develop a simple decentralized gain optimization method in which each bus updates its control gain using only its local data trajectory. Simulation results demonstrate that the proposed approach effectively updates the Volt/VAR control gains in response to changing operating conditions while requiring only local information.
Read the full abstract

Information Retrieval

Embedding Subspace Partitioning for Dynamic Multi-Objective Retrieval

Shaobo Zhang, Alice Leung, Yunxiang Ren et al. (+13 more)

Modern industrial recommender systems must optimize across competing objectives, balancing semantic relevance with business metrics such as engagement and revenue. While bi-encoders dominate large-scale retrieval due to their efficiency, they collapse these heterogeneous signals into a single static embedding space. This design creates a fundamental limitation: once trained, the retriever cannot adapt to shifting objective priorities at serving time without retraining. Moreover, joint optimization with multi-objective losses often induces interference between objectives, leading to suboptimal trade-offs. We propose Embedding Subspace Partitioning (ESP), a retrieval framework that decomposes the embedding into task-aware subspaces and replaces the single dot product with a weighted sum of per-subspace similarities, whose weights are tunable at serving time. For Transformer bi-encoders, ESP uses the model's native end-of-sequence token as a segment delimiter, with segment-aware attention masking and position encoding resets to guarantee subspace isolation in a single forward pass. Serving is performed via GPU-accelerated exhaustive kNN over one concatenated index, eliminating the need for per-objective Approximate Nearest Neighbor (ANN) infrastructure required by multi-head approaches. We evaluate ESP on an open-source benchmark built from MS MARCO. A single ESP model traces a broad Pareto frontier, consistently outperforming strong multi-task baselines across diverse operating points. In LinkedIn's job matching platform (70M+ weekly users), ESP enabled dynamic retrieval reconfiguration and delivered significant key business metric lifts.
Read the full abstract

All authors Shaobo Zhang, Alice Leung, Yunxiang Ren, Ping Liu, Yuchin Juan, Qianqi Shen, Benjamin Le, Jianqiang Shen, Chengming Jiang, Ko-Cheng Wang, Vidya Krishnamurthy, Caleb Johnson, Fedor Borisyuk, Luke Simon, Jingwei Wu, Wenjing Zhang

Functional Analysis

Best constants in noncommutative Burkholder inequalities

Yong Jiao, Adam Osekowski, Lian Wu et al. (+1 more)

The paper contains the study of noncommutative Burkholder inequalities. We prove that the constant involved is of order $O(p/\log p)$ when $p\to \infty$, which is known to be optimal even in the commutative setting. This answers an open problem raised by Randrianantoanina in 2007, and by Junge and Xu in 2008.
Read the full abstract

All authors Yong Jiao, Adam Osekowski, Lian Wu, Yahui Zuo

Robotics

Learning-Accelerated Narrow-Phase Collision Detection via Check Ordering for Sampling-Based Motion Planning

Hao Jiang, Yinghan Wang, Jianping He et al. (+1 more)

Collision detection is critical for ensuring the safety of planned paths. However, it imposes a non-negligible computational burden on motion planners, motivating extensive studies on collision-detection acceleration. In commonly used phase-based collision-detection methods, the broad phase employs hierarchical structures to rapidly discard object pairs that are clearly collision-free, while the subsequent narrow phase performs detailed collision checks on the remaining object pairs whose collision status cannot be determined by the broad phase. Although these methods effectively reduce the number of detailed checks through broad-phase pruning, the narrow phase is usually executed in the default order returned by the broad phase, with little explicit optimization of the check order. This leaves room for further acceleration, especially in cluttered environments where many object pairs may remain after the broad phase and the narrow phase can account for a significant portion of the total detection time. In this work, we propose a learning-based method to accelerate phase-based collision detection by optimizing the check order in the narrow phase. We first formulate the expected time cost of the narrow phase and derive an optimal check-ordering criterion that minimizes this expectation. Since the priors required by this criterion are difficult to obtain in advance, we design a hypernetwork-based model to predict collision probabilities, which are then used to approximate the optimal check order. The resulting order guides the execution of exact mesh checks in the narrow phase, thereby reducing detection time without replacing the underlying geometric collision checker. Simulation results show that our method effectively accelerates phase-based collision detection and improves the efficiency and success rate of sampling-based motion planning, especially in cluttered environments.
Read the full abstract

All authors Hao Jiang, Yinghan Wang, Jianping He, Xiaoming Duan

Astrophysics of Galaxies

The JWST Proto-PAH project: Aromatic backbones with extensive aliphatic substitution account for both the aromatic and aliphatic emission

N. Clark, E. Peeters, J. Cami et al. (+18 more)

We find that the aromatic and aliphatic emission bands in four carbon-rich post-asymptotic giant branch stars with class D emission arise from a single carrier population: extensive aliphatic decoration of an aromatic skeleton, i.e. molecules with a coherent aromatic backbone and an unusually high aliphatic content. This conclusion is based on JWST NIRSpec and MIRI/MRS spectroscopy, covering 2.88-28.7um. The aliphatic 3.4um feature is stronger than the aromatic 3.3um feature, reversing the ratio seen in sources with typical polycyclic aromatic hydrocarbon (PAH) emission. If the aromatic emission arose from nominal PAHs, the 3.3um feature - unavailable in previous observations - together with the mid-infrared aromatic features, would constrain them to be large (greater than about carbon atoms), compact, and at least partially ionized. These properties predict a strong 8.6um feature. However, the 8.6um feature is weak or absent in all four sources, ruling out PAHs as the source of the aromatic emission. Instead, the emission must arise from the same carriers as the aliphatic emission. The position of the 6.9um feature and the broadened 11-14um out-of-plane emission independently support this. The CH2 and CH3 feature strengths vary across the sample, with the class D2 sources showing higher CH3 content. These carriers are far more aliphatic than the PAHs of planetary nebulae and the interstellar medium. However, they retain the aromatic skeleton that produces their aromatic features.
Read the full abstract

All authors N. Clark, E. Peeters, J. Cami, G. C. Sloan, K. Volk, K. E. Kraemer, H. L. Dinerstein, M. A. Gomez-Munoz, G. M. Wahlgren, N. C. Sterling, A. Ricca, A. A. Zijlstra, J. Li, M. Matsuura, D. A. Garcia-Hernandez, R. Sahai, B. Aringer, C. Bhatt, J. Bernard-Salas, D. Das, A. Manchado

Statistical Mechanics

Emergent Wigner magnon crystals in a fully frustrated Heisenberg four-leg tube

Azam Zoshki, Hamid Arian Zad and Jozef Strecka

The ground state, magnetization curves, low-temperature thermodynamics, and heat-engine performance of the fully frustrated spin-$1/2$ Heisenberg four-leg tube with antiferromagnetic inter- and intra-plaquette coupling constants J1 and J2 are examined using exact diagonalization, density matrix renormalization group, and localized-magnon theory. In the unfrustrated to weakly frustrated regime J2/J1 << 2, the system exhibits a continuous field-driven quantum phase transition between the gapped Haldane phase and a gapless Tomonaga-Luttinger quantum spin liquid. In the highly frustrated regime J2/J1 > 2, the system contrarily displays discontinuous field-driven quantum phase transitions between the Wigner magnon crystals, which are manifested in zero-temperature magnetization curves as intermediate plateaus at zero, one-quarter, one-half, and three-quarters of the saturation magnetization. The low-temperature magnetic and thermodynamic properties in this regime are accurately captured by an effective interacting lattice-gas model of two monomer quasi-particle species constructed from localized one- and two-magnon states. Finally, we explore a quantum Stirling heat engine using the fully frustrated four-leg tube as the working medium. The work output and efficiency are strongly suppressed near discontinuous field-induced transitions and reach pronounced local maxima well inside the stability regions of the Wigner magnon crystals.
Read the full abstract

Computer Vision

Action Forcing: Training World Models on Unsupervised Video by Recovering Underlying Egomotion Bases

Ashish Sundar, Tiankuo Hou, Zhong Fan et al. (+2 more)

Synchronised action annotations are needed to train controllable world models and these datasets remain elusive. Existing approaches make use of instrumented platforms with calibrated sensors, costly manual annotation, or latent-action models which lack grounding. We instead turn ordinary unlabelled video into action-supervised training data by recovering (without training) a data-derived egomotion basis. We track pixel displacements across frames and exploit the recurring coherent structure induced by egomotion to obtain grounded control signals directly. Using a method as simple as principal components analysis perform this, we find that the leading components provide signed, scalable, and composable throttle–yaw controls, although the method can recover only motion axes represented in the data. To prevent a high-capacity video DiT from exploiting pixel-level supervision, an online latent critic distils a frozen decoder–tracker–PCA (Principal Components Analysis) teacher without backpropagating through the decoder or tracker. Finally we critique the use of video generation metrics to evaluate WMs and introduce an example of an alternative, reference-free evaluation method. We measure controllability, plausibility, conjuring (creating objects out of thin air) and geometric integrity, revealing failures that conventional video metrics miss. We show that most baselines follow familiar action directions but struggle to reverse or remain stationary. Our model handles both while retaining compositional control and generation quality. Despite backwards actions being less than $1\%$ of our training data, we find that the model learns to reverse, scale its response linearly, and compose throttle with steering, all simply by learning through a grounded action space.
Read the full abstract

All authors Ashish Sundar, Tiankuo Hou, Zhong Fan, Chunbo Luo, Xiaoyang Wang

Robotics

HuGo: LLMs as Whole-Body Policy Code Designers for Humanoid Loco-Manipulation

Seoyeon Choi, Shizhao Ye, Nicholas Bui et al. (+5 more)

For humanoids to be useful in everyday environments, they must perform a wide range of tasks that couple locomotion and manipulation. Existing approaches commonly acquire a loco-manipulation policy through reward engineering or demonstrations followed by task-specific training, making it costly to scale to new tasks. In this work, we propose a hierarchical approach to humanoid loco-manipulation that eliminates these per-task requirements. HuGo, Humanoid policy code Generation, uses a Large Language Model (LLM) to generate executable, closed-loop high-level policy code from a task description on top of a frozen low-level whole-body policy. Given the task, observation, and command specifications, the LLM constructs the task logic in code. HuGo then refines the policy from its rollouts using numerical trajectories and selected video frames to produce feedback and targeted code updates. Across five simulation tasks, using two different low-level policies, HuGo substantially outperforms a high-level reinforcement learning baseline and approaches the performance of a demonstration-based baseline. We achieve this level of performance without task-specific reward design or demonstration collection. We further demonstrate zero-shot transfer of simulation-generated policies to hardware and show that applying the same refinement loop to real-world rollouts can further improve transfer performance without expert demonstrations or policy retraining. Project website is https://iconlab.negarmehr.com/HuGo/
Read the full abstract

All authors Seoyeon Choi, Shizhao Ye, Nicholas Bui, Aayushi Shrivastava, Kanghyun Ryu, Dhruva Tirumala, Markus Wulfmeier, Negar Mehr

Software Engineering

EA-Ops: Git-Native Architecture as Code for Continuous Enterprise Architecture Governance

Vahid Tavakkoli, Kabeh Mohsenzadegan and Kyandoghere Kyamakya

Enterprise architecture (EA) repositories frequently separate architecture models from the engineering workflow used to change software and infrastructure. This article presents EA-Ops, an open-source Git-native Enterprise Architecture-as-Code framework that represents architecture facts as YAML, validates typed relationships against an ArchiMate 3.2 profile, enforces organization-specific governance rules, performs graph-based change-impact analysis, and publishes human-facing reports and a static interactive portal from the same reviewed source. We evaluate EA-Ops with a reproducible GitHub Actions harness. Eight independently injected structural, semantic, and governance fault classes were executed across 30 trials each; all 240 trials matched ground truth exactly, with precision, recall, and $F_1$ of 1.000. Scalability experiments with 30 measured repetitions reached 50,000 objects and 100,000 relationships: median validation time was 52.582~s, median impact traversal was 627.000~ms, and peak resident-set size was 919.2~MB. A ten-scenario Metroville digital-permit reference architecture produced exact validation outcomes and exact impact-set agreement with an independent breadth-first-search oracle in every scenario. The configured 100,000-object end-to-end benchmark generated its model successfully but exceeded the 180-minute CI budget during the performance stage; no timing result is extrapolated. At 50,000 objects, Markdown report generation rather than semantic validation is the dominant scaling bottleneck. The results support Git-native continuous governance as a practical EA operating model at tens-of-thousands-of-object scale while defining clear limits and optimization targets for larger repositories.
Read the full abstract

Machine Learning

QSV: Quat-Sphere-Vision for Coupled Quaternion Attention on Spherical Lattices

Nicholas Foley, Devin Marinelli, Donny Moore et al. (+2 more)

In standard attention, three separately learned projections decide how strongly a token attends to each neighbor ($W_Q$, $W_K$) and how the attended features are transformed before aggregation ($W_V$). We study Quat-Sphere-Vision (QSV), a sparse spherical vision model that replaces this projection triple with a single learned unit quaternion per token: the relative quaternion $r_{ij} = q_i^{*} \otimes q_j$ supplies both the attention logit $\operatorname{Re}(r_{ij})$ and a sandwich-product feature transport $x \mapsto r_{ij} \otimes x \otimes r_{ij}^{*}$, with messages passed over sparse kNN graphs on concentric Fibonacci spheres. Ablations that change only the targeted component show the two roles to be asymmetric. Removing the transport reduces test accuracy by about four percentage points on CIFAR-10 and CIFAR-100 (single runs per CIFAR-100 variant), while replacing the learned attention weights with uniform averaging leaves it essentially unchanged. Parameter-matched controls then remove the geometry itself: standard attention on the same graph exceeds QSV (mean $87.3\%$ vs. $85.9\%$), and the same model on a flat 2D lattice reaches $91.1\%$, within $2.1$ points of a ResNet-20 trained under the same pipeline (single run). In the coupled kernel, nearly all of the learned pairwise computation resides in the transport channel.
Read the full abstract

All authors Nicholas Foley, Devin Marinelli, Donny Moore, Diego Enriquez, Amanda Fernandez

Algebraic Topology

Rational homotopy theory of flag manifolds

Haibao Duan

Let $G/P$ be a flag manifold with $G$ simple. We determine the rational homotopy groups $\pi _{\ast }(G/P)\otimes \mathbb{Q}$ and describe the rational cohomology ring $H^{\ast }(G/P;\mathbb{Q})$ solely in terms of the rational homotopy types of $G$ and $P$, respectively. The proof is based on the restrictions of basic Weyl invariants of $G$ to suitable simple factors of $P$, which also address a key nonvanishing issue not settled by earlier approaches of Meier, Shiga-Tezuka, Kotschick, and Terzic.
Read the full abstract

Probability

Potential Hessian Ascent IV: Sampling the Sherrington-Kirkpatrick model at $\beta < 1$

Holden Lee, Juspreet Singh Sandhu and Jonathan Shi

We give a polynomial-time algorithm to sample from the Gibbs measure of the Sherrington-Kirkpatrick (SK) model with $o_n(1)$ error in total-variation distance (TVD) at any inverse-temperature $\beta < 1$. The algorithm combines algorithmic stochastic localization (ASL) with rejection sampling over path-space via Jarzynski's equality (JE). The analysis extends the authors' prior $\beta < 1/2$ result [arXiv:2605.03718] by replacing all global regularity requirements in the stochastic differential equation (SDE) error analysis with local regularity around likely trajectories. The relaxed regularity is established using Celentano's proof of the local strong convexity of the TAP free energy [arXiv:2208.09550]. The analysis utilizes the cavity interpolation theory and free probability toolkit developed in the authors' previous result, where the former applies nearly verbatim and the latter applies supplemented with Lipschitz and $C^2$ extensions of various functions. The ASL and JE analysis arises from using the TAP free energy as an efficiently computable proxy for the actual free energy of the stochastically localized Gibbs measure [$ \S $ 3, arXiv:2605.03718]. We give a list of \vocab{desiderata} encapsulating the approximation and regularity properties required of the TAP free energy, relaxing those of [$ \S $ 2.5, arXiv:2605.03718] to only require local regularity. These generic desiderata are potentially applicable in other settings where a free energy surrogate exists, giving algorithmic sampling guarantees while bypassing the usual functional inequalities based approach.
Read the full abstract