arXiv subjects

Search papers

Search indexed arXiv papers on artificial intelligence, large language models, computer vision and robotics. Read source abstracts and follow links to arXiv.

At least 1,729 records · Page 96Linked to original sources

arXiv2026

Projections in reduced group C*-algebras with prescribed trace values in (0,1)

Building on OpenAI's preprint, we construct, for every real number $r\in(0,1)$, a finitely generated discrete group $G_r$ and a projection $e_r\in C_r^*(G_r)$ whose canonical trace is $r$. The construction encodes the geometry of expanders in the dynamics of a semidirect product of groups. For irrational $r$, the $K$-theory class $[e_r]$ is not in the image of the reduced Baum--Connes map. The groups $G_r$ have torsion and satisfy the maximal strong Novikov conjecture.

math.OA↗

arXiv2026

The design and flight of the Payload for Ultrahigh Energy Observations (PUEO)

Sources of cosmic rays at ultrahigh energies (UHE, $\gtrsim$ 1 EeV) are anticipated to give rise to neutrinos at similar energies by production via propagation interactions either within the source region or on cosmological scales. Measurements of the neutrino flux, flavor, and energy provide significant insight into those underlying processes governing UHE cosmic ray acceleration and propagation. Detecting neutrinos at these energies requires immense detector volumes due to their low predicted flux. We present the balloon-borne mission PUEO (Payload for Ultrahigh Energy Observations) which is an antenna array designed to detect the Askaryan emission from UHE neutrino interactions in the Antarctic ice. This paper reports on the design considerations and implementations for constructing PUEO, from the design parameters to the as-built validations.

astro-ph.HE↗

arXiv2026

Connectivity rewiring and filament formation by active region emergence observed with Solar Orbiter

Using an 11-day dataset from Solar Orbiter, combining EUV imaging, H$α$ observations, vector magnetograms and nonlinear force-free field extrapolations, we track the evolution from the emergence of a new active region to the reconfiguration of coronal connectivity. We find that the restructuring proceeds through a sequence of localized reconnection events that propagate along the filament channel, producing successive brightenings and bidirectional flows. This process leads to the gradual erosion of the original filament and the simultaneous formation of a new, highly sheared filament connecting the two regions. Magnetic modeling reveals a transition from a sheared arcade supporting the initial filament to a twisted flux rope underlying the newly formed structure. Our results demonstrate that flux emergence can reorganize filament channels through a spatially and temporally ordered sequence of reconnection events, rather than a single impulsive episode. This sequential reconnection front provides a natural mechanism for filament formation in evolving active regions and highlights the importance of continuously changing magnetic connectivity in the corona.

astro-ph.SR↗

arXiv2026

Estimating Treatment Effects in Complex Disease Dynamics: Immortal Time, Varying Hazards, and Multiple Time Scales

In observational survival studies, delayed treatment initiation can induce immortal time bias in effect estimation when treatment groups are defined by future uptake. Despite originating from distinct predictive or causal inference paradigms, time-dependent approaches (e.g. the time-dependent Cox model, landmark supermodel, multi-state model, and g-formula) are often treated as interchangeable in applied research when addressing immortal time bias. These models effectively reclassify person-time according to time-varying exposure, yet the interpretation of their target estimands depends on assumptions about the underlying disease process and clinically relevant time scales. We study effect estimation for allogeneic hematopoietic stem cell transplantation under baseline intention-to-transplant strategies in acute myeloid leukemia (i.e. transplantation is planned for a prespecified time, subject to the patient's eligibility), within a multi-state framework allowing time-inhomogeneous Markov, semi-Markov, and state arrival extended semi-Markov dynamics. Using target trial emulation, we benchmark, on the probability scale, estimates derived from Cox-type time-dependent models and the g-formula against the corresponding theoretical values implied by the stochastic data-generating process. Decomposing the partial likelihood score, we show that single-time-scale models can be systematically biased when death intensities from different originating states are not proportional on a common time scale, as can occur when the clock is reset after intermediate events. We quantify these findings in a phase III simulation study and illustrate their implications in an application to data from three acute myeloid leukemia trials.

stat.ME↗

arXiv2026

CounterFlow: Target-Conditioned Flow Matching for Visual Counterfactuals

In eXplainable AI, counterfactual explanations have gained significant traction by offering users direct, intuitive, and actionable insights into model decisions. We introduce CounterFlow, a novel framework based on Conditional Flow Matching that generates visual counterfactual explanations by directly steering a source image toward a user-specified target class. By leveraging target-conditioned vector fields, CounterFlow identifies highly realistic semantic modifications while requiring substantially lower computational overhead than traditional GAN- or diffusion-based architectures. Empirical evaluations on standard benchmarks demonstrate that CounterFlow achieves state-of-the-art performance in both visual realism and target class switching. By working directly in the raw pixel space, CounterFlow offers a more practical solution for real-world deployment.

cs.LG↗

arXiv2026

Verse: Semantics and Verification of Hybrid Multi-Agent Scenarios

Verifying cyber-physical systems with many interacting agents remains difficult because interactions are mediated by sensors, individual decisions are constrained by environment (map) structure, and the resulting mode space scales combinatorially. We present Verse, a framework for writing agents, maps, and sensors in a familiar programming language syntax, with a formal semantics for the resulting scenarios. A compiler translates each scenario to a hybrid automaton whose executions correspond exactly to the runs of the scenario. Reachability analysis on this automaton builds an execution tree from two post operators, one for continuous flow over a time step and one for a single agent's mode change, and checks safety assertions at every vertex. The tree expands only the modes that are reached, without enumerating the global mode space. Experiments highlight expressive modeling across cars and drones, 2D/3D maps (including OpenDRIVE imports), noisy sensors, and multiple simultaneous safety assertions. Verse verifies a scenario with 7 agents and 37 modes in 90 seconds on a standard laptop; it is competitive with CORA and JuliaReach on ARCH benchmarks, and on multi-agent scenarios is an order of magnitude faster than JuliaReach at the cost of looser reachable sets.

cs.FL↗

arXiv2026

Supply Chain Information Sharing among Competitors: Profit-Based Participation under Market Shocks

Sharing information can help coordinate supply chain decisions in response to market shocks. Yet, firms must weigh these benefits against the risk of strengthening competitors. We study when competing firms voluntarily join an information-sharing platform to exchange information about market shocks while retaining operational autonomy. Firms differ in information capabilities and decide whether to participate before private signals are realized, creating stochastic information asymmetry. We model participation and quantity decisions as a two-stage dynamic game with incomplete information. For two horizontal competitors, we derive closed-form equilibrium participation strategies that reveal how information power and production efficiency shape participation. Numerical analysis examines more complex supply chain networks. We find that supply chain characteristics impact participation and the level of information asymmetry is a key factor in the equilibrium platform profiles. Firms with similar information power may benefit from joining forces to reduce their competitive disadvantage against higher information power competitors. Cost-inefficient firms may be less willing to participate because their competitors can exploit shared information, and uncertainty can serve as a competitive shield by restricting the actions of more efficient competitors. However, when several competitors join the platform, a cost-inefficient firm may also choose to participate to avoid an informational disadvantage.

math.OC↗

arXiv2026

Printed and Scanned: Exposing the Fragility of Single-Image Morphing Attack Detectors under Print-Scan Degradation

Face morphing attacks, in which the biometric traits of two or more identities are blended into a single face image, pose a severe threat to identity document issuance and automated border control. Consequently, a large body of work has proposed Single-image Morphing Attack Detection (S-MAD) algorithms and benchmarked them on datasets such as SYN-MAD22. However, the majority of these benchmarks operate exclusively in the digital domain, which contrasts with the real-world passport issuance pipeline in some countries, which still requires an applicant-supplied photograph to be printed and subsequently re-scanned before it is embedded in the identity document or fed into an automated verification system. This print-scan step introduces halftoning, resampling, color-space, and sensor noise that are absent from digital pipelines, and may mask the high-frequency artifacts that S-MAD models rely on. In this work, we introduce a print-scan version of the SYN-MAD22 dataset, obtained by printing and re-scanning all bona fide and morphed images at 300 and 600 DPI, and we use it to evaluate eight state-of-the-art S-MAD models trained exclusively on digital data. We additionally repeat the acquisition with a second scanner to assess cross-device robustness. Our results show that detectors that achieve near state-of-the-art performance in the digital domain suffer a more than threefold increase in average equal error rate, with the same qualitative degradation observed across both scanning devices, exposing a critical, previously under-reported gap between digital-benchmark performance and robustness to print-scan degradation.

cs.CV↗

arXiv2026

Nuclear recoil effects with a realistic nuclear charge form factor

We introduce a novel two-parameter Gamma model of the nuclear charge distribution as an alternative to the widely used Fermi distribution. The model yields simple analytical formulas for all moments of the charge distribution and closed-form expressions for the electrostatic potential and momentum-space form factor, the latter satisfying the correct analyticity requirements for a physical charged particle. We apply the model to calculate the nuclear recoil correction to the Lamb shift in electronic and muonic atoms and analyze its nuclear-model dependence. We find that the nuclear-size recoil corrections to energy levels of heavy muonic atoms are strongly sensitive to higher moments of the nuclear charge distribution, highlighting the importance of realistic nuclear modeling. To facilitate further calculations, we also provide the Fortran code RECFNS, designed for computing the nuclear recoil correction within the QED framework. The present work establishes a basis for future extensions to higher orders in the electron-to-nucleus mass ratio, hyperfine splitting, and radiative corrections.

physics.atom-ph↗

arXiv2026

OmniShowcase: Omnidirectional Subject-Consistent Video Generation from Multi-View Images

Generating videos of a specific subject from a few reference images remains challenging when the camera reveals surfaces that are not visible in the input. Early subject-to-video (S2V) methods condition on a single reference image and must therefore infer the subject's unseen appearance and geometry as the viewpoint changes. To address this limitation, recent methods extend S2V to multiple reference views, providing more information about the subject from different viewpoints. However, they do not explicitly specify the 3D structure that connects these observations. In addition, existing training data do not guarantee that viewpoint changes preserve the same underlying 3D subject. We present OmniShowcase, built on two components: (i) a fully automated data synthesis pipeline that produces training videos along diverse camera trajectories, with viewpoint variation driven by underlying subject geometry, even spanning full 360-degree orbits; and (ii) a geometry-conditioning method that provides the video model with a turntable of surface-normal maps reconstructed from the reference views without requiring temporal alignment with the generated frames. We construct Showcase-4K, a corpus of 3D-asset-based videos covering twelve camera motions across 88 trajectory configurations with reconstruction-derived surface-normal conditioning, and ShowcaseBench, a held-out evaluation set with the source meshes and camera parameters used to render each video. Experiments show state-of-the-art subject identity and 3D consistency while maintaining competitive video quality and motion. Project page: justin4ai.github.io/OmniShowcase

cs.CV↗

arXiv2026

Adapting LLM-SC to Living Sacred Heritage: Domain-Specific Extraction, Semantic Networks and Human Evaluation on Pashupatinath Temple Reviews

Large language model (LLM) pipelines can turn visitor reviews into structured data, but generic schemes may miss rituals, sacred spaces, cultural restrictions and emotionally mixed experiences at living sacred heritage sites. We adapted the LLM-Social Computing (LLM-SC) framework to Pashupatinath Temple, Nepal, by specifying a 12-type entity taxonomy (four original LLM-SC types plus eight sacred-heritage types), four relation types and nine aspect-sentiment dimensions in a GPT-4o extraction prompt, which was applied to a fixed random sample of 2,500 English-language Tripadvisor reviews from a processed corpus of 3,939. The exports contain 11,286 entity, 6,232 aspect-sentiment and 5,590 relation records, covering 2,485, 2,487 and 2,433 reviews, respectively; the eight added sacred-heritage types account for 64.8% of entity records. Against an adjudicated two-annotator reference for 200 reviews, exact normalized surface-and-type matching gave provisional entity scores of 41.0% precision, 21.8% recall and 28.4% micro-F1. Aspect-presence F1 was 53.6% and end-to-end aspect-plus-polarity F1 was 42.6%. Human adjudication supported 153 of 194 first-listed relation candidates (one per review; 78.9%), a candidate-level rate that does not estimate relation recall or whole-graph precision. The contribution is a culturally grounded, domain-specific adaptation with component-wise human evaluation and explicit reproducibility and generalization limitations.

cs.CL↗

arXiv2026

FISRF: Flexible Instructional Scaffolding Reflection Framework for Robotics Task Planning with Large Language Model

Large Language Models (LLMs) are widely adopted for reflective error correction in task planning; however, existing approaches primarily rely on autonomous exploration, rendering them prone to accumulating erroneous experiences during early interactions with new environments. To address this limitation, we propose the Flexible Instructional Scaffolding Reflection Framework (FISRF), an agent--mentor architecture that estimates the robot's Zone of Proximal Development (ZPD) in task planning and adaptively selects reflection strategies and mentor guidance, thereby facilitating the accumulation of effective error-correction experience. Experiments in ALFWorld and on real-world robotic platforms demonstrate that FISRF reduces the average number of error-correction rounds by 37% and redundant reasoning by 25.1%, while improving the overall task success rate by 8%. Project webpage : https://fisrfplanning.github.io/

cs.RO↗

arXiv2026

TrySee: Modular Perception-Enhanced VLM Grounding for Closed-Loop Robot Task Planning in Specialized Domains

Vision Language Models (VLMs) have shown impressive capabilities in high-level reasoning and task planning for robots. However, specialized domains remain a challenge for VLMs' image interpretation and grounding abilities. We introduce TrySee, a modular closed-loop robot task planning pipeline integrating general-purpose VLM reasoning with domain-specific perception and established robotics components. The system consists of an inherently reactive cyclic planning pipeline, augmented with domain-specific modular vision preprocessing, to provide online grounding. Finally, a dynamic preprocessing selection mechanism contributes to the scalability of the pipeline. The system requires neither VLM fine-tuning nor end-to-end demonstrations and adapts through replaceable perception modules and motion planning. The approach is evaluated on planning tasks across 5 widely different domains, including a real-world robot disassembly task. The field tests in a brownfield environment demonstrate the approach's practicality in integrating VLM planning into specialized robotic systems with limited domain-specific training data.

cs.RO↗

arXiv2026

Sharp volume tests and entropy thresholds for quasilinear elliptic inequalities with gradient reactions

We investigate positive weak solutions of $-Δ_m u\ge u^p|\nabla u|^q$ on complete noncompact Riemannian manifolds without curvature, volume doubling, or Poincaré assumptions. For the superhomogeneous mixed reaction we derive a curvature-free integral volume test by combining two first-order estimates with a recurrence for the mass of a \emph{fixed} density. The same approach organizes the pure-gradient and capacity criteria and yields sharp finite hierarchies of iterated-logarithmic growth bounds, with counterexamples on smooth model manifolds. In the subhomogeneous region, vanishing lower growth of $\log V(R)/R^{m-q}$ forces the Liouville property. On the homogeneous line $p+q=m-1$, we obtain the optimal strict lower-volume-entropy threshold $(p+1)((m-1)/p)^{p/(p+1)}$ for $p>0$ (and $1$ for $p=0$), together with examples at equality. The existing six-region classification is not claimed as new; the advances concern refined volume hypotheses and their model sharpness. The tests remain valid for smooth weighted measures and for locally bounded positive Sobolev solutions. Sharpness is understood within the stated model families; no characterization from volume alone on an individual manifold is asserted.

math.AP↗

arXiv2026

Facet normals of Minkowski sums of staircase simplices

Staircase simplices are the simplices of the standard (Freudenthal--Kuhn) triangulation of $[0,1]^d$ and of Stanley's triangulations of order polytopes. We study the facet normals of their Minkowski sums and how the number of summands constrains the size of primitive integer normals. We characterize the normals attainable with a prescribed number of summands in terms of $d-1$ linearly independent zero-one vectors orthogonal to the normal, whose supports have an intersection graph that can be colored with one color per summand. This gives a combinatorial characterization of the least number of summands needed to realize a normal. For two summands, all primitive normals have entries in $\{-1,0,1\}$, and we determine which of them occur. For three summands we bound every entry in absolute value by $3^{(d-2)/3}$ and construct normals with entries of order $θ^d$, $θ$ the plastic number; for four summands a second construction gives entries of order $ψ^d$, $ψ$ the supergolden ratio. Since $ψ>3^{1/3}$, the largest coefficients for three and four summands are exponentially separated. For vectors with at most two distinct nonzero entry values, we determine the least number of summands exactly, using a matching-rank formula proved by exterior algebra and representations of symmetric groups. With no restriction on the number of summands, the maximum coefficient equals the classical maximum for facets of $0/1$-polytopes in the same dimension.

math.CO↗

arXiv2026

EvoRouter: Cost-Efficient Multi-Agent Routing via Insight-Guided On-Policy Distillation

Serving web queries at scale with multi-agent systems requires jointly configuring large language models, collaboration modes, agent counts, interaction rounds, and skills. However, the resulting configuration space grows combinatorially, while sparse execution rewards provide limited fine-grained supervision for interdependent planning decisions. We propose EvoRouter, a self-evolving multi-agent routing framework driven by insight-guided on-policy distillation. A dedicated expert periodically synthesizes performance-cost feedback from configuration trajectories across tasks into reusable conditional routing insights about capability requirements, returns from resource scaling, and interactions among configuration choices. A privileged teacher combines these evolving insights with current execution feedback to provide dense token-level supervision for the router's own planning trajectories through on-policy distillation. Cost-aware reinforcement learning jointly optimizes task performance and computational cost. Together, these training signals internalize execution insights into the routing policy. The updated policy generates new trajectories that support subsequent insight extraction, forming an execution-driven self-improvement loop. We evaluate EvoRouter on 19 in-domain datasets spanning five domains and two out-of-domain benchmarks. EvoRouter achieves an average in-domain accuracy of 68.69%, outperforming the strongest routing baseline by 7.23 percentage points while reducing inference compute cost by 50.95%. Our code is open-sourced at https://github.com/CenYangtze/EvoRouter.

cs.AI↗

arXiv2026

Computing Data-Aware Diagonal Frame Decompositions

We present a method for computing a specific type of matrix decomposition that combines a structured framework with data-driven adaptation. This framework allows for an efficient and stable representation of signals or images using frames by identifying sets of vectors that naturally capture the essential features of the given data. The resulting decomposition enables the development of regularization techniques that enhance the quality of reconstruction by preserving important signal characteristics while suppressing noise. We develop an optimization-based algorithm to compute the decomposition and establish convergence properties of the resulting iterates. Numerical experiments on synthetic data and tomographic image reconstruction illustrate the behavior of the proposed method and its performance compared to conventional spectral regularization methods.

math.NA↗

arXiv2026

Self-Confirming Equilibrium under Model Misspecification

We study the steady states of games played recurrently within large anonymous populations of players who receive partial ex-post feedback about the play of their opponents. Departing from the classical Bayesian account, players do not trust their conjectures as a correct description of the strategic environment but consider them as approximations of a reality that is only partially modeled and partially observed. We introduce games with feedback and uncertainty attitudes, in which the normal-form structure is enriched with a profile of feedback functions and, for each player, a preference functional that encodes aversion to misspecification and to ambiguity. The induced misspecification sensitive criterion contains the variational, maxmin, and smooth-ambiguity evaluations as special cases and collapses to expected utility when beliefs are unambiguous and the player is neutral to misspecification. The appropriate steady-state notion for such games is the Self-Confirming Equilibrium under Model Misspecification (SCEM), which requires each player to best respond to a set of structured conjectures that the feedback generated by equilibrium play cannot falsify. In contrast with the classical theory and with its ambiguity-averse extensions, mixed-strategy Nash Equilibria need not be self-confirming. We therefore develop a companion notion of Robust Nash Equilibrium (R-NE), in which every supported action remains optimal under a variational evaluation. Robust Nash equilibria refine SCEM, exist whenever the statistical distances are continuous, thereby delivering existence of SCEM, and coincide with the latter under fully informative feedback. Under observationally consistent divergences and observable payoffs, R-NE and SCEM reduce, respectively, to Nash equilibrium and Self-Confirming Equilibrium under Ambiguity (SCEA), which generalizes smooth self-confirming equilibrium.

econ.TH↗