Searcharxiv⌕ Search

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 325 records · Page 18Linked to original sources
arXiv2026

Finite-Sample Selected Covariance Spectra in Classical Shadows

Classical shadows use a single measurement snapshot to estimate many observables. These estimates are jointly distributed, and their covariance is not determined by their expectation values or individual variances. It also differs from physical correlations between observables because it depends on the measurement and reconstruction protocol. We study the covariance matrix for a fixed set of shadow estimates and its spectrum. For general shadow protocols, we give a finite-sample operator-norm bound for the empirical covariance computed using the sample mean. This bound also controls the eigenvalues and isolated spectral subspaces of the covariance matrix. For local product shadow protocols with fixed local dimension, observables supported on a bounded number of sites lead to bounds that do not depend on the total system size, provided that the number of selected observables and the relevant local reconstruction factors remain bounded. For biased local Pauli shadows, we obtain explicit finite-sample bounds and an exact covariance formula determined by Pauli compatibility and local basis probabilities. A comparison with global Clifford shadows shows that dimension-independent covariance estimation depends on the measurement protocol and does not follow from the general theory alone.

quant-ph↗
arXiv2026

Giant enhancement in spin-to-charge conversion in Bi2Se3/NiFe heterostructure via interface engineering

Topological insulators provide a promising platform for spintronic applications owing to their spin-momentum-locked surface states and efficient spin-to-charge conversion. Among these, Bi2Se3 has been the subject of intensive investigation due to its large bulk bandgap and single Dirac cone band structure. However, spin-to-charge conversion strongly depends on the quality of the topological insulator/ferromagnet interface. Here, we investigate spin transport and spin-to-charge conversion in sputter-deposited Bi2Se3/Ti/NiFe heterostructures at room temperature. The topological insulator layer is deposited on a CMOS-compatible silicon substrate. Low-temperature magnetoresistance measurements are conducted to establish the existence of a surface conducting channel in the deposited topological insulator layer. Pure spin current is injected into the Bi2Se3 layer through the titanium spacer layer via spin pumping induced by the spin precession in microwave-driven ferromagnetic resonance of the ferromagnetic film. Spin pumping studies are carried out by varying the thickness of the Bi2Se3 layer. The Bi2Se3 thickness dependence of the Gilbert damping reveals a pronounced 55% enhancement at a thickness of 4 nm, consistent with the hybridization of the top and bottom surface states of the topological insulator layer. The spin Hall angle, a parameter that quantifies the spin-to-charge conversion efficiency, exhibits an approximately one-order-of-magnitude enhancement upon insertion of the Ti spacer compared with Bi2Se3/NiFe heterostructures without the spacer. The significant enhancement of the spin Hall angle is attributed to Ti, which inhibits interdiffusion between the Bi2Se3 and NiFe layers, thereby protecting the topological surface states. Our findings highlight the role of titanium spacers in topological spintronics applications.

cond-mat.mtrl-sci↗
arXiv2026

Toward Efficient End-to-End Quantum Elliptic PDE Solvers: a Multilevel Correction Algorithm for Direct Observable Estimation

Spatial derivatives pose a basic challenge for quantum algorithms for partial differential equations: their standard discretizations have operator norms that grow as the mesh is refined, leading to large block-encoding normalization factors. Even when preconditioning improves access to the solution, estimating derivative-dependent physical observables can reintroduce polynomial dependence on the inverse mesh width. We develop a multilevel quantum algorithm for estimating linear and quadratic observables of elliptic equations that addresses this readout cost. The algorithm decomposes the output into corrections across nested discretizations and realizes fine--coarse cancellation coherently, before measurement. A Ritz-complement factorization connects the $O(h^2)$ size of the corrected response with a block encoding at the same scale, allowing this decay to offset growing readout normalizations. Under efficient quantum access to the corrected coordinates, input data, and readouts, observables with readout scale $O(h^{-χ})$, $0\leχ\le2$, can be estimated to additive accuracy $ε$ at cost $\widetilde O(ε^{-1})$ using amplitude estimation or $\widetilde O(ε^{-2})$ using direct sampling, with only polylogarithmic dependence on the inverse finest mesh width $h_L^{-1}$. We give explicit constructions of the required operator oracles for one-dimensional piecewise linear finite elements with diffusion piecewise constant on a fixed coarsest partition, and for Fourier spectral hierarchies. Output-specific bias estimates connect the discrete results to continuum accuracy.

quant-ph↗
arXiv2026

TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languages

Safety evaluation of Large Language Models (LLMs) remains heavily English-centric, leaving Low-Resource Languages (LRLs), particularly African ones, critically underexplored. We introduce TUKABENCH, a jailbreak benchmark for seven African languages that extends JailbreakBench (JBB) beyond direct translation through four settings: human translation of JBB prompts, English adaptation to African contexts followed by human translation, human-curated prompts validated through interactions with GPT-5.2, and code-switched prompts combining English and African languages, isolating the effect of language, cultural grounding, and prompt evasiveness on model safety. Across closed and open models, prompting in African languages reduces refusal relative to English, with culturally adapted prompts leading to least refusal. The evaluation also surfaces two structural limitations: model comprehension failures and reduced LLM-as-a-judge reliability in LRLs. To capture the first, we introduce Deflection alongside Refused and Jailbroken; to assess the second, we validate outputs with human annotations, showing that judge-human agreement drops in lower-resource languages and less commonly supported scripts.

cs.CL↗
arXiv2026

SAMatcher: Dense Co-Visibility Modeling via Cross-View Fusion for Scale-Imbalance Image Matching

Reliable correspondence estimation supports image processing and 3D vision tasks, including Structure from Motion, visual localization, and image registration. Wide-baseline matching is difficult under cross-view scale imbalance. The same content may appear at different resolutions and spatial extents, creating unequal observation quality. Such pairs constitute low-quality multi-view data, even when each image is clear. Image information that is distinctive in one view may be weak or unavailable in the other. Existing methods rely on pixel- or patch-level appearance and do not explicitly identify where reliable evidence is shared across views. We propose SAMatcher, a modular co-visibility framework for scale-imbalanced image matching. It uses symmetric cross-view interaction to fuse the two feature representations and identify their shared spatial support. From the fused features, SAMatcher predicts dense co-visible masks and bounding boxes. These predictions define view-specific cropping windows for downstream point-level matching. Built upon the Segment Anything Model (SAM), SAMatcher extends monocular region modeling to cross-view co-visibility reasoning. Point-sampled mask learning, box regression, and mask--box consistency jointly supervise its multi-granularity predictions. Extensive experiments follow an evaluation protocol centered on cross-view scale imbalance. The results show that SAMatcher consistently improves matching robustness across diverse matching pipelines. When integrated with RoMa, SAMatcher improves AUC@5 and mAA@5 by 5.58 and 4.97 points, respectively. The results demonstrate that cross-view fusion and dense co-visibility modeling provide reliable priors for matching under scale-induced quality imbalance. Code and project page are available at https://xupan.top/projects/samatcher

cs.CV↗
arXiv2026

A Common Antimatter Response in AMS-02 Positrons and Antiprotons

I present a common finite-history picture for the contrasting spectra of cosmic-ray positrons and antiprotons. Particles and antiparticles retain opposite Dirac phase orientations relative to ordinary matter clocks, while local interactions and positive physical energies remain unchanged. The finite-history description assigns their accumulated response a different overlap with the matter-defined Galactic environment. Its leading effect is represented by a single, approximately shape-preserving rescaling. For positrons, an illustrative overlap benchmark moves the broad structure of an empirical electron reference near 10 GeV into the few-hundred-GeV region, without introducing a separate dominant source specifically to generate that feature. A number-conserving spatial-dilution example provides an order-of-magnitude interpretation of the relative amplitude. For antiprotons, ordinary secondary production supplies an approximately power-law reference; a scale-neutral response preserves its index and gives a nearly constant antiproton-to-proton ratio. Direct comparisons with the published AMS-02 spectra illustrate these two outcomes using fixed, rounded scales. The resulting characteristic-scale displacement and spectral-index preservation provide a unified physical organization of two otherwise different antimatter observations.

hep-ph↗
arXiv2026

AgenticDiffusion: Multi-View Reasoning with View-Conditioned Diffusion Planning for Vision-Based UAV Navigation

Vision-based UAV navigation becomes challenging when navigation targets are distributed across complementary camera views and cannot be reliably observed from a single viewpoint. We propose AgenticDiffusion, an agentic multi-view UAV navigation framework that semantically coordinates first-person-view (FPV) and top-view observations for mission-level navigation. Given a natural-language instruction, AgenticDiffusion identifies the requested targets, selects the most appropriate camera view for each navigation task, determines the corresponding navigation goal, and invokes the appropriate view-conditioned diffusion planner for trajectory generation. The resulting trajectories are executed using Nonlinear Model Predictive Control (NMPC). AgenticDiffusion was evaluated in four real-world indoor scenarios, achieving an overall mission success rate of 80% across 40 physical-flight trials. In mixed-visibility scenarios, where the requested targets were distributed across FPV and top-view observations, coordinated multi-view navigation reduced average mission time by 50.8% relative to FPV-only navigation and by 26.8% relative to Top-only navigation. The semantic view-selection mechanism was also robust to lexical variation in target descriptions, achieving 100% accuracy across 66 test cases, compared with 63.64% for a confidence-based view-selection baseline. In a substantially larger Gazebo environment, AgenticDiffusion achieved a 90% mission success rate and completed the multi-stage mission, whereas the FPV-only and Top-only variants were unable to complete all requested navigation tasks.

cs.RO↗
arXiv2026

SoK: Post-Quantum Cryptography Implementation in Software: Approaches, Challenges and the PQC-HOT Framework

Secure implementation of post-quantum cryptography (PQC) requires attention to cryptographic mechanisms, software integration, developer capability, and organisational support. Understanding how available approaches address these requirements is important for preparing software systems for quantum threats. This Systematisation of Knowledge (SoK) synthesises 30 publications and analyses PQC implementation approaches and challenges using a Human, Organisational, and Technological (HOT) perspective. We identify four approach categories: guidelines, frameworks, tools and libraries, and educational interventions. Technological support receives greater representation in the extracted mapping, while no approach is classified primarily as organisational, despite secondary organisational contributions in some approaches. The challenge synthesis identifies five layers covering implementation security, system integration and lifecycle, tooling, organisational governance, and human factors. Together, these findings highlight a difference between the primary support functions of the mapped approaches and the breadth of the reported implementation challenges. We propose PQC-HOT, an evidence-informed analytical framework connecting implementation tasks with technical resources, practitioner capabilities, and organisational arrangements. We derive research priorities and engineering implications to guide framework evaluation and support secure, maintainable PQC-enabled software.

cs.CR↗
arXiv2026

Euler Scheme for Stochastic Functional Differential Equations Driven by Fractional Brownian Motion via Fractional Calculus Techniques

We study a stochastic functional differential equation (SFDE) with memory driven by a fractional Brownian motion (fBm) with Hurst parameter H>1/2. An Euler-type numerical scheme is proposed and analyzed under suitable regularity conditions on the drift and diffusion coefficients using tools from fractional calculus. We prove the convergence of the scheme and derive the corresponding rate in terms of the discretization step. Numerical simulations illustrate the theoretical results and confirm the accuracy of the proposed method.

math.NA↗
arXiv2026

Galaxy morphology dependent (black hole mass)-(velocity dispersion) relations: implications for gravitational wave forecasts and cosmological simulations

The correlation between black hole mass, $M_{\rm bh}$, and stellar velocity dispersion, $σ_0$, is revisited using 137 galaxies with quantitative bar strengths and enhanced morphological awareness. Interpreted within the `Triangal' evolutionary framework, gas-rich and gas-poor assembly pathways emerge in the $M_{\rm bh}$-$σ_0$ diagram. To quantify these scaling relations, a symmetric Bayesian hierarchical regression code, dubbed the Symmetric COvariance Population Estimator (SCOPE), is introduced. Unlike conditional estimators (e.g., LINMIX), SCOPE derives the intrinsic population covariance, natively accommodating asymmetric measurement errors while guaranteeing directional invariance between axes. Primeval, dust-poor S0 galaxies (including dwarf early-type galaxies with $R_{\rm e,gal}$ ~ 1 kpc) follow a shallow relation ($M_{\rm bh}\proptoσ_0^{2.5\text{--}3.1}$). Explained via the virial theorem, this flattening reframes expectations for intermediate-mass black holes. In contrast, tracing the `Disc Down-sizing' sequence - where dry mergers erase discs - yields a steep relation for massive elliptical and ellicular galaxies ($M_{\rm bh}\proptoσ_0^{7.8\pm1.4}$). Applying a single, monolithic scaling relation across all morphologies inadvertently averages over different formation histories, potentially skewing AGN virial $f$-factor calibrations and systematically under-predicting the ultra-massive black holes needed to generate the nanohertz gravitational wave background. Furthermore, strongly barred, dust-poor S0 galaxies appear offset to higher $σ_0$, while this dynamical signature is lost in the complexities of spiral galaxies. Ultimately, these morphology-dependent relations provide physically-motivated benchmarks for cosmological simulations and a framework for disentangling regimes driven by AGN feedback from those driven by mergers.

astro-ph.GA↗
arXiv2026

Agent Memory: Characterization and System Implications of Stateful Long-Horizon Workloads

LLM agents are increasingly deployed on long-horizon tasks requiring sustained reasoning over extended interaction histories. Realizing this at scale requires agents to persistently store, retrieve, and update their own memory across sessions. A rich ecosystem of agent memory systems has emerged spanning flat retrieval, LLM-mediated extraction, consolidating fact stores, and agentic control flows. Yet, their system-level behavior remains uncharacterized. We present the first systems characterization of agent memory. First, we introduce a system-oriented taxonomy classifying agent memory systems along four axes. Second, we build a phase-aware profiling harness attributing cost to construction, retrieval, and generation. Third, we characterize ten representative systems across two benchmark suites, uncovering how design choices shift cost across the write and read paths. Finally, we derive 10 system recommendations covering construction scheduling, capability floors, amortization via query volume, freshness-latency tradeoffs, and fleet-scale management.

cs.AI↗
arXiv2026

Beyond Agent Architecture: Execution Assumptions and Reproducibility in LLM-Based Trading Systems

Large language models (LLMs) and agentic systems are increasingly proposed for financial trading, yet their reported performance remains difficult to compare because studies vary in data provenance, temporal split discipline, execution timing, turnover treatment, and transaction-cost modeling. This article presents a targeted topical review and reproducibility audit of execution realism in LLM-based trading research. A coded evidence matrix covering 30 trade-relevant primary studies is used to assess point-in-time controls, split transparency, held-out evaluation, cost and turnover treatment, execution semantics, universe definition, and artifact release. Across the audited sample, architecture reporting is generally clearer than the evaluation assumptions needed to judge whether a trading result is economically interpretable or reproducible. A 10-equity worked example is included only as a methodological scaffold to illustrate how explicit friction and timing choices can materially compress active-strategy results. The main conclusion is that the next useful step for LLM trading research is not only better agent design, but also clearer reporting standards for execution realism, reproducibility, and evaluation comparability.

cs.AI↗
arXiv2026

Recovering the Zipfian Distribution in Unsupervised Term Discovery

Unsupervised term discovery involves segmenting unlabelled speech into word- or syllable-like units and clustering these into a lexicon of candidate types. True lexicons follow a Zipfian distribution, yet the dominant centre-based clustering approach -- K-means -- produces a more uniform distribution due to an inductive bias toward spherical clusters. In this paper we revisit graph-based clustering as a bottom-up alternative, where segment embeddings are connected by pairwise similarity and partitioned using the Leiden algorithm. We show that graph clustering substantially outperforms centre-based approaches (K-means, GMM, BIRCH) in both word- and syllable-level lexicon discovery across three languages, producing more Zipf-like distributions. Another bottom-up approach, agglomerative clustering with average linkage, also performs well, although it is computationally less efficient and allows for less control over the resulting distribution. Our work calls into question the dominance of centre-based clustering for term discovery, and promotes graph clustering as an attractive alternative.

eess.AS↗
arXiv2026

A Stochastic Maximum Principle for Partially Observed Jump-Diffusion Systems with State-Dependent Counting-Process Observations

This paper studies a partially observed stochastic optimal control problem for jump-diffusion states observed through multivariate counting processes with state-dependent intensities. Because the observation jumps carry information about the latent state through their intensities, the variational analysis involves coupled state and likelihood-ratio perturbations. By introducing a reference probability measure and augmenting the state with the counting-process likelihood ratio, we establish a stochastic maximum principle whose first-order necessary condition is expressed as a conditional Hamiltonian stationarity relation with respect to the observation filtration. Finally, an LQ example is used to examine the compatibility of a linear adjoint ansatz and to illustrate numerically the coupled partial-information stationarity system.

math.OC↗
arXiv2026

Application of the Skyrme Hartree-Fock-Bogoliubov Theory to WIMP-Nucleus Interactions in 40Ar

WIMP scattering from 40Ar is investigated using a self-consistent Skyrme Hartree-Fock-Bogoliubov (HFB) approach. Nuclear form factors relevant to dark matter direct detection are calculated from the resulting one-body density matrix elements and compared with shell-model predictions. Good agreement is found for the spin-independent response, while significant differences are observed for the spin-orbit response due to variations in single-particle occupancies. The effects of particle-number projection are shown to be small for 40Ar. These results demonstrate the sensitivity of certain dark matter response channels to the underlying nuclear structure model and establish a framework for extending mean-field calculations to nuclei beyond the reach of large-scale shell-model studies.

nucl-th↗
arXiv2026

Learning Urban Access Costs from Origin-Destination Flows via Inverse Optimal Transport

Cities deliver basic services through mixed public-private facility networks, including schools, clinics, transit providers, and subsidized service points. In these systems, planners often observe where households go, but not the latent cost function through which they trade off factors such as distance, price, and institutional access. We study this urban problem through school choice in the Philippines, where the country's largest national education subsidy is intended to redirect learners from congested public schools to participating private schools. Treating school-to-school enrollment flows as an entropic optimal transport plan, we recover latent choice costs using two complementary inverse optimal transport models: an interpretable distance-banded model with a subsidy term, and a neural cost model trained through a differentiable Sinkhorn forward pass. Applied to 283{,}016 learner trips across 23{,}820 observed flows in the most populated region, the framework estimates a subsidy-equivalent distance, $λ^{(k)}$, interpreted as the kilometers of perceived travel cost offset by the subsidy. The case demonstrates how administrative origin-destination data can be transformed into interpretable planning metrics for accessibility-aware subsidy design, facility siting, and urban service allocation.

cs.LG↗
arXiv2026

Physics-Informed Variational Quantum Classifier for Phase Detection in Strongly Correlated Matter

The characterisation of quantum phases in strongly correlated systems is a crucial milestone for the deployment of quantum sensors. In this work, we present a Physics-Informed Variational Quantum Classifier (VQC) designed to detect the topological phase transition between the Fermi polaron quasiparticle and the molecular bound state. Unlike conventional Machine Learning approaches, our quantum architecture is constructed via the Trotterised time-evolution of an effective Hamiltonian, ensuring that the learnable parameters correspond to interpretable physical quantities. We show that the VQC efficiently discovers the optimal interferometric protocol, specifically the evolution time and effective bath interactions required to maximise the visibility of Ramsey fringes, thereby clearly distinguishing the Bose-Einstein Condensate (BEC) and Bardeen-Cooper-Schrieffer (BCS) regimes. Furthermore, we report the validation of this classifier on the QRed superconducting quantum processor (BSC-CNS). Despite the intrinsic hardware noise and decoherence, the VQC preserves the relative ordering of the topological phases. We demonstrate that the physics-informed architecture achieves a linear gate complexity $\mathcal{O}(N)$, bypassing the exponential memory wall of classical simulation and ensuring scalability to many-body regimes.

quant-ph↗
arXiv2026

On a necessary condition for removing singularities of solutions of nonlinear elliptic inequalities

We study solutions of the differential inequality $$ Δ^{m / 2} u \ge f (x) g (u) \quad \mbox{in } B_1 \setminus \{ 0 \}, $$ where $m \ge 2$ is an even integer, $f$ and $g$ are some functions, and $B_1$ is an open unit ball in $R^n$, $n \ge 2$, centered at zero. Our aim is to obtain a necessary condition for a singularity at zero to be removable for any solution of this inequality.

math.AP↗