Search arXiv⌕ Search

SEARCH · Search arXiv

Search Search arXiv

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

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,153 records · Page 64Linked to original sources
arXiv2026

Revisiting $B_c^{(*)}$ production in $Z$-boson decays at NLO in QCD

We study $B_c^{(*)}$ production in $Z$-boson decays at next-to-leading order (NLO) in QCD within the color-singlet nonrelativistic QCD framework. The analysis covers both the $1S$ and $2S$ pseudoscalar and vector states, extending the phenomenological study to radially excited $B_c$ mesons. We calculate direct production widths and energy-fraction and invariant-mass distributions, including both charm- and bottom-quark production topologies. For the central inputs and $μ_R=2m_b$, the $K$ factors are $1.52$ for $B_c$ and $1.33$ for $B_c^*$. We examine uncertainties from the renormalization scale, heavy-quark masses, and wave functions at the origin, and apply the principle of maximum conformality to fix the renormalization scale from the $β_0$-dependent NLO contribution. The NLO corrections enhance the small- and intermediate-energy-fraction regions and modify the invariant-mass spectra. We discuss these features in relation to heavy-quark fragmentation. The results provide production-rate and kinematic inputs for studies of ground-state and radially excited $B_c$ mesons at high-statistics $Z$ factories.

hep-ph↗
arXiv2026

Neural Computation Runs on Dissipative Structures Under Cyclic Calibration: A Systems Perspective on Sleep

Biological function operates far from thermodynamic equilibrium, sustained by metastable structures that continuously cycle through synthesis, function execution, and degradation--consuming energy and exporting entropy in the process. Neural computation exemplifies this dissipative organization, relying on low-entropy structures engaged in dynamic processes that define key information transfer and processing parameters such as gain, threshold, and integration time. This perspective proposes that sleep serves as a systemic recalibration phase required to reset crucial signaling parameters that drift during dissipative processing. During wakefulness, neural activity progressively consumes and disrupts the organization of the very substrate from which it emerges, leading to parameter drift and reduced computational precision. Sleep maintains computational accuracy across activity cycles by recalibrating function-structure coupling--that is, by counterbalancing entropic drift. Rather than explaining the experimentally observed sleep need through homeostatic regulation of unknown physiological variables, this perspective reframes it as an emergent property of neural computation under thermodynamic constraints.

q-bio.NC↗
arXiv2026

Approximating the Value of Time-Bounded Weighted Timed Games

The Value Problem for weighted timed games, an extension of timed automata with weights and adversity, consists in determining, given a weighted timed game with a reachability objective and a rational threshold, whether or not the value of the game exceeds the threshold. This problem is known to be undecidable for weighted timed games, i.e., there is no algorithm that can compute the value exactly. A natural alternative is to approximate the value; however, the Approximation problem is also undecidable in general. In this paper, we prove that the Approximation problem is decidable in 4-ExpTime for weighted timed games whose executions are time-bounded.

cs.GT↗
arXiv2026

Smoothness of Shortest Isogenies to Frobenius Conjugates

Let $E$ be chosen uniformly from the geometric isomorphism classes of supersingular elliptic curves over $\overline{\mathbb{F}}_p$ and $δ_E$ be the minimum degree of isogenies from $E$ to its Frobenius conjugate $E^{(p)}$. We prove the smoothness of $δ_E$ and hence affirm Wesolowski's $p^{1/3+o(1)}$ attack on the supersingular endomorphism ring problem is unconditional.

math.NT↗
arXiv2026

Dynamical Conservation Laws in Quantum Spin Systems

We study dynamical conservation laws and the propagation of thermodynamic structure in translation-invariant lattice quantum spin systems. Using Lieb-Robinson bounds [CMP 28 (1972), 251-257] in arbitrary dimension, and Araki's bounds [CMP 14 (1969), 120-157] in dimension one, we prove conservation of the specific entropy, specific energy, and specific relative entropy under suitable assumptions on the interactions. The transported local Hamiltonians admit surface energies that vanish per unit volume in the van Hove limit, and local observables remain a core for the transported generators at every finite time. Weak Gibbsianity is preserved for sufficiently small times in arbitrary dimension and for all times in dimension one. These results provide the analytical foundation for the structural theory of approach to equilibrium developed in a companion paper, and establish partial results toward the conjectures formulated there.

math-ph↗
arXiv2026

Hidden-charm and hidden-beauty hybrid mesons

The mass spectra of $Π_u$, $Σ_u^-$, $Π_g$, $Δ_g$, $Σ_g^-$, $Δ_u$, and $Σ_u^+$ hybrid charmonium and hybrid bottomonium states are computed using a non-relativistic potential model. The hybrid potential is modelled as a linear combination of quarkonium potential and gluonic excited potential. The gluonic excited potential is computed by fitting the potential difference between hybrid potential and quarkonium potential obtained from the lattice quantum chromodynamics. Experimental masses and $J^{PC}$ values of $Ψ(4230)$, $Ψ(4500)$, $X(4500)$, and $Ψ(4660)$ are compatible with hybrid charmonium states, and $Υ(10753)$ and $Υ(11050)$ are compatible with hybrid bottomonium states obtained from our model, respectively.

hep-ph↗
arXiv2026

NEST: Stale-Robust Tokenized-Stock Measurement for a Crypto-Treasury LETF Stack

Tokenized equities trade while listed markets are closed, so their marks could improve off-hours measurement along layered exposures that link Bitcoin, crypto-treasury stocks, and leveraged ETFs. A token return and the next opening gap, however, share the price move from the previous close to the token mark, so part of their apparent fit is mechanical. We develop NEST, a stale-robust protocol that separates this shared-history fit from conditional measurement value. A shared-history identity and a variance-clock bound show how much fit timing alone can generate. The partial determination index (PDI), the squared partial correlation between token and gap given public controls, measures what remains, and a finite-sample frontier expresses the partial-F decision in PDI units. Product, provenance, timestamp, and execution audits decide whether an informative mark is admissible. For MSTRx, a tracker certificate on MicroStrategy (MSTR), a token-only $R^2$ of 0.916 shrinks to a modest increment once Bitcoin, mark age, and a pre-09:00 MSTR premarket proxy are controlled: the token explains 15.7% of the remaining gap variation (1% frontier 0.031), and 14.0% to 17.0% under stricter clock variants, including one in which every token mark precedes 09:00. A trade-level reconstruction on Gate gives 25.2%; NVDAX and AAPLX clear their frontiers, whereas TSLAX does not. An admitted mark updates, rather than replaces, BTC-unit exposure diagnostics for the BTC-MSTR-LETF chain.

q-fin.TR↗
arXiv2026

MetaExp: Learning to Self-Improve through Experience via Verifiable Cross-Task Environment Scaling

Language-model agents can accumulate interaction feedback, but reliably turning that feedback into reusable experience remains challenging. Existing methods either optimize external experience scaffolds or train experience-related behavior on specific task distributions, leaving open whether experience learning itself can become a transferable model capability. Training such a capability calls for supervision beyond isolated tasks, which mainly support in-episode problem solving rather than revealing what knowledge remains useful across episodes. In this work, we introduce MetaExp, a scalable framework that trains this capability through verifiable task families with controlled cross-task structure. Each family preserves a reusable factor while varying task-specific realizations, enabling transfer without direct solution reuse. We propose two complementary designs: Shared-Tool families keep black-box tools fixed across tasks to elicit operational experience, while Shared-Workflow families preserve latent procedures across changing realizations to elicit procedural experience. We scale environments under both designs, collect and filter sequential interaction trajectories, and fine-tune Qwen3.5-122B-A10B on 4,956 trajectories from 1,000 task families. With parameters frozen during downstream evolution, MetaExp improves by +8.5 points on SpreadsheetBench, peaks at +2.3 on APEX-Agents, and improves by +14.8 on AutomationBench. The gains persist under domain-excluded training, supporting experience learning as a transferable capability for continued non-parametric improvement after deployment.

cs.AI↗
arXiv2026

Phase Kernel Lifts Capacity of Dense Associative Memory

Dense associative memory underlies modern attention mechanisms; however, Modern Hopfield Networks (MHNs) with linear kernels yield overlapping attractor basins and spurious states under high memory load, limiting their theoretically exponential storage capacity in practice. We construct a class of nonlinear phase kernels that modify the energy landscape and retrieval dynamics, and prove that these kernels preserve exponential storage capacity and local fixed-point convergence under given conditions on kernel smoothness, memory separation, and memory distributions. The framework is instantiated by the Sigmoidal Phase Hopfield Network (S-Hop), which guarantees monotone energy descent and mitigates practical capacity loss. S-Hop has $8.3\times$ and $4\times$ gains in one-step noisy retrieval load over Uniform Hopfield Networks (U-Hop) on CIFAR-10 and TinyImageNet respectively, and reaches a $50\times$ gain in first-failure load for ImageNet. The results demonstrate that phase kernels provide a principled route to robust dense associative memory.

cs.LG↗
arXiv2026

Continuous representations of the meta Hahn algebra and special functions

Infinite families of biorthogonal rational functions are constructed using infinite dimensional representations of the three-generated meta Hahn algebra. Continuous Hahn and continuous dual Hahn polynomials are also recovered along the way using an embedding of the Hahn algebra inside the meta Hahn. Rational functions and polynomials arise as transition coefficients between eigenvectors and generalized eigenvectors, and their algebraic properties follow naturally from the relations of the meta Hahn algebra. Finally, biorthogonality relations are established for the rational functions introduced in the paper.

math.RT↗
arXiv2026

Identification of ultra-high-energy neutrinos in vertical air showers with surface detectors

We investigate the potential of surface detector arrays to identify ultra-high-energy neutrinos using vertical air showers with zenith angle $θ\leq 60^\circ$, a region largely unexplored in current searches. The method exploits the enhanced electromagnetic component at ground level arising from energetic leptons produced in neutrino interactions. Using detailed air-shower and detector simulations, we show that observables sensitive to the electromagnetic and muonic content provide genuine discrimination between neutrino-induced and hadron-induced showers. This is demonstrated both with idealized shower quantities and with realistic detector-level signals based on water-Cherenkov and scintillator detectors. The corresponding neutrino identification efficiency is used to derive the effective mass and, primarily, the effective area of the array under background-rejection requirements appropriate to targeted multi-messenger searches. Although not competitive with inclined-shower searches under the stringent background rejection factors required for diffuse-flux analyses, this approach establishes the feasibility of a vertical detection channel, extending the field of view of surface arrays and providing a complementary tool for multi-messenger observations of transient sources.

astro-ph.HE↗
arXiv2026

OPRD: On-Policy Reminder Distillation for Long-Horizon Coding Agents

Reliable repository-level coding requires agents not only to generate correct code but also to preserve constraints and respond to tool feedback across long interactions. However, our diagnostic analysis shows that agents often fail to act on relevant information already available in context, while final Pass/Fail outcomes provide only sparse process supervision. We introduce On-Policy Reminder Distillation (OPRD), which combines Multi-Scope Reminder supervision with outcome-based routing on reminder-free student rollouts. Successful rollouts provide autoregressive self-imitation targets, whereas failed rollouts receive Reminder-conditioned reverse-KL supervision from a teacher that re-evaluates the same student-generated prefixes. A Global Reminder converts final-test diagnostics into trajectory-level guidance, while Turn-local Reminders apply only to the next decision following the relevant environment observation. The teacher and student share a two-stage initialization from Coding SFT and Instruction-Following (IF) SFT. At inference, the student operates without Reminders. On SWE-bench Verified, SWE-bench Pro, and SWE-bench Multilingual, OPRD improves the benchmark-averaged task resolution rate by 11.3 points over the open-source Qwen3.5-4B base model. Under the same diagnostic protocol, it reduces the aggregate error-event count from 446 to 222, a 50.2% reduction.

cs.CL↗
arXiv2026

Complete homogeneous varieties over a field

The main objects of this paper are complete homogeneous varieties under the action of a semisimple group $G$ over field $k$ of characteristic $p>0$. When the isotropy groups of the geometric points are smooth parabolics, these are classically known as twisted flag varieties and are well studied in the literature. If the smoothness assumption is removed, we get a larger class of varieties which we call as \emph{twisted parabolic varieties}. The main goal of this paper is to study the structure, classification and properties of these varieties. We first understand their structure over a separable closure of $k$ by classifying non-smooth parabolics containing a Borel; this builds on the recently completed classification of such parabolics over an algebraically closed field. Then, via Galois descent we get a combinatorial criterion for such a variety to descend over $k$. We compute the Picard group of twisted parabolic varieties by embedding them into products of twisted flag varieties, study their local rigidity properties and show that they are rational if they have a $k$-rational point for $p \geq 3$. When $k$ is perfect, we show many interesting properties such as existence of a universal purely inseparable cover, transitivity of $k$-rational points under $G(k)$ as well as the existence of a smooth parabolic in the stabilizer of every such point. However, these properties fail if $k$ is imperfect and we demonstrate this with illuminating examples.

math.AG↗
arXiv2026

GRASE: Geometry-native Real-to-sim Agentic Scene Modeling from Video

Real-to-sim conversion aims to turn a captured video of a real environment into an editable 3D scene while preserving the geometry and layout of what was observed. Existing methods often follow a one-pass, loosely coupled paradigm: each stage receives only limited feedback from previous stages, allowing errors to propagate without reliable correction and leaving the final scene inconsistent. We present GRASE, an agentic framework that closes this loop by treating the captured multi-view video observations as a standing reference throughout scene modeling. Instead of committing to intermediate estimates, GRASE revises its current hypothesis whenever rendered evidence disagrees with the capture, overcoming the fixed intermediate assumptions of conventional pipelines. Specifically, GRASE first fits the architectural shell and its openings from depth and multi-view cues. An agent then places generated assets in a shared coordinate frame while checking their spatial relationships. Finally, each asset's geometry and placement are refined iteratively to improve agreement across the observed views. By coupling these steps through feedback, GRASE localizes structural discrepancies and revises only the responsible components, while preserving cross-view consistency. The resulting 3D scenes are faithful to the input, consistent across views, and suitable for simulations. In our experiments, GRASE substantially outperforms other state-of-the-art methods, including a direct real-to-sim conversion using GPT6-Astra.

cs.CV↗
arXiv2026

Polarization Properties of Thomson-Scattering Triaxial Ellipsoids

Polarimetry is a powerful tool for constraining the geometry of distant astrophysical objects. Continuum polarization induced by Thomson scattering can probe the photospheric geometry of transients and provide insight into their explosion mechanisms. We systematically study continuum polarization from spheroids over a broader range of deformations and viewing angles than considered previously, and generalize the calculation to triaxial ellipsoids. We find that strongly deformed spheroids exhibit a non-monotonic dependence of polarization on viewing angle, a behavior overlooked in previous studies. Triaxial ellipsoids show more diverse polarization signatures, but their qualitative behavior can be understood from the area of surface elements viewed nearly edge-on, where the local polarization is strongest. Our results provide a simple physical understanding of the dependence of polarization on photospheric geometry and help constrain the geometry of astrophysical transients through polarimetric observations. The computational results of this study will be made publicly available for use in future research.

astro-ph.HE↗
arXiv2026

When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers for Context-Parallel Prefill

Reusing the attention KV cache across requests is a standard way to cut prefill cost in long-context LLM serving, and production systems increasingly back the per-engine cache with a remote tier shared by every engine. We study such a tier in a disaggregated deployment of GLM-5.3, an FP8 mixture-of-experts model with multi-head latent attention (MLA), on 8xB300 nodes running prefill context parallelism (PCP8). Enabling the remote tier (LMCache with a Valkey store) left median time-to-first-token unchanged but raised the 16k-token p99 from 3.96 s to 17.3 s. Replaying a production request stream on an isolated engine, we find that a remote hit took 0.88 s against 0.31 s to recompute the prompt, and that synchronous loading turned a hit-heavy stream into an unstable queue: p99 83.32 s with the GPUs 7% busy. A two-term cost model explains both: a hit pays off only if the tier delivers more than B* = sR (2.53 GB/s for this model and GPU), and a fixed per-hit engine cost (T0 ~ 0.33 s) means that hits on prompts shorter than about 17k tokens cannot beat recompute at any bandwidth. Guided by the model, we (i) shard the MLA KV store across PCP ranks, with a proof that the existing minimum-over-ranks lookup returns the longest cached prefix and a load protocol that can never desynchronize collectives; (ii) partition the remote tier by rank, one server instance per rank with no coordination, which raises GET bandwidth 6.3x to 10.76 GB/s; (iii) add asynchronous loading to the sharded store; and (iv) fix three client bugs that caused unbounded stalls. On the hit-heavy stream, with 98.5% of prompt tokens served from the remote tier, p99 falls from 83.32 s to 0.67 s. The sharded store raises prefill throughput by 14%, and by 34% together with an upstream kernel fix. All changes are submitted upstream to LMCache.

cs.DC↗
arXiv2026

Ehrhart positivity for paving positroids

A central conjecture of Ferroni, Jochemko, and Schröter posits that every positroid is Ehrhart positive. While Ehrhart positivity has recently been verified for notable classes such as lattice path matroids and series-parallel matroids, the general conjecture remains open. In this paper, we prove that the base polytope of any paving positroid has positive Ehrhart coefficients. This establishes a new class of Ehrhart positive positroids and provides substantial evidence supporting this conjecture.

math.CO↗
arXiv2026

ISW-Agent: Integrating LLM-Driven Dialogue with n8n Workflow for Internal Solitary Wave Remote Sensing Image Processing

Internal solitary wave (ISW) remote sensing requires task parameters and spatial information to be transferred across image acquisition, detection, stripe extraction, and propagation prediction. Here, we present ISW-Agent, which integrates large language model (LLM) interpretation with n8n workflows and independent scientific models through structured task constraints, geospatial transformations, and stage-specific output validation. In tests of 120 Chinese instructions per backend, GPT-5.6 Sol and DeepSeek both achieved a strict task-match rate of 94.2%, with all intents and execution decisions matching the reference. Sentinel-1 cases demonstrated continuous processing from dialogue requests to wavefront prediction for the specified scene. Of 50 timed scenes, 39 completed detection, extraction, and prediction with a median turnaround time of 48.04 s; 11 stopped after empty extraction. Across 3,408 controlled fault-injection runs, retries recovered 234 transient-failure conditions and output validation rejected 234 structural defects, while 61 provenance or completeness errors indicated the need for content-consistency checks. Overall, ISW-Agent thus connects natural-language requests to traceable spatial processing chains for multimodel ISW analysis.

cs.AI↗