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 1,387 records · Page 77Linked to original sources

arXiv2026

Hardware-Attributed Operator Profiling for PyTorch

Framework profilers expose operator timing without hardware counters; GPU profilers expose hardware counters without operator attribution. Bridging this gap manually is error-prone and does not scale. We present Operator Profiler, a hardware attribution pipeline that automatically links hardware metrics to PyTorch operators via three complementary attribution paths: PyTorch profiler CUPTI correlation, NVTX temporal enclosure with per-stream interval trees, and Inductor fusion- map enrichment from debug artifacts. NVIDIA Nsight Compute (ncu) hardware counters are matched to NVIDIA Nsight Systems (nsys) kernel records via invocation-order matching, avoiding timestamp joins across incompatible clock domains. A curated 20-counter metric set with duration-weighted aggregation covers all hardware bottleneck axes, layer deduplication reduces ncu replay time by a factor of N/K for models with N layers across K unique structural classes, and GPU clock locking controls the kernel-duration aggregates used for operator-level comparison. On an NVIDIA RTX PRO 6000 Blackwell, Operator Profiler attributes 95-100% of kernel runtime for compiled workloads (GPT-2, SDPA Attention); black-box library backends such as cuDNN RNN are correctly surfaced as greater than 85% unattributed rather than silently dropped. Applied to profile-guided FX graph optimization, attributed profiles yield 1.76x-2.24x profiled-kernel-time speedups on the two compiled optimization case studies; a third LSTM diagnostic case identifies cuDNN re-dispatch as a structural fix rather than an FX graph rewrite.

cs.AR↗

arXiv2026

Nuclear Recoils from Invisible Neutron-Pair Annihilation and the LZ event

Two bound neutrons can annihilate into a single invisible scalar carrying baryon number two, producing a monochromatic nuclear recoil. Pair removal connects the $0^+$ ground states of even-even nuclei and, for several detector isotopes, leaves a stable daughter where single-neutron removal would leave a radioactive one. Below the first daughter excitation, the leading transition produces neither nuclear de-excitation nor subsequent daughter decay. We identify the corresponding mass windows in argon and xenon, including argon recoils up to $108\,\mathrm{keV}$, and obtain an approximate partial-lifetime bound of $8.5 \times 10^{26}\,\mathrm{yr}$ from DEAP-3600 data. We also discuss this mechanism as a possible origin of the $248\,\mathrm{keV}$ recoil candidate reported by LZ. A ground-state interpretation points to a scalar mass near $1.85\,\mathrm{GeV}$, where daughter excitations are also accessible, and motivates searches for correlated higher-energy recoil lines in argon. Recoil lines in different isotopes reconstruct a common invisible-particle mass. The emitted scalar can itself be dark matter, making these searches a probe of ordinary matter converting into a dark sector.

hep-ph↗

arXiv2026

Dephasing-driven suppression of superradiance and metastable dynamics in the anisotropic open Rabi model

Finite-component light-matter systems realize dissipative phase transitions in a single controllable atom-cavity setup, but how atomic dephasing - ubiquitous in real cavity- and circuit-QED devices - affects this criticality remains unknown. We study the anisotropic open Rabi model under cavity decay, spontaneous emission, and atomic dephasing together. We show that when spontaneous emission stabilizes a long-lived metastable superradiant phase, atomic dephasing actively competes with it - eroding its coherence and shortening the lifetime. This direct competition reveals that a dissipation channel's microscopic character, not its strength, controls its nonequilibrium criticality - a distinction directly tunable via independent spontaneous-emission and dephasing rates in circuit-QED and trapped-ion platforms.

quant-ph↗

arXiv2026

Optimal regularity for vectorial, higher order and non-minimizing Bernoulli problems

We prove the Lipschitz regularity of solutions for a wide class of generalized Bernoulli free boundary problems, in vectorial, higher order, and non-minimizing settings. Our method does not rely on notions of viscosity solutions or comparison methods, which allows us to reach the optimal regularity for Bernoulli-type problems associated to elliptic operators which do not satisfy any maximum principle, namely the elasticity, biharmonic and Stokes equation. With a similar method we obtain the optimal Lipschitz regularity for stationary, non-minimizing solutions of the standard Bernoulli problem in two dimensions.

math.AP↗

arXiv2026

Exponential tails for factors and the chromatic number of random graphs

The celebrated result of Johansson, Kahn and Vu determined the threshold order for clique factors in random graphs, and subsequent work identified the sharp threshold and the corresponding hitting-time phenomenon. In this paper we study the probability that there is no $K_r$-factor above the threshold and, more generally, the probability that the largest $K_r$-matching covers less than $n-s$ vertices of $G(n,p)$. For every fixed $r\ge3$, throughout the range $$n^{-2/r}(\log n)^{1/\binom r2}\ll p\ll n^{-2/(r+1)},\qquad n-s\in r\mathbb Z,\qquad s=o(n),$$ we prove $$\mathbb P\bigl(ϕ_r^s(G(n,p))=0\bigr)=\exp\left(-Θ_r\!\left((s+1)\frac{μ_r(n,p)}n\right)\right),$$ where $ϕ_r^s(G)$ is the number of $K_r$-matchings covering exactly $n-s$ vertices and $μ_r(n,p):=\binom nrp^{\binom r2}$. The lower bound is given by $s+1$ vertices which lie in no copy of $K_r$. For the upper bound we develop an iterable one-root version of the Johansson--Kahn--Vu method. As a structural consequence, we show that the remainder of $G(n,p)$ outside every maximal $K_r$-matching has an almost-perfect $K_{r-1}$-matching throughout the sparse clique window. Independently, we prove a central limit theorem for the maximum $K_r$-matching number. Combining these inputs and a structural theorem for $r=2$ from our earlier work, we prove a central limit theorem for the chromatic number of very dense random graphs: for every $r\ge2$ and $n^{-2/r}(\log n)^{1/\binom r2}\ll p\ll n^{-2/(r+1)},$ $$\frac{χ(G(n,1-p))-\mathbb Eχ(G(n,1-p))}{\sqrt{μ_{r+1}(n,p)}/r}\xrightarrow{\mathrm d}\mathcal N(0,1),\qquad\operatorname{Var}\bigl(χ(G(n,1-p))\bigr) \sim\frac{μ_{r+1}(n,p)}{r^2}.$$ This settles the Surya--Warnke conjecture throughout the interior of every clique window with $r\ge2$, strengthening its concentration prediction to a Gaussian limit with asymptotically exact variance.

math.CO↗

arXiv2026

Density-matrix quantum kinetics of spin-mode crossover and ac Edelstein response in spin--orbit-coupled chiral metals

To clarify spin relaxation and precession across the weak-to-strong SOC crossover in chiral conductors, we formulate a density-matrix quantum kinetic theory for a three-dimensional isotropic chiral metal with hedgehog SOC and nonmagnetic impurity scattering. The formulation retains interband coherence, and its collision integral conserves charge, energy, and spin during impurity scattering. We identify three spin modes that evolve continuously from a long-lived D'yakonov--Perel' relaxation mode and two strongly damped precessional modes at weak SOC to one relaxational and two coherent precessional modes at strong SOC. A closed multipole decomposition maps the spin dynamics onto an effective Bloch equation, providing a unified interpretation of relaxation and precession across the crossover. We further derive the ac Edelstein susceptibility and the reciprocal current response to a time-dependent Zeeman field, show that their poles coincide with the spin modes, and verify Onsager reciprocity. The theory thus provides a unified analytic description of spin relaxation, precession, and spin--charge conversion across the weak-to-strong SOC crossover.

cond-mat.mes-hall↗

arXiv2026

PhysPlan: Grounded Physical State Reasoning and Graph-Guided Optimization for Physically Plausible Video Generation

Video diffusion models (VDMs) synthesize photorealistic content, yet they often fail to follow the course that a physical phenomenon should take within a given scene. Recent training-free methods let a vision-language model (VLM) plan the phenomenon and guide a frozen VDM toward the plan; however, such plans are derived from the prompt and consumed as whole keyframes or trajectories, which leaves unspecified where the consequences land in the observed scene and turns incidental visual details into optimization targets. We observe that a phenomenon specified in words unfolds as sparse, local changes to the physical state of the observed scene. Building on this observation, we present PhysPlan, a training-free image-to-video framework that represents a phenomenon as a grounded state graph and uses this graph to decide what, where, and when the guidance constrains. Grounded Physical State Reasoning decomposes the phenomenon into physical deltas, each stating which objects change, to what state, and by which physical rule, and translates each delta into graph edits, verified by deterministic checks, that leave all other objects unchanged. Graph-Guided Test-Time Optimization renders a keyframe for each state, measures the denoised estimates only along the properties selected by the edits, and concentrates the update on the edited objects. On PhyGenBench and Physics-IQ, PhysPlan raises its base model from 0.52 to 0.77 and from 27.1 to 38.2, surpassing the strongest prior I2V method (0.60 and 34.6), and lowers FVD by over 20%. Project page: https://physplan.github.io

cs.CV↗

arXiv2026

Dominant Young Diagrams in Matrix Models and Partial Deconfinement

We discuss dominant representations (or Young diagrams), in thermal matrix models with gauge symmetry from the perspective of partial deconfinement. We propose a prescription for defining the dominant representations for thermal matrix models with interaction terms. As an explicit example, we consider the large-$N$ Gaussian matrix model. We obtain the Vershik--Kerov--Logan--Shepp (VKLS) shape of the dominant Young diagrams through a new analytic saddle-point analysis based on a mapping of the representation theory of U($\infty$) to free fermions in two spacetime dimensions. This computation provides a direct derivation of the previously observed functional relation between the shape of the dominant Young diagrams and the eigenvalue distribution of the thermal holonomy: the position of the complex saddle point is naturally identified with the eigenvalue. The dominant Young diagrams admit a natural interpretation in terms of partial deconfinement: the number of rows in the dominant Young diagrams matches the size of the submatrix corresponding to the deconfined subsector.

hep-th↗

arXiv2026

Large-scale bioacoustic detection using semantic segmentation: a deep learning framework applied to fin whale calls in ocean-bottom seismometer recordings

Ocean-bottom seismometers (OBS), originally deployed for geophysical research, continuously record low-frequency sound for months to years across broad areas of ocean, offering a largely untapped resource for passive acoustic monitoring (PAM) of baleen whales. Realising this potential requires automated detection methods that operate reliably across the varied conditions in large sensor networks. We present a deep learning semantic segmentation framework that detects the 20-Hz notes of fin whales (Balaenoptera physalus) in OBS spectrograms, assigning each pixel a probability of belonging to a call and converting the resulting probability maps into time-frequency bounding boxes describing individual detections. We trained the model on hydrophone data from one OBS deployment in the Azores-Madeira-Canaries region and applied it without retraining to vertical-component seismometer data from a second, geographically distinct deployment, showing that a single trained model generalises across sensor types and recording environments. Applied to 378,912 h of recordings from 46 OBS sites, the detector identified 6.3 million calls, forming the largest fin whale call catalogue assembled to date, with high precision (~97%) across both deployments. The resulting catalogue resolves call timing and spectral structure accurately enough to support ecological analyses, revealing coherent seasonal shifts in three persistent inter-note interval (INI) groups across the singing season and basin-scale patterns in calling activity. By transforming existing geophysical infrastructure into a scalable sensing network, our approach substantially expands the spatial and temporal reach of PAM without new hardware investment, offering a transferable framework for tracking other low-frequency vocalising species and informing conservation planning, marine spatial management, and abundance estimation across large scales.

physics.geo-ph↗

arXiv2026

Attention Is All You Need (to Avoid Spurious Oscillations)

Can attention move a shock across several cells in one update without breaking it? We develop a conservative, fixed grid finite-volume scheme in which a CFL-conditioned attention flux selects upstream information according to the transport required by the current time step. One-dimensional inviscid Burgers transport is used as the central mechanism test: the same learned flux remains reliable in the conventional small-step regime and, with a time step four times larger, preserves sharp shocks while using one stage per update. A standard fifth-order WENO scheme with third-order strong-stability-preserving Runge-Kutta time integration (WENO-5+SSP-RK3) is included alongside controlled Forward Euler comparisons to separate flux selection from time integration. The learned attention shifts upstream with the local transport reach and becomes more selective near shocks; inference-time interventions and retrained ablations show that transport-scale information and state-dependent selection contribute directly to performance. Directional two-dimensional scalar Burgers transport and the one-dimensional shallow-water system then test whether the conservation-scale-selection principle transfers beyond the original scalar setting. The results support attention as a learnable information stencil for conservative large-step shock transport, while identifying finite candidate reach and problem-dependent robustness as the present limits.

cs.LG↗

arXiv2026

Oblivious Self-Distance Symmetric Rendezvous on the Integer Line

Symmetric rendezvous on the line is a search problem in which two agents, initially placed at distance $2d$, must follow the same randomized strategy to meet as quickly as possible. In the standard model, agents may condition their actions on the entire execution history, and both the known- and unknown-distance variants admit expected rendezvous time $Θ(d)$. We study the role of memory by introducing oblivious self-distance strategies, in which an agent's decision depends only on her position relative to her own starting location. For an initial separation of $2d$, let $R_d$ denote the optimal oblivious expected rendezvous time in the known-distance setting. We develop two finite-state frameworks based on absorbing Markov chains. Truncated chains give computable upper bounds through finite-support strategies, while weak-peek chains give lower bounds through a revealed-information relaxation. Together, they provide a mechanism for certifying optimality. Using that mechanism, we determine $R_1$ exactly and prove that it is attained by a finite-support strategy. For $d=2,\ldots,6$, numerical optimization gives the same truncation structure and objective values, yielding rigorous upper bounds below $7.83d^2$. We do not prove that the computed weak-peek minimizers are global, but the stability of the computations leads us to conjecture that they are, in which case the corresponding truncated strategies are optimal. We also prove that $R_d=Θ(d^2)$. In the unknown-distance setting, we construct a universal strategy, independent of $d$, with expected rendezvous time $O(d^{2+η})$ for every fixed $η>0$. Thus, under the memory restriction, the known-distance rendezvous time becomes quadratic, while near-quadratic performance remains possible even without knowing $d$. The asymptotic analysis uses birth-death Markov chains and their electrical-network interpretation.

cs.DM↗

arXiv2026

How Should Reasoning Be Organized in a Transformer's Latent Space?

Continuous reasoning has emerged as a promising way to improve reasoning in large language models (LLMs). Yet we still lack a clear principle for deciding what a latent state should preserve. Reasoning by superposition shows that a single latent state can encode several search alternatives and expand them in parallel. We ask how those states should be weighted as reasoning proceeds. A natural choice is to preserve only the states active at the frontier step, since keeping every reached state appears to spread a limited hidden width too thin. We show that the opposite can hold. When later computation draws on several reached states, a cumulative state can guide attention correctly at a smaller hidden width than a frontier state that stores fewer states. At the same width, the cumulative state therefore keeps more intermediate states available for later reasoning. More generally, equal cumulative weights are optimal when future queries are unknown and remain close to the best task-specific weights when those queries are known. Experiments with two-layer and GPT-2 Transformers reproduce the predicted width advantage and show that unequal weights fail first on the states that receive the least weight. This suggests a important principle: keep reached states equally weighted, and restore equal weights as computation proceeds.

cs.AI↗

arXiv2026

Bilinear magnon--exciton coupling in biased ferromagnetic electron--hole bilayers

The hybridization of magnons and excitons would combine magnetic and optical degrees of freedom in a single composite quasiparticle. Such a hybridization is however difficult to achieve, because of their inherent energy mismatch. We propose that in biased bipolar ferromagnetic electron--hole bilayers the excitons and magnons can be brought into resonance, with the exciton energy lowered through the voltage bias to match the magnon energies. We demonstrate the linear hybridization of magnons and spin-flip excitons in this regime, starting from the microscopic exchange interactions between electrons and localized magnetic moments. We show that further increasing the gate voltage softens the hybrid magnon--exciton mode and realizes a magnon--exciton condensate, which manifests in both the magnon and exciton sectors and is associated with spin-superfluid transport. Ferromagnetic electron--hole bilayers therefore provide a new platform for the study of composite magnon--exciton quasiparticles and the realization of spinful condensates and associated spin superfluidity.

cond-mat.str-el↗

arXiv2026

Tabby: An Open Pretraining Recipe for Time Series Foundation Models

In this report, we release Tabby, a long context probabilistic time series foundation model, together with a complete and open recipe of how it was built. Tabby adopts an encoder-only patch Transformer architecture and concentrates the contributions on the data and the training procedure. The pretraining corpus combines an extended real-world collection, GIFT-Eval-Pretrain+ and BLAST, with synthetic data from KernelSynth and CauKerV2, an online generator that composes temporal dynamics through randomly sampled structural causal models. Training couples a progressive convergence schedule, which yields reusable intermediate checkpoints, with a deep quantile supervision objective for intermediate layers. The resulting 145M parameter backbone supports contexts of up to 8,192 observations and serves forecasting, classification, and anomaly detection, while a prompt-tuning module further improves in-distribution forecasting performance with the pretrained weights frozen. Tabby achieves competitive zero-shot forecasting performance on GIFT-Eval and the out-of-distribution TIME benchmark, while the same pretrained backbone also supports classification on the UCR Archive and zero-shot anomaly detection on TSB-AD-U. We release training pipeline and model as open source at huawei-noah/trustworthyAI.

cs.LG↗

arXiv2026

Limit Cycles in a Discontinuous Generalized Liénard Systems with Bivariate Perturbation

We study the number of limit cycles of the piecewise smooth differential system \[ \dot{x}=y, \quad \dot{y}=-x-\varepsilon\bigl(f(x)y+\operatorname{sgn}(x)g(x,y)\bigr),\] where $f(x)$ is a polynomial of degree $n\geq 1$, $g(x,y)$ is a bivariate polynomial of degree $m\geq 2$, and $\varepsilon$ is a sufficiently small parameter. Using the first-order averaging theorem for discontinuous systems, we obtain that the number $H_{m,n}=\lfloor n/2\rfloor+\lfloor m/2\rfloor$ is a lower bound for the number of limit cycles bifurcating from the linear center $\dot{x}=y, \ \dot{y}=-x$. The bound is sharp, and we provide an explicit example that attains $H_{m,n}$ limit cycles.

math.DS↗

arXiv2026

New Proofs of Weak Normalization for Propositional Logic

We present new proofs of weak normalization for intuitionistic and classical propositional logics (with the full set of operators -- falsum, implication, conjunction and disjunction). These proofs work with cuts rather than cut segments, and they provide explicit ``local'' rules for determining whether to contract a whole proof or reduce one of its subproofs, and in the latter case, which subproof to reduce. Interestingly, much of the complication in the case of intuitionistic logic is due to the disjunction elimination rule, while our version of the same rule for classical logic has falsum as conclusion always, and so is much easier to handle. All the complication in the case of classical logic shifts to cuts involving the reductio ad absurdum rule. We also discuss a formalization of the entire proof in Lean, and present a deterministic algorithm for weak normalization.

cs.LO↗

arXiv2026

The Matroid Secretary Conjecture is True

We resolve the matroid secretary conjecture, giving an online algorithm that accepts each element of the offline optimum with probability at least $1/4$. The algorithm only needs the number of elements in advance and independence-oracle access to subsets of already-arrived elements; it does not need to know the matroid upfront.

cs.DS↗

arXiv2026

Forecasting Coupled Dark Energy Parameters with the One-Loop Galaxy Power Spectrum

We forecast constraints on the parameters of the coupled dark energy model using DESI and Euclid data with the one-loop galaxy power spectrum. We investigate the distinguishability of our model from the zero-coupling scenario at the $1σ$ level and explore how the parameter constraints depend on the fiducial coupling $β_{\rm fid}$, the fixed potential slope parameter $ν$, the maximum wavenumber $k_{\max}$, and different prior choices. We find that the inclusion of mildly non-linear scales improves the constraints on the coupling by roughly a factor of five. Then, we address the question of which is the minimum value of $β$ that can be distinguished from zero at $1σ$. We find that for our reference case $k_{\rm max}=0.2 h/$Mpc, $β=0.15$ lies roughly $1σ$ above zero. In the most optimistic case with $k_{\rm max}=0.3 h/$Mpc and including a Planck prior on $Ω_{m0}$, this value can be reduced to $0.05$. These values are substantially larger than the current constraints on $β$, but the latter have been obtained assuming $β$ to be constant from at least the decoupling epoch to today, while we only employ late-time data. We conclude therefore that only models that allow for time-varying couplings can be detected with late-time datasets.

astro-ph.CO↗