Searcharxiv⌕ Search

SEARCH · Searcharxiv

Search Searcharxiv

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 1,639 records · Page 91Linked to original sources

Hagendorf Orders

Call a linear order type $φ$ Hagendorf if it shares two properties with additively indecomposable ordinal numbers without being one itself: $φ$ is strictly indecomposable to the right and whenever $ψ< φ$, then $ψ$ can be embedded into a proper initial segment of $φ$. In the 1970s, J. Hagendorf asked whether such types exist. F. Galvin observed that they must be uncountable, and soon thereafter, J. Larson proved that they cannot be scattered. We provide a simplification of her argument that might give additional insight into the $σ$-scattered case. We then show that consistently Hagendorf types exist and that this holds in many models of set theory, for instance under $\mathsf{BA}$ or $\mathsf{MA}_{\aleph_1}$. In fact, their non-existence has large cardinal strength. Furthermore, we construct real Hagendorf types from $\diamondsuit$ and from $\mathsf{PFA}$. This marks the first progress on this interesting problem since its description by Larson four dozen years ago.

math.LO↗

Correcting CondOT: Exact Finite-Step Sampling in Gaussian Flow Matching

Flow matching generates samples by gradually transforming noise into data. In practice, using a finite number of sampling steps introduces a numerical error that depends on the chosen schedule. We study this dependence for Gaussian targets and the explicit midpoint sampling method, using the exact flow field. We measure sampling error by the squared Wasserstein distance between the target distribution and the final distribution produced by the midpoint sampler. We show that the standard conditional optimal transport (CondOT) schedule cancels the leading midpoint error and improves the general convergence bound, even when the sampling steps are unequally spaced. On a uniform grid of $S$ sampling steps, we fix the signal schedule at $α_t=t$ and prove the existence of scalar noise schedules $β_t$ that approach the CondOT noise schedule $1-t$ at rate $1/S$ and yield exact Gaussian sampling for every sufficiently large $S$. Controlled Gaussian experiments illustrate the convergence rates and exact calibration.

cs.LG↗

Towards Robust Time Series Learning via Capacity-Centric Modulation

Sample-level reliability heterogeneity is common in deep time series learning. Standard training pipelines apply a uniform regularization setting to all samples, which can under-regularize corrupted samples and over-restrict clean samples. Common robustness approaches filter observations in data space or impose priors on latent representations. We propose Capacity-Centric Modulation (CCM) as a complementary, sample-adaptive regularization principle. Under this principle, we introduce SACM (Sample-Adaptive Capacity Modulation), a task-agnostic framework that exploits spectral sparsity to assign sample-wise dropout probabilities along internal activation paths. SACM integrates into existing backbones without architectural redesign and preserves the deterministic inference pipeline. Across 301 real-world dataset-backbone pairs covering 9 forecasting, 32 classification, and 4 anomaly-detection datasets, SACM reduces forecasting MSE by 6.7% on average and improves classification accuracy and point-adjusted F1 by 3.04% and 17.05%, respectively, relative to unmodified backbones, with zero test-time overhead.

cs.LG↗

OmniReasoning: Pushing the Limits of Audio-Visual Joint Reasoning

Recent advances have enabled unified omni-modal models in understanding audio, vision, and language. However, existing benchmarks, training data, and learning methods largely treat the modalities independently, leaving the capability of audio-visual joint reasoning poorly evaluated and insufficiently elicited. We address this gap with a benchmark, data engine, and learning method. First, we introduce OmniReasoningBench, a benchmark where both audio and visual evidence are indispensable. It comprises 1,150 multiple-choice and open-ended questions across two tasks, reasoning over video and reasoning beyond video. Second, we develop a data engine OmniQA. It automatically constructs evidence-grounded QA pairs that explicitly necessitate audio-visual joint reasoning, together with time-stamped clue chains that guide the annotation of thinking process. Besides our benchmark, this engine produces training data OmniReasoning-SFT-112K and OmniReasoning-RL-19K. Finally, we propose an on-policy self-distillation method Modality-Factored Self-Distillation (MFSD). It evaluates each sampled response under modality-specific clue contexts, disentangling the contributions of individual clues and their cross-modal interactions for token-level credit assignment. With our training data and learning method, our model OmniReasoning-30B-A3B achieves 50.0% on OmniVideoBench and 42.5% on OmniReasoningBench, improving the base model Qwen3-Omni-30B-A3B-Thinking by 12.8 and 9.3 percentage points, respectively. Moreover, it delivers substantial gains on general and long-video benchmarks, including Video-MME-v2. We hope our work offers a solid step for facilitating future research in omni-modal joint reasoning.

cs.CV↗

Perspectivity and the perspective-Schröder-Bernstein Property

In this paper, we study various aspects of perspectivity in modules. We prove that the following classes of modules satisfy the perspective-Schröder-Bernstein property: modules with transitive perspectivity, quasi-continuous modules, Harada (and hence discrete) modules and quasi-discrete modules with the (finite) exchange property. Furthermore, we prove that for a semiregular ring, the Schröder-Bernstein property implies the perspective-Schröder-Bernstein property. We prove that for $A,B \subseteq ^{\oplus} M$, if all complements of $A$ are perspective with $B$, then all complements of $B$ are perspective with $A$. We also provide new characterizations of weakly perspective modules and modules in which perspectivity is transitive. Some applications of these results are given. We finally prove that perspectivity is transitive in a ring $R$ if and only if every special clean element of $R$ is perspective.

math.RA↗

Front-to-Back: Benchmarking Vision-Language Models for Asymmetric Cross-View Vehicle Re-Identification

Matching the same vehicle across front and rear cameras is difficult because the cameras do not share a view and the vehicle's appearance changes substantially. We introduce Front2Back-ReID, a benchmark of 500 manually verified vehicle handovers from 20 recording sequences in South Africa. Each example asks a model to match a vehicle highlighted in a front-camera image to the same vehicle among at least three candidates in a later rear-camera image. We evaluate seven zero-shot vision-language models, four image-retrieval baselines, and 25 human participants. Models are tested using full front RGB images, cropped target vehicles, and binary silhouettes. The strongest VLM achieved 76.6 percent Rank-1 accuracy on target crops, compared with 74.0 percent for the frozen SigLIP2 baseline; this difference was not statistically clear. Human participants achieved 94.0 percent accuracy with full images and 92.2 percent with target crops. Under our evaluation setup, enabling reasoning improved accuracy across all three input conditions for every model evaluated in both modes. We also found that VLMs generally performed worse on full scenes than on target crops. These results show that general-purpose VLMs do not yet consistently outperform strong visual retrieval for front-to-rear vehicle matching, while humans remain substantially more reliable.

cs.CV↗

Driving Spintronic Terahertz Emitters at GHz Repetition Rates

Spintronic terahertz emitters (STEs) are versatile sources of terahertz (THz) frequency radiation that offer high electric field strengths and gap-free bandwidths covering the THz (0.1-30 THz) band. In addition, they are low-cost to produce and easy to use in a variety of optical setups. However, an obstacle to their wider applicability is the recently reported decrease in optical fluence damage threshold when driven with high-repetition-rate lasers. Here, we use an asynchronous optical sampling (ASOPS) THz time-domain spectrometer (THz-TDS) to observe the degradation of the THz signal in real-time when driven by a 1 GHz repetition-rate laser. We demonstrate that the loss of THz emission is permanent and is likely the result of thermal damage to the STE structure. However, by utilizing substrates with high thermal conductivities, an increase of the damage threshold of over three orders of magnitude is observed enabling stable emission from a STE for the first time when driven by a high-repetition-rate laser.

physics.optics↗

Disentangling Self-Distillation: Measuring and Modeling Acquisition and Retention

Self-distillation with privileged context adapts a language model from demonstrations by letting the model, once conditioned on a reference response, teach its context-free copy token by token. Our taxonomy reveals existing methods differ along three entangled axes: (i) the rollout source (student or teacher), (ii) the teacher coupling (frozen, or an exponential moving average of the student at some coupling rate) and (iii) the KL direction (reverse or forward), yet these axes are usually studied in fixed combinations and have led to conflicting conclusions. We formalize a unifying framework to encompass all self-distillation methods vs classic supervised fine-tuning: we train every combination of the three axes, on Qwen2.5-7B and Ministral-3-3B across ordinary and contradictory tasks, totaling 1,200 adaptation runs, to systematically investigate the impact of the above axes. We propose a controlled model of the same objective to explain the resulting acquisition-retention trade-offs. We find that (i) the rollout source matters mostly where the task contradicts the pretrained behavior: there teacher rollouts raise acquisition well above what student rollouts achieve, with almost no change in retention; (ii) the teacher coupling changes acquisition most, on every task: acquisition rises with the coupling rate, then falls past a task-specific rate; (iii) switching the KL direction costs retention in one model but not the other so which axis to tune first depends on the model. The controlled model reproduces the three trends.

cs.AI↗

On quantum interactive proofs with a laconic prover

Interactive proof systems with a laconic prover, studied by Goldreich, Vadhan, and Wigderson (CC, 2002), capture problems verifiable with logarithmic prover communication in the classical setting. For two-message quantum analogs, even a single-bit prover response contains quantum statistical zero-knowledge ($\sf QSZK$), introduced by Watrous (FOCS 2002). However, restricting the verifier's question to classical public coins collapses the corresponding class to $\sf BQP$, as shown by Beigi, Shor, and Watrous (ToC, 2011). We further study two-message quantum interactive proof systems with a laconic prover. To this end, we introduce the class ${\sf QIP}_{\ell\text{-}{\rm bit}}(2)$, where $\ell$ is the length of the prover's response, and establish: 1. A natural complete characterization of ${\sf QIP}_{\ell\text{-}{\rm bit}}(2)$ by Multi-State Distinguishability. In particular, Quantum State Distinguishability (QSD) is ${\sf QIP}_{\rm bit}$-complete. Since QSD is $\sf QSZK$-hard, our result places ${\sf QIP}_{\ell\text{-}{\rm bit}}(2)$, for $\ell\geq 2$, in a landscape "just above" $\sf QSZK$. 2. Easy regimes for ${\sf QIP}_{\ell\text{-}{\rm bit}}(2)$ collapsing to $\sf QSZK$. We prove that QSD$[a,b]$ (and thus ${\sf QIP}_{\rm bit}[a,b]$) is in $\sf QSZK$ when $a(n)-b(n)\geq 1/O(\log n)$, and combine this with an answer compression from ${\sf QIP}_{\ell\text{-}{\rm bit}}[2,c,s]$ to ${\sf QIP}_{\rm bit}$ to obtain another easy regime when $2c>(1+2^{\ell/2})s$. Remarkably, our improved polarization applies to SD and $\sf SZK$, resolving an open problem in Sahai and Vadhan (JACM, 2003). 3. Quantum public coins also make the interaction useless: ${\sf qc}\text{-}{\sf QAM}[O(\sqrt{\log{n}})]$ with constant gap is in $\sf BQP$, where ${\sf qc}\text{-}{\sf QAM}[\ell]$ is a subclass of ${\sf QIP}_{\ell\text{-}{\rm bit}}(2)$ in which the verifier's question is exactly halves of EPR pairs.

quant-ph↗

When the Right Answer Is Missing: An Arithmetic-Dependent Rejection Bottleneck in Jev

Typed decision models such as Jev offer an efficient alternative to generative LLMs in decision-making workflows by selecting directly from predefined options. When candidate sets contain no valid answer, TypeSafe recommends including an "other" or "none-of-the-above" option to enable rejection. In this report, however, we identify an arithmetic-dependent rejection bottleneck: Jev reliably selects correct numerical answers when available but frequently accepts incorrect alternatives when they are absent despite an explicit rejection option. On paired arithmetic problems, answer-present accuracy reaches 99%, while correct rejection falls to 7%. Moreover, this gap persists across numerical magnitudes, operation depths, contextual formulations, and rejection labels, and extends to scenarios such as time calculation and capacity rounding. Yet native Boolean verification achieves 99% exact-match accuracy on the same answer-absent arithmetic cases, showing that categorical rejection can fail even when the model successfully verifies candidate correctness. Finally, we show that a simple decision threshold selected on separate development problems raises arithmetic rejection accuracy from 7% to 79% while retaining 97% answer-present accuracy, substantially mitigating the failure without retraining or additional inference.

cs.LG↗

The Geometry of Randomized Smoothing on Feasible Sets

Randomized smoothing certifies the probability of a fixed output event as the center of Gaussian noise moves. Feasibility or confidence filtering reports label probabilities only among retained proposals, producing a ratio. Its numerator is a fixed Gaussian event mass, while its denominator is the probability of retention and can change with the center. Substituting this ratio into the ordinary smoothing formula can therefore certify a ball that contains a decision boundary. We separate the problem into a geometric question and a certification question. Geometry determines when conditioning preserves Gaussian comparisons. Convex retained sets preserve the full comparison, while general sets require geometric control of the retained law as the center moves. Without such control, conditional probabilities imply no positive universal radius. Joint retention-and-label probabilities always yield a valid certificate for the same filtered predictor. A uniform covariance bound transfers divergence certificates to the retained law and can yield larger radii even when the Gaussian event comparison fails. Both methods admit finite-sample bounds. For a learned image classifier with a training-selected nonconvex filter, conditional Rényi bounds certify more images than joint-mass bounds without additional model evaluations. A released confidence filter exhibits verified label changes inside radii obtained by conditional substitution. An application of adaptive Gaussian composition covers causal finite-horizon executions with history-dependent center shifts under a pathwise energy bound.

cs.LG↗

Quantum double lock-in detection via sequential orthogonal quantum mixing

High-precision measurement of oscillating signal is a ubiquitous issue in fundamental science and a critical task in practical technologies. In quantum metrology, quantum lock-in detection provide an efficient method for measuring such signal. In general, when the initial phase of the oscillating signal is unknown,quantum double lock-in detection can effectively extract complete information about the signal's amplitude, frequency, and initial phase. Conventional quantum double lock-in detection requires two individual quantum interferometry, each of which must undergo state preparation and readout. However, the time for state preparation and readout need not be negligible in practical experiments. In particular, the time for state preparation is longer than the time for sensing. To save experimental resources, it is challenging to achieve quantum double lock-in detection just via a single quantum interferometry while still extracting complete information about the signal's amplitude, frequency, and initial phase. Here, we present a general protocol for achieving a quantum double lock-in detection just via a single quantum interferometry under a sequential orthogonal periodic multipulse sequences. In particular, if the input state is a Greenberger-Horne-Zeilinger state and two interaction-based operations are applied during interferometry, the measurement precisions for frequency, amplitude, and initial phase can both approach the Heisenberg limit. Our study paves a new way for measuring oscillating signals with a single quantum interferometry, and provides a feasible method for achieving Heisenberg-limited detection of alternating signals.

quant-ph↗

The Maximal Robust Positively Invariant Set for Linear Difference Inclusions

This article develops a systematic characterization and exact computational framework for the maximal robust positively invariant set of linear difference inclusions subject to hard state constraints. The set is characterized as the greatest fixed point of the robust predecessor operator, leading to three equivalent decreasing set iterations---the standard, self-restricted, and incremental iterations---and exact stopping tests. Under uniform exponential stability of the matrix family and bounded disturbances, the limiting disturbance-reachable set is characterized and used to derive conditions for nonemptiness and finite determination. When the state constraint set is polyhedral, the three iterations admit exact half-space implementations for bounded disturbances and finite or polytopic matrix families. Although these implementations generate the same set sequence, they distribute computational effort differently. The complexity analysis and numerical study make these differences explicit and reveal how the underlying problem structure informs the choice of implementation.

math.OC↗

Emergent Quantum Geometric Phases in Holey Graphene

In graphene and other two-dimensional materials, periodic modulations of the electron density can significantly alter the energy spectrum and transport properties. Here, we report magnetotransport measurements in encapsulated monolayer graphene with ultra-high-quality patterned periodic antidot lattices that preserve the intrinsic electronic properties of the material. This lithographically defined platform enables controlled access to commensurability and superlattice phenomena at length scales otherwise difficult to achieve. By systematically tuning the lattice dimensions, we reveal a hierarchy of classical commensurability features arising from cyclotron orbits with comparable radii that follow multiple classical trajectories, resulting in broadened resistance peaks beyond the conventional single-orbit picture. Superimposed on these features, we observe pronounced Brown-Zak oscillations arising from the quantum commensurability between the magnetic flux quantum and the unit cell of the engineered Bravais lattices. We demonstrate that the intrinsic geometric phase of our system is directly measurable and show a precise matching of the magnetic field periodicity to the lithographic periodic patterning, where moiré-like electronic spectra can be geometrically generated in single-layer graphene without the need for twist, lattice mismatch, or multilayer stacking. Our results establish nanopatterned graphene as a clean, tunable, and scalable platform for realizing and exploring moiré physics through on-demand real-space design

cond-mat.mes-hall↗

Early Planet Formation in Embedded Disks (eDisk). XXV. Inclination-Induced Minor-Axis Brightness Asymmetries Reveal Limited Dust Settling in Embedded Protostellar Disks

How and when dust settles in young protostellar disks is a key open question for the dust concentration needed to form planetesimals and, ultimately, planets. However, directly measuring the vertical dust distribution in embedded (Class 0/I) systems remains challenging. We show that brightness asymmetry along the minor axis of highly inclined disks provides a simple, powerful geometric diagnostic of vertical dust structure. Using radiative transfer modeling with RADMC-3D, we generate synthetic continuum images showing that the observed asymmetry arises naturally from disk inclination, optical depth, and dust scale height. We apply this framework to nine Class 0 and I disks from the ALMA Large Program, Early Planet Formation in Embedded Disks (eDisk), using Markov Chain Monte Carlo (MCMC) fitting. Outflow observations independently validate the inferred near- and far-side geometries: all eight sources with useful outflow constraints agree with the orientations predicted by the dust continuum modeling. Our results indicate that most embedded disks show no strong evidence of significant dust settling, with dust scale heights comparable to the gas scale height. Given that the literature indicates Class II disks tend to be well settled, our results reinforce the notion that significant dust settling occurs during the Class I phase, when deeply embedded Class 0 disks transition to their more revealed Class II counterparts. Intriguingly, the timing of dust settling appears to broadly coincide with the development of widespread dust substructures, suggesting that gravitationally driven vertical dust concentration may have triggered substructure formation.

astro-ph.EP↗

Role of flame localization in wavemaker regions on the suppression of combustion instability in turbulent partially-premixed methane flames

The interaction between flame dynamics & hydrodynamic instabilities plays a fundamental role in determining the stability of swirl-stabilized combustors. In the present study, we investigate the hypothesis that spatial overlap between flame & wavemaker region is a necessary prerequisite for combustion stability in flows characterized by a wavemaker region. To test the hypothesis, a low-momentum secondary methane injection was introduced through circumferential holes located on the centerbody of a swirl-stabilized burner. The injection velocity was maintained below 5\% of bulk flow velocity to minimize momentum-induced modifications of flow field while selectively redistributing heat release. In addition, an equivalent amount of fuel was diverted from primary fuel supply to secondary injection ports, known as fuel-staging, to isolate the effects of flame relocation from those of the total fuel flow rate. The introduction of secondary injection produced a transition of the flame from M-shaped to V-shaped structure, while fuel-staging yielded a similar flame response, demonstrating that observed behavior results primarily from redistribution of heat release. Linear stability analysis revealed that wavemaker region remains near inlet of the combustion chamber. Moreover, flame root transitions from a lifted M-flame to attached V-flame and consistently stabilizes at radial position corresponding to the identified wavemaker region. The coincidence of flame attachment & wavemaker location under stable operating conditions provides strong experimental evidence that flame stabilization is governed by their spatial overlap rather than by a modification of the underlying hydrodynamic instability. The proposed framework provides new insight into the coupling between flame stabilization & hydrodynamic instability & offers practical guidance for the design of stable, low-emission combustion systems.

physics.flu-dyn↗

Central exclusive production of $η$ and $η'$ mesons in diffractive proton-proton collisions at the LHC

We discuss central exclusive production (CEP) of $η$ and $η'(958)$ mesons in high-energy proton-proton collisions using the tensor-pomeron approach, incorporating absorption effects at the amplitude level. Model parameters are constrained using WA102 experimental data measured at $\sqrt{s} = 29.1$ GeV. We present cross sections and differential distributions for the LHC energy of $\sqrt{s} = 13$ TeV. For $pp \to pp η$, the upper limit of the total cross section is 2.5 $μ$b for $|η_{M}| < 1$ and 5.6 $μ$b for $2 < η_{M} < 5$. For $pp \to pp η'$, the predicted cross sections are 0.3--0.7 $μ$b and 0.9--2.1 $μ$b in the respective pseudorapidity regions. These results demonstrate the feasibility of studying diffractive flavour singlet pseudoscalar meson production to probe the nature of the pomeron at the LHC.

hep-ph↗

PartiCam: Camera Controlled Video Generation with Reward Guidance

We present PartiCam, a training-free Particle filtering rooted method for improved Camera controlled video generation. Generating videos that follow a precisely specified camera trajectory remains challenging for large video diffusion models. Training-free approaches are backbone-agnostic and avoid the need to construct large camera-annotated datasets by steering pretrained models toward the desired camera motion at test time. This enables the generation of camera-controlled video data that can subsequently be used to train camera-conditioned video diffusion models. Existing sampling-based guidance approaches often suffer from unstable trajectories: they either explore too broadly and fail to respect the target camera motion or collapse early and lose visual diversity over time. We introduce a global-local refinement framework for diffusion reward guidance, enabling accurate and consistent camera control during video generation. Our method builds on Sequential Monte-Carlo (SMC) guidance, but introduces a local refinement stage based on particle filtered resampling. Experiments show large improvements in camera trajectory adherence, reduced drift, and better visual quality, without requiring model retraining.

cs.CV↗