Search arXiv⌕ Search

arXiv subjects

Yifan Chen

Publications and source records attributed to Yifan Chen.

At least 19 recordsLinked to original sources
arXiv2026

AOI: Turning Failed Trajectories into Training Signals for Autonomous Cloud Diagnosis

Large language model (LLM) agents offer a promising data-driven approach to automating Site Reliability Engineering (SRE), yet their enterprise deployment is constrained by three challenges: restricted access to proprietary data, unsafe action execution under permission-governed environments, and the inability of closed systems to improve from failures. We present AOI (Autonomous Operations Intelligence), a trainable multi-agent framework formulating automated operations as a structured trajectory learning problem under security constraints. Our approach integrates three key components. First, a trainable diagnostic system applies Group Relative Policy Optimization (GRPO) to distill expert-level knowledge into locally deployed open-source models, enabling preference-based learning without exposing sensitive data. Second, a read-write separated execution architecture decomposes operational trajectories into observation, reasoning, and action phases, allowing safe learning while preventing unauthorized state mutation. Third, a Failure Trajectory Closed-Loop Evolver mines unsuccessful trajectories and converts them into corrective supervision signals, enabling continual data augmentation. Evaluated on the AIOpsLab benchmark, our contributions yield cumulative gains. (1) The AOI runtime alone achieves 66.3% best@5 success on all 86 tasks, outperforming the prior state-of-the-art (41.9%) by 24.4 points. (2) Adding Observer GRPO training, a locally deployed 14B model reaches 42.9% avg@1 on 63 held-out tasks with unseen fault types, surpassing Claude Sonnet 4.5. (3) The Evolver converts 37 failed trajectories into diagnostic guidance, improving end-to-end avg@5 by 4.8 points while reducing variance by 35%.

cs.LG↗
arXiv2026

PICKLES: Covariance-Preconditioned Kinetic Langevin Ensemble Sampler

We propose a covariance-preconditioned kinetic Langevin ensemble sampler named PICKLES. This is a kinetic analogue of the affine invariant interacting Langevin dynamics (ALDI). PICKLES uses a rectangular covariance square root as a preconditioner to avoid computing matrix inverses or square roots, leading to position and momentum variables of different dimensions. Theoretically, the continuous-time dynamics are ergodic under mild regularity and growth conditions on the potential, assuming $N>d+1$ particles and a full-rank initial ensemble. We discretize PICKLES using BAOAB or OBABO splitting, with the latter convenient for Metropolization. For Gaussian targets at stationarity, we demonstrate that Metropolized PICKLES achieves an integration step size scaling of $d^{-1/4}$ to achieve fixed acceptance probabilities, as the ensemble size and dimension increase at a fixed ratio. This matches the scaling of single particle Metropolized Hamiltonian and kinetic samplers. Numerical experiments compare sampling efficiency and estimation error with non-preconditioned, overdamped, and single-chain counterparts, demonstrating improved efficiency per gradient evaluation. On a Dark Energy Survey Year 1 3x2pt cosmology benchmark, we obtain competitive effective sample size per gradient evaluation compared to a densely preconditioned No-U-Turn Sampler.

stat.CO↗
arXiv2026

Who Speaks, Who Authorizes, Who Acts? Understanding Trust Boundaries in Multi-Agent LLMs

Multi-agent LLM systems exchange messages before acting through tools, but message delivery, signing authority, and tool-side enforcement are separate trust boundaries. We represent them as communication, authorization, and execution structures, and use the authority cut to characterize the smallest independently compromised signer coalition that can construct a policy-valid false-claim proof under stated assumptions. Controlled paired hidden-state studies reveal four distinctions. First, no fixed-message routing contrast across four models meets the prespecified large-effect criterion; when relays rewrite messages, executed false-state action rates vary by model and route configuration. Second, presented source-identity cues do not uniformly increase protected-action selection across three models. Third, on invalid-proof cases, Nemotron and Llama request protected actions in 52.1% and 54.2% of trials, respectively, yet a deterministic tool gate commits none: request-level and commitment-level safety readouts diverge. Finally, a separate single-model authorization check shows that a false claim with a policy-valid compromised-signer proof can pass the gate, since proof validity is not factual truth. These findings motivate evaluating delivered information, qualified authorization, model proposals, and committed actions separately.

cs.AI↗
arXiv2026

Acceleration and its Limits in Gaussian Variational Inference

We study whether the classical momentum-based acceleration that improves the condition-number dependence of gradient descent from $κ$ to $\sqrtκ$ for strongly convex, smooth objectives extends to Gaussian variational inference (GVI) for strongly log-concave, log-smooth target distributions. With exact Gaussian expectations, a convex composite formulation of GVI yields the classical accelerated $\sqrtκ$ convergence rate. However, when only pointwise derivatives are available, we construct a family of strongly log-concave targets for which any randomized algorithm requires $κ^{1-o(1)}$ queries to achieve a fixed objective accuracy, even when the initial objective gap is bounded independently of $κ$. Thus curvature bounds alone cannot guarantee uniform accelerated total-query complexity. We then ask which additional structure permits acceleration and give positive examples: under a Hessian-Lipschitz assumption, variance-reduced stochastic methods yield accelerated guarantees when the Hessian variation is sufficiently small. Finally, experiments on real datasets show substantial empirical speedups from acceleration across a broad range of methods.

stat.ML↗
arXiv2026

Critical Interval MSE: Toward Reliable Offline Validation for Robot Manipulation Policies

Real-world evaluation is the gold standard for robot manipulation policies because it tests them against the physical conditions and deployment challenges they are ultimately designed to handle. However, real-world evaluation is also the bottleneck for iterating on robot imitation learning policies: it is costly, difficult to reproduce, and often too sparse to reliably compare nearby model variants. A straightforward proxy for performance is validation loss on expert demonstrations, but this proxy is often poorly correlated with real-world performance. In this paper, we introduce Critical Interval MSE (CI-MSE), an intuitively simple yet effective offline validation metric. CI-MSE uses a zero-shot agent to localize task-critical segments, restricts error computation to those segments, and applies simple action-alignment procedures that better match rollout-time behavior. Across simulation and real-world experiments, CI-MSE generally yields a stronger correlation between validation error and rollout performance than raw MSE. Across 27 distinct policy checkpoints, CI-MSE achieves a Spearman rank correlation of $-0.87$, closer to the ideal value of $-1$ than raw MSE's $-0.61$, demonstrating a significant improvement. Sensitivity analysis shows that our metric is robust across a wide range of hyperparameter settings. We further study the effectiveness of CI-MSE under evaluation distribution shifts and identify practical boundaries for its use. In summary, this paper provides a simple and reliable offline validation tool for accelerating policy iteration. Project webpage: https://ci-mse.github.io/

cs.RO↗
arXiv2026

Analytic Criteria for Detecting Bernstein-Sato Roots

We study roots of Bernstein-Sato polynomials of ideals. Our first main result gives a general partial differential equation criterion for detecting roots of the Bernstein-Sato polynomial of an ideal. For homogeneous case, we recast this criterion as an analytic condition, connecting the algebraic problem of locating roots with analytic properties of the defining data. As an application, we give a series of roots for the general determinantal ideal prove the strong monodromy conjecture for them. Keywords. Bernstein-Sato polynomial, determinantal variety, monodromy conjecture.

math.AG↗
arXiv2026

Geometric Ergodicity of Affine Invariant Ensemble Langevin and its Discrete Time Variants

Affine-invariant ensemble samplers are widely used in Bayesian applications. However, their quantitative convergence theory, in particular geometric ergodicity, remains a basic open question. We study the affine invariant ensemble Langevin dynamics, an interacting particle system that uses the empirical covariance of the whole ensemble as a preconditioner. While effective in practice, theoretical understanding of this method is not available beyond plain qualitative convergence in total variation, with a central difficulty being that the empirical covariance can become degenerate. For potentials with bounded Hessian that are strongly convex outside a ball, we prove geometric ergodicity using a novel Lyapunov function that combines the inverse covariance with a coercive exponential energy. We then show that directly applying the Euler--Maruyama scheme can diverge with positive probability, even for a one-dimensional Gaussian target. This motivates a covariance-trace regularization in continuous time. We prove geometric ergodicity of the regularized diffusion and, for sufficiently small step size, of its unadjusted Euler--Maruyama discretization. We also show that the invariant distributions of the discretization converge weakly to the product target distribution as the step size tends to zero. Lastly, we provide an explicit discretization that is both affine-invariant and geometrically ergodic, based on an adaptive step size clamping.

math.ST↗
arXiv2026

RoboCoach: World Models as Active Coaches for Compositional Robot Skills

Long-horizon robot manipulation reuses skills across many task compositions, but improving these compositions with additional end-to-end demonstrations is costly. A practical self-improving system must decide both what to teach next and where to apply that supervision. We present ROBOCOACH, a world-model-guided coaching framework that uses imagined failures to guide demonstration requests and expert updates. Its Route-Imagine-Diagnose-Improve (RIDI) loop executes reusable skill experts inside COACHWORLD, our shared action-conditioned world model, and uses a progress judge to record the first subtask that fails to complete. Aggregated records select which subtask demonstrations to acquire and which expert adapters to update. Across two simulation suites and two real-robot platforms, imagined and deployed success correlate over 22 task-policy pairs (rho = 0.840). Controlled comparisons show that our coaching method outperforms matched baselines under matched data budgets and update schedules. With only 150 additional subtask demonstrations, success rises from 13.3% to 75.0% on Franka and from 40.0% to 83.8% on AgileX. The coached experts also transfer to four held-out compositions, achieving an average success of 35.0%, compared with 0% for a shared-policy baseline updated with uniformly acquired demonstrations. Together, these results show that world models can serve as active coaches, turning imagined failures into targeted supervision for modular policy improvement. Project Page: https://robocoach-ai.github.io/

cs.RO↗
arXiv2026

Krylov complexity of a tilted extended Bose-Hubbard chain with Rydberg-dressed interactions

We investigate Krylov state complexity in a tilted extended Bose-Hubbard chain in which both the on-site interaction $U$ and nearest-neighbor interaction $V$ are present. The tilted extended Bose-Hubbard model can be realized with Rydberg-dressed interactions in optical lattices. Using exact diagonalization and Lanczos recursion from three physically motivated Fock states, we compute the time-dependent complexity, its long-time saturation value, and the fluctuations of the Lanczos coefficients for an open chain at unit filling. Crucially, we identify a pronounced quasi-chaotic diagonal regime near $U\simeq V$ in which the Krylov complexity saturation exhibits a sharp ridge while the spectral level-spacing statistics remain partially chaotic. This dissociation arises from the competition between the on-site and nearest-neighbor interaction, which generates a dense manifold of near-resonant many-body configurations that is efficiently coupled by the state-dependent dynamics but retains residual spectral structure. Our results show that Krylov state complexity provides complementary, state-sensitive information about many-body chaos that is not captured by the spectral diagnostics alone, establishing Krylov complexity as a useful diagnostic for studying thermalization and information scrambling in experimentally accessible quantum simulators.

cond-mat.quant-gas↗
arXiv2026

Physics-Guided Spectral Distillation for Underwater Image Enhancement on Resource-Constrained Devices

Underwater image enhancement is crucial for improving visual perception in marine applications. Existing underwater image enhancement studies mainly focus on enhancement quality and visual fidelity, while rarely considering real-time deployment capability, which is essential for resource-constrained underwater robots. To this end, we introduce a physics-guided spectral distillation (PSD) method, which reduces model capacity for real-time applications while maintaining the high performance of underwater image enhancement models. To decompose the outputs of teacher and student models, PSD adopts a multilevel Haar discrete wavelet transform. It transfers low-frequency color and illumination information as well as high-frequency structural details through band-specific objectives. Moreover, the distillation process of PSD is degradation-aware. We estimate degradation-aware weights through a physical head and combine them with ground-truth-guided reliability masks to selectively retain valuable teacher guidance. Experiments on the UIEB, LSUI, and EUVP datasets validate the effectiveness of the proposed method. Furthermore, we demonstrate the benefits of enhanced images for downstream perception tasks, including object detection. Deployment on a self-developed ROV further demonstrates its practical applicability in real-world underwater scenarios.

cs.CV↗
arXiv2026

ALH Gravitational Instantons and Rational Surfaces

We construct complete ALH hyperkähler metrics on complements of anticanonical elliptic curves with zero self-intersection in smooth rational surfaces. This answers the existence question of Filip--Tosatti, related to the picture proposed by Haskins--Hein--Nordström. Conversely, we prove that every hyperkähler ALH gravitational instanton admits, for each induced complex structure, a compactification by an anticanonical elliptic curve in a rational surface.

math.DG↗
arXiv2026

Cavity Multimodes as an Array for High-Frequency Gravitational Waves

Microwave cavities operated in the presence of a background magnetic field provide a promising avenue for detecting high-frequency gravitational waves (HFGWs). We demonstrate for the first time that the distinct antenna patterns of multiple electromagnetic modes within a single cavity enable localization and reconstruction of key properties of an incoming HFGW signal, including its polarization ratio and frequency drift rate. Using a 9-cell cavity commonly employed in particle accelerators as a representative example, we analyze the time-domain response of 18 nearly degenerate modes, which can be sequentially excited by a frequency-drifting signal. The sensitivity is further enhanced by the number of available modes, in close analogy to the scaling achieved by a network of independent detectors, enabling sensitivity to astrophysically plausible binary sources.

hep-ph↗
arXiv2026

Particle Physics Driven by Quantum Technology - Quantum Simulations and Quantum Sensing

We review recent advances in particle physics enabled and motivated by the rapid progress of quantum technologies. The continued development of quantum computing toward large-scale, fault-tolerant systems has the potential to address dynamical problems that remain extremely challenging for classical computational approaches, opening new avenues for studying nonperturbative dynamical processes. Tabletop detectors and quantum-enhanced technologies have reached unprecedented sensitivities, enabling novel searches for ultralight particles, gravitational waves, and other physics beyond the Standard Model in parameter regimes previously inaccessible to conventional instrumentation. Taken together, these developments highlight a rapidly evolving landscape in which quantum technologies are beginning to reshape fundamental physics. We aim to synthesize recent progress and illuminate the opportunities they present for advancing particle physics in the coming years.

hep-ph↗
arXiv2026

Compressive sensing inspired self-supervised single-pixel imaging

Single-pixel imaging (SPI) is a promising imaging modality with distinctive advantages in strongly perturbed environments. Existing SPI methods lack physical sparsity constraints and overlook the integration of local and global features, leading to severe noise vulnerability, structural distortions and blurred details. To address these limitations, we propose SISTA-Net, a compressive sensing-inspired self-supervised method for single-pixel imaging. SISTA-Net unfolds the Iterative Shrinkage-Thresholding Algorithm (ISTA) into an interpretable network consisting of a data fidelity module and a proximal mapping module. The fidelity module adopts a hybrid CNN-Visual State Space Model (VSSM) architecture to integrate local and global feature modeling, enhancing reconstruction integrity and fidelity. We leverage deep nonlinear networks as adaptive sparse transforms combined with a learnable soft-thresholding operator to impose explicit physical sparsity in the latent domain, enabling noise suppression and robustness to interference even at extremely low sampling rates. Extensive experiments on multiple simulation scenarios demonstrate that SISTA-Net outperforms state-of-the-art methods by 2.6 dB in PSNR. Real-world far-field underwater tests yield a 3.4 dB average PSNR improvement, validating its robust anti-interference capability.

cs.CV↗
arXiv2026

A Polynomial PDE Criterion for the $-n/d$ Root of the Bernstein-Sato polynomial of Homogeneous Ideals

Let $I\subseteq\mathbb C[x_1,\ldots,x_n]$ be an ideal generated by homogeneous polynomials of a common degree $d$. We give a polynomial partial differential equation criterion guaranteeing that $-n/d$ is a root of the Bernstein-Sato polynomial $b_I(s)$. We apply this criterion to the ideal of maximal minors of a generic $m\times n$ matrix and obtain the distinguished root $-n$; combined with local divisibility along determinantal strata, this allows us to obtain the strong monodromy conjecture in the maximal-minor case. Finally, we prove that the criterion is stable under enlarging the linear span of the generators, adjoining generators in disjoint variables, products satisfying the natural slope condition, and Thom-Sebastiani sums. These stability results provide new classes of homogeneous ideals and polynomials for which the distinguished Bernstein-Sato root can be detected. Keywords. Bernstein-Sato polynomial, monodromy conjecture.

math.AG↗
arXiv2026

Nearly Isotropic Vortex Solid in $\mathbf{(La,Pr)_{3}Ni_{2}O_{7}}$ Thin Films

The discovery of superconductivity in bulk bilayer nickelates has established a new platform for exploring high-$T_c$ superconductivity beyond the cuprates. The role of the Ni $3d_{z^2}$-derived $γ$ band in the superconductivity of bilayer nickelates remains unresolved. By performing simultaneous resistance and diamagnetism measurements on (La,Pr)$_3$Ni$_2$O$_7$ thin films, we map the vortex melting phase diagram for both in-plane and out-of-plane magnetic fields. For $H\parallel c$, the geometric confinement effect gives rise to pancake vortices. Remarkably, the anisotropy parameter of the vortex melting field $γ_{H_m} \equiv H_m^{ab}/H_m^c$ decreases monotonically with decreasing temperature and approaches unity at low temperatures. Within the anisotropic Ginzburg--Landau scaling, $H_m^{ab}/H_m^c = \sqrt{ρ_s^{ab}/ρ_s^c}$ tracks the superfluid-density anisotropy. Such a vortex solid implies a nearly isotropic superfluid density, which is irreconcilable with the strictly two-dimensional $3d_{x^2-y^2}$-derived bands, but naturally explained by a substantial interlayer superfluid contribution from the $3d_{z^2}$-derived $γ$ band. Our results provide thermodynamic evidence for a substantial contribution of the $γ$ band to superconductivity in bilayer nickelate thin films.

cond-mat.supr-con↗
arXiv2026

Dynamics and Frequency Conversion of Accreting Axion Clouds

Axion fields can form exponentially growing gravitational clouds around compact objects through self-interaction-driven relaxation of ambient axion waves. As the field amplitude approaches the axion decay constant, nonlinear effects become important. We identify two distinct regimes of late-time evolution, determined by the gravitational fine-structure constant and the cloud growth rate: a Bosenova regime, characterized by collapse accompanied by explosive axion bursts, and a saturation regime, in which self-interaction-induced axion emission balances accretion. In the latter regime, the emitted axion radiation exhibits stable discrete spectral lines at odd multiples of the bound-state energy, directly probing the global structure of the axion potential beyond its quadratic minimum. We show that single-cosine potentials and QCD axion-like potentials predict distinct emission spectra, enabling probes of the underlying axion self-interaction structure and its ultraviolet completion through terrestrial detection of relativistic axion fluxes from compact objects.

hep-ph↗