arXiv subjects

Bowen Li

Publications and source records attributed to Bowen Li.

At least 37 records · Page 2Linked to original sources

arXiv2026

Accelerated stochastic first-order method for convex optimization under heavy-tailed noise

We study convex composite optimization problems, where the objective function is given by the sum of a prox-friendly function and a convex function whose subgradients are estimated under heavy-tailed noise. Existing work often employs gradient clipping or normalization techniques in stochastic first-order methods to address heavy-tailed noise. %In this paper, we demonstrate that a vanilla stochastic algorithm---without additional modifications such as clipping or normalization---can achieve optimal complexity for these problems. In this paper, we analyze the first-order oracle complexity of vanilla stochastic algorithms---without additional modifications such as clipping or normalization---for solving these problems. In particular, we establish that an accelerated stochastic proximal subgradient method achieves a first-order oracle complexity for finding an approximate optimal solution in expectation that is universally optimal for smooth, weakly smooth, and nonsmooth convex optimization, as well as for stochastic convex optimization under heavy-tailed noise. Moreover, we derive high-probability first-order oracle complexity bounds for the accelerated stochastic proximal subgradient method under heavy-tailed and sub-Weibull noise, respectively. Numerical experiments are further provided to illustrate the numerical behavior of the methods.

math.OC↗

arXiv2026

Automatic Optical Alignment Using Projective Geometry

Aligning and maintaining complex optical beam paths is a central challenge across experimental science, because it is a high-dimensional task with strong cross-coupling between controls, often in systems with limited physical access. We present an automated hardware-software framework that resolves this alignment challenge using low-cost, retro-fittable motorized mounts driven by projective-geometry models and a photodiode-fed optimizer. A compact forward model describes the beam path to paraxial order with only the physical mirror angles left free, so it can be rapidly ($\sim$ms) numerically inverted to return the required mirror angles for a desired beam trajectory. A photodiode-fed optimizer then fine-tunes this geometric starting point, and converged mirror settings are tabulated for retrieval in milliseconds and actuation in seconds. We experimentally demonstrate the performance of this approach on a retro-reflected lattice atom-transport system, yielding improvements in both speed and precision over manual alignment. This framework reduces the manual effort required to align complex beam paths, enables programmable optical control in experiments with limited physical access, and enhances the scalability of complex optical architectures.

physics.optics↗

arXiv2026

Performance Guarantees for Data-Driven Sequential Decision-Making

The solutions to many sequential decision-making problems are characterized by dynamic programming and Bellman's principle of optimality. However, due to the inherent complexity of solving Bellman's equation exactly, there has been significant interest in developing various approximate dynamic programming (ADP) schemes to obtain near-optimal solutions. A fundamental question that arises is: how close are the objective values produced by ADP schemes relative to the true optimal objective values? In this paper, we develop a general framework that provides performance guarantees for ADP schemes in the form of ratio bounds. Specifically, we show that the objective value under an ADP scheme is at least a computable fraction of the optimal value. We further demonstrate the applicability of our theoretical framework through several applications: data-driven robot path planning, pendulum stabilization, and multi-agent sensor coverage.

eess.SY↗

arXiv2026

Zing-0.5: Toward Playable Worlds with Real-Time Joint Action and Text Control

We introduce Zing-0.5, a 5B autoregressive world model designed for playability: users can explore generated worlds, influence unfolding events, and respond to the resulting feedback through joint keyboard and online text control. Our approach brings together three technical contributions: (1) Unified action and text conditioning, combining magnitude-aware keyboard inputs with temporally aligned text instructions and jointly annotated videos to learn navigation and event control within the same sequence; (2) Event-scale supervision for incremental generation, using a segment-level teacher trained on connected multi-prompt videos to supervise a block-level causal student through distribution-matching distillation; and (3) Low-cost real-time interaction, combining four-step generation with context-preserving streaming to support 832 x 480 inference at 24 FPS at an estimated server rental cost of approximately USD 0.009 per stream-minute. Zing-0.5 achieves an overall score of 81.0 and a consistency score of 88.5 across 158 WBench Navigation cases. A joint-control demonstration shows a text-directed event change during continued navigation without restarting generation. We release the model weights, inference code, and Zing-SGLang serving implementation to support further work on playable generated worlds.

cs.CV↗

arXiv2026

Detect Anything in Graphic Design: Element-Level Rewards for Autoregressive Detection

Graphic designs, such as posters, advertisements, and infographics, are an important medium for communicating information and shaping understanding. Unlike natural images, they consist of layered elements with explicit compositional order. However, existing object detection models treat these elements as an unordered set, leaving compositional order unexploited. To address this limitation, we present Detect Anything in Graphic Design (DAD), a model that formulates graphic design detection as compositional deconstruction. It decodes elements in compositional order, using lower-layer elements to better detect higher-layer ones. The key feature of DAD is amodal detection, which predicts the full bounding box of each element, including regions occluded by elements placed above it. Building on this formulation, we propose Element Relative Policy Optimization (EleRPO), which extends GRPO from sequence-level supervision to element-level optimization. EleRPO provides fine-grained training signals that capture how each detected element contributes to overall detection quality, and works synergistically with compositional order to improve detection performance. To support training and evaluation, we build a dataset of 10 million graphic designs. Experiments show that DAD outperforms all baselines and achieves human-level performance in amodal detection, supporting effective image-to-layer decomposition. EleRPO consistently improves over GRPO across nine detection benchmarks.

cs.CV↗

arXiv2026

EEG-Driven Decoding Framework for Passenger Hazard Perception in Highly Automated Vehicles

Reliable risk assessment remains a central challenge for Autonomous Vehicles (AVs). Despite advances in automation, passenger cognition provides a non-intrusive auxiliary signal that improves both objective and perceived safety without requiring active human intervention. We introduce an Electroencephalogram (EEG)-based Brain-Computer Interface (BCI) that decodes passenger neural responses for both Risk Prediction (RP) and Danger Identification (DI), explicitly modeling humans as passengers to match real-world AV use. To achieve this, we propose the Passenger Cognitive Model (PCM), Risk-aware Sequential Labeling (RSL), and the Passenger EEG Decoding Strategy (PEDS), which integrates a 3D Convolutional Recurrent Neural Network (3D-CRNN) model for joint EEG decoding. Experimental results show that 3D-CRNN achieves a Balanced Accuracy (BA) of $95.3\% \pm 2.7\%$ in RP and improves single-subject DI from $80.9\% \pm 3.9\%$ to $85.0\% \pm 3.2\%$ with RSL. Event-wise analyses further show that 3D-CRNN consistently outperforms other models across different event types in RP and DI. In generalization experiments, 3D-CRNN achieves $77.0\% \pm 5.3\%$ BA in cross-session DI and $77.4\% \pm 1.1\%$ BA on seen subjects in cross-subject evaluation, while maintaining a $64.9\% \pm 8.5\%$ BA on unseen subjects, demonstrating promising generalizability and transferability across both intra-subject and inter-subject variability. These findings establish an Electroencephalogram (EEG) decoding framework for AV passenger hazard perception and suggest that passenger cognitive signals can provide auxiliary supervision for future AV decision-making and Safety of the Intended Functionality (SOTIF) support.

cs.AI↗

arXiv2026

Joint Accuracy and Confidentiality in Semantic-Aware Secure Remote Reconstruction

In this paper, we consider remote reconstruction over wireless networks when simultaneous accuracy at the legitimate receiver and confidentiality against eavesdropping are required. These two objectives are often treated separately, even though they arise from the same update process and are marginals of a joint reconstruction event. This paper introduces confidential reconstruction accuracy (CRA), a metric to capture the joint event in which the legitimate receiver reconstructs correctly while the eavesdropper fails. Under randomized stationary policies, we develop a three-dimensional stationary analysis and derive closed-form expressions for the long-term average CRA and the optimal transmission probability. The results show that conventional marginal analysis can misidentify the optimal policy and misestimate the achievable simultaneous accuracy-confidentiality performance. They also reveal nontrivial behaviors: more frequent transmissions or better legitimate channels do not necessarily improve joint accurate and confidential reconstruction, and when the eavesdropping channel is strong, improving the legitimate channel alone may be insufficient. Finally, the framework induces the spatial safety boundary in a geofencing setting for secure remote reconstruction.

cs.IT↗

arXiv2026

Waveguiding in systems of high contrast resonators: Theory and fast computations

In this work, we study guided modes in systems of high-contrast resonators near nonzero interior Neumann frequencies, beyond the subwavelength regime. In the regular exterior regime, where the exterior Dirichlet problem is well-posed at the reference wavenumber, we introduce an infinite-dimensional frequency-dependent capacitance operator obtained by compressing the exterior Helmholtz Dirichlet-to-Neumann map to the traces of the interior resonant Neumann eigenspaces. We prove the norm-resolvent convergence of the continuous problem to this discrete effective operator as the contrast $δ\to0$, and derive first-order asymptotic formulas for compact-defect frequencies and line-defect band functions. We then establish exponential off-diagonal decay of the capacitance coefficients by a Combes--Thomas argument, yielding an exponentially accurate truncation of the discrete operator, and show that its retained coefficients can be computed from local Helmholtz problems. At the physical frequency, this local approximation converges exponentially under a uniform stability assumption for the growing finite-cluster problems. The stability assumption can be removed by introducing a vanishing complex absorption together with a Hermitian symmetrization. In particular, an absorption parameter of order $\sqrtδ$, together with interaction truncation and patch radii of order $|\logδ|$, suffices to preserve the $O(δ^2)$ accuracy of the first-order high-contrast expansion of the defect eigenfrequencies, yielding a fast computational method. Numerical experiments for dipole and quadrupole resonances illustrate the accuracy, exponential locality, and applicability of the discrete model to straight and bent waveguides generated by material or geometric detuning.

math.NA↗

arXiv2026

Nonlinear Modal Reduction for Subwavelength Dielectric Scattering

We study three-dimensional wave scattering by high-index dielectric resonators with Kerr-type nonlinearity under plane-wave incidence, together with the associated nonlinear dielectric scattering resonances, through a nonlinear Lippmann-Schwinger equation. Using a Lyapunov-Schmidt reduction near a simple eigenmode of the Newtonian potential, we obtain a local decomposition of the scattered wave into a resonant contribution and a controlled remainder and derive an explicit nonlinear equation for the resonant modal coefficient for sufficiently small contrast parameter and locally small resonant amplitude and incident field. A second reduction covers the regime of incident fields of order one and weak nonlinearity. Our results extend for the first time the linear modal decomposition to nonlinear wave-scattering problems. We complement them by developing a numerical framework based on Nyström discretization, real Newton iteration, and pseudo-arclength continuation. For single resonators of several geometries, our computations confirm the predicted high-contrast resonance scaling and the nonlinear modal approximation, and exhibit the multivalued incident-wave response. For a mirror-symmetric resonator dimer, we track symmetric, antisymmetric, and symmetry-broken resonance branches over a broad range of separations. Our computations show that the symmetry-breaking threshold increases as the gap between the resonators decreases, in agreement with the leading-order bifurcation theory.

math.AP↗

arXiv2026

Long Directed Cycles in Vertex-Transitive Digraphs

The search for Hamiltonian cycles in vertex-transitive graphs and digraphs is a classical problem at the interface of graph theory and group theory. In the undirected setting, this goes back to the well-known conjectures of Lovász and Thomassen concerning Hamiltonian paths and cycles in connected vertex-transitive graphs. Dating back to Rankin's 1946 work, the directed analogue has an even longer history, linking the search for long cycles to classical group-rearrangement problems. Trotter and Erdős showed in 1978 that connected vertex-transitive digraphs need not be Hamiltonian. In light of this result, Alspach asked in 1981 whether there exist connected vertex-transitive digraphs whose longest directed cycle misses arbitrarily many vertices. This question was only recently resolved by Bucić, Hendrey, Mohar, Steiner and Yepremyan, who constructed connected vertex-transitive digraphs on $n$ vertices whose longest directed cycle omits $(1-o(1))\log n$ vertices. They conjectured that the number of omitted vertices can grow linearly with $n$, remarking that it would already be interesting to improve their logarithmic lower bound to a polynomial bound. In this paper, we confirm their conjecture in a strong form by constructing infinitely many connected vertex-transitive digraphs on $n$ vertices whose longest directed cycle omits at least $n/12$ vertices. In the same work, Bucić, Hendrey, Mohar, Steiner and Yepremyan also proved that every connected vertex-transitive digraph on $n$ vertices contains a directed cycle of length $Ω(n^{1/3})$, giving the first lower bound for this problem that grows with $n$. We improve this to $Ω(\sqrt n)$, matching the order of Babai's classical theorem from 1979 for undirected vertex-transitive graphs.

math.CO↗

arXiv2026

Governance at the Boundary: How Agent Decomposition Degrades Policy Compliance

Existing agent benchmarks ask whether the agent finished the task. We ask whether it finished it within policy. We introduce Fiducia-bench, a benchmark for the governability of financial agents---whether they escalate when obligated, abstain when required, and leave an auditable trail---and use it to study a question no prior benchmark addresses: does decomposing an agent into components degrade its governance? It does, and the mechanism is specific. Policy-relevant facts discovered by one component are attenuated at the handoff boundary before reaching the component that must act on them. In a 626-episode experiment across 100 KYC/AML task variants, two models, and three architectures, a 32B open-weights model attenuated 0% of discovered facts under a single-loop baseline, 56% under a fixed pipeline, and 85% under an orchestrator-subagent architecture (all at constraint distance 2). A stronger model (gpt-4.1-mini) attenuated 3-6% under the same conditions, suggesting the governance cost of decomposition is partly a function of model capability. Critically, the same mechanism produces both under-escalation and over-escalation, depending on whether the dropped fact was a risk signal or an exculpating one. The benchmark, all tasks, and the verification harness are open-source

cs.AI↗

arXiv2026

Sharp hypocoercive convergence estimates for underdamped Langevin dynamics via the modified $L^2$ method

In this note, we consider the underdamped Langevin dynamics with invariant measure $μ(\mathrm{d}x\,\mathrm{d}v) \propto e^{-U(x)-|v|^2/2}\,\mathrm{d}x\,\mathrm{d}v$. Assume that the position marginal $μ_x(\mathrm{d}x)\propto e^{-U(x)}\,\mathrm{d}x$ satisfies a Poincaré inequality with constant $m>0$, and that $\nabla^2 U\ge -K\,\mathrm{Id}$ for some $K\ge 0$. We revisit the modified $L^2$ method of Dolbeault--Mouhot--Schmeiser, employing a shifted corrector \begin{equation*} A_α=(α- L_{\mathrm o})^{-1}(L_aΠ_v)^*, \qquad α\ge0, \end{equation*} where ${L}_{\mathrm{o}}=Δ_x-\nabla U\cdot\nabla_x$ is the overdamped generator, ${L}_a$ is the generator of the Hamiltonian flow, and $Π_v$ denotes averaging over the velocity variable. We establish an explicit hypocoercive $L^2$-convergence rate $Λ_{α,γ}$ for every shift $α\ge0$ and friction coefficient $γ>0$, and show that, for each fixed $γ$, the rate is maximized at $α=0$. Optimizing further over $γ$ gives \begin{equation*} Λ_{0,γ_*} = \frac{\sqrt m} {2\left(2+\sqrt{2+\frac{2K}{m}} +\sqrt{6+\frac{2K}{m}}\right)}\,, \qquad \text{at} \quad γ_*=\sqrt{6m+2K}. \end{equation*} For convex $U$, this recovers the optimal $O(\sqrt m)$ rate.

math.AP↗

arXiv2026

Reaction-Transformation-Aware Flow Matching for Generalizable Transition State Generation

Transition-state (TS) structures define the energetic barriers and mechanistic pathways of elementary chemical reactions, yet their identification remains computationally demanding because conventional saddle-point searches require expensive quantum-mechanical calculations. Recent machine-learning approaches have accelerated TS generation by predicting structures from reaction endpoint information, but they primarily learn geometric correspondence between endpoints and TSs, leaving the structural transformations underlying elementary reactions implicitly represented. To address this limitation, we introduce TransTS, a reaction-transformation-aware framework for generalizable TS generation from atom-mapped reactant-product pairs. TransTS explicitly learns atom-level structural transformations between reaction endpoints and integrates them with a unified atom-aligned geometric representation of reactants, TSs and products, enabling reaction-aware equivariant generation of TS geometries. TransTS is designed to provide reliable TS initial guesses for subsequent quantum-chemical refinement, where generated structures are evaluated not only by geometric similarity but also by their ability to converge to validated saddle points and recover the intended reaction pathways. Across IID and zero-shot OOD benchmarks, TransTS demonstrates improved TS initialization quality, with particularly strong generalization to unseen reaction distributions. On the challenging GDB-10-rxn and GDB-17-rxn OOD benchmarks, TransTS generates TS candidates that more frequently converge to validated saddle points and recover the intended elementary reactions after refinement than existing approaches under the same training regime. Scaling reaction coverage and model capacity further improves both geometric fidelity and refinement outcomes.

physics.chem-ph↗

arXiv2026

On the Cartan Graphs of Nichols Algebras over Coquasi-Hopf Algebras

Over an algebraically closed field of characteristic zero, let $H$ be a coquasi-Hopf algebra with bijective antipode, and let $M$ be a tuple of finite-dimensional simple Yetter--Drinfeld modules over $H$. We prove that, if $M$ admits all reflections, then its associated semi-Cartan graph is a Cartan graph. We characterize the finiteness of this Cartan graph by tensor decomposability of $\mathcal B(M)$ and obtain a finite-dimensionality criterion of Nichols algebras. We also show that braided monoidal equivalences preserve reflections and the associated Cartan graphs. As applications, we prove that every Cartan graph associated with a diagonal type tuple over a finite abelian group equipped with an abelian $3$-cocycle is covered by one arising from a diagonal type tuple over some finite abelian group $G$ with trivial associator, and that their real-root sets agree at corresponding objects. We also construct a Cartan graph of the former kind that cannot be obtained from any diagonal tuple in ${}_G^G\mathcal{YD}$.

math.QA↗

arXiv2026

Frequency-dependent capacitance matrix formulation for Fabry-Perot resonances in two and three dimensional systems

We study scattering resonances of finite and infinite periodic two- and three-dimensional systems of high-contrast resonators beyond the subwavelength regime. At each fixed admissible nonzero reference frequency, we introduce frequency-dependent capacitance matrices and derive quantitative asymptotic expansions of hybridized Fabry--Pérot resonant frequencies and their corresponding eigenmodes in terms of the material contrast parameter. We provide a partial differential equation (PDE) formulation of the frequency-dependent capacitance matrix analogous to the one for the capacitance matrix in the subwavelength regime. Based on this PDE formulation, we establish key properties of the frequency-dependent capacitance matrix and estimate its norm at high frequencies for a single smooth resonator with nontrapping exterior in three dimensions, identifying a regime in which the reduced residual remains perturbative. For infinite periodic resonator arrays, we prove bandgap opening under a uniform exterior non-resonance condition and generic Dirac cones for a honeycomb scaling family. Our results extend the use of discrete approximations as a powerful tool for characterizing the resonant properties of finite and infinite periodic systems of high-contrast resonators at arbitrarily high frequencies and understanding their anomalous localization and transport properties arising from strong coupling to a discrete set of eigenfrequencies.

math.AP↗

arXiv2026

FocusMem: Factorizing Content, Readout, and Trust in Latent GUI Memory

GUI agents must remember both useful experience from earlier tasks and unfinished progress in the current interaction. Latent memory offers a compact solution by compressing multimodal trajectories into a few continuous tokens. Existing methods, however, usually map each trajectory to one fixed memory block and train it mainly through next-action supervision. This creates three practical problems: important details may be lost during compression, the same memory block must serve different decision stages, and irrelevant retrieved trajectories may still mislead the agent. We introduce FocusMem, which separates these responsibilities within a compact latent-memory interface. A role-aware content basis encourages episodic memory to retain reusable experience and working memory to retain task progress. A state-conditioned readout generates a decision-specific view of the same stored evidence, while a lightweight trust gate can suppress memory blocks that appear irrelevant to the current step. All components are trained while the GUI policy remains frozen. Across five GUI-agent benchmarks, FocusMem consistently outperforms a fully matched action-only fixed-memory baseline and prior latent memory adaptations. Further analysis shows that semantic and functional supervision preserve complementary information, state-conditioned readout is more robust as surrounding trajectory context grows, and the trust gate reduces the harm caused by injected irrelevant episodic evidence. These results show that effective latent memory depends not only on compressing past interaction, but also on what is retained, what is exposed, and what is allowed.

cs.CV↗

arXiv2026

Answer Presence Drives RAG Rewriting Gains

Retrieval-augmented QA pipelines often route retrieved passages through an LLM \emph{rewriter} before a smaller reader, lifting F1 by tens of points on multi-hop benchmarks; this gain is typically credited to improved evidence quality. We ask whether that lift is causally driven by the gold answer string appearing in the rewritten context rather than by curation per se, using a controlled intervention audit. For each rewritten context we re-run the reader after one of four controlled edits to the compile output: removing the gold answer span, replacing a length-matched random non-answer span (placebo), or injecting the gold into rewrites where it was absent (at the prefix or at a midpoint sentence boundary). Across twelve completed (cell, baseline) intervention runs spanning three reader families (Qwen2.5-7B, Qwen3.5-35B, GLM-4.7), two datasets (HotpotQA, 2WikiMultihopQA), and three compiler arrangements (MA-only, MB-only, MA$+$verify), removing the gold answer drops reader F1 by $28$ to $64$ points beyond the length-matched placebo on paired \texttt{answer-in-compile} strata, and prepending the gold into rewrites that lacked it raises F1 by $+0.7$ to $+9.7$ points in $10$ of $12$ (cell, baseline) combinations. A companion five-sentinel audit shows the conventional single-\texttt{[MASK]} probe is itself sentinel-fragile: on 2Wiki it reports a $+4.12$~F1 ``non-leakage residual'' that flips to $-3.33$ to $-7.81$~F1 under four alternative sentinels and fails an equivalence test for three of those four ($1/4$~pass). We do not propose a new rewriter or mitigation; we release the intervention runner and the sentinel panel so that other rewriter-gain claims can be tested against the same standard.

cs.AI↗

arXiv2026

Value of Communication in Goal-Oriented Semantic Communications: A Pareto Analysis

Emerging cyber-physical systems increasingly operate under stringent communication constraints that preclude reliable transmission of all available machine-type data. Motivated by this challenge, goal-oriented semantic communication advocates a minimalist design principle: transmit only what is necessary to achieve the system's goal. In this work, we formulate optimal semantic communication design as a bi-objective Markov decision process (MDP) that trades off two competing objectives: system performance and communication cost. In contrast to classical approaches that seek to optimize system performance by exhausting a prescribed communication budget, we propose a minimalist design that answers: What is the marginal value of communication, and what is the minimum communication required to achieve the goal? Our approach is based on a Pareto analysis that characterizes the complete set of policies achieving optimal tradeoffs between these two objectives. The value of communication is defined as the absolute slope of the resulting Pareto front. A key result of this paper shows that this front admits a tractable structure: it is convex and piecewise linear, and its corner points correspond to simple deterministic policies. The entire front can be constructed by mixing the deterministic policies at neighboring corner points. Leveraging these geometric properties, we introduce SPLIT, an efficient and provably optimal algorithm for computing the Pareto front. Numerical results demonstrate the efficiency of SPLIT, the diminishing returns of over-provisioning in communication, and the effectiveness of Pareto-optimal semantic communication design.

cs.IT↗