Search arXiv⌕ Search

arXiv subjects

Xi Chen

Publications and source records attributed to Xi Chen.

At least 19 recordsLinked to original sources
arXiv2026

When Do Discrete Actions Make Good Continuous Controllers?

Though modern AI models like LLMs predict discrete tokens, today's robotic foundation models represent actions as continuous vectors. In this paper, we challenge this common consensus by asking if continuous action parametrizations are superior to their discrete counterparts. We conduct an extensive comparison of single-step regression and multi-step flow variants of both continuous and discrete policies in both from-scratch single-task and multi-task VLA training, in both simulation and on real hardware. In all regimes, we find that discrete action prediction, even the most naive, uniform-bin quantization, can match or even outperform continuous-action policies in task success rate. Moreover, we find that the challenge to discrete action policies is not action precision, but rather factorization error introduced by product-distribution predictions across time and action dimensions. We show that, when paired with the correct interventions (iterative computation through discrete flows, or distribution sharpening through classifier-free guidance), discrete flow policies can turn this weakness into a strength, enabling more responsive action changes during a predicted action chunk. Lastly, we show that attempts to explicitly inject geometric information into discrete policies, e.g. through corruption processes that emulate continuous flow matching, do not yield consistent improvements. Taken together, our findings question existing assumptions that robotic action prediction requires distinct design decisions compared to other AI domains, and motivate the possibility of discrete action prediction as a compelling alternative to today's continuous action parametrizations. Project page: https://iscoyizj.github.io/discrete-action-website/

cs.RO↗
arXiv2026

What Does World Modeling Actually Improve in Robot Action Learning?

World Action Models (WAMs) combine future-video prediction with robot action learning, yet improvements in task success alone do not reveal how predictive supervision manifests in execution behavior or whether its benefits persist under task and environmental shifts. We conduct controlled future-video prediction loss ablations within FastWAM and OpenWAM, keeping model architecture, initialization, robot training data, and training steps fixed within each model family. Under clean conditions on seen tasks, predictive supervision increases success rates by 24.38 and 72.91 percentage points, respectively. Trajectory clustering further reveals higher success rates within several coarse trajectory clusters, including cases where cluster frequencies change little, indicating that the gains are not solely reflected in changes in the prevalence of trajectory patterns. However, these benefits do not consistently persist under distribution shifts: improvements on unseen tasks vary across models and tasks, while both models exhibit low success rates under the evaluated environmental randomizations, without a consistent advantage from predictive supervision. We further introduce future-video prediction as an auxiliary training objective for $π_{0.5}$, where a lightweight Transformer predictor used only during training increases success from 82.94% to 89.19%. These findings support future-video prediction as effective auxiliary supervision for action learning, while distinguishing improved execution on familiar tasks from reliable transfer and showing that such gains need not rely on a large pretrained video diffusion transformer (DiT) in the evaluated setting.

cs.RO↗
arXiv2026

Auto-Research Arena: Benchmarking and Advancing Auto-Research in Language Modeling

As autonomous research systems become more capable, benchmarks must both evaluate their performance and support meaningful research progress. We introduce Auto-Research Arena, a multi-target benchmark for evaluating autonomous research systems and discovering ideas in language modeling. Arena supports diverse research objectives within a shared experimental environment by separating admissibility from improvement: model quality and resource constraints determine whether a result qualifies, while the selected target measures progress among admissible results. We instantiate Arena in language model pretraining with nine targets spanning model, data, training, and inference efficiency, and evaluate seven research harnesses using two foundation models. Arena clearly differentiates existing harnesses across research objectives, with several sophisticated methods outperformed by our simple sequential search baseline, indicating substantial headroom in current auto-research systems. Beyond evaluation, Arena reveals a substantially broader frontier of model and system ideas aligned with active research directions. Reusing these discoveries in a new quality-driven search yields lower training loss under matched data and FLOPs and lower held-out BPB at the same model size. Together, Arena provides a common framework for benchmarking autonomous research and expanding its scope in language modeling. The code is available at https://github.com/Auto-Research-Arena/auto-research-arena

cs.LG↗
arXiv2026

Efficient Remote Sensing Image-Text Retrieval via Interaction-Free Recall and Text-Guided Reranking

Remote sensing (RS) image-text retrieval plays a critical role in understanding massive RS imagery. However, the dense multi-object distribution and complex backgrounds in RS imagery make it difficult to simultaneously achieve fine-grained cross-modal alignment and efficient retrieval. Existing methods either rely on complex cross-modal interactions that lead to low retrieval efficiency, or depend on large-scale vision-language adaptation, which requires massive data and computational resources. To address these issues, we develop a fast-then-fine (FTF) retrieval framework centered on interaction-free multi-granular recall and parameter-free text-guided cross-modal reranking with the Balanced Text-guided Interaction Block (BTIB). Specifically, interaction-free multi-granular recall pre-computes gallery embeddings and matches them with the corresponding query embedding without full-gallery token-level text-image interaction, while text-guided cross-modal reranking with BTIB refines local semantic correspondence within the top-K candidate set. Furthermore, an inter- and intra-modal loss is designed to jointly optimize cross-modal alignment across multi-granular representations. Extensive experiments on public benchmarks demonstrate that the FTF achieves competitive retrieval accuracy while significantly improving retrieval efficiency compared with existing methods.

cs.CV↗
arXiv2026

Evaluating and improving AI agents for sequencing protocol understanding

Existing scientific-agent benchmarks rarely evaluate whether agents can reconstruct the exact molecular structures and generation workflows specified by sequencing protocols, a capability critical for analyzing customized sequencing assays. We introduce LibStructBench, a document-grounded benchmark for reconstructing sequencing-library structures and molecular workflows. Our experiments show that frontier agents often produce approximately correct reconstructions but remain unreliable at exact reconstruction, particularly in molecular-state representation and workflow connectivity. Switching the underlying language model produced larger average performance differences than switching agent harnesses, with a statistically significant difference only for molecular-state reconstruction. Additional knowledge improved oligo-family recovery in a small pilot without improving mean molecular-workflow scores. Autonomous accumulation of procedural rules and cross-protocol examples showed non-monotonic validation performance and no mean held-out molecular-workflow improvement in the tested trajectory. These findings highlight the need for more controlled and verifiable approaches to scientific reasoning about molecular processes in complex experimental protocols.

cs.AI↗
arXiv2026

Temporal-Difference Calibration of Black-Box Simulators

We develop a derivative-free temporal-difference (TD) framework for calibrating black-box simulators using one-step transitions from a target system and noisy simulator-generated continuation values. The calibration parameter is defined through Bellman orthogonality, recovering the physical parameter under correct simulator specification and yielding a near-optimal $L_2$ approximation of the policy value under misspecification. We propose three TD algorithms based on random-direction finite differences, distinguished by their query costs and reuse of noisy simulator evaluations. We establish almost-sure convergence, last-iterate mean-square error bounds, and central limit theorems for Polyak-Ruppert averages across four query-noise regimes. Our analysis reveals a specification-dependent bias phenomenon: for two algorithms, the finite-difference bias of the TD estimating equation vanishes at the calibration target under correct specification, even when the gradient surrogate remains biased. This permits a fixed perturbation radius and rate exponents approaching $1/2$ in the simulator rollout budget, whereas misspecification generally recovers the classical Kiefer-Wolfowitz rates. We also develop a few-replicate online bootstrap that yields Student-$t$ calibrated confidence intervals under negligible query noise, and identify a covariance mismatch that can invalidate conventional multiplier inference along Markov trajectories. Finally, a second-stage TD recursion corrects the remaining value discrepancy without additional simulator queries.

stat.ML↗
arXiv2026

RobustRLlib: A Unified Library and Benchmark for Robust Reinforcement Learning Algorithms

Robust reinforcement learning (RL) is critical for reliable real-world deployment, where policies may fail under diverse shifts and disruptions. Existing robust RL algorithms offer a rich set of mechanisms and principles, yet two major barriers limit their broader potential: limited algorithm reusability and an unclear empirical capability boundary. We introduce RobustRLlib, an algorithm-centric library and benchmark that addresses both challenges. It integrates 17 representative robust online, offline, and safe RL algorithms under a unified interface for substantially improved reusability, together with an enhanced evaluation platform spanning diverse real-world shifts and advanced humanoid and vision-language-action tasks. Systematic evaluation over 273 perturbation configurations provides an up-to-date capability map of current robust RL algorithms, revealing where they remain effective and where substantial robustness gaps persist. As an open-source, user-friendly platform, RobustRLlib offers a reproducible and extensible codebase for applying existing methods to new problems and developing new robust RL algorithms.

cs.LG↗
arXiv2026

Numerical reconstruction of polygonal obstacles by diffracted waves

We present a numerical method for recovering rigid polygonal obstacles in two dimensions, based on the diffraction-based inverse scattering theory of Bao, Chen, Lu, and Xiong. A single pulse is emitted from a point source, and the resulting wave field is measured on a receiver curve. After isolating the diffracted signals, we recover the visible vertices from arrival times and the corner orientations and opening angles from amplitudes. The use of amplitude information permits local reconstruction in the reflection shadow, where reflected waves do not reach the receiver. The amplitudes also determine the direction of a shadowed edge adjacent to a visible vertex, enabling local non-line-of-sight imaging. Numerical experiments for a convex corner, a concave corner, a triangle, a pentagon and a star-shaped obstacle confirm that the visible portion of the boundary can be recovered stably from diffracted signals. A further convex-corner test with additive noise at level $0.2\%$ on the single arc $L_1$ confirms that the tip and opening remain recoverable.

math.NA↗
arXiv2026

DraMaze: A Multi-Agent Pipeline with Future-Aware Memory for Diverse Microdrama Script Generation

The rapid growth of microdramas has increased demand for high-quality script production. Long-form microdrama script generation with LLMs requires both diverse plot progression and long-range narrative coherence. Our observations reveal that candidate continuations increasingly concentrate on fewer narrative mechanisms as generation progresses, while conventional retrieval may reinforce trope reuse and overlook historical information essential to future plots. To address these challenges, we propose DraMaze, a multi-agent pipeline that explores diverse causal paths through trope memory-guided evolutionary tree search and mutation. It further constructs a Backward Narrative Dependency Tree to organize, retain, and retrieve historical memories based on future narrative goals. We also introduce MiDramaBench, a bilingual benchmark for evaluating script logic and creative quality across different script lengths, together with causal-pattern metrics for narrative repetition and diversity. Experiments show consistent improvements across three backbone LLMs. For 60-episode scripts of up to 60k words, DraMaze reduces causal-pattern repetition by 32.1% and achieves a Logic score of 82.5, outperforming the best-performing baseline by 8.5 points.

cs.CL↗
arXiv2026

Infinite cascades for the defocusing cubic half-wave equation on the torus

We construct global solutions of the defocusing cubic half-wave equation on the torus whose $H^s$ norms grow exponentially for every $1/2<s<3/2$. Such solutions exist at every positive mass. To the best of our knowledge, this provides the first example of a genuine infinite cascade for a defocusing fractional nonlinear Schrödinger equation on the torus. In particular, our result resolves an open question posed by Gérard and Grellier concerning the existence of unbounded Sobolev trajectories for the defocusing cubic half-wave equation on the torus.

math.AP↗
arXiv2026

Beyond Experience: Knowledge-Guided Evolution for NPU Kernel Generation

LLM agents have become capable kernel engineers on mainstream GPU stacks, but they struggle on emerging accelerators such as Ascend NPUs, whose programming model is barely represented in pretraining data. Recent NPU agents address this gap largely through experience: they accumulate memory from their own generation and optimization trials and reuse it on later tasks. We argue that experience alone is not enough for kernel generation. An agent can only remember what it has tried, and it rarely tries a mechanism it does not know. As a result, such memory offers little help when a new operator calls for a different mechanism, and even familiar operators raise unseen situations as optimization proceeds. We present ASCENDEVOLVE, which grounds every step of an evolutionary kernel search in a large external knowledge base built from platform documentation, toolkit headers, and operator repositories. Because this corpus far exceeds any context window, ASCENDEVOLVE indexes it and routes each agent to a small, task-specific slice of evidence instead of the corpus itself. The retrieved knowledge seeds diverse implementation strategies, guides repair when a candidate fails, and constrains how strong candidates are recombined. On 53 Ascend C operators from CANN-Bench spanning four difficulty levels, ASCENDEVOLVE achieves an overall score of 71.39, improving over Reflexion and OpenEvolve by 11.6% and 17.2%.

cs.AI↗
arXiv2026

Cumulative locality governs collective quantum charging

Quantum batteries can outperform their classical counterparts through collective quantum effects, yet the conditions enabling such advantage remain unclear. Using a generic class of free-fermionic chargers we prove that global operators are insufficient for scalable superextensive charging and that a cumulative hierarchy of many-body operators is required, with the structure of the battery determining the degree of advantage. This is obtained by relating the charging power to the norm of the commutator between the battery and charger Hamiltonians and deriving an upper bound only capturing noncommuting terms. Moreover, we revisit the concept of participation number and propose an effective definition that accounts for the relative strength of the contributing terms, correlating it with the charging power scaling. Not limited to quadratic Hamiltonians, our results establish cumulative locality as a key resource for quantum charging advantage.

quant-ph↗
arXiv2026

Herculean: An Agentic Benchmark for Financial Intelligence

As AI agents advance, the central question is no longer whether they can solve isolated financial tasks, but whether their capabilities generalize across complete professional workflows. Existing benchmarks cannot answer this question because they evaluate either static competencies or agents within a single interaction structure. We introduce Herculean, a benchmark of end-to-end agentic financial intelligence that crosses agent architecture, backbone model, and four structurally distinct workflows: Trading, Hedging, Market Insights, and Auditing. Each workflow is implemented as an MCP skill that preserves its native state, actions, constraints, and outcome criteria, allowing intact agent systems to be evaluated under matched external conditions. Across 30 architecture--backbone configurations, we find that agent capability is not globally ordered. Workflow rankings are weakly aligned, changing the architecture frequently reverses performance under the same backbone, and each architecture concentrates failure in different forms of workflow control. These results establish that agent performance is an interaction between architecture and workflow demand, rather than an intrinsic property of the underlying model. Consequently, success on a single financial workflow provides little evidence of dependable, general financial intelligence.

cs.AI↗
arXiv2026

StopWatch: Goal Grounding for Object-goal Navigation

Pretrained VLMs bring rich world knowledge and semantic priors to object-goal navigation, and fine-tuning a VLM into a navigation policy has become a common recipe. Despite this progress, success rates remain low, and it is unclear which capability these fine-tuned policies lack. Our diagnosis identifies goal grounding as the dominant failure. These policies explore well, yet often fail to notice that the goal is already in view or to estimate how far away it is. To address this failure, we ground the goal explicitly. Explicit grounding substantially improves success and reaches state-of-the-art results on six benchmarks across three scene collections. We also correct annotation defects in these benchmarks, so that future methods are evaluated more faithfully. Our analysis on the clean benchmarks shows that, with ground-truth grounding, our policy nearly matches the expert that generated its demonstrations, leaving the headroom in perception rather than policy learning. Our work points to goal grounding as a missing piece of object-goal navigation and offers a simple and effective solution to the task.

cs.CV↗
arXiv2026

AlgoPlasma: Open Algorithms for Plasma Modeling

AlgoPlasma is an open-source library in which core numerical algorithms for plasma modeling are implemented as modular, well-documented, and independently testable components. Rather than offering a complete simulation code, it allows researchers to select, adapt, and assemble the required components into application-specific workflows. The current release is centered on particle-based simulation, while AlgoPlasma is designed to encompass a broader range of approaches to plasma modeling. It provides components for particle initialization and advancement, particle-grid coupling, field solution, collision modeling, neutral transport, selected continuity updates, probe signal analysis, and input/output, together with Message Passing Interface (MPI) exchange routines and selected Open Multi-Processing (OpenMP) kernels. Documentation links mathematical formulations to source implementations, interfaces, and usage, while verification and validation cases evaluate numerical accuracy and physical behavior. AlgoPlasma thus establishes a shared algorithmic foundation for plasma modeling, transforming repeatedly reimplemented numerical methods into open, reusable, tested, and explainable components for research, verification, education, and collaborative development.

physics.plasm-ph↗
arXiv2026

VeriVLA: Adaptive Test-Time Verification for Flow-Based VLAs with Latent World Models

Flow-based Vision-Language-Action (VLA) policies are inherently stochastic: different noise seeds can produce action chunks of varying quality from the same observation. At test time, however, the policy typically commits to a single sample without an explicit estimate of its quality. In fine-grained manipulation, executing a suboptimal action can irreversibly disrupt task progress and lead to failure. To exploit policy stochasticity, we introduce a training-free two-axis trigger that uses two initial samples to measure within-chunk temporal variation and cross-sample variance, detecting the critical states. Once triggered, a JEPA-based latent world model ranks candidate action chunks by predicting action-conditioned future latent states and the corresponding changes in task progress. By operating in latent space, the verifier avoids pixel reconstruction and enables efficient and more accurate ranking. Across three real-world tasks, $\textit{i.e.,}$ pen insertion, pipetting, and apple placement), our method raises the average completion score from 50.8% to 65.5%, while adding 326 ms for sampling and verification per triggered decision on a single RTX 4090. Consistent gains are also observed in simulation, where the average success rate rises from 66.4% to 73.6% in the clean setting of RoboTwin 2.0, from 62.8% to 69.6% in its randomized setting, and from 96.9% to 97.6% on LIBERO.

cs.RO↗
arXiv2026

Failure of fixed-profile modified scattering and generalized modified scattering for the 1D defocusing cubic NLS

We study the long-time behavior of the one-dimensional defocusing cubic nonlinear Schrödinger equation with initial data in $L^2(\mathbb R)$. We first prove that the standard fixed-profile modified-scattering ansatz fails at the unweighted $L^2$ endpoint for the one-dimensional defocusing cubic nonlinear Schrödinger equation. More precisely, there exists a real-valued datum $$u_*\in L^1(\mathbb R)\cap\bigcap_{m\in\mathbb N}H^m(\mathbb R), \qquad xu_*\notin L^2(\mathbb R).,$$ for which the corrected Fourier profile has no strong $L^2$ limit. Secondly, for every $L^2$ datum, we prove generalized modified scattering: in the physical ray coordinates $x=2tz$, the rescaled solution admits a strong $L^2$ asymptotic representation whose amplitude is the square root of the conserved transmission density and whose unit complex phase is determined by the initial datum.

math.AP↗
arXiv2026

Fast spectral methods for Volterra integral equations

We develop novel methods for the fast construction of spectral approximations to smooth linear and weakly singular Volterra integral operators and for the fast evaluation of smooth nonlinear Volterra integrals. The proposed methods employ the Duffy mapping, which transforms the original indefinite integral into a definite integral and maps the original causal triangular domain to a square, thereby avoiding any artificial extension of the kernel function. For smooth linear and weakly singular Volterra operators, the resulting approximations are almost-banded and lower-banded, respectively. The $N\times N$ principal finite section of these approximations can be constructed in $O(N)$ and $O(N^2)$ flops, respectively, via recurrence relations. For Volterra integrals with a smooth kernel that depends nonlinearly on a given function, recurrence relations lead to fast evaluation with a complexity of $O(N^2\log N)$ when the function is approximated by a degree-$N$ Chebyshev approximant. These spectral approximations and evaluation procedures give rise to fast and stable spectral methods for solving Volterra integral equations, as well as a broad class of problems involving Volterra integral operators. Extensive numerical experiments show that these spectral methods significantly outperform existing methods in terms of stability, speed, accuracy, and ease of implementation.

math.NA↗