arXiv subjects

Xi Chen

Publications and source records attributed to Xi Chen.

At least 37 records · Page 2Linked to original sources

arXiv2026

Component-wise outer similarity of turbulent boundary layers with strong adverse-pressure gradient

The adverse-pressure-gradient (APG) effect challenges the classical scaling of turbulent boundary layers (TBLs) over a flat plate, and it remains unclear whether outer-layer similarity holds across a wide range of APG strengths.We assemble a benchmark comprising the present direct numerical simulation of a suction--blowing-induced separation bubble and published profiles, spanning Clauser parameters $β_e=0.04$--$180$ and Reynolds numbers $Re_θ=530$--$19000$. Across this broad $(β_e,Re_θ)$ parameter range, existing APG normalizations do not simultaneously collapse the two mean-velocity defects and four Reynolds-stress components, leaving open whether outer similarity persists. Instead of defining new scalings, we normalize each quantity by its value at a common location $y_s$. Results are shown using $y_s=0.5δ$, beyond the outer stress peaks. For each of the six quantities, the normalized profiles exhibit a common outer shape across the conditions, with collapse emerging for $y/δ\gtrsim0.2$ for the velocity defects and $y/δ\gtrsim0.4$ for the Reynolds stresses. The collapse-error analysis shows that outer self-similarity is retained over a broad range of $y_s$ within the outer region and is not restricted to $y_s=0.5δ$. Furthermore, explicit expressions are provided for the normalized outer profiles. These results support component-wise outer similarity under strong APGs and provide compact references for APG turbulence-model assessment and conditional profile reconstruction.

physics.flu-dyn↗

arXiv2026

OmniVBench: A Benchmark and Large-Scale Dataset for Omni Reference-to-Video Generation

Omni reference-to-video (R2V) generation is emerging as a general paradigm that enables flexible video customization from diverse and compositional visual references. However, existing benchmarks fall short of these emerging capabilities: their test cases cover only a limited range of reference types and compositions, and their evaluation protocols largely assess holistic reference consistency, overlooking whether reference factors are properly preserved, disentangled, and routed. Meanwhile, the high cost of constructing omni R2V training data makes suitable training resources scarce. To address these gaps, we introduce OmniVBench and the Omni-R2V Dataset, establishing a shared foundation for evaluating and training omni R2V models. OmniVBench expands R2V evaluation across broader reference types, richer reference compositions, and more flexible instruction-driven customization, covering 7 task families and 18 fine-grained tasks. Complementing this task design, we introduce factor-grounded evaluation with 12,172 case-specific checklist items, explicitly assessing whether intended reference factors are faithfully preserved, correctly disentangled and bound to their targets, and realized according to the instruction. We further introduce the Omni-R2V Dataset, bringing industrial-grade training resources for diverse R2V tasks to the broader research community. Drawing primarily on a large-scale corpus of professional video footage, it comprises 340K processed training samples spanning diverse reference types and multi-reference compositions. Specifically, we develop task-specific pipelines for reference-target pair construction, offering a practical and scalable recipe for omni R2V data construction. Extensive evaluation of advanced open- and closed-source R2V models reveals substantial performance variation across diverse tasks and distinct capability patterns across evaluation dimensions.

cs.CV↗

arXiv2026

Scattering for the KP-II equation

We prove scattering in $L^2(\mathbb{R}^2)$ for real-valued solutions of the Kadomtsev--Petviashvili II (KP-II) equation with localized initial data. No smallness assumption is made, no zero mean condition is imposed, and the admissible data include all Schwartz functions. The proof uses the inverse scattering transform attached to the heat operator. It rests on three ingredients: an a priori bound for the scalar $\bar\partial$ problem which depends on the size of the scattering coefficient but not on its oscillating phase, a nonlinear Plancherel identity proved by analytic continuation from small data, and an estimate showing that the nonlinear part of the reconstruction is negligible on the propagation cones of free waves. To the best of our knowledge, this is the first result establishing scattering for the KP-II equation for every real-valued Schwartz initial datum.

math.AP↗

arXiv2026

Coupled Harness--Model Evolution for Scientific Equation Discovery

Large language models (LLMs) are increasingly used to discover scientific equations from experimental data. Such a discovery system has two parts that must work together: a library of executable computational tools and the model's tool-use policy. The tools supply the solving capability, from diagnosing structure in the data to fitting candidate equations; the policy plans and orchestrates their use. Yet existing systems adapt only one part, keeping either the toolset or the model fixed, which leaves the corresponding failures uncorrectable. We develop a coupled evolution framework that improves both parts from the same verifiable fit-error signal. In the tool-evolution phase, an agent grows an executable tool library from a small set of generic numerical operations by diagnosing its own failures: it authors a new tool when a capability is missing and revises its selection rules when the right tool went unused. In the policy-evolution phase, a small open model (roughly 3B active parameters) is fine-tuned on successful tool-use trajectories collected with the frozen library, on training formulas screened to be disjoint from all evaluation benchmarks. Ablations show the two phases are complementary: with no tool library at all the same model solves almost nothing, the tools account for most of the gain, and trajectory training adds a further gain that is largest on the most difficult problems. The full system solves 79.1/71.3\% of LSR-Synth problems at fit tolerances $10^{-2}$/$10^{-3}$, reaches 63.8\% pooled accuracy on the distributionally different SRSD-Feynman panel, and obtains a normalized CP3-28 aggregate of $-7.27$. These results suggest a broader recipe for scientific agents: let the model and its instruments improve each other under verifiable feedback.

cs.AI↗

arXiv2026

BRIDGE: Selective Boundary Repair with Dependency Graph Evidence in Multi-Agent Systems

Multi-agent collaboration has emerged as a critical paradigm for tackling complex, long-horizon tasks; however, while coordination protocols continue to advance, systemic failures at the inter-agent collaboration interface remain a pervasive bottleneck. In collaborative workflows, systems frequently suffer from \emph{broken handoffs}---where critical task context is omitted, misrouted, or ignored despite each agent emitting locally plausible responses. To address this challenge, we present \method, a verifiable runtime control framework that treats handoff repair as selective intervention over explicit dependency graphs. Central to \method is an integrated fault-attribution and calibration pipeline, \dataset, which decouples execution dependencies and injects counterfactual handoff perturbations to provide empirical supervision for runtime intervention. Leveraging this diagnostic foundation, \method dynamically compiles observable interaction traces into a provenance graph to pinpoint unresolved dependencies, without accessing private agent states. It then evaluates candidate repairs through an empirical gate that weighs expected gains against disruption risks, and executes repairs only when formal local postconditions hold. Extensive evaluations across demanding benchmarks demonstrate that \method boosts task success by up to 12.3 absolute percentage points on SILO-BENCH and 11.1 on TeamBench. Across both backbones, \method maintains strict cost-effectiveness, adding as little as 2.0\% marginal token overhead. Furthermore, on 562 handoff-relevant traces from the MAST benchmark, \method achieves 69.8\% F1 in a single pass, establishing a reliable operational boundary for verifiable coordination repair.

cs.CL↗

arXiv2026

FedGraph: Defending Federated Large Language Model Fine-Tuning Against Backdoor Attacks via Graph-Based Aggregation

Federated fine-tuning of large language models (LLMs) allows multiple clients to collaboratively adapt models without sharing raw data, offering strong privacy benefits. However, this setting is highly vulnerable to backdoor attacks, in which malicious clients manipulate their updates to induce targeted behaviors while maintaining normal performance on benign inputs. In this work, we make a key observation specific to federated LLM fine-tuning: malicious clients tend to exhibit strong consistency in their model updates, whereas benign clients produce more diverse and dispersed updates. This structural difference is not well captured by existing defense methods, which typically rely on direct anomaly detection in high-dimensional parameter space and therefore struggle in the presence of entangled LLM updates. Motivated by this observation, we propose FedGraph, a graph-based aggregation framework that explicitly models relationships among client updates. By representing clients as nodes in a dynamically constructed graph and extracting structural features that characterize their relative positions and interactions, FedGraph derives compact client representations that effectively separate malicious and benign behaviors. These representations are then used to identify and suppress malicious contributions during aggregation. Extensive experiments demonstrate that FedGraph significantly improves backdoor defense. It consistently reduces attack success rates to below 10% while achieving high detection accuracy (95.5% on average) and a low false positive rate (2.33%), substantially outperforming existing federated learning defenses.

cs.CR↗

arXiv2026

Rollout-Time Intervention: Unlocking Agentic Reinforcement Learning Beyond Policy Limits

Reinforcement learning (RL) has emerged as a key paradigm for improving large language model (LLM) agents, enabling stronger reasoning, planning, and tool-use in interactive environments. However, existing agent RL methods remain fundamentally constrained by an on-policy reasoning bottleneck: when tasks exceed the model's current capability boundary, effective trajectories become extremely sparse and training often collapses. We introduce Rollout-Time Intervention (RTI), a technique designed to elicit successful trajectories beyond the policy's intrinsic exploration capacity. RTI operates by leveraging reward-side evidence (i.e., evaluation rubrics) during rollout to convert erroneous trajectories into successful ones, providing effective learning signals for tasks that the policy rarely solves on its own. Although intervention trajectories are inherently off-policy, we find that they can effectively guide policy optimization when trajectories with excessive distributional shifts are filtered out. Extensive experiments across five benchmarks demonstrate that RTI consistently outperforms RL baselines, enabling the post-trained policy to approach the performance of substantially stronger contemporary LLMs and, more importantly, transcend its original capability boundary. The code is available at https://github.com/Alibaba-NLP/qqr.

cs.AI↗

arXiv2026

DepthTile: Depth-Guided Existing-Image Self-Tiling with Input-Side Tileability Assessment

Existing-image self-tiling aims to transform a given image into a seamlessly repeatable tile while preserving its original semantics and local details. This task requires weak editing to retain the source but leave periodic seams, whereas strong editing improves tileability at the large risk of content drift.Existing diffusion-based tiling pipelines alleviate boundary discontinuities, yet lack awareness of which regions are geometrically important to preserve. In this work, we propose DepthTile, a depth-guided self-tiling method whose generation module requires no retraining of the diffusion backbone. It converts monocular depth into a spatial preservation prior and integrates it into the timestep-dependent masking mechanism of Tiled Diffusion. DepthTile converts depth into a spatial editability map and fuses it with the differential diffusion mask. This strengthens source preservation in near-field regions indicated by the monocular depth prior while retaining editing flexibility near compatible background boundaries. To access when self-tiling is feasible, we introduce the Self-Tiling Ability Score (STAS), a lightweight input-side predictor based on RGB, gradient, depth, and depth-gradient cues. On a 335-image benchmark, relative to the default Tiled Diffusion pipeline, DepthTile reduces MAE, LPIPS, and Tiling Score by 12.8%, 26.7%, and 8.4%, respectively. STAS reliably enables difficulty-aware analysis and further supports compatibility-aware pairwise image tiling.

cs.CV↗

arXiv2026

From Existing Methods to New Research Ideas: Two Perspectives on Recursive Self-Improvement

Recursive self-improvement (RSI) offers a route to AI systems that participate in their own continued development. Recent methods allow language agents to retain changes to their parameters, memory, skills, and code for subsequent execution and improvement. We examine how these methods can inform new research ideas through two complementary perspectives. First, we introduce a unified taxonomy with nine design dimensions organized into four groups: update targets, update organization, update schedule and scope, feedback and selection. This taxonomy places existing methods in a shared design space, where examining individual dimensions and their interactions can inspire new research ideas. Second, we examine how existing RSI methods draw on traditional deep learning, for example through optimization, regularization, data augmentation, continual learning, and adversarial learning. We develop further ideas by examining more specific learning mechanisms and drawing on evolutionary computation. We also review how existing benchmarks assess self-improvement under online, offline, and offline-to-online settings. Together, these perspectives help researchers develop new RSI research ideas from existing designs and learning mechanisms. RSI papers and resources are available at https://prism-shadow.github.io/awesome-rsi/.

cs.AI↗

arXiv2026

AdaSwitch: An Adaptive Switching Meta-Algorithm for Learning-Augmented Bounded-Influence Problems

We study history-dependent online problems with a possibly inaccurate prediction of the future request sequence. Motivated by several real-world applications, we introduce a \emph{bounded-influence} framework in which past decisions and requests affect the future optimal value by only a bounded amount. Within this framework, we develop AdaSwitch, a meta-algorithm that adaptively switches between suitable offline and online oracles. AdaSwitch provides explicit guarantees on expected performance that tighten as prediction error decreases or the offline optimum increases. With perfect predictions, its guarantee approaches the offline oracle's guarantee as the offline optimum grows. It also retains a worst-case guarantee close to that of the online oracle under arbitrary predictions. Applications to online lead-time quotation, $k$-server and caching, and online reusable resource allocation demonstrate the framework's applicability to both reward maximization and cost minimization.

cs.LG↗

arXiv2026

Shortcuts for Adiabatic and Variational Algorithms in Molecular Simulation

Quantum algorithms offer a promising route toward computational advantage, but current implementations remain constrained by limited coherence and gate errors, making circuit complexity and noise sensitivity critical considerations. Here, we develop a shortcuts-to-adiabaticity strategy for molecular ground-state preparation and apply it to both digitized adiabatic quantum computing and the variational quantum eigensolver. In the adiabatic framework, we construct an approximate counterdiabatic (CD) Hamiltonian using a nested-commutator expansion. For the molecular systems considered, the CD correction accelerates state preparation and improves energy convergence in the fast-evolution regime, enabling a target accuracy to be reached with a lower implemented-operator count. In the variational setting, we introduce the adiabatic gauge ansatz ($\mbox{AGA}$), obtained by promoting the Pauli-string components generated by the approximate adiabatic gauge potential (AGP) to independent variational generators. We further propose a reduced variant, $\mbox{AGAR}$, that retains only Pauli generators with weight not larger than two, substantially reducing the ansatz size while maintaining competitive accuracy over the regimes considered. Benchmarks on $\mathrm{LiH}$ and $\mathrm{BeH}_2$ show that the proposed ansätze achieve accuracies competitive with established approaches such as UCCSD. Experiments on the \texttt{IBM\_basquecountry} superconducting quantum processor, together with simulations under depolarizing noise, show that the reduced CD-inspired ansatz yields smaller energy discrepancies than UCCSD under the noise conditions considered. Overall, our results demonstrate how AGP-inspired operator structures can be used to balance accuracy and implementation cost in molecular simulations on current noise-limited quantum hardware.

quant-ph↗

arXiv2026

Speaking the Navigator's Language: Trajectory-Grounded Instruction Translation for Frozen Aerial VLN Agents

Aerial vision-and-language navigation (VLN) agents are typically trained on detail-rich, trajectory-aligned commands, whereas users issue short, intent-driven instructions; on a frozen OpenFly navigator, this \emph{instruction gap} drops success rate (SR) from $31.03\%$ to $11.33\%$. To scale translator training, we prompt a language model with human-written style examples to convert original commands into paired, intent-centered Weak commands, which yield $15.27\%$ SR. We introduce the \textbf{Trajectory-Grounded Instruction Translator (TGIT)}, a front-end that keeps the navigator frozen and translates Weak inputs into agent-executable commands by learning from its trajectory outcomes. The resulting Weak-trained translator raises Weak-input SR to $37.93\%$ and transfers zero-shot to real human instructions ($11.33\%{\rightarrow}32.51\%$); it also improves held-out OpenFly ($4.95\%{\rightarrow}20.79\%$) and yields recovery on CityNav and AirVLN.

cs.AI↗

arXiv2026

LVSPM: Long Sequence View Synthesis and Pose Estimation Model

We present LVSPM, a generalizable model that jointly estimates camera poses and synthesizes novel views from uncalibrated image collections. Trained with only RGB images and pose supervision, LVSPM avoids dense 3D ground truth and employs test-time training (TTT) layers to scale seamlessly to hundreds of input views. On RealEstate10k, Co3Dv2, and DL3DV, LVSPM surpasses VGGT in pose estimation across 16-256 views, with especially large margins at strict thresholds. For novel view synthesis under a practical protocol where more views cover larger scenes, LVSPM achieves state-of-the-art pose-free quality---surpassing even pose-dependent models in PSNR---and still maintains high quality as scene scale grows, while baselines collapse. The code is available at https://burningdust21.github.io/Projects/LVSPM .

cs.CV↗

arXiv2026

Model-agnostic distribution recovery in high dimension via Fourier measurements

We study the problem of reconstructing an unknown probability distribution to high accuracy, given noisy information about its low-degree Fourier coefficients. Prior results for this basic problem have typically imposed strong modeling assumptions on the distribution, often requiring that it is a linear combination of spatially separated point sources. In this work, we analyze a very general version of the high-dimensional recovery problem, by considering completely general distributions over the $d$-dimensional torus $[0,1)^d$; we do not assume any spatial separation between point sources, or even that the distribution is a finite linear combination of point sources. The question naturally arises: what can be said about distribution recovery in such a general setting? - As a warm-up, we first give results for reconstructing distributions with respect to the Wasserstein distance. We establish essentially matching upper and lower bounds on the cutoff frequency $T$ and the magnitude $κ$ of the noise for which accurate reconstruction is possible. Roughly speaking, our results here show that for $d$-dimensional distributions, estimates of $\approx \exp(d)$ many Fourier coefficients are necessary and sufficient for accurate reconstruction under the Wasserstein distance. - As our main result, we define a new notion of "heavy hitter" reconstruction for distributions, which essentially amounts to achieving high-accuracy reconstruction of all "sufficiently dense" regions of the distribution. We give essentially matching upper and lower bounds on the cutoff frequency $T$ and the magnitude $κ$ of the noise for which accurate reconstruction is possible under this notion. Our results show that (in sharp contrast with Wasserstein reconstruction) accurate estimates of only $\approx \exp(\sqrt{d})$ many Fourier coefficients are necessary and sufficient for heavy hitter reconstruction.

cs.DS↗

arXiv2026

On the Complexity of Mixed Equilibria in First-Price Auctions with Correlated Priors

We show that computing an approximate mixed Bayes-Nash equilibrium in a discrete first-price auction with correlated priors is PPAD-complete. The key intermediate step in our reduction is the PPAD-completeness of computing an approximate Nash equilibrium in a new normal-form game, the hypergraph discrete first-price auction, which may be of independent interest.

cs.GT↗

arXiv2026

Inertial Counterdiabatic Driving

We introduce an inertial counterdiabatic (CD) protocol that reformulates shortcut-to-adiabaticity design as a moving-frame diagonalization problem. The method employs the Magnus expansion to construct a variational propagator generated by an accessible CD ansatz. In the new rotating frame, the controls are optimized to suppress the off-diagonal Hamiltonian terms, and the CD terms are defined to counteract the corresponding inertial contribution. The propagator generated with the Magnus expansion naturally exploits additional control directions generated through its nested commutator, while the sequential design of the propagator introduces temporal nonlocality into the optimization. We benchmark the method on a two-qubit system and on the one-dimensional quantum Ising chain, where it consistently outperforms variational CD protocols based on action minimization. The largest improvements are observed near gap-closing regions, demonstrating the advantage of combining temporal and algebraic nonlocality for suppressing diabatic excitations.

quant-ph↗

arXiv2026

Tarski Fixed Points in Quasi-FPT Queries

We study the query complexity of finding a Tarski fixed point over $[n]^k$. Previous work has left a large gap between $\smash{Ω(\log^2 n)}$ and $\smash{\log^{O(k)}n}$. We show that both of the previous upper and lower bounds were far from tight: for every $k\geq 3$, \[ Ω\left((\log n)^{\frac{1}{2}\lceil \log k\rceil}\right)\le \operatorname{Tarski}(n,k)\le O\left(5^k(\log n)^{\lceil \log k\rceil}\right). \] Succinctly, up to the fixed-parameter factor of $5^k$, the complexity is settled at $(\log n)^{Θ(\log k )}$. In particular, we obtain the first super-polynomial query lower bound for this problem.

cs.CC↗

arXiv2026

SPECTRUM: Proximal Spectral Modulation for Looped Self-Distillation

A model that learns from its own outputs inherits more than their correctness: it inherits which solutions it produces. We formulate Looped Self-Distillation, a self-evolution framework for code generation in which a model repeatedly generates and learns from its own raw outputs, under a fixed information budget, without ongoing external assessment or test-based selection of the generated samples. We identify a consequential separation: correctness can improve while the breadth of correct implementations contracts. We introduce SPECTRUM, which re-estimates loss-sensitive key/value geometry from a fixed reference anchor at each round and converts it into full-rank proximal spectral modulation. All generated completions train a single student, whose subsequent inference requires no intervention. After five rounds of experiments on MBPP, SPECTRUM retains 89.9% of the initial model's 64-sample correct AST richness, compared with 66.4% for Vanilla self-distillation and 65.5% for a subspace-projection control. The advantage persists at matched correct-sample counts. Without further training or recalibration, the resulting student also achieves higher matched-correct richness than Vanilla SD on HumanEval+ and APPS Intro, demonstrating transfer of the diversity benefit. These findings establish correct-solution retention as a complementary objective of recursive self-improvement (RSI) and show that generation-time intervention can improve the solution repertoire retained by subsequent students.

cs.AI↗