Searcharxiv⌕ Search

arXiv subjects

Shuai Shao

Publications and source records attributed to Shuai Shao.

At least 19 recordsLinked to original sources
arXiv2026

A Dichotomy for Boolean Complex Holant Problems with Conjugate-Closed Signature Sets

We study Boolean Holant problems with complex-valued signature sets closed under conjugation. Such sets arise naturally in tensor-network expressions for classical strong simulation of quantum circuits. We prove a complexity dichotomy for such problems with an explicit tractability criterion. This extends the dichotomy for real-valued Holant problems, with the same four tractability conditions. Our proofs use Xia's projective binary group framework and quantum entanglement theory. The conjugate closure assumption precisely makes $k$-uniformity, directly applicable to the classification of Holant problems, by realizing reduced density matrices via Holant gadgets. We also use the classification of absolutely maximally entangled states to resolve a particular $6$-ary obstruction in our inductive proof of the \#P-hardness.

cs.CC↗
arXiv2026

Classification of Small-size Quantum Secret Sharing Schemes using Uniform States

We study the connection between quantum secret sharing (QSS) schemes and $k$-uniform states of qubits beyond the equivalence between threshold QSS schemes and AME states. Specifically, we show that $3$-uniformity is a necessary but not sufficient condition for constructing a $3$-homogeneous QSS scheme using states of qubits. To the best of our knowledge, this is the first result connecting \emph{non-threshold} QSS schemes with $k$-uniform states. As an application of our result, we classify QSS schemes for up to 7 players and provide explicit characterizations of their existence. Our results offer new insights into the role of $k$-uniform states in the design of QSS schemes (not necessarily threshold) and provide a foundation for future classifications of QSS schemes with more complex structures.

quant-ph↗
arXiv2026

ATBench: A Diverse and Realistic Agent Trajectory Benchmark for Safety Evaluation and Diagnosis

Evaluating the safety of LLM-based agents is increasingly important because risks in realistic deployments often emerge over multi-step interactions rather than isolated prompts or final responses. Existing trajectory-level benchmarks remain limited by insufficient interaction diversity, coarse observability of safety failures, and weak long-horizon realism. We introduce ATBench, a trajectory-level benchmark for structured, diverse, and realistic evaluation of agent safety. ATBench organizes agentic risk along three dimensions: risk source, failure mode, and real-world harm. Based on this taxonomy, we construct trajectories with heterogeneous tool pools and a long-context delayed-trigger protocol that captures realistic risk emergence across multiple stages. The benchmark contains 1,000 trajectories (503 safe and 497 unsafe), averaging 9.01 turns and 3.95k tokens, with 1,954 invoked tools drawn from pools spanning 2,084 available tools. Data quality is supported by rule-based and LLM-based filtering plus full human audit. Experiments on frontier LLMs, open-source models, and specialized guard systems show that ATBench is challenging even for strong evaluators, while enabling taxonomy-stratified analysis, cross-benchmark comparison, and diagnosis of long-horizon failure patterns.

cs.AI↗
arXiv2026

OmniCamera: A Unified Framework for Multi-task Video Generation with Arbitrary Camera Control

Video fundamentally intertwines two crucial axes: the dynamic content of a scene and the camera motion through which it is observed. However, existing generation models often entangle these factors, limiting independent control. In this work, we introduce OmniCamera, a unified framework designed to explicitly disentangle and command these two dimensions. This compositional approach enables flexible video generation by allowing arbitrary pairings of camera and content conditions, unlocking unprecedented creative control. To overcome the fundamental challenges of modality conflict and data scarcity inherent in such a system, we present two key innovations. First, we construct OmniCAM, a novel hybrid dataset combining curated real-world videos with synthetic data that provides diverse paired examples for robust multi-task learning. Second, we propose a Dual-level Curriculum Co-Training strategy that mitigates modality interference and synergistically learns from diverse data sources. This strategy operates on two levels: first, it progressively introduces control modalities by difficulties (condition-level), and second, trains for precise control on synthetic data before adapting to real data for photorealism (data-level). As a result, OmniCamera achieves state-of-the-art performance, enabling flexible control for complex camera movements while maintaining superior visual quality.

cs.CV↗
arXiv2026

SkillGate: Training In-Policy Skill Selection in Long-Horizon Agents

Agent frameworks increasingly package procedural knowledge as skills: instruction files an agent reads on demand, while public libraries now hold thousands of them. Which skill to read has thus become a decision the policy itself makes in the middle of an episode, yet no existing signal trains it. We show that the default remedy, outcome-rewarded RL over the candidate slate, cannot teach it, for a structural reason we identify and name selector credit starvation: under a broadcast, sequence-level advantage, the few tokens that name the chosen skill carry a vanishing share of the loss, and the credit they inherit is increasingly wrong-signed as trajectories lengthen. A correct choice is punished whenever the execution after it fails, even though the choice itself is among the most valuable decisions in the trajectory. Auditing a completed run's own training artifacts confirms all three properties, each worsening monotonically with horizon. SkillGate removes the failure by construction: it partitions the token support into two disjoint credit channels, outcome credit reaching only execution tokens, and a separate action-local advantage reaching exactly the skill-naming tokens, positive only when a trajectory's single read is the correct one. On five agentic benchmarks under a 16-candidate slate, SkillGate lifts a 9B policy from 40.8% to 53.2% trial success, well ahead of the identical budget spent on outcome reward alone, while cutting exposure to misleading candidates by two thirds and reading fewer skills.

cs.AI↗
arXiv2026

Rethinking Generalization in Reasoning SFT: A Conditional Analysis on Optimization, Data, and Model Capability

A prevailing narrative in LLM post-training holds that supervised finetuning (SFT) memorizes while reinforcement learning (RL) generalizes. We revisit this claim for reasoning SFT with long chain-of-thought (CoT) supervision and find that cross-domain generalization is not absent but conditional, jointly shaped by optimization dynamics, training data, and base-model capability. Some reported failures are under-optimization artifacts: cross-domain performance first degrades before recovering and improving with extended training (a dip-and-recovery pattern), so shorttraining checkpoints can underestimate generalization. Data quality and structure both matter: low-quality solutions broadly hurt generalization,while verified long-CoT traces yield consistent cross-domain gains. Model capability is essential: stronger models internalize transferable procedural patterns (e.g., backtracking) even from a toy arithmetic game, while weaker ones imitate surface verbosity. This generalization is asymmetric, however: reasoning improves while safety degrades, reframing the question from whether reasoning SFT generalizes to under what conditions and at what cost.

cs.AI↗
arXiv2026

MedClaw: Heuristic Agent Harness for Long-Horizon Surgical Video Reasoning

Understanding tens-of-minutes surgical videos requires long-horizon temporal reasoning, answering what happens before, after, or across stages of a procedure by grounding the question in visual evidence spread across time. Existing approaches handle this poorly: a one-shot vision-language model (VLM) compresses the whole procedure to fit its context window and loses the detail a "before" or "after" question depends on, while video agents that train the model where to look are data-hungry and transfer poorly to out-of-domain surgery. We build an agent harness that separates reasoning from perception and improves by evolving context rather than optimizing weights. A text-only orchestrator plans which evidence to gather and issues an auditable sequence of tool calls, while frozen vision-language sub-agents execute each call over the pixels, viewing, cropping, inspecting frames, and retrieving external knowledge. We further propose a gradient-free, reward-gated Heuristic Skill Distillation loop that mines the agent's own low-scoring traces and keeps a candidate skill only when it raises a validation reward, yielding reusable retrieval skills, notably directed re-look. Growing an external skill library rather than tuning weights, the loop adapts from only about 100 labeled examples, far fewer than supervised or reinforcement fine-tuning requires. To evaluate this agent, we introduce MedClawBench, a de-leaked, doctor-grounded benchmark of 1,123 questions over self-built long neurosurgery recordings and a held-out public lecture-video test split. Across both datasets and all four evaluation dimensions, our agent consistently outperforms one-shot VLMs and general video-agent frameworks, with the largest gains on the long, out-of-domain neurosurgery videos. Project page: https://fyycs.github.io/medclaw/.

cs.CV↗
arXiv2026

HyperFL: Query-Adaptive Representation Learning for Software Fault Localization

Software fault localization identifies the code locations responsible for reported issues and is a fundamental step toward automated debugging and program repair. Recent retrieval-based approaches formulate fault localization as a dense retrieval task by learning a shared embedding space between issue reports and source code. However, these methods encode all issue reports using a fixed query representation, despite the substantial diversity of real-world issue reports in length, structure, and debugging information. To address this limitation, we propose HyperFL, a query-adaptive representation learning framework for software fault localization. HyperFL employs a lightweight hypernetwork to generate query-specific LoRA parameters for the query encoder, enabling dynamic query adaptation while keeping the code encoder fixed and reusable. Experiments on a real-world issue localization benchmark demonstrate that HyperFL consistently improves retrieval performance across multiple embedding backbones, achieving up to 13.3% relative improvement in function-level MRR@10 and 16.7% relative improvement in Hit@1 over the state-of-the-art method SweRank. Further analysis shows that HyperFL learns distinct adaptation patterns for different issue characteristics, highlighting the effectiveness of query-adaptive representations for software issue localization.

cs.SE↗
arXiv2026

ConFL: Explainable Concurrent Fault Localization via Hierarchy-Guided LLM Reasoning

Localizing concurrent bugs from bug reports alone is challenging due to incomplete information, misleading program-entity mentions, and complex cross-thread interactions, causing existing LLM-based approaches to suffer from unstable reasoning and limited explainability. We propose ConFL, an explainable concurrent fault localization framework that augments LLM reasoning with structured concurrency knowledge. ConFL constructs a Concurrent Knowledge Base (CKB) from source code and performs LLM-guided hierarchical retrieval to progressively narrow the search space from components to interaction-level concurrency contexts. An interaction-level DSL explicitly encodes cross-thread interactions over shared resources, enabling focused reasoning without traversing deep call chains. Experiments on real-world concurrent bugs from eight large-scale Java projects show that ConFL significantly outperforms state-of-the-art IR-based and LLM-based baselines, achieving an MRR of 0.503 and a MAP of 0.486, while remaining robust to noisy bug reports, unseen bugs, and different LLM backbones.

cs.SE↗
arXiv2026

Harness-R1: Learning to Edit Executable Runtime Harnesses from Agent Failure Trajectories

Agents built around large language models continually accumulate interaction trajectories during deployment, yet their behavior typically remains fixed. Beyond updating model weights, these trajectories can improve the agent harness that constructs context, mediates tools, validates actions, and recovers execution. We introduce Harness-R1, the first method, to our knowledge, that makes failure-conditioned, lifecycle-wide editing of an existing executable runtime a learned capability. It post-trains a dedicated harness engineer with online reinforcement learning so that its edits are optimized for the realized task success they produce, rather than proposed by a fixed editor. A separate 9B engineer converts batches of target-agent failures into validated executable patches; fresh same-batch reruns of the frozen target provide outcome rewards, so training updates only the engineer. Cold-start supervised fine-tuning initializes this editing policy, which is then trained online with group-relative policy optimization. Across WebShop, ALFWorld, and DBBench, Harness-R1 raises vanilla Qwen3.5-9B success from 44.3% to 53.6% (+9.3 percentage points). After direct target-agent fine-tuning, a target-specific engineer raises the average further from 59.2% to 64.2% (+5.0 points); because these gains hold both before and after fine-tuning the target, Harness-R1 points toward co-evolving the harness engineer and the target agent.

cs.AI↗
arXiv2026

An LP Algorithm for Counting Eulerian Orientations Through the Lens of Quasi-polymorphism

The weighted Eulerian orientation counting problem ($\#\mathrm{EO}$) plays a key role in the complexity classification program for Holant problems. A recent result established an $\mathrm{FP}^{\mathrm{NP}}$ versus $\#\mathrm{P}$-hard dichotomy for $\#\mathrm{EO}$ problems. The tractable side of this dichotomy can be characterized by functions admitting quasi-polymorphisms of the ternary XOR operation, leaving open whether these cases on the $\mathrm{FP}^{\mathrm{NP}}$ side are in fact in FP. In this paper, we settle this question by giving a polynomial-time algorithm for all cases on the $\mathrm{FP}^{\mathrm{NP}}$ side. Consequently, we obtain a complete FP versus $\#\mathrm{P}$ dichotomy for counting weighted Eulerian orientations, and further for complex-valued Holant problems with an odd-arity signature. Our algorithm is based on a linear programming relaxation, but we use it in a nonstandard way. Instead of proving that the relaxation is integral and solving the problem directly from an optimal LP solution, we use the relaxation as a structural tool to lift the quasi-polymorphism condition to an ordinary polymorphism condition. This reveals an affine local structure of the constraint functions, which leads to tractability.

cs.CC↗
arXiv2026

VAD: Attributing Visual Evidence for Target Reconstruction in Multimodal On-Policy Distillation

Multimodal on-policy distillation (OPD) transfers fine-grained visual knowledge by supervising student-generated trajectories with a privileged-view teacher. Yet its next-token corrections are source-mixed, combining visual signals with linguistic priors and teacher-specific effects. The key challenge is to estimate which corrections are supported by visual evidence, not merely where or how strongly to distill. We introduce Visual Attribution Distillation (VAD), a counterfactual target-reconstruction algorithm that estimates the visually attributable part of a teacher correction. At each student-generated prefix, VAD evaluates the same fixed teacher with the relevant evidence present and removed. The corresponding change in centered log-probabilities defines ut, a signed proxy for the visual evidence direction that estimates how revealing the evidence supports or refutes candidate tokens. VAD projects the original correction onto this proxy to obtain an intervention-aligned component and a proxy-unexplained residual, then reconstructs a student-anchored target from the former. During training, this reconstructed target supplies the primary supervision signal, while the privileged teacher contributes a weak regularizer. Across six fine-grained visual benchmarks at 4B and 9B scales, VAD outperforms direct privileged-view distillation and visual-advantage weighting. Token- level and controlled-target analyses show that the proxy-aligned component is enriched in task-relevant visual corrections and yields stronger target shifts, especially when evidence refutes a mistaken answer. These results support counterfactual target reconstruction as an effective alternative to source-mixed supervision.

cs.CV↗
arXiv2026

From Single- to Cross-Document: Benchmarking Multi-Granularity Event Analysis of Large Language Models

Event analysis is an essential and fundamental direction of information extraction, involving various event-centric tasks at different granularity of documents. While large language models (LLMs) have preliminarily achieved promising performance in part of these tasks individually, their capability in event analysis still lacks comprehensive understanding due to restricted document granularity, task designs, and data source of existing benchmarks. To address these limitations, we introduce MiGUE-Bench, a systematic benchmark for assessing the performance of LLMs in multi-granularity event analysis. To support large-scale evaluation, we first develop an LLM-driven self-correcting annotation framework called MiGUE-Pipeline, enabling scalable acquisition of high-quality source data of events with automatic labels. Then, we design four core tasks in our benchmark, i.e., event detection, relation reasoning, structure induction, and future prediction, to probe model competence at different levels, from atomic event details to complex cross-document narratives. Extensive experiments on state-of-the-art LLMs and retrieval-augmented generation (RAG) methods delineate the current capability boundary and identify critical deficiencies, providing insights into the future improvement of LLMs in challenging event analysis tasks.

cs.CL↗
arXiv2026

ACM: Agentic Context Management for Long Horizon Tasks

Agentic tasks are inherently long-horizon and multi-turn, constantly accumulating context through interactions with the environment. Existing context compression methods inevitably incur information loss and are triggered by rigid heuristic rules, leaving them misaligned with the agent's evolving reasoning focus. We propose Agentic Context Management (ACM), a framework that equips agents with purpose-built context editing tools for lossless context management. Inspired by the interaction between short-term and long-term human memory, the agent autonomously decides when to compress its context, offloads discarded content to an external memory system, and queries it on demand for later retrieval. Building on this framework, we further develop a post-training pipeline that constructs high-quality demonstrations of context management and improves model performance on both agentic search and coding tasks. Further analysis reveals that effective context management reduces peak token pressure, enables extended explorations, and yields more consistent solutions across independent trials. Code, data, and model checkpoints are available at https://github.com/lixiaochuan2020/agentic-context-management.

cs.AI↗
arXiv2026

Uniform-in-time rational approximation of the matrix exponential with real poles

We propose two new approaches for constructing families of rational functions with shared real poles that nearly uniformly approximate the functions $\exp(-tz)$ for $z\geq 0$ and $t$ in a positive time interval. The first result concerns the case where all real poles coalesce into a single point. With an appropriate choice of a weight function we are able to derive a closed formula for the asymptotically optimal location of such a pole. We then discuss the more general case where all real poles are distinct. Using Zolotarev's construction of certain optimal rational functions, we present a simple algorithm to derive nearly optimal poles efficiently. We analyze the stability of the numerical evaluation of the resulting rational matrix functions in floating-point arithmetic. By controlling the growth of potential ill-conditioning arising from partial fractions, reliable and highly parallelizable exponential propagators are obtained.

math.NA↗
arXiv2026

Strong Spatial Mixing for General 2-Spin Systems: A Unified Approach from Zero-Freeness

We study the algorithmic implications of zero-free regions for the partition functions of 2-spin systems. While Barvinok's algorithm yields FPTASes in such regions, the applicability of Weitz's algorithm is limited to parameter regimes where strong spatial mixing (SSM) can be established. It remains open whether Weitz's algorithm can be applied to general zero-free regions, particularly in settings where no standard tree-recurrence-based proof of SSM is known. We establish new SSM results and thereby extend the applicability of Weitz's FPTAS to all currently known zero-free regions of 2-spin systems with pinned vertices. We achieve this through a unified approach to deriving SSM from zero-freeness in the most general settings of 2-spin systems. Our work features two key innovations. 1.Our SSM results cover parts of the celebrated Lee-Yang zero-free region for the ferromagnetic Ising model, where no tree-recurrence-based proof of SSM is currently known or considered feasible. The tree recurrence method typically relies on carefully designed potential functions, the construction and analysis of which can be highly challenging. For ferromagnetic 2-spin systems, it remains an open challenge whether such potential functions can be constructed. We circumvent this difficulty by deriving SSM directly from zero-freeness. 2.The prior approach to deriving SSM from zero-freeness relies on cluster expansions, which are model-specific and known only for a few restricted parameter settings such as the hard-core model near vertex activity $λ=1$. We overcome this obstacle by introducing a purely combinatorial approach based on a novel Christoffel-Darboux-type identity that holds universally for 2-spin systems. This provides a broadly applicable framework for handling general 2-spin systems with arbitrary multivariate parameters and zero-free regions of arbitrary shape in a unified manner.

math-ph↗
arXiv2026

Eulerian orientations and Hadamard codes: A novel connection via counting

We discover a novel connection between two classical mathematical notions, Eulerian orientations and Hadamard codes by studying the counting problem of Eulerian orientations (\#EO) with local constraint functions imposed on vertices. We present two special classes of constraint functions and a chain reaction algorithm, and show that the \#EO problem defined by each class alone is polynomial-time solvable by the algorithm. These tractable classes of functions are defined inductively, and quite remarkably the base level of these classes is characterized perfectly by the well-known Hadamard code. Thus, we establish a novel connection between counting Eulerian orientations and coding theory. We also prove a \#P-hardness result for the \#EO problem when constraint functions from the two tractable classes appear together.

cs.CC↗
arXiv2026

Clarus: Coordinating Autonomous Research Agents toward Web-Scale Scientific Collaboration

Existing autonomous research agents can support parts of the research process, but most systems still treat research as either an isolated assistant task or a closed workflow. Therefore, autonomous science needs a collaboration infrastructure that coordinates projects, agents, and digital and physical resources. We identify this as a shift from code-centered execution loops to research-oriented collaboration processes, where questions, evidence, participants, and resources must be coordinated under uncertainty. In this framing, an agent may be an AI system, a human researcher, a team, a laboratory, or an organization-backed participant. To this end, we present Clarus, a collaboration infrastructure for coordinating autonomous research agents toward web-scale scientific collaboration. Clarus reformulates research as an open, auditable, attributable, and resource-aware multi-phase collaboration process. It defines a minimal project-agent-resource object model and organizes scientific collaboration through four layers including Research Application, Digital Collaboration, Physical Substrate, and Physical World. Core modules are implemented as pluggable mechanisms, allowing Clarus to adapt to task risk, collaboration structure, and resource constraints. Through a controlled paper-generation case study, we show that Clarus can organize a research goal into a traceable, reviewable, attributable, and accumulative collaboration network across phases, tasks, and participants. Together, the object model, collaboration protocol, trust mechanisms, and prototype validation provide an initial foundation for open research networks. Clarus is now available at clarus.holosai.io.

cs.AI↗