Search arXiv⌕ Search

arXiv subjects

Chen Zhang

Publications and source records attributed to Chen Zhang.

At least 163 records · Page 9Linked to original sources
arXiv2026

NPG-Muse: Scaling Long Chain-of-Thought Reasoning with NP-Hard Graph Problems

Reasoning Large Language Models (RLLMs) have recently achieved remarkable progress on complex reasoning tasks, largely enabled by their long chain-of-thought (Long CoT) capabilities. However, developing these Long CoT behaviors relies heavily on post-training with high-quality datasets, which are typically costly and human-curated (e.g., mathematics and code), leaving scalable alternatives unexplored. In this work, we introduce NP-hard (NPH) graph problems as a novel synthetic training corpus, as they inherently require deep reasoning, extensive exploration, and reflective strategies, which are the core characteristics of Long CoT reasoning. Building on this insight, we develop a two-stage post-training framework: (i) Long-CoT Supervised Fine-Tuning (SFT) on rejection-sampled NPH graph instances, which substantially enhances reasoning depth, and (ii) Reinforcement Learning (RL) with a fine-grained reward design, which sharpens reasoning efficiency. The resulting NPG-Muse-series models exhibit substantially enhanced Long CoT reasoning capabilities, achieving consistent gains across mathematics, coding, logical, and graph reasoning benchmarks. NPG-Muse-7B even surpasses QwQ-32B on NPH graph problems in both accuracy and reasoning efficiency. These results position NPH graph problems as an effective and scalable resource for advancing Long CoT reasoning in LLM post-training. Our implementation is available at https://github.com/littlewyy/NPG-Muse.

cs.CL↗
arXiv2026

AudioRAG: A Challenging Benchmark for Audio Reasoning and Information Retrieval

Due to recent advancements in Large Audio-Language Models (LALMs) that demonstrate remarkable performance across a range of sound-, speech- and music-related tasks, there is a growing interest in proposing benchmarks to assess these models. Existing benchmarks generally focus only on reasoning with internal knowledge, neglecting real-world scenarios that require external information grounding. To bridge this gap, we introduce AudioRAG, a novel benchmark designed to evaluate audio-based reasoning augmented by information retrieval in realistic web environments. This benchmark comprises both LLM-generated and manually curated question-answer pairs. Our evaluations reveal that even the state-of-the-art LALMs struggle to answer these questions. We therefore propose an agentic pipeline that integrates audio reasoning with retrieval-augmented generation, providing a stronger baseline for future research.

eess.AS↗
arXiv2026

CoMAS: Co-Evolving Multi-Agent Systems via Interaction Rewards

Self-evolution is a central research topic in enabling large language model (LLM)-based agents to continually improve their capabilities after pretraining. Recent research has witnessed a transition from reinforcement learning (RL)-free to RL-based methods. Current RL-based methods either rely on dense external reward signals or extract intrinsic reward signals from LLMs themselves. However, these approaches diverge from the self-evolution mechanisms observed in human intelligence, where individuals learn and improve through mutual discussion and collaboration. In this work, we introduce Co-Evolving Multi-Agent Systems (CoMAS), a novel framework that enables agents to improve autonomously by learning from inter-agent interactions without external supervision. CoMAS generates intrinsic rewards from rich discussion dynamics, employs an LLM-as-a-judge mechanism to formulate these rewards, and optimizes each agent's policy through RL, thereby enabling decentralized and scalable co-evolution. Experimental results demonstrate that CoMAS consistently outperforms untrained agents and achieves state-of-the-art performance across most evaluation settings. Ablation studies confirm the necessity of interaction-based reward signals and reveal promising scalability as the number and diversity of agents increase. These findings establish CoMAS as a novel and effective paradigm for self-evolution in LLM-based agents.

cs.CL↗
arXiv2026

DA-RAG: Dynamic Attributed Community Search for Retrieval-Augmented Generation

Owing to their unprecedented comprehension capabilities, large language models (LLMs) have become indispensable components of modern web search engines. From a technical perspective, this integration represents retrieval-augmented generation (RAG), which enhances LLMs by grounding them in external knowledge bases. A prevalent technical approach in this context is graph-based RAG (G-RAG). However, current G-RAG methodologies frequently underutilize graph topology, predominantly focusing on low-order structures or pre-computed static communities. This limitation affects their effectiveness in addressing dynamic and complex queries. Thus, we propose DA-RAG, which leverages attributed community search (ACS) to extract relevant subgraphs based on the queried question dynamically. DA-RAG captures high-order graph structures, allowing for the retrieval of self-complementary knowledge. Furthermore, DA-RAG is equipped with a chunk-layer oriented graph index, which facilitates efficient multi-granularity retrieval while significantly reducing both computational and economic costs. We evaluate DA-RAG on multiple datasets, demonstrating that it outperforms existing RAG methods by up to 40% in head-to-head comparisons across four metrics while reducing index construction time and token overhead by up to 37% and 41%, respectively.

cs.IR↗
arXiv2026

SRR-Judge: Step-Level Rating and Refinement for Enhancing Search-Integrated Reasoning in Search Agents

Recent deep search agents built on large reasoning models (LRMs) excel at complex question answering by iteratively planning, acting, and gathering evidence, a capability known as search-integrated reasoning. However, mainstream approaches often train this ability using only outcome-based supervision, neglecting the quality of intermediate thoughts and actions. We introduce SRR-Judge, a framework for reliable step-level assessment of reasoning and search actions. Integrated into a modified ReAct-style rate-and-refine workflow, SRR-Judge provides fine-grained guidance for search-integrated reasoning and enables efficient post-training annotation. Using SRR-annotated data, we apply an iterative rejection sampling fine-tuning procedure to enhance the deep search capability of the base agent. Empirically, SRR-Judge delivers more reliable step-level evaluations than much larger models such as DeepSeek-V3.1, with its ratings showing strong correlation with final answer correctness. Moreover, aligning the policy with SRR-Judge annotated trajectories leads to substantial performance gains, yielding over a 10 percent average absolute pass@1 improvement across challenging deep search benchmarks.

cs.CL↗
arXiv2026

An Archival Optical Counterpart Search for Extragalactic Fast X-Ray Transients Discovered by Einstein Probe

Extragalactic fast X-ray transients (eFXTs) represent a rapidly growing class of high-energy phenomena, whose physical origins remain poorly understood. With its wide-field, sensitive all-sky monitoring, the Einstein Probe (EP) has greatly increased the discovery rate of eFXTs. The search and identification of the optical counterparts of eFXT are vital for understanding their classification and constraining their physical origin. Yet, a considerable fraction of eFXTs still lack secure classifications due to the absence of timely follow-up observations. We carry out a systematic search of publicly available optical survey data and transient databases (including the Zwicky Transient Facility, ZTF, and the Transient Name Server, TNS) for optical counterparts to eFXT candidates detected by EP. In this paper, we describe our ongoing program and report the first results. Specifically, we identified the eFXT EP240506a to be associated with a UV/optical counterpart, AT 2024ofs. Spectroscopy of its host galaxy with VLT yields a redshift of $z = 0.120 \pm 0.002$. By combining archival survey data with early-time multiwavelength observations, we find that the luminosity and light-curve evolution of AT~2024ofs are consistent with a core-collapse supernova origin. From detectability simulations, we estimate a local event rate density $ρ_{0}=8.8^{+21.2}_{-3.9}\ \mathrm{yr^{-1}\, Gpc^{-3}}$ for EP240506a-like events, and completeness-corrected rate of about $36$--$78\ \mathrm{yr^{-1}\ Gpc^{-3}}$ for EP-detected X-ray transients associated with supernovae. Our results demonstrate the potential of EP to uncover prompt high-energy emission from core-collapse supernovae and underscore the critical importance of timely follow-up of future eFXT events.

astro-ph.HE↗
arXiv2026

VGGT-Motion: Motion-Aware Calibration-Free Monocular SLAM for Long-Range Consistency

Despite recent progress in calibration-free monocular SLAM via 3D vision foundation models, scale drift remains severe on long sequences. Motion-agnostic partitioning breaks contextual coherence and causes zero-motion drift, while conventional geometric alignment is computationally expensive. To address these issues, we propose VGGT-Motion, a calibration-free SLAM system for efficient and robust global consistency over kilometer-scale trajectories. Specifically, we first propose a motion-aware submap construction mechanism that uses optical flow to guide adaptive partitioning, prune static redundancy, and encapsulate turns for stable local geometry. We then design an anchor-driven direct Sim(3) registration strategy. By exploiting context-balanced anchors, it achieves search-free, pixel-wise dense alignment and efficient loop closure without costly feature matching. Finally, a lightweight submap-level pose graph optimization enforces global consistency with linear complexity, enabling scalable long-range operation. Experiments show that VGGT-Motion markedly improves trajectory accuracy and efficiency, achieving state-of-the-art performance in zero-shot, long-range calibration-free monocular SLAM.

cs.CV↗
arXiv2026

Millimeter-scale rigid diamond probe for high sensitivity endoscopic-magnetometry applications

Magnetometry based on diamond nitrogen-vacancy (NV) centers has been extensively studied for applications requiring diverse capabilities, spanning from nanometer spatial resolution to subpicotesla sensitivity. Among various applications, diamond magnetometers can demonstrate high sensitivity magnetic sensing within millimeter-scale size for endoscopic applications. However, the trade-off between sensitivity and spatial resolution of diamond magnetometry makes it difficult to achieve such a probe. In this study, we present a millimeter-scale rigid diamond magnetometer probe with enhanced sensitivity via optimizing the optical design. By coupling the frustum diamond with the miniaturized compound parabolic concentrator (CPC) lens, we enhance the fluorescence collection efficiency by 37% within 4 mm diameter, and the achieved sensitivity is 200 pT/Hz1/2 based on the sample with the resonance linewidth of ~8 MHz. With this verified structure, endoscopes with mm-size probe and picotesla sensitivity can be projected for surgical and industrial applications in the future.

physics.ins-det↗
arXiv2026

Understanding and Detecting Flaky Builds in GitHub Actions

Continuous Integration (CI) is widely used to provide rapid feedback on code changes; however, CI build outcomes are not always reliable. Builds may fail intermittently due to non-deterministic factors, leading to flaky builds that undermine developers' trust in CI, waste computational resources, and threaten the validity of CI-related empirical studies. In this paper, we present a large-scale empirical study of flaky builds in GitHub Actions based on rerun data from 1,960 open-source Java projects. Our results show that 3.2% of builds are rerun, and 67.73% of these rerun builds exhibit flaky behavior, affecting 1,055 (51.28%) of the projects. Through an in-depth failure analysis, we identify 15 distinct categories of flaky failures, among which flaky tests, network issues, and dependency resolution issues are the most prevalent. Building on these findings, we propose a machine learning-based approach for detecting flaky failures at the job level. Compared with a state-of-the-art baseline, our approach improves the F1-score by up to 20.3%.

cs.SE↗
arXiv2026

Planning-Augmented Sampling with Early Guidance for High-Reward Discovery

Generative Flow Networks (GFlowNets) enable structured generation with inherent diversity, but existing sampling strategies often rely on weak guided exploration, slowing early discovery of high-reward candidates. In tasks such as molecular design, rapid and consistent generation of high-reward solutions can outweigh faithful distribution matching. We propose a planning-augmented framework in which Monte Carlo Tree Search using polynomial upper confidence bounds provides online value estimates, and a controllable soft-greedy mechanism integrates these planning signals into the GFlowNets forward policy. This design fosters early exploration of high-reward trajectories and gradually shifts to policy-driven exploitation as experience accumulates. Empirical results show that our method accelerates early high-reward discovery, sustains top-quality sample generation, and preserves diversity across representative tasks. All implementations are available at https://github.com/ZRNB/PLUS.

cs.LG↗
arXiv2026

LongCat-Flash-Thinking-2601 Technical Report

We introduce LongCat-Flash-Thinking-2601, a 560-billion-parameter open-source Mixture-of-Experts (MoE) reasoning model with superior agentic reasoning capability. LongCat-Flash-Thinking-2601 achieves state-of-the-art performance among open-source models on a wide range of agentic benchmarks, including agentic search, agentic tool use, and tool-integrated reasoning. Beyond benchmark performance, the model demonstrates strong generalization to complex tool interactions and robust behavior under noisy real-world environments. Its advanced capability stems from a unified training framework that combines domain-parallel expert training with subsequent fusion, together with an end-to-end co-design of data construction, environments, algorithms, and infrastructure spanning from pre-training to post-training. In particular, the model's strong generalization capability in complex tool-use are driven by our in-depth exploration of environment scaling and principled task construction. To optimize long-tailed, skewed generation and multi-turn agentic interactions, and to enable stable training across over 10,000 environments spanning more than 20 domains, we systematically extend our asynchronous reinforcement learning framework, DORA, for stable and efficient large-scale multi-environment training. Furthermore, recognizing that real-world tasks are inherently noisy, we conduct a systematic analysis and decomposition of real-world noise patterns, and design targeted training procedures to explicitly incorporate such imperfections into the training process, resulting in improved robustness for real-world applications. To further enhance performance on complex reasoning tasks, we introduce a Heavy Thinking mode that enables effective test-time scaling by jointly expanding reasoning depth and width through intensive parallel thinking.

cs.AI↗
arXiv2026

Optimal Engagement of Residential Battery Storage to Alleviate Grid Upgrades Caused by EVs and Solar Systems

The integration of distributed energy resources has ushered in a host of complex challenges, significantly impacting power quality in distribution networks. This work studies these challenges, exploring issues such as voltage fluctuations and escalating power losses caused by the integration of solar systems and electric vehicle (EV) chargers. We present a robust methodology focused on mitigating voltage deviations and power losses, emphasizing the allocation of a Permitted Percentage (PP) of battery-based solar systems within residential areas endowed with storage capabilities. A key facet of this research lies in its adaptability to the changing landscape of electric transportation. With the rapid increase of electric trucks on the horizon, our proposed model gains relevance. By tactically deploying PP to oversee the charging and discharging of batteries within residential solar systems, utilities are poised not only to assist with grid resilience but also to cater to the upcoming demands spurred by the advent of new EVs, notably trucks. To validate the efficacy of our proposed model, rigorous simulations were conducted using the IEEE 33-bus distribution network as a designed testbed. Leveraging advanced Particle Swarm Optimization techniques, we have deciphered the optimal charging and discharging commands issued by utilities to energy storage systems. The outcomes of these simulations help us understand the transformative potential of various PP allocations, shedding light on the balance between non-battery-based and battery-based solar residences. This research underscores the need for carefully crafted approaches in navigating the complexities of modern grid dynamics amid the anticipated increase in electric vehicles.

cs.AR↗
arXiv2026

M2XFP: A Metadata-Augmented Microscaling Data Format for Efficient Low-bit Quantization

Existing low-bit Microscaling (MX) formats, such as MXFP4, often suffer from substantial accuracy degradation due to the use of a shared scaling factor with the Power-of-Two format. In this work, we explore strategies that introduce minimal metadata to recover accuracy lost during quantization while maintaining high bit efficiency across a wide range of large language models. We propose a complete algorithm-hardware co-design based on flexible metadata, featuring an online quantization with simple encoding. To support the proposed method efficiently, we implement a lightweight hardware unit and integrate it into the accelerator. Evaluation results demonstrate that our method substantially narrows the accuracy gap, achieving on average a 70.63% reduction in accuracy loss compared to MXFP4 and a 37.30% reduction relative to the latest NVFP4 on LLM benchmarks. Furthermore, our design delivers up to 1.91$\times$ speedup and 1.75$\times$ energy savings over state-of-the-art accelerators. Our code is available at https://github.com/SJTU-ReArch-Group/M2XFP_ASPLOS26.

cs.AR↗
arXiv2026

MultiVis-Agent: A Multi-Agent Framework with Logic Rules for Reliable and Comprehensive Cross-Modal Data Visualization

Real-world visualization tasks involve complex, multi-modal requirements that extend beyond simple text-to-chart generation, requiring reference images, code examples, and iterative refinement. Current systems exhibit fundamental limitations: single-modality input, one-shot generation, and rigid workflows. While LLM-based approaches show potential for these complex requirements, they introduce reliability challenges including catastrophic failures and infinite loop susceptibility. To address this gap, we propose MultiVis-Agent, a logic rule-enhanced multi-agent framework for reliable multi-modal and multi-scenario visualization generation. Our approach introduces a four-layer logic rule framework that provides mathematical guarantees for system reliability while maintaining flexibility. Unlike traditional rule-based systems, our logic rules are mathematical constraints that guide LLM reasoning rather than replacing it. We formalize the MultiVis task spanning four scenarios from basic generation to iterative refinement, and develop MultiVis-Bench, a benchmark with over 1,000 cases for multi-modal visualization evaluation. Extensive experiments demonstrate that our approach achieves 75.63% visualization score on challenging tasks, significantly outperforming baselines (57.54-62.79%), with task completion rates of 99.58% and code execution success rates of 94.56% (vs. 74.48% and 65.10% without logic rules), successfully addressing both complexity and reliability challenges in automated visualization generation.

cs.CL↗
arXiv2026

Self-bound hybrid stars with strong phase transitions can relieve major compact star observation tensions

Some recent pulsar observations cannot naturally fit into the conventional picture of neutron stars: the compact objects associated with HESS J1731-347 and XTE J1814-338 have too small radii in the low-mass regime, while the secondary component of GW190814 is too massive for neutron stars to be compatible with constraints from the GW170817 event. In this study, we demonstrate that all these anomalous observations and tensions, together with other conventional ones such as recent NICER observations of PSR J0740+6620, J0030+0451, and PSR J0437-4715, can be naturally explained simultaneously by a new general type of self-bound hybrid stars with large density discontinuities, and thus are radially stable in either the slow or rapid phase transition context. As a proof of concept, we use hybrid quark stars, inverted hybrid stars, and hybrid strangeon stars as benchmark examples to explicitly demonstrate the advantage and feasibility of self-bound hybrid stars with strong phase transitions in relieving all tensions related to compact stars' masses, radii, and tidal deformabilities.

astro-ph.HE↗
arXiv2026

AgentAsk: Multi-Agent Systems Need to Ask

Multi-agent systems (MAS) built on large language models promise improved problem-solving through collaboration, yet they often fail to consistently outperform strong single-agent baselines due to error propagation at inter-agent message handoffs.In this work, we conduct a systematic empirical analysis of such failures and introduce an edge-level error taxonomy that identifies four dominant error types: Data Gap, Signal Corruption, Referential Drift, and Capability Gap, as primary sources of failure in multi-agent interactions. Building on this taxonomy, we propose AgentAsk, a lightweight clarification module designed to intervene at the edge level in MAS to prevent cascading errors. The module operates by strategically applying minimal clarifications at critical points within the system, improving the accuracy and efficiency of the overall task. AgentAsk is trained to balance the trade-offs between clarification cost, latency, and accuracy, while it is also architecture-agnostic and can be easily integrated into existing systems. Evaluated across five benchmarks, AgentAsk consistently improves accuracy by up to 4.69%, while keeping latency and extra costs below 10% compared to baseline MAS, showcasing its high efficiency and minimal overhead.

cs.AI↗
arXiv2026

Bayesian constraints on quark stars from multi-messenger observations

We perform a systematic Bayesian analysis of quark star equations of state under current multimessenger constraints, investigating the impact of prior assumptions and extreme-mass observations. Quark matter is modeled within an interacting MIT bag framework that consistently accommodates color-superconducting phases (2SC, 2SC+s, and CFL) and perturbative QCD corrections. We find that quark star models exhibit a distinct advantage in naturally accommodating the ultra-low mass object HESS J1731-347, a configuration that is challenging for standard neutron star models. In the high-mass regime, the interpretation of the secondary component of GW190814 is shown to be strongly prior-dependent: only broad priors allow for the substantial stiffness required to support such a massive object ($\sim$2.6 M$_\odot$), while more restrictive priors favor a softer equation of state consistent with standard pulsar populations. Microscopically, we demonstrate that current data tightly constrain the effective bag constant and the overall stiffness, but cannot distinguish between different color-superconducting phases. Furthermore, we validate a reduction of the model to two effective parameters without loss of information. Our results indicate that if quark stars exist, their sound speeds consistently exceeds the conformal limit ($c_s^2>1/3$) at stellar densities.

astro-ph.HE↗
arXiv2026

Bridging External and Parametric Knowledge: Mitigating Hallucination of LLMs with Shared-Private Semantic Synergy in Dual-Stream Knowledge

Retrieval-augmented generation (RAG) aims to mitigate the hallucination of Large Language Models (LLMs) by retrieving and incorporating relevant external knowledge into the generation process. However, the external knowledge may contain noise and conflict with the parametric knowledge of LLMs, leading to degraded performance. Current LLMs lack inherent mechanisms for resolving such conflicts. To fill this gap, we propose a Dual-Stream Knowledge-Augmented Framework for Shared-Private Semantic Synergy (DSSP-RAG). Central to it is the refinement of the traditional self-attention into a mixed-attention that distinguishes shared and private semantics for a controlled knowledge integration. An unsupervised hallucination detection method that captures the LLMs' intrinsic cognitive uncertainty ensures that external knowledge is introduced only when necessary. To reduce noise in external knowledge, an Energy Quotient (EQ), defined by attention difference matrices between task-aligned and task-misaligned layers, is proposed. Extensive experiments show that DSSP-RAG achieves a superior performance over strong baselines.

cs.CL↗