arXiv subjects

Dan Zhang

Publications and source records attributed to Dan Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Chiral and $U(1)_A$ symmetries in background magnetic fields from lattice QCD

We study chiral symmetry and singlet $U(1)_A$ symmetry in QCD in a background magnetic field using lattice QCD. We first clarify the neutral-sector symmetry structure in a pure magnetic background, where the unequal electric charges of the light quarks explicitly reduce the non-singlet flavor symmetry. We identify the neutral-pion--sigma susceptibility difference, $χ_{π^0}-χ_σ$, as the chiral-partner splitting associated with the surviving neutral non-singlet axial symmetry, and the neutral-pion--delta susceptibility difference, $χ_{π^0}-χ_{δ^0}$, as the singlet $U(1)_A$ partner splitting. We also discuss the disconnected contribution to the neutral-pion susceptibility and its continuum constraint. Numerical results are obtained on fixed-scale $(2+1)$-flavor HISQ ensembles with $m_l=m_s^{\rm phys}/10$, corresponding to a pion mass of about $220~{\rm MeV}$ at vanishing magnetic field. We find that the neutral chiral-partner splitting increases with the magnetic field strength $eB$ at low temperature and decreases at sufficiently large $eB$ near the crossover, providing susceptibility-splitting counterparts of magnetic catalysis and inverse magnetic catalysis, respectively. The singlet $U(1)_A$ partner splitting shows an analogous low-temperature enhancement and large-field suppression near the crossover, with the suppression setting in at larger $eB$ and remaining milder than in the chiral channel. These results provide a first lattice-QCD study of neutral-sector probes of chiral and singlet $U(1)_A$ partner susceptibility splittings in background magnetic fields.

hep-lat↗

arXiv2026

GUI-AC: Enhancing Continual Learning in GUI Agents

Graphical User Interfaces (GUIs) serve as the dominant medium for human-computer interaction, yet building GUI agents that generalize across the vast diversity of real-world interface environments, with the same flexibility and robustness that humans naturally exhibit, remains unsolved. Notably, GUI data are inherently non-stationary: the continual emergence of previously unseen interface instances (e.g., novel domains and resolutions) induces persistent distribution shifts, significantly impeding the continual learning of existing GUI agents. Reinforcement fine-tuning (RFT) has attracted considerable attention as a promising approach. Nevertheless, RFT exhibits pronounced instability in its grounding capability, manifested as sharp reward discontinuities and high-variance oscillations. The imbalanced distribution of rollout outcomes introduces substantial noise into advantage estimation, leading to policy overconfidence. The fixed clipping bound suppresses the increase in policy probabilities needed to adapt to new distributions, leading to a collapse in exploration capacity. To address these challenges, we propose GUI-AC, a method that enhances the continual learning capability of GUI agents. GUI-AC introduces grounding certainty to support two core mechanisms: (i) Adaptive Advantage, which down-weights noisy advantage estimates to prevent policy overconfidence; and (ii) Dynamic Clipping, which relaxes the clipping bound to encourage exploration range. Extensive experiments show that these mechanisms jointly improve performance, enabling our method to surpass state-of-the-art baselines. Code is available anonymously at https://github.com/Can-Lin/GUI-AC.

cs.CV↗

arXiv2026

R$^2$PO: Decoupling Rollout and Inference Policies for LLM Reasoning

Existing reinforcement learning methods for LLM reasoning implicitly assume that the policy generating training trajectories should coincide with the one producing inference responses. We argue that this is a misleading inductive bias: the optimization-optimal trajectory distribution favors informative gradients, whereas the inference-optimal response distribution emphasizes accuracy and consistency. Forcing both into a single policy entangles their gradients and suppresses exploration. We propose R$^2$PO (Residual Rollout Policy Optimization), which attaches a lightweight Residual Rollout-Head atop the policy to decouple training trajectories from inference responses, diversifying rollouts during training while keeping inference generation intact. Experiments show that R$^2$PO consistently outperforms baselines, with average accuracy gains of 3.4% on MATH-500 and 1.3% on APPS, alongside more diverse rollouts and reduced length bias. Our code is available at https://github.com/RRPO-ARR/Code.

cs.LG↗

arXiv2026

Re-evaluating Confidence Remasking in Masked Diffusion Language Models

Masked diffusion language models (dLLMs) have recently emerged as a competitive alternative to autoregressive language models, with the promise of faster inference via parallel token generation. A notable limitation of the masked formulation, however, is that once a token has been unmasked it can no longer be revised, leaving dLLMs vulnerable to early sampling mistakes. To address this, a growing body of work has sought to extend masked dLLMs with self-correcting (remasking) capabilities. One appealing subset of these methods does so in a training-free, post-hoc manner based on token confidences, with encouraging early reported results. In this work, we revisit the empirical evaluation of a representative post-hoc remasking method, WINO [Hong et al., 2026], and find that under standard decoding settings (shorter block lengths) it brings little-to-no benefit over confidence-based unmasking alone [Wu et al., 2025]. Extending the evaluation to non-greedy decoding, we find that while confidence-based remasking can mitigate errors introduced by increased stochasticity to some extent, it also exacerbates the diversity collapse previously reported for confidence-based unmasking. Overall, our results show that the benefits of post-hoc confidence-based remasking are highly setting-dependent, underscoring the need for a more comprehensive evaluation framework.

cs.LG↗

arXiv2026

Report on CHIIR 2026 Workshop on Generative AI and Academic Search (GAI&AS)

This report summarizes the CHIIR 2026 Workshop on Generative AI and Academic Search (GAI\&AS), which examined how GenAI is reshaping academic search systems and research practices. The workshop brought together researchers in human information interaction and information retrieval to explore key challenges and opportunities in designing and evaluating future academic search systems that integrate GenAI, moving beyond traditional document retrieval to support summarization, recommendation, synthesis, and conversational interaction. Participants' interests and discussions focused on three thematic clusters: foundations and principles, applications and opportunities, and search-as-learning. Across these themes, the workshop highlighted the importance of academic search systems in supporting transparency, credibility, research integrity, and long-term scholarly needs, as well as in fostering higher-order cognitive processes. Participants discussed guiding theories, design principles, methodological approaches, partnerships, and community-building efforts aimed at advancing human-centered GenAI-enhanced academic search systems. Overall, the workshop demonstrated strong community interest and a diverse range of ongoing and emerging research initiatives at the intersection of GenAI and academic search.

cs.IR↗

arXiv2026

TRACE: Trajectory Risk-Aware Compression for Long-Horizon Agent Safety

Long-horizon LLM agents produce safety evidence across long trajectories, where sparse, delayed, and compositional risk signals often escape local moderation. Existing turn-level or short-context detectors struggle to reliably retain and aggregate such evidence over extended horizons. We reframe long-horizon agent safety detection as trajectory-level evidence compression and propose Trajectory Risk-Aware Compression for Long-Horizon Agent Safety (TRACE). TRACE uses a Compressor-Reader design: the Compressor encodes the full trajectory into a compact latent evidence state under trajectory-level supervision, and the Reader judges the raw trajectory with this latent evidence state as a safety reference. This design helps aggregate dispersed risk cues and reduce premature evidence loss. Across ASSEBench, Pre-Ex-Bench, and R-Judge, TRACE achieves the best accuracy on all evaluated backbones, improving over strong baselines by up to 12.6 percentage points. On LongSafety, TRACE shows smaller performance degradation as context length grows. Attention visualizations and case studies suggest that the compressed reference helps the Reader focus on risk-critical segments and recover cross-step evidence. Code is available at https://github.com/Peregrine123/TRACE_official.

cs.AI↗

arXiv2026

CodecCap: High-Fidelity Codec-Inspired Residual Modeling for Dense Video Captioning

Existing video captioning methods struggle to balance visual fidelity and redundancy: holistic captions are compact but lose fine-grained evidence, whereas segment-wise captions improve coverage but introduce heavy redundancy. We propose CodecCap, a codec-inspired framework for high-fidelity dense video captioning. Analogous to video codecs, CodecCap represents videos using keyframe and residual captions. Keyframe captions exhaustively encode stable visual context, while residual captions capture temporally only localized actions, motions and changes. This effectively preserves fine-grained visual evidence while reducing redundant descriptions. To quantify the fidelity of captions, we introduce VidCapQA, a caption-then-QA benchmark with 1,000 questions across 14 capability dimensions. Results on VidCapQA show that captions directly generated by strong VLMs still miss many visual details, highlighting caption representation as a critical bottleneck. Experiments show that CodecCap significantly surpasses direct captioning with the same underlying VLMs, suggesting keyframe-residual captioning a way for high-fidelity video-language supervision. We further use CodecCap to construct CodecVDC-100K, a large-scale dense captioning dataset with anchor, residual, scene-level, and video-level supervision.

cs.CV↗

arXiv2026

RotMoLE: Enhancing Mixture of Low-Rank Experts through Rotational Gating Mechanism

While Large Language Models (LLMs) are commonly fine-tuned to handle domain-specific tasks before being applied to vertical applications, adapting them to complex scenarios with diverse specialized knowledge remains challenging. Meanwhile, Mixture-of-Experts (MoE) architecture has risen as a crucial paradigm for training LLMs, and some recent works have also incorporated MoE into Parameter-Efficient Fine-Tuning (PEFT) to propose the Mixture of Low-rank Experts (MoE-LoRA), to enhance the power of low-rank adapters for learning complicated knowledge. However, conventional gating mechanisms in MoE typically apply only a scalar reweighing to selected experts, thereby limiting their underlying capacity of representation and generalization. Motivated and enabled by the low-rank structures in MoE-LoRA, we propose RotMoLE, a specialized MoE framework for low-rank experts featuring an additional rotation gate. Beyond simple scaling, RotMoLE implements a rotation mechanism for each selected expert, enabling superior expert exploitation and specialization for learning diverse data, especially when expert candidates are limited. Empirical results on complex multi-task and multilingual training scenarios validate our effectiveness.

cs.LG↗

arXiv2026

How to Steer Your Multi-Agent System: Human-LLM Collaborative Planning

In orchestrated multi-agent systems, humans often struggle to manage plans due to their complexity and limited transparency. Existing approaches rely on outcome-level supervision, where users verify only final outputs without visibility into intermediate reasoning. We formalize a design space for human-LLM co-planning interactions along three axes: mode (semantic vs. structural), scope (global vs. targeted), and level (low vs. high-level edits). We realize it in AMBIPOM, a prototype supporting process-level supervision through both semantic and structural interactions. Through a user study, we characterize how users navigate this space, revealing hybrid workflows and effort-control-risk trade-offs; through a controlled benchmark, we analyze how LLMs revise plans under varying scope and revision strategies. Our findings yield design insights for more transparent, controllable, and effective human-AI co-planning. We release code and data at https://github.com/megagonlabs/ambipom.

cs.MA↗

arXiv2026

Chiral Properties of $(2\!+\!1)$-Flavor QCD in Magnetic Fields at Zero Temperature

We present a lattice QCD study of the chiral properties of $(2\!+\!1)$-flavor QCD in background magnetic fields at zero temperature with physical pion masses. Simulations are performed using the highly improved staggered quark action across four different lattice spacings to enable a controlled continuum extrapolation. We compute the renormalized chiral condensates together with pseudoscalar meson masses and decay constants for pions, kaons, and the fictitious $η^0_{s\bar{s}}$ pseudoscalar as functions of the magnetic-field strength $eB$ up to $eB\simeq1.2$ $\mathrm{GeV}^2$. The chiral condensates exhibit clear magnetic catalysis, increasing monotonically with the field strength. In the meson sector, neutral pseudoscalar masses decrease steadily with $eB$, whereas charged pseudoscalar masses display a nonmonotonic response: They rise at small fields, consistent with the lowest-Landau-level expectation, but then saturate and slightly decrease at larger fields, signaling sizable internal-structure effects. At the same time, neutral pseudoscalar decay constants are strongly enhanced by the magnetic field. To quantify deviations from chiral symmetry relations, we isolate the magnetic-field-induced shift in the Gell-Mann--Oakes--Renner corrections and find it to remain small for the neutral pion but to become sizable for the neutral kaon. To elucidate the origin of the magnetic response, we separately analyze the sea- and valence-quark contributions to both neutral and charged meson masses, finding that valence effects dominate at zero temperature. These results provide new insights into the interplay between QCD chiral symmetry breaking and strong magnetic fields.

hep-lat↗

arXiv2026

Quasinormal modes of a rotating loop quantum black hole

We investigate the quasinormal modes of a massless scalar field on an effective rotating loop quantum black hole background, constructed from a covariant spherical model via an improved Newman-Janis algorithm. Using the continued fraction method, we compute the spectrum for both fundamental and overtone modes, and systematically analyze how the frequencies depend on the quantum correction, spin, and angular structure of the perturbation. For all fundamental modes, increasing the quantum gravity correction monotonically reduces both the oscillation frequency and the damping rate, signaling slower oscillations and prolonged decay. Rotation imprints a nontrivial modulation: for a spherically symmetric perturbation, the real frequency displays a crossover as the spin grows, whereas this feature is suppressed once angular momentum is turned on; further activating the azimuthal component enhances the frequency and reduces the damping even more strongly. In the overtone sector, the rotating solution retains the hallmark quantum gravitational signatures of the spherical case - overtone outbursts and non-monotonic evolution - with rotation shifting these phenomena to weaker quantum corrections. Nonzero orbital angular momentum suppresses the outbursts, while the azimuthal degree of freedom boosts the frequency, giving rise to novel spectral inversions among higher overtones. Our results confirm that the effective rotating metric captures essential loop quantum gravity features, providing clear theoretical benchmarks for black hole spectroscopy and future gravitational-wave observations.

gr-qc↗

arXiv2026

GLM-5V-Turbo: Toward a Native Foundation Model for Multimodal Agents

We present GLM-5V-Turbo, a step toward native foundation models for multimodal agents. As foundation models are increasingly deployed in real environments, agentic capability depends not only on language reasoning, but also on the ability to perceive, interpret, and act over heterogeneous contexts such as images, videos, webpages, documents, GUIs. GLM-5V-Turbo is built around this objective: multimodal perception is integrated as a core component of reasoning, planning, tool use, and execution, rather than as an auxiliary interface to a language model. This report summarizes the main improvements behind GLM-5V-Turbo across model design, multimodal training, reinforcement learning, toolchain expansion, and integration with agent frameworks. These developments lead to strong performance in multimodal coding, visual tool use, and framework-based agentic tasks, while preserving competitive text-only coding capability. More importantly, our development process offers practical insights for building multimodal agents, highlighting the central role of multimodal perception, hierarchical optimization, and reliable end-to-end verification.

cs.CV↗

arXiv2026

Rubric-based On-policy Distillation

On-policy distillation (OPD) is a powerful paradigm for model alignment, yet its reliance on teacher logits restricts its application to white-box scenarios. We contend that structured semantic rubrics can serve as a scalable alternative to teacher logits, enabling OPD using only teacher-generated responses. To prove it, we introduce ROPD, a simple yet foundational framework for rubric-based OPD. Specifically, ROPD induces prompt-specific rubrics from teacher-student contrasts, and then utilizes these rubrics to score the student rollouts for on-policy optimization. Empirically, ROPD outperforms the advanced logit-based OPD methods across most scenarios, and achieving up to a 10x gain in sample efficiency. These results position rubric-based OPD as a flexible, black-box-compatible alternative to the prevailing logit-based OPD, offering a simple yet strong baseline for scalable distillation across proprietary and open-source LLMs. Code is available at https://github.com/Peregrine123/ROPD_official.

cs.LG↗

arXiv2026

Do Agents Need to Plan Step-by-Step? Rethinking Planning Horizon in Data-Centric Tool Calling

Explicit planning is a critical capability for LLM-based agents solving complex data-centric tasks, which require precise tool calling over external data sources. Existing strategies fall into two paradigms based on planning horizon: (1) full-horizon (FH), which generates a complete plan before execution, and (2) single-step horizon (SH), which interleaves each action (tool call) with incremental reasoning and observation. While step-by-step execution is a common default under the assumption that eager execution monitoring is necessary for adaptability, we revisit this assumption for well-defined data-centric tasks. Our controlled empirical study isolates planning horizon as the key architectural feature and systematically analyzes the effects of topological complexity and tool robustness on both paradigms. Our experiments across Knowledge Base Question Answering and Multi-hop QA show that FH planning with lazy replanning achieves accuracy parity with SH across varying depths, breadths, and robustness levels, while using 2-3x fewer tokens. These findings suggest that for well-defined data-centric tasks, eager step-wise monitoring is often unnecessary, and full-horizon planning with on-demand replanning can offer a more efficient default.

cs.CL↗

arXiv2026

Effects of Cosmic Muons on $μ$eV-to-meV Scale Axion Dark Matter Searches

We estimate the synchrotron radiation of cosmic muons in a uniform magnetic field in the $μ$eV-to-meV energy scale. Such events can potentially bring backgrounds to the axion dark matter searches. The GEANT4 software package is utilized to simulate the muon tracks in a cylindrical region of interest with an 8~T solenoid magnetic field applied. We further develop an analytical estimation of the angular-frequency-differential synchrotron radiation power spectra in this work as the cosmic muons span a wide range of Lorentz factor $γ$ and pitch angle $α$. We verify that the cosmic muons are not the dominant noise background for the current axion dark matter experiments on the $μ$eV scale because of the high quality factor $Q$ and fine energy resolution in the readout. However, without sufficient energy resolution in the detector readout, future broadband axion dark matter experiments will be vulnerable to the synchrotron radiation of these charged particles.

physics.ins-det↗

arXiv2026

Learning from Supervision with Semantic and Episodic Memory: A Reflective Approach to Agent Adaptation

We investigate how agents built on pretrained large language models (LLMs) can learn target classification functions from labeled examples without parameter updates. While conventional approaches like fine-tuning are often costly, inflexible, and opaque, we propose a memory-augmented framework that leverages LLM-generated critiques grounded in labeled data. Our framework uses episodic memory to store instance-level critiques - capturing specific past experiences - and semantic memory to distill these into reusable, task-level guidance. Across a diverse set of tasks and models, our best performing self-critique strategy (utilizing both memory types) yields an average improvement of 8.1 percentage points over the zero shot baseline, and 4.6pp over a RAG-based baseline that relies only on labels. However, improvements vary substantially across models and domains. To explain this variation, we introduce suggestibility - a novel metric capturing how receptive a model is to external reasoning provided in context. We use suggestibility to illuminate when and why memory augmentation succeeds or falls short. Beyond accuracy gains, we find pre-computed critiques substantially reduce inference-time computation for reasoning models, cutting thinking tokens by an average of 31.95% across all datasets by substituting for reasoning that the model would otherwise perform independently. Our findings highlight the conditions under which memory-driven, reflective learning can serve as a lightweight, interpretable, and efficient strategy for improving LLM adaptability.

cs.CL↗

arXiv2026

Assessing EMRI Detectability of the Rotating Quantum Oppenheimer-Snyder Black Hole

This letter presents an assessment of quantum gravity effects on extreme-mass-ratio inspirals (EMRIs) for the rotating quantum Oppenheimer-Snyder (qOS) black hole. Employing the adiabatic evolution, we compute the gravitational wave (GW) dephasing, which quantifies the cumulative phase shift induced by the quantum correction α . We further generate the augmented analytic kludge (AAK) waveform and investigate the faithfulness between the waveforms with and without the quantum parameter α for different values of a. Our results reveal that the quantum gravity effect induces detectable imprints in LISA, while the presence of rotation suppresses these signatures. This suggests that rotational degrees of freedom must be carefully accounted for when probing quantum gravity with EMRI observations.

gr-qc↗

arXiv2026

Learning from Demonstration with Failure Awareness for Safe Robot Navigation

Learning from demonstration is widely used for robot navigation, yet it suffers from a fundamental limitation: demonstrations consist predominantly of successful behaviors and provide limited coverage of unsafe states. This limitation leads to poor safety when the robot encounters scenarios beyond the demonstration distribution. Failure experiences, such as collisions, contain essential information about unsafe regions, but remain underutilized. The key difficulty lies in the fact that failure data do not provide valid guidance for action imitation, and their naive incorporation into policy learning often degrades performance. We address this challenge by proposing a failure-aware learning framework that explicitly decouples the roles of success and failure data. In this framework, failure experiences are used to shape value estimation in hazardous regions, while policy learning is restricted to successful demonstrations. This separation enables the effective use of failure data without corrupting policy behavior. We implement this design within an offline reinforcement learning (RL) setting and evaluate it in both simulation and real-world environments. The results show that our framework consistently reduces collision rates while preserving the task success rate, and demonstrate strong generalization across different environments and robot platforms.

cs.RO↗