Searcharxiv⌕ Search

arXiv subjects

Qiang Zhu

Publications and source records attributed to Qiang Zhu.

At least 19 recordsLinked to original sources
arXiv2026

Benchmarking Universal Machine Learning Force Fields for Crystal Structure Prediction of High-Energy Molecular Systems

Recent developments of universal machine learning interatomic potentials (UMLIPs) offer a fast route for screening molecular crystals based on geometry relaxation and energy ranking, but their reliability across chemically diverse energetic materials remains elusive. In particular, it is unclear whether or not these UMLIPs are over-sensitive to break the desired molecular connectivity for relaxing the periodic crystals. Herein we tested the hypothesis that classical force-field pre-relaxation can provide a more suitable starting geometry for subsequent UMLIP relaxation on a large database of high energy molecular crystals. Three models (MACE, MACE-OFF and UMA) in conjunction with the General Amber Force Field (GAFF) were applied to test this hypothesis. Among them, direct MACE-OFF and UMA showed very high relaxation success and preserved the reference geometries most closely, but they still exhibit failures for some rare cases. Using GAFF pre-relaxation can systematically reduce the number of failed relaxations with lower computational costs. Our comparative failure and robustness analyses revealed distinct trade-offs among the evaluated models. Among them, MACE-OFF achieves a better compromise between potential energy surface smoothness, structural fidelity, and stress convergence, serving as a good choice to provide a reliable foundation for automated structural optimization.

cond-mat.mtrl-sci↗
arXiv2026

The Illusion of $\textit{What If}$: Evaluating the Breakdown of Counterfactual Reasoning in LLMs

Counterfactual reasoning requires models to reason beyond the observed world and explain how altered conditions propagate through downstream consequences. Existing benchmarks largely target bounded settings with fixed variables or single gold outcomes, overlooking open-domain scenarios requiring causal-process evaluation. To this end, we present $\textbf{WhatIfBench}$, a diagnostic benchmark for open-domain, open-form, long-horizon counterfactual causal reasoning, containing 220 what-if questions across STEM, HSS, and Hybrid scenarios. To evaluate free-form responses, we further propose $\textbf{PRISM}$, which first converts each natural-language explanation into a Response-Derived Semantic Causal Graph of events, states, and mechanisms. On top of this graph, PRISM then jointly applies a Process Metric assessing graph-level causal validity and a Rubric Metric assessing answer-level explanatory adequacy. Evaluating six frontier LLMs with this framework, we find that WhatIfBench remains far from saturated: even the strongest model reaches only a 64.62% final score. Further analysis reveals persistent causal gaps, premise drift, and topology fragmentation, suggesting that fluent counterfactual narratives often mask fragile causal processes. The benchmark, code, and evaluation scripts are available at $\href{https://github.com/zju-gt/WhatIfBench}{WhatIfBench}$.

cs.AI↗
arXiv2026

High-throughput Discovery of Magnetic Rare Earth Transition Metal Alloys

We present an accelerated materials discovery framework that combines diffusion-based crystal structure generation with hierarchical screening to identify new rare-earth--transition-metal magnets simultaneously achieving high magnetization and thermodynamic stability. Using this workflow, we systematically explored over 3000 binary (R-T) and ternary (R-T-T$'$) compositions spanning R~$\in \{\text{Y, Sm}\}$, T~$\in \{\text{Fe, Co, Ni}\}$, and T$' \in \{\text{Ti, V, Cr, Mn, Cu, Zn}\}$, and filtered approximately 240{,}000 generated crystal structures through machine-learning interatomic potential prescreening and spin-polarized density functional theory validation. We identify 300+ low-energy magnetic candidates within 0.1~eV/atom above the convex hull at the DFT level, including 5 thermodynamically stable phases. The highest saturation magnetization reaches ${\sim}1.8$~T in Fe-rich binary and ternary phases (SmFe$_{12}$, YFe$_{12}$, YFe$_{18}$Ti and Sm$_2$Fe$_{16}$Mn). Symmetry analysis reveals that the majority of ternary candidates are subgroup derivatives of known binary prototypes through Wyckoff site splitting that accommodates T$'$ substitution. Site-resolved magnetic moment analysis further shows that Mn aligns ferromagnetically with the Fe sublattice with minimal magnetization loss, whereas Cr couples antiferromagnetically, providing systematic guidance for dopant selection. These findings demonstrate a generalizable strategy for targeted magnetic materials discovery and suggest that extending generative searches to larger unit cells ($>$20 atoms) with higher Fe fractions is a promising route toward stable phases with saturation magnetization exceeding 1.8~T.

cond-mat.mtrl-sci↗
arXiv2026

MetaRAG: Belief-Action Aligned Policy Optimization for Agentic RAG

Agentic retrieval-augmented generation (RAG) requires language models to decide when to continue searching and when to answer. Existing RL-based methods rely on external supervision and overlook the agent's internal belief about whether the current evidence is sufficient. To address this problem, we reformulate the search decision quality as belief-action alignment and propose MetaRAG, a belief-action aligned policy optimization framework for agentic RAG. MetaRAG uses Verify-first Action Generation to elicit an explicit verification process before each actual action, and Internal Belief Probing to estimate the policy model's own answerability belief from the same question-history context. Based on these, MetaRAG derives a consistency reward that is further gated by answer correctness, avoiding reinforcement of internally consistent but incorrect trajectories. The belief probe is used only during training and introduces no inference-time overhead. Experiments on seven public QA benchmarks show that MetaRAG consistently improves the accuracy-efficiency trade-off over strong RL-based agentic RAG baselines, with gains that transfer to deep research settings, different optimizers, and multiple model backbones.

cs.AI↗
arXiv2026

NGS-Marker: Robust Native Watermarking for 3D Gaussian Splatting

With the rapid development and adoption of 3D Gaussian Splatting (3DGS), the need for effective copyright protection has become increasingly critical. Existing watermarking techniques for 3DGS mainly focus on protecting rendered images via pre-trained decoders, leaving the underlying 3D Gaussian primitives vulnerable to misuse. In particular, they are ineffective against Partial Infringement, where an adversary extracts and reuses only a subset of Gaussians. In this paper, we propose NGS-Marker, a novel native watermarking framework for 3DGS. It integrates a jointly trained watermark injector and message decoder, and employs a gradientbased progressive injection strategy to ensure full-scene coverage. This enables robust ownership decoding from any local region. We further extend NGS-Marker with hybrid protection (combining native and indirect watermarks) and support for multimodal watermarking. Extensive experiments demonstrate that NGS-Marker effectively defends against partial infringement while offering practical flexibility for real-world deployment.

cs.CV↗
arXiv2026

$D^{2}R^{2}$: Discrete Diffusion with Regulation Reinforcement for Single-Cell Perturbation Prediction

Predicting single-cell transcriptomic responses to genetic perturbations is central to functional genomics and virtual-cell modeling. Existing approaches, however, typically predict an entire expression profile as a whole, leaving the order in which individual gene responses are generated unmodeled. To address this problem, we introduce \textbf{$D^{2}R^{2}$} (\textbf{D}iscrete \textbf{D}iffusion with \textbf{R}egulation \textbf{R}einforcement), which reformulates perturbation prediction as regulation-guided gene-wise progressive generation. A Masked Discrete Diffusion Model represents expression as ordinal tokens and reconstructs a fully masked profile step by step, allowing generated gene responses to condition those that remain masked. A Regulatory Policy Module initializes the generation policy from a gene regulatory network inferred from control cells and adapts it to the perturbation and current partially generated state. Then, group-relative policy optimization refines only the ordering policy using final perturbation-effect agreement as reward. Across Norman19 and VCC-H1, $D^{2}R^{2}$ achieves the best performance on all five metrics on Norman19 and remains competitive on H1. Controlled ablations holding the generator and generation budget fixed show that biological-prior ordering improves over random ordering and is more reliable than uncertainty-based heuristics, whereas reversing the biological-prior ordering degrades every metric. Biological analyses further show that the refined policy prioritizes regulatory genes early while promoting perturbation-specific transcription factors and responsive genes. These results establish gene generation order as an effective, controllable, and biologically interpretable dimension of single-cell perturbation prediction.

cs.AI↗
arXiv2026

CARE: Confidence-Aware Reasoning for Reliable Medical VQA

Reinforcement Fine-Tuning (RFT) has enabled medical Multimodal Large Language Models (MLLMs) to produce Chain-of-Thought (CoT) reasoning for visual question answering, yet these models suffer from $\textit{confidence miscalibration}$---a systematic gap between expressed certainty and actual diagnostic accuracy that undermines clinical trust. We propose $\textbf{CARE}$, a $\textbf{C}$onfidence-$\textbf{A}$ware medical $\textbf{RE}$asoning framework that jointly optimizes accuracy and calibration through a dual-stage pipeline. First, a scalable Medical-CoT synthesis provides structured cold-start data for Supervised Fine-Tuning. Second, Group Relative Policy Optimization (GRPO) with a novel $\textbf{Confidence-Aware Reward (CAR)}$ mechanism ties the model's confidence to diagnostic correctness within the reward signal. Across three Medical VQA benchmarks, $\textbf{CARE}$ achieves the highest diagnostic accuracy while obtaining the lowest Expected Calibration Error and Hallucination Rate, establishing a foundation for trustworthy clinical decision support. Our code is available at https://github.com/anotherbricki/CARE.

cs.CV↗
arXiv2026

CrystalGRPO: Target-Aligned and Coverage-Preserving Reinforcement Learning for Flow-Based Crystal Structure Prediction

Flow-based generative models can efficiently produce candidate structures for crystal structure prediction (CSP), but their pretrained objectives do not directly optimize downstream target recovery. Reinforcement-learning post-training offers a flexible solution, yet existing approaches rely primarily on energy rewards and coordinate-only stochastic policies. Predicted energy does not identify the reference polymorph, while reward-driven concentration can reduce the candidate coverage required for Top-N recovery. We introduce CrystalGRPO, a CSP-aligned post-training framework that extends existing ODE-to-SDE policy constructions to the joint coordinate--lattice state. CrystalGRPO combines MACE-predicted energy with a StructureMatcher-based recovery score and provides two operating modes: CrystalGRPO-Q, which prioritizes single-draw recovery, and CrystalGRPO-C, which combines full-trajectory reference regularization with a coverage-aware group advantage to preserve finite-budget target recovery. Across MP-20 and MPTS-52 with PXRDGen and OMatG backbones, both variants reduce one- and twenty-sample RMSE relative to coordinate-only reinforcement in all four backbone--dataset settings. CrystalGRPO-Q consistently improves Top-1, whereas CrystalGRPO-C achieves a higher Top-20 across all settings.

cs.LG↗
arXiv2026

CLEAR: Conflict-aware Learning via Evidence-guided Adaptive Routing for Unified Sparse-View 3D Gaussian Super-Resolution

Sparse-view 3D Gaussian Splatting Super-resolution is highly challenging since the sparse and low-resolution (LR) inputs lack sufficient geometric and high-frequency information for accurate reconstruction. To achieve high-quality reconstruction, existing sparse-view super-resolution methods adhere to two-stage pipeline that performs LR Gaussian reconstruction and then high-resolution (HR) Gaussian refinement, which directly results in stage-wise Gaussian transfer and reconstruction error accumulation. To this end, we propose CLEAR, a Conflict-aware Learning via Evidence-guided Adaptive Routing, as the first unified single-stage framework for Sparse-view 3D Gaussian Splatting Super-resolution. Specifically, CLEAR performs joint the optimization of authentic LR observations and external HR priors within a unified Gaussian representation. To mitigate the gradient conflicts introduced by sparse supervision during training, we propose a Gaussian-wise conflict-aware optimization strategy that regards the LR gradient as a reliable anchor and applies evidence-conditioned soft correction only to severe HR conflicts. Moreover, to recover high-frequency details, we introduce an evidence-guided Patch-to-Gaussian routing mechanism which estimates patch reliability and detail demand, lifts them into Gaussian space, and selectively routes high-frequency gradients and densification. Finally, we employ shared Gaussian dropout and a detached mid-training anchoring to enhance the robustness of training framework. Extensive experiments on both synthetic and real-world $4\times$ super-resolution benchmarks demonstrate that CLEAR consistently achieves state-of-the-art rendering quality and superior geometric fidelity.

cs.CV↗
arXiv2026

Living-Harness Is an Interactive-Agent Evolver

Large language model (LLM) agents may recover from a failure within an episode or after a retry, yet the same execution failure can recur in later tasks because post-episode feedback rarely revises the persistent harness that guides future interactions. Static harnesses improve reliability through fixed tools, context, memory, and workflow structures, but remain unchanged after deployment. We propose $\textbf{Living-Harness}$, a self-evolving agent harness that converts each completed trajectory and its evaluator signals into posterior evidence for bounded harness updates. Guided by a domain-level $\textbf{Evolution-SOP}$ ($\textbf{S}$tandard $\textbf{O}$perating $\textbf{P}$rocedure), Living-Harness extracts an episode abstraction and structured update evidence, and writes two complementary forms of procedural knowledge: episodic memory that records trigger conditions, failure patterns, and recovery actions, and a state graph that records state nodes, repair edges, and transition rules. The updated harness state is retrieved to guide future interactions, while tools and base context remain frozen, allowing procedural repairs to accumulate across evolution cycles. On eight interactive environments derived from $\tau^2$-Bench and MultiWOZ-2.4, Living-Harness improves average Pass@1 over the strongest interactive baseline by 10.07 and 9.91 percentage points, respectively, and supports retrieval-only reuse of the evolved harness state across model backbones. Our code will be made publicly available soon at https://github.com/anotherbricki/Living-Harness.

cs.MA↗
arXiv2026

LOTAPO: Leave-One-Turn Attribution for Self-Generated Process Rewards in Multi-Turn Search Reasoning

Reinforcement learning for multi-turn search reasoning typically relies on terminal outcome rewards, which cannot distinguish useful, redundant, and harmful intermediate interactions. We propose LOTAPO , a self-generated process-supervision method based on backward leave-one-turn attribution. For each search turn, LOTAPO replaces the turn and its retrieval observation with a fixed [DELETE] placeholder and measures the resulting change in the current policy's mean log-likelihood of the gold answer. This Answer-Likelihood Gain estimates the turn's contribution while preserving all downstream interactions, allowing early evidence to be evaluated in the complete reasoning context. LOTAPO further applies sign-consistency gating, retaining only normalized process advantages whose directions agree with their raw attribution scores. The method requires no additional reward model, teacher, verifier, or LLM-as-a-Judge. Across seven knowledge-intensive question-answering datasets with local retrieval, LOTAPO achieves an average exact-match score of 0.326, outperforming the strongest step-reward baseline, IGPO, by 0.053. Ablations show complementary benefits from backward attribution and sign-consistency gating, demonstrating that policy-derived retrospective attribution can provide effective process supervision for multi-turn search agents.

cs.AI↗
arXiv2026

CARL: Constraint-Aware Reinforcement Learning for Planning with LLMs

Despite their strong reasoning capabilities and extensive world knowledge, Large Language Models (LLMs) frequently generate plans that violate task constraints, undermining their reliability in real-world applications. This deficiency arises from a lack of systematic mechanisms to incorporate constraint information during the generation process. While existing approaches attempt to mitigate this by relying on external tools or task decomposition, they fail to enhance the model's intrinsic constraint awareness. To address this, we propose Constraint-Aware Reinforcement Learning (CARL), a novel RL framework designed to strengthen LLMs' intrinsic focus on constraints. CARL introduces a constraint-aware reward by comparing the model's output distributions under constrained and unconstrained inputs, encouraging constraint focus and penalizing neglect. Compatible with various RL frameworks and requiring no external solvers or top models, CARL enables scalable, end-to-end constraint-aware planning. Extensive experiments on BlocksWorld, TravelPlanner, and T-Eval demonstrate that CARL significantly outperforms standard Reinforcement Fine-Tuning (RFT) baselines and state-of-the-art reasoning models, exhibiting a markedly increased focus on constraints.

cs.AI↗
arXiv2026

STAPO: Selective Trajectory-Aware Policy Optimization for LLM Agent Training

Reinforcement Learning (RL) is the dominant paradigm for training Large Language Model (LLM) agents on long-horizon tasks. However, sparse and delayed rewards often lead to trajectory neglect, in which agents lose focus on the task goal and interaction history at intermediate steps. Prior work has explored step-level supervision using Shannon-entropy-based uncertainty signals, which conflate inherent state complexity with agent confidence and therefore provide unreliable estimates of decision reliability. To address this issue, we propose normalized entropy, which measures confidence deviations relative to an agent's average behavior under a given state, thereby strengthening the association between low-quality actions and trajectory neglect. Building on this insight, we introduce Selective Trajectory-Aware Policy Optimization (STAPO), a hierarchical group-based RL framework. STAPO leverages normalized entropy to locate outlier steps associated with trajectory neglect and optimizes them via a joint mechanism of trajectory-aware reward and trajectory-independent penalty, enhancing trajectory awareness while preserving training stability. Extensive experiments on ALFWorld, WebShop, and Search-Augmented QA demonstrate that STAPO achieves state-of-the-art performance while substantially alleviating trajectory neglect, validating its effectiveness and robustness for agentic tasks.

cs.AI↗
arXiv2026

Confidence Calibration for Multimodal LLMs: An Empirical Study through Medical VQA

Multimodal Large Language Models (MLLMs) show great potential in medical tasks, but their elicited confidence often misaligns with actual accuracy, potentially leading to misdiagnosis or overlooking correct advice. This study presents the first comprehensive analysis of the relationship between accuracy and confidence in medical MLLMs. It proposes a novel method that combines Multi-Strategy Fusion-Based Interrogation (MS-FBI) with auxiliary expert LLM assessment, aiming to improve confidence calibration in Medical Visual Question Answering (VQA). Experiments demonstrate that our method reduces the Expected Calibration Error (ECE) by an average of 40\% across three Medical VQA datasets, significantly enhancing MLLMs' reliability. The findings highlight the importance of domain-specific calibration for MLLMs in healthcare, offering a more trustworthy solution for AI-assisted diagnosis.

cs.CV↗
arXiv2026

Ytterbium lattice clock with uncertainty of $1.1\times 10^{-18}$ and instability of low $10^{-19}$

We report an optical lattice clock based on $^{171}$Yb atoms with a total systematic uncertainty of $1.1\times 10^{-18}$. In-vacuum buildup cavity was employed to enhance the lattice light power. Differential frequency measurement between two identical clocks facilitate the evaluation of systematic shifts. Synchronous comparison of the two clocks reached a stability level of $2.7\times 10^{-19}$ in an averaging time of 216,000 s. The magic frequency $\nu_{\mathrm{zero}}$ was determined to be 394 798 258.3(1) MHz. Under typical operating conditions, the lattice light shift is controlled at an uncertainty level of $3\times 10^{-19}$. The blackbody radiation (BBR) shield which is placed in vacuum provides a well-characterized BBR environment, enabling an uncertainty contribution of $8.7\times 10^{-19}$ from the BBR Stark shift. Other systematic shifts have also been evaluated. The two clocks will be used for remote frequency comparisons between Shanghai and Wuhan.

physics.atom-ph↗
arXiv2026

Ab-initio Crystal Structure Determination from Powder X-Ray Diffraction

Determining crystal structures from powder X-ray diffraction (PXRD) has been a significant challenge in materials science, particularly when experimental data contain noise or the target structure has a high complexity. While recent AI generative models show promise for rapid structure generation, they predominantly employ data-driven approaches to learn direct mappings between PXRD patterns and crystal structures, often failing on complex or out-of-distribution cases. In this work, we present a hybrid ab-initio approach that decomposes structure determination into a two-stage optimization problem: (1) discrete selection of space group symmetry, unit cell parameters, and Wyckoff site combinations; and (2) continuous optimization of atomic coordinates within the selected Wyckoff positions. By integrating AI-based techniques for peak profile analysis, density estimation and energy minimization with physics-informed constraints, our method systematically overcomes limitations of purely data-driven PXRD solvers. We demonstrate that this hierarchical optimization framework enables robust structure determination even for challenging cases with high structural complexity or limited experimental data quality. Our approach provides a principled pathway for incorporating crystallographic knowledge into AI models for more reliable and generalizable crystal structure determination.

cond-mat.mtrl-sci↗
arXiv2026

PairDropGS: Paired Dropout-Induced Consistency Regularization for Sparse-View Gaussian Splatting

Dropout-based sparse-view 3D Gaussian Splatting (3DGS) methods alleviate overfitting by randomly suppressing Gaussian primitives during training. Existing methods mainly focus on designing increasingly sophisticated dropout strategies, while they overlook the resulting inconsistencies among different dropped Gaussian subsets. This oversight often leads to unstable reconstruction and suboptimal Gaussian representation learning.In this paper, we revisit dropout-based sparse-view 3DGS from a consistency regularization perspective and propose PairDropGS, a Paired Dropout-induced Consistency Regularization framework for sparse-view Gaussian splatting. Specifically, PairDropGS first constructs a pair of the dropped Gaussian subsets from a shared Gaussian field and designs a low-frequency consistency regularization to constrain their low-frequency rendered structures. This design encourages the shared Gaussian field to preserve stable scene layout and coarse geometry under different random dropouts, while avoiding excessive constraints on ambiguous high-frequency details. Moreover, we introduce a progressive consistency scheduling strategy to gradually strengthen the consistency regularization during training for stability and robustness of reconstruction. Extensive experiments on widely-used sparse-view benchmarks demonstrate that PairDropGS achieves superior training stability, significantly outperforms existing dropout-based 3DGS methods in reconstruction quality, while exhibiting the simplicity and plug-and-play nature for improving dropout-based optimization.

cs.CV↗
arXiv2026

DOC-GS: Dual-Domain Observation and Calibration for Reliable Sparse-View Gaussian Splatting

Sparse-view reconstruction with 3D Gaussian Splatting (3DGS) is fundamentally ill-posed due to insufficient geometric supervision, often leading to severe overfitting and the emergence of structural distortions and translucent haze-like artifacts. While existing approaches attempt to alleviate this issue via dropout-based regularization, they are largely heuristic and lack a unified understanding of artifact formation. In this paper, we revisit sparse-view 3DGS reconstruction from a new perspective and identify the core challenge as the unobservability of Gaussian primitive reliability. Unreliable Gaussians are insufficiently constrained during optimization and accumulate as haze-like degradations in rendered images. Motivated by this observation, we propose a unified Dual-domain Observation and Calibration (DOC-GS) framework that models and corrects Gaussian reliability through the synergy of optimization-domain inductive bias and observation-domain evidence. Specifically, in the optimization domain, we characterize Gaussian reliability by the degree to which each primitive is constrained during training, and instantiate this signal via a Continuous Depth-Guided Dropout (CDGD) strategy, where the dropout probability serves as an explicit proxy for primitive reliability. This imposes a smooth depth-aware inductive bias to suppress weakly constrained Gaussians and improve optimization stability. In the observation domain, we establish a connection between floater artifacts and atmospheric scattering, and leverage the Dark Channel Prior (DCP) as a structural consistency cue to identify and accumulate anomalous regions. Based on cross-view aggregated evidence, we further design a reliability-driven geometric pruning strategy to remove low-confidence Gaussians.

cs.CV↗