Search arXiv⌕ Search

arXiv subjects

Yijie Wang

Publications and source records attributed to Yijie Wang.

At least 19 recordsLinked to original sources
arXiv2026

TreeRef-BFN: Equivariance-Free De Novo Molecule Generation based on 2D Topology and Internal 3D Geometry

De novo 3D molecular generation jointly models molecular size, topology, and geometry. Most methods pre-sample molecular size and generate Cartesian coordinates, limiting variable-size conditional tasks such as fragment completion and scaffold decoration while often relying on equivariant architectures. Internal-coordinate methods avoid rigid-body redundancy but typically require a known molecular graph or autoregressive construction, which may accumulate errors. We propose TreeRef, a tree-based molecular representation that assigns molecular topology and topology-dependent local 3D geometry to a naturally variable-size tree. RingRef nodes encode ring closures while preserving the tree structure, while Null nodes allow molecular size to emerge directly from node occupancy. Based on TreeRef, we develop TreeRef-BFN, a Bayesian Flow Network with a standard Transformer backbone that globally couples these locally defined variables and jointly generates discrete molecular variables and continuous local geometry. A single pretrained TreeRef-BFN supports unconditional generation and variable-size structure-conditioned 3D generation through masking alone, without retraining. Empirical studies demonstrate strong chemical validity, molecular stability, and diversity, accurate local geometric distributions, fast sampling, and competitive property-conditioned generation, establishing TreeRef-BFN as an efficient and flexible framework for 3D molecular generation.

cs.AI↗
arXiv2026

Evolutionary Stability Does Not Guarantee Learning Accessibility: A Multi-Agent Reinforcement Learning Perspective on Cooperation Emergence

Cooperation emergence is a central problem in multi-agent systems because decentralized agents must coordinate while adapting to the changing behavior of others. Evolutionary game theory identifies strategically stable outcomes, but stability under a population adjustment dynamic need not imply that finite-sample learning agents can reach the same outcome through local reward feedback. We study this distinction in a transparent three-agent governance-motivated game involving a government, a platform firm, and users. We derive replicator dynamics for the fixed stage-game incentives, evaluate the cooperative evolutionary basin on a symmetric initial-condition grid, and compare it with learning-basin estimates for three decentralized value-based learners. The learning analysis uses independent Q-learning with $\varepsilon$-greedy action selection, scaled Boltzmann exploration, and SA--EA BQL under the same payoff environment and outcome criterion. The evolutionary basin has volume $V_E=1.00$ on the sampled grid. The empirical learning basin is $0.88$ for $\varepsilon$-IQL and $0.00$ for both scaled Boltzmann and SA--EA BQL. Diagnostic traces show that broader action diversity and nonzero value separation can coexist with failure to sustain the cooperative joint action in this fixed configuration. These results indicate that evolutionary stability and learning accessibility are distinct properties of a coupled game--learning system. The shared-bike setting is a motivating application; the broader contribution is a framework for comparing population-level stability with the finite-sample accessibility of cooperation under specified multi-agent learning dynamics.

cs.AI↗
arXiv2026

Data-Driven Contextual Optimization with Gaussian Mixtures: Flow-Based Generalization, Robust Models, and Multistage Extensions

Contextual optimization enhances decision quality by leveraging side information to improve predictions of uncertain parameters. However, existing approaches face significant challenges when dealing with multimodal or mixtures of distributions. The inherent complexity of such structures often precludes an explicit functional relationship between the contextual information and the uncertain parameters, limiting the direct applicability of parametric models. Conversely, while non-parametric models offer greater representational flexibility, they are plagued by the "curse of dimensionality," leading to unsatisfactory performance in high-dimensional problems. To address these challenges, this paper proposes a novel contextual optimization framework based on Gaussian Mixture Models (GMMs). This model naturally bridges the gap between parametric and non-parametric approaches, inheriting the favorable sample complexity of parametric models while retaining the expressiveness of non-parametric schemes. By employing normalizing flows, we further relax the GM assumption and extend our framework to arbitrary distributions. Finally, inspired by the structural properties of GMMs, we design a novel GMM-based solution scheme for multistage stochastic optimization problems with Markovian uncertainty. This method exhibits significantly better sample complexity compared to traditional approaches, offering a powerful methodology for solving long-horizon, high-dimensional multistage problems. We demonstrate the effectiveness of our framework through extensive numerical experiments on a series of operations management problems. The results show that our proposed approach consistently outperforms state-of-the-art methods, underscoring its practical value for complex decision-making problems under uncertainty.

math.OC↗
arXiv2026

Direct Bayesian Inference of Helicity Amplitudes from Detector-Level Scattering Data

Helicity amplitudes give the most complete description of a variety of scattering reactions used in studies of strong interactions, but they cannot be measured directly: experiments record bilinear combinations of them folded through a detector response, and conventional analyses recover them through multiple stages that introduce discrete ambiguities and require a separate extraction of the absolute cross sections. Here we replace that chain with a single Bayesian inference that determines the experimentally identifiable amplitude parameters directly from detector-level measurements. A score-based diffusion model, trained on a forward simulator, provides the full posterior in each kinematic bin, with the detector response carried by the forward model and positivity of the spin-density matrix guaranteed by the parameterization. The observable amplitude content in electroproduction of final particles grows with polarization of beams and targets, providing additional sensitivity to underlying phases. In simulation, the posterior achieves empirical coverage at or above the nominal level and delivers the angular observables and the separated contributions from longitudinal and transverse photons with their correlations retained. It transfers without retraining to a realistic detector response absent from training and remains reliable in the weakly constrained nucleon-helicity-flip sector, where per-bin likelihood maximization degrades. The result is a general framework for a broad class of inverse problems, phase retrieval, quantum-state tomography, and partial-wave analysis are further instances. Its core requires only a forward simulation of the complete measurement process, instrumental effects are handled within one statistically consistent posterior, applicable across exclusive vector-meson programs at Jefferson Lab, COMPASS, HERMES, and the future Electron-Ion Collider.

nucl-ex↗
arXiv2026

Agent-MD: Selective LLM Intervention with Event-Driven Escalation for Stateful GCMC--MD Campaigns

Long-running molecular simulation campaigns require repeated continuation from saved states, provenance-aware progression, adaptive assessment, and occasional interpretation of workflow conditions that cannot be resolved safely by fixed rules. Here, we present Agent-MD, a framework that places large language model (LLM) reasoning selectively at campaign construction and event-triggered review, while routine simulation, analysis, continuation, archiving, and state progression are handled by a persistent rule-based campaign agent using approved policies and explicit state records. Agent-MD was demonstrated in a grand canonical Monte Carlo-molecular dynamics (GCMC-MD) water-vapor desorption campaign comprising five montmorillonite systems and three sequential relative-humidity states (RH = 0.9-0.3-0.1). Across 15 system-RH states, the workflow completed 120 segmented simulation cycles with state-specific sampling lengths and provenance-aware restart inheritance. Routine production required no live reasoning-agent invocation, while one state reached a review boundary; two preserved incidents were subsequently evaluated through blinded reasoning-agent replay, which identified the underlying workflow problems and recommended appropriate follow-up actions. The simulations also revealed distinct composition-dependent low-RH responses, with Ca-bearing montmorillonite retaining more interlayer water and maintaining a larger basal spacing than the Na- and K-bearing systems, while the highest-charge Na system retained more residual water under dry conditions. These results demonstrate that long-running scientific workflows need not place every operation inside an LLM reasoning loop: selective reasoning can instead be combined with deterministic execution, structured evidence, and validated control handoffs to provide reproducible and auditable agent-assisted molecular simulation.

cs.AI↗
arXiv2026

Unveiling the Role of Friction in Coarse-Grained Clay: A Hybrid Framework Integrating Long-Range Interactions and Granular Contact Mechanics

Given the predominant role of inter-particle physicochemical forces in governing clay behavior, researchers have increasingly utilized coarse-grained molecular dynamics (CGMD) simulations. However, inter-particle friction has been historically overlooked due to methodological limitations, and the extent to which this omission influences simulation accuracy remains an unresolved question. This study proposes a novel hybrid CGMD framework explicitly coupling long-range Buckingham potential with Hertzian granular contact mechanics. A baseline model was validated via isotropic compression, where the resulting compressibility and derived compression index (Cc) aligned with macroscopic geotechnical observations. Parametric analyses revealed that viscoelastic damping of particle contacts governs structural evolution. Elevated damping suppresses densification, trapping platelets in disorganized, high-void-ratio configurations. Furthermore, evaluating the interplay with thermal fluctuations underscores the necessity of precise temperature control to prevent such unphysical kinetic trapping. Finally, uniaxial compression tests demonstrate the critical importance of inter-particle friction. Explicit friction locks sliding interfaces and sustains significantly higher loads compared to frictionless models; the latter rely solely on geometric interlocking and ultimately exhibit unphysical fluid-like yielding. By bridging atomistic potentials with contact mechanics, this framework highlights the fundamental role of the inter-particle friction and offers essential guidelines for future multi-scale simulations of clay assemblies.

cond-mat.mtrl-sci↗
arXiv2026

The Continuous Latent Ornstein-Uhlenbeck Dynamics Framework: A Scalable Latent Process Model for Multivariate Longitudinal Categorical Data

Longitudinal biomedical studies increasingly collect irregularly sampled, multivariate categorical data that imperfectly reflect disease progression. This presents three key analytical challenges: highly heterogeneous disease progression across subjects; the presence of unobserved, co-evolving latent variables driving multiple measurements; and highly irregular sampling intervals both within and across patients. To address those challenges, we present the Continuous Latent Ornstein-Uhlenbeck Dynamics (CLOUD) framework for modeling complex disease trajectories from multivariate longitudinal categorical data. CLOUD links multivariate categorical observations to underlying latent functional domains and characterizes their coupled temporal evolution via the integration of a measurement component adjusted from item response theory (IRT) with a dynamic component based on multivariate Ornstein-Uhlenbeck (OU) processes. Methodologically, we introduce a time-inhomogeneous OU process that incorporated covariate-dependent components into the shifting mean function of the latent dynamics, allowing baseline biomarkers to modulate individual-level disease trajectories. We further propose a scalable parameterization of the OU drift matrix that enabled valid interaction modeling without restricting the number of latent functional domains. We establish theoretical properties of the proposed framework, including the analytical tractability of the model, the generality of the drift matrix reparameterization, and the identifiability of the entire model. Through simulation studies and an application to longitudinal amyotrophic lateral sclerosis (ALS) clinical data, we demonstrate that CLOUD provided a principled and flexible tool for characterizing subject-specific disease evolution across multiple interacting functional domains.

stat.ME↗
arXiv2026

Smooth Learning with Hard Constraints via Legendre-Regularized Policies

We revisit contextual optimization from the perspective of policy class design. A desirable policy class should be expressive enough to learn rich context-decision relationships, should enforce hard feasibility constraints rather than soft penalty terms, and should remain smooth enough for gradient-based training on downstream decision losses. Existing approaches usually emphasize only part of these requirements. We propose Legendre-regularized policies, which parameterize decisions as solutions of regularized optimization problems over the original feasible region. This construction yields policies that are feasible by construction and differentiable with respect to learned latent parameters. We prove that the associated optimizer map is single-valued, maps onto the relative interior of the feasible set, admits an explicit Jacobian, is Lipschitz continuous, and can be made arbitrarily smooth. We also establish a universal approximation result showing that the proposed class can approximate any continuous feasible policy on compact context sets. The framework unifies explicitly regularized optimizers and implicit perturbation-based smooth optimizers. Experiments on contextual newsvendor and resource allocation problems show that our approach improves prescriptive performance relative to the benchmark methods.

math.OC↗
arXiv2026

Decoupling Constraints from Two Directions for Evolutionary Constrained Multi-objective Optimization

Real-world constrained multi-objective optimization problems (CMOPs) commonly involve multiple constraints, and understanding and exploiting their coupling relationships is crucial for efficient optimization. Recent constraint-decoupling methods handle individual constraints separately, but they generally search only in the evolutionary direction to approximate single-constraint Pareto fronts (SCPFs). In this study, we show that part or all of the constrained Pareto front (CPF) may be unrelated to any SCPF and instead be shaped by the boundaries of infeasible regions. We refer to such a portion as the independent CPF (ICPF) and introduce the reverse CPF (RCPF) to characterize its associated informative infeasible boundaries. Based on these observations, we propose a bidirectional constraint-decoupling coevolutionary algorithm named DCF2D. DCF2D dynamically identifies the constraints obstructing the main population and activates constraint-specific auxiliary populations. These populations adaptively search in the evolutionary direction for the corresponding SCPFs or in the reverse evolutionary direction for the corresponding RCPFs. Its three-stage framework integrates unconstrained global exploration, event-driven bidirectional coevolution, and final convergence refinement. Experiments on 87 benchmark instances from seven test suites and 28 real-world engineering CMOPs demonstrate that DCF2D achieves the best overall performance among nine algorithms. Code available at: https://github.com/RuiqingS/DCF2D.

cs.NE↗
arXiv2026

Distributionally Robust Optimization via Targeted Integral Probability Metrics for General Data Processes

Distributionally robust optimization (DRO) provides a principled framework for decision-making under distributional uncertainty. Classical data-driven DRO frameworks typically construct ambiguity sets from distributional information, such as moment constraints, divergence neighborhoods, or Wasserstein balls, specified before the downstream loss is considered. We propose a task-aware DRO framework based on targeted integral probability metrics. The ambiguity set is defined directly through the loss functions induced by feasible decisions, thereby controlling the loss discrepancy between an adversarial distribution and a data-driven reference distribution. This construction leads to an expected hinge-constrained formulation that is equivalent to an infinitely constrained loss-discrepancy formulation. It also yields finite-sample guarantees that bypass the ambient curse of dimensionality: whenever an appropriate scalar pointwise concentration inequality is available for the induced loss estimator, the ambiguity radius can be calibrated at the canonical $\widetilde{\mathcal O}(N^{-1/2})$ rate after uniformization over the decision class. As a result, the framework applies broadly to settings including heavier-tailed sub-Weibull losses, Markovian data, outlier-corrupted data, and incomplete data. We derive exact infinite-dimensional dual reformulations, establish out-of-sample and excess-risk guarantees, and develop a conservative Monte Carlo approximation scheme with convergence and suboptimality guarantees. For piecewise affine losses, the sampled problems admit tractable conic reformulations. Numerical experiments in inventory management under heavy-tailed demand and regression with outlier corruption demonstrate strong out-of-sample performance relative to existing approaches.

math.OC↗
arXiv2026

Magnon-Mediated Superconductivity in a 2D Itinerant Ferromagnet with Weak Easy-plane Magnetic Anisotropy

Motivated by recent observations of superconductivity in a quarter-metal state of spin- and valley- polarized graphene multilayers, we investigate pairing within a ferromagnetic phase of a single-valley model of itinerant two-dimensional (2D) electrons with Hubbard-type interaction and no artificial high-energy cutoff. In 2D, the Stoner transition is first-order into a fully-polarized state wherein the only gapless collective excitations are transverse magnons. We find that in a spin-SU(2) symmetric model, this magnon-mediated pairing interaction between equal-spin fermions vanishes at $T=0$. We show that a small easy-plane magnetic anisotropy $Ω_0 \ll E_F$, where $E_F$ is the Fermi energy, breaks the SU(2) symmetry and generates an attractive interaction for equal-spin $p-$wave pairing. We explicitly derive the corresponding coupling constant $λ_p$ as the scaling function of both the relative strength of the easy-plane anisotropy, $Ω_0/E_F$, and the proximity to the ferromagnetic transition. While $λ_p$ is parametrically small in $Ω_0/E_F$ deep inside the ferromagnetic phase, it becomes enhanced near the ferromagnetic transition, reaching order unity regardless of how small $Ω_0/E_F$ is. This mechanism yields a sizable $T_c$, peaked near the onset of ferromagnetism.

cond-mat.supr-con↗
arXiv2026

ParetoPilot: Zero-Surrogate Offline Multi-Objective Optimization via Infer-Perturb-Guide Diffusion

Offline multi-objective optimization (Offline MOO) seeks Pareto-optimal designs from static datasets without additional environment interactions. Existing generative methods typically guide sampling with external surrogate or preference models, which adds training complexity and may provide unreliable guidance. We propose ParetoPilot, a plug-and-play method that guides designs to Pareto front at inference time using a pre-trained conditional diffusion model without any surrogate. ParetoPilot introduces an Infer-Perturb-Guide (IPG) engine within the reverse diffusion process. IPG first infers the individual conditional target for each sample in the batch by aligning its conditional and unconditional predictions. It then perturbs these targets collectively across the batch, balancing convergence toward the Pareto front and diversity among samples. Finally, the engine guides the generative trajectory toward the Pareto front by injecting these perturbed targets via standard Classifier-Free Guidance (CFG). Experiments on 51 tasks demonstrate that ParetoPilot achieves the best overall ranking among 16 methods and competitive hypervolume improvement.

cs.LG↗
arXiv2026

Calculation of Particle Pair Correlation Functions with Classical Trajectory Approximation

Femtoscopic interferometry is a powerful tool for probing the spatio-temporal evolution of emission sources in heavy-ion collisions. A major challenge in the field is formulating a self-consistent description of the source function, final-state interactions between the particle pair, and interactions inherent to the source itself. To address this, we have developed a novel Monte Carlo model for calculating two-particle correlation functions in a classical trajectory approximation (CTA-I). The model incorporates self-consistently the emission source of thermal equilibrium and three-body final state interactions. Application of the model shows satisfactory fit to experimental data, revealing that the correlation function is highly sensitive to the source's spatio-temporal extent. In contrast, the temperature parameter governing the emitted particles' energy spectra has a negligible influence. Our approach offers the potential to extract the spatio-temporal information from the emission source, thereby advancing the applicability of femtoscopic interferometry in the Fermi energy domain.

nucl-th↗
arXiv2026

Structure-Aware Compound-Protein Affinity Prediction via Graph Neural Networks with Group Lasso Regularization

Explainable artificial intelligence approaches accelerate drug discovery by improving molecular representation learning, identifying key molecular structures, and rationalizing drug property prediction. However, developing end-to-end explainable models for target-specific structure-activity relationship modeling remains challenging because compound-protein interaction data are often limited for individual targets, and small changes in chemical substituents or local structural motifs can cause large differences in molecular properties. Therefore, effectively leveraging structural and property information to identify key moieties associated with compound-protein affinity is essential. We propose a graph neural network (GNN) framework that uses property and structural information from activity-cliff molecule pairs targeting specific proteins to predict compound-protein affinity, measured by half-maximal inhibitory concentration (IC50), and explain property differences. To improve explainability, we trained GNNs with structure-aware loss functions using group lasso and sparse group lasso regularization, which prune and highlight molecular subgraphs relevant to activity differences. We applied this framework to activity-cliff data from molecules targeting six tyrosine-protein kinases across the Src, Abl, and Tec families, as well as anaplastic lymphoma kinase. Integrating common- and uncommon-node information with sparse group lasso improved target-specific molecular property prediction, producing lower root mean square errors and higher Pearson correlation coefficients. Regularization also enhanced GNN feature attribution by improving graph-level global direction scores and atom-level coloring accuracy. These results support more interpretable drug discovery pipelines, particularly for identifying critical molecular substructures during lead optimization.

cs.LG↗
arXiv2026

Potential-Guided Flow Matching for Vision-Language-Action Policy Improvement

Large vision-language-action (VLA) policies are increasingly trained as conditional generative models over action chunks. Yet deployment produces mixed-quality experience-successful demonstrations, partial completions, recoverable mistakes, and failures-that is difficult to use with standard imitation. Full behavior cloning (BC) imitates failures, filtered BC discards useful sub-trajectories, and offline reinforcement learning adds a large critic. We introduce ForesightFlow, a self-guided flow-matching policy that augments each generated action chunk with a learned success-potential trajectory. The same flow proposes and scores candidate actions, enabling best-of-$K$ inference without an external critic. The key issue is that policy improvement and value calibration require different supervision: advantage weighting should emphasize high-quality actions, but applying the same weights to potential coordinates suppresses failure gradients and creates overconfident scores. We address this with decoupled advantage-weighted flow matching, applying exponentiated advantage weights only to action velocities while training potential velocities uniformly. We further derive a one-step boundary estimator for conditional flow matching, allowing advantage computation with a single stop-gradient forward pass. Across five BEHAVIOR-1K simulation tasks and five real-world bimanual tasks, ForesightFlow improves over imitation baselines, matches the strongest separate-critic baseline in simulation success, improves real-world success, and reduces training compute by $38\%$. Ablations show that decoupling prevents value hallucination, the one-step estimator preserves candidate-ranking fidelity, and self-guided sampling improves long-horizon execution.

cs.RO↗
arXiv2026

IMPACT: Influence Modeling for Open-Set Time Series Anomaly Detection

Open-set anomaly detection (OSAD) is an emerging paradigm designed to utilize limited labeled data from anomaly classes seen in training to identify both seen and unseen anomalies during testing. Current approaches rely on simple augmentation methods to generate pseudo anomalies that replicate unseen anomalies. Despite being promising in image data, these methods are found to be ineffective in time series data due to the failure to preserve its sequential nature, resulting in trivial or unrealistic anomaly patterns. They are further plagued when the training data is contaminated with unlabeled anomalies. This work introduces $\textbf{IMPACT}$, a novel framework that leverages $\underline{\textbf{i}}$nfluence $\underline{\textbf{m}}$odeling for o$\underline{\textbf{p}}$en-set time series $\underline{\textbf{a}}$nomaly dete$\underline{\textbf{ct}}$ion, to tackle these challenges. The key insight is to $\textbf{i)}$ learn an influence function that can accurately estimate the impact of individual training samples on the modeling, and then $\textbf{ii)}$ leverage these influence scores to generate semantically divergent yet realistic unseen anomalies for time series while repurposing high-influential samples as supervised anomalies for anomaly decontamination. Extensive experiments show that IMPACT significantly outperforms existing state-of-the-art methods, showing superior accuracy under varying OSAD settings and contamination rates. Code is available at https://github.com/mala-lab/IMPACT.

cs.LG↗
arXiv2026

CA3D: Computing Accessibility-Aware Cooperative 3D Deployment of Multiple UAVs

This letter investigates computing-accessibility-aware cooperative 3D deployment of multiple UAVs for task completion enhancement, termed CA3D. We first provide a theoretical analysis showing that computing accessibility is the key mechanism linking UAV deployment to delay-constrained task completion, and that UAV inter-spacing creates a fundamental tradeoff between computing-resource accessibility and task completion. We then develop a cooperative 3D deployment design that jointly balances accessible computing capacity, task completion probability, and redundant UAV overlap. Simulation results under heterogeneous computing node capacities show that CA3D consistently outperforms Random, Fixed, and Greedy deployment baselines under both hotspot and random ground user (GU) distributions. Under the hotspot GU distribution, CA3D achieves nearly full task completion, improving the task completion probability by about 3.3x over Random deployment when the number of UAVs is 8. Under a more challenging random GU distribution, CA3D still achieves about 35% higher task completion probability than the best baseline when the number of UAVs is 12. These results demonstrate that computing-accessibility-aware cooperative 3D deployment improves not only task completion but also robustness to GU distribution changes.

cs.NI↗
arXiv2026

CiteVQA: Benchmarking Evidence Attribution for Trustworthy Document Intelligence

Multimodal Large Language Models (MLLMs) have significantly advanced document understanding, yet current Doc-VQA evaluations score only the final answer and leave the supporting evidence unchecked. This answer-only approach masks a critical failure mode: a model can land on the correct answer while grounding it in the wrong passage -- a critical risk in high-stakes domains like law, finance, and medicine, where every conclusion must be traceable to a specific source region. To address this, we introduce CiteVQA, a benchmark that requires models to return element-level bounding-box citations alongside each answer, evaluating both jointly. CiteVQA comprises 1,897 questions across 711 PDFs spanning seven domains and two languages, averaging 40.6 pages per document. To ensure fidelity and scalability, the ground-truth citations are generated by an automated pipeline-which identifies crucial evidence via masking ablation-and are subsequently validated through expert review. At the core of our evaluation is Strict Attributed Accuracy (SAA), which credits a prediction only when the answer and the cited region are both correct. Auditing 20 MLLMs reveals a pervasive Attribution Hallucination: models frequently produce the right answer while citing the wrong region. The strongest system (Gemini-3.1-Pro-Preview) achieves an SAA of only 76.0, and the strongest open-source MLLM reaches just 22.5. Ultimately, towards trustworthy document intelligence, CiteVQA exposes a reliability gap that answer-only evaluations overlook, providing the instrumentation needed to close it. Our repository is available at https://github.com/opendatalab/CiteVQA.

cs.CL↗