arXiv subjects

Yanzhi Zhang

Publications and source records attributed to Yanzhi Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Modeling Unknown Nonlocal PDE Systems via Flow Map Learning

Nonlocal partial differential equations arise in many applications but are often difficult to model and learn because of the presence of nonlocal operators. We present a flow-map learning (FML) framework for modeling unknown nonlocal PDEs directly from solution data. Rather than learning or approximating the underlying nonlocal operators, the proposed approach learns the finite-time evolution operator in either modal or nodal space. Two complementary formulations are developed for spectral and grid-based solution representations. Numerical experiments on one- and two-dimensional fractional diffusion and wave equations demonstrate accurate and stable long-time prediction using only short observation windows. The proposed approach provides an effective data-driven framework for learning unknown nonlocal dynamics without explicit evaluation of nonlocal operators.

cs.LG↗

arXiv2026

M3: A State-Event Generative Foundation Model for Market Microstructure Dynamics

Market microstructure simulation aims to model how liquidity, prices, and order flow evolve in electronic financial markets. Since market data reveal only one realized trajectory, many important questions are inherently counterfactual and require realistic trajectory-level simulation. Existing financial generative models, however, often model order events and market states, such as the LOB, in isolation, overlooking the dynamic interaction between order flow and liquidity in market microstructure. We propose the \textbf{M3} (\underline{M}arket \underline{M}icrostructure \underline{M}odel), a state-event generative foundation model for market microstructure dynamics. \textbf{M3} learns to generate future order-flow trajectories, while accounting for the evolving interaction between order events and limit-order-book liquidity. Trained on large-scale order-level real stock market data, \textbf{M3} exhibits predictable scaling behavior, reproduces key market stylized facts, and enables practical simulation-based applications including forecasting, stress testing, and market-impact analysis. These results suggest a scalable foundation-model paradigm for counterfactual market simulation at the microstructure level.

q-fin.CP↗

arXiv2026

One Tool Is Enough: Reinforcement Learning for Repository-Level LLM Agents

Locating files and functions requiring modification in large software repositories is challenging due to their scale and structural complexity. Existing LLM-based methods typically treat this as a repository-level retrieval task and rely on multiple auxiliary tools, which often overlook code execution logic and complicate model control. We propose RepoNavigator, an LLM agent equipped with a single execution-aware tool: jumping to the definition of an invoked symbol. This unified design reflects the actual flow of code execution while simplifying tool manipulation. RepoNavigator is trained end-to-end via Reinforcement Learning (RL) directly from a base pretrained model, without relying on closed-source distillation. Experiments demonstrate that RL-trained RepoNavigator achieves state-of-the-art performance, with the 7B model outperforming 14B baselines, the 14B model surpassing 32B competitors, and the 32B model exceeding closed-source models such as GPT-5 on most metrics. These results confirm that integrating a single, structurally grounded tool with RL training provides an efficient and scalable solution for repository-level issue localization.

cs.SE↗

arXiv2026

FutureWorld: A Live Reinforcement Learning Environment for Predictive Agents with Real-World Outcome Rewards

Live future prediction refers to the task of making predictions about real-world events before they unfold. This task is increasingly studied using large language model-based agent systems, and it is important for building agents that can continually learn from the real world. It can provide a large number of prediction questions grounded in diverse real-world events, while preventing answer leakage. To leverage the advantages of future prediction, we present FutureWorld, a live agentic reinforcement learning environment that closes the training loop between prediction, outcome realization, and parameter updates. Specifically, we modify and extend verl-tool, resulting in a new framework that we call verl-tool-future. Unlike standard reinforcement learning training frameworks that rely on immediate rewards, verl-tool-future stores prediction-time rollouts, backfills rewards after real-world outcomes become available, and then replays the completed trajectories for policy update. Across three open-source agents, successive FutureWorld training rounds lead to consistent improvements in prediction accuracy, probabilistic scoring, and calibration, demonstrating that delayed real-world outcome feedback can serve as an effective reinforcement learning signal.

cs.AI↗

arXiv2026

Harnessing Pre-Resolution Signals for Future Prediction Agents

Many high-stakes decisions depend on forecasts made before outcomes are known. In this future prediction setting, the central challenge is that public evidence evolves over time, while the main supervision signal arrives only after resolution: the realized outcome mainly assesses final correctness, offering only coarse guidance on what to track, what to verify, and which judgments to leave uncertain along the way. Our key observation is that revisiting the same unresolved question over time creates informative temporal contrasts across evolving evidence and repeated forecasts, exposing what earlier attempts missed before resolution and yielding a diagnostic signal we call the pre-resolution signal. We instantiate this idea in Milkyway, a future prediction agent with a persistent future prediction harness, an editable external state that stores reusable procedural guidance across revisits to the same unresolved question. As the same unresolved question is revisited, Milkyway extracts pre-resolution signals from evolving evidence and repeated forecasts, uses them to update the harness, and improves later forecasts on that question before resolution. After resolution, the realized outcome serves as a post-resolution check of provisional updates. On the FutureX and FutureWorld benchmarks, Milkyway achieves strong performance against competitive baselines, and a mechanism study suggests that the gains stem from harness evolution driven by pre-resolution signals rather than repeated prediction alone.

cs.AI↗

arXiv2026

Targeted Exploration via Unified Entropy Control for Reinforcement Learning

Recent advances in reinforcement learning (RL) have improved the reasoning capabilities of large language models (LLMs) and vision-language models (VLMs). However, the widely used Group Relative Policy Optimization (GRPO) consistently suffers from entropy collapse, causing the policy to converge prematurely and lose diversity. Existing exploration methods introduce additional bias or variance during exploration, making it difficult to maintain optimization stability. We propose Unified Entropy Control for Reinforcement Learning (UEC-RL), a framework that provides targeted mechanisms for exploration and stabilization. UEC-RL activates more exploration on difficult prompts to search for potential and valuable reasoning trajectories. In parallel, a stabilizer prevents entropy from growing uncontrollably, thereby keeping training stable as the model consolidates reliable behaviors. Together, these components expand the search space when needed while maintaining robust optimization throughout training. Experiments on both LLM and VLM reasoning tasks show consistent gains over RL baselines on both Pass@1 and Pass@$k$. On Geometry3K, UEC-RL achieves a 37.9\% relative improvement over GRPO, indicating that it sustains effective exploration without compromising convergence and underscoring UEC-RL as a key for scaling RL-based reasoning in large models. Our code is available at https://github.com/597358816/UEC-RL.

cs.AI↗

arXiv2026

Can a Lightweight Automated AI Pipeline Solve Research-Level Mathematical Problems?

Large language models (LLMs) have recently achieved remarkable success in generating rigorous mathematical proofs, with "AI for Math" emerging as a vibrant field of research (Ju et al., 2026). While these models have mastered competition-level benchmarks like the International Mathematical Olympiad (Huang et al., 2025; Duan et al., 2025) and show promise in research applications through auto-formalization (Wang et al., 2025), their deployment via lightweight, natural-language pipelines for research problems remains underexplored. In this work, we demonstrate that next-generation models (e.g., Gemini 3 Pro, GPT-5.2 Pro), when integrated into a streamlined automated pipeline optimized for citation-based verification, can solve sophisticated research-grade problems. We evaluate our pipeline on two novel datasets: (1) the ICCM (2025) problem sets (comparable to the S.-T. Yau College Student Mathematics Contest) proposed by leading mathematicians (Shanghai Math Challenge, 2026), and (2) the "First Proof" problem set (Abouzaid et al., 2026), consisting of previously unpublished research questions. Our pipeline generated candidate proofs for all problems in the first two ICCM sets and the "First Proof" set. The solutions for the first two ICCM sets and Problem 4 of the "First Proof" set have been fully verified by our team. All generated proofs have been submitted to the official organization, and our generated results are publicly available at https://github.com/ml1301215/question_sets-test_results. We have open-sourced the code and developed a user-friendly UI for this workflow, accessible at https://github.com/ml1301215/research-math-assistant.

cs.AI↗

arXiv2025

Population-Evolve: a Parallel Sampling and Evolutionary Method for LLM Math Reasoning

Test-time scaling has emerged as a promising direction for enhancing the reasoning capabilities of Large Language Models in last few years. In this work, we propose Population-Evolve, a training-free method inspired by Genetic Algorithms to optimize LLM reasoning. Our approach maintains a dynamic population of candidate solutions for each problem via parallel reasoning. By incorporating an evolve prompt, the LLM self-evolves its population in all iterations. Upon convergence, the final answer is derived via majority voting. Furthermore, we establish a unification framework that interprets existing test-time scaling strategies through the lens of genetic algorithms. Empirical results demonstrate that Population-Evolve achieves superior accuracy with low performance variance and computational efficiency. Our findings highlight the potential of evolutionary strategies to unlock the reasoning power of LLMs during inference.

cs.AI↗

arXiv2025

Mathematical and numerical studies on ground states of trapped unitary Fermi gases

We mathematically and numerically study the ground states of unitary Fermi gases. Starting from the three-dimensional nonlinear Schrödinger equation that contains a quantum pressure term and an angular momentum rotation term, we first nondimensionalize the equation and then obtain its one-dimensional and two-dimensional counterparts in some limit regimes of the external potentials. Existence and uniqueness of the ground states of the unitary Fermi gases are studied with/without the angular momentum rotation term. We present a regularized normalized gradient flow method to compute the ground states of trapped unitary Fermi gases. Our numerical results show that the quantum pressure term has a significant effect on the ground state properties. Specifically, with the presence of the quantum pressure term, the vortex lattices are very different from those obtained in conventional Bose-Einstein condensation.

math-ph↗

arXiv2025

EFRame: Deeper Reasoning via Exploration-Filter-Replay Reinforcement Learning Framework

Recent advances in reinforcement learning (RL) have significantly enhanced the reasoning capabilities of large language models (LLMs). Group Relative Policy Optimization (GRPO), a lightweight variant of Proximal Policy Optimization (PPO), improves efficiency but suffers from limited exploration and training instability, limiting its effectiveness on complex reasoning tasks. To address these challenges, we introduce EFRame, an Exploration-Filter-Replay framework that augments GRPO across three dimensions: additional rollouts enable deeper and more targeted exploration, online filtering removes low-quality samples to stabilize gradients and accelerate training, and experience replay amplifies rare yet informative trajectories for stable convergence. This unified framework establishes a principled training cycle that balances exploration, efficiency, and stability. Experiments on diverse reasoning benchmarks demonstrate that EFRame achieves consistent gains, including a 37.9\% relative improvement on Geometry3K over GRPO. EFRame further supports fine-grained sample categorization and precise entropy control, highlighting it as a robust solution for advancing deeper reasoning in LLMs. Our code is available at https://github.com/597358816/EFRame.

cs.LG↗

arXiv2025

Nonlocal modeling of spatial fractional diffusion with truncated interaction domains and truncated kernel function singularity

Parabolic partial differential equations (PDEs) are in ubiquitous, very effective use to model diffusion processes. However, there are many applications (e.g., such as in hydrology, animal foraging, biology, and light diffusion just do name a few) for which results obtained through the use of parabolic PDEs do not agree with observations. In many situations the use of fractional diffusion models has been found to be more faithful to that which is observed. Specifically, we replace the Laplacian operator in the PDE by a fractional Laplacian operator ${\mathcal L}$ which is an integral operator for which solutions are sought for on all of space, has an unbounded domain of integration, and for a given point $x$ the integrand contains a kernel function $ϕ(y-x)$ that is infinite whenever $y=x$. These three features pose impediments not only for the construction of efficient discretization methods but also because all three involve one or more sort of "infinity''. To overcome these impediments we choose to invoke one or more of the following strategies. (a) We seek solutions only within a chosen bounded domain $Ω$. (b) For every $x\inΩ$, we choose a bounded domain of integration such as, e.g., an Euclidean ball $B_δ(x)$ having finite radius $δ$. (c) We truncate the singularity of $ϕ(y-x)$ by setting, for a given constant $\varepsilon>0$, $ϕ(y-x)= ϕ(\varepsilon)$ whenever $|y-x|\le\varepsilon$. We then provide extensive illustrations of the possible combinations chosen from among (a), (b), and (c). We also illustrate, for the various models defined for each of these combinations, their limiting behavior of solutions such as showing that as $δ\to0$ we recover the PDE model and also showing that in the limit of some other parameters we recover the fractional Laplacian model.

math.NA↗

arXiv2025

Discretization-independent multifidelity operator learning for partial differential equations

We develop a new and general encode-approximate-reconstruct operator learning model that leverages learned neural representations of bases for input and output function distributions. We introduce the concepts of \textit{numerical operator learning} and \textit{discretization independence}, which clarify the relationship between theoretical formulations and practical realizations of operator learning models. Our model is discretization-independent, making it particularly effective for multifidelity learning. We establish theoretical approximation guarantees, demonstrating uniform universal approximation under strong assumptions on the input functions and statistical approximation under weaker conditions. To our knowledge, this is the first comprehensive study that investigates how discretization independence enables robust and efficient multifidelity operator learning. We validate our method through extensive numerical experiments involving both local and nonlocal PDEs, including time-independent and time-dependent problems. The results show that multifidelity training significantly improves accuracy and computational efficiency. Moreover, multifidelity training further enhances empirical discretization independence.

cs.LG↗

arXiv2025

No Free Lunch: Rethinking Internal Feedback for LLM Reasoning

Reinforcement learning has emerged as a powerful paradigm for post-training large language models (LLMs) to improve reasoning. Approaches like Reinforcement Learning from Human Feedback (RLHF) and Reinforcement Learning with Verifiable Rewards (RLVR) have shown strong results, but they require extensive external supervision. We investigate an alternative class of methods, Reinforcement Learning from Internal Feedback (RLIF), which relies solely on intrinsic model-derived signals instead of external rewards. In particular, we leverage unsupervised reward proxies such as token-level entropy, trajectory-level entropy, and self-certainty. Our theoretical analysis shows these internal objectives are partially equivalent, and we empirically evaluate various RLIF strategies on challenging math reasoning benchmarks. Experimental results demonstrate that RLIF can boost the reasoning performance of base LLMs at the beginning phase of the training, matching or surpassing RLVR techniques on these tasks. However, when training progresses, performance degrades even below the model before training. Moreover, we find that RLIF yields little improvement for instruction-tuned models, indicating diminishing returns of intrinsic feedback once an LLM is already instruction-tuned. We further analyze this limitation by mixing model weights and explain the reason of RLIF's training behaviors, providing practical guidelines for integrating internal feedback signals into LLM training. We hope our analysis of internal feedback will inform more principled and effective strategies for LLM post-training.

cs.LG↗

arXiv2024

A novel and simple spectral method for nonlocal PDEs with the fractional Laplacian

We propose a novel and simple spectral method based on the semi-discrete Fourier transforms to discretize the fractional Laplacian $(-Δ)^\fracα{2}$. Numerical analysis and experiments are provided to study its performance. Our method has the same symbol $|\boldsymbolξ|^α$ as the fractional Laplacian $(-Δ)^\fracα{2}$ at the discrete level, and thus it can be viewed as the exact discrete analogue of the fractional Laplacian. This {\it unique feature} distinguishes our method from other existing methods for the fractional Laplacian. Note that our method is different from the Fourier pseudospectral methods in the literature which are usually limited to periodic boundary conditions (see Remark \ref{remark0}). Numerical analysis shows that our method can achieve a spectral accuracy. The stability and convergence of our method in solving the fractional Poisson equations were analyzed. Our scheme yields a multilevel Toeplitz stiffness matrix, and thus fast algorithms can be developed for efficient matrix-vector multiplications. The computational complexity is ${\mathcal O}(2N\log(2N))$, and the memory storage is ${\mathcal O}(N)$ with $N$ the total number of points. Extensive numerical experiments verify our analytical results and demonstrate the effectiveness of our method in solving various problems.

math.NA↗

arXiv2024

Fourier pseudospectral methods for the spatial variable-order fractional wave equations

In this paper, we propose Fourier pseudospectral methods to solve the variable-order space fractional wave equation and develop an accelerated matrix-free approach for its effective implementation. In constant-order cases, our methods can be efficiently implemented via the (inverse) fast Fourier transforms, and the computational cost at each time step is ${\mathcal O}(N\log N)$ with $N$ the total number of spatial points. However, this fast algorithm fails in the variable-order cases due to the spatial dependence of the Fourier multiplier. On the other hand, the direct matrix-vector multiplication approach becomes impractical due to excessive memory requirements. To address this challenge, we proposed an accelerated matrix-free approach for the efficient computation of variable-order cases. The computational cost is ${\mathcal O}(MN\log N)$ and storage cost ${\mathcal O}(MN)$, where $M \ll N$. Moreover, our method can be easily parallelized to further enhance its efficiency. Numerical studies show that our methods are effective in solving the variable-order space fractional wave equations, especially in high-dimensional cases. Wave propagation in heterogeneous media is studied in comparison to homogeneous counterparts. We find that wave dynamics in fractional cases become more intricate due to nonlocal interactions. Specifically, dynamics in heterogeneous media are more complex than those in homogeneous media.

math.NA↗

arXiv2024

Variable-order fractional Laplacian and its accurate and efficient computations with meshfree methods

The variable-order fractional Laplacian plays an important role in the study of heterogeneous systems. In this paper, we propose the first numerical methods for the variable-order Laplacian $(-Δ)^{α({\bf x})/2}$ with $0 < α({\bf x}) \le 2$, which will also be referred as the variable-order fractional Laplacian if $α({\bf x})$ is strictly less than 2. We present a class of hypergeometric functions whose variable-order Laplacian can be analytically expressed. Building on these analytical results, we design the meshfree methods based on globally supported radial basis functions (RBFs), including Gaussian, generalized inverse multiquadric, and Bessel-type RBFs, to approximate the variable-order Laplacian $(-Δ)^{α({\bf x})/2}$. Our meshfree methods integrate the advantages of both pseudo-differential and hypersingular integral forms of the variable-order fractional Laplacian, and thus avoid numerically approximating the hypersingular integral. Moreover, our methods are simple and flexible of domain geometry, and their computer implementation remains the same for any dimension $d \ge 1$. Compared to finite difference methods, our methods can achieve a desired accuracy with much fewer points. This fact makes our method much attractive for problems involving variable-order fractional Laplacian where the number of points required is a critical cost. We then apply our method to study solution behaviors of variable-order fractional PDEs arising in different fields, including transition of waves between classical and fractional media, and coexistence of anomalous and normal diffusion in both diffusion equation and the Allen-Cahn equation. These results would provide insights for further understanding and applications of variable-order fractional derivatives.

math.NA↗

arXiv2023

Gaussian radial basis functions collocation for fractional PDEs: methodology and error analysis

The paper introduces a new meshfree pseudospectral method based on Gaussian radial basis functions (RBFs) collocation to solve fractional Poisson equations. Hypergeometric functions are used to represent the fractional Laplacian of Gaussian RBFs, enabling an efficient computation of stiffness matrix entries. Unlike existing RBF-based methods, our approach ensures a Toeplitz structure in the stiffness matrix with equally spaced RBF centers, enabling efficient matrix-vector multiplications using fast Fourier transforms. We conduct a comprehensive study on the shape parameter selection, addressing challenges related to ill-conditioning and numerical stability. The main contribution of our work includes rigorous stability analysis and error estimates of the Gaussian RBF collocation method, representing a first attempt at the rigorous analysis of RBF-based methods for fractional PDEs to the best of our knowledge. We conduct numerical experiments to validate our analysis and provide practical insights for implementation.

math.NA↗

arXiv2022

The Development and Prospect of Code Clone

The application of code clone technology accelerates code search, improves code reuse efficiency, and assists in software quality assessment and code vulnerability detection. However, the application of code clones also introduces software quality issues and increases the cost of software maintenance. As an important research field in software engineering, code clone has been extensively explored and studied by researchers, and related studies on various sub-research fields have emerged, including code clone detection, code clone evolution, code clone analysis, etc. However, there lacks a comprehensive exploration of the entire field of code clone, as well as an analysis of the trend of each sub-research field. This paper collects related work of code clones in the past ten years. In summary, the contributions of this paper mainly include: (1) summarize and classify the sub-research fields of code clone, and explore the relative popularity and relation of these sub-research fields; (2) analyze the overall research trend of code clone and each sub-research field; (3) compare and analyze the difference between academy and industry regarding code clone research; (4) construct a network of researchers, and excavate the major contributors in code clone research field; (5) The list of popular conferences and journals was statistically analyzed. The popular research directions in the future include clone visualization, clone management, etc. For the clone detection technique, researchers can optimize the scalability and execution efficiency of the method, targeting particular clone detection tasks and contextual environments, or apply the technology to other related research fields continuously.

cs.SE↗