arXiv subjects

Xian Tang

Publications and source records attributed to Xian Tang.

5 recordsLinked to original sources

arXiv2026

Positron annihilation lifetime and Doppler broadening spectral calculations of oxygen-doped 3C-SiC

Based on density functional theory (DFT), the formation energies of intrinsic vacancy defects (VC, VSi, and VSi+C) and oxygen-related defects (OC, OSi, OCVSi, and OSiVC) in 3C-SiC are systematically investigated. The results indicate that all defects considered, except for OC, possess neutral or negative charge states, thereby making them suitable for detection by positron annihilation spectroscopy (PAS). Furthermore, the electron and positron density distributions and positron annihilation lifetimes for the perfect 3C-SiC supercell and various defective configurations are computed. It is found that the OSi and OSiVC complexes act as effective positron trapping centers, leading to the formation of positron trapped states and a notable increase in annihilation lifetimes at the corresponding defect sites. In addition, coincidence Doppler broadening (CDB) spectra, along with the S and W parameters, are calculated for both intrinsic and oxygen-doped point defects (OC, OSi, OCVSi, and OSiVC). The analysis reveals that electron screening effects dominate the annihilation characteristics of the OSi defect, whereas positron localization induced by the vacancy is the predominant contributor in the case of OSiVC. This distinction results in clearly different momentum distributions of these two oxygen-related defects for different charge states. Overall, the PAS is demonstrated to be a powerful technique for distinguishing intrinsic vacancy-type defects and oxygen-doped composites in 3C-SiC. Combining the analysis of electron and positron density distributions, the electron localization and positron trapping behavior in defect systems with different charge states can be comprehensively understood. These first-principles results provide a solid theoretical foundation for identifying and characterizing the defects in oxygen-doped 3C-SiC by using PAS.

cond-mat.mtrl-sci↗

arXiv2026

Reinforcement Learning with Promising Tokens for Large Language Models

Reinforcement learning (RL) has emerged as a key paradigm for aligning and optimizing large language models (LLMs). Standard approaches treat the LLM as the policy and apply RL directly over the full vocabulary space. However, this formulation includes the massive tail of contextually irrelevant tokens in the action space, which could distract the policy from focusing on decision-making among the truly reasonable tokens. In this work, we verify that valid reasoning paths could inherently concentrate within a low-rank subspace. Based on this insight, we introduce Reinforcement Learning with Promising Tokens (RLPT), a framework that mitigates the action space issue by decoupling strategic decision-making from token generation. Specifically, RLPT leverages the semantic priors of the base model to identify a dynamic set of promising tokens and constrains policy optimization exclusively to this refined subset via masking. Theoretical analysis and empirical results demonstrate that RLPT effectively reduces gradient variance, stabilizes the training process, and improves sample efficiency. Experiment results on math, coding, and telecom reasoning show that RLPT outperforms standard RL baselines and integrates effectively across various model sizes (4B and 8B) and RL algorithms (GRPO and DAPO).

cs.LG↗

arXiv2026

EDCO: Dynamic Curriculum Orchestration for Domain-specific Large Language Model Fine-tuning

Domain-specific large language models (LLMs), typically developed by fine-tuning a pre-trained general-purpose LLM on specialized datasets, represent a significant advancement in applied AI. A common strategy in LLM fine-tuning is curriculum learning, which pre-orders training samples based on metrics like difficulty to improve learning efficiency compared to a random sampling strategy. However, most existing methods for LLM fine-tuning rely on a static curriculum, designed prior to training, which lacks adaptability to the model's evolving needs during fine-tuning. To address this, we propose EDCO, a novel framework based on two key concepts: inference entropy and dynamic curriculum orchestration. Inspired by recent findings that maintaining high answer entropy benefits long-term reasoning gains, EDCO prioritizes samples with high inference entropy in a continuously adapted curriculum. EDCO integrates three core components: an efficient entropy estimator that uses prefix tokens to approximate full-sequence entropy, an entropy-based curriculum generator that selects data points with the highest inference entropy, and an LLM trainer that optimizes the model on the selected curriculum. Comprehensive experiments in communication, medicine and law domains, EDCO outperforms traditional curriculum strategies for fine-tuning Qwen3-4B and Llama3.2-3B models under supervised and reinforcement learning settings. Furthermore, the proposed efficient entropy estimation reduces computational time by 83.5% while maintaining high accuracy.

cs.LG↗

arXiv2024

Efficient k-step Weighted Reachability Query Processing Algorithms

Given a data graph G, a source vertex u and a target vertex v of a reachability query, the reachability query is used to answer whether there exists a path from u to v in G. Reachability query processing is one of the fundamental operations in graph data management, which is widely used in biological networks, communication networks, and social networks to assist data analysis. The data graphs in practical applications usually contain information such as quantization weights associated with the structural relationships, in addition to the structural relationships between vertices. Thus, in addition to the traditional reachability relationships, users may want to further understand whether such reachability relationships satisfy specific constraints. In this paper, we study the problem of efficiently processing k -step reachability queries with weighted constraints in weighted graphs. The k -step weighted reachability query questions are used to answer the question of whether there exists a path from a source vertex u to a goal vertex v in a given weighted graph. If it exists, the path needs to satisfy 1) all edges in the path satisfy the given weight constraints, and 2) the length of the path does not exceed the given distance threshold k. To address the problem, firstly, WKRI index supporting k -step weighted reachability query processing and index construction methods based on efficient pruning strategies are proposed. Secondly, the idea of constructing index based on part of the vertexs is proposed to reduce the size of the index. We design and implement two optimized indexes GWKRI and LWKRI based on the vertex coverage set. Finally, experiments are conducted on several real datasets. The experimental results verify the efficiency of the method proposed in this paper in answering k -step weighted reachability queries.

cs.DB↗

arXiv2022

Efficient Reachability Ratio Computation for 2-hop Labeling Scheme

As one of the fundamental graph operations, reachability queries processing has been extensively studied during the past decades. Many approaches followed the line of designing 2-hop labels to make acceleration. Considering that the index size cannot be bounded when using all nodes to construct 2-hop labels, researchers proposed to use a part of important nodes to construct 2-hop labels (partial 2-hop labels) to cover as much reachability information as possible. Then, we may achieve better query performance with limited index size and index construction time. However, partial 2-hop labels do not always perform well on different graphs. In this paper, we focus on the problem of how to efficiently compute reachability ratio, such that to help users determine whether partial 2-hop labels should be used to answer reachability queries for the given graph. Intuitively, reachability ratio denotes the ratio of the number of reachable queries that can be answered by partial 2-hop labels over the total number of reachable queries involved in the given graph. We discuss the difficulties of reachability ratio computation, and propose an incremental-partition algorithm for reachability ratio computation. We show by rich experimental results that our algorithm can efficiently get the result of reachability ratio, and show how the overall query performance is affected by different partial 2-hop labels. Based on the experimental results, we give out our findings on whether partial 2-hop labels should be used to the given graph for reachability queries processing.

cs.DB↗