arXiv subjects

Zhijie Sun

Publications and source records attributed to Zhijie Sun.

16 recordsLinked to original sources

arXiv2026

A High-Performance Pauli-Algebra Framework for Large-Scale Quantum Simulations

Efficient manipulation of Pauli-algebraic objects is a key bottleneck in the classical emulation and benchmarking of quantum algorithms for chemistry and many-body physics. This bottleneck appears in Hamiltonian construction, variational ansatz preparation, expectation-value and gradient evaluation, and real-time propagation, all of which require repeated Pauli-algebra operations. Here, we present a high-performance Pauli-algebra framework tailored to quantum many-body and quantum-chemical simulations. The framework combines compact binary symplectic encoding, canonical coefficient reduction, and grouped sparse operator representations that exploit shared bit-flip patterns among Pauli strings. The resulting Julia/C\texttt{++} implementation accelerates Pauli multiplication, Hamiltonian construction, and operator--state multiplication in sparse and symmetry-adapted many-electron spaces. Benchmarks demonstrate efficient Hamiltonian construction, large-active-space VQE and ADAPT-VQE calculations, and real-time variational dynamics on modern multicore CPU and GPU architectures. These results show that structure-aware Pauli-algebra engines provide a scalable classical backend for developing and benchmarking quantum algorithms in quantum chemistry and many-body simulation.

quant-ph↗

arXiv2025

Data Trajectory Alignment for LLM Domain Adaptation: A Two-Phase Synthesis Framework for Telecommunications Mathematics

General-purpose large language models (LLMs) are increasingly deployed in verticals such as telecommunications, where adaptation is hindered by scarce, low-information-density corpora and tight mobile/edge constraints. We propose Data Trajectory Alignment (DTA), a two-phase, model-agnostic data curation framework that treats solution processes - not only final answers - as first-class supervision. Phase I (Initializing) synthesizes diverse, high-coverage candidates using an ensemble of strong teachers. Phase II (DTA) rewrites teacher solutions to align intermediate steps and presentation style with the target student's inductive biases and then performs signal-aware exemplar selection via agreement checks and reflection-based judging. Instantiated on telecommunications mathematics (e.g., link budgets, SNR/AMC selection, and power-control feasibility), DTA yields state-of-the-art (SOTA) accuracy on TELEMATH without enabling explicit "thinking" modes: 72.45% pass@1, surpassing distilled-only training by +17.65 points and outperforming a strong baseline (Qwen3-32B with thinking enabled) by +2.94 points. Token-shift analyses indicate that DTA concentrates gains on logical-structural discourse markers rather than merely amplifying domain nouns, indicating improved reasoning scaffolding. Under edge-like inference settings, DTA improves efficiency by reducing reliance on multi-sample voting and disabling expensive reasoning heuristics, cutting energy per output token by ~42% versus Qwen3-32B (thinking mode enabled) and end-to-end latency by ~60% versus Qwen3-32B (thinking mode disabled). These results demonstrate that aligning how solutions are produced enables compact, high-yield supervision that is effective for both accuracy and efficiency, offering a practical recipe for domain adaptation in low-resource verticals beyond telecom.

cs.LG↗

arXiv2025

Improved time-translationally invariant tensor network influence functional method for Anderson impurity problems

The Anderson impurity model (AIM) is of fundamental importance in condensed matter physics for studying strongly correlated phenomena. However, accurately simulating its long-time dynamics still remains a significant numerical challenge. A class of recently developed numerical approaches represents the Feynman-Vernon influence functional (IF), which encodes all the bath effects on the impurity, as a matrix product state (MPS) in the temporal domain. The computational cost of this approach is largely determined by the bond dimension $\chi$ of the temporal MPS. In this work, we propose an efficient and accurate method that, when the hybridization function in the IF can be approximated as a sum of $n$ exponential functions, systematically constructs the IF as an MPS by multiplying $O(n)$ small MPSs, each with bond dimension $2$. Our method yields a worst case scaling of $\chi$ as $2^{8n}$ and $2^{2n}$ for real- and imaginary-time evolution respectively. We demonstrate the performance of our method for two commonly used bath spectral functions, and show that the required bond dimensions are significantly smaller than the worst case.

cond-mat.str-el↗

arXiv2025

HES-SQL: Hybrid Reasoning for Efficient Text-to-SQL with Structural Skeleton Guidance

We present HES-SQL, a novel hybrid training framework that advances Text-to-SQL generation through the integration of thinking-mode-fused supervised fine-tuning (SFT) with Group Relative Policy Optimization (GRPO). Our approach introduces three key innovations: (1) a skeleton-completeness scoring mechanism that enhances preference alignment between generated queries and optimal SQL structures; (2) a query-latency-aware reward system that incentivizes the generation of computationally efficient SQL queries; (3) a self-distillation process for thinking-mode completion that prevents degradation of the model's reasoning capabilities. This framework enables hybrid thinking models to switch between reasoning and non-reasoning modes while improving SQL query accuracy and execution efficiency. Experimental evaluation, conducted on MySQL 8.0 and SQLite 3.42 under controlled single-user conditions, demonstrates that HES-SQL achieves competitive performance with execution accuracies of 79.14\% and 54.9\% on the BIRD and KaggleDBQA benchmarks, respectively. Query latency is measured as the end-to-end execution time of generated queries on the DBMS, averaged over multiple runs to mitigate variance. Efficiency gains range from 11\% to 20\% relative to supervised baselines. Our results establish a new paradigm for Text-to-SQL systems that effectively balances semantic accuracy with computational efficiency through execution-informed reinforcement learning (RL). The proposed methodology has significant implications for developing robust natural language interfaces to databases and can be extended to broader structured generation tasks requiring both correctness and efficiency optimization.

cs.DB↗

arXiv2025

Logits Replay + MoClip: Stabilized, Low-Cost Post-Training with Minimal Forgetting

Large language models (LLMs) often face a trade-off in post-training: improvements on specialized domains frequently come at the expense of general capabilities. Existing solutions attempt to mitigate this tension via regularization, selective parameter updates, or data-centric replay, but each imposes significant costs in computation, data access, or adaptability. Recent work has shown that training signals can be compressed to subsets of logits without severe accuracy loss, suggesting a path toward efficient adaptation. However, naive truncation destabilizes optimization and exacerbates forgetting. We introduce Logits Replay + MoClip, a two-stage framework that compresses supervision in the logit space and stabilizes optimization at the update level. In Stage 0, we record dynamic Top-K token subsets that cover a probability threshold, always including the gold label. In Stage 1, we replay these compact subsets to compute exact renormalized losses, avoiding full softmax computation and implicitly regularizing. To ensure stability, we design MoClip, an optimizer that caps gradient-momentum rotation and applies an arctan2-based rescaling of updates. Empirically, our method improves domain performance on Communication Technology (CT) and NL2SQL tasks while mitigating forgetting on general benchmarks (MMLU, BBH, GPQA, MATH), and reduces training cost by over 40%. Together, these contributions offer a scalable, architecture-agnostic path for domain adaptation of LLMs without sacrificing generalization.

cs.LG↗

arXiv2025

Agentic-KGR: Co-evolutionary Knowledge Graph Construction through Multi-Agent Reinforcement Learning

Current knowledge-enhanced large language models (LLMs) rely on static, pre-constructed knowledge bases that suffer from coverage gaps and temporal obsolescence, limiting their effectiveness in dynamic information environments. We present Agentic-KGR, a novel framework enabling co-evolution between LLMs and knowledge graphs (KGs) through multi-round reinforcement learning (RL). Our approach introduces three key innovations: (1) a dynamic schema expansion mechanism that systematically extends graph ontologies beyond pre-defined boundaries during training; (2) a retrieval-augmented memory system enabling synergistic co-evolution between model parameters and knowledge structures through continuous optimization; (3) a learnable multi-scale prompt compression approach that preserves critical information while reducing computational complexity through adaptive sequence optimization. Experimental results demonstrate substantial improvements over supervised baselines and single-round RL approaches in knowledge extraction tasks. When integrated with GraphRAG, our method achieves superior performance in downstream QA tasks, with significant gains in both accuracy and knowledge coverage compared to existing methods.

cs.LG↗

arXiv2025

Expert-Token Resonance MoE: Bidirectional Routing with Efficiency Affinity-Driven Active Selection

Mixture-of-Experts (MoE) architectures enable efficient scaling of large language models by activating only a subset of parameters per input. However, existing MoE models suffer from two critical limitations: (1) inefficient token-to-expert routing that causes excessive communication overhead, and (2) expert homogenization that leads to redundant computations. Current approaches address these challenges separately, failing to achieve simultaneous improvements in both training efficiency and model performance. We present Expert-Token Resonance (ETR), a theoretically-grounded bidirectional routing mechanism that fundamentally reimagines expert-token interactions in MoE architectures. Our key insight is that optimal routing requires adaptive coordination between token-choice routing (TCR) during early training phases and expert-choice routing (ECR) in later stages. We prove that this dynamic approach maximizes training success rate (the probability of correct token-expert assignments) while reducing the expert capacity lower bound by up to 40%. ETR incorporates three technical innovations: (1) an affinity-based routing architecture using Grouped Average Pooling (GrAP) that reduces computational complexity from O(d^2) to O(d^2/D) while maintaining orthogonality to prevent expert homogenization; (2) a bidirectional selection mechanism that enables both tokens and experts to actively participate in the routing process based on cosine similarity scores; and (3) an adaptive capacity strategy that dynamically adjusts expert bounds based on training progress, eliminating communication bubbles in All-to-All operations. Extensive experiments on Ascend NPU clusters demonstrate that ETR achieves 5.4%-46.6% improvements in end-to-end training efficiency compared to baseline MoE implementations, with 9.7%-14.5% performance gains across GDAD, GPQA, HumanEval, and TeleQnA benchmarks.

cs.CL↗

arXiv2025

Scalable tensor network algorithm for quantum impurity problems

The Grassmann time-evolving matrix product operator method has shown great potential as a general-purpose quantum impurity solver, as its numerical errors can be well-controlled and it is flexible to be applied on both the imaginary- and real-time axis. However, a major limitation of it is that its computational cost grows exponentially with the number of impurity flavors. In this work, we propose a multi-flavor extension of it to overcome this limitation. The key insight is that to calculate multi-time correlation functions on one or a few impurity flavors, one could integrate out the degrees of freedom of the rest flavors before hand, which could greatly simplify the calculation. The idea is particularly effective for quantum impurity problems with diagonal hybridization function, i.e., each impurity flavor is coupled to an independent bath, a setting which is commonly used in the field. We demonstrate the accuracy and scalability of our method for the imaginary time evolution of impurity problems with up to three impurity orbitals, i.e., 6 flavors, and benchmark our results against continuous-time quantum Monte Carlo calculations. Our method paves the way of scaling up tensor network algorithms to solve large-scale quantum impurity problems.

cond-mat.str-el↗

arXiv2025

DeepJSONEval: Benchmarking Complex Nested JSON Data Mining for Large Language Models

The internet is saturated with low-density, high-redundancy information, such as social media comments, repetitive news, and lengthy discussions, making it difficult to extract valuable insights efficiently. Multi-layer nested JSON structures provide an effective solution by compressing such information into semantically rich, hierarchical representations, which organize data into key-value pairs, arrays, and nested objects, preserving contextual relationships and enabling efficient storage, retrieval, and semantic querying. For instance, in news aggregation, a JSON object can nest an article's metadata (title, author, date), content (text, multimedia), and multimedia information (multimedia type, caption) hierarchically. Large Language Models (LLMs) play a transformative role in web data mining by parsing unstructured text and outputting structured results directly into complex JSON schemas. However, current benchmarks for evaluating LLMs' JSON output capabilities overemphasize pure JSON generation rather than assessing data comprehension and extraction abilities, a limitation that lacks relevance to practical web data mining tasks. To address this, we introduce DeepJSONEval, a novel benchmark featuring 2100 multi-domain instances with deep nested structures, categorized by difficulty. Experiments show significant performance gaps among LLMs in handling such complexity. Our benchmark and datasets are open-sourced to advance research in structured JSON generation.(https://github.com/GTS-AI-Infra-Lab-SotaS/DeepJSONEval).

cs.AI↗

arXiv2025

Infinite Grassmann time-evolving matrix product operators for quantum impurity problems after a quench

An emergent numerical approach to solve quantum impurity problems is to encode the impurity path integral as a matrix product state. For time-dependent problems, the cost of this approach generally scales with the evolution time. Here we consider a common non-equilibrium scenario where an impurity, initially in equilibrium with a thermal bath, is driven out of equilibrium by a sudden quench of the impurity Hamiltonian. Despite that there is no time-translational invariance in the problem, we show that we could still make full use of the infinite matrix product state technique, resulting in a method whose cost is essentially independent of the evolution time. We demonstrate the effectiveness of this method in the integrable case against exact diagonalization, and against existing calculations on the L-shaped Kadanoff-Baym contour in the general case. Our method could be a very competitive method for studying long-time non-equilibrium quantum dynamics, and be potentially used as an efficient impurity solver in the non-equilibrium dynamical mean field theory.

cond-mat.str-el↗

arXiv2025

Dynamical multipartite entanglement in a generalized Tavis-Cummings model with XY spin interaction

Multipartite entanglement is a long-term pursuit in the resource theory, offering a potential resource for quantum metrology. Here, we present the dynamical multipartite entanglement, which is in terms of the quantum Fisher information, of a generalized Tavis-Cummings (TC) model introducing the XY spin interaction. Since our model cannot be solved exactly, we theoretically derive and numerically examine the effective description of our model. By the Holstein-Primakoff transformation, we show the bridge from the generalized TC model to the central spin model. Furthermore, the reduced density matrix of the central spins is presented, which is the prerequisite for calculating multipartite entanglement. We also discuss the effect of the temperature, the coupling constant, and the magnetic field on the dynamical multipartite entanglement in the central spin model, where the central spin is initially unentangled. Strong coupling and low temperature are necessary conditions for a genuine multipartite entanglement in the XY model, and together with the magnetic field, they govern the modulation of both the entanglement period and amplitude. Our results unveil the deep link between the TC model and the central spin model, allowing for a better comprehension of their dynamical multipartite entanglement.

quant-ph↗

arXiv2024

Circuit-Efficient Qubit-Excitation-based Variational Quantum Eigensolver

The wave function Ansatze are crucial in the context of the Variational Quantum Eigensolver (VQE). In the Noisy Intermediate-Scale Quantum era, the design of low-depth wave function Ansatze is of great importance for executing quantum simulations of electronic structure on noisy quantum devices. In this work, we present a circuit-efficient implementation of two-body Qubit-Excitation-Based (QEB) operator for building shallow-circuit wave function Ansatze within the framework of Adaptive Derivative-Assembled Pseudo-Trotter (ADAPT) VQE. This new algorithm is applied to study ground- and excited-sate problems for small molecules, demonstrating significant reduction of circuit depths compared to fermionic excitation-based and QEB ADAPT-VQE algorithms. This circuit-efficient algorithm shows great promise for quantum simulations of electronic structures, leading to improved performance on current quantum hardware.

quant-ph↗

arXiv2024

LocMoE: A Low-Overhead MoE for Large Language Model Training

The Mixtures-of-Experts (MoE) model is a widespread distributed and integrated learning method for large language models (LLM), which is favored due to its ability to sparsify and expand models efficiently. However, the performance of MoE is limited by load imbalance and high latency of All-to-All communication, along with relatively redundant computation owing to large expert capacity. Load imbalance may result from existing routing policies that consistently tend to select certain experts. The frequent inter-node communication in the All-to-All procedure also significantly prolongs the training time. To alleviate the above performance problems, we propose a novel routing strategy that combines load balance and locality by converting partial inter-node communication to that of intra-node. Notably, we elucidate that there is a minimum threshold for expert capacity, calculated through the maximal angular deviation between the gating weights of the experts and the assigned tokens. We port these modifications on the PanGu-Sigma model based on the MindSpore framework with multi-level routing and conduct experiments on Ascend clusters. The experiment results demonstrate that the proposed LocMoE reduces training time per epoch by 12.68% to 22.24% compared to classical routers, such as hash router and switch router, without impacting the model accuracy.

cs.LG↗

arXiv2022

A blockchain-based secure storage scheme for medical information

Medical data involves a large amount of personal information and is highly privacy sensitive. In the age of big data, the increasing informatization of healthcare makes it vital that medical information is stored securely and accurately. However, current medical information is subject to the risk of privacy leakage and difficult to share. To address these issues, this paper proposes a healthcare information security storage solution based on Hyperledger Fabric and the Attribute-Based Access Control (ABAC) framework. The scheme first utilizes attribute-based access control, which allows dynamic and fine-grained access to medical information, and then stores the medical information in the blockchain, which can be secured and tamper-proof by formulating corresponding smart contracts. In addition, this solution also incorporates IPFS technology to relieve the storage pressure of the blockchain. Experiments show that the proposed scheme combining access control of attributes and blockchain technology in this paper can not only ensure the secure storage and integrity of medical information but also has a high throughput when accessing medical information.

cs.CR↗

arXiv2021

Fabric-SCF: A Blockchain-based Secure Storage and Access Control Scheme for Supply Chain Finance

Supply chain finance(SCF) is committed to providing credit for small and medium-sized enterprises(SMEs) with low credit lines and small financing scales. The resulting financial credit data and related business transaction data are highly confidential and private. However, traditional SCF management schemes mostly use third-party platforms and centralized designs, which cannot achieve highly reliable secure storage and fine-grained access control. To fill this gap, this paper designs and implements Fabric-SCF, a secure storage and access control system based on blockchain and attribute-based access control (\textbf{ABAC}) model. This scheme uses distributed consensus to realize data security, traceability, and immutability. We also use smart contracts to define system processes and access policies to ensure the efficient operation of the system. To verify the performance of Fabric-SCF, we designed two sets of simulation experiments. The results show that Fabric-SCF achieves dynamic and fine-grained access control while maintaining high throughput in a simulated real-world operating scenario.

cs.CR↗

arXiv2012

Partial Sliced Inverse Regression for Quality-Relevant Multivariate Statistical Process Monitoring

This paper introduces a popular dimension reduction method, sliced inverse regression (SIR), into multivariate statistical process monitoring. Provides an extension of SIR for the single-index model by adopting the idea from partial least squares (PLS). Our partial sliced inverse regression (PSIR) method has the merit of incorporating information from both predictors (x) and responses (y), and it has capability of handling large, nonlinear, or "n<p" dataset. Two statistics with their corresponding distributions and control limits are given based on the X-space decomposition of PSIR for the purpose of fault detection in process monitoring. Simulations showed PSIR outperformed over PLS and SIR for both linear and nonlinear model.

stat.AP↗