LQCDMaster: Agentic Scientific Computing for Lattice Quantum Chromodynamics Research

📅 2026-07-16
📈 Citations: 0
✨ Influential: 0
📄 PDF
🤖 AI Summary
This work addresses the heavy reliance on expert knowledge and low automation in lattice quantum chromodynamics (LQCD) workflows by introducing the first end-to-end LQCD intelligent agent system. Integrating agent-based planning, an expert-curated skill library, a deterministic Wick contraction tool, and the high-performance PyQUDA framework, the system automatically translates natural language physics tasks into fully executable computational workflows. Evaluated on 70 state-of-the-art tasks, the approach achieves machine-precision agreement with expert implementations in 63 cases, with discrepancies in three attributable to gauge conventions. The method reduces workflow execution time from hours to minutes and, for the first time, enables automated computation of non-standard observables such as light-cone distribution amplitudes, substantially lowering the barrier to entry and expanding the frontier of LQCD research.
📝 Abstract
Lattice quantum chromodynamics (LQCD) provides a first-principles framework for computing hadronic observables, but its practical use remains limited by the substantial expertise required to turn research motivation into reliable computing workflows. Here we present \textsc{LQCDMaster}, a tool-augmented, skill-guided and domain-specialized scientific computing agent that converts natural-language LQCD research tasks into executable PyQUDA computing workflows, including measurement scripts, job-submission artifacts, execution logs and numerical outputs. The system combines agentic planning, expert-annotated LQCD skills and a deterministic Wick-contraction tool to constrain the algebraically fragile components of code generation. We evaluate \textsc{LQCDMaster} on a benchmark at the forefront of scientific research, comprising 70 LQCD computing tasks, with observables covering local and nonlocal two-point functions, Wilson loops, meson and baryon three-point functions. The generated workflows exactly reproduce expert-written implementations in 63 of 70 tasks at machine precision, with three additional discrepancies attributable to convention mismatches. Across representative observables, the agent reduces implementation time from hours to minutes while preserving end-to-end numerical validation. Further, we present a typical case of \textsc{LQCDMaster}-driven exploration: a lattice computation of light-cone distribution amplitudes with diagonal Wilson-line, a quantity accessible with standard methods but never before computed, and computation of the spectrum of proton, deuteron, triton, hyperon, hyperdeuteron and hypertriton. This work pioneers the paradigm of agentic scientific computing by automating the end-to-end scientific computing workflows in lattice QCD research, lowering its barrier and facilitating the exploration and verification of non-standard scientific ideas.
Problem

Research questions and friction points this paper is trying to address.

Lattice Quantum Chromodynamics
scientific computing
workflow automation
hadronic observables
computational barrier
Innovation

Methods, ideas, or system contributions that make the work stand out.

agentic scientific computing
Lattice QCD
code generation
Wick contraction
PyQUDA
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
Authors to Follow
H
Haofei Gao
School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, 200240, China.; Tsung-Dao Lee Institute, Shanghai Jiao Tong University, Shanghai, 200240, China.; State Key Laboratory of Dark Matter Physics, Shanghai Jiao Tong University, Shanghai, 200240, China.
T
Tingjia Miao
School of Artificial Intelligence, Shanghai Jiao Tong University, Shanghai, 200030, China.; Zhiyuan College, Shanghai Jiao Tong University, Shanghai, 200240, China.
W
Wenkai Jin
School of Artificial Intelligence, Shanghai Jiao Tong University, Shanghai, 200030, China.; SciLand, Shanghai, 200030, China.
Muhua Zhang
Muhua Zhang
School of Electrical Engineering, Southwest Jiaotong University
RoboticsMobile Robots
H
Hanzhang Wang
School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, 200240, China.; Tsung-Dao Lee Institute, Shanghai Jiao Tong University, Shanghai, 200240, China.; State Key Laboratory of Dark Matter Physics, Shanghai Jiao Tong University, Shanghai, 200240, China.
Jie Ran
Jie Ran
The University of Hong Kong
J
Jinxin Tan
School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, 200240, China.; Tsung-Dao Lee Institute, Shanghai Jiao Tong University, Shanghai, 200240, China.; State Key Laboratory of Dark Matter Physics, Shanghai Jiao Tong University, Shanghai, 200240, China.
Z
Zhentao Zhang
School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, 200240, China.; Zhiyuan College, Shanghai Jiao Tong University, Shanghai, 200240, China.
B
Bo Tang
School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, 200240, China.; Zhiyuan College, Shanghai Jiao Tong University, Shanghai, 200240, China.
L
Leiyi Li
School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, 200240, China.; Tsung-Dao Lee Institute, Shanghai Jiao Tong University, Shanghai, 200240, China.; State Key Laboratory of Dark Matter Physics, Shanghai Jiao Tong University, Shanghai, 200240, China.
J
Jun Hua
Institute of Quantum Matter, South China Normal University, Guangzhou, 510006, China.
X
Xiangyu Jiang
Department of Physics, Indiana University, Bloomington, Indiana 47405, USA.
Q
Qi-An Zhang
School of Physics, Beihang University, Beijing, 102206, China.
Siheng Chen
Siheng Chen
Shanghai Jiao Tong University
Collective intelligenceLLM agentgraph signal processingcollaborative perception
Wei Wang
Wei Wang
Shanghai Jiao Tong University
speech recognitionspeech enhancementtext-to-speech