Accelerating lattice gauge theory studies with Agentic AI
Lattice gauge theory research, with its computationally intensive simulations and complex multi‑stage workflows, is well positioned to benefit from agentic AI systems. We demonstrate how such tools can support key components of lattice gauge theory research, including novel simulation code development using standard LQCD frameworks, HPC job orchestration, simulation data analysis, and expert‑guided tuning of algorithmic parameters such as Hasenbusch mass preconditioning and multigrid solvers. Our results show that agentic AI can reduce manual effort, improve productivity, and accelerate the research cycle while maintaining essential human oversight.
Ayyar, Venkitesh [Fermilab]↗