Abstract: Large language model agents have excelled at mathematical and coding benchmarks, but can they accelerate scientific discovery in quantum physics [1]? I present a multi-agent framework that integrates LLM reasoning with computational backends at increasing levels of rigor, and demonstrate three applications: (i) co-design of quantum error-correcting codes with transversal diagonal gates via multi-agent numerical search [3], (ii) automated discovery of matrix product operator symmetries through symbolic computation in Wolfram Mathematica [4], and (iii) formal verification of tensor network theory in Lean 4 [5]. The underlying system being developed, TeXRA [2], orchestrates specialized agents and builds research and verification harnesses that connect informal physical reasoning to computer algebra and machine-checked proof. I will share lessons learned from building and deploying this system, illustrating both current capabilities and open challenges for AI-assisted theoretical physics.

References:
[1] S. Lu, Z. Jin, T. J. Zhang, P. Kos, J. I. Cirac, and B. Schölkopf, "Can Theoretical Physics Research Benefit from Language Agents?", NeurIPS 2025 AI4Science Workshop (Oral), arXiv:2506.06214.
[2] TeXRA, https://texra.ai
[3] X. He, S. Lu, and B. Zeng, "Co-Designing Quantum Codes with Transversal Diagonal Gates via Multi-Agent Systems", arXiv:2510.20728.
[4-5] In preparation, with Y. Liu, E. Toja, J. I. Cirac.

Short Bio: Sirui Lu is a final-year doctoral researcher at the Max Planck Institute of Quantum Optics, supervised by Prof. J. Ignacio Cirac. His current research lies at the intersection of many-body physics and machine learning — on one hand leveraging AI to drive advances in quantum algorithm and quantum information, and on the other drawing on quantum and information-theoretic principles to provide new foundations for machine learning. He obtained his bachelor's degree from Tsinghua University, supervised by Prof. Luming Duan.

Contact: Lei Wang 9853