AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-26
QuantumMatterExplorer: An Autonomous Agent for the Discovery of Quantum Many-Body Phases of Matter
Mapping the phase diagram of a new quantum Hamiltonian normally requires expert judgement at every step. QuantumMatterExplorer is an LLM-based autonomous agent that takes only a Hamiltonian and its parameter ranges, runs its own DMRG simulations cross-checked against exact diagonalization, and returns a phase diagram. Applied to a frustrated spin-1 chain without hints, it identified candidate Néel, Haldane and dimerized phases broadly consistent with the reference, detected incommensurate correlations, and finished within an eight-hour compute budget.