TRANSFER LEARNING FOR NEURAL-NETWORK QUANTUM STATES

Abstract

Neural-network quantum state trains a neural network to be a surrogate of the wave function of a quantum many-body system. An important question in quantum many-body physics is the identification of the phases. Phase transitions occur at quantum critical points. Empirically, a quantum critical point can be identified by finding the inflection points of a physical quantity in the ground state. Our first and second contributions consist of several physics-informed transfer learning protocols to scale neural-network quantum states to larger sizes and explore the space of parameters for the effective and efficient identification of quantum critical points. We also devise transfer learning protocols to larger, deeper, and broader networks as the third contribution. Quantum critical points formally corresponding to closing spectral gap can be relevant to approximating the wave function of excited states. We devise and adopt four transfer learning protocols to achieve this purpose as our fourth contribution.