Di Luo, Massachusetts Institute of Technology
Title
Machine Learning meets Quantum Many-body Physics
Abstract
The simulation of quantum many-body physics, pivotal in uncovering ground state properties and real-time dynamics, is essential in the study of quantum science. In this talk, I will focus on how neural network quantum states, enriched with symmetries and physics principles, provide new opportunities for tackling challenges in quantum many-body simulations. I will introduce the pioneering work of designing anti-symmetric and gauge equivariant neural wavefunctions, which provides new tools for exploring exotic phases of quantum matter in two-dimensional quantum materials and quantum gauge theories. Furthermore, I will discuss how neural network generative models can be used to simulate real-time open quantum systems based on quantum information theory, and applied in quantum experiments and computation. I will conclude with a discussion on the new possibilities of AI for physics, as well as how physics theories can help advance AI.