题目：Simulating Quantum Dynamics by Dynamic Mode Decomposition and Operator Learning

报告人：印佳 研究员（复旦大学）

地点：致远楼108室

时间：2026年10月30日 星期五 15：30-16：30

Abstract：Long-time simulations of nonequilibrium quantum dynamics are often limited by high-dimensional state spaces, the complexity of two-time Green’s functions, and the increasing cost of evaluating memory integrals. This talk explores how short-time data can be used to predict long-time behavior. I will first introduce dynamic mode decomposition (DMD), which extracts dominant dynamical modes from short trajectories. Applications include forecasting observables in quantum many-body systems, extrapolating diagonal and off-diagonal components of nonequilibrium Green’s functions, and accelerating electron–phonon dynamics. I will then turn to operator learning as a complementary approach to modeling memory-dependent evolution. Finally, I will discuss a hybrid method that combines short-time quantum-computer measurements with classical DMD.

欢迎各位参加！