ITP|
|
WS23/24 – Abstract
Alexander Rothkopf
(University of Stavanger)
18.1.2024
Tackling the sign problem with machine learning assisted real-time simulations
Accurately predicting the quantum many-body dynamics of strongly correlated quantum systems holds the key to understanding phenomenologically relevant phenomena on widely different energy scales. One example is the dynamics of impurities in hot nuclear matter (quarkonia) or in a cold quantum gas (polarons). The ab-initio determination of real-time dynamics in such strongly correlated systems is however severely limited due to the notorious sign problem. In this talk I will discuss some of the successes and limitations of using conventional Monte-Carlo simulations to indirectly compute the real-time dynamics in strongly interacting matter (lattice QCD) and present a recently developed strategy, inspired by reinforcement learning, to improve the direct simulation of strongly correlated field theories.
