Title: Simulating the real-time dynamics of strongly correlated quantum fields from first principles
일시: 2025년 3월 18일(화), 17시
Speaker: Alexander Rothkopf (Korea University)
Abstract:
The vast amount of data collected in relativistic heavy-ion collision experiments at the Large-Hadron-Collider at CERN amounts to a treasure trove for theory to understand nuclear matter under extreme conditions, such as in the form of a quark-gluon-plasma (QGP) at temperatures above 10¹²K. While significant progress has been made in uncovering the static properties of the strongly correlated QGP (phase structure, equation of state), its dynamical properties, such as interaction with external impurities (e.g., quark-antiquark pairs) and its transport properties (viscosity/conductivity), are much less understood.
A central hurdle obstructing progress is the so-called sign problem, which prevents well-established simulation techniques on classical computers from directly accessing real-time dynamics. So far, dynamical information is obtained only indirectly as an ill-posed inverse problem.
Over the past years, my group has focused on developing novel simulation techniques to directly simulate the dynamics of strongly correlated quantum fields. We focus on the stochastic quantization approach, one of the competing strategies proposed to tackle the sign problem. In stochastic quantization, quantum fluctuations are incorporated by placing the classical system in an additional fictitious time dimension in which it evolves randomly. This approach, called complex Langevin, has shown promise in correctly describing several model systems but has suffered from two key drawbacks: divergent trajectories and convergence to incorrect solutions.
By taking inspiration from financial engineering and machine learning, we have improved complex Langevin by incorporating prior information available for the simulated system (e.g., symmetries) to significantly extend the reach of this first-principles simulation technique. I will discuss the current limitations of our approach and directions for future work.