Center for Advanced Systems Understanding (CASUS), Görlitz, Germany.
Phys Rev E. 2019 Aug;100(2-1):023307. doi: 10.1103/PhysRevE.100.023307.
The ab initio thermodynamic simulation of correlated Fermi systems is of central importance for many applications, such as warm dense matter, electrons in quantum dots, and ultracold atoms. Unfortunately, path integral Monte Carlo (PIMC) simulations of fermions are severely restricted by the notorious fermion sign problem (FSP). In this paper, we present a hands-on discussion of the FSP and investigate in detail its manifestation with respect to temperature, system size, interaction-strength and -type, and the dimensionality of the system. Moreover, we analyze the probability distribution of fermionic expectation values, which can be non-Gaussian and fat-tailed when the FSP is severe. As a practical application, we consider electrons and dipolar atoms in a harmonic confinement, and the uniform electron gas in the warm dense matter regime. In addition, we provide extensive PIMC data, which can be used as a reference for the development of new methods and as a benchmark for approximations.
从头算相关费米体系的热力学模拟在许多应用中都非常重要,例如温稠密物质、量子点中的电子和超冷原子。不幸的是,费米子的路径积分蒙特卡罗(PIMC)模拟受到著名的费米子符号问题(FSP)的严重限制。在本文中,我们将对 FSP 进行实际讨论,并详细研究其与温度、系统大小、相互作用强度和类型以及系统维度的关系。此外,我们还分析了费米子期望值的概率分布,当 FSP 严重时,它可能是非高斯和长尾的。作为一个实际应用,我们考虑了在谐和约束下的电子和偶极原子以及温稠密物质状态下的均匀电子气。此外,我们还提供了广泛的 PIMC 数据,这些数据可用于开发新方法的参考和近似的基准。