arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.11704hep-lat
收藏

关于动力学费米子算法的综述

Review on Algorithms for dynamical fermions

Jacob Finkenrath

AI总结:

本文综述了大规模格点QCD模拟中动力学费米子算法的研究进展,涵盖物理点系综生成、HMC算法模拟大格点的步骤,并重点探讨了克服拓扑电荷冻结的算法改进与全局校正、生成模型及多层算法等前沿发展。

摘要:

本综述概述了用于大规模格点QCD模拟的动力学费米子算法的研究。首先简要概述了在物理点上进行系综生成的最新技术进展。随后概述了使用 Hybrid Monte Carlo 算法模拟大格点的必要步骤,其中讨论了分子动力学中迭代求解器和数值积分器的调优程序。接着回顾了算法的持续发展,重点关注可能有助于在极细格距下模拟规范理论的方法,即非常适合克服拓扑电荷冻结的方法。这包括对 HMC 算法的修改,以及通过全局校正步骤引入费米子权重的算法讨论。部分讨论涉及通过规范等变流应用生成模型以及多层算法。

英文摘要:

This review gives an overview on the research of algorithms for dynamical fermions used in large scale lattice QCD simulations. First a short overview on the state-of-the-art of ensemble generation at the physical point is given. Followed by an overview on necessary steps towards simulation of large lattices with the Hybrid Monte Carlo algorithm. Here, the status of iterative solvers and tuning procedures for numerical integrators within the molecular dynamics are discussed. This is followed by a review on the on-going developments for algorithms, with a focus on methods which are potentially useful to simulate gauge theories at very fine lattice spacings, i.e. well suited to overcome freezing of the topological charge. This includes modification of the HMC algorithm as well as a discussion of algorithms which includes the fermion weight via global correction steps. Parts of the discussions are on the application of generative models via gauge equivariant flows as well as multi-level algorithms.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑