Network-Initialized Monte Carlo Based on Generative Neural Networks

作     者：Hongyu Lu Chuhao Li Bin-Bin Chen Wei Li Yang Qi Zi Yang Meng 卢虹宇;李楚豪;陈斌斌;李伟;戚扬;孟子杨

作者机构：Department of Physics and HKU-UCAS Joint Institute of Theoretical and Computational PhysicsThe University of Hong KongHong KongChina Beijing National Laboratory for Condensed Matter Physicsand Institute of PhysicsChinese Academy of SciencesBeijing 100190China School of Physical SciencesUniversity of Chinese Academy of SciencesBeijing 100190China Institute of Theoretical PhysicsChinese Academy of SciencesBeijing 100190China School of PhysicsBeihang UniversityBeijing 100191China State Key Laboratory of Surface PhysicsFudan UniversityShanghai 200438China Center for Field Theory and Particle PhysicsDepartment of PhysicsFudan UniversityShanghai 200433China 

出 版 物：《Chinese Physics Letters》 (中国物理快报（英文版）)

年 卷 期：2022年第39卷第5期

页      面：23-28页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 07[理学] 081104[工学-模式识别与智能系统] 08[工学] 0714[理学-统计学（可授理学、经济学学位）] 070201[理学-理论物理] 0835[工学-软件工程] 0701[理学-数学] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 0702[理学-物理学] 

基　　金：support from the RGC of Hong Kong SAR of China(Grant Nos.17303019,17301420,17301721,and Ao E/P-701/20) the National Natural Science Foundation of China(Grant Nos.11974036,11874115,and 11834014) the Strategic Priority Research Program of the Chinese Academy of Sciences(Grant No.XDB33000000) the K.C.Wong Education Foundation(Grant No.GJTD-2020-01) supported by the Seed Funding“Quantum-Inspired explainable-AI”at the HKU-TCL Joint Research Centre for Artifcial Intelligence,Hong Kong 

主　　题：process critical chains 

摘      要：We design generative neural networks that generate Monte Carlo configurations with complete absence of autocorrelation from which only short Markov chains are needed before making measurements for physical observables,irrespective of the system locating at the classical critical point,fermionic Mott insulator,Dirac semimetal,or quantum critical *** further propose a network-initialized Monte Carlo scheme based on such neural networks,which provides independent samplings and can accelerate the Monte Carlo simulations by significantly reducing the thermalization *** demonstrate the performance of our approach on the two-dimensional Ising and fermion Hubbard models,expect that it can systematically speed up the Monte Carlo simulations especially for the very challenging many-electron problems.