摘要
非线性光学晶体是全固态激光器的核心器件,在信息技术、国防安全等方面具有广泛且重要的应用。随着高性能计算技术的发展,计算辅助实验的“自上而下”靶向设计逐渐成为非线性光学材料设计的重要组成部分。同时,基于高通量计算获取的大规模结构性能信息,进一步为数据挖掘、机器学习的算法训练提供了坚实的数据基础,促进了材料设计的第四范式的发展。本文从非线性光学晶体的计算材料设计入手,探讨了数据驱动的非线性光学理论设计的新范式,对本团队近年来在高通量筛选、晶体结构预测以及机器学习加速非线性光学材料设计等方向的研究进展进行了综述。
Nonlinear optical crystals are the core devices of all-solid-state lasers,and have extensive and important applications in information technology and national security.With the development of high-performance computing,the“top-down”computer-aided design methods have gradually become an important part of nonlinear optical materials design.In addition,the large-scale structural and properties information obtained based on high-throughput computing provides a solid data foundation for data mining and machine learning algorithm training,accelerating the development of the fourth paradigm of material design.This paper starts with the computational design of nonlinear optical materials,and then,discusses the new paradigm of data-driven nonlinear optical materials theoretical design.Finally,the recent research progress of our team in high-throughput screening,crystal structure prediction,and machine learning accelerated nonlinear optical materials are reviewed.
作者
储冬冬
杨志华
潘世烈
CHU Dongdong;YANG Zhihua;PAN Shilie(Research Center for Crystal Materials,Xinjiang Technical Institute of Physics and Chemistry,Chinese Academy of Sciences,Urumqi 830011,China)
出处
《人工晶体学报》
CAS
北大核心
2024年第9期1475-1493,共19页
Journal of Synthetic Crystals
基金
国家重点研发计划(2021YFB3601502)
国家自然科学基金(22193044,61835014,51972336)。
关键词
晶体结构预测
高通量筛选
机器学习
非线性光学材料
材料设计
crystal structure prediction
high-throughput screening
machine learning
nonlinear optical material
material design
