Cluster Expansion of Alloy Theory: A Review of Historical Development and Modern Innovations
Cluster Expansion of Alloy Theory: A Review of Historical Development and Modern Innovations
复制标题
DOI：
10.1007/s11837-021-04840-6
复制
发表时间：
2021-09
期刊：
影响因子：
2.6
通讯作者：
S. Kadkhodaei;Jorge A. Muñoz
中科院分区：
文献类型：
--
作者：
S. Kadkhodaei;Jorge A. Muñoz
从一组高度精确但昂贵的计算或测量中对物理或经验模型进行参数化，以产生不太精确但便宜的预测，这在许多学科中都很常见。在计算材料科学和信息学支持的材料设计中，簇展开（CE）方法提供了晶格自由能或任何其他热力学变量的直接近似，以离散簇函数的形式，使其成为相图计算最广泛使用的方法之一，包括有序-无序相变。在本文中，我们回顾了最终形成CE方法的理论发展，目前用于拟合和优化CE模型参数的众多统计技术，CE方法与现代机器学习和数据科学技术的融合，以及推动该领域超越传统CE的最新发展，包括结构合金设计。
The parameterization of a physical or empirical model from a set of highly accurate but expensive calculations or measurements to generate less precise but cheaper predictions is common in many disciplines. In computational materials science and informatics-enabled design of materials, the cluster expansion (CE) method provides a direct approximation of the free energy of a lattice, or any other thermodynamic variable, in terms of a discrete cluster function, making it one of the most widely used approaches for phase diagram calculations, including order–disorder phase transitions. In this article, we review the theoretical developments that culminated in the formulation of the CE method, numerous statistical techniques currently used to fit and optimize the parameters of the CE model, the convergence of the CE method with modern machine learning and data science techniques, and recent developments that push the field beyond the conventional CE, including for structural alloy design.