Models of configurationally-complex alloys made simple

We present a Python package for the efficient generation of special quasi-random structures (SQS) for atomic-scale calculations of disordered systems. Both, a Monte-Carlo approach or a systematic enumeration of structures can be used to carry out optimizations to ensure the best optimal configuration is found for given cell size and composition. We present a measure of randomness based on Warren-Cowley short-range order parameters allowing for fast analysis of atomic structures. Hence, optimal structures are found in a reasonable time for several dozens or even hundreds of atoms. Both SQS optimizations and analysis of structures can be carried out via a command-line interface or a Python API. Additional features, such as optimization towards partial ordering or independent sublattices allow the generation of atomistic models of modern complex materials. Moreover, hybrid parallelization, as well as distribution of vacancies, are supported. The output data format is compatible with ase, pymatgen and pyiron packages to be easily embeddable in complex simulation workflows.本研究提出一款用于高效生成特殊准随机结构（Special Quasi-Random Structures，SQS）的Python软件包，可用于无序体系的原子尺度计算。该软件包支持采用蒙特卡洛（Monte-Carlo）方法或系统的结构枚举方式执行优化操作，以确保在给定晶胞尺寸与组分下找到最优原子构型。我们提出了一种基于沃伦-考利短程有序参数（Warren-Cowley short-range order parameters）的随机性度量方法，可实现原子结构的快速分析。因此，针对包含数十乃至数百个原子的体系，可在合理时间内获得最优结构。特殊准随机结构优化与结构分析均可通过命令行界面或Python应用程序编程接口（Application Programming Interface，API）实现。此外，该软件包还具备多项附加功能，例如针对部分有序或独立亚晶格的优化，可用于生成现代复杂材料的原子级模型。不仅如此，该软件包还支持混合并行计算以及空位分布功能。其输出数据格式与ase、pymatgen及pyiron软件包兼容，可轻松嵌入复杂的模拟工作流中。