AMP2: A fully automated program for ab initio calculations of crystalline materials

Ab initio calculations based on the density functional theory (DFT) become a vital tool in material science for understanding and predicting material properties. However, DFT calculations involve several parameters and procedures that call for deep understanding on underlying theories and preceding knowledge on certain properties of target materials. Such technicalities cost a significant amount of human time and expose practitioners to mistakes. Here, we introduce a fully automated package for DFT calculations, automated ab initio modeling of materials property package (AMP2), which aims to produce key DFT properties of crystalline materials with essentially no user intervention except for initial structural information. This is achieved through algorithms that automatically determine various technical parameters and make self-decisions during computational workflow. As results, AMP2 is material-agnostic and provides a highly accurate band structure, band gap, effective mass, density of states and dielectric constant for the given material. Notably, the package finds the antiferromagnetic ground state by applying a genetic algorithm to effective Ising models. AMP2 also addresses band-gap underestimation in semilocal functionals with help of a hybrid functional, thereby producing a more accurate band gap, even if the material turns out to be metallic within the semilocal functional. We believe that the present package will significantly expand usage of DFT calculations by making them more accessible.基于密度泛函理论（Density Functional Theory，DFT）的从头算（ab initio）计算，现已成为材料科学领域中理解与预测材料物性的核心工具。然而，DFT计算涉及诸多参数与流程，要求研究者深度掌握其底层理论，同时需具备目标材料特定物性的前置知识储备。此类技术细节会耗费大量人力时间，且易使从业者出现操作失误。在此，我们推出一款面向DFT计算的全自动化工具包——材料物性从头算自动化建模工具包（AMP2），该工具包旨在仅需用户提供初始结构信息，即可在几乎无需人工干预的前提下，计算得到晶体材料的关键DFT物性。这一功能通过内嵌算法实现：该算法可在计算流程中自动确定各类技术参数，并自主完成决策环节。由此，AMP2具备材料无关性，可为目标材料提供高精度的能带结构、带隙、有效质量、态密度以及介电常数等物性参数。值得注意的是，该工具包通过将遗传算法应用于有效伊辛模型，可自动搜索反铁磁基态。此外，AMP2借助杂化泛函，可修正半局域泛函计算中存在的带隙低估问题；即使目标材料在半局域泛函框架下被判定为金属，也能得到更为准确的带隙值。我们相信，本工具包将通过降低DFT计算的使用门槛，大幅拓展其应用场景。