Pyiron—Simulation Workflows for the Design of Sustainable Materials

One aspect of designing sustainable materials is to reduce the reliance on rare earth elements. A critical step in achieving this is to have efficient and accurate tools to screen the entire periodic table for potential candidates. We have therefore developed automated simulation workflows that predict thermodynamic properties with ab-initio accuracy and allow quantitative comparison with experiments. These workflows combine a wide range of different tools and utilities, with a recent focus on machine-learned interatomic potentials. This typically starts with the generation of atomistic structures, continues with the evaluation of these structures with an ab-initio reference method such as Density Functional Theory (DFT), the fitting of the interatomic potential and finally the validation of the resulting potential.<br/><br/>To maintain the provenance of the individual steps in these workflows and guarantee the reproducibility of the development process of interatomic potentials, we developed pyiron an integrated development environment (IDE)[1, 2] for the development and assessment of interatomic potentials. Rather than implementing its own DFT simulation code or interatomic potential fitting code, the pyiron IDE combines existing codes developed in the community and provides one unified high-level interface to access these codes in an generic way. To highlight the features of pyiron, we present a specific workflow that combines the VASP DFT simulation code, the FitSNAP interatomic potential fitting code, and the LAMMPS molecular dynamics to develop and validate an interatomic potential.<br/><br/>[1]: http://pyiron.org<br/>[2]: J. Janssen, et al., Comp. Mat. Sci. 161 (2019)