Tools for atomic-scale model construction: Cluster expansions and force constants

Modeling the thermodynamic and kinetic properties of materials commonly requires taking into account both vibrational and compositional degrees of freedom. While electronic structure calculations strictly speaking can provide the necessary information, they are computationally too demanding for many practical purposes. One therefore requires models that can reproduce the relevant parts of the potential energy landscape at a much smaller computational cost. Here, we will present a set of software packages that we developed over the course of the last years that allow one to rapidly construct and sample models that map vibrational (hiphive [1]) and/or compositional degrees of freedom (icet [2]). These tools can be integrated in a homogeneous Python based workflow, which combines structure generation, electronic structure calculations, model construction, thermodynamic sampling, and subsequent analysis. Example applications include the prediction of alloy phase diagrams, temperature driven phase transitions, thermal conductivities, and ordering phenomena.

References:

1. F. Eriksson, E. Fransson, and P. Erhart, Adv. Theory Simul. 1800184 (2019). doi:10.1002/adts.201800184 https://hiphive.materialsmodeling.org/
2. M. Ångqvist, W. A. Muñoz, J. M. Rahm, E. Fransson, C. Durniak, P. Rozyczko, T. H. Rod, and P. Erhart, Adv. Theory Simul. 1900015 (2019). doi:10.1002/adts.201900015 https://icet.materialsmodeling.org/