Machine-learning potentials for structurally and chemically complex MAB phases: strain hardening and ripplocation-mediated plasticity

The data coresponds to the publication Machine-learning potentials for structurally and chemically complex MAB phases: strain hardening and ripplocation-mediated plasticity , by Nikola Koutná , Shuyao Lin , Lars Hultman , Davide G. Sangiovanni , Paul H. Mayrhofer accessible at doi.org/10.1016/j.matdes.2025.114307 Methodology The methods used to produce the data are described in the publication Contents The zip file contains a README file and 3 folders with various text files: MABs_structures : relaxed structures in the VASP POSCAR format ( vasp.at/wiki/index.php/POSCAR ) MLIPs : machine-learning interatomic potentials in the mlip-2 format ( gitlab.com/ashapeev/mlip-2 ) and the corresponding training sets in the cfg format (compatible with the mlip-2 package) Raw_data_from_tables : calculated lattice parameters, elastic constants, and mechanical properties, as listed in Tab.1-3 in the publication ( doi.org/10.1016/j.matdes.2025.114307 )