Neuroevolution machine learning potentials (NEP MLPs) for accurately and efficiently sampling short-range order (SRO) of GeSn alloys using large-scale atomistic simulations

Two NEP MLPs for GeSn alloys trained based on different training datasets

• NEP_MLP_GeSn_Full_V1.txt is a NEP MLP trained based on a training dataset comprising 276009 structures (17664576 atoms), 90% of a full training dataset of 306677 structures (19627328 atoms).

• NEP_MLP_GeSn_Active_V1.txt is a NEP MLP trained based on a active learning (farthest point sampling) training dataset comprising 137 structures (8768 atoms).

Reference

S. Chen, X. Jin, W. Zhao, and T. Li, "Intricate short-range order in GeSn alloys revealed by atomistic simulations with highly accurate and efficient machine-learning potentials", submitted.