Fig. 2
From: Predicting phase behavior of grain boundaries with evolutionary search and machine learning
Evolutionary search and clustering analysis identify GB phases. The evolutionary search and clustering analysis identify three grain boundary phases of Σ5(210)[001]. The evolutionary algorithm explores different atomic densities and identifies multiple grain boundary phases: a Kites, Split Kites, and Filled Kites. The three phases correspond to the energy minima as a function of number of atoms. b Energy of grain boundary configurations generated by the evolutionary search as a function of number of atoms. d, e The generated structures are automatically clustered into three grain boundary phases according to similarities in their excess properties. c Grain boundary energy plot same as in b with data points colored according to the clustering
