Principal Component Analysis for Molecular Dynamics Simulation of the Crystal Phase Transition

抄録

Principal component analysis (PCA) which is one of the dimensionality reduction techniques is adapted to find the common or major factors from sampling big data. In order to be clear the trigger motion of phase transition/change, PCA performed the MD sampling coordinates of crystal structure around phase transition temperature. In this paper, we especially focus on the α − β transition using the PCA analysis to the MD trajectories. The first principal component of PCA can be approximated as a mixed motion of the rotational mode of one SiO₃-OSiO₃ bond and the bending mode of Si-O-Si angle. The result indicated that the PCA can show the trigger motion of phase transition/change as found in previous studies. We could suggest that the PCA is useful tool to understand the importance information of phase transfer/change of crystal structure.