Presentation Information
[8p-P11-3]Machine Learning of Hamiltonian Matrices by Linear Combination of Rotation-Equivariant Matrices for Rapid Quantum Chemistry Calculations
〇(B)Jintaro Endo1, Koki Ozawa1, Hiroyuki Matsui1 (1.ROEL, Yamagata Univ)
Keywords:
Materials Informatics,Hamiltonian Matrix,Quantum Chemistry Calculations
In recent years, research on the prediction of Hamiltonian matrices using machine learning has been actively conducted. In this study, we propose a method for predicting Hamiltonian matrices by expressing them as a linear combination of five rotationally equivariant matrices: the overlap matrix S, its square S2, the kinetic energy matrix K, the nuclear-electron attraction matrix V, and the identity matrix I. The coefficients of this linear combination are predicted by a machine learning model.