[16a-M_B07-1~9] 9.1 Dielectrics, ferroelectrics

11:30 AM - 11:45 AM JST(2:30 AM - 2:45 AM UTC)Presentation by Applicant for JSAP Young Scientists Presentation Award

[16a-M_B07-9] Development of simulation method for dielectric properties using GNN model

〇(M1)Koki Yoshimochi¹, Alex Kutana¹, Ryouji Asahi¹ (1.Nagoya Univ.)

Keywords:

dielectric constants,molecular dynamics,machine learning

In this study, we aimed to develop a dielectric permittivity prediction method based on a rotation-equivariant graph neural network model, and validated the approach using CaTiO₃ as a test system. By optimizing the sampling methods and accounting for nonlinear effects with respect to the applied electric field, high-accuracy evaluation of temperature-dependent dielectric permittivity was achieved. The changes in dielectric permittivity and its anisotropy associated with phase transitions in CaTiO₃ were analyzed in terms of the crystal structure and the Born effective charge tensor. This work demonstrates applicability of the proposed method for finite-temperature dielectric response in a large-scale materials system.