Daisuke Sugita | Searcharxiv

arXiv2021

Robust and efficient calculation of activation energy by automated path search and density functional theory

Because inorganic solid electrolytes are one of the key components for application to all-solid-state batteries, high-ionic-conductivity materials must be developed. Therefore, we propose a method of efficiently evaluating the activation energy of ionic diffusion by calculating a potential-energy surface (PES), searching for the optimal diffusion path by an algorithm developed using dynamic programming (DP), and calculating the corresponding activation energy by the nudged elastic band (NEB) method. Taking beta-Li3PS4 as an example, the activation energy of Li-ion diffusion was calculated as 0.43, 0.25, and 0.40 eV in the a-, b-, and c-axis directions, respectively, which is in good agreement with previously reported values. By comprehensively searching for the lowest energy path by PES-DP, the arbitrariness of the path selection can be eliminated, and the activation energy must only be calculated using the NEB method a few times, which greatly reduces the computational cost required for evaluating activation energy and enables the high-throughput screening of solid state electrolytes.

cond-mat.mtrl-sci↗

arXiv2019

Active learning for enumerating local minima based on Gaussian process derivatives

We study active learning (AL) based on Gaussian Processes (GPs) for efficiently enumerating all of the local minimum solutions of a black-box function. This problem is challenging due to the fact that local solutions are characterized by their zero gradient and positive-definite Hessian properties, but those derivatives cannot be directly observed. We propose a new AL method in which the input points are sequentially selected such that the confidence intervals of the GP derivatives are effectively updated for enumerating local minimum solutions. We theoretically analyze the proposed method and demonstrate its usefulness through numerical experiments.

stat.ML↗