Speaker
Yu Xie
State Key Lab of Superhard Materials, College of Physics, Jilin University, Changchun, P. R. China
International Center for Computational Method & Software, College of Physics, Jilin University, Changchun, P. R. China
International Center of Future Science, Jilin University, Changchun, P. R. China
Date＆Time
Thursday, Oct. 8, 2020 at 16:15（CST）
Location
Join the seminar：
https://us02web.zoom.us/j/86588958999?pwd=aHlhd3hvN1Q3dVgvcGlVRUQxK0NBdz09
Join Zoom Meeting:
Meeting ID: 865 8895 8999 | Password: NOMAD
Abstract
Ab initio electronic structure calculations within Kohn-Sham density functional theory requires a solution for the Kohn–Sham equation. However, the traditional self-consistent field (SCF) approach of solving the equation using iterative diagonalization exhibits an inherent cubic scaling behavior and becomes prohibitive for large systems. The Chebyshev-filtered subspace iteration (CheFSI) method holds considerable promise for large-system calculations by substantially accelerating the SCF procedure. Here, we employed a combination of the real space finite-difference formulation and CheFSI to solve the Kohn–Sham equation, and implemented this approach in ab initio Real-space Electronic Structure (ARES) software in a multi-processor, parallel environment. An improved scheme was proposed to generate the initial subspace of CheFSI in ARES efficiently, making it suitable for large-scale simulations. We also proposed a new algorithm for the generalized eigenvalue problem, which can be easily extended the CheFSI method to the ultrasoft pseudopotentials.
