Accuracy and computational efficiency of real-time subspace propagation schemes for the time-dependent density functional theory

Arthur Russakoff; Yonghui Li; Shenglai He; Kalman Varga

doi:10.1063/1.4952646

Accuracy and computational efficiency of real-time subspace propagation schemes for the time-dependent density functional theory

J Chem Phys. 2016 May 28;144(20):204125. doi: 10.1063/1.4952646.

Authors

Arthur Russakoff¹, Yonghui Li¹, Shenglai He¹, Kalman Varga¹

Affiliation

¹ Department of Physics and Astronomy, Vanderbilt University, Nashville, Tennessee 37235, USA.

PMID: 27250297
DOI: 10.1063/1.4952646

Abstract

Time-dependent Density Functional Theory (TDDFT) has become successful for its balance of economy and accuracy. However, the application of TDDFT to large systems or long time scales remains computationally prohibitively expensive. In this paper, we investigate the numerical stability and accuracy of two subspace propagation methods to solve the time-dependent Kohn-Sham equations with finite and periodic boundary conditions. The bases considered are the Lánczos basis and the adiabatic eigenbasis. The results are compared to a benchmark fourth-order Taylor expansion of the time propagator. Our results show that it is possible to use larger time steps with the subspace methods, leading to computational speedups by a factor of 2-3 over Taylor propagation. Accuracy is found to be maintained for certain energy regimes and small time scales.