A divide-and-conquer implementation of the discrete variational DFT method for large molecular and solid systems

Warschkow, Oliver (1999) A divide-and-conquer implementation of the discrete variational DFT method for large molecular and solid systems. University of Southampton, Doctoral Thesis.

Record type: Thesis (Doctoral)

Abstract

A novel density functional theory (DFT) code is described that implements Yang's divide-and-conquer approach in the framework of the discrete variational method (DVM). The limitations of the DVM embedded cluster model were explored, and it is argued that, as a natural extension of this model, a superior scheme would describe a large system not by a single large quantum calculation but by many smaller, overlapping and mutually interacting clusters instead. This concept, being the essence of Yang's divide-and-conquer technique, was combined with the discrete variational approach towards solving the DFT equations and implemented into a novel, general-purpose computer code. The primary purpose of this software is the rapid computation of approximate electron densities and density of states for a given arrangement of atoms. By using moderately sized grids and compact basis and density fit sets, a high degree of efficiency is achieved. Algorithmic details of various sub-tasks of the method are discussed. Benchmark calculations on two systems - linear alkane chains and globular proteins - demonstrate that the computational performance of the method scales linearly with respect to system size for up to more than 1000 atoms. Calculations on some example structures demonstrate the convergence of calculated properties with respect to the size of the divide-and-conquer fragments. These calculations also illustrate the many potential fields of application of this new code to systems in both the molecular and solid state.

This record has no associated files available for download.