Large-scale ab initio quantum chemistry: Selected applications and algorithmic development for parallel computers.

摘要

The present thesis is concerned with two different aspects of large-scale ab initio quantum chemistry: Part I considers reformulation and new efficient computational implementations of ab initio methods with the purpose of making these methods applicable to larger molecular systems, and Part II comprises applications of high-level ab initio quantum chemistry methods in the study of chemical reactions and molecular potential energy surfaces.;In Part I the conventional formulation of Moller-Plesset (MP) perturbation theory is reviewed along with new formulations of second-order MP theory (MP2) which reduce the computational complexity of the method. The MP2 method plays an important part in ab initio quantum chemistry because of its relative computational simplicity and its ability to accurately predict, e.g., energetics, molecular structures and vibrational frequencies. New algorithms are presented for computing energies and gradients on parallel computers with the MP2 method and other perturbation methods. The algorithms have been implemented on massively parallel supercomputers and allow calculations to be performed for larger systems by reducing computational times by up to several orders of magnitude and by reducing the requirements for computer memory.;Part II comprises high-level ab initio studies of the ground state potential energy surface of silicon dicarbide (SiC