| This thesis describes new methods for incorporating quantum electronic structure information into classical molecular dynamics simulations.;In the first part, we discuss an adaptation of the Car-Parrinello parallel molecular dynamics method that uses the Hartree-Fock approximation and an atom centered basis set, rather than the density functional method with a plane wave basis. This allows us to compute the forces on the nuclei using an extensible approximation based on the electronic structure of the studied system, as opposed to a more traditional empirical internuclear potential function. We find that the utility of the method is limited by problems that arise in the computation of the forces on the nuclei. The problems are related to the orbital orthonormality constraints.;Various ways of circumventing these problems are described, including using Hellmann-Feynmann forces and removing exact orthonormality constraints and replacing them with a penalty function. Examples comparing the speech of the Hartree-Fock parallel MD method to converging the wave-function each MD timestep are discussed.;The second part deals with new conjugate gradient methods for electronic wave function minimization in variational calculations of the ground state wave function within the Hartree-Fock approximation. Our first section deals with a generalization of a method previously applied in density functional theory that optimizes the wave function one orbital at a time. When applied to Hartree-Fock calculations with atom centered basis sets, the method is found in general to be noncompetitive with other optimization techniques. The second section describes an augmented Lagrangian method for energy minimization, for which we have several methods for computing the Lagrange multipliers. The method is found to be extremely sensitive to errors in the multipliers. We discuss the optimization of various parameters in the method. Overall, however, this method is also found to be noncompetitive in our systems as compared to other minimization methods.;In the final section, we describe the group of computer codes we have developed and apply them to some test problems. |
