PROJECTS

Development of electron structure quantum Monte Carlo methods

PRINCIPAL INVESTIGATOR

PROJECT RESEARCH ASSISTANT

ABSTRACT

The diffusion Monte Carlo method (DMC) is capable of solving the full 3n-dimensional electronic Schrödinger equation numerically even for larger molecules. Stabilization energies, reaction energies, and excitation energies among other properties can be calculated from first principles. The most efficient variant is exact (with statistical error only) within boundary conditions given implicitly by the nodes of a trial function which is usually an ab initio wave function such as a Hartree-Fock or Kohn-Sham Slater determinant. The results of the DMC method depend only on the quality of the nodal hypersurface of the trial wave function. While many excellent results have been obtained, the method lacks a way of systematically improving the nodes – and thus the DMC energy – of the trial function. This proposal aims at constructing trial wave functions with optimal nodes through the use of global and local criteria of the accuracy of the nodes. With optimized nodes, an increased accuracy of the method should be obtained without considerably more computational effort.