Design of New Materials Using Computational Chemistry and Materials Science (CHSSI)

Abstract

Abstract : The objective of this work was to develop a broad array of methods for electronic structure theory, with special emphasis on improved parallel codes. A new highly scalable code for frozen core second order perturbation theory (MP2) gradients for closed and open shell molecules has been developed and is on-line for general use at several IBM SP2 and Cray T3E systems. The key element of this approach is our distributed data interface (DDI) that facilitates distribution of very large arrays across all available nodes. This code has been ported to the cluster environment, so it can now be used on cluster of PCs or Macs running Linux, as well as on clusters of high performance workstations. The development of analogous codes for molecules with unpaired electrons has been completed and is in the most recent GAMESS release. The derivation for the preferred restricted open shell ZAPT method is complete and a paper describing this method has been published. The corresponding code is now in GAMESS. The code for implementing Hartree-Fock with DDI has been completed, and a paper describing this code has been published. We are now turning to analogous developments for MCSCF and CI codes. A scalable code for the multi-reference second order perturbation theory has been completed, a paper published, and the code is in GAMESS. Other important developments include: the derivation of gradients for multi-reference second order perturbation theory, further developments of our effective fragment potential (EFP) method for studying solvation and liquid behavior, the development of molecular dynamics and Monte Carlo methods to facilitate the study of solvation and liquid behavior, the development and implementation of a new method for producing global energy surfaces from sets of ab initio points, the development and implementation of both grid and gridless approaches to destiny functional theory.