Peter Pulay of the University of Arkansas is supported by the Theoretical and Computational Chemistry Program to develop methods using alternatives to the generally employed Gaussian basis sets in electronic structure calculations. Plane waves, wavelets, Slater functions, and semi-numerical orbitals will be explored for their potential implementation in chemical applications. A Gaussian-based local electron correlation method will be developed, with goals of reduction of the local basis size for weakly interacting orbital pairs by using pseudo-natural orbitals, inclusion of gradients, and programming of high-level correlation methods for selected localized parts of a molecule. Application of these improved techniques include the structures and vibrational spectra of metalloprotein models. Also, accurate hydrogen bond potential functions based on correlated calculations for hydrocarbons will be developed. It is anticipated that the new methods developed in this project will significantly improve computational modeling of chemical and biochemical systems.
The research in this project takes advantage of the rapid technological developments in the computer industry in the past five years. The dramatic expansion in fast memory and disk capacity, in particular, makes it possible to offer viable alternatives to the computational methods presently used in chemistry and biochemistry. Successful outcomes from this project could change the way computational chemistry is done, with substantial industrial impacts in areas such as pharmaceutical chemistry and materials science.