NSF Award
9561833
This SBIR Phase I project will result in a multiconfigurational density functional approach to electronic structure problems that combines ab initio generalized valence bond (GVB) or restricted configuration interaction (RCI) techniques with density functional theory (DFT) methods. The GVB DFT and GVB RCI DFT software will use the pseudospectral techniques of this firm's PS GVB software to allow rapid calculations without loss of accuracy. In Phase I, this firm will write the software necessary to implement the proposed GVB DFT and GVB RCI DFT methods in PS GVB's pseudospectral framework, examine results from various functionals and hybrid parameters, and use the methods to perform some preliminary calculations of molecular properties to illustrate its feasibility, speed, and accuracy. Extensions of Becke's half half and three parameter methods that employ densities from GVB and GVB RCI wave functions in the density functional evaluation of the correlation energy will be examined, and various parameters for the three parameter methods will be tested. Calculations of various properties of large molecules will be performed using medium and large basis sets, including optimized molecular structures, bond dissociation energies, ionization potentials, proton affinities, solvation energies, multipole moments, polarizabilities, and hyperpolarizabilities. These results will be compared to experiment or to timing and accuracy obtained using less rapid or less accurate ab initio and DFT methods. An extensive market already exists for electronic structure codes, with researchers performing ab initio and DFT studies for chemical, biological, pharmaceutical, and materials applications. The extension of PS GVB's correlation methods to GVB DFT, GVB RCI DFT, and GVB RCI MP2 DFT techniques with the same efficient scaling already present for other pseudospectral methods would permit rapid calculation of chemically accurate dissociation energies, reaction barriers, conformational energy differences, force fields, and multiple other molecular properties. Consequently, this work will enhance the market for PS GVB significantly.
