Electronic Structure of Large Molecules using Parallel Computers

Modern methods of drug discovery rely heavily on computer modeling of potential therapeutic agents to calculate their geometric and electronic properties and optimize them to interact specifically with target receptors with high potency. This process is extremely compute intensive due to the large sizes of the molecular Systems involved. Recently, new linear-scaling algorithms based on fast multipole methods (FMM) have been introduced for computing the electronic structures of such large systems. Energies, forces, and force constants can now be computed efficiently even on workstations. However, study of biological systems requires new linear-scaling methods capable of modeling solvation, computing additional important properties, such as excitation energies, and optimizing geometry. Moreover, to enable routine application, it is necessary to exploit highly parallel computer systems. We propose to address all of these issues by investigating parallel linear-scaling methods suitable for drug design. In Phase I we will develop prototype parallel versions of FMM-based methods for basic electronic structure calculations and geometry optimization, and we will design algorithms for calculating solvation effects and other properties, such as the lowest-unoccupied and highest-occupied molecular orbitals. Phase II will include implementation of the latter algorithms along with significant extension, refinement, and testing of all the proposed methods. BENEFITS: This project will lead to enhanced software that will dramatically improve the ability of researchers in the DOD, universities, and commercial industry to carry out the kinds of modeling required for advanced drug discovery and other molecular analysis of biological systems. Since the software will be based on widely used commercial computing and chemistry technologies, it will have broad appeal and be readily accessible to research groups involved in the design of future medicinal and agricultural products as well as to material scientists designing new alloys, propellants, and explosives.