Prof Peter Knowles - MA PhD CChem FRSC FLSW
- The development of new approximations and computational methods for improving the accuracy and reliability of first-principles molecular electronic structure.
- The implementation of ab initio methods for large molecules, including linear-scaling methodology, and hybrid embedding methods.
- High-performance computing, including parallel computing, and the development of interoperating chemistry components for the Grid.
Computation — whether from first principles or through simple models — has in recent years emerged as an equal partner of experiment in elucidating the structure, energetics and reactivity of materials. Our research efforts are focused on applying theory, through computation, to the prediction of the electronic structure of molecules, which determine molecular properties and the forces between atoms.
The key physical effect that presents a challenge for accurate representation is the correlated motion of electrons that is absent in the standard molecular-orbital picture of electronic structure. We are developing mathematical methods that support both the long-range correlation of electrons that is manifest when chemical bonds break, and the short-range correlated motion of two nearby electrons. A key aspect of such work is providing practical computer implementations that exploit available computational resources to the full, and that can be used by other researchers. Thus, much of the work of our group is directed towards the development of computer software. Challenges include computational methods that are applicable to large molecules, and the effective exploitation of high-performance parallel computers with thousands of processors. An important feature of this endeavour is the widespread distribution of our software package (Molpro) to the chemical community.