Determining the structures of materials at the atomic level is fascinating and important. We have discovered the structures of new materials using density-functional-theory calculations and our ab initio random structure searching method. Some of these materials have been synthesised experimentally. This approach has been very successful. It is a fast moving field with many possibilities, and therefore there is no point in providing a list of the materials that we might study. (In collaboration with Prof Chris Pickard, Department of Materials Science, Cambridge)
The anharmonicity of atomic vibrations often becomes important at high temperatures, for light atoms, when weak bonding is present, near phase transitions, etc. We have developed methods in which the low-energy part of the Born-Oppenheimer surface is mapped out using density-functional-theory methods. We calculate the coupling between vibrations and other observables such as band gaps. When strong coupling between modes is present we face a many-body problem, which we solve using Monte Carlo methods. We can calculate properties such as the temperature dependence of the electronic bandstructure, phonon dispersion curves, heat capacity, the stress tensor, infra-red and Raman intensities, and the NMR shielding. We often combine vibrational calculations with structure searching, and sometimes also with our quantum Monte Carlo approaches.
We have developed the widely-used CASINO code for performing quantum Monte Carlo calculations for many-body systems. Diffusion quantum Monte Carlo is the most accurate method known for calculating the energies of large assemblies of interacting quantum particles. These calculations are subject to the so-called "localization error", which is often the largest error in the calculations. In this project we will develop a new approach for greatly reducing localization errors.