"Understanding is a Three-Edged Sword"
My main research interest at the moment is the determination of crystal structure using genetic algorithms.
I initially wrote a program to calculate the structure of Lennard-Jones clusters using the Deaven and Ho (PRL 75 288 (1995)) Genetic Algorithm. Now I am working on a new periodic crossover operation which is suitable for use in bulk and surface systems. This method also allows different parents to have different cells during the calculation.
Genetic Algorithms were first proposed by John Holland in 1975 in his book Adaption in Natural and Artificial Systems (a 1992 reprint is available, ISBN 0-262-58111-6) and use natural selection to search for a minimum configuration. A good review article on this subject is by R. L. Johnston (Dalton Trans., pages 4193-4207, 2003) and applies specfically to real-space cluster calculations.
Application to Periodic Systems
A poster which has some of the details of our new method is available for download, as is a talk that I presented on our method at the 2006 APS March Meeting in Baltimore, which can be found in the talks section of my downloads page. A paper which gives technical details of this method has been accepted for publication in Physical Review B. A preprint of which can be downloaded from arXiv.org at cond-mat/0605066.