Exploring new computing paradigms in theoretical chemistry

Department of Physical Chemistry, Indian Association for the Cultivation of Science, Jadavpur,

Kolkata-700 032, India

Department of Chemistry, Indian Institute of Technology, Powai, Mumbai-400 076, India

E-mail : pcspb@chem.iitb.ac.in

Manuscript received 25 March 2013, accepted 26 March 2013

Computing in theoretical chemistry has been largely and traditionally based on purely numerical 'non-intelligent' computing techniques. The tools of 'Artificial Intelligence' (AI) or 'Computational Intelligence' have been little explored and exploited in the context of research in theoretical chemistry. Over the last decade and a half we had been experimenting with 'evolutionary computing techniques' like the Genetic Algorithms and Random Mutation Hill Climbing in the general context of computing electronic structure of atoms and molecules. These methods have the underpinning of certain microscopic low-level biological processes and are supposed to be endowed with 'Artificial Intelligence'. We trace the evolution of the AI-based techniques developed by us and review some of the rather non-trivial applications. In particular, we focus on an Adaptive Random Mutation Hill Climbing (ARMHC) method for locating global minima on the complex potential energy landscapes of 3-D Coulomb clusters and assessing the possibilities of structural phase transitions in them. Possible directions of future developments are indicated.