Quantum molecular modeling with simulated annealing-A distributed processing and visualization application

Abstract

A quantum-mechanical method for calculating the geometrical structure of small molecules is presented. A combination of the density functional method with finite-temperature molecular dynamics and simulated annealing techniques provides a reliable scheme for sampling energy surfaces, without being trapped in energetically unfavorable local minima. Its value and efficiency are demonstrated by computing previously unknown geometries of small clusters composed of phosphorous, sulphur, selenium and oxygen atoms. The implementation combines distributed processing with real-time visualization and allows instantaneous interaction with the simulation. The computationally demanding quantum mechanics problem is solved on a Cray supercomputer, communicating with a Silicon Graphics Incorporated (SGI) workstation in a UNIX environment using SGI's distributed graphics library (DGL).<>