Abstract
Algorithms for the symmetry-adapted energy minimisation of solids using analytical first and second derivatives have been devised and implemented in a new computer program GULP. These new methods are found to lead to an improvement in computational efficiency of up to an order of magnitude over the standard algorithm, which takes no account of symmetry, the largest improvement being obtained from the use of symmetry in the generation of the hessian. Accelerated convergence techniques for the dispersion energy are found to be beneficial in improving the precision at little extra computational cost, particularly when a one centre decomposition is possible or the Ewald sum weighting towards real-space is increased.