APPROACHES TO LARGE-SCALE PARALLEL SELF-CONSISTENT-FIELD CALCULATIONS

The availability of massively parallel computers with high computation rates but limited memory and input/output bandwidth prompts the reeva luation of appropriate solution schemes for the self-consistent field (SCF) equations. Several algorithms are considered which exhibit betwe en linear and quadratic convergence using various approximations to th e orbital Hessian. A prototype is developed to understand the computat ional expense of each approach. The optimal choice is found to be a co njugate-gradient method preconditioned with a level-shifted approximat ion to the orbital Hessian. This is a compromise between efficiency, s tability, and low memory usage. Sample benchmarks on two parallel supe rcomputers are also reported. (C) 1995 by John Wiley & Sons, Inc.