Adaptation of asymptotic close-coupling methods to massively parallel computers Original Research Article

A standard asymptotic close-coupling method based on a perturbation technique is adapted for use on massively parallel computers. The calculation of K-matrices and cross-sections is partitioned over the processors according to incident electron energy. Use of external disk during the calculation is eliminated. Linear scaling with the number of processors is achieved for a 128-node Intel iPSC/860 and a small farm of Sun Sparc 10 workstations.