Revisiting and parallelizing SHAKE
Y Weinbach and R Elber, JOURNAL OF COMPUTATIONAL PHYSICS, 209, 193-206 (2005).
An algorithm is presented for running SHAKE in parallel. SHAKE is a widely used approach to compute molecular dynamics trajectories with constraints. An essential step in SHAKE is the solution of a sparse linear problem of the type Ax = b, where x is a vector of unknowns. Conjugate gradient minimization (that can be done in parallel) replaces the widely used iteration process that is inherently serial. Numerical examples present good load balancing and are limited only by communication time. (c) 2005 Elsevier Inc. All rights reserved.
Return to Publications page