Improved parallelization techniques for the density matrix renormalization group
arXiv:0909.5662 · doi:10.1016/j.cpc.2010.03.018
Abstract
A distributed-memory parallelization strategy for the density matrix renormalization group is proposed for cases where correlation functions are required. This new strategy has substantial improvements with respect to previous works. A scalability analysis shows an overall serial fraction of 9.4% and an efficiency of around 60% considering up to eight nodes. Sources of possible parallel slowdown are pointed out and solutions to circumvent these issues are brought forward in order to achieve a better performance.
8 pages, 4 figures; version published in Computer Physics Communications