Research on tally data decomposition algorithms based on reactor Monte Carlo code RMC

The applications of Monte Carlo method in reactor physics analysis are restricted due to excessive memory demand in solving large-scale problems, while data decomposition is supposed to be a remedy. In this paper quantitative memory requirements in MC simulation are analyzed. Two types of tally data decomposition algorithms, which utilize synchronous and asynchronous communications, are designed and implemented in Reactor Monte Carlo code (RMC). Numerical tests are carried out to evaluate performance of new algorithms. It is shown that tally data decomposition algorithm can reduce memory size effectively while parallel efficiency of the code is not affected. (authors)