| Issue |
2014
SNA + MC 2013 - Joint International Conference on Supercomputing in Nuclear Applications + Monte Carlo
|
|
|---|---|---|
| Article Number | 04201 | |
| Number of page(s) | 8 | |
| Section | 4. Advanced Parallelism and HPC Strategies: b. Monte Carlo Methods, Parallelism and HPC | |
| DOI | https://doi.org/10.1051/snamc/201404201 | |
| Published online | 06 June 2014 | |
Implementation of Domain Decomposition and Data Decomposition Algorithms in RMC code
Department of Engineering Physics, Tsinghua University, Beijing, 100084, China
* Corresponding Author, Tel.: +8610-62783087, Fax: +8610-62782658 E-mail: wangkan@tsinghua.edu.cn (Kan Wang)
The applications of Monte Carlo method in reactor physics analysis is somewhat restricted due to excessive memory demand in solving large-scale problems. Domain decomposition and data decomposition are two remedies. Memory demand in MC simulation is analyzed firstly. Domain decomposition and tally data decomposition algorithms separately are designed and implemented based on reactor Monte Carlo code RMC. Numerical tests are executed to evaluate performance of new algorithms respectively. Domain decomposition algorithm shows potentials to speed up MC simulation as a space parallel method. As for tally data decomposition algorithms, memory size is reduced obviously.
Key words: domain decomposition / data decomposition / memory problem / Monte Carlo
© Owned by the authors, published by EDP Sciences, 2014