A comparison of the FW-CADIS and MR-CADIS variance reduction methods

Research output: Contribution to journalConference articlepeer-review

1 Scopus citations

Abstract

For some problems with a small number of responses, the MR-CADIS method may be slightly faster than the FW-CADIS method. For global or semi-global problems or for problems with many tallies, the time required to make the variance reduction parameters using the MR-CADIS method may be so large as to lower the overall FOM of the problem. The TN24P problem highlighted the difficulty of determining how many adjoints are needed and how they are selected. Clearly, using 2 x 10s adjoint calculations would not be reasonable, but it is not certain how many calculations would be reasonable. An increase from three to nine adjoints in the MR-CADIS method improved the Monte Carlo performance some, but overall performance decreased. For global problems with the goal of optimizing the calculation of the space and energy flux <p(r,E), would the MR-CADIS method require an adjoint for every combination of energy group and voxel of the tally? For the TN24P problem using 19 energy groups and 2 x 10s voxels, would 3.8 million adjoint calculations be required? FW-CADIS can make variance reduction parameters for the optimization of (p(r,E) over many groups and large volumes using just two deterministic calculations.

Original languageEnglish
Pages (from-to)540-543
Number of pages4
JournalTransactions of the American Nuclear Society
Volume116
StatePublished - 2017
Event2017 Transactions of the American Nuclear Society, ANS 2017 - San Francisco, United States
Duration: Jun 11 2017Jun 15 2017

Fingerprint

Dive into the research topics of 'A comparison of the FW-CADIS and MR-CADIS variance reduction methods'. Together they form a unique fingerprint.

Cite this