Advanced Sensitivity Analysis of the Danish Eulerian Model in Parallel and Grid Environment
Sobol sequence
IBM
DOI:
10.1063/1.3659924
Publication Date:
2011-12-01T09:26:14Z
AUTHORS (7)
ABSTRACT
A 3‐stage sensitivity analysis approach, based on of variances technique for calculating Sobol's global indices and computationaly efficient Monte Carlo integration techniques is considered applied to a large‐scale air pollurion model, the Danish Eulerian Model. On first stage it necessary carry out set computationally expensive numerical experiments extract data. The output used construct mesh‐functions ozone concentration ratios be in next stages evaluating variances. Here we use specially adapted purpose version called SA‐DEM. It has been successfully implemented run most powerful parallel supercomputer Bulgaria—IBM Blue Gene/P. more advanced version, capable using efficiently full capacity this supercomputer, described paper, followed by some performance experiments. Another source computational power solving such tuff problem grid. That why another SA‐DEM exploit our Grid infrastructure. results from both implementation are presented, compared analysed.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (0)
CITATIONS (19)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....