Probabilistic methods applied to 2D electromagnetic numerical dosimetry
Résumé
Probabilistic approaches are performed on electromagnetic numerical dosimetry problems in order to take into account the variability of the input parameters. These approaches are based on an expansion of the random parameters in two different ways: a spectral description and a nodal description. Compared to the Monte Carlo method, these methods are attractive since they exploit determinist codes in a more efficient way. The number of calculations can be further reduced using a regression technique, sparse grids computed from Smolyak's algorithm or a suited coordinate system. It is shown in a simple scattering problem that only 100 calculations are required applying these methods while the Monte Carlo method uses 10,000 samples.
Origine : Fichiers produits par l'(les) auteur(s)