Premium
MLMC method to estimate propagation of uncertainties in electromagnetic fields scattered from objects of uncertain shapes
Author(s) -
Litvinenko Alexander,
Yucel Abdulkadir,
Bagci Hakan,
Oppelstrup Jesper,
Michielssen Eric,
Tempone Raúl
Publication year - 2021
Publication title -
pamm
Language(s) - English
Resource type - Journals
ISSN - 1617-7061
DOI - 10.1002/pamm.202000064
Subject(s) - monte carlo method , discretization , polygon mesh , fidelity , scattering , computer science , uncertainty quantification , domain (mathematical analysis) , hierarchy , algorithm , mathematical optimization , statistical physics , mathematics , physics , geometry , mathematical analysis , optics , machine learning , statistics , telecommunications , economics , market economy
We estimate the propagation of uncertainties in electromagnetic wave scattering problems. The computational domain is a dielectric object with uncertain shape. Since classical Monte Carlo (MC) method is too expensive, we suggest to use a modified multilevel Monte Carlo (MLMC) method. This method uses a hierarchy of spatial meshes and optimally balances the statistical and discretisation errors. MLMC performs most of the simulations using low‐fidelity models and only a few simulations using high‐fidelity models. As a result, the final computational cost is becoming significantly smaller.