Abstract: 

We present the idea of multi-index Monte Carlo methods in the frame of  PDEs with random coefficients. We show error estimates and prove that the multi-index approach can be more efficient than the multi-level approach. We give examples which connect the theory with fast random field generation by use of H-matrices.

 

Speaker
Michael Feischl
Research Area
Computational Maths
Affiliation
UNSW, Sydney.
Date
Tue, 21/11/2017 - 1:05pm
Venue
OMB-G32, Old Main Building, UNSW