Siegel der Universität Augsburg

Universität Augsburg
Institut für Mathematik

Siegel der Universität Augsburg


Oberseminar Numerische Mathematik


Dr. Michael Feischl
Karlsruher Institut für Technologie

spricht am
Montag, 5. Februar 2018
16 Uhr s.t.
Raum 1007 (L1)
über das Thema:

»Multi-index Monte Carlo and H-Matrices«

Abstract: We consider a new method to generate normal or log-normal random fields which builds on fast matrix-vector multiplication via H-matrices. The method proves to be robust with respect to the covariance length of the random field, and is particularly efficient for very smooth and very rough random fields. Moreover, the method applies to a fairly general class of covariance functions and is not limited to the stationary case. We use this new method in combination with Monte Carlo integration, to solve a Poisson equation with random coefficient. Moreover, to exploit the inherent sparsity of the approximation, and to obtain an efficient algorithm, we use the Multi-Index quasi-Monte Carlo approach in three coordinate directions: the finite-element approximation error, the approximation error of the random field, and the integration error of the Monte Carlo rule. This allows us to significantly reduce the computational time.


Hierzu ergeht herzliche Einladung.
Prof. Dr. Daniel Peterseim

wwwadm@Math.Uni-Augsburg.DE,    Do 1-Feb-2018 09:24:32 MEZ