The talk Preconditioning of stochastic Galerkin method, by Professor Ivana Pultarová, will be held on Monday January 5, 2015 at 8:30 in room K4.
Abstract of the talk
One of the popular methods for numerical solution of partial differential equation with uncertain data is the stochastic Galerkin method. Approximation spaces used are tensor products of finite elements of spatial variables and of sets of polynomials of random variables. Studying relations among subspaces of these tensor products is important for efficient preconditioning and for a posteriori error estimates.
Especially, small values of the strengthened Cauchy-Bunyakowski-Schwarz constants between some subspaces with respect to the energy scalar product lead to efficient two-by-two block preconditioning. We introduce some sharp upper bounds to these constants.