Numeerisen analyysin ja laskennallisen
tieteen seminaari
14.11.2005 klo
14.15
U356
Teemu Pennanen, Kauppakorkeakoulu
A Galerkin method
for multistage stochastic programs
In Galerkin methods, one approximates infinite-dimensional
optimization problems by finite-dimensional ones that are obtained by
restricting the search for optimal solutions to subspaces spanned by
finite sets of ``basis functions''. In this talk, we show how this idea
leads to numerically solvable approximations of multistage stochastic
programs where uncertainty is modeled by general probability
distributions.