Mat-1.3656 Numeerisen
analyysin ja laskennallisen tieteen seminaari.
Ma 12.11. 2007, sali U356
Teemu Pennanen, Helsinki School of
Economics
Galerkin methods in
stochastic control
Galerkin methods are a classical technique for approximating infinite-dimensional optimization problems by finite-dimensional ones. When applied to convex problems of stochastic control, it yields computationally attractive alternatives to scenario tree based discretizations. We describe its implementations for dynamic portfolio optimization problems and report some encouraging numerical results.