Speaker
William McEneaney
(University of California, San Diego)
Description
Abstract: We consider nonlinear control problems of both deterministic and stochastic type. In the latter class, we consider only the case of dynamics driven by Brownian motion. We seek fundamental solutions of the associated Hamilton-Jacobi PDE problems, where here, these are defined to be solutions such that changes in the terminal-cost data require only a stat-convolution of the fundamental solution with an object determined by the specific terminal-cost data. The fundamental solutions are generated by solution of first-order Hamilton-Jacobi PDE problems over a certain dual space.