Numerical evaluation of resolvents and Laplace transforms of Markov processes using linear programming (Q1396975)

For the computation of Laplace transforms of first-exit time distributions and resolvents of moments of standard Markov processes in bounded regions, a linear programming method based on the associated martingale is proposed, which leads to lower and upper bounds. The Poisson process, Brownian motion and the Ornstein-Uhlenbeck process are used as examples.