Two Simulated Annealing Algorithms for Noisy Objective Functions
Andrei A. Prudius, Sigrún Andradóttir · 2006
We present two new variants of the simulated annealing algorithm (with a decreasing cooling schedule) that are designed for solving discrete simulation optimization problems. We also provide conditions under which our methods converge almost surely to the set of global optimal solutions, discuss the implications of our results for both transient and steady-state simulations, and provide some numerical results.