Simultaneous perturbation Newton algorithms for simulation optimization (Q2260692)

The authors present a balanced three-simulation Hessian estimator based on the simultaneous perturbation technique. Using this Hessian estimator, two Newton-based algorithms that work around the normally tedious procedure of inverting the Hessian are also presented. The proofs of convergence for both algorithms are given, too. An interesting application of the algorithms on a problem of road traffic control is given.