Global Optimization of Stochastic Multivariable Functions

This paper presents a new methodology for global optimization of the stochastic multivariable functions subjet to stochastic, possibly nonlinear constraints. The least squares parametric estimation is applied as an intermediate step in the stochastic optimizer which uses a special transformation to capture the global optima estimate. A comparative study of implementing estimation in polynomial least squares, versus spline fitting is also presented together with the illustrative examples.