Yeomans [38] further extends the work by Yeomans et al. [37]
combining the use of penalty functions with GLP into an Evolutionary
Simulation-Optimization (ESO) procedure. The computationally
efficient GLP procedure [36] is used to quickly generate a
starting population to initialize the evolutionary algorithm. Then,
in order to allow ESO to explore regions beyond its feasible
boundaries, infeasible solutions are considered in a penaltybased
fashion, transforming the MSW model into an unconstrained
stochastic minimization problem