Hellow Abdul,

I did not uderstant your question. Is it about the optimization algorithm or the control llaw (cost function)?

When you cast your problem as an optimisation of an objective function with some constraints, there are various ways of finding its solution (approximate usually). Classical optimisation apporaches can have no stoachstic element; however, all modern approaches I am aware of have a stochastic element, for example, GA uses probability when making mutation and cross-over.