Multi-objective optimization is an active field today and there's no F# library to do it.
Metaheuristical techniques are considered like heuristics for generalized problems. Polymorphism in F# allow to avoid assumptions about the Data Types.
Most of the mathematical definitions can be translated to working code effortlessly.
Side effects in F# allows to generate new random solution candidates without having to use monadic wrapping.
Function composition and partial defined functions allow certain transformations of problems that can't be done (at least in an elegant way) using for example Java or C++.