There are many (numeric) optimizers which solve some subset of the set of multi-objective, black-or-grey-or-white box, robust, stochastic, mixed-integer, non-linear, non-smooth, manifold-constrained programming problems.
Some of them are well known. Such as all the solvers that appear in the Mittelmann benchmarks. I don't want to hear about those, I want to hear about obscure, old, bleeding edge, unknown solvers or algorithms that you chose over the more popular well known ones:
What made you choose it over more well known solvers?
Where can I learn more about it?
What is the underlying principle?
What kind of problems can it solve?
Under what license/pricing is it (or an implementation of it) available for hobbyist, academics or professionals?
From what programming languages can it (or an implementation of it) be called?
When describing an algorithm please be specific (not just "evolutionary algorithm") about it.
Background
If I find any solver/algorithms intriguing I might decide to wrap/implement them for a modeling language. Others might find a solver that fits their use case in the replies or learn something new and interesting.
