Real world problems typically have several objectives and constraints that must be solved in feasible time for decision making. Trial and error or exhaustive solutions can take too much time and be boring.

A computational library of deterministic multi-objective non-linear optimization algorithms with theorical guarantees of improving the solution and reducing its uncertainty after each iteration.

The following figure depicts the convergence of the ellipsoid algorithm towards the minimum optimal point of the Svanberg optimization problem whose objective function is represented by level curves.