Section: New Results

On Translating MiniZinc Constraint Model into Fitness Functions: Application to Continuous Placement Problems.

Participants : François Fages, Thierry Martinez, Bao Duy Tran.

MiniZinc is a solver-independent constraint modeling lan- guage which is increasingly used in the constraint programming com- munity. It can be used to compare different solvers which are currently based on either constraint programming, Boolean satisfiability or mixed integer linear programming. In [12] , we show how MiniZinc models can be compiled into fitness functions for evolutionary algorithms. More specifically, we describe the translation of FlatZinc models into fitness functions over the reals and their use in the Covariance Matrix Adapta- tion Evolution Strategy (CMA-ES) solver. We illustrate this approach, and evaluate it, on the modeling and solving of complex shape continuous placement problems.