OptLets: A Generic Framework for Solving Arbitrary Optimization Problems

Meta-heuristics are an effective paradigm for solving large-scale combinatorial optimization problems. However, the development of such algorithms is often very time-consuming as they have to be designed for a concrete problem class with little or no opportunity for reuse. In this paper, we present a generic software framework that is able to handle different types of combinatorial optimization problems by coordinating so-called OptLets that work on a set of solutions to a problem. The framework provides a high degree of self-organization and offers a generic and concise interface to reduce the adaptation effort for new problems as well as to integrate with external systems. The performance of the OptLets framework is demonstrated by solving the well-known Traveling Salesman Problem.