Leopold Luley

A Framework for Modular Construction and Evaluation of Metaheuristics

Tech Report - PDF

This paper presents MAHF, a software framework for the highly flexible construction of metaheuristics from individual components and the subsequent evaluation of these algorithms. At that, MAHF is developed specifically for the experimental analysis of the algorithmic behaviour during the optimization process with a focus on the influences of the algorithm’s components.

Furthermore, uncommon and incompletely examined operators or frameworks of “novel” metaheuristics are included as well, so that their usefulness can be assessed. In the following, we will elaborate on MAHF’s structure and its general goals and application possibilities. Concerning MAHF’s component structure, we will provide examples of its usage and extension to ensure that it is reusable by others as well.