EasyAnalyzer: An Object-Oriented Framework for the Experimental Analysis of Stochastic Local Search Algorithms

One of the aspects of applying software engineering to Stochastic Local Search (SLS) is the principled analysis of the features of the problem instances and the behavior of SLS algorithms, which —because of their stochastic nature— might need sophisticated statistical tools. In this paper we describe EasyAnalyzer, an object-oriented framework for the experimental analysis of SLS algorithms, developed in the C++ language.EasyAnalyzer integrates withEasyLocal++, a framework for the development of SLS algorithms, in order to provide a unified development and analysis environment. Moreover, the tool has been designed so that it can be easily interfaced also with SLS solvers developed using other languages/tools and/or with command-line executables. We show an example of the use of EasyAnalyzer applied to the analysis of SLS algorithms for the k-GraphColoring problem.