BIG DATA Laboratory

The research group Complex Intelligent Systems (CIS) of the Department of Mathematics and Computer Science, Università degli Studi di Catania organizes the seminar entitled Software Engineering Abstractions for Metaheuristics: a 20 years journey. The seminar is taught by prof. Luca Di Gaspero, Università degli Studi di Udine.

Abstract: Differently from other search and optimization paradigms, e.g., branch and bound, the supposed simplicity of metaheuristic methods and the limited availability of modeling tools induces the researchers to build their metaheuristics applications from scratch. On the other hand, the evolution of the techniques (local search, genetic and evolutionary algorithms, hybrids, etc.) seems to make the development of general tools quite impractical.

In 2003, together with Andrea Schaerf, we decided to subvert these stereotypes and we proposed EasyLocal++, an open-source object-oriented framework for local search metaheuristics, with the aim of engineering the development of solvers both at a research and at an industry-ready level. After 20 years the system has evolved (according to the new capabilities of the underlying programming language, i.e., C++) and extended (incorporating further modules to be industry-level) but most of the design choices (or in other words the software abstractions) are still valid nowadays.

In this presentation I will briefly survey the tools for metaheuristic development and I will present the abstractions that are incorporated into EasyLocal++, exhibiting some software engineering principles behind them. I will show how useful is to work with an abstract framework, also in terms of providing different access points to the solver.

The seminar will be on January 13, 15:30, with link for Microsoft Teams available upon request.

News by prof. Mario F. Pavone, Department of Mathematics and Computer Science, Università degli Studi di Catania.

Share This

S5 Box

Cini Single Sign ON

This site only stores technical/functional cookies. If you want to know more, go to the Cookie Policy section.