In this paper, after a beginning on the concept of ant algorithms, a brief survey of the ant-based methods proposed for optimization of problems with continuous design spaces is presented. As a common approach in continuous domains, discretizing the search space is the model presented to be appended to the original ant colony system (ACS) algorithm. Evaluating this method and comparing it to the standard simulated annealing shows that it is robust enough not to fall in local minima. However, when higher resolution is required, the algorithm fails to capture the global optimums and the computational costs rapidly increase. Therefore, it can be safely proposed for the problems in which a trade-off between time, solution accuracy and algorithm intricacy is needed.

Keywords