Multi-threaded genetic-algorithm (MTGA)-based optimization for the one dimensional electromagnetic inverse scattering problem

Although genetic-algorithm (GA) based inverse scattering algorithms are powerful and robust optimization techniques because of their independence of the initial guess and their few constrains on the solution domain, their main drawback is their overall run time which can easily become unacceptable. Many techniques have been proposed to overcome this problem. One of the promising techniques is to parallelize the GA algorithm which is compatible with the parallel nature of many parts of the algorithm itself. For instance, the algorithm may be implemented on a multi-machine, i.e. cluster. However, this solution may be costly unacceptable. In this paper, multi-core platform is proposed to accelerate the one dimensional GA-based inverse scattering problem. The algorithm is parallelized using OpenMF programming models.