The optimization process of large-scale computational problems presents various issues that need to be solved in order to achieve favorable results. One of the most significant challenges faced is the computational costs of these problems. These costs force researchers to use large-scale computational resources. Parallel implementations of different algorithms may be considered. However, these techniques present issues when dealing with problems in which a high variability in the execution time of the evaluation function takes place. This paper presents an algorithm designed to efficiently optimize these large-scale optimization problems with different execution times for the evaluation of the candidate solutions. Production grid computing infrastructures are used to carry out these optimizations. The own paradigm of the grid introduces new challenges in the process that must to be taken into account in the design of the algorithm.
展开▼
