Research of a grid-enabled parallel computational model and algorithm implementation

In this paper, a grid computational model and algorithm based on mind evolutional computation is constructed and implemented, which supports the dynamically resource allocation under grid environment. We investigate the grid enabled parallel computational model performance metrics, and proposed the Mind Evolutionary Computation based space decomposition parallel evolutionary algorithm, which simulates the human behavior and divides the population into the superior sub-populations and substitution sub-populations. The MEC based parallel evolutionary algorithm (MEPEA) has been successfully applied to Shanghai High Education Grid-realistic case studies. MEPEA algorithm included splicing/decomposable encoding scheme can solve computation intensive problems by using low-dimension algorithms. The proposed algorithm is experimentally testified with a test suit containing four complex function optimization benchmarks. The experiments all demonstrate that the proposed algorithm outperforms other algorithms in both scalability and solution quality.