design an Evolutionary algorithm with GPU/CUDA programming + phthon

All Operations used a thread block of CUDA. First randomly population is saved in global memory. And the assigned block checks whether the individual fitness is evaluated or not. If not then randomly select one individual and evaluate it. Crossover, mutation, selection, and termination assign blocks used global memory for In and Out solution process. When this global memory in out of range then assign one block that removes the worst (low) solution from global memory and replace this randomly to create a new individual