Evolution program for deterministic and stochastic optimizations

This paper presents an evolution program for deterministic and stochastic optimizations. To overcome premature convergence and stalling of the solution, we suggest an exponential-fitness scaling scheme. To avoid the chromosomes jamming into a corner, we introduce mutation-1 which mutates the chromosomes in a free direction. To improve the chromosomes, we introduce mutation-2 which mutates the chromosomes in the gradient direction or its negative, according to the kind of problem. Monte Carlo simulation will be employed to solve the multiple integral which is the most difficult task in the stochastic optimization. Finally, some numerical examples are discussed.