| We develop an unconstrained continuous genetic algorithm (GA) that asymptotically converges in probability to a global solution. The algorithm is based on Darwin's theory of natural evolution and the random search method. We also design a continuous version of tabu search and show its convergence. We then incorporate the idea of tabu search in our genetic algorithm.; For solving the constrained optimization problems, we introduce a repair operator, which is a mapping from an infeasible point to the nearest feasible point. In this dissertation we only study a linear constraint case. We apply our repair operator to our GA for solving linearly constrained optimization problems.; Our algorithms are implemented for computer simulation. A large number of examples have been adopted from the optimization literature. They are used to test our algorithms. Numerical results show that our stochastic algorithms can successfully find at least one globally optimal solution for most of the examples. |
