Finite state machine optimization using genetic algorithms

We present the results we have obtained after applying techniques on a basis of genetic methodology to the resolution of problems related with the automatic synthesis of digital circuits. We tackle the minimization of the number of states in incompletely specified finite state machines and the optimal state assignment on two level logic. Both class of problems involves the resolution of NP problems. In the first case, we have used a classical genetic algorithm. In the second one have been used new types of operators and ways of representation to avoid the problems that appear. Finally, we try to find the optimal mutation probability which guarantees the exploration of new regions of solution space without search becoming aleatory