Solving synthesis problems with genetic algorithms

Genetic algorithms (GAs) have several important features that predestinate them to solving logic synthesis problems. The main aim of this paper is to show how to apply the GAs for solving such synthesis problems. We propose a number of concepts for enhancement of a GA´s effectiveness and efficiency. These concepts include the mixed selection mechanisms, deterministic crossover and mutation operators, pseudo-random construction of the initial population, evolution of the operator application probabilities as the computation progresses, etc. In the paper, an effective and efficient GA scheme is proposed and applied for solving an important design problem: the minimal input support problem (MISP). In almost all cases, our GA produces strictly optimal results and realizes the best trade-off between effectiveness and efficiency. Experimental results clearly demonstrate that the proposed GA scheme is suitable for solving synthesis problems and its application results in very effective and efficient genetic synthesis algorithms.