Logic synthesis using a genetic algorithm
Kenji Ohmori, T. Kasai · 2002
The paper describes logic synthesis using a genetic algorithm, which succeeds in obtaining good results for many combinatorial problems. However, logic synthesis has difficulty in directly mapping its essential entities to genes. Several new ideas are introduced. In our GA, genes represent where product terms are placed in the circuit. Two product terms are transformed into a product term when placed in the same location by the crossover or mutation operation in order that a new circuit, closer to an optimal solution, is created. In the synthesis process, reduced forms are firstly obtained and optimal solutions of multiple level circuits are sought. Our genetic algorithm is applied to benchmark circuits and has succeeded in obtaining good results.