Cartesian Genetic Programming with Guided and Single Active Mutations for Designing Combinational Logic Circuits

Abstract

The design of digital circuits using Cartesian Genetic Programming (CGP) has been widely investigated but the evolution of complex combinational logic circuits is a hard task for CGP. We introduce here a new mutation operator for CGP that aims to reduce the number of evaluations needed to find a feasible solution by modifying the subgraph of the worst output of the candidate circuits. Also, we propose a variant of the standard evolutionary strategy commonly adopted in CGP, where (i) the Single Active Mutation (SAM) and (ii) the proposed mutation operator is used in order to improve the capacity of CGP in generating feasible circuits. The proposals are applied to a benchmark of combinational logic circuits with multiple outputs and the results obtained are compared to those found by a CGP with SAM. The main advantages observed when both mutation operators are combined are the reduction of the number of objective function evaluations required to find a feasible solution and the improvement in the success rate.