Parallel grammatical evolution for circuit optimization

Abstract

This paper describes application of Parallel Grammatical Evolution (PGE) algorithm to combinatorial logic circuit generation. The grammar and algorithms used are described. To increase the efficiency of Grammatical Evolution (GE) the backward processing algorithm was used. Different approaches to create multiobjective fitness functions are described and tested. Specifically the fitness functions are defined as set of rules incorporating different comparison methods in each stage of the computation. The algorithm is internally parallel and consists of three different interconnected populations. © 2008 Springer-Verlag Berlin Heidelberg.