Semantically-oriented mutation operator in cartesian genetic programming for evolutionary circuit design

Cartesian genetic programming (CGP) represents the most efficient method for evolution of digital circuits. Despite many successful applications, however, CGP suffers from limited scalability, especially when used evolutionary circuit design, i.e. design circuits a randomly initialized population. Considering multiplier problem, example, 5\(\times\)5-bit complex designed by scratch. The efficiency highly depends on performance point mutation operator, this operator is purely stochastic. This contrasts with recent developments in (GP), where advanced informed approaches such as semantic-aware operators are incorporated to improve search space exploration capability GP. In paper, we propose semantically-oriented (\(\mathrm {SOMO}^k\)) suitable combinational contrast standard modifying values mutated genes randomly, proposed uses semantics determine best value each gene. Compared common and its variants, converges Boolean benchmarks substantially faster while keeping phenotype size relatively small. successfully evolved instances presented paper include 10-bit parity, 10 + adder multiplier. were less than one hour single-thread implementation running CPU.