Multistage Multi-population Species Conservation Evolutionary Algorithm for Reversible Logic Circuit Synthesis

This paper aims to propose a multistage multi-population species conservation evolutionary algorithm for reversible logic circuit synthesis (MMSCEA_RLC), which substitutes the species updating schemes in our SCEA_RLC by a new population updating one. After deleting the species falling into evolutionary stagnation, SCEA_RLC reinitializes the species using random chromosome, whereas, MSSCEA_RLC redefines a new fitness function for a reinitialized population according to the seed of the deleting species. The population updating intends to conquer the side effect brought by parsimony pressure originally used to balance the object optimization and constraint violation decreasing in variable length representation, increase the feasible solution rate through a more detailed exploration and save the fitness evaluation. Dynamic tree-structured populations with species are formed through population updating. The solution composition is conducted from the root to a leaf. Experiments carried out over a series of benchmark problems hard for SCEA_RLC show the superior performance of MSSCEA_RLC.