Synthesis of reversible logic circuit using a species conservation method

This paper aims to propose a modified species conservation technique for reversible logic circuits synthesis which is characterized by multimodal and large search space. The species conservation technique is tailored to adapt the uncertainty caused by the variable length representation. The different species is divided according to a new similarity definition and the similarity threshold is dynamically adjusted with the increasing of the chromosome length to ensure the search space exploring. A species elimination and restart search are conducted to avoid redundant search when a species converged. The same reproduction probability, other than that proportionate to its ranking, is given to different species. Experiments have been performed on a series of benchmark test functions. Comparison is primarily conducted to show the superior performance different to the basic evolutionary algorithm without species conservation mechanism and the original species conservation method.