Evolutionary Design Method Of Reversible Logic Gates And Its Implementation Based On CUDA

With the deepening of the research about reversible logic gates, there have emerged many comprehensive approaches of reversible circuits. However, these approaches are overwhelmingly targeting at the overall optimized design of logic circuits, but it focuses little on quantum logic gates which is the foundation of logic circuits, and the optimization of the basic logic gates will directly affect the optimized extent of quantum logic gates. If it can automatically evolved and gain gate structure with better function, more complete functions and least quantum cost, it will play an ineligible role in overall circuit optimization.In the circuit evolutions design algorithm (Genetic algorithm in particular), evolutionary objects are circuits’structures and parameters. It can explore large design space and gain novel or better result without any prepared knowledge and regulations, and it can even achieve completely automation of the complex and large circuits. Moreover, genetic algorithm has parallel parameter; therefore, changing to parallel genetic algorithm can raise the speed of solving.The paper is mainly about evolutionary design approach of reversible logic gates and its CUDA realization. Firstly, based on analysis and comparison, we choose basic gates like NOT、 CNOT、Controlled-V、Controlled-V+to construct complete and general gate base. Secondly, to build genetic algorithm model and scheme for reversible logic gates, and we use specific example to show it is feasible and effective; based on these, we change the genetic algorithm to parallel algorithm on CUDA platform. Finally, the paper achieves NCV parallel realization of reversible logic gates and compares its efficiency. Preliminary result shows the combination of genetic algorithm and CUDA and the application in reversible logic gates can boost the overall optimization of circuit evolutionary design, and it can also improve its search speed. Moreover, to achieve expected circuits and structure without knowledge and artificial intervention will provide reference to new (semi-)automatic discovery and the enlargement of current base.