Automatic Design of Cell-like P Systems through Tuning Membrane Structures, Initial Objects and Evolution Rules
Zhu Ou, Gexiang Zhang, Tao Wang and Xiaoli Huang
Membrane computing is an important and newly-emerging branch of natural computing. As the basic type of membrane computing models, a cell-like P system consists of three basic elements: membrane structure, initial objects and evolution rules. The automatic design of a cell-like P system in the existing literature considered a simplified case, i.e., the selection of evolution rules from a predefined set of redundant evolution rules on the condition that the membrane structure and initial objects were fixed. How to automatically design a cell-like P system by simultaneously considering its three basic elements is a challenging and ongoing issue. In this paper, we propose an automatic design approach for a cell-like P system, by tuning its three key elements, membrane structures, initial objects and evolution rules, simultaneously and in an interrelated manner. In this method, a binary encoding technique is used to codify the P system with variable membrane structures, initial objects and evolution rules; an elitist genetic algorithm is applied to evolve a population of P systems toward a successful P system for fulfilling a specific task; an effective fitness function is employed to evaluate each candidate P system by using a P-lingua simulator. The parameter setting is also discussed. Experimental results show that the introduced method can successfully accomplish the automatic design of a cell-like P system for calculating the square of 4.
Keywords: Membrane computing, P systems, cell-like P systems, automatic design, genetic algorithm