A Method of Representing Cyclic Program Structures in Artificial Chemistry Model

Abstract

The need for automation of software development processes makes it necessary to search for forms of program representation that can undergo automatic transformations without violating the integrity and semantic significance of the results of such transformations. Previously, we have proposed a notation for programs that permits the use of automatic transformation methods, namely methods of evolutionary development used, in particular, in genetic pro­gramming. But we have considered only linear and tree-like structures. In this article, we expand the list of available types of structures by adding cyclic structures, as well as complex structural compositions obtained by combining structures of simpler types. We also propose a rule excluding possible anomalies with cyclic structures representation. In general, the proposed methods are based on the concept of artificial chemistry, where programs are considered as analogues of molecules, and program transformations are considered as analogs of reactions. We illustrate the application of proposed notation using examples of the Kekule formula and the cyclic program, automatically obtained in our previous studies. The results obtained demonstrate that the proposed notation and methods make it possible to compose formulas representing computational structures of various and even mixed types.