An Approach for Evolving Transformation Sequences Using Hybrid Genetic Algorithms

Abstract

The digital transformation revolution has been crawling toward almost all aspects of our lives. One form of the digital transformation revolution appears in the transformation of our routine everyday tasks into computer executable programs in the form of web, desktop and mobile applications. The vast field of software engineering that has witnessed a significant progress in the past years is responsible for this form of digital transformation. Software development as well as other branches of software engineering has been affected by this progress. Developing applications that run on top of mobile devices requires the software developer to consider the limited resources of these devices, which on one side give them their mobile advantages, however, on the other side, if an application is developed without the consideration of these limited resources then the mobile application will neither work properly nor allow the device to run smoothly. In this paper, we introduce a hybrid approach for program optimization. It succeeded in optimizing the search process for the optimal program transformation sequence that targets a specific optimization goal. In this research we targeted the program size, to reach the lowest possible decline rate of the number of Lines of Code (LoC) of a targeted program. The experimental results from applying the hybrid approach on synthetic program transformation problems show a significant improve in the optimized output on which the hybrid approach achieved an LoC decline rate of 50.51% over the application of basic genetic algorithm only where 17.34% LoC decline rate was reached.

Keywords

• Program analysis
• Program transformation
• Genetic algorithms
• Particle swarm optimization