Exploring Evolutionary Search Strategies to Improve Applications' Energy Efficiency.

Energy consumption have become an important nonfunctional requirement for applications running on battery powered devices through data centers. Despite the increased interest on detecting and understanding what causes an application to be energy inefficient, few works focus on helping developers to automatically make their applications more energy efficient based on developers' design and implementation decisions. This paper explores how search strategies based on genetic algorithms can help developers automatically find an energy efficient version of an application based on transformations corresponding to developers' high level decisions (e.g., selecting API implementations). Our results show how different search strategies can help to improve the energy efficiency for nine Java applications.