Challenges on applying genetic improvement in JavaScript using a high-performance computer

Abstract

Genetic Improvement is an area of Search Based Software Engineering that aims to apply evolutionary computing operators to the software source code to improve it according to one or more quality metrics. This article describes challenges related to experimental studies using Genetic Improvement in JavaScript (an interpreted and non-typed language). It describes our experience on performing a study with fifteen projects submitted to genetic improvement with the use of a supercomputer. The construction of specific software infrastructure to support such an experimentation environment reveals peculiarities (parallelization problems, management of threads, etc.) that must be carefully considered to avoid future research threats to validity such as dead-ends, which make it impossible to observe relevant phenomena (code transformation) to the understanding of software improvements and evolution.

Enhancing Genetic Improvement of Software with Regression Test Selection

“…Genetic improvement (GI), an Artificial Intelligence technique, has been successfully used to improve various software properties, ranging from reduction of software's runtime [5]- [7], optimisation of energy [10]- [12] and memory [8], [9] consumption, through to bug fixing [2]- [4] and addition of new software features [33]. However, it has yet to see wider uptake.…”

“…The most common type of functional improvement is APR [13], [14], [25], in which a faulty program is modified until the failing test suite passes. In non-functional GI however, the goal is to improve the software's memory usage [8], [9], execution time [5]- [7], energy consumption [10]- [12], and other non-functional properties, whilst maintaining the functional properties of the software, measured with the use of the program's test suite, i.e., test cases should pass after the program transformation. Either way, the GI process is guided by a fitness function that measures the level of functional or non-functional improvement.…”

“…Genetic Improvement (GI) is an Artificial Intelligence technique used to improve a given property of an existing software in an automated way [1]. The software property can be either functional (e.g., bug fixing [2]- [4]) or non-functional (e.g., runtime [5]- [7], memory usage [8], [9], energy consumption [10]- [12]), provided such a property can be formulated as a fitness function. The fitness function guides the iterative search process by quantitatively measuring the level of improvement of the software.…”

“…At each iteration, multiple variants of the software are produced and tested in order to assess whether the transformations preserve the overall software functionality, and then their level of improvement is measured. As one can infer, this process of constantly testing different versions of the software is timeconsuming [5], [13]- [15], especially when the software is accompanied by a costly test suite. Even for toy programs with relatively small test suites, GI executions can take several hours or even days [16].…”