Abstract. In this work we use metaheuristic search to improve Google's Guava library, finding a semantically equivalent version of com.google. common.collect.ImmutableMultimap with reduced energy consumption. Semantics-preserving transformations are found in the source code, using the principle of subtype polymorphism. We introduce a new tool, Opacitor, to deterministically measure the energy consumption, and find that a statistically significant reduction to Guava's energy consumption is possible. We corroborate these results using Jalen, and evaluate the performance of the metaheuristic search compared to an exhaustive search-finding that the same result is achieved while requiring almost 200 times fewer fitness evaluations. Finally, we compare the metaheuristic search to an independent exhaustive search at each variation point, finding that the metaheuristic has superior performance.

We discuss the useful role that can be played by a subtype of improvement programming, which we term 'Embedded Dynamic Improvement'. In this approach, developer-specified variation points define the scope of improvement. A search framework is embedded at these variation points, facilitating the creation of adaptive software that can respond online to changes in its execution environment.

No abstract

Abstract. There is an emerging trend towards the automated design of metaheuristics at the software component level. In principle, metaheuristics have a relatively clean decomposition, where well-known frameworks such as ILS and EA are parameterised by variant components for acceptance, perturbation etc. Automated generation of these frameworks is not so simple in practice, since the coupling between components may be implementation specific. Compositionality is the ability to freely express a space of designs 'bottom up' in terms of elementary components: previous work in this area has used combinators, a modular and functional approach to componentisation arising from foundational Computer Science. In this article, we describe Haiku, a combinator toolkit written in the Scala language, which builds upon previous work to further automate the process by automatically composing the external dependencies of components. We provide examples of use and give a case study in which a programatically-generated heuristic is applied to the Travelling Salesman Problem within an Evolutionary Strategies framework.

Abstract. Apache Spark is a popular framework for large-scale data analytics. Unfortunately, Spark's performance can be difficult to optimise, since queries freely expressed in source code are not amenable to traditional optimisation techniques. This article describes Hylas, a tool for automatically optimising Spark queries embedded in source code via the application of semantics-preserving transformations. The transformation method is inspired by functional programming techniques of "deforestation', which eliminate intermediate data structures from a computation. This contrasts with approaches defined entirely within structured query formats such as Spark SQL. Hylas can identify certain computationally expensive operations and ensure that performing them creates no superfluous data structures. This optimisation leads to significant improvements in execution time, with over 10,000 times improvement observed in some cases.