A Flexible Framework for Analysing Genetic Algorithms In Go
Kahembwe, E. (2014). A Flexible Framework for Analysing Genetic Algorithms In Go (BSc (Hons) Games Development Dissertation). Edinburgh Napier University (Kerridge, J.,
This dissertation documents the planning, research, development and implementation of a Go framework for analysing genetic algorithms. It begins by describing the motivation, aim & objectives, research questions, and scope thereof before defining the constraints and sources of information that form the basis of this project.
The work in this project can be subdivided into two main components, the investigation of Go and the study of the Hierarchical Fair Competition (HFC) algorithm.
By developing the analysis framework in Go, this provides for a good opportunity to investigate the Go language while concurrently studying the principles behind the HFC. The final deliverable is a Go library for genetic algorithms that supports single and multi-population genetic algorithm (GA) models such as the Island and HFC model.
This paper also identifies and addresses flaws in previously proposed HFC algorithms. Additionally, a new algorithm for minimisation based on the HFC principles is also presented.
It is concluded that the proposed algorithm performs significantly better than the traditional Island and single population GAs. It is also argued that the proposed algorithm is more stable and reliable than previously proposed adaptive HFC algorithm.
It is also concluded that Go is a suitable language for scientific computing.