Dynamic search space characterisation

Optimisation involves finding the best solution to an unsolved problem. In problems that change over time, studies have shown that the problem's behaviour fundamentally affects the performance of the optimisation technique. Therefore, determining the behaviour of an unknown dynamic problem would allow the chosen approach to be tailored accordingly and improve the overall optimisation procedure. This thesis proposes a set of characterisation metrics that gather information on the behaviour of a dynamic problem. These metrics can successfully characterise a wide range of problems based on their behaviour and their topology, allowing for fast and efficient bottom-up characterisation of dynamic problems.