Trajectories and attractors as specification for the evolution of behaviour in cellular automata

In this paper evolvability of uniform and nonuniform cellular automata is investigated in relation to the level of detail in the behavioural description. An experimental approach was taken to test how increased information, from only initial and final state to an approach including intervals of interest in the state space, influenced the evolutionary process. The results are promising as it seems that a more detailed specification of the behaviour, i.e. trajectory, was not annihilating regarding evolvability. The presented work is part of a larger goal of applying artificial evolution and development to achieve the goal of designing cellular machines that can be applied to complex computation and modelling. The specification of the sought problems for such machines is often of a nature that includes incomplete knowledge of the behaviour.