List of Figures

• 2.1. The Sante Fe Trail for the Artificial Ant Problem.
• 2.2. Truth table for the Even-3-Parity problem.
• 2.3. The Quartic and Rastrigin functions.
• 4.1. Example of entropy distributions.
• 4.2. Example of population visualisations on a circular lattice.
• 4.3. Examples of ranked correlation scatter plots between fitness and diversity.
• 4.4. Fitness vs. generation for Ant, Parity, Quartic and Rastrigin experiments.
• 4.5. Average number of nodes vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.6. Average depth vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.7. Phenotype diversity vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.8. Average entropy vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.9. Edit distance One diversity vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.10. Edit distance Two diversity vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.11. Genotype diversity vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.12. Pseudo-isomorph diversity vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.13. Correlation coefficient vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.14. Fitness vs. entropy for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.15. Fitness vs. edit distance One diversity for the Ant, Parity, Quartic and Rastrigin experiments.
• 4.16. Fitness vs. edit distance One diversity vs. generation for the Ant, Parity, Quartic and Rastrigin experiments.
• 5.1. Example of genetic lineage loss during recombination.
• 5.2. An example of lineage selection.
• 5.3. Average mean and average best fitness vs. generation for the control and lineage selection experiments.
• 5.4. Average diversity measures and size vs. generation for the control and lineage selection experiments.
• 5.5. Edit distance between successive best-fit individuals for the control and lineage selection experiments.
• 5.6. Generation where best fitness was found in the Ant experiments.
• 5.7. Average size of an individual in the generation where the best fitness was found in the Parity experiments.
• 5.8. Behaviour definition example for the Regression domain, with standard fitness calculation (mean squared error) and the behaviour definition.
• 5.9. Ant results, cumulative sampling of unique structures and behaviours.
• 5.10. Parity results, cumulative sampling of unique structures and behaviours.
• 5.11. Regression results, cumulative sampling of unique structures and behaviours.
• 5.12. Confidence bars for the average cumulative structure and behaviour sampling distributions.
• 6.1. The Binomial-3 and random polynomial functions.
• 6.2. Discretized final population results for the Binomial-3 and random polynomial experiments.
• 6.3. Size vs. fitness, generation-of-best-fitness vs. fitness and generation-of-best-fitness vs. depth for the Binomial-3 experiments.
• 6.4. Size vs. fitness, generation-of-best-fitness vs. fitness and generation-of-best-fitness vs. depth for the random polynomial experiments.
• 6.5. Generation vs. (fitness, depth, nodes, diversity and entropy) for the Binomial-3 experiments.
• 6.6. Generation vs. (fitness, depth, nodes, diversity and entropy) for the random polynomial experiments.
• 6.7. Spearman correlation between size and fitness and between edit distance and entropy for the Binomial-3 experiments.
• 6.8. Spearman correlation between size and fitness and between edit distance and entropy for the random polynomial experiments.
• 6.9. Hypothesis of a causal relationship between code growth and instance difficulty.
• 6.10. The model of difficulty and growth experiment results.
• 7.1. The division of phenotype and genotype spaces to define genetic outliers.
• 7.2. The target binary tree structure for the Tree-String problem.
• 7.3. The pareto front and points visited for the Tree-String experiments.
• 7.4. The average number in the population, the number of times selected and the survivability of the outliers, in-liers and un-fit for the Tree-String experiments. Outliers are defined by the (fitness, similarity) tuple as (better-than, 2 standard deviations).
• 7.5. The average number in the population, the number of times selected and the survivability of the outliers, in-liers and un-fit for the Tree-String experiments using the (better-than $\lor$ equivalent-to , 2 standard deviation) definition of outliers.
• 7.6. The average number in the population, the number of times selected and the survivability of the outliers, in-liers and un-fit for the Tree-String experiments using the (better-than, 1 standard deviation) definition of outliers.
• 7.7. The average number in the population, the number of times selected and the survivability of the outliers, in-liers and un-fit for the Tree-String experiments using the (better-than $\lor$ equivalent-to , 1 standard deviation) definition of outliers.
• 7.8. The average number in the population, the number of times selected and the survivability of the outliers, in-liers and un-fit for the Ant experiments.
• 7.9. The average number in the population, the number of times selected and the survivability of the outliers, in-liers and un-fit for the Parity experiments.
• 7.10. The average number in the population, the number of times selected and the survivability of the outliers, in-liers and un-fit for the Binomial-3 experiments.
• 7.11. Two possible views of outliers in the genotype space.