Populations and Population Dynamics

• Note that there isn't just one gene at a time, but a population of genes.
• Typically, the simulations run through generations.
• One generation is entirely replaced by the next.
• The best of the current generation are preferentially selected for breeding.
• You can keep old genes, or sock them away for later.
• What often happens is the gene pool settles into a space where everything is similar (both in genes and evaluation).
• You can keep different sub-populations.
• You can run it several times, and keep the results.
• You can then combine them.
• In general you want to find some good solutions. Since the problem space is (usually) so big, you don't normally get the best solution.