An evolutionary reincarnation algorithm

As there is little or no experimental experience of reincarnation in the natural world, attempts to add a reincarnation metaphor to an evolutionary algorithm must of necessity proceed cautiously. In previous work the authors have established that the reintroduction of previously stored gene values into the population can have a noticeable effect on the progress of evolution, this paper now considers a range of options for deciding which gene values to store, which to return and which individuals in the current population should receive the returned gene values. Consistent experimental results on three well known functions allow a suggestion to be made of, if not the best choices, at least a good choice selection to use on initial experiments on other problems.