next up previous contents
Next: Detailed Analysis of a Up: Analysis and Visualisation Techniques Previous: Population-based Measures

   
A Neutral Shadow

In order to gauge the extent to which the observed dynamics of Cosmos are due to the adaptive properties of individual genotypes rather than to any other features of the system, we also ran a ``neutral shadow'' of the system (as suggested by Bedau et al. [Bedau et al. 98], [Bedau & Brown 97], [Bedau et al. 97]).

The neutral shadow works as follows. During a standard run of the system, various details are recorded at each time slice (if the parameter record_neutral_model_data is set appropriately). These details include the number of organism births during that time slice, the number of these new births which resulted in new genotypes (through the action of flaws or mutations), the number of cell divisions (to form parallel processes), the number of cell deaths, and any movements made by individual cells.

The idea of the neutral shadow is to re-run the system using this recorded data, but in such a way that that no particular organism or genotype has any adaptive advantage over any other. Nominal ``programs'' exist at grid locations, reproduce, and die. At each time slice, a number (specified by the data recorded from the real run) of these programs are chosen at random to reproduce and to mutate. (Cell divisions in the real run to form parallel programs are treated as reproductions in the neutral shadow, so that all programs are unicellular.)5.6 Similarly, a number of cells are chosen at random to be killed, and to move around the grid, determined by the numbers recorded from the standard run. In this way, we get the same number of events occurring in the neutral shadow as in the standard run, but these events happen to randomly chosen individuals. By applying the same measurement techniques to the neutral shadow as we do to the standard run, we can get an idea of the ``raw'' dynamics of the system, independent of the adaptive success of any particular genotype.

The results of the neutral shadow are discussed alongside the results of the standard run in the following sections.


next up previous contents
Next: Detailed Analysis of a Up: Analysis and Visualisation Techniques Previous: Population-based Measures
Tim Taylor
1999-05-29