I am experimenting with convergence of Galapagos. To get my data, I am comparing two methods:
- Getting the Species record data and porcessing the data in such a way, that:
Generation [1]: Genome[0], Genome[1]
Generation [2]: Genome[0], Genome[1]
gets converted to
iteration[1]
iteration[2]
iteration[3]
iteration[4]
- directly recording the fitness values with a
recordcomponent, then partitioning the list into the population size (in this case 10), sorting each generations values (like the species record) and flattening the list
this is the result of the two methods:
The green line indicates a new generation. In the first generation, the two curves fit perfectly, afterwards strange shifts start to happen.
My observation: The species record seems to always paste the current optimum as genome[0], while the actual output delivers another value at that iteration count, a value that the species record seem to cull. A small detail, but a bit confusing.
@DavidRutten