Closed schrum2 closed 2 years ago
Now the type_count
works with values other than 1000 (when the RANGES are values other than 10). As for type_target
, I tested it twice with two different values and different dimensions and it also seemed to work. The fitness_threshold
seemed to accurately change to match the changes made with different command line parameter values.
The NEAT config requires that a fitness threshold be specified. This is the value at which evolution will terminate in success, and is generally the maximum possible fitness. We need to make it so that each fitness function can tell the simulation what the
fitness_threshold
is. It is currently specified inevolution.py
as 1000, but this only works for thetype_count
function, and ONLY when the shape is 10 by 10 by 10!Make it so that all fitness functions, starting with
type_count
, return their maximum fitness threshold if theclient
parameteris None
. This means you can call whatever the fitness function is (which will be able to vary once #39 is resolved) with aNone
for theclient
(though you will still need to provide accurateposition_information
) and it will return the fitness_threshold for you (assign it inevolution.py
).Basically, each fitness function just starts with an if statement that returns the max value
if client is None: