Creating the initial population

[Max Ploner]

Hello,

I think there is an issue with how the initial population is created at the moment. In this line np.tile should be replaced with np.repeat.

Let's assume we have 3 input (+1 bias) nodes and 4 output nodes. Then, the algorithm will do the following steps (a little bit simplified):

ins = np.arange(4)
outs = np.arange(4) + 4
conn = np.empty((2, 16))
conn[0,:] = np.tile(ins, len(outs))
conn[1,:] = np.tile(outs,len(ins))

This will result in these connections:

array([[0., 1., 2., 3., 0., 1., 2., 3., 0., 1., 2., 3., 0., 1., 2., 3.],
       [4., 5., 6., 7., 4., 5., 6., 7., 4., 5., 6., 7., 4., 5., 6., 7.]])

The first input node is connected 4 times to the first output node, but no other. Using np.tile once and np.repeat once leads to the desired results (each input connected to each output):

array([[0., 1., 2., 3., 0., 1., 2., 3., 0., 1., 2., 3., 0., 1., 2., 3.],
       [4., 4., 4., 4., 5., 5., 5., 5., 6., 6., 6., 6., 7., 7., 7., 7.]])

[adam]

you're right, good catch! fixed.