Boolean representation outputs boolean values instead of binary strings.

[markcoletti]

LEAP binary representation uses a numpy binary vector. (See, binary_rep.initializers.random_bernoulli_vector()). The string representation returns a vector of booleans instead of the expected binary string. E.g.,

n = np.random.rand(*shape) < 0.5
n
Out[6]: array([False, False, False, False])

That should be [0,0,0,0].

You can cast that to a vector of integers:

n.astype(int)
Out[7]: array([0, 0, 0, 0])

But then you presumably lose the compressed and optimized binary representation used by numpy.

So, ideally we'd like to keep that internal optimized binary representation while also printing ones and zeroes.

[markcoletti]

@lukepmccombs has a solution that involves extending his context manager in probe.py that entails use of tailoring a numpy string format for binary genomes.