Mutual inforamtion loss normaliser 1d & 2d redundant?

[qiuhuaqi]

This step computing the joint distribution in MI pairwise loss:

airlab/airlab/loss/pairwise.py

Line 381 in 80c9d48

p_joint = th.mm(p_f, p_m.transpose(0, 1)).div(self._normalizer_2d)

The marginal distributions are already normalised by self._normalizer_1d in:

airlab/airlab/loss/pairwise.py

Line 339 in 80c9d48

def _compute_marginal_entropy(self, values, bins):

I might be mistaken. But doesn't the joint distribution no longer need normalisation (self._normalizer_2d) as its the outer product of the normalised marginal distributions?