Related to #89 and #88. I can't remember if the non-vectorization in estimate_mi was a conscious decision or an oversight (as pairwise_mi is vectorized). This might need some design thinking, therefore did not include in 1.1.1.
I'm concerned that this might introduce some new edge cases that I forget to test for... This should be implemented (or at least started) by somebody who actually needs the feature.
Related to #89 and #88. I can't remember if the non-vectorization in
estimate_mi
was a conscious decision or an oversight (aspairwise_mi
is vectorized). This might need some design thinking, therefore did not include in 1.1.1.