SciPy does this, and when you think about computing the distance between vectors, this interpretation makes the Bray-Curtis dissimilarity invariant to reflection. Taking the absolute value does not make a difference if the vector elements are all positive, so this change would not alter the behavior of the function for species counts or proportions. Must look into the symmetry properties of manhattan distance and see how many are maintained if we take the absolute value.
SciPy does this, and when you think about computing the distance between vectors, this interpretation makes the Bray-Curtis dissimilarity invariant to reflection. Taking the absolute value does not make a difference if the vector elements are all positive, so this change would not alter the behavior of the function for species counts or proportions. Must look into the symmetry properties of manhattan distance and see how many are maintained if we take the absolute value.