Distance between vectors with missing values with no overlapping values

Question

I am trying to cluster, with Python, some sparse data, using agglomerative clustering on a precomputed distance matrix. My data are vectors with missing (NaN) values. I can calc