Just a note that getting familiar with jupyter notebooks can get you
   into machine learning really fast.
   I've actually written a general equation solver for histograms, it's in
   the gr-blocks subfolder of my openemissions repo.  Given histograms of
   known values it computes a histogram of an unknown, using the
   expression I crazily posted.
   I'm very curious how and why the reference quantifies the relation
   between properties of sampling a sample and properties of a population.