Last year, I had two talented Bachelor students. The project I had in mind was to just MCMC on the M-dwarfs found in the BoRG survey. It was a nice way to reuse HST data taken already to find high-redshift galaxies. Honestly I thought they might improve the fit I had made earlier using my M-dwarf catalog somewhat.
Well! they took to the project and rapidly a) determined that there had been a bug in my original catalog, b) corrected it, and c) did an expert series of MCMC runs (using the emcee implementation).
The results were much better than I had expected. Not only did they manage to fit the disk of our Milky Way using the numbers of M-dwarfs found, they concluded they needed to use a Halo model as well. It worked beautifully: they inferred 60 billion M-dwarfs in the Milky Way, 7% in the MW halo, all with very nice errors, courtesy of the MCMC modelling.
The press release was yesterday (MNRAS press release) and (Dutch Press Release). Now we see how often it will be picked up by Dutch and foreign media...