Analysis of van Dokkum et. al. 2018

An analysis of the star cluster velocities published in van Dokkum et. al 2018 (https://arxiv.org/abs/1803.10237) using python and pymc3. This was prompted by this twitter thread, by the astronomers Nicholas Martin and Nicholas Longeard.

I find the most probable value of $\sigma=10.3^{+5.7}_{-4.1}$ km/s, with the posterior shown in Plots/pdf.pdf. Allowing for some of the data to be classed as an "outlier", as in the excellent paper Hogg, Bovy and Lang 2010 (https://arxiv.org/pdf/1008.4686) leads to a very similar posterior (with the most probable $\sigma=8.7 ^{+6.0}_{-4.2}$ km/s). This posterior is shown in Plots/pdf_outliers.pdf.

You can read some of my thoughts about this paper and my analysis on Medium, here