Bayes in astronomy

David Schminovich points me to this paper by Yu Lu, H. Mo, Martin Weinberg, and Neal Katz:

We believe that a wide range of physical processes conspire to shape the observed galaxy population but we remain unsure of their detailed interactions. The semi-analytic model (SAM) of galaxy formation uses multi-dimensional parameterisations of the physical processes of galaxy formation and provides a tool to constrain these underlying physical interactions. Because of the high dimensionality, the parametric problem of galaxy formation may be profitably tackled with a Bayesian-inference based approach, which allows one to constrain theory with data in a statistically rigorous way. In this paper we develop a SAM in the framework of Bayesian inference. . . .

And here’s another from the same authors, this time on “Bayesian inference of galaxy formation from the K-band luminosity function of galaxies: tensions between theory and observation.”

I haven’t actually looked at the papers but I thought some of you out there might be interested.

3 thoughts on “Bayes in astronomy

  1. There have been a fair number of papers in astronomy recently which exploit Bayesian Inference and MCMC (e.g. it’s being used for estimating parameters of extrasolar planets too). Astronomical data is extremely sparse if you think about it; all we get is photons to constrain very complex processes. I can see Bayesian methods becoming very popular in this field soon, though it would take more training and computing power. (PS: Very interesting blog you have!)

  2. I fear that soon “Bayesian” will get a bad name in astronomy: It is used to mean “we multiplied by a completely made-up prior and used MCMC”. It *should* mean “we regularized in the situations where the data are uninformative and we marginalized out our nuisance parameters” but it rarely does. At worst it means “we performed model comparison via Bayes integrals and our results depend incredibly sensitively on what we put in for the limits on (or support of) our made-up priors”. I have stopped using “Bayesian” in favor of “probabilistic” but that won’t really save us: Bad Bayesian inference is just as bad (it turns out) as bad made-up heuristics.

    • oops! I just had coffee with Schiminovich and I realized that my comments here *look* like they are applied to the above-quoted articles: They aren’t! The above-quoted articles are *excellent*. I was responding to the implicit question about Bayes generally in astronomy.

Comments are closed.