Stargazing with statistics

19 July 2017

Dr Brendon Brewer
Dr Brendon Brewer

In his office just off Princes St, Dr Brendon Brewer is cataloguing stars. He’s an astrostatistician, using statistics to determine precisely what stars exist, exactly where they are, and how bright they are.

This star cataloguing is trickier than you might think. The raw data in much of astronomy is an image – essentially, a photo taken with a long exposure time through a telescope. Due to atmospheric effects and the properties of the telescope, stars blur into smudges – and there’s also what Dr Brewer calls “noise”, or speckling that interferes with the ability to determine what’s there.

In an effort to get around this, astronomers have tried all sort of ways over the years to extract data from these fuzzy images, which tended to be “heuristic methods – rules of thumb – with a bit of stats sprinkled in,” says Dr Brewer, a senior lecturer in the Department of Statistics. But he and colleagues David Hogg and Dan Foreman-Mackey eventually came up with a reliable method that delivers impressive results, and their paper detailing the method, Probabilistic Catalogs for Crowded Stellar Fields, was published in 2013.

The method takes a “fairly purist Bayesian stance”, says Dr Brewer – that means embracing the idea that we can and should represent human uncertainty using probability theory. So the method aims to find all the possible solutions that are plausible, given the constraint of the data, rather than just finding a single guess for the solution.

“For example, with this method, we can pick up on objects in an image that other methods miss,” explains Dr Brewer, “and when two stars are very close together in an image so they look like they might be a single star, we can quantify how sure we ought to be that it is one star or two.” The downside, he adds, is that such work takes a lot of computing power, so can be applied to only small images at this stage.

As a young child, Australia-born Dr Brewer loved science. “I lost that in high school but then found it again in later years when I had really good science and maths teachers who got me to do the homework,” he says.  Dr Brewer took physics, applied maths and statistics at the University of Sydney, “but I didn't dream of doing astrophysics in particular; rather, I ended up there because one of my favourite lecturers happened to be an astrophysicist and I wanted to work with him.” That was Welshman Geraint Lewis, under whose supervision Dr Brewer completed his PhD in astrophysics in 2008; the pair have worked together since. After his doctorate, Dr Brewer was a research associate at the University of New South Wales and then a post-doctoral scholar at the University of California, Santa Barbara, before relocating to the University of Auckland in 2012.

When he’s not spending time with wife Lianne and the couple’s bird and puppy, Dr Brewer is a keen science blogger. You can find his thoughtful posts on his own blog, Plausibility Theory, and also occasionally at Quillette.