Six key trends in contemporary statistics that really could revolutionise astronomical data analysis …

I’ve just come across this very interesting astrostatistics site, and I thought I’d reblog a piece from it. In fact I did make a very crude attempt back in the 90s to do something very like the SPDE analysis described here, but it came to nothing and I dropped the idea. Now it seems that there’s been a great deal of more recent activity in this area which I knew nothing about so it might be worth reviving interest in it.

Now. Where did I put those notes?

Another Astrostatistics Blog

(2) the SPDE approach to large-scale random fields.
Cosmologists have long been familiar with the statistics of random fields, owing in part to the prominent role of the Gaussian and log-Gaussian in models of the initial density perturbations (Bardeen et al. 1986) and large-scale structure (Coles & Jones 1991), respectively—and indeed cosmologists have in the past led the way to development of key techniques for excursion set analysis and efficient simulation from Gaussian fields (e.g. Arnoud Doucet credits Hoffman & Ribak 1991 for introducing one of the most widely used tricks for sampling the posterior at indirectly constrained locations).  However, in the past decade the thrust of new ‘applied statistics’ ideas in random field theory have been driven by geostatistical research, where the embedding of Gaussian fields within the Bayesian hierarchical modelling framework has proven an exceptionally powerful technique for mapping continuous spatial signals (cf. Diggle et…

View original post 627 more words

One Response to “Six key trends in contemporary statistics that really could revolutionise astronomical data analysis …”

  1. I’ve lately been struggling to understand what I think is a convergent evolution between some ideas in cosmological statistics and INLA. Not the SPDE connection between GPs and GMRFs, but rather the strategy for fast approximate integration over nuisance parameters: https://astrostatistics.wordpress.com/2015/07/24/on-path-integral-evidence-marginal-likelihood-computations/?preview_id=2145

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: