After taking a holiday in August with Lianne, I’ve spent the last three and a half weeks in Cambridge, UK, and have enjoyed it immensely. It’s been refreshing to focus entirely on research, as you can see from my Github green square thingy:
During this stay I’ve had many interesting discussions and ideas, which is the point. Here is a subset of highlights. Some of these may one day lead to interesting finished projects, but others may not:
i) Last year Geraint Lewis suggested I write a book, and I wasn’t too keen on the idea because I was very busy, but now I’ve started. The idea is to demystify and present fundamental Bayesian ideas in a simple way, accessible to any interested reader who is happy to do thinking and arithmetic.
ii) It’s been a number of years since I last worked on quasar microlensing, but I’ve always had in the back of my mind a few ideas for how to do the full inference problem (given the measured fluctuations of some quasar images, infer something about the stars in the lens galaxy, and the quasar source itself, etc). The ideas have been on the backburner for years because I’m out of touch with more recent developments in the field. Anyway, I wanted to suggest this to Nick Bate, who is still in the field. Coincidentally, he had an idea to pitch to me during my visit, which happened to be exactly the same thing. This was one of the best coincidences I can remember!
iii) Various other astronomy chats and data sharing with the folks at the Insitute of Astronomy.
iv) I’ve been discussing various new and old ideas related to Nested Sampling with three Cambridge “theoretical chemists”: Gábor Csányi, Livia B. Pártay, and Robert Baldock. I had met the first two at MaxEnt 2009 and they were my co-authors on Diffusive Nested Sampling. They have been doing fascinating and impressive research on doing classical statistical mechanics from first principles using NS, with applications in materials science. This is an area I’ve been wanting to get into. They have also built up a lot of experience with hard high dimensional sampling issues that make some of our astronomy inference problems look simple by comparison. There’s also a PhD student, Will Handley, working on next-generation NS algorithms at Cavendish. Will wasn’t aware of the chemistry group before but now is — hopefully this leads to some good innovations!
iv) I got to hang out with Steve Gull a few times and learn all sorts of stories about the history of Bayesianism in Cambridge (a lot of which centers around Steve!).