Simon J Murphy (Sydney) is in town for two weeks of hacking with Dan Foreman-Mackey (Flatiron). On arrival last week, the two of them implemented something I have been wanting to do for a long time, which is use asteroseismic phase shifts to find binary companions (yes, people have done this for a while now) but without binning the data up or ever explicitly measuring any time delays in bins or at times. This week (having solved that) they are looking at radial-velocity predictions from those discoveries, and testing them with HIRES spectra. They teamed up with Megan Bedell (Flatiron) to use her wobble system to make these measurements. All I did was cheer-lead.
In the afternoon, Alex Malz (NYU) and I discussed what we might do in an upcoming LSST transient classification challenge. I am interested in the following question: Say you have two sparsely and irregularly sampled light-curves of two transient events that are intrinsically similar but maybe at different redshifts, and you want to see that they are similar. How do you construct a relevant, useful, and tractable similarity or distance metric? I have lots of ideas; if we can solve this, we might have something to contribute.