Wow, it really feels like no research gets done (by me) any more. But I do go to a lot of good seminars! Today it was Dani Maoz (Tel Aviv) who gave a nice talk about inferring the type-Ia supernova delay function from observations. He came up with four methods (really five if I count one that was in his talk but he didn't separate out as a separate method), and they all (but that one) agree that most of the type Ia supernovae go off promptly (within less than a billion years) after star formation. He has additional evidence—though weaker because it is more dependent on star-formation history fitting—for a longer tail to long delays. That is, there appear, observationally, to be two populations. There are many remaining puzzles and interesting bits of phenomenology worth following up there. This is all very important to the chemical history of the Universe and the formation of metal-rich systems like rocky planets and the like.
One of Maoz's methods involves a classical inverse problem, with two things related by a linear transformation. As my loyal reader knows, I think all the interesting things about these inverse problems lie in the transformation itself, not the inversion, but that is the subject for a beer-in-hand conversation with Maoz this weekend.