The day started with Josh Tucker (NYU) talking about the SMaPP lab at NYU, where they are doing observational work in Politics and Economics using data science methods and in a lab-like structure. The science is new, but so is the detailed structure of the lab, which is not a standard way of doing Political Science! He pointed out that some PIs in the audience have larger budgets for their individual labs than the entire NSF budget for Political Science! He showed very nice results of turning political-science theories into hypotheses about twitter data, and then performing statistical falsifications. Beautiful stuff, and radical. He showed that different players (protesters, opposition, and oppressive regimes) are all using diverse strategies on social media; the simple stories about twitter being democratizing are not (or no longer) correct.
In the afternoon, we returned from Cle Elum to UW, where I discussed problems of inference in exoplanet science with Foreman-Mackey, Elaine Angelino (UCB), and Eric Agol (UW). After we discussed some likelihood-free inference (ABC) ideas, Angelino pointed us to the literature on probabilistic programs, which seems highly relevant. In that same conversation, Foreman-Mackey pointed out the retrospectively obvious point that you can parameterize a positive-definite matrix using its LU decomposition and then never have to put checks on the eigenvalues. Duh! And awesome.
"In that same conversation, Foreman-Mackey pointed out the retrospectively obvious point that you can parameterize a positive-definite matrix using its LU decomposition and then never have to put checks on the eigenvalues. Duh! And awesome."
ReplyDeleteI wonder how many people have rediscovered this over the years. In magnetronhierarchy, Murray suggested it, but I wanted to use an alternative that I had thought of. A day later, while improving my alternative, Murray pointed out that I was just writing down the LU decomposition.
With respect to the parameterisation of a positive-definite matrix, I have found the Stan manual to be a treasure-trove of constrained parameterisations like this: http://mc-stan.org/documentation/ (Chapter 55). If you want code, I have implemented a number of these parameterisations in Python at https://github.com/farr/plotutils/blob/master/plotutils/parameterizations.py .
ReplyDelete