Coalitions are central to politics, at all levels. We discuss some mathematical results relating to the stability of coalitions and the probability of a decisive vote, with connections to the prisoner’s dilemma, agent-based modeling, and probability distributions on trees. Our empirical analysis suggests that the votes from small countries are overrepresented in E.U. elections.
La conférence sera ici.
You might wonder why I’m [ed. Andrew is] giving a political science talk to the group “Statistique et imagerie.” It’s because I’ve been told they’re more mathematicians than statisticians, and my research on coalitions and voting (see here and here) has fun mathematical content and interesting open problems as well as relevance to politics (as discussed in our paper, Standard voting power indexes don’t work: an empirical analysis).
Et après ma conférence, Bob Carpenter parlera sur Stan:
I [Bob] will describe the mechanisms underlying Stan’s probabilistic programming language, including
– blocks for data, parameter, and predictive quantities
– transforms of parameters to unconstrained spaces with Jacobians
– automatic computation of first- and higher-order derivatives
– operator, function, and linear algebra library
– vectorized density functions and cumulative distributions
– applications to MCMC and penalized MLE
And I’ll discuss our newest two features, rolling out soon:
– user-defined functions
– ordinary differential equation solvers
If anyone’s interested, I can also provide a bit of background on how the project is managed with a team of 11 developers and numerous user contributions. I can also go into arbitrary amounts of detail about the underlying implementations.
Update by Bob: This is very confusing with multiple first-person points of view in these posts. After my (Bob’s) talk, someone asked for the slides, so here they are:
I also rearranged the earlier text so Andrew’s comments come after his abstract.