Bernstein polynomial angular densities of multivariate extreme value distributions

To model the angular measure of a multivariate extreme value distribution, we develop a mean-constrained Bernstein polynomial over the (p - 1)-dimensional simplex, along with a generalization that places mass on the simplex boundaries. Simple componentwise adaptive Markov chain Monte Carlo algorithms for fitting the models to data derived from multivariate extremes is provided; the algorithms are implemented in code provided in the online supplementary content.