We define three classes of multivariate extreme value distributions (EVD) that are defined in arbitrary dimensions. We show that each class is dense, in the sense that any multivariate EVD can be approximated in distribution. We then show how to simulate from two of the classes and discuss estimation techniques using max linear combinations.

Keywords: extremes; multivariate extremes; simulation

Biography: John Nolan is a Professor of Mathematics and Statistics at American University in Washington, DC. He earned his PhD from the University of Virginia, and has taught at the University of Zambia, Kenyon College and American Univesity. His primary interest is in heavy tailed stable distributions. This work on multivariate exteme value laws is a consequence of the parallels between them and multivariate stable laws. Prof. Nolan is also one of the organizers of Math for America DC, a training program for math teachers in the Washington, DC schools.