CMStatistics 2017: Start Registration
View Submission - CMStatistics
B0710
Title: Efficient simulation of Brown-Resnick processes by means of locally equivalent log-Gaussian representations Authors:  Kirstin Strokorb - Cardiff University (United Kingdom) [presenting]
Marco Oesting - University of Siegen (Germany)
Abstract: Among max-stable processes, the class of Brown-Resnick processes seems particularly attractive for practical use. Several algorithms have been suggested for its simulation. While the efficiency of such algorithms can be viewed as a function of the number of points on which the process is simulated, it is more difficult to understand the role of the simulation domain in which they lie, on which we focus in this project. Our suggestion improves the efficiency and/or accuracy of a stopping-time based algorithm. We show that a relatively simple and non-costly adjustment of previous algorithms can be very beneficial.