Computational challenges of t and related copulas

Abstract

<p>The present paper addresses computational and numerical challenges when working with <em>t</em> copulas and their more complicated extensions, the grouped <em>t</em> and skew <em>t</em> copulas. We demonstrate how the R package nvmix can be used to work with these copulas. In particular, we discuss (quasi-)random sampling and fitting. We highlight the difficulties arising from using more complicated models, such as the lack of availability of a joint density function or the lack of an analytical form of the marginal quantile functions, and give possible solutions along with future research ideas.</p>