Journal of Mathematical Sciences: Advances and Applications Volume 43, 2017, Pages 89-101 Available at http://scientificadvances.co.in DOI: http://dx.doi.org/10.18642/jmsaa_7100121761 2010 Mathematics Subject Classification: 62G32, 62N99, 62G20. Keywords and phrases: Archimax copula, generator, extreme values, generalized Pareto distribution. Received November 28, 2016  2017 Scientific Advances Publishers ON ARCHIMAX COPULAS AND MULTIVARIATE GPD MODELS BARRO DIAKARYA and SOUMAÏLA MOUSSA Université Ouaga II 12 BP: 417 Ouagadougou Burkina Faso e-mail: dbarro2@gmail.com FST, Dépt de Maths et Info Université AM BP 10662 Niamey Niger Abstract This paper develops an extension to three dimensional study some properties of Archimax copulas. Moreover, multivariate generalized Pareto distributions are characterized by a pseudo-dependence function while usual multivariate extremal models are given with the corresponding dependence function. 1. Introduction The concept of Archimedean copula is inherently related to Archimedean generator. So, the class of Archimedean copulas form a parameterized by real-valued functions. An n-dimensional copula C is one-dimensional generator functions (see [4]). Arising in the context of