Sampling Techniques for Halphen Distributions

Abstract

In this paper we present algorithms based on the acceptance-rejection (A-R), and Markov chain Monte Carlo methods to draw samples from Halphen distributions. Quantiles computed with these generating techniques are compared to those given by the method of importance sampling. Results show that our choice of the instrumental distribution considered in the A-R, produce an efficient method to generate samples from Halphen distributions. The availability of such procedures makes it possible to approximate all integral quantities and to study statistical properties of parameters estimators in the case of small sample sizes usually encountered in hydrology.