In this dissertation we investigate Lagrange interpolation. Our first result will deal with a hierarchy of interpolation schemes. Specifically, we will show that given a triangular array of points in a regular compact set K, such that the corresponding Lebesgue constants are subexponential, one always has the uniform convergence of Ln(f) to f for all functions analytic on K. We will then show that uniform convergence of Ln(f) to f for all analytic functions f is equivalent to the fact that the probability measures γn = 1/n Σn j=1 δzn,j , which are associated with our triangular array, converge weak star to the equilibrium distribution for K. Motivated by our hierarchy, we will then come to our main result, namely that the Lebesgue constants associated with Leja sequences on fairly general compact sets are subexponential. More generally, considering Newton interpolation on a sequence of points, we will show that the weak star convergence of their corresponding probability measures to the equilibrium distribution, together with a certain distancing rule, implies that their corresponding Lebesgue constants are sub-exponential.

We present a numerical inversion method for generating random variates from continuous distributions when only the density function is given. The algorithm is based on polynomial interpolation of the inverse CDF and Gauss-Lobatto integration. The user can select the required precision which may be close to machine precision for smooth, bounded densities; the necessary tables have moderate size. Our computational experiments with the classical standard distributions (normal, beta, gamma, t-distributions) and with the noncentral chi-square, hyperbolic, generalized hyperbolic and stable distributions showed that our algorithm always reaches the required precision. The setup time is moderate and the marginal execution time is very fast and the same for all distributions. Thus for the case that large samples with fixed parameters are required the proposed algorithm is the fastest inversion method known. Speed-up factors up to 1000 are obtained when compared to inversion algorithms developed for the specific distributions. This makes our algorithm especially attractive for the simulation of copulas and for quasi-Monte Carlo applications. (author´s abstract) / Series: Research Report Series / Department of Statistics and Mathematics

We present a numerical inversion method for generating random variates from continuous distributions when only the density function is given. The algorithm is based on polynomial interpolation of the inverse CDF and Gauss-Lobatto integration. The user can select the required precision which may be close to machine precision for smooth, bounded densities; the necessary tables have moderate size. Our computational experiments with the classical standard distributions (normal, beta, gamma, t-distributions) and with the noncentral chi-square, hyperbolic, generalized hyperbolic and stable distributions showed that our algorithm always reaches the required precision. The setup time is moderate and the marginal execution time is very fast and nearly the same for all distributions. Thus for the case that large samples with fixed parameters are required the proposed algorithm is the fastest inversion method known. Speed-up factors up to 1000 are obtained when compared to inversion algorithms developed for the specific distributions. This makes our algorithm especially attractive for the simulation of copulas and for quasi-Monte Carlo applications. <P> This paper is the revised final version of the working paper no. 78 of this research report series. / Series: Research Report Series / Department of Statistics and Mathematics

This Online Supplement summarizes our computational experiences with Algorithm NINIGL presented in our paper "Random Variate Generation by Numerical Inversion when only the Density Is Known" (Report No. 90). It is a numerical inversion method for generating random variates from continuous distributions when only the density function is given. The algorithm is based on polynomial interpolation of the inverse CDF and Gauss-Lobatto integration. The user can select the required precision which may be close to machine precision for smooth, bounded densities; the necessary tables have moderate size. Our computational experiments with the classical standard distributions (normal, beta, gamma, t-distributions) and with the noncentral chi-square, hyperbolic, generalized hyperbolic and stable distributions showed that our algorithm always reaches the required precision. The setup time is moderate and the marginal execution time is very fast and nearly the same for all these distributions. Thus for the case that large samples with fixed parameters are required the proposed algorithm is the fastest inversion method known. Speed-up factors up to 1000 are obtained when compared to inversion algorithms developed for the specific distributions. Thus our algorithm is especially attractive for the simulation of copulas and for quasi-Monte Carlo applications. / Series: Research Report Series / Department of Statistics and Mathematics