Arithmetization of distributions and linear goal programming
This is an interesting article about arithmetization of probability distribution functions (PDF). It can be viewed as an extension of Panjer's algorithm used in the numerical evaluation of PDFs [see \textit{H. H. Panjer} and \textit{G. E. Willmot}, Insurance risk models. Society of Actuaries, Schaumburg (1992)]. When trying to replace a discrete or mixed type PDF by an arithmetic one, it can happen that the local moments matching method up to the second order fails for every appropriate span, giving some negative mass. A possible solution to this problem is retreating to first order local moment matching. This article explores an alternative approach, consisting in looking for the arithmetic probability function (PF) with equal expectation and closest second moment to the original PDF. The determination of this arithmetic PF is done by means of linear goal programming, which makes possible the junction of both the ideas of moment matching and closest second moment in a single method called the method of the nearest second moment.
- scientific article; zbMATH DE number 5929604
- Linear Approximations of Probability Density Functions
- Moment-based approximations of discrete probability distributions using rational functions
- Matching moments to phase distributions: nonlinear programming approaches
- Discrete Approximations to Continuous Univariate Distributions—an Alternative to Simulation
This page was built for publication: Arithmetization of distributions and linear goal programming
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1584590)
