Accuracy estimation of approximated Gaussian distribution obtained from Fast Forward Selection scenario reduction algorithm

Abstract

Probability distributions are used to represent uncertainty. One area of application of probability distribution is optimization under uncertainty more specifically known as Stochastic Integer Programming. Distributions with large number of scenarios increase computational complexity. Fast Forward Selection scenario (FFS) reduction algorithm provides a way to approximate the probability distribution. The paper applies FFS to a gaussian distribution and estimates the original distribution with lower number of scenarios while maintaining the overall variation of probability curve similar to the original curve. New probability density of the approximated distribution is close to the original distribution.