Showing a limited preview of this publication:
Abstract
The standard Monte Carlo algorithm for estimating global sensitivity indices may be spoilt by loss of accuracy if the index is very small. Two approaches were proposed for eliminating the loss of accuracy: reduction of the mean value and correlated sampling. In the present paper both approaches are investigated and a third combined approach is suggested.
Received: 2007-05-02
Revised: 2007-07-15
Published Online: 2008-02-13
Published in Print: 2008-01
© de Gruyter 2007