Speaker: Rayna Georgieva, IPP, Bulgarian Academy of Sciences
Location: Room G74, Lyle Building
Time/Date: 13:00-14:00 on 22nd March 2007
Abstract: Sensitivity analysis is a procedure for investigation the dependence of a given model output upon the input parameters. It is applied in various fields where mathematical models are developed as environmental modeling, financial applications, risk analysis, signal processing, neural networks. Sensitivity analysis is a powerful technique used to determine robustness, reliability and efficiency of a model. The main problem in this procedure is the evaluating global sensitivity indices that measure a parameter’s main effect and all the interactions (of any order) involving that parameter. A brief review of known approaches for evaluating global sensitivity indices will be given. Their computational complexity (in terms of required operations) has been compared. The talk focuses on the Sobol’ sensitivity indices, whose evaluating requires multidimensional integration. Various algorithms (numerical quadratures, Monte Carlo and Quasi-Monte Carlo techniques) for multidimensional integration have been investigated. Some numerical experiments for evaluating integrals of different dimensions have been performed.
<a href=”slides_reading07.pdf”>Slides</a>
Valid XHTML | Copyright | Last Modified: 22/Mar/2007 |