Date of Award

4-2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Mathematical Sciences

First Advisor

Munevver Mine Subasi

Second Advisor

Luis Daniel Otero

Third Advisor

Jewgeni Dshalalow

Fourth Advisor

Muzaffar Shaikh

Abstract

The contribution of the shape information of the underlying distribution in probability bounding problem is investigated and an efficient linear programming based bounding methodology, which takes advantage of the advanced optimization techniques, probability theory, and the state-of-the-art tools, to obtain robust and efficiently computable bounds for the probabilities that at least k and exactly k-out-of-n events occur is developed. The k-out-of-n type probability bounding problem is formulated as linear programs under the assumption that the probability distribution is unimodal. The dual feasible bases structures of the relaxed versions of linear programs involved are fully described. The bounds for the probability that at least k and exactly k-out-of-n events occur are obtained in the form of formulas. A dual based linear programming algorithm is proposed to obtain bounds as the customized algorithmic solutions of the LP’s formulated. Numerical examples are presented to show that the use of shape constraint significantly improves on the bounds for the probabilities that at least k and exactly k-out-of-n events occur when only first a few binomial moments are known. An application in PERT, where the shape of the underlying probability distribution can be used to obtain bounds for the distribution of the critical path length, is presented.

Included in

Mathematics Commons

Share

COinS