Uncertainty quantification for Markov chain models

Hadi Meidani, Roger Ghanem

Research output: Contribution to journalArticlepeer-review

Abstract

Transition probabilities serve to parameterize Markov chains and control their evolution and associated decisions and controls. Uncertainties in these parameters can be associated with inherent fluctuations in the medium through which a chain evolves, or with insufficient data such that the inferential value of the chain is jeopardized. The behavior of Markov chains associated with such uncertainties is described using a probabilistic model for the transition matrices. The principle of maximum entropy is used to characterize the probability measure of the transition rates. The formalism is demonstrated on a Markov chain describing the spread of disease, and a number of quantities of interest, pertaining to different aspects of decision-making, are investigated.

Original languageEnglish (US)
Article number043102
JournalChaos
Volume22
Issue number4
DOIs
StatePublished - Oct 4 2012
Externally publishedYes

ASJC Scopus subject areas

  • Statistical and Nonlinear Physics
  • Mathematical Physics
  • Physics and Astronomy(all)
  • Applied Mathematics

Fingerprint Dive into the research topics of 'Uncertainty quantification for Markov chain models'. Together they form a unique fingerprint.

Cite this