Robust Estimation of Mixture Constants
Abstract
The problem of estimating the mixture constant for a mixture process is investigated. The asymptotic properties of M-Estimators are discussed. Conditions are given under which such estimators are asymptotically normal and consistent. It is shown that the maximum likelihood estimate and estimates based on moments satisfy these conditions. For instances in which there is uncertainty about the distributions composing the mixture distribution the robust estimator is found to be a censored version of the nominal MLE nonlinearity. Some numerical results on the existence of robust solutions are presented.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 27, 1987
- Accession Number
- ADA204890
Entities
People
- Douglas J. Warren
- John B. Thomas
Organizations
- Princeton University