The estimation of parameters in nonlinear, implicit measurement error models with experiment-wide measurements

Technical Report · Sun May 01 00:00:00 EDT 1994

DOI:https://doi.org/10.2172/10156430· OSTI ID:10156430

Anderson, K K

Measurement error modeling is a statistical approach to the estimation of unknown model parameters which takes into account the measurement errors in all of the data. Approaches which ignore the measurement errors in so-called independent variables may yield inferior estimates of unknown model parameters. At the same time, experiment-wide variables (such as physical constants) are often treated as known without error, when in fact they were produced from prior experiments. Realistic assessments of the associated uncertainties in the experiment-wide variables can be utilized to improve the estimation of unknown model parameters. A maximum likelihood approach to incorporate measurements of experiment-wide variables and their associated uncertainties is presented here. An iterative algorithm is presented which yields estimates of unknown model parameters and their estimated covariance matrix. Further, the algorithm can be used to assess the sensitivity of the estimates and their estimated covariance matrix to the given experiment-wide variables and their associated uncertainties.

• Research Organization: Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
• Sponsoring Organization: USDOE, Washington, DC (United States)
• DOE Contract Number: AC06-76RL01830
• OSTI ID: 10156430
• Report Number(s): PNL-9794; ON: DE94012759
• Resource Relation: Other Information: PBD: May 1994
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL PROTECTION
PARAMETRIC ANALYSIS
MATHEMATICAL MODELS
ARMS CONTROL
VERIFICATION
ERRORS
STATISTICS
ALGORITHMS
DATA COVARIANCES
MAXIMUM-LIKELIHOOD FIT
990200
350300
MATHEMATICS AND COMPUTERS