Some equivalences in linear estimation (in Russian)

Under normality, the Bayesian estimation problem, the best linear unbiased estimation problem, and the restricted least-squares problem are all equivalent. As a result we need not compute pseudo-inverses and other complicated functions, which will be impossible for large sparse systems. Instead, by reorganizing the inputs, we can rewrite the system as a new but equivalent system which can be solved by ordinary least-squares methods.

MoreLess

Year of publication:	2007
Authors:	Danilov, Dmitry ; Magnus, Jan R.
Published in:	Quantile. - Quantile. - 2007, 3, p. 83-90
Publisher:	Quantile
Subject:	Linear Bayes estimation \| best linear unbiased \| least squares \| sparse problems \| large-scale optimization

More details

Extent:	application/pdf
Type of publication:	Article
Language:	English
Classification:	C11 - Bayesian Analysis ; C61 - Optimization Techniques; Programming Models; Dynamic Analysis ; C63 - Computational Techniques
Source:	RePEc - Research Papers in Economics

Persistent link: https://www.econbiz.de/10005569955