General Gauss–Markov Models
- 1.2k Downloads
The standard linear model assumes the data vector has a covariance matrix of \(\sigma ^2 I\). Sections 2.7 and 3.8 extended the theory to having a covariance matrix of \(\sigma ^2 V\) where V was known and positive definite. This chapter extends the theory by allowing V to be merely nonnegative definite.
- Searle, S. R., & Pukelsheim, F. (1987). Estimation of the mean vector in linear models, Technical Report BU-912-M, Biometrics Unit. Ithaca, NY: Cornell University.Google Scholar