0
$\begingroup$

Suppose we have the following mixed model:

$y = \boldsymbol{X \beta} + \boldsymbol{Z \gamma} + \epsilon$

Where:

$\mathbf{X}$ is the fixed effect design matrix.

$\boldsymbol{Z}$ is the design matrix of RVs.

$\boldsymbol{\gamma}$ is the vector of random effect parameters.

$\epsilon$ Is not assumed to be independent and homogenous.

In my SAS notes when they calculate the solution to the beta coefficients in this mixed model, we see the following notation.

$\hat{\boldsymbol{\beta}} = (\boldsymbol{X'\hat{V}^{-1}X})^{-} \boldsymbol{X'\hat{V}y}$

Where:

$\hat{\boldsymbol{V}} = \boldsymbol{Z \hat{\boldsymbol{G}}Z' + \hat{R}} $

But im unsure about the notation $\boldsymbol{X^{-}}$. What does the minus in the exponent of the matrix $(\boldsymbol{X'\hat{V}^{-1}X})^{-}$ mean?. Or is simply an error and should be the inverse?

$\endgroup$
1
  • 1
    $\begingroup$ It's a generalized inverse, used here because $\boldsymbol{X'\hat{V}^{-1}X}$ may not be invertible. $\endgroup$ Nov 24 at 18:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.