0
$\begingroup$

Is it conventional to use superscript notation to denote the inverse of a block matrix. The context is the quadratic expression in the exponential of a multivariate Gaussian probability density distribution:

$$-\frac{1}{2}({\bf X - \mu})^T\, \Sigma^{-1}\, ({\bf X - \mu})$$

The covariance matrix $\Sigma$ is a block matrix:

$$\Sigma = \begin{bmatrix} \Sigma_{11} & \Sigma_{12} \\ \Sigma_{21} & \Sigma_{22} \end{bmatrix}$$

The question is how accepted or conventional is it to express

$$\Sigma^{\color{red}{-1}} = \begin{bmatrix} \Sigma^{\color{red}{11}} & \Sigma^{\color{red}{12}} \\ \Sigma^{\color{red}{21}} & \Sigma^{\color{red}{22}} \end{bmatrix}$$

?

Equivalently, I guess I could be asking about common notation for the precision matrix.

$\endgroup$
0
$\begingroup$

I have never seen superscripts to denote blocks (or elements) of the inverse matrix. This does not indicate that it has never been used.

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.