Notation in linear algebra and rule for square of matrix norm

2 replies · 4K views · last reply
lishrimp
Messages
2
Reaction score
0
Hi.

I have a few simple questions.

Printed equation chain: E_CWLS = sum over i of ||r_i||^2 with a red circle around the subscript Sigma_i^-1, equals sum over i of r_i^T Sigma_i^-1 r_i, equals sum over i of sigma_n^-2 r_i^T A_i r_i.
(<- sorry, please click this image.)

1. What does the notation in the red circle mean?

2. Is there a rule for expanding square of norm? (e.g. || A*B*C ||^2)
I don't really understand how the first eq. changes to the second eq.

Thanks. :)
 
Hi lishrimp! :smile:

lishrimp said:
Hi.

I have a few simple questions.

View attachment 37815 (<- sorry, please click this image.)

1. What does the notation in the red circle mean?

2. Is there a rule for expanding square of norm? (e.g. || A*B*C ||^2)
I don't really understand how the first eq. changes to the second eq.

Thanks. :)

My guess:

If [itex]\Sigma[/itex] is a matrix, then we can define a "norm" [itex]\|~\|_\Sigma[/itex] by setting

[tex]\|\mathbf{x}\|_\Sigma=\sqrt{\mathbf{x}^T\Sigma \mathbf{x}}[/tex]

In your case, the matrix is [itex]\Sigma^{-1}[/itex], so the norm is

[tex]\|\mathbf{x}\|_{\Sigma^{-1}}=\sqrt{\mathbf{x}^T\Sigma^{-1} \mathbf{x}}[/tex]

So

[tex]\|\mathbf{x}\|_{\Sigma^{-1}}^2=\mathbf{x}^T\Sigma^{-1} \mathbf{x}[/tex]

So that equality is true by definition.