Negative inverse transposed matrix

[PhilipLeitch]

I have a text book example I can't translate into Mathcad.

They provide a 2x2 matrix as follows:

1 0.051
0.051 1

They then say to transform it by "transpose of the inverse of the square root" of the above matrix.

They use notation that is M followed by a superscript of -1/2

(M^-(1/2))`


The result should be:
1 -0.03
-0.03 1


I don't know how to square root a matrix, and square root of individual values doesn't provide the "correct" answer.

Any ideas?

Thank you.

Philip
___________________
Nobody can hear you scream in Euclidean space.

[AlvaroDíaz]

Use eigenvalues for generalize functions definitions from complex to matrix domain.

Regards. Alvaro.

[PhilipLeitch]

Thanks. That worked perfectly.

I wouldn't have been able to work that one out.

I'm using it to find the canonical coefficients of a canonical correlation. I can then use these correlation coefficients as part of a descriminant function analysis.

The book I have skims over the calculations because it assumes you just plug the figures into SPSS and get the answer.

Philip
___________________
Nobody can hear you scream in Euclidean space.

[alnstevens]

You could also use a Given ... Find solve block if you want to avoid using eigenvalues/vectors - see attached.

stv

[StuartBruff]

On 1/5/2010 9:20:26 AM, stv wrote:
== You could also use a Given ... Find solve block if you want to avoid using eigenvalues/vectors

Or the Denman-Beavers algorithm ...

Stuart

[PhilipLeitch]

Thanks for the various options.


Philip
___________________
Nobody can hear you scream in Euclidean space.

[ptc-1368288]

Maths-matix aren't always "true" and clear cut.
Book notation make them worst.
Your example seems to come out indirectly from:



jmG