Singular Value Decomposition syntax - use output from svd2() ?

[JXBWk]

To all,

I am playing around the 'Singular Value Decomposition' in mcad 15 (function sdv2). Follow the example and all good . Get S, U and V matrices What is the correct syntax to use the data to solve {y}=[A]{f}? see attached I know it’s a dumb question but the obvious syntax f = svd2(H)y does not work !

Thanks

[-MFra-]

You have to consider the fact that matrix A is not square (it is a (4 x 3) matrix) and y is a vector consisting of 4 rows. The product can only be done by considering the transpose of A. The sdv2 function does not exist. For example:

[JXBWk]

Thanks for the reply. So what is the point of function svd2()? I am reading some notes on structural dynamic testing and the use of SVD is common because of lack of data (or data incompletness)

If I use the transpose as per your example one does indeed get a f vector (solution) but if you check A.f one does not get the y vector one stared with !

Same with the geninv() function the check A.f does not return the y vector one started with. As far as I can tell they are not even a scale away form each other

[-MFra-]

The result is different from y because the matrix A is not square.