Large Scale Equality Constrained Optimization

[Metz]

Hi everybody,

I have got a  question concerning the large scale optimization capabilites of LabView:

I want to solve a problem of the form

MIN |D-CS|

S.T.: CA=b

where C is unknown and is not only a vector but a nxm dimensional matrix.

D and S are also higher-dimensional matrices, and A is a mx1 matrix.

Is there a possibility to use those matrices as parameters for a built-in optimization function directly, or maybe with a transformation via the Kronecker-product and the vec-operator?

Or does anybody know which external add-on package can handle such a high-dimensional problem directly?

Many thanks in advance for any helpful comments

Christian

[DSPGuy]

Christian,

You mention that the problem is really large.  How large?  Is your problem sparse?  Also, just want to understand your problem description better:  It looks like you are trying to minimize a matrix norm subject to equality constraints, is that correct?  Is the the 2Norm?  Can you share your application?

-Jim