More fun with Derivatives of Matrices

[PoulsenL at capecon.com]

The problem is that I do not have a single function for which I can take the
derivative.  I have broken the problem up into a series of subfunctions that
are called and put together.  So I used the Scientific package to "cheat"
and save some time and avoid writing out the single monolithic matrix
operation. This is extremely convenient for me because my first order
condition can change slightly within any one of the subfunctions and it
avoids me having to write several different cases in matrix notation. This
was a perfect solution except for my inability to invert the matrix when it
contained a DerivVar object.  

Also, I cannot use the formula for the derivative of a matrix, because it is
only part of the larger equation that I am trying to avoid working out the
analytical derivative of in the first place.  Also I have broken up taking
the first derivative in parts because there are over 300 parameters and I
can't do it all at once.  Again DerivVar work very well for this, but for
the inversion problem.

Thank you for your help so far,

Loren