Is Matrix a Hilbert vector? 
what is the correct dot product? 
what is the correct norm? 

I definitely want a Matrix container that contains 
several RnVecs (columns and rows). 
It should provide pivoting, linear combinations of rows and 
colums, and other steps needed for 
the standard direct Matrix inversion methods. 
The matrix space should always be a deep copy and should keep track 
of the pivoting and linear combinations. Should it? 

