Fri Feb 22 22:13:27 CET 2013
For 1-1 functions, gaussian elimination is not necessary, as there is
always a way to write the transfer function a rational function,
i.e. without matrix invsersion / equation solving.
How does that work? This is a computation on the coefficients.. It
seems that in general this is still a matrix operation, i.e. to move
to the all-pole form:
a b c d
1 0 0 0
0 1 0 0
0 0 1 0
I.e. given a program
Another thing: in practice, most feedback is going to have 1 or 2
state vectors. Can't we just exploit that and special-case the
Even if it's special-cased, gaussian elimination might be a simpler
way to express it!