Linear and
non-linear fits usually end with
At the end of the minimization, BJ the first
derivative of c2(c) with respect to cJ, is zero. The
non-linear equations are linearized making them
equivalent to linear equations in this limit.
The matrix A-1 is closely related to the error matrix which
can be used to find the standard deviations in any function of c.
A test of the standard deviation is detailed in ..\Weights\ArtDat\Sum.doc
and
..\nlfit\FermiFit\FermiStd.htm
– For full details go to ..\nlfit\FermiFit\Welcome.htm
This last link contains a zip with the wpj, codes and inputs needed to find the standard deviation
in a Fermi function after fitting the derivative of
the Fermi function to data.