Possible inconsistency in gsl_multifit_covar

[Giulio Bottazzi <giulio dot bottazzi at libero dot it>]

Hello,
I've noticed that the matrix returned by the
function "gsl_multifit_covar" is defined as

(J^T J)^{-1}


Now the problem is that this matrix IS NOT the
(approximated) estimated variance-covariance matrix of the asymptotic
normal distribution of the estimates. This should indeed be defined as

\sigma^2 (J^T J)^{-1}

where

\sigma^2 = \sum_j res_j / (N-K)

res being the residuals, N-K the degrees of freedom.

This behavior of gsl_multifit_covar is strange, first because calling a
"covariance matrix" what the function returns sound a little bit
deceiving and second because it is inconsistent with the covariance
matrix returned by the GSL linear routines (which contain the proper
factor \sigma^2). 

Is this a feature or a bug?

Best,
	Giulio.

Attachment: pgp00000.pgp
Description: PGP signature