Mercurial > hg > ltpda
view m-toolbox/classes/+utils/@math/jr2cov.m @ 43:bc767aaa99a8
CVS Update
author | Daniele Nicolodi <nicolodi@science.unitn.it> |
---|---|
date | Tue, 06 Dec 2011 11:09:25 +0100 |
parents | f0afece42f48 |
children |
line wrap: on
line source
% JR2COV Calculates coefficients covariance matrix from Jacobian and Residuals. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % DESCRIPTION % % Calculates coefficients covariance matrix from Jacobian and % Residuals. The algorithm uses the QR factorization of J to perform % the calculation inv(J'J)*s wehre J is the Jacobian matrix and s is % the mean squared error. % % CALL: % % covmat = jr2cov(J,resid) % % INPUT: % % J - Jacobian of the function with respect to the given coefficients % resid - Fit residuals % % % OUTPUT: % % covmat - covariance matrix of the fit coefficients % % Note: Resid should be a column vector, Number of rows of J must be equal % to the number of rows of Resid. The number of columns of J defines the % number of corresponding fit parameters for which we want to calculate the % covariance matrix % Estimate covariance from J and residuals. Look at the nlparci.m function % of the stats toolbox for further details % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % VERSION: $Id: jr2cov.m,v 1.2 2009/03/27 14:48:54 luigi Exp $ % % HISTORY: 26-03-2009 L Ferraioli % Creation %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% function covmat = jr2cov(J,resid) missing = isnan(resid); if ~isempty(missing) resid(missing) = []; end n = length(resid); J(missing,:) = []; if size(J,1)~=n error('The number of rows of J does not match the number of rows of RESID.'); end [n,p] = size(J); v = n-p; % degrees of freedom of the parameters estimation % Approximation when a column is zero vector temp = find(max(abs(J)) == 0); if ~isempty(temp) J(temp,:) = J(temp,:) + sqrt(eps(class(J))); end % Calculate covariance matrix [Q,R] = qr(J,0); Rinv = R\eye(size(R)); mse = norm(resid)^2 / v; % mean square error covmat = mse * Rinv*Rinv'; end