We have these two models :
z = h(x) + e, r1 = z - z_hat = h(x) - h(x_hat)
z = Bx + e, r2 = z - z_hat = Bx - Bx_hat
in the first equation x is estimated by nonlinear weighted least square and second one is estimated by linear weighted least square.
1) How the covariance matrix of r1 and r2 are calculated?('e' follows a Gaussian distribution with zero mean and covariance matrix R).
2)In Linear weighted least square, if B and x are complex matrix and vector respectively, what happens for residual covariance matrix?