No confidence bounds in a nonlinear fit

I am using the fit function to perform a nonlinear general fit. The fit model is a user defined model with 8 parameters and it usually works well. The problem is that in some cases I can't get confidence bounds for the fit. I obtain confidence intervals with confint function and I check all the outputs of the fit: fit, gof and output, and check output to see what happened. I have different cases without confidence intervals: cases with exitflag = 3 cases with exitflag = 4 cases with exitflag = 0 (success, the fit process converged) And I also have cases with exitflag = 3, in which I can obtain the confidence intervals with confint.

I don't understand why I can't get the conf. intervals in some cases, while in other similar situations, with the same exitflag, I got them.

Any idea? Thanks in advance,

Edi

Answer 1

[Star Strider]

I don’t have the Curve Fitting Toolbox, but my first thought would be to check the covariance matrix. The condition number of the covariance matrix will probably explain it.