TMinuit and Asymmetric Errors

Dear Rooters,
In my fitter, I call TMinuit and extract the asymmetric Minos errors associated with the different parameters. If one of the parameters is fixed in the fit, I notice that the index of the parameters is not correct. Here is an example:
Fit parameters: a, b, and c, with index: 0, 1, 2 if I fix "b" in the fit, then TMinuit->fErp[1] is corresponding to the negative Minos error associated with the parameter "c". Have you seen this before?
Thanks,

             Arafat Received on Thu Mar 11 2010 - 19:03:55 CET