Error Determination

If errors are required the command:

$mxf_do_errors $on

should be used. To determine the error on the fit for distribution 1, for example, the value for distribution 1 is set away from the best fit value and the minimisation is repeated allowing all the other distributions to vary. In iteration is carried out searching for the point where the returned maximum likelihood is $(L_{max}-0.5)$, though the error contours can be changed through the command:

$mxf_error_cont  0.6   ! look for L-0.6 contours 
                       ! instead of L-0.5 for errors

At present the algorithm for this iteration is not particularly efficient.