How to draw Minuit contour?

I have a fit function with two free parameters. For the fit I’m using user defined FCN which is given to Minuit for the minimization. At the end I’m using gMinuit->Contour(80,0,1); to draw 1 sigma confidence contour for parameter 1 vs parameter 2.

I would like to compare my contour to a contour in a article. My question is how to draw a confidence contour for two parameters, if I know values and errors of both parameters, and correlation coefficient between the parameters?
I was looking how to give those external parameters to Minuit and then call “Contour” for calculation points of ellipse.

Please help.

Thanks in advance,


Minuit, to get the correct contour, uses the full likelihood (or chi2) function.
In your case, if you have just errors and correlation coefficient, you could assume that the log-likelihood (or chi2) is parabolic and derive the contour from that parabola.
For a given confidence level, the contour is obtained from the values for which the log-likelihood (or chi2) has a certain value.
See for example … nerror.pdf
or chapter 2 of this book, Data Analysis in High Energy Physics: A Practical Guide to Statistical Methods.

Best Regards