Estimated distance to minimum

Dear experts,

I would like to understand what is exactly
“estimated distance to minimum” and how it is calculated.

Second questions is the following:

I run unbinned fit and see in my printout lines lines like

Info in : VariableMetricBuilder 3 - FCN = -0.002022590529 Edm = 0.0009089154139 NCalls = 958

after some iterations fit is converged and print me best fit parameters after the line

Info in : VariableMetricBuilder 6 - FCN = -0.003814485944 Edm = 7.009636942e-05 NCalls = 1758

The after parameters printout I have this following

RooFitResult: minimized FCN value: -0.00381449, estimated distance to minimum: 0.000176492
covariance matrix quality: Full, accurate covariance matrix
Status : MINIMIZE=0 HESSE=0

so, there is a difference between “estimated distance to minimum: 0.000176492” and “Edm = 7.009636942e-05…” Why it is so ?

Question 3. Sometime, very rare I have very large value of Edm (for example 50) and unphysical result of fit : Delta NLL is very large which can’t be in my case. Is it due to fit can’t find a good minimum ?

Thank you, Sasha.

Thank you for your question. Let me add @jonas who is our RooFit expert.

Cheers,
Dev

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.