sPlot / RooFit - Impact of defined range on sWeigths outside this range and fit-results in general

[hviemann]

Hi there,
I think I do have some kind of followup to this solved issue:

SPlot does not work with fixCoefRange

After doing a fit, I want to extract s-weights. I’ve noticed that when I apply the SPlot class it changes my yields; usually, the yields change very slightly and all is well. However, if I call model.fixCoefRange('fitrange'), the yields change dramatically and the s-weights no longer sum to 1. This is true even if the range boundaries equal that of the fit parameter. Is there a workaround? Details below. Here is the relevant output of a fit to 10,000 entries: Without calling fixCoefRange: b…

The questions concerns the Range() behaviour of RooFit and its impact on analysis results and sWeights.

Imagine some mass-distribution in a RooDataSet with a defined ‘fit-range’ as basis. The pdf is already fit. And with the same pdf the range-patched RooStats::SPlot is run.
Now I enrolled some tests

• If I use ‘reduce()’ of RooAbsPdf on my dataset before my initial fit I get different results. Shouldn’t this have no impact on the result, as a defined range is basically the same as the cut dataset? This leads me to the conclusion that events outside of the defined fit-range do have a strong impact on the fit that should only be enrolled inside the range.
• Checking my sWeights in the output dataset for events outside the defined fit-region of the RooStats::SPlot I see weights deviating from 0. I’m not sure if I might have mistaken the theory, but wouldn’t I expect them to be forced to 0 as my fitted function could strongly deviate from the dataset outside the fit-range?
• Testing both previous quotes on a sub-range of the fit-range strangely results into the same fit-results and sWeights

UPDATE: How should I handle weights for events outside the fit-region?

[StephanH]

At first look, this could be another range bug. Are you using multiple ranges? If yes, this could be the problem:
https://sft.its.cern.ch/jira/browse/ROOT-10518

[hviemann]

Hi Stephan,
no, I think my problem was more on the user side. I’m using a RooSimultaneous to fit. At some point I must have come across to fit like such:
pdf.fitTo(…, RooFit::SplitRange(), …)
I just checked with the example fit of the tutorials (https://root.cern.ch/root/html/tutorials/roofit/rf501_simultaneouspdf.C.html) but this SplitRange() option is not coded there. Not sure when I picked this up, but this somehow let the fit ignore the given range via the ‘Range()’ option.

Cheers
Harald