Editing the SPlots in RooStats

RooStats currently offers the possibility of creating sPlots from models with fixed yields in them. However, the current implementation does not differentiate between varying and fixed yields, which results in calculating the wrong covariance matrix (that is, one with also the fixed yields in them).
This issue has been documented and a fixed has been proposed in an internal LHCb note (https://cds.cern.ch/record/2139626?ln=fr for those who can see it).
How can I propagate this to the base ROOT code and are there further tests needed?

Cheers,
Louis

Hi,

Unfortunately the node is not visible. If possible it would be great to upload the PDF file or just send directly to me if you don’t want to make it public.
Then the best way to commit fixes in ROOT is to submit PR in GitHub

Cheers

Lorenzo

Hi,
Whereas there is no sensitive information in that note, I will ask more knowledgeable people than me if that can be shared, and come back to you.
Cheers,
Louis.

main.pdf (258.7 KB)
So here is the note relative to that issue.

Hi,

Thank you very much for the note ! How do you want to proceed ? If you have the time to provide me a PR for the required changes will be perfect. Also having as well a test program to be committed in GitHub showing the issue will be very good.

Best Regards

Lorenzo

Hi,
Apologies for the delay. Ideally I would have provided the code that is documented in the note, as I have no experience with PRs, would that suit you?
Concerning tests, the smoking gun for that problem is that sum of coefficients is one in cases where it should not be. I do not know however if that is enough for a test program, as the issue is rather subtle.
Cheers,
Louis

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