NNPDF / nnpdf

An open-source machine learning framework for global analyses of parton distributions.
https://docs.nnpdf.science/
GNU General Public License v3.0
29 stars 6 forks source link

Dressed Lepton QCD k-factor #1859

Open comane opened 7 months ago

comane commented 7 months ago

Hi @enocera , @cschwan I am looking at this dataset: ATLAS_DY_7TEV_CC, and in particular at the 66-116 GeV (central) region.

Problem: It seems that the "QCD" k-factor that we have is NNLO QCD + NLO EWK (see for instance on hep-data: https://www.hepdata.net/record/ins1502620 Table 48 and compare to CF_QCD_ATLAS_DY_7TEV_CC)

Why is this (possibly) a problem: The data central values are supposedly Born level data value (no FSR QED radiation), hence, no EWK corrections should be needed (https://www.wiki.ed.ac.uk/display/nnpdfwiki/2021+Gargnano+meeting+agenda?preview=/503329049/505521689/talk-schwan.pdf)

What is strange: If I multiplying the data central values with a dressing C factor and then recompute the chi2 with the same NNLO QCD + NLO EWK kfactor, the chi2 does not improve, rather it gets a bit worse (from 1.2 to 1.4)

cschwan commented 7 months ago

@comane to which C-factor file are you comparing exactly? Could you please link it here?

comane commented 7 months ago

The C-factor is CF_QCD_ATLAS_DY_7TEV_CC in theory 400, and in particular from row 7 to 19

****SetName: ATLASWZRAP11CC Author: Emanuele R. Nocera enocera@ed.ac.uk Date: August 2021 CodesUsed: MCFM v8 TheoryInput: NNPDF4.0 NNLO alphas(mZ)=0.118 PDFset: 210713-n3fit-001 Warnings: cfactor splitted, originally part of a concatenation of cfactors


1.03940 0.00000 1.03786 0.00000 1.03990 0.00000 1.03905 0.00000 1.02616 0.00000 1.02491 0.00000 0.99276 0.00000 0.99367 0.00000 0.99345 0.00000 0.99456 0.00000 0.99331 0.00000 0.99137 0.00000 0.99414 0.00000 1.00205 0.00000 1.00891 0.00000 1.01283 0.00000 1.01638 0.00000 1.02225 0.00000 0.99447 0.00000 0.99811 0.00000 0.99867 0.00000 1.00853 0.00000 1.01704 0.00000 1.02312 0.00000

cschwan commented 7 months ago

I don't see these numbers in https://www.hepdata.net/record/ins1502620?version=1&table=Table%2048, but rather numbers that are smaller, in agreement with my expectation that NNLO QCD is usually positive. Am I reading your information incorrectly?

comane commented 7 months ago

Hi @cschwan, sorry for the confusion I realise my comment might have been unclear. I am looking at row 7 to 19 of the CFactor I have printed. Those values seem to be the same as those of the table you have just linked above.

cschwan commented 7 months ago

Now I see it. That looks indeed very bad for us ...