CMS_Z0J_13TEV

[enocera]

This PR incorporates the commondata implementation for the data set in the title.

[scarlehoff]

I think there's something wrong with the kinematics.

This might be a good opportunity to start using m_ll instead of the square...

RE data Vs theory something is wrong with the first bin (this is theory / data), it is systematically 30% off (probably some normalization or I did wrong a cut in my theory predictions).

The second bin and third bins seem ok.

Third and fourth a bit worse, I guess the statistics are all around the place but so is the data:

ratio = array([0.63142557, 0.69191467, 0.72222455, 0.72407745, 0.76986966,
       0.82255564, 1.04158693, 1.03487063, 1.01877841, 1.01176393,
       1.01778132, 1.00585585, 1.00522577, 0.99522485, 0.99550279,
       1.00101172, 1.00696618, 1.01148373, 1.02241088, 1.0466769 ,
       1.06189633, 1.07604775, 1.06241513, 1.1156114 , 1.03361906,
       0.98315902, 0.97514231, 0.97673345, 0.93802319, 1.00462591,
       0.98311333, 0.97416855, 0.88055417, 1.07630434, 1.00228661,
       0.8514168 ])

And the plots for the 2nd, 3rd and 5th bins (disregard the value of the mass which is wrong and filled just to get different plots)

[enocera]

I think there's something wrong with the kinematics.

This might be a good opportunity to start using m_ll instead of the square...

RE data Vs theory something is wrong with the first bin (this is theory / data), it is systematically 30% off (probably some normalization or I did wrong a cut in my theory predictions).

The second bin and third bins seem ok.

Third and fourth a bit worse, I guess the statistics are all around the place but so is the data:

ratio = array([0.63142557, 0.69191467, 0.72222455, 0.72407745, 0.76986966,
       0.82255564, 1.04158693, 1.03487063, 1.01877841, 1.01176393,
       1.01778132, 1.00585585, 1.00522577, 0.99522485, 0.99550279,
       1.00101172, 1.00696618, 1.01148373, 1.02241088, 1.0466769 ,
       1.06189633, 1.07604775, 1.06241513, 1.1156114 , 1.03361906,
       0.98315902, 0.97514231, 0.97673345, 0.93802319, 1.00462591,
       0.98311333, 0.97416855, 0.88055417, 1.07630434, 1.00228661,
       0.8514168 ])

And the plots for the 2nd, 3rd and 5th bins (disregard the value of the mass which is wrong and filled just to get different plots)

Let me check this.

[scarlehoff]

I'm a bit worried about this dataset still because the agreement of the second bin is quite good (which is also special since it is the one that contains the Z mass).

But, here's the problem, I'm not normalising by the bin width, and according to the hepdata info, it is normalized. So I'd like to have a second look.

fwiw, the differences I see (the sames as I put above) are clearly not a factor of the bin width either.

[scarlehoff]

Suggested change

	conversion_factor: 1000.0
	conversion_factor: 0.001

Data is in pb and theory in fb

[scarlehoff]

Suggested change

	description: "Z boson mass squared"
	label: '$m^2_{\ell \ell}$'
	units: "GeV^2"
	description: "invariant mass of the dilepton system"
	label: '$m_{\ell \ell}$'
	units: "GeV"

[enocera]

This PR has become very messy, as it incorporates changes well beyond those in the title. This PR is going to be split in three sub-PRs:

• one for the CMS_Z0J_13TEV data set Implementation of CMS_Z0J_13TeV - general cleanup #2468
• one for the fixed ATLAS_Z0J_13TEV data set - ATLAS_Z0J_13TEV - bugfix #2469
• one for variants of both data sets (not needed for NNDPF4.1) - to appear

[enocera]

@scarlehoff Concerning the second point above, as you may remember, we discovered that the commondata implementation was bugged because only a fraction of the pT bins were implemented and the uncertainty was one order of magnitude too large. So we really need a brand new data set implementation (this is not a variant, this is not a new observable of the existing data set, this is really a dataset implementation version 2, the bugged version 1 having been used in the pheno paper). In this case, I have to confess that I don't remember what the workflow is. Is it sufficient to erase the previous implementation and have a new one (with version: 2 in the metadata) or should I create a new folder called, say ATLAS_Z0J_13TEV_V1 or what? Thanks.

[scarlehoff]

Version 2 in the metadata, comment: fix to bug in version 1. If you want to keep the bug as a variant you can do that.

[enocera]

Version 2 in the metadata, comment: fix to bug in version 1. If you want to keep the bug as a variant you can do that.

Sorry @scarlehoff : are you sure that I can keep the variant? The variant will also differ for the kinematics (the number of pT bins has changed).

[scarlehoff]

Ah, no, in that case no. And then also fktables will differ.

I would just replsce it. The pheno paper version lives in the tags of the repo.

[enocera]

Very good. That's what I wanted to hear.