NCD TTB CMS

[t7phy]

No description provided.

[scarlehoff]

Thanks! Below the comparison looking as usual to predictions / data / covmat / t0 covmat. To ensure that I didn't pick up differences in the fktables, I used the new ones always.

For CMS_TTBAR_8TEV_2L_DIF I see differences of a few %
As with the #1834, it might be that the new numbers are expected to be correct and the old ones are wrong but we need to understand why if that's the case.

Note that not all of these entered in NNPDF4.0 so I'm obviously less worried about the ones that didn't...

 > Testing old: CMS_TTBAR_2D_DIFF_PT_TRAP_NORM vs new: CMS_TTBAR_8TEV_2L_DIF_PTT-YT-NORM
# Problem in the data comparison of old: CMS_TTBAR_2D_DIFF_PT_TRAP_NORM vs new: CMS_TTBAR_8TEV_2L_DIF_PTT-YT-NORM
Relative differences above 1e-2
 > Testing old: CMS_TTBAR_2D_DIFF_MTT_TRAP_NORM vs new: CMS_TTBAR_8TEV_2L_DIF_MTTBAR-YT-NORM
# Problem in the data comparison of old: CMS_TTBAR_2D_DIFF_MTT_TRAP_NORM vs new: CMS_TTBAR_8TEV_2L_DIF_MTTBAR-YT-NORM
Relative differences above 1e-2
 > Testing old: CMS_TTBAR_2D_DIFF_MTT_TTRAP_NORM vs new: CMS_TTBAR_8TEV_2L_DIF_MTTBAR-YTTBAR-NORM
# Problem in the data comparison of old: CMS_TTBAR_2D_DIFF_MTT_TTRAP_NORM vs new: CMS_TTBAR_8TEV_2L_DIF_MTTBAR-YTTBAR-NORM
Relative differences above 1e-2

For CMS_TTBAR_8TEV_LJ_DIF there are some differences in the t0 chi2 which points to maybe a difference in MUL / ADD treatment

 > Testing old: CMSTOPDIFF8TEVTRAPNORM vs new: CMS_TTBAR_8TEV_LJ_DIF_YT-NORM
The t0 chi2 is different: 878.34 vs 894.87
 > Testing old: CMSTOPDIFF8TEVTTRAPNORM vs new: CMS_TTBAR_8TEV_LJ_DIF_YTTBAR-NORM
 > Everything ok!
 > Testing old: CMSTOPDIFF8TEVTTMNORM vs new: CMS_TTBAR_8TEV_LJ_DIF_MTTBAR-NORM
The t0 chi2 is different: 131.07 vs 119.87

Instead perfect agreement for CMS_TTBAR_13TEV_2L_DIF

 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TPT vs new: CMS_TTBAR_13TEV_2L_DIF_PTT
 > Everything ok!
 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TPTNORM vs new: CMS_TTBAR_13TEV_2L_DIF_PTT-NORM
 > Everything ok!
 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TTM vs new: CMS_TTBAR_13TEV_2L_DIF_MTTBAR
 > Everything ok!
 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TTMNORM vs new: CMS_TTBAR_13TEV_2L_DIF_MTTBAR-NORM
 > Everything ok!
 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TRAP vs new: CMS_TTBAR_13TEV_2L_DIF_YT
 > Everything ok!
 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TRAPNORM vs new: CMS_TTBAR_13TEV_2L_DIF_YT-NORM
 > Everything ok!
 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TTRAP vs new: CMS_TTBAR_13TEV_2L_DIF_YTTBAR
 > Everything ok!
 > Testing old: CMS_TTB_DIFF_13TEV_2016_2L_TTRAPNORM vs new: CMS_TTBAR_13TEV_2L_DIF_YTTBAR-NORM 
 > Everything ok!

There's of course the problem with the fktables. As I said before, please put also in the OP a list with the dataset that are new (and thus not included in dataset_names.yaml)
Then I will produce a report 1) comparing old and new where appropriate 2) showing only the new ones

[t7phy]

@scarlehoff the newly implemented ones are: CMS 13 TEV LJ (all in the metadata) and CMS_TTBAR_8TEV_LJ_DIF_PTT-NORM

[scarlehoff]

Report for the new datasets https://vp.nnpdf.science/iooLV2_jQQOMQrX3LnnJTA==

[scarlehoff]

Of the data-problematic ones, only CMS_TTBAR_2D_DIFF_MTT_TRAP_NORM is used in NNPDF4.0

Coincidentally, this is the one for which the new pipeline produces different results so not sure whether something needs to be done.