Cross section study and Time Instability

I plot the cross sections after unfolding for each years, using the command getPtY.

I compared each year with 2016 EOY datasets ak4, where the cross sections, pt binning and error are taken from HepDATA.

The plots are reported below. The red curves represent my UL data for each year in each plot, while the black one represents the EOY 2016 in all the plots. In my plots I have applied all the corrections for the data (in particular, Average Map correction for the prefiring) and all the corrections besides PU reweighting for MC used in the unfolding procedure. A pseudo inversion wo regularisation was used. Flat samples were used.

ANYear16Pre.pdf ANYear16Post.pdf ANYear17.pdf ANYear18.pdf

To have a better overview of the difference between each UL year with EOY 2016, I report below also their ratios:

2018 and 2016 Post are in more or less good agreement with 2016 EOY, but 2016Pre and 2017 are not. Especially 2017 seems completely out. I reported below also the ratio of the inputs in which I have done the ratio of the sample in which I have applied the average map and the smooth function respectively for 2016 Pre, 2016Post and 2017. A difference of 3% is expected using the different methods

16Pre:

16Post:

17:

I compared the first binning with Laura results. A completely different behaviour is shown for 2017. Her results are more in agreement with the time stability study that we did. This shows a different behaviour between 2017 Flat and 2017 Slices, maybe a bug? I should expect in principle the same behaviour.