Early Run-3 data/MC comparison to study CMS Tracking Performance

Link to the public DP note

Results are published in the DP note CMS-DP-2022-064

Data/MC comparisons using ZeroBias events

The figures in the following show a comparison between data and MC of the reconstructed track properties. Events selected are ZeroBias events, the tracks which are considered are tracks which pass the highPurity selection [1], with pT>1GeV. MC distributions are normalized to the number of vertices in data. The so-called ZeroBias events are triggered using only the information on the beam-beam coincidence, and were collected from July 19th, 2022 to October 17th, 2022 (with the exception of the period from August 23rd to September 27th). The trigger which is used collects only a fraction of delivered events.

Figures in PNG format (click on plot to get PDF)	Description
	The figures show the distribution of the reconstructed Z position of the tracks passing the selection for data and MC outlined in the previous slide. Comparisons are shown for the different periods of time shown in the figures, after the indicated luminosity was delivered since the installation of the new Barrel Pixel (BPix) layer 1 [2]. Fits using a Gaussian distribution summed with a constant are superimposed on data and MC. Distributions in data are corrected to account for the position of the pixel detector. The MC sample is simulated with pixel detector in (0,0,0) and the beam is displaced, while in data both the pixel detector center and the luminous region centroid are free floating parameters. Distributions in data are narrower and the peak tends to move towards 0 with time.
	The figures show the pT and pseudorapidity (η) distributions of the tracks from ZeroBias events, passing the selection described above. All events from the combined data taking period described above are used in these histograms. An agreement in the pT distribution is found within 10%. The η distribution in MC is wider than in data: possibly due to the tuning of the MC generator used, which can probably be improved. Also asymmetry in absolute value, due to beamspot Z.
	The figure shows the distribution of the distance of closest approach to the Primary Vertex of tracks from ZeroBias events, passing the selection described above. All events from the combined data taking periods described in above are used in these histograms. The MC distribution is narrower than the data distribution, indicating a better quality of alignment in MC.
	The figures show the distributions of the significance of 3D impact parameters with respect to the Primary Vertex of tracks from ZeroBias events, passing the selection described above. Comparisons are shown for the different periods of time shown in the figures, after the indicated luminosity was delivered since the installation of the new BPix layer 1 [2]. Agreement between data and MC gets worse over time, indicating aging of BPix layer 1 due to accumulated irradiation. Improvement in agreement in the latter data taking period due to an update in the high-voltages and in the alignment which has been implemented later in the data-taking.
	The figures show the distributions of the significance of 3D impact parameters with respect to the Primary Vertex of tracks from ZeroBias events, passing the selection described above. In this case only those events which have been re-reconstructed are considered here. Re-reconstruction includes updates to pixel local reconstruction and the alignment of the tracker, leading to better performance. The figure on the left shows the prompt reconstruction, the figure on the right shows the re-reconstruction pass, for the period indicated and after the indicated luminosity was delivered since the installation of the new BPix layer 1 [2]. Variables connected to impact parameters (hence used for b/tau tagging, etc.) are the ones most improved by the re-reconstruction conditions, as expected from the updates previously indicated. The agreement between data and MC is much better for re-reconstructed data.
	The figures show the distributions of the significance of 3D impact parameters with respect to the Primary Vertex of tracks from ZeroBias events, passing the selection described above. In this case only those events which have been re-reconstructed are considered here. These are a subset of the events of the middle histogram in slide 6. Re-reconstruction includes updates to pixel local reconstruction and the alignment of the tracker, leading to better performance. The figure on the left shows the prompt reconstruction, the figure on the right shows the re-reconstruction pass, for the period indicated and after the indicated luminosity was delivered since the installation of the new BPix layer 1 [2]. Variables connected to impact parameters (hence used for b/tau tagging, etc.) are the ones most improved by the re-reconstruction conditions, as expected from the updates previously indicated. The agreement between data and MC is much better for re-reconstructed data.
	The figures above show the number of valid hits in the pixel and strip detectors of tracks from ZeroBias events, passing the selection described in slide 2. All events from the combined data taking periods described in slide 2 are used in these histograms. For the pixel detector a trend with number of hits, the number of tracks reconstructed with 4 hits good is well-reproduced. For the strip detector the number of hits in general overestimated in MC, with agreement at the 20% level. These distributions are sensitive to the bad components of the tracker. The number of bad components in simulation was set to correspond to the data taking period from August 20th, 2022 to August 23rd, 2022.

Data/MC comparisons using Z → µµ events

The figures in the following show a comparison between data and MC of the reconstructed track properties using events which are compatible with decay of the Z boson into two muons. MC events are weighted to match the distribution of number of vertices in data. Tracks are selected with the following requirements:

● They must come from global PF muons [3]

● They must have |η|<2.4, pT>5 GeV

● Normχ2<10, |dxy,BS|<0.02 cm, |dZ,BS|<20 cm

● At least one pixel hit, at least 8 strip hits, has at least hits in 2 muon stations

● Relative isolation within a cone of radius ΔR=0.4 less than 0.3 [3]

● The two highest pT tracks must have 60 GeV<mμμ<120 GeV

The events are selected by a High Level Trigger having at least one isolated muon with pT>24 GeV. The data analyzed here were collected from July 19th, 2022 to October 17th, 2022 (with the exception of the period from August 23rd to September 27th)

Figures in PNG format (click on plot to get PDF)	Description
	The figures above show the pT and η distributions in data and MC for the tracks compatible with Z→μμ selection as outlined above. All events from the combined data taking period described above are used in these histograms. Disagreement in pT between data and MC, is due to improper modelling of Z boson production mechanism (Z pT, helicity, rapidity, etc.). For the η distribution the agreement is good, except at high values. The sharp increase at 24 GeV is due to the trigger threshold. Since muons reconstructed in the muon system are used, the holes of the muon spectrometer can be seen.
	The figure above shows the distribution of the distance of closest approach to the Primary Vertex of the tracks compatible with Z→μμ as outlined above for data and MC. All events from the combined data taking period described above are used in these histograms. The first and the last bin are respectively underflow and overflow bins. Distribution wider in data than in MC, shifted. Same considerations about quality of alignment as above apply here.
	The figures show the distributions of the significance of 3D impact parameters with respect to the Primary Vertex for tracks compatible with Z→μμ selection as outlined above. Comparisons are shown for the different periods of time shown in the figures, after the indicated luminosity was delivered since the installation of the new BPix layer 1 [2]. Agreement between data and MC gets worse over time, indicating aging of BPix layer 1 due to accumulated irradiation. Improvement in agreement in the latter data taking period due to an update in the high-voltages and in the alignment which has been implemented later in the data-taking.
	The figures show the number of valid hits in the pixel and strip detectors for tracks compatible with Z→μμ selection as outlined above. All events from the combined data taking period described in slide 10 are used in these histograms. For the pixel detector the agreement between data and MC is at 10% above nHits≥4, while it breaks at lower values. For the strip detector the agreement between data and MC is at 10% above nHits>10 while it breaks at lower values. These distributions are sensitive to the bad components of the tracker. The number of bad components in simulation was set to correspond to the data taking period from August 20th, 2022 to August 23rd, 2022.

References

[1]: Description and performance of track and primary-vertex reconstruction with the CMS tracker, CMS Collaboration, JINST 9 (2014) P10009

[2]: CMS Phase-1 pixel detector refurbishment during LS2 and readiness towards the LHC Run 3, CMS Collaboration, CMS-CR-2021-255 (2021)

[3]: Performance of the CMS muon detector and muon reconstruction with proton-proton collisions at √s = 13 TeV, CMS Collaboration, JINST 13 (2018) no.06, P06015

-- DavideBruschini - 2022-11-21