thesis_sci_2022_barreiros andre s.pdf

The TRD trigger was replicated in the MC data set, allowing comparison of relative cutting efficiencies between the two data sets. This allowed further identification of inconsistencies between the dataset, limiting its use for TRD.

Standard model

Analysis of data on proton-proton (pp) collisions collected in 2018 will be presented in Chapter 4, as well as a study of the performance of TRD as a trigger for the conversion of photons into electron-positron pairs. The QGP is a dense, hot soup of freely moving mostly quarks and gluons that, as it cools, condenses into all sorts of unique hadrons, such as short-lived quarkonium states.

Photons

Decay Photons

In QED, photons as vector mediator carry no charge, whereas the gluon has color charge leading to interesting characteristics.

Direct photons

Quarkonium states

Photon interactions with materials

Pair production (Conversion method)

The LHC was built as a successor to the Large Electron-Positron Collider (LEP) which already reached a fundamental energy ceiling in electron-positron collisions of 114 GeV in beam energy in 2000 [18]. CM S- Compact Muon Solenoid (CMS) is another general purpose detector within the LHC with similar goals to ATLAS.

Figure 6: Aerial photo of Geneva with the LHC highlighted [Credit: CERN].

ALICE

Detectors of ALICE

The solenoid magnet in ALICE, although not a detector, is nevertheless important for both the identification of charged particles and the determination of the momentum of charged particles in the plane parallel to the magnetic field. As such, the magnet is able to provide a field uniformity along the beam tube up to 2×10−3T, where the range of operation of a magnet reaches 1.5 T [36].

Triggering system

This thesis deals with 3 trigger classes, two of them are minimum bias triggers, CINT7-B for high statistics and CINT7-T for calculating TRD trigger efficiency.

TRD

Transition radiation

The light emission of a charged particle with a velocity v is defined by the change in the ratio v/vp =vn/c. In the case of TRD, there is variation in a clearly defined boundary between two media.

Geometry of the TRD

A charged particle moving through a chamber ionizes the electrons of the gas, due to the electric field in the chambers, drives the electrons towards the anode wires as shown in Figure 11. The combined signal from the energy loss in the gas chamber (< dE/dx >) and TR -photon ionizes the gas mixture in a TRD layer, the electrons in the gas flow upwards towards the high voltage anode lines (~1520 V) [45].

Figure 9: Stacks running parallel to the z-direction. C- side in Figure 8 is on the right-hand side of the image, towards the muon arm, while A- side is on the opposite side.

TRD track reconstruction

The distance of closest approach between the line and the primary peak is inversely proportional to the transverse moment of the track TRD. But if the neutral particle decays (or produces a pair) into charged particles closer to the TRD.

Figure 14: Hierarchy structure of the GTU. The Front-end electronics from each half-chamber sends tracklet data to its respective Track Matching Unit (TMU).

Data and MC Reconstruction

Between the two event cuts are all the cuts made to the photon's daughter particles (each of them). The efficiency of the HQU booster is found by taking the ratio of the amount of photons presented to the TRD and how many of them meet all the requirements of the HQU.

Figure 17: Description of how the data and MC events are stored. MC generates digits which can then be transformed into ESD events or AOD events, while the Data produces raw information which is also reconstructed into AODs or ESDs.

Efficiency of track matching

Introduction to predicted and optimal tracks
Properties of predicted tracks
Matching to TRD tracks
Matching efficiency of TRD stacks
Optimal TRD stacks

Consider the following illustration of a TRD stack projected onto the xy plane in global coordinates, with the TRD stack in blue (not to scale) and the predicted path in green. In Fig. 23, it is interesting that none of the predicted traces reaching the 5th layer of TRD stack 52 pass through the TRD stacks immediately adjacent to it in the η direction (for photon conversions within −0.8 < η < 0.8). This is because the TRD 52 stack, which is in the middle of the supermodule, is directly above the primary node.

Figure 26 shows the total number of predicted traces that succeed in reaching layer 0 (yellow line labeled layer 0), and how many of them reach layer 5 of the same TRD stack. Now look at the electron efficiency and compare it to the maximum number of tracklets on a TRD track produced for a given TRD stack.

Figure 20: Sketch of an example optimal track (in green) going through all 6 layers of the TRD.

Efficiency of PID cut

Although it remains unclear from this whether the decrease in efficiency of the positrons can be accounted for by their shorter average TRD track length (in tracks) or some other factor in its reconstruction. Most of the positron layers are higher than the electrons for PID < 50 except the positron 0th layer. To better understand the difference in PID values between the two particles, the ratio of the electrons and positrons in the same layer will more easily demonstrate how they are related.

The normalization of the electrons is divided by the normalization of the positrons, the fraction of the two is called normalization in this figure. It seems that the layers are consistent in the way they relate the two particles regardless of the layer, suggesting that electron traces are more likely to be classified as such than positrons.

Figure 33: TRD PID spectrum of electrons and positrons from conversion pho- pho-tons.

Efficiency of the Sagitta cut

The green and blue graphs in Figure 38 are mirror images of each other and both represent positron sagitta values. While the red represents the values of the sagitta electrons, and the purple line continues to be the sum of the two (similar to Figure 37). Even before taking the ratio of the two, it is clear that the distributions between electrons and (inverted) positrons differ, with positrons exceeding electrons at counts below −0.5.

Taking the electrons and dividing them by the reversed sign positrons would present the relative performance of the two particles to each other. The shape of the relationship remains interesting, although the reasons for it remain beyond the scope of this thesis.

Figure 38: Sagitta spectrum of TRD tracks of conversion photon electrons (red) and the flipped saggita of positrons (green).

Possibilities of calculating yield using TRD as a triggering detector 41

Isolating the photons with all HQU cuts and the photons without cuts indicated in Figure 45 allows a comparison with the photons from Figure 18. To verify that the cuts in both datasets are all performed the same way, taking the ratio from the Figure 46 above, NNγw cuts.

Figure 46: The spectrum contains conversion photons from both LCH18o (Data) and LHC18k2 (MC), with MC being coloured red while the data has the colour blue

R spectrum

Sagitta

Calculating the sagitta

Sagitta is not stored by itself in the ESD ROOT files in both datasets, instead the B and C parameters are stored and the sagitta is calculated from these two parameters. Parameters B and C are respectively the slope in the transverse plane and the slope in the r-z plane. As expected from figures sagitta49 and50 in the previous section, the parameters b and c are different for both MC and Data.

Due to the large differences in sagitta values between the two data sets, the associated shear will be ignored for the remainder of this thesis. However, as will become clear in the next section, there are still other discrepancies between the two data sets to observe.

Difference in TRD track cuts

It is important to note that the reductions are cumulative (as in Figure 18, 45) as you move through the rows. This indicates that the TRD track matching appears to be more efficient in the MC than in the data. It should be noted that the ratio for TRD tracks and layer 0 is not the same, there is a slight discrepancy between the two cuts.

It is observed that the MC cuts for pT and PID are tighter than their counterparts, while the track length cut is too lenient. The result of these small differences in the TRD rail cuts is that the resulting final cut coincides.

Table 5: Demonstration of how the different cuts from the HQU trigger (ex- (ex-cluding the Sagitta cut) performed on the conversion photons from both Data and MC, affect the total number of photons present.

Difference in matching efficiency

Presenting the problem
How TRD matching differs per TRD stack
Positrons vs Electrons
TRD tracking efficiency in other periods

Figure 56 shows how early and late conversions are defined by taking the spectrum of the radius of conversion (R) of MC photons. Figure 59 is the ratio of the efficiencies (similar to Figure 55) seen in Figure 58 for early photons and all conversion photons. In Figure 60, the x-axis shows the TRD stack number, while the y-axis shows the efficiency of the corresponding detector.

However, it is difficult to verify because of the large uncertainty present for the MC and even smaller uncertainties present in the data. Figure 64 similar to Figure 55, has the matching efficiency of the MC photon track divided by their data counterpart to verify that the efficiency ratio is the same for all periods mentioned.

Figure 54 would therefore use the ratio N γ w/T RDtrack(s)

TRD Track matching

Track rating
Position rating
p T rating
Extending the rating range

The global (ESD) track is extended from the outer TPC to the TRD layer. After examining how track matching selects valid candidates, one can proceed with comparisons of the two datasets. After checking the position estimates in the previous section, it would be prudent to check whether the pT estimates of the track match are similar in the two data sets.

This problem can be seen if a ratio is obtained between the estimates of two sets of data. There are two hypotheses as to why the difference in efficiency persists: the first is that the deviation (abbreviated to Dy) of the TRD trace is not well implemented in the MC and this may lead to a different reconstruction of the TRD trace.

Figure 65: TRD track match rating between electrons and positrons from con- con-version photons with their associated TRD track.

All global tracks

Displaying the full rating in Figure 77 does not seem to have resolved the issue of the MC outperforming the data. The second hypothesis is that the MC lack of pile-up simulation is to blame for the difference. The MC appears to be biased towards less predicted tracks relative to the data, which would indicate that the MC has a low pT bias, and this relationship is shown by Figure 22.

Second, MC TRD tracking is more efficient than in the data, see Fig. 54 , so it is expected that there will be more predicted trackers with assigned TRD trackers. Similar to Figure 84, Figure 87 is a combination of the RMS lines from Figures 85 and 86 and also shows how MC RMS varies with Data RMS.

Figure 78: Sketch of predicted track through the drift region of the TRD with all points of interest indicated.

DCA values and Pile-up

Thus, the TRD track matching efficiency of conversion photons with different DCA cuts from LHC18o and LHC18k2 is shown. The figure also suggests that the |DCA r| cut does not significantly improve the efficiency of the TRD matching for the data photons. And the DCA reductions do not affect the TRD tracking efficiency of photons in the MC above pγT >1.75 GeV/c.

The study found that the Sagitta value of TRD traces was not well represented in MC. 18] CERN.History of LEP.url:https://home.cern/news/press-release/. 19] CERN Accelerator Complex.url:https://home.cern/science/accelerators/. url: https://home.cern/science/physics/higgs-boson.