Trigger

As described in section5.3.4, the L1 trigger shows nozγ dependence and a conservative systematic of ±2% is adopted in the 7 TeV and 8 TeV analyses to cover any potential deviations. In the 7 TeV analysis an additional systematic of ±0.5% is assigned to account for uncertainty in the high level trigger efficiency, as described in [112].

8.1.3 7 TeV Photon Identification Efficiency

Diphoton Identification: As described in section 5.3.3, any differences in data and MC relating to the photon identification efficiency and shower shape variables were independent of zγ in the 7 TeV analysis. Therefore any uncertainty due to differences between the efficiency of identifying photons in data and MC will be the same for prompt and non-prompt photons. This systematic was calculated to be ±4.4% [112] and covers the effects of photon ET scale and resolution as well as the efficiency of both photons

passing the Tight IsEM. Due to the stricter cuts, the uncertainty of a photon passing the Tight IsEM cuts is larger than for a photon passing the Loose IsEM, implying that this is a conservative estimate for the Tight-Loose selection used in this analysis. Isolation: To calculate the systematic uncertainty due to isolation the difference in the isolation cone around electrons in theZ →eetag and probe MC and data samples were compared. The mean of the isolation in a cone of ∆R < 0.2 around the electrons in MC and data was found to differ by 0.4 GeV [112]. Therefore, for a range of signal samples the isolation cut was shifted by ±0.4 GeV and the efficiency of photons from that sample passing the new isolation cut was recorded in table 8.3. It can be seen that the maximum deviation from the nominal efficiency is 1.4%. Since this is negligible compared to other systematic uncertainties in the analysis, this value was taken as a flat symmetric systematic for every sample.

Λ τ Efficiency difference (%) when Efficiency difference (%) when (TeV) (ns) lowering isolation cut to 4.6 GeV raising isolation cut to 5.4 GeV

90 2 -0.8 +0.8 90 6 -1.4 +0.6 100 2 -1.0 +0.7 100 6 -0.8 +0.9 120 2 -0.7 +0.6 120 10 -1.1 +0.9 150 2 -0.9 +0.6 200 2 -0.9 +0.6

Table 8.3: Change in signal efficiency for a sample of signal grid points as the isolation

cut is changed by ±0.4 GeV [57].

8.1.4 8 TeV Photon Identification Efficiency

Energy Scale and Resolution: Any difference in the photon energy scale in data and MC creates uncertainty on the expected signal yield by changing the efficiency of photons passing the 50 GeVpTcut. Similarly, any uncertainty on the energy resolution of the EM

calorimeter will create uncertainty on the efficiency of photons passing the 50 GeV pT

cut. The photon ET scale and resolution systematic uncertainty was calculated using

the standard method developed by the ATLAS Egamma Working Group [92]. This method involves smearing the photon ET distribution within the known uncertainties

and recalculating the new efficiency of photons passing the 50 GeVpTcut. The difference in efficiency compared to the nominal value was found to be±1%.

IsEM Identification Efficiency: The ATLAS Egamma Working Group [134] gives the uncertainty on the efficiency of all photons with energy greater than 40 GeV passing the Tight IsEM cuts as±1.5%, except for unconverted photons detected with|η|>1.81 which have an uncertainty of ±2.5%. This uncertainty was calculated for photons that pass the standard isolation cut and also includes any uncertainties relating to the iso- lation requirements. This uncertainty is considered to be conservative due to requiring photons to pass the Tight IsEM cuts rather than the Loose IsEM cuts used in this analysis.

Shower Shape VariablesAs described in section5.3.3no additional systematic is re- quired to account for any potential difference between data and MC in the reconstruction of SSVs at largezγ.

IsolationIn the 8 TeV analysis, the uncertainty due to the isolation cut is included in the IsEM identification systematic. However, this was calculated for prompt photons and does not take into account any uncertainties that may arise due to photons being non-pointing. Figure 8.1 compares the efficiency of probe electrons from Z → ee data and MC samples passing the isolation requirement and shows that they are in good agreement even as zγ increases. Also shown in figure 8.1 is the efficiency of photons

from a prompt and non-prompt signal sample passing the isolation requirements and it can be seen that they are in good agreement in thezγ range where the prompt sample

has statistics. The right hand plot in figure 8.1 shows the isolation of a signal sample extended to much larger zγ values and it can be seen that the efficiency remains flat,

therefore no additional uncertainty is applied.

E ff ic ie n c y 0 0.2 0.4 0.6 0.8 1 =0ns τ =160TeV, Λ =2ns τ =160TeV, Λ Zee MC sample Zee Data sample Zee MC/Data ratio

| [mm] γ |z 0 20 40 60 80 100 120 140 160 180 200 Ratio 0.9 0.92 0.94 0.96 0.981 1.02 1.04 1.06 1.08 1.1 | [mm] γ |z 0 200 400 600 800 1000 1200 1400 1600 1800 2000 E ff ic ie n c y 0.7 0.72 0.74 0.76 0.78 0.8 0.82 0.84 0.86 0.88 0.9 =2ns τ =160TeV, Λ

Figure 8.1: The left plot shows the isolation efficiency of electrons in data and MC

Z →eetag and probe samples and photons in a prompt and 2 ns SPS8 sample as a

function ofzγ. The figure on the right shows the isolation efficiency of photons in a

2 ns SPS8 sample over an expanded range ofzγ.

Object Quality and LAr Hole Flags - 7 TeV analysis: As described in section

6.1.1, events are rejected if any particle fails the object quality cuts or if any particle

is reconstructed in the EM calorimeter hole. These effects were included in certain MC samples and it was found that the efficiency of a single photon passing this veto in data and MC differed by 0.1% [115]. Incorrectly assuming that both photons in the final state are completely correlated would give a, negligible and conservative, estimate of ±0.2% for this systematic uncertainty in diphoton events.