Until further notice, the contents of this page are under construction, and subject to change. It is a working document.
Introduction
In examining the latest NLO and NNLO cross section predictions for various processes, it is important not to forget to compare all of these with the standard MC production used by members of ATLAS on a daily basis. For these, leading-order generators such as Pythia, Herwig and Alpgen are used, as well as MC@NLO. It turns out that leading-order does not, in this sense, mean the same thing as LO in programs such as FEWZ or MCFM, some higher-order and nonperturbative effects are already accounted for. This page shows results for Standard Model Z->ee events generated using Pythia, and comparisons with other physics tools detailed in
WZProductionMonteCarlos.
Most comments (eg about the Pythia flags) should apply to other processes, with appropriate reinterpretation, but I will talk specifically about Z->ee only.
Pythia flags
The most commonly used sample for ATLAS studies of Z->ee is sample number 5144. The job options for this sample (which set process-specific Pythia switches) are
here, but these hide most of the subtleties. The ATLAS MC tuning is in
this file, especially lines 220-248. All of this can, in principle, be overidden.
The most important non-LO effects are (in some order - many of these are under ongoing investigation):
- Inclusion of Photos: This allows electrons (in the final state) to radiate photons.
- "Primordial kT": The incoming partons have a kT taken from a Gaussian distribution with a width of 2 GeV. This is a parameterised non-perturbative correction.
- ISR: ie QCD radiation of the partons "before" collision. This can also give the Z boson non-zero p_T.
- Parton Shower: This again allows non-perturbative QCD corrections to be made.
- Multiple Interactions: Not investigated yet.
- FSR: QCD radiation in the final state - this should not have too much effect in this channel.
- Hadronisation: Ditto.
Plots as a function of electron cuts
Various quantities are plotted here as a function of cuts on the electrons. All events generated have M_Z > 60
GeV.
- Black/Solid circle: No selection.
- Blue/Solid square: Standard ATLAS filter for sample 5144: at least one electron with pTe > 10 GeV and |eta| < 2.7.
- Red/Solid triangles: Both electrons have |eta| < 2.5.
- Purple/Open circle: Both electrons have pTe > 20 GeV.
- Green/Open squares: Both electrons have pTe > 20 GeV and |eta| < 2.5.
In the summary lines, the colours and marker styles belong instead to the particular settings of Pythia used for that run, with each line having the same selection from the list above.
Acceptances (Filter efficiencies)
The table below shows the acceptances of various cuts for each physics case considered. The acceptance (which can also be regarded as an efficiency for a filter based on those cuts) is simply #events(with cuts)/#events(no cuts). The expected uncertainty in each case should be < O(0.05%).
The last three columns show the efficiency of the various selections with respect to the ATLAS filter. These are the acceptances that should be seen in the bulk MC production.
| Generator settings |
ATLAS filter |
Eta cut |
pT cut |
Eta&pT cuts |
Eta cut wrt ATLAS |
pT cut wrt ATLAS |
Eta&pT cuts wrt ATLAS |
| |
|
|
|
|
|
|
|
| Leading Order |
82.12% |
44.43% |
85.20% |
42.11% |
54.00% |
86.66% |
51.28% |
| LO + Photos |
81.57% |
44.55% |
80.31% |
40.10% |
54.51% |
82.04% |
49.16% |
| LO + Parton Shower |
82.13% |
44.52% |
85.20% |
42.18% |
54.11% |
86.64% |
51.36% |
| LO + primodial kT |
82.24% |
44.52% |
84.06% |
41.82% |
54.08% |
85.15% |
50.86% |
| LO + ISR |
84.59% |
47.26% |
78.60% |
41.72% |
55.85% |
78.46% |
49.32% |
| |
|
|
|
|
|
|
|
| ATLAS default |
84.02% |
47.27% |
74.40% |
39.64% |
56.21% |
74.56% |
47.17% |
| ATLAS - Photos |
84.56% |
47.34% |
78.62% |
41.73% |
55.96% |
78.41% |
49.35% |
| ATLAS - PS |
84.08% |
47.28% |
74.42% |
39.71% |
56.20% |
74.59% |
47.24% |
| ATLAS - primordial kT |
84.02% |
47.29% |
74.44% |
39.69% |
56.25% |
74.64% |
47.24% |
| ATLAS - ISR |
81.66% |
44.60% |
79.37% |
39.81% |
54.52% |
80.84% |
48.75% |
This is the same information, in a slightly easier-to-digest form. For each variable, the row labeled "wrt LO" shows
(LO+X)/(LO) - 1, while the "wrt ATLAS" row shows
(ATLAS)/(ATLAS-X) - 1. If an effect completely factorises, these ratios should be equal, and I have chosen the sign so that if an effect increases the acceptance, the figure is positive.
| Effect |
Comparison |
ATLAS filter |
Eta cut |
pT cut |
Eta&pT cuts |
Eta cut wrt ATLAS |
pT cut wrt ATLAS |
Eta&pT cuts wrt ATLAS |
| |
|
|
|
|
|
|
|
|
| Photos |
wrt LO |
-0.67% |
0.27% |
-5.74% |
-4.77% |
0.94% |
-5.33% |
-4.13% |
| |
wrt ATLAS |
-0.64% |
-0.15% |
-5.37% |
-5.01% |
0.45% |
-4.91% |
-4.42% |
| Parton Shower |
wrt LO |
0.01% |
0.20% |
0.00% |
0.17% |
0.20% |
-0.02% |
0.16% |
| |
wrt ATLAS |
-0.07% |
-0.02% |
-0.03% |
-0.18% |
0.02% |
-0.04% |
-0.15% |
| kT |
wrt LO |
0.15% |
0.20% |
-1.34% |
-0.69% |
0.15% |
-1.74% |
-0.82% |
| |
wrt ATLAS |
0.00% |
-0.04% |
-0.05% |
-0.13% |
-0.07% |
-0.11% |
-0.15% |
| ISR |
wrt LO |
3.01% |
6.37% |
-7.75% |
-0.93% |
3.43% |
-9.46% |
-3.82% |
| |
wrt ATLAS |
2.89% |
5.99% |
-6.26% |
-0.43% |
3.10% |
-7.77% |
-3.24% |
Nomenclature note
In the above, "ATLAS" refers to the standard Pythia settings used in the bulk Monte Carlo production (eg data sample 5144 itself, but also for other processes). Something like "ATLAS - Photos" means exactly what it says - Photos is switched off, all other settings are as per the default.
"Leading Order" or "LO" is designed to be as close to what the other calculational tools (MCFM, FEWZ, Horace, etc) call Leading Order. All the effects detailed above are off, the job options snippet that does this is
here. With some cuts (the ATLAS filter, detailed below), this compares well with Horace, as shown
here. Something like "LO+Parton Shower" again means exactly what it says.
Because of FSR, plots have been split into those for the boson itself ("*** of Z") and those referring to the final state electrons ("*** of ee pair"). For LO plots without FSR, these plots will be the same.
- M_Z: The mass distribution of the boson is nearly a Breit-Wigner shape, slightly modified by interference with the photon and the pdfs ("parton luminosity"). None of the flags above change either of these effects, leaving the M_Z distribution unchanged throughout.
- M_ee: In ATLAS, this is different from M_Z through the action of FSR, ie photon radiation from the final state electrons. The other flags only seem to have very minor effects.
- pT_Z and pT_ee: At LO, these are both strictly zero. In the default ATLAS settings, the Z boson can have a significant pT, mostly arising from ISR. The primordial kT of the partons also adds a contribution, limited to 10 GeV by construction. Finally, addition of FSR through Photos affects pT_ee, making it different from pt_Z.
- It is interesting to compare pT_Z for ATLAS-ISR and LO+kT: they are very nearly identical.
- y_Z and y_ee: These are always very similar to each other, as FSR typically changes the energy but not the direction of final state electrons. In the ATLAS settings, the Z appears to be a little more central than at LO, chiefly due to ISR.
- pT_e: Here, the difference between LO and the ATLAS settings is startling. While at LO, kinematics imply that there are very few electrons with pT > 45 GeV, they are common with the ATLAS settings. This difference is largely, but not entirely, down to ISR. (the plot of ATLAS-ISR is interesting for its shape here). The effects of primordial kT and FSR also contribute to the shape of this plot.
- eta_e: This is kinematically related to y_ee, and similar comments apply - the difference between LO and ATLAS settings can by and large be attributed to ISR.
Photos is a separate Monte Carlo program that is used to simulate the effect of QED in the final state. The Z->ee channel, with electrons in the final state, can be significantly altered through this effect. As the energy of one particle is split amongst many, this effect will reduce the average electron pT. To test this effect, a K-factor can be defined, using some set of Pythia parameters, S, as a benchmark:
K = (S-Photos)/(S+Photos). Taking the electron pT as an example, K(pT_e) will typically be less than 1 at low pT and greater than 1 at high pT due to the migration of electrons from high pT bins to lower pT bins.
To test the independence (or otherwise) of Photos from the other effects studied, the table below shows plots comparing K when S = LO Pythia and S = default ATLAS settings. pT_ee has been omitted because it is zero in Leading Order making this comparison meaningless. The only significant difference arises with pT_e, where the inclusion of ISR in the ATLAS defaults has already changed the distribution beyond recognition. At low pT, where these distributions are similar, the K factors agree within statistical precision.
| |
Mass of ee pair |
rapidity of ee pair |
pT of single electron |
eta of single electron |
| No cuts |
Link |
Link |
Link |
Link |
The frankly spiky nature of the LO electron pT distribution also clearly affects the equivalent plots for primordial kT and ISR. The other
(S-kT)/(S+kT)
(S-ISR)/(S+ISR)
The other distributions "agree", in the sense that (ATLAS - X)/(ATLAS) is the same as (LO)/(LO + X) within statistical precision.
--
MikeFlowerdew - 24 Apr 2008
ATLAS
Copyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.