This tutorial is aimed at (new and old) members of the ATLAS collaboration interested in learning the basics of ATLAS physics analysis tools. There are several introductory lectures aimed at introducing the topics followed by practical sessions.
You need ONLY an atlas UI account on Naples machine

Missing Energy exercise

Go to your accounts on atlas UI
ssh -XY -p5022 <your_user>
or on lxplus
ssh -XY <your_cern_user>

and setup all you need to use ROOT

export ATLAS_LOCAL_ROOT_BASE=/cvmfs/
source ${ATLAS_LOCAL_ROOT_BASE}/user/

Input files can be found here

the directories are public, so you can read the files from here...don't copy it on your directories, they're big and require time!
Now we are ready for the exercises.

Selection criteria

  1. trigger: trigger_HLT_xe70==1
  2. good events: last>=X
  3. leading jet pt: > X We divide to have it in GeV
  4. eta jet TMath::Abs(<X
  5. tight leading jet / > X
  6. MET: met_nomuon_tst_et/1000 > X
  7. lepton veto n_el==X   =n_el_baseline==X  n_mu_baseline==X n_mu==X ALL TOGETHER
  8. number of jet n_jet<X
  9. jet-MET mimim phi: jet_met_dphi_min > XX

Suggestion : use a MakeClass to read the TTree. You can create it with the following commands:

root -l -b
TChain c
for running
root -l -b
.L myAnalysis.C
myAnalysis pippo

Suggestion 2: use a TSelector to read the TTree. You can create it with the following commands:

root -l -b
TChain c
MakeSelector command produces an output similar to MakeClass. There are some differences, which Giuliano will show you.
See here for more information: Root Guide, and Root Guide 2 and and Root Guide Online

Edit myAnalysis.C to book your histograms and using the variable you need. In myAnalysis.C remember to add in Process() method the following line
for running
root -l -b


1. Implement the Signal Region selection reading the data TTree and plot the leading jet pt and MET distribution with the 3.21 fb-1 available (you have to count 21447 events).

2. Add to (1) the background MC simulations to make a data/MC comparison
Remember to set the weight correctly. Use the variable weight but remember that it is normalized to 1 fb-, so you need to scale the distributions for the proper luminosity.

3. Do the same but in the range 150 GeV < MET < 250 GeV
What do you note? which could be the explanation?

4. Test the behavior of the MET with the pileup plotting the 2D histogram of the average interaction per bunch crossing VS MET and make the histogram profile.

5. Repeat (1) and (2) in the Control Region with one muon following the same selection of the Signal Region
Add the cut on the () transverse mass 30 GeV < mT < 100 GeV. The transverse mass definition is

In this case the muon is a signal muon and not a baseline one.
Since that we are using the MET with the muons treated as invisible particles, which informations does the MET distribution give to us?

Solutions: don't cheat!

The macros and the results are in

N.B. Don’t use those macros because are complex and you don’t need so many features in your plots.

Lepton exercise

With this exercise you will work with lepton variables.


Go to your accounts on atlas UI
ssh -XY -p5022 <your_user>

Now copy on your local folder the code for the analysis:

cp -r /data/cirotto/TreeReader_tutorial .

Go in the directory and do de setup

cd TreeReader_tutorial 
Now that you completed setup, let's see how to use the code


You will work with TreeReader code, it's a simple class that reads trees and do plots. You have to create a run macro which exploits TreeReader methods. You have a template in your directory, run.C. What you need to do is
  1. Have a selection to use: this means a trigger and a selection for your region of interest
  2. Set your luminosity
  3. Using the doPlot() method for plotting your variable

This is a possible code that you will use
void run()
  //Don't modify this line
  gROOT->ProcessLine(".L TreeReader.cxx+");

  TStopwatch p;

  //Write your selection and the trigger you want to use
  TString selection = " my selection";
  TString trigger =  " my trigger";
  //Tag for your plot
  TString tag = " my tag";

  TreeReader plot("my path); 

  //set LUMI for MC
  //Set weight for bkg
  plot.SetWeight("weight"); //deafult is "weight"

  //set weight for signal
  plot.SetSigWeight("weight"); //default is weight, set weight for signal as appropriate

  //rescale signal (ONLY signal!)
  plot.SetSigScale(1); //default is 1

  //Set Bkg error on significance calculation
  plot.SetBkgError(0.15); //default is 0.

  //Set LogScale

  //Draw Plot
  plot.doPlot("n_jet","n_jet","n jet", selection + trigger, tag,
         false,"n_jet",3,0,6, false);




Let's focus on doPlot method. It has the following structure

void TreeReader::doPlot(TString histoTitle, TString var, TString title, TString selection, TString selTag,
                         Int_t nbins, Double_t xmin, Double_t xmax,
                         Bool_t doScan, TString cut, Int_t div, Double_t min, Double_t max, Bool_t doRev)
  • The variable present in the TTree is the second one, var. Be careful to use a variable present in the tree, or you will get a crash!
  • histoTitle and title are strings for histogram name and its title
  • selTag just a string which will give a tag to your file
  • nbins, xmin, and xmax are the number of bins, minimum and maximum in the histograms (you should be familiar since you are experts of TH1!)
  • We won't use the last line for now so please set doScan = false

Some advice

  1. The access to vector variables it's not like in yesterday's exercises. For example, to get leading jet pt you have to write jet_pt[0] and NOT This is because we're passing the string to the tree.
  2. You will get a file with all the histos you defined in your run macro. (Discard about the histos without name) Suggestion: maybe i can produce my histos with TreeReader when I do this Exercise

How to run

  1. Open and edit run.C
  2. Update your selection and your trigger
  3. Don't forget to set luminosity! We're working with 2.5 fb-1
  4. Write the variable you want to plot, by using doPlot method
  5. run the code root -l -b -q run.C
  6. You can find your plots in a folder called data
  7. You also have a file with the histogram of variables you plotted (divided for each contribution)


If you open a file and type TBrowser you can see al the variables.
Trigger Variables: trigger_HLT_+name variable for triggers. xeXX refers to MET triggers, for leptons you can find e or mu in the name
Lepton variables: there are several variable of interest:
  • n_lep and n_lep_baseline: where lep can be e or mu it's the number of lepton
  • mumu_m, ee_m: dilepton invariant mass
  • enu_mT, munu_mT: transverse mass on lepton-MET system
  • lep_pt, lep_eta,  where lep can be =e or mu, vector variables (pt is in MeV)
MET: met_nomuon_tst_et, met_noelectron_tst_et they're in MeV, remind to divide if you want to have GeV


Trigger items

Check if the following trigger items are prescaled in run 267638
  • Level1 items: L1_2MU4, L1_2EM15, L1_TAU8, L1_J12, L1_XE70, L1_XE55
  • High Level Trigger items: HLT_2mu4, HLT_2mu14, HLT_e17_lhloose, HLT_j15, HLT_j25, HLT_xe70, HLT_xe80

In case some of them are prescaled check if the prescale is constant in all the lumiblocks
Fill a table with trigger items, the physical object(s) to which is related, the trigger items seeds and prescales when there is the condition “stable beam”

Trigger efficiencies

Evaluate trigger efficiency of the following trigger items as a function of the variable X:
  • HLT_xe70 and X is the missing energy without muons (X=met_nomuon_tst_et) (hint: use all the event as reference): use Wmunu (/data/cirotto/ NtupleTutorial_monob/SUSY7_tutorial_bkg_Wmunu.root) and compare the results with the data
  • HLT_mu26_imedium and X is the leading muon pT (X=mu_pt[0]) (hint: use a proper orthogonal trigger): use Wmunu (/data/cirotto/NtupleTutorial_monob/ SUSY7_tutorial_bkg_Wmunu.root)

Are there eta/phi regions where the efficiency is higher?

You define 3 histograms, numerator, denominator and ratio. See Here for dividing histograms

h_num1d->Sumw2(); //numerator
h_den1d->Sumw2();      //denominator 
h_ratio->Divide(h_num1d, h_den1d, 1, 1, "B") //B option guarantees the correct errors! 

Dilepton invariant mass reconstruction

Use the variables described before to see data/MC comparison of dilepton invariant mass.
What about Z->mumu? We need to selects events with 2 muons and no electrons. The opposite for Z->ee.
Use the following selections:
  • Zmumu
    • Trigger trigger_HLT_mu26_imedium
    • Selection: number of muons and electrons, 66 < dilepton mass < 116 (in GeV)
  • Zee
    • Trigger trigger_HLT_e24_lhmedium_L1EM18VH|| trigger_HLT_e24_lhmedium_L1EM20VH|| trigger_HLT_e60_lhmedium || trigger_HLT_e120_lhloose
    • Selection: number of muons and electrons, 66 < dilepton mass < 116 (in GeV)

Lepton-neutrino transverse mass reconstruction

Use the variables described before to see data/MC comparison of transverse mass.
You can reconstruct W->munu and W->enu mass
Use the following selections:
  • Wmunu
    • Trigger trigger_HLT_xe70 and MET > 200 GeV
    • Selection: number of muons and electrons
  • Wenu
    • Trigger trigger_HLT_xe70 and MET > 200 GeV
    • Selection: number of muons and electrons

Working with histograms

  1. What are the main backgrounds for these selections? You can check it by using the root files produced after each run of your code. Open it and retrieve the integral of each histogram.
  2. Do a ratio plot, divide your data histogram by the TOTAL MC histogram, and see how is the agreement bin per bin

You define 3 histograms, numerator, denominator and ratio. You can clone ratio histogram

h_num1d->Sumw2(); //numerator
h_den1d->Sumw2();      //denominator 
TH1F *h_ratio = (TH1F*)h_num1d->Clone(); 
h_ratio->Divide( h_den1d) 

Bonus exercise

Can you reproduce mono jet SR by using TreeReader? Plot the leading jet pt using the proper cuts. You should find the same number for data!
Suggestion: you should edit the input file in /config/ where to put path to Giuliano's file.

Actually there's already in your folder, so write down the selection in your run.C!

Major updates:
-- ArturoS - 2016-04-15

Responsible: ArturoS
Subject: outreach

Edit | Attach | Watch | Print version | History: r5 < r4 < r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r5 - 2016-04-20 - FrancescoCirotto
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Atlas All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright &© 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback