TWiki> Sandbox Web>AzzahTTH (revision 16) Edit Raw edit Attach Print version


This twiki is a place to collect all the plans, tasks, studies and results related to Azzah's PhD. The overall plan is here: AzzahPlan.pdf

The studies to be carried out (and those already completed) are included in this twiki:

1. Understanding the KLFitter performance in ttbar events

How often do we give the correct jets to KLFitter?

KLFitter runs over the permutations of the 4 jets that we specify it should use: currently the 4 highest pT jets. The assumption we make is that these are the 3 jets from the hadronic top and one from the leptonic top. However, it is perfectly possible that one of these jets is the 5th or 6th jet in order of pT. In this case we have given one or more "non-ttbar" jets to the fitter and so the fitter will never find the correct permutation. This results in a low efficiency overall. Aim: to measure the efficiency that first 4 jets are actually the ttbar jets that we want:

<latex>\epsilon = \frac{\rm{4 ~matches}}{\rm{all ~events}}</latex>

  • For the 4 leading jets in the event, access the "jet_truthmatch" variable for each one and count how many matches you have to the decay products of the tops. You can find more information about the variable here.
  • Make a histogram of the number of matches per event
  • Calculate the efficiency of having all 4 jets matching a decay product from the top.
  • Calculate the efficiency of 3 matches, 2 matches, 1 match and no matches.

Is there a better set of jets to give to KLFitter?

We currently use the leading 4 jets (in order of pT) as input to KLFitter. Are there other sets of 4 jets which would give us a higher efficiency for 4 matches (as measured above)? For example, we could give priority to b-tagged jets. To do this we will select events with b-tags, so first we must look at the change in 4-match efficiency when we change the selection.

What is the change in efficiency for 4 matches when we require b-tagged jets in the event selection?

  • Change the event selection in cuts.txt to require at least one b-tagged jet (>=4 jets, >=1 b-tags)
  • Calculate the efficiency of 4 matches, 3 matches, 2 matches, 1 match and no matches in these events.
  • Change the event selection in cuts.txt to require at least two b-tagged jets (>=4 jets, >=2 b-tags)
  • Calculate the efficiency of 4 matches, 3 matches, 2 matches, 1 match and no matches in these events.

What is the effect (on matching efficiency) of changing the set of jets in >=4 jets, >=2 b-tags?

Even in events with >=4 jets, >=2 b-tags, if we consider only the first 4 jets (in order of pT) then there will be times where we are not using the b-tagged jets in KLFitter. This is because they may be the 5th or 6th jet in pT. Since we are studying the best set to give to KLFitter, it might be better to choose the set in another way. Next you should study the efficiency of 4 matches, 3 matches, 2 matches, 1 match and no matches for different choices of sets of jets.

  • Select events with >=4 jets, >=2 b-tags using cuts.txt.
  • Using this ntuple, repeat the study above for a different set of 4 jets, e.g.
    • 1 leading (in pT) b-tagged jet + 3 more jets (highest in pT, b-tagged on not)
    • 2 leading (in pT) b-tagged jets + 2 more jets (highest in pT, b-tagged on not)
    • Any other options you can think of!

What is the effect on the KLFitter reconstruction efficiency of changing the set of jets in >=4 jets, >=2 b-tags?

  • Investigate how to change the set of jets to give to KLFitter using twikis and documentation
  • Change the jets used by KLFitter to perform the reconstruction by requiring that is using one or two b-jets for example (investigate the different possibilities)
  • Find the reconstruction efficiency as you did before and compare with the results of using KLFitter in the default configuration.

Results with one file (left) and increasing the statistics (right):


Update to newer versions of the samples and repeat the KLFitter study

  • Get a newer derivation of the ttbar sample (mc15_13TeV.410000.PowhegPythiaEvtGen_P2012_ttbar_hdamp172p5_nonallhad.merge.DAOD_TOPQ1.e3698_s2608_s2183_r7267_r6282_p2516/)
  • Make the plot comparing the efficiencies for the different sets of jets (as above) with this sample and compare to what you had before. Make sure you use sufficient statistics.
  • Does your conclusion change?

Efficiency of reconstructing the rest of the event

Now you have measured the efficiency of reconstructing the hadronic top correctly, you can do the same for the leptonic top and for both:

  • What is the efficiency for reconstructing the leptonic top in the events?
  • What is the efficiency for reconstructing both the hadronic and leptonic tops?

Reconstruction efficiency in ttH events

We would like to know what the efficiency of reconstructing the tops in ttH events using exactly the same KLFitter technique as you did in ttbar:

  • Download a few files of the ttH sample: mc15_13TeV.341270.aMcAtNloHerwigppEvtGen_UEEE5_CTEQ6L1_CT10ME_ttH125_semilep.merge.DAOD_TOPQ1.e4277_s2608_s2183_r6869_r6282_p2516/
  • What is the reconstruction efficiency for the hadronic top in these events
  • Also, measure the efficiency of reconstructing the leptonic top and then both the hadronic and leptonic tops together.

2. Extending KLFitter to be used with ttH events

Adjust KLFitter likelihood

More details to come soon....

3. Understanding the ttH analysis strategy through review of the Run-1 publication

For the first results, the 13TeV ttH analysis will follow the Run-1 ttH analysis very closely. Therefore, having a very thorough understanding of the 8TeV analysis is crucial for successfully completing the tasks outlined in your thesis plan. For this reason, we recommended studying the 8TeV ttH analysis very carefully. Having carried out this review, you will be able to give a presentation about ttH analysis in a Glasgow ATLAS meeting.

List of questions to answer

Here is a list of questions designed to guide you to what are the most important points to understand about the analysis: TTH_Questions.pdf

The list is not complete but covers most of the important points. Prepare a document with your answers to these questions in a LaTeX document, as you will certainly find this a useful thing to refer back to in the future.

4. Make a complete set of control plots for all resolved ttH analysis regions

For this you will need to run plot factory:

Set-up and compile the code:

%CODE{ lang="sh"}%

export ATLAS_LOCAL_ROOT_BASE=/cvmfs/ alias setupATLAS='source ${ATLAS_LOCAL_ROOT_BASE}/user/' setupATLAS --quiet

setupATLAS mkdir PlotFactory_MayProduction_20.1_devbranch cd PlotFactory_MayProduction_20.1_devbranch export AnalysisTopVersion ="2.3.45a" rcSetup Top,"${AnalysisTopVersion}"

rc checkout_pkg svn+ssh:// rc checkout_pkg atlasoff/PhysicsAnalysis/TopPhys/TopPhysUtils/TopDataPreparation/trunk rc checkout_pkg atlasoff/PhysicsAnalysis/TopPhys/xAOD/TopFakes/tags/TopFakes-00-00-08

rc find_packages rc compile


Useful Links


KLFitter twiki:

TTHbbLeptonic SVN:

TTHbbLeptonic twiki:

KLFitter paper:

Run-1 ttH analysis Links


conf note:

int note:


Topic attachments
I Attachment History Action Size Date Who Comment
PDFpdf AzzahPlan.pdf r1 manage 105.2 K 2016-05-19 - 15:41 SarahBoutle  
PDFpdf TTH_Questions.pdf r1 manage 101.0 K 2016-05-19 - 15:31 SarahBoutle  
PNGpng efficiency2.png r1 manage 24.7 K 2016-02-18 - 11:39 SarahBoutle  
PNGpng efficienvy1.png r1 manage 25.2 K 2016-02-18 - 11:39 SarahBoutle  
Edit | Attach | Print version | History: r17 < r16 < r15 < r14 < r13 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r16 - 2016-09-20 - ChrisCollins
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback