====== Recasting/fastsim comparison ======

//Members: Ben F, Andy B, Sezen S, Jon B, R. Ruiz, **Anders K**, Are R., Nishita D., Philippe G., Tomas G., Jack Araz//

  * Finish comparisons on particularly leptonic channels and busy environments
      * Benj: The analysis is now implemented, but not yet validated, in MA5. **Are there common events available somewhere (google drive or CERNBox)?**
  * Comparisons between MA5 (Delphes/Superfast)/Rivet/CheckMATE/ColliderBit
  * Compressed SUSY: ATLAS soft dilepton + soft tracks? ATLAS-CONF-2019-014?
  * CMS Razor (Sezen)


=== Recast comparison with CMS SUS-16-048 ===

We'll use the published version of CMS SUS-16-048: [[http://cms-results.web.cern.ch/cms-results/public-results/publications/SUS-16-048/index.html]]

Our benchmark point will be the (150,142.5) test point from CMS, i.e. degenerate and pure wino Chi20/Chi+1 at 150 GeV, and a pure bino Chi10 at 142.5 GeV.

{{:2019:groups:tools:cms-pas-sus-16-048_figure_001-a.png?300|}}


The CMS signal simulation uses Pythia (8.212) to decay the Chi20 and Chi1+. They assume SM branching ratios  for the resulting virtual Z and W, but take into account the Breit-Wigner shapes. To make this work in Pythia we assume that they have adjusted the "23:mMin" and "24:mMin" settings. We'll try with
  * 23:mMin = 1
  * 24:mMin = 1

-> values used by CMS: 23:mMin = 24:mMin = 0.1

SLHA file: {{:2019:groups:tools:cms_150_143_slha.txt|}} 

250k events will be produced and stored in HepMC format:

Dropbox link to event file (~4.5 GB zipped): [[https://www.dropbox.com/s/wcvufynwymenq86/tag_1_pythia8_events.hepmc.gz?dl=1|tag_1_pythia8_events.hepmc.gz]]

Preliminary ColliderBit results: {{:2019:groups:tools:ColliderBit_preliminary_results.txt|}}

Preliminary MadAnalysis 5 results: [[ https://phystev.cnrs.fr/wiki/_media/2019:groups:tools:ma5_cms_sus_16_048_a.png | MA5-results ]]

Preliminary Rivet results: {{ :2019:groups:tools:rivet_preliminary_results.txt |}}

ADL results (usinng adl2tnm or CutLang): {{ :2019:groups:tools:cmssus16048_adlresults.txt |}}

Delphes card used for the ADL exercise, where efficiencies from the CMS public lepton efficiency plots provided by the analysis were implemented on the default Delphes CMS PileUp card: {{ :2019:groups:tools:delphes_card_cms_pileup_sus-16-048.tcl.txt |}}

=== Cut flow table ===
{{:2019:groups:tools:cms-sus-16-048_cutflow.png?600|}}

Comparing ATLAS and CMS soft dilepton analysis preselections: [[https://indico.cern.ch/event/746178/contributions/3394684/attachments/1850142/3037118/SekmenCompressedEWKSUSY19.pdf|Sezen's slides]]

=== Comparison with experiment simulation software ===

CMS software (simulation and reconstruction) is public and can be used to be compared with results from Delphes and smearing-based simulation.

  * public repository: https://github.com/cms-sw/cmssw
  * Instructions from opendata: http://opendata.cern.ch/docs/about-cms, http://opendata-dev.web.cern.ch/docs/cms-guide-event-production
  * simulation mode:
      * fast simulation (called fastsim), more precise than Delphes
      * full simulation based on Geant 4
  * Run environment:
      * in virtualbox using provided virtual machine image
      * natively. Works with SLC and SLC7. More complicated than virtualbox
      * using a docker image
      * natively from CERN lxplus
  * processing time:
      * 1s/event with the fast simulation on a virtual machine
      * 1mn/event with the full simulation on a virtual machine 

Events will be stored in Delphes format.

=== Analysis description using the ADL ===

Implemented the analysis using a domain specific [[2019:groups:tools:adl|analysis description language]] approach:

[[2019:groups:tools:adl#The ADL file|Link to the analysis description file]]

[[2019:groups:tools:adl#The external function Mtautau|Link to the fMtautau external function]]

Interpretation of the language done by **adl2tnm** and **CutLang**.

=== Tasks and responsabilities ===

Project contact: Anders K.

^ Task                                           ^ Responsible ^
|Sample production                               | Roberto R.  |
|Cutflow numbers from Gambit                     | Anders K. + Tomas G. |
|Cutflow numbers from MA5                        | Benjamin F., Jack Araz |
|Cutflow numbers from CheckMate                  | Nishita D.  |
|Cutflow numbers from Rivet                      | Andy B.     |
|Cutflow numbers with CMS simulation + CheckMate | Philippe + Nishita |
|Cutflow description with ADL, numbers from adl2tnm, CutLang | Sezen |