• 6b Analysis
  • Table of Contents
  • Install Instructions
  • Gridpack Generation
  • Running Instructions
    • LHE step
      • Hadronisaton and ntuple step
    • Content and description of the ntuples
  • Perform Skim on NTuples
  • HiggsCombine

This repository contains three projects:

1. Sample skimming for Multi-Higgs analyses

This project is used to skim both private and centrally-produced samples. For instructions on how to perform skims, see analysis/MultiHAnalysis.

2. Gridpack production with MadGraph5

This project is used to produce gridpacks, which are used for gen-level event generation (located in MadGraph/). See installation and running instructions below.

3. Full simulation private sample generation

This project is used to generate event simulations that propagate the gen-level events through the detectors. For instructions on how to perform skims, see FullSim.

Note:

• CMSSW recommended release from here
• MadGraph version as in the gen scripts here

Follow these steps to install CMSSW, MadGraph, and the BSM model

# install CMSSW release
cmsrel CMSSW_10_6_28_patch2
cd CMSSW_10_6_28_patch2/src
cmsenv
git cms-addpkg CommonTools/Utils CondFormats/JetMETObjects CondFormats/Serialization FWCore/MessageLogger FWCore/Utilities JetMETCorrections/Modules PhysicsTools/TensorFlow PhysicsTools/ONNXRuntime
scram b -j 4
git clone https://github.com/UF-HH/MultiHiggs

# install MadGraph
cd MadGraph
wget https://cms-project-generators.web.cern.ch/cms-project-generators/MG5_aMC_v2.6.5.tar.gz
tar xzf MG5_aMC_v2.6.5.tar.gz

# now install the BSM model for the process
cd MG5_aMC_v2_6_5
wget https://cms-project-generators.web.cern.ch/cms-project-generators/NMSSMHET_UFO_fermioncouplings.zip
cd models
unzip ../NMSSMHET_UFO_fermioncouplings.zip

Sample event production can be done through the manual production of LHE events through MadGraph and hadronization of LHE events through Pythia. See Running Instructions (Manual).

Sample event production can also be done by generating a gridpack (consisting of several mass points) and using genproductions. See MadGraph/gridpacks/README.md for instructions to generate the cards and Quick tutorial on how to produce a gridpack

After running python generate_grid.py, navigate to the directory MultiHiggs/MadGraph/gridpacks/genproductions/bin/MadGraph5_aMCatNLO/ and modify and run sh generate_6b_gridpacks.sh. This will generate the tarballs, which you can copy to FullSim/Summer20UL18/Template/, in which you can modify genSim_step.py and crabConfig.py and submit each Full Sim sample to CRAB via the command crab submit crabConfig.py.

These steps allow for generating an LHE file and run Pythia (within the CMSSW framework) on top of it.

To generate the MadGraph process (needed only once) see below. It will generate the diagrams and copy updated run and param cards to the destination folder called X_YH_HHH_6b.

cd MadGraph/scripts
source setup_process.sh

To generate a LHE file see below. Note that this takes approx 30s for 10 000 events, so can be reasonably run locally with minimal parallelisation.

NOTE: to run a parallelisation of this step, one has to copy the X_YH_HHH_6b and run multiple copies of the script, adapting the folder name (MadGraph can generate one LHE at the time in a given folder)

source generate_LHE.sh <mX> <mY> [nevents: default 10000]

This will create a folder with the zipped LHE file under MG5_aMC_v2_6_5/X_YH_HHH_6b/Events/gen6b_<mX>_<mY>

To uncompress the file for running the Pythia step, use gunzip filename.tar.gz.

This step runs Pythia within CMSSW on top of the previously produced LHE file. First you need to compile the code with scram b -j 4 (just once).

To run on a single file locally:

cd PythiaAndNtuples
cmsRun pythia_and_ntuples.py inputFiles=file:/path/to/the/LHE outputFile=output_file_name.root

This step takes approximately 90s to hadronise and ntuplise 1000 events (a few error messages from Pythia may appear) when running locally.

For test purposes you can add at the end maxEvents=<num_events> (e.g. to run on 100 events)

Conventions:

• The particles in the process are labeled as X → Y HX, and Y → HY1 HY2. The three H bosons (HX, HY1, HY2) decay to b1 b2.
• Objects are ordered in pt, so that pt(HY1) > pt(HY1) and pt(b1) > pt (b2) .
• The first and last copy of the bosons in the hadronisation are labelled fc and lc respectively. Note that this only matters for X, as Y, HX, HY1, HY2 are identical between lc and fc

Content:

• A vector of pt/eta/phi/m of all the genjets satisfying minimal pt requirements (both in the cases where the neutrinos are clustered and are not clustered in the jet)
• The pt/eta/phi/m of the gen particles as individual floats X, Y, HY, HY1, HY1, HY_b1, HY_b2, HY1_b1, HY1_b2, HY1_b1, HY1_b2
• For every gen b the index of the genjet matched (matched = the closest within a cone dR = 0.4). Value -1 if no match found

For instructions on how to perform skims, see analysis/MultiHAnalysis.

See this page for instructions on how to install and run Combine (HiggsAnalysis-CombinedLimit GitHub)

Notes:

• Documentation recommends using CMSSW_11_3_X to run Combine v9 but this requires architecture slc7_amd64_gcc900, which is not available on lpc machines. Instead, CMSSW_12_X is acceptable.
• Must start a new shell to run Combine (i.e., no cms environments)