lpc-hh / combine-hh

combine repo for HH

combine-hh

CMSSW+Combine Quickstart

export SCRAM_ARCH=slc7_amd64_gcc700
cmsrel CMSSW_10_2_13
cd CMSSW_10_2_13/src
cmsenv
git clone https://github.com/cms-analysis/HiggsAnalysis-CombinedLimit.git HiggsAnalysis/CombinedLimit
cd HiggsAnalysis/CombinedLimit

cd $CMSSW_BASE/src/HiggsAnalysis/CombinedLimit
git fetch origin
git checkout v8.2.0
scramv1 b clean; scramv1 b

cd $CMSSW_BASE/src/
git clone https://github.com/cms-analysis/CombineHarvester.git CombineHarvester
scram b

pip install --user flake8
pip install --user --upgrade numpy
pip install --user https://github.com/jmduarte/rhalphalib/archive/coefsq_rebase.zip
pip install --user --upgrade uproot # use uproot4

Get the HH model

wget https://gitlab.cern.ch/hh/tools/inference/-/raw/master/dhi/models/hh_model.py -O $CMSSW_BASE/src/HiggsAnalysis/CombinedLimit/python/hh_model.py

For reference, consult

• https://cms-analysis.github.io/HiggsAnalysis-CombinedLimit/
• https://cms-analysis.github.io/CombineHarvester/

Get input file

The latest input file for the datacard can be found here:

/storage/af/user/nlu/hh/looper_output/datacard_hist/

and latest VBF card: https://gitlab.cern.ch/hh/results/datacards_run2/-/blob/master/bbbb_boosted_vbf/v2/datacard.txt

Checkout this repo and create datacards:

git clone https://github.com/LPC-HH/combine-hh
cd combine-hh/
./make_cards.sh v8p2yield_AN_sr_sys_0830_fix2017trigSF0913 v1 /storage/af/user/nlu/hh/looper_output/datacard_hist/ True

Here the first argument should match the input histogram file name.

e.g. here this histogram is used for the preapproval result:

/storage/af/user/nlu/hh/looper_output/datacard_hist/HHTo4BPlots_Run2_BDTv8p2yield_AN_sr_sys_0830_fix2017trigSF0913.root

The second argument v1 is a version number in case we want to try different versions using the same input file with different systematic uncertainties etc.

The third argument is the input directory containing the ROOT file with template histograms.

The fourth argument is blinded (True) or not (False)

Command to run unblinded result:

./make_cards.sh v8p2yield_AN_sr_sys_1111 v1 /storage/af/user/idutta/HH/CMSSW_9_4_2/src/V2/HHLooper/hists/result1111/ False
./make_cards.sh v8p2yield_AN_sr_sys_1123 v1 /storage/af/user/nlu/hh/looper_output/datacard_hist False

Run F-test (1st vs 2nd order):

For Bin1,

python runFtest.py --v1n1=0 --v1n2=1 -t 1000 -s 1 --passBinName Bin1 -i /storage/af/user/nlu/hh/looper_output/datacard_hist/HHTo4BPlots_Run2_BDTv8p2yield_AN_sr_sys_1123.root --blinded 0

produce ttbar CR plot figure 27 in ANv4

python create_datacard_TTCR.py --inputfile /storage/af/user/nlu/work/HH/CMSSW_9_4_2/src/HHLooper_sysTest/python/HHTo4BPlots_Run2_ttbarSkim_BDTv8p2.root
cd cards_shapes_TTBarCR/HHModel
source build.sh 
combine -M FitDiagnostics model_combined.root --setParameters r=1 --rMin 0 --rMax 2 --skipBOnlyFit --saveNormalizations --saveShapes --saveWithUncertainties --saveOverallShapes -n SBfitonly --ignoreCovWarning

the output is fitDiagnosticsSBfitonly.root, which will be the input of makePostFitPlot_TTCR.py in the HHLooper directory

Kl and mu scan

Go to Coupling directory

Systematic uncertainty ranking and impact

Go to SystImpact directory

Higgs boson self-coupling (kl) and signal strength scan

Go to Coupling directory

Upper limit as a function of kl

Go to xsUpperLimit directory

Script not done yet

combine-hh's Issues

F-test

@zhangzc11 can you add your scripts for the F-test etc.

Nan's data cards

@nanlu06 I've been using this repo for making data cards based on a standard format of the root files... Maybe we can merge just the combine-relevant code here?

Update to using HH inference

• Make inference and datacards_run2 repos dependencies
• Update/synchronize datacard preparation to be consistent with datacards_run2
• Add script to generate all results using inference (like https://gist.github.com/jmduarte/41dddd641f28d217592bbf93b3784854)