First, install 2DAlphabet from my fork, with the binningFloatFix branch that addresses some of the issues with using non-integer bin widths.

Log on to the LPC and ensure that your Github SSH keys are added to the agent.

eval "$(ssh-agent -s)"
ssh-add ~/.ssh/id_XYZ

Then, install 2DAlphabet following the modified instructions, shown here:

export SCRAM_ARCH=slc7_amd64_gcc700
cmsrel CMSSW_10_6_14
cd CMSSW_10_6_14/src
cmsenv
git clone --branch binningFloatFix https://github.com/ammitra/2DAlphabet.git
git clone https://github.com/cms-analysis/HiggsAnalysis-CombinedLimit.git HiggsAnalysis/CombinedLimit
curl -s https://raw.githubusercontent.com/lcorcodilos/CombineHarvester/master/CombineTools/scripts/sparse-checkout-ssh.sh | bash
scram b clean; scram b -j 4
cmsenv

Finally, create the virtual environment into which we install 2DAlphabet:

python -m virtualenv twoD-env
source twoD-env/bin/activate
cd 2DAlphabet
python setup.py develop

Now, you are ready to use 2DAlphabet. Note that, every time you log on to the LPC you must run these steps to use 2DAlphabet:

cd /path/to/CMSSW_10_6_14/src
cmsenv
source twoD-env/bin/activate

The HSCP input histograms are created from the templates found in the crab_Analysis_2018_*.root files. The script makeRegions.py will find the appropriate histogram from the background (crab_Analysis_2018_AllBackground_CodeV40p4_v1.root) and signal (crab_Analysis_2018_HSCPgluino_M-1800_CodeV40p4_v1.root) files and create new template files for 2DAlphabet.

The script creates two new ROOT files - one for signal and one for data (we just treat the background as data for this example). Each file contains an hpass and hfail histogram, where the Pass region is defined as 0 < ProbQ < 0.1 and the Fail as 0.1 < ProbQ < 0.7.

The Pass and Fail histograms are 20x1 (I_as vs ProbQ), where I've moved I_as to the X-axis because at the moment 2DAlphabet only allows blinding on the X-axis.

The 2DAlphabet config file is test.json. The binning is defined such that LOW, SIG and HIGH each have events in them (otherwise the fit will fail). 2DAlphabet should be modified so that we can pass an empty dict to HIGH

More explanations to come

To make the workspace and run the fit, run

python test.py

explanation to come

export SCRAM_ARCH=slc7_amd64_gcc700
cmsrel CMSSW_12_3_0
cd CMSSW_12_3_0/src
cmsenv
python3 -m virtualenv plotting-env
source plotting-env/bin/activate
pip install numpy root_numpy mplhep matplotlib

cd CMSSW_12_3_0/src
cmsenv
source plotting-env/bin/activate