Fit and bias studies

  1. Try to rewrite all the files that are necessary to run the bias study with the combine tool. README is here https://github.com/CMSDIJET/DijetRootTreeAnalyzer/blob/master/README_caloscouting_analysis.md, but it is not very informative, so I need to dig through code to understand better the framework.
  2. I found the best bkg function: fit_mjj_full_calotrijet2016_3.pdf - it the 6 parametric functions that was used by atlas and ua2 collaboration to fit the dijet mass spectrum.
  3. The script RunBias.py executes all the commands that are needed for the bias study:
    1. First command: writes the data card need to run combine tool in the following steps using the data file (prepared), nominal signal file (prepared), signal files with systematic uncertainties (prepared).
    2. Second command: generates N pseudo experiments using as input:
      1. the output data card from the first step
      2. the function to generate toys
      3. parameters from the dijet mass spectrum fit using the same function as for the generation
      4. user-defined signal strength
    3. Third command: fit using the combine tool with the maximum likelihood method all of the toys using user-defined function and compute signal strength from the fit.