Skip to content

Latest commit

 

History

History
70 lines (50 loc) · 3.16 KB

README.md

File metadata and controls

70 lines (50 loc) · 3.16 KB
Raw

EFT-Fitter

Run the fit(s)

The main tool is the runFit.py script. This will run a fit of the pois, followed by a profiled and a fixed scan of each poi in the range given. Note that the default values in the fixed scans will be fixed to those given in the params file unless you add the option --setParamsToNominal in which case they will instead be fixed to the global maximum likelihood estimates.

An example run is

python runFit.py --inputs inputs.HIG_19_015.ttH_HIG_19_015 --pois params.tth_hgg_kVkF --functions functions.tth_hgg_kVkF --output tth_hgg_fits_kVkF --npoints 50

The output is a .pkl file with the results of the $\Delta \chi^{2}$ vs the poi values from each scan. In this case its called resultstth_hgg_fits_kVkF_observed.pkl. You can take a look at the results, for example to pull out the profiled scan of kappa_F, with

  • python utils/pickle2text.py resultstth_hgg_fits_kVkF_observed.pkl kappa_F output.txt

  • python utils/overlay_scans.py show output.txt

This will produce the following plot

Example

Or you can plot either POI (along with the function values, and profiled parameter values) directly from the results via

python utils/poi_profile.py resultstth_hgg_fits_kVkF_observed.pkl kappa_F

This will produce the following plot

Exampleprof

Options

python runFit.py --help                                                                                            [15:44:30]
Usage: runFit.py [options]

Options:
  -h, --help            show this help message and exit
  --pois=POIS           Name of json file storing pois
  --scanpois=SCANPOIS   comma separated list of pois to scan (if empty,
                        default is all pois)
  --output=OUTPUTSTR    Identifier string for output results
  --functions=FUNCTIONS
                        Name of json file storing functions
  --inputs=INPUTS       Comma separated list of input files
  --npoints=NPOINTS     number of points in the scan
  --theory_uncert=THEORY_UNCERTS
                        config for theory uncertainties
  --doAsimov            Do asimov fit (i.e. set all best-fit to nominal)
  --doReset             Reset poi values each step in profiled scan
  --doFlip              Start scan from max val of poi
  --doLinear            Also run the scan using linear terms of functions
                        (defined in --functions) -- only appropriate for EFT
                        models with SM+linear+BSM terms)
  --setParamsToNominal  Set nominal values of the POIs to those at the global
                        minimum

To-do : add proper description of runFit.py

Settings

Inside tools/fitter_2.py there are 2 settings, that you may need to edit

SCIPY_MINIMIZE=True
USE_GRADIENT=True

The first sets the minimizer to scipy.minimize. Setting this to False will use the iminuit interface to Minuit instead.

The second option tells the fitter to use the analytic gradient. For this, you just need to make sure your functions include the gradients too. See this example