HEP Fitting Tutorial

The material

Jupyter notebooks preview is available here.

To run the tutorial, just click on:

Contents

• Fitting basics: Minmize a cost function using scipy.optimize.minimize
• Fitting polynomials to points using np.polyfit
• Fitting arbitrary functions to points using scipy.optimize.curve_fit
• Template fits using pyroofit
• Histogram template fits using pyhf
• Hypothesis tests with pyhf

Setup instructions

If you want to use what you learned in the tutorial on your projects later, here are some examples how you can setup the nescessary software:

Generic

If you don't have anything setup and need both ROOT and the python packages, the easiest way is through Anaconda or Miniconda. After installing, create your environment using

conda env create -f environment.yml

and activate it via

conda activate HEPFittingTutorial

LMU Bachelor projects

On kuhrios (for Belle II projects)

Most tutorials should run on top of the latest Belle II release which you can set up with

source ~/setup_belle2.sh

You can then install the additional packages used in this tutorial on top with

pip3 install --user pyroofit pyhf iminuit mplhep

You should then be able to run the examples in a jupyter notebook or in jupyter hub (on Jupyter hub, make sure you use the Belle II Kernel with the latest version).

Note: Some of the packages have stopped supporting python3.6, so if you need the latest versions, maybe go for the Generic installation with conda, e.g install Miniconda into your home directory.

On gar-ws-etp* (for ATLAS projects)

You can install the packages on top of the ROOT + python3.7 setup:

module load root/6.20.04_py3.7

To give newer versions of python packages you install as a user priority over the system environment, reset the PYTHONPATH to just contain ROOT

export PYTHONPATH=/software/opt/bionic/x86_64/root/6.20.04_py3.7/lib

Then you can install what is used in the tutorial with

pip install --user --upgrade pyroofit pyhf iminuit mplhep

Then you should be able to run the examples in a jupyter notebook.

Contributors ✨

Thanks goes to these wonderful people (emoji key):

Nikolai Hartmann 💻

ThomasLueck 💻

This project follows the all-contributors specification. Contributions of any kind welcome!