Skip to content

Latest commit

 

History

History
82 lines (56 loc) · 2.39 KB

README.md

File metadata and controls

82 lines (56 loc) · 2.39 KB

PyPWA Build Status Coverage Status

A python based software framework designed to perform Partial Wave and Amplitude Analysis with the goal of extracting resonance information from multi-particle final states. Is constantly tested to work with Python Version 3.7+

Has support for multiple likelihoods, including:

  • Extended Log Likelihood
  • Standard Log Likelihood, Optionally Binned
  • Binned ChiSquared Likelihood
  • Standard ChiSquared Likelihood

You can even define your own likelihood, or calculate entirely without one if you chose to do so!

Features

Generic Fitting Tools

  • Fitting
    • Can fit to a log-likelihood, chi-square, or you can define your own
    • Supports Binned Data
    • Supports a quality factor per event
  • Simulation using Monte Carlo Rejection
  • Easy to use Yaml based configuration for command line operation
  • Jupyter Integration
  • Supports using all the threads on the machine

Installing into Anaconda

We've setup an user channel on Anaconda so that you can install PyPWA into your Anaconda installation with the following command

conda install -c markjonestx pypwa

Notes for Apple Users

With testing, we've found that PyPWA's environment in Anaconda doesn't work well with Xterm in Mac OS X, however, Terminal found in your Utilities folder does not seem to have the same issues.

Using from GitHub

Clone the master branch onto your computer, or if you are daring clone the development branch

 git clone https://github.com/JeffersonLab/PyPWA

Setup and activate a virtualenv:

 virtualenv --system-site-packages venv
 source venv/bin/activate

Install the package inside the virtualenv:

 pip install .

Contribute or Support

If you have any issues, or would like to see any features added to the project, let us know!

License

The project is licensed under the GPLv3 license.

Funding

This project is partially supported by NSF Grants #1507208 and #1820235