Skip to content
Switch branches/tags
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Oslo Method Python - OMpy

Build Status Binder Code Climate maintainability DOI DOI

| Installation | Troubleshooting | General usage |

Package structure summary | Citing | License


This is ompy, the Oslo method in python. It contains all the functionality needed to go from a raw coincidence matrix, via unfolding and the first-generation method, to fitting a level density and gamma-ray strength function. It also supports uncertainty propagation by Monte Carlo. If you want to try the package before installation, you may simply click here to launch it on Binder.

This is a short introduction, see more at

NB! This repo is currently under development. Use it at your own risk.


Please cite the following (more info below):

  • The code, by using the DOI documenting this github
  • The article describing the implementation
  • If the unfolding / first-generation methods are used, cite the corresponding articles
  • If you are unfolding, make sure to document also the response, e.g. through the citation guide on OCL_response_functions
  • For the decomposition / normalization, you may cite the previous implementation by Schiller (2000).

The code: If you cite OMpy, please use the version-specific DOI found by clicking the Zenodo badge above; create a new version if necessary. The DOI is to last published version; the master branch may be ahead of the published version.

The full version (including the git commit) can also be obtained from ompy.__full_version__ after installation.

The article: The article describing the implementation is now published in Comp. Phys. Comm. (2021): A new software implementation of the Oslo method with rigorous statistical uncertainty propagation DOI: 10.1016/j.cpc.2020.107795.

Other methods: We have reimplemented the unfolding [Guttormsen (1996)] and first generation method [Guttormsen (1987)], see also documentation in the corresponding classes. The decomposition/normalization is subject to the same degeneracy as shown in [Schiller (2000)], but the minimizer and the normalization procedure are different, which is explained in detail in the OMpy article.


Start off by downloading ompy:

git clone --recurse

where the --recurse flag specifies, that all submodules shall be downloaded as well.


  • Get and compile MultiNest (use the cmake version from The goal is to create lib/

    git clone
    cd MultiNest/build
    cmake ..
    sudo make install

    ⚠️ If the make steps above fails it might be that you have gcc/gfortran version 10 or higher. To fix this issue use the following steps instead

    git clone
    cd MultiNest/build
    cmake -DCMAKE_Fortran_FLAGS="-std=legacy" ..
    sudo make install

    Multinest has following hard dependencies: lapack and blas. To use MPI, additionally openmp has to be installed (probably does not work for MAC users, see below.). With apt-get you may fix the dependencies by:

    sudo apt-get install liblapack-dev libblas-dev libomp-dev

    If you still get an error like:

    OSError: cannot open shared object file: No such file or directory

    visit .

  • We require python>=3.7. Make sure you use the correct python version and the correct pip. You may need to replace python by python3 and pip by pip3 in the examples below. Run python --version and pip --version to check whether you have a sufficient python version.

  • All other dependencies can be installed automatically by pip (see below). Alternatively, make sure to install all requirements listed in requirements.txt, eg. using conda or apt-get. You may try following in conda (untested)

    conda install --file requirements.txt
  • For openMP support (optional), install libomp. Easiest on linux/ubuntu: sudo apt-get install libomp-dev or MAC brew install libomp.

  • Many examples are written with jupyter notebooks, so you probably want to install this, too.

OMpy package

There are two main options on how to install OMpy. We will start off with our recommendation, that is with the -e flag is a local project in “editable” mode. This way, you will in principal not have to reinstall ompy if you pull a new version from git or create any local changes yourself.

Note: If you change any of the cython modules (*.pyx files), you will have to reinstall/recompile anyways. As they may have changed upstream, the easiest is probably if you install again every time you pull.

pip install -e .

If you want to install at the system specific path instead, use

pip install .

For debugging, you might want to compile the cython modules "manually". The first line here is just to delete any existing cython modules in order to make sure that they will be recompiled.

rm ompy/*.so
rm ompy/*.c
python build_ext --inplace


Try to reinstall

If you changed / if after a git pull there have been any changes to one of the cython modules, you will have to reinstall/recompile anyways: pip install -e ..

Docker container

If you don't succeed with the above, we also provide a Docker container via dockerhub, see However, for everyday usage, we think it's easier to install the package normally on your machine. The dockerfile is in the .binder folder.

Python version

If you had some failed attempts, you might try to uninstall ompy before retrying the stepts above:

pip uninstall ompy

Note that we require python 3.7 or higher. If your standard python and pip link to python 2, you may have to use python3 and pip3.


If you don't have OpenMP / have problems installing it (see above), you can install without OpenMP. Type export ompy_OpenMP=False in the terminal before the setup above.

Cloned the repo before September 2019

NB: Read this (only) if you have cloned the repo before October 2019: We cleaned the repository from old comits clogging the repo (big data files that should never have been there). Unfortunetely, this has the sideeffect that the history had to be rewritten: Previous commits now have a different SHA1 (git version keys). If you need anything from the previous repo, see ompy_Archive_Sept2019. This will unfortunately also destroy references in issues. The simplest way to get the new repo is to rerun the installation instructions below.

General usage

All the functions and classes in the package are available in the main module. You get everything by importing the package

import ompy

Package structure summary

Below you can find a summary of the most important files and directories of this repository. You can find a full documentation of the packages functionality here.

.ompy - main repository
├── .binder - Dockerfile for easy & reproducible installation and hooks for MyBinder ├── Dockerfile - Dockerfile link
├── docs - documentation, rendered at
│   └── ...
├── example_data - some example data for calculations
│   └── ...
├── - License file
├── notebooks - usage example(s)
│   └── ...
├── OCL_response_functions - submodule facilitating the unfolding method
│   └── ...
├── ompy - package code, see also packaging projects
│   └── ...
├── - Readme with short instructions. See also the online documentation
├── requirements.txt - dependencies, see also pip
├── resources - miscellaneous files
│   └── ...
├── - setup, see also packaging projects
└── tests unit test files, see also packaging projects
   └── ...