This version of the repo is DEPRECATED and only used for archiving purposes (for those who started using ompy before the new layout).
- Why was this neccesary? The old repository was filled rather fill of large files that were not connected directly to the code. However, any filtering/rebase/etc of the history changes the
sha1s of the previous commits. This may be inconvennient for some people who started using ompy before Oktober 2019. However, it will make the repo cleaner in the future. - We tried also to preserve copy over the issues, just for reference.
Please visit https://github.com/oslocyclotronlab/ompy for the current repository.
Installation | Unfolding | First Generation | Ensemble | Rhosig
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.
This is a short introduction, see more at https://ompy.readthedocs.io/
NB! This repo is currently under development. Many features do not work correctly.
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.
An article describing the implementation more detailled will follow shortly. A draft can be read on arXiv: A new software implementation of the Oslo method with complete uncertainty propagation.
Start off by downloading ompy:
git clone https://github.com/oslocyclotronlab/ompy/- Get and compile MultiNest (use the cmake version from https://github.com/JohannesBuchner/MultiNest). The goal is to create lib/libmultinest.so
git clone https://github.com/JohannesBuchner/MultiNest cd MultiNest/build cmake .. make - We require
python>=3.7. Make sure you use the correct python version and the correctpip. You may need to replacepythonbypython3andpipbypip3in the examples below. Runpython --versionandpip --versionto 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 inrequirements.txt, eg. usingcondaorapt-get. You may try following inconda(untested)conda install --file requirements.txt
- Many examples are written with jupyter notebooks, so you probably want to install this, too.
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 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, you will have to reinstall/recompile anyways.
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 setup.py build_ext --inplaceIf you had some failed attempts, you might try to uninstall ompy before retrying the stepts above:
pip uninstall ompyNote 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, you can install without OpenMP. Type export ompy_OpenMP=False in the terminal before the setup above. For attempts to solve this issue, see also #30.
All the functions and classes in the package are available in the main module. You get everything by importing the package
import ompyThe overarching philosophy is that the package shall be flexible and transparent to use and modify. All of the "steps" in the Oslo method are implemented as classes with a common structure and call signature. If you understand one class, you'll understand them all, making extending the code easy.
As the Oslo method is a complex method involving dozen of variables which can be daunting for the uninitiated, many class attributes have default values that should give satisfying results. Attributes that should be modified even though it is not strictly necessary to do so will give annoying warnings. The documentation and docstrings give in-depth explanation of each variable and its usage.
Still working on a nice interface for the implementation. Test implementation only through norm_nldand norm_gsf classes. Do not have the same calling signatures yet.
An important feature of physics programs is the ability to validate that the program works as intended. This can be achieved by either running the program on problems whose solutions are already known, or by inspecting the program and confirming that each step is working as expected. OMpy uses both methods. Integration tests are performed both on artificial data satisfying the minimal assumptions required of each method (unfold, first generation method, etc.), as well as experimental data which has already been analyzed using other programs (MAMA).
In addition, the methods themselves are written in a way
which separates the uninteresting "book keeping" of each method, such as constructing arrays and normalizing rows, from the actual interesting steps performing the calculations. All parts of a method, its
initial set up, progression and tear down, can be separately inspected using the ompy.hooks submodule and logging framework. This allows the user to not only verify that each method works as intended,
but also get a visual understanding of how they work beyond their mere equational forms.
OMpy is written with modularity in mind. We want it to be as easy as possible for the user to add custom functionality and interface OMpy with other Python packages. For example,
it may be of interest to try other unfolding algorithms than the one presently implemented. To achieve this,
one just has to write a wrapper function that has the same input and output structure as the function Unfolder.__call__(),
found in the file ompy/unfolder.py.
It is our hope and goal that OMpy will be used, and we are happy to provide support. Feedback and suggestions are also very welcome. We encourage users who implement new features to share them by opening a pull request in the Github repository.
The contributors of this project are Jørgen Eriksson Midtbø, Fabio Zeiser and Erlend Lima.
This project is licensed under the terms of the GNU General Public License v3.0 license. You can find the full license here.