(Build tests on Travis CI currently being done for Python versions 2.7, 3.3 and 3.4.)
ExoSOFT is an Open Source project. Contributions are very welcome. Create a branch off 'master', make your changes and submit a pull request.
Documentation of the code can be found HERE (Note: Documentation is still a work in progress)
The Exoplanet Simple Orbit Fitting Toolbox
ExoSOFT is a toolbox for the orbital analysis of exoplanets and binary star systems. It is free to compile, open source, fits any combination of astrometric and radial velocity data, and offers 4 parameter space exploration techniques, including MCMC. Additionally, it is packaged together with an automated set of post-processing and plotting routines to summarize the results . Verifications were performed by fitting noisy synthetic data produced with an independent Keplerian model for a variety of systems ranging the entire parameter space.
The release paper reviewing ExoSOFT and its verification are available here.
Examples of its use on real data include:
-
An investigation of the V450And system Helminiak & Kuzuhara & Mede et. al. (2016),
-
and the Tau AB binary Mede & Brandt (2014).
ExoSOFT would be a suitable choice for performing orbital analysis during surveys with new RV and direct imaging instruments.
Option 1:
$ pip install ExoSOFT
An example of a full installation from scratch on a Mac, using conda:
$ conda create --name ExoSOFTcondaEnv python Cython gcc scipy numpy matplotlib
$ source activate ExoSOFTcondaEnv
$ pip install ExoSOFT
Option 2: $ conda create --name ExoSOFTcondaEnv python Cython gcc scipy numpy matplotlib
$ source activate ExoSOFTcondaEnv
-
clone this repository
-
from its directory on your local machine, run:
$ python setup.py install
Make a valid settings.yaml file, matching the format in the repository's examples directory. Depending on the system you are investigating, you will also need to have data files of the correct format (data format examples here).
** Make sure to update the directory where ExoSOFT should write all output files to, as indicated by the 'outDir' setting in your local settings.py file. **
Then, from the directory containing the settings.yaml, simply start ExoSOFT with:
$ ExoSOFT
Should you wish to provide custom priors, a proper priors.py file must also be in the same directory. To make one, copy the default file and edit accordingly.
There are two working examples, one for that runs the stages Simulated Annealing then emcee (settings_SAemcee.py). The other runs Simulated Annealing, Sigma Tuning, then MCMC (settings_SASTMCMC.py). Either copy their settings file in the examples directory to a new file or rename them to settings.py and run it the same as above.
Additionally, there are two basic examples of how to interact with the model more directly. These examples are given in:
- model_example_basic.py
- model_example_emcee.py
These will show the user how to write their own scripts that use the ExoSOFT model instead of using ExoSOFT itself.
Solution to 'binary incompatible' error:
Depending on the install of the scientific python stack on your machine, you may get a 'binary incompatibility' error.
This arrises from the copy of scipy currently installed having being built against a version of numpy<1.8, which does not catch the error thrown by Cython.
The solution is to uninstall and re-build numpy, scipy. It takes a while to build scipy especially, but this worked for me.
The commands in order are:
$ pip uninstall -y numpy
$ pip uninstall -y scipy
$ pip install numpy --ignore-installed --no-cache-dir --no-binary :all:
$ pip install scipy --ignore-installed --no-cache-dir --no-binary :all:
Solution to problems with matplotlib
Again, depending on your installation of matplotlib, some errors can arise.
These include missing init.py in one of the matplotlib directories (mpl_toolkits in particular).
Another is the inability to import matplotlib.pyplot.
To fix this:
-
Make sure there is now locally installed version of matplotlib in your home directory that could confuse the python path.
-
re-install matplotlib from scratch
$ pip uninstall -y matplotlib
$ pip install matplotlib
There are some additional modules available in the git repository that are not part of the base ExoSOFT package. These can be found in the extra directory and include:
- customPost.py
- artificialDataMaker.py
customPost.py is for re-performing the post analysis of ExoSOFT with the ability to customize the plotting settings.
This is in order to produce the exact versions of the plots and statistical calculations necessary for publication.
artificialDataMaker.py is to produce simulated/artificial data with pre-determined orbital elements.
For more details, please see the documentation in each module.
Please cite our paper if you find this code useful in your research.
The Bibtex entry for this paper is::
@article{1538-3881-153-3-135, author={Kyle Mede and Timothy D. Brandt}, title={The Exoplanet Simple Orbit Fitting Toolbox (ExoSOFT): An Open-source Tool for Efficient Fitting of Astrometric and Radial Velocity Data}, journal={The Astronomical Journal}, volume={153}, number={3}, pages={135}, url={http://stacks.iop.org/1538-3881/153/i=3/a=135}, year={2017}, }
Copyright 2016 Kyle Mede and contributors.
ExoSOFT is free software made available under the GNU GPLv3 license. For details see the license.txt file.