Updated version of magnetron using DNest 4.
Magnetron is a code to model one-dimensional data (e.g. time series) with a superposition of simple shapes. The number of components in the superposition is part of the model and need not be known beforehand.
Note that this version of magnetron2 uses a Poisson Likelihood, since it was originally written for X-ray data of magnetar bursts. For a version that works on data with normally distributed uncertainties (like optical or radio data of, say blazars), check out ongoing work in the blazars branch!
- DNest4
- python 3.5
- numpy
- scipy
First, make sure DNest4 is installed and the DNEST4_PATH
variable set correctly.
Also, please run python setup.py install
in the python directory of DNest4.
In this case, installing magnetron2
should be as simple as typing make
in the /code/
directory!
magnetron2
unfortunately doesn't have a super convenient user interface. File names of
data files to be run generally need to be changed directly in main.cpp
. For automatization,
you can use the run_dnest.py
script, which automatically reads all files with a certain
matching string in a given directory and will run magnetron2 on each file one after the other.
Be aware that the script is a bit hacky and not the ideal version to do this, but this is
what we have.
Coming soon!
All content © 2017 the authors. The code is distributed under the MIT license.
Pull requests are welcome! If you are interested in the further development of this project, please get in touch via the issues!