pyUPMASK is an unsupervised clustering method for stellar clusters that builds upon the original UPMASK package. Its general approach makes it plausible to be applied to analyses that deal with binary classes of any kind, as long as the fundamental hypotheses are met.
The core of the algorithm follows the method developed in UPMASK but introducing several key enhancements that make it not only more general, they also improve its performance considerably.
We recommend that the packages are installed inside a conda environment. To do this, first follow the steps to install Miniconda. After this, install the required packages into a conda environment using the following command in a terminal:
$ conda create -n pyupmask numpy scikit-learn scipy astropy
To activate the environment (called pyupmask), use:
$ conda activate pyupmask
The (pyupmask) before the $ symbol in the terminal indicates that the environment is activated.
Alternatively you can install the packages in your system directly using pip, but this method is not recommended.
The latest version of the code can be obtained as a compressed file in Releases. Once downloaded, simply extract its contents anywhere in your system.
Once the package is uncompressed and the environment activated, the user needs to set the desired input parameters in the params.ini file. The cluster data files that will be processed must to be stored in the input/ sub-folder.
The code is run simply with:
(pyupmask) $ python pyUPMASK.py
Notice that you need to activate the environment before running pyUPMASK, every time you want to run it.
The code comes with a synthetic cluster to test it. The results will be stored in the output/ sub-folder.
The accompanying article describing the code in detail can be accessed [via A&A][xxxx], and referenced using the following BibTeX entry:
@ARTICLE{2021A&A...650A.109P,
author = {{Pera}, M.~S. and {Perren}, G.~I. and {Moitinho}, A. and {Navone}, H.~D. and {Vazquez}, R.~A.},
title = "{pyUPMASK: an improved unsupervised clustering algorithm}",
journal = {\aap},
keywords = {open clusters and associations: general, methods: data analysis, open clusters and associations: individual: NGC 2516, methods: statistical, Astrophysics - Astrophysics of Galaxies},
year = 2021,
month = jun,
volume = {650},
eid = {A109},
pages = {A109},
doi = {10.1051/0004-6361/202040252},
archivePrefix = {arXiv},
eprint = {2101.01660},
primaryClass = {astro-ph.GA},
adsurl = {https://ui.adsabs.harvard.edu/abs/2021A&A...650A.109P},
adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}