smoh / kinesis

kinematic modelling of clusters with gaia

kinesis

Kinematic modelling of clusters with the Gaia data.

Kinesis is a package for fitting the internal kinematics of a star cluster with astrometry and (incomplete) radial velocity data of its members. In the most general model, the stars can be a mixture of background (contamination) and the cluster, for which the (3,3) velocity dispersion matrix and velocity gradient (dv_x/dx, dv_y/dx, ...) are included. Please refer to Oh & Evans 2020 for full details. There are also simpler versions of the most general model and utilities to generate mock clusters and mock observations.

Check out the documentation (it is under development and may be incomplete).

Attribution

If you make use of this code, please cite Oh & Evans 2020. The bibtex entry is:

@ARTICLE{2020MNRAS.498.1920O,
       author = {{Oh}, Semyeong and {Evans}, Neil Wyn},
        title = "{Kinematic modelling of clusters with Gaia: the death throes of the Hyades}",
      journal = {\mnras},
     keywords = {methods: data analysis, software: data analysis, astrometry, stars: distances, stars: fundamental parameters, open clusters and associations: individual: Hyades, Astrophysics - Solar and Stellar Astrophysics, Astrophysics - Astrophysics of Galaxies},
         year = 2020,
        month = oct,
       volume = {498},
       number = {2},
        pages = {1920-1938},
          doi = {10.1093/mnras/staa2381},
archivePrefix = {arXiv},
       eprint = {2007.02969},
 primaryClass = {astro-ph.SR},
       adsurl = {https://ui.adsabs.harvard.edu/abs/2020MNRAS.498.1920O},
      adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}

kinesis's Issues

ArviZ related experiments

Hi,

I am a participant of the Google Summer of Code with ArviZ, and I recently finished rewriting and extending the diagnostics and model comparison and checking functionalities. Now, as the last part of my project, I am trying out this new functions in real use cases. I saw this repository in ArviZ's dependencies page while searching for suitable examples and I saw that the model uses multivariate normal variables. Due to their shapes of the data and its pointwise likelihood they are very interesting to test for implementation bugs I may have missed in the CI tests. I hope you don't mind I experiment on my fork.

I am creating this issue basically to inform you about it, so that you may check my changes and (if you have time and energy) maybe comment on them to see if they would be useful at all for you. Also, I can send some pull requests for some minor things (i.e. I had some troubles installing the package that I managed to solve modifying slightly __init__.py and setup.py if it is of your interest I can create a PR).

For now, I have just started using ArviZ everywhere possible on the example notebook (I think it looks quite good in general, but I am not convinced with comparing models with plot_posterior), and I will implement the new diagnostics and stats functions in the following days.

Thanks for everything.