Notes for getting started with Starfish

[gully]

Hi folks, I recently emailed some Starfish quick start notes to a colleague, and thought I'd post them here in case they are useful to anyone else. More tips/tricks for newcomers are welcomed below.

Here are some general tips for getting started with Starfish:

All the code is on GitHub, either in Ian Czekala's original version, or my fork: Ian's original version: https://github.com/iancze/Starfish My fork: https://github.com/gully/Starfish

My active branch for the last year supports all sorts of extensions to Starfish, including:

• The ability to sample all stellar and nuisance parameters at once with the robust "emcee" algorithm.
• Nominal support for using the Marley et al. synthetic spectral models (in prep.)
• Several extensions to support "separable" model physics: mixture models from starspots, spectroscopic binaries, disk veiling, extinction, etc.
• Correctly supports flux-calibrated spectra

All of these extensions are undocumented here: https://github.com/gully/Starfish/tree/mix_model_omega2

The best place to get started on understanding the code is:

• Read Ian's paper http://adsabs.harvard.edu/abs/2015ApJ...812..128C
• Read Section 3 of my paper http://adsabs.harvard.edu/abs/2017ApJ...836..200G
• Review these slides on how the emcee MCMC algorithm works: https://speakerdeck.com/dfm/data-analysis-with-mcmc
• Review this slideshow on probabilistic programming in Python:
  https://speakerdeck.com/dfm/pyastro16
• Review these slides on Gaussian Processes for Astronomy, only up to slide 60: https://speakerdeck.com/dfm/an-astronomers-introduction-to-gaussian-processes-v2
• Read the documentation and demos here: http://iancze.github.io/Starfish/current/index.html

The most up-to-date info and conversations about the code occur in the open in the GitHub issues. I recommend a newcomer read through these to get a sense for the limitations and opportunities of the code. https://github.com/iancze/Starfish/issues

All my code for my 2017 paper is free/public here, including some helpful Jupyter-notebooks: https://github.com/BrownDwarf/welter/tree/master/notebooks

[iancze]

Hi @gully, thanks for compiling this list, this is very helpful! Perhaps we can use this as a thread to collect various sources of documentation, for eventual incorporation into the documentation (embed or link).