PHAT code for individual Bayesian stellar SED fitting (aka Stellar Nut Cracker)
Python Prolog
Latest commit d05cbfb Oct 10, 2012 Morgan First python version
Something is wrong with the units in the model grid. I will figure that out next

demo script works on generated data from the model grid. Should be internally

See notes from Dan to run the demo script.

Stellar Nut Cracker Code

Bayesian stellar SED fitting for PHAT.


Karl Gordon (STScI)


Help on Bayesian techniques for this code has come from:

  • David Hogg (NYU)
  • Dan Weisz, Morgan Fouesneau (UofWa)

Help on the stellar models for this code has come from:

  • Danny Lennon (STScI)

Support for this work from the Panchromatic Hubble Andromeda Treasury (PHAT)

  • Julianne Dalcanton (PI, UofWa)


Copyright 2011, 2012 the author. All rights reserved.

If you have interest in using or re-using any of this content, get in touch with Karl Gordon.

Revision history:

7 Aug 2012: Work enabling non-HST bandpasses. Getting ready for predicting fluxes!

7 Aug 2012: Basic testing and bugfixing done. Code runs and produces reasonable output. More extensive testing still needed. Enhancements: 1) Major changes to the output files output now down in sets of stars (chunks) to save creating a new file for every stars full likelihood output done as a sparse matrix - [indices of 4D matric,log likelihood values] output now much smaller 2) using log likelihoods instead of linear likelihoods 3) major update of variable names to be correct and clearer 4) streamlining of code to cache when possible to improve speed 5) output catalog has most probable values of model parameters instead of best fit (max likelihood)

6 Aug 2012: Finished cleanup and enhancement. Now to test the code.
Commiting to make sure nothing gets lost.

2 Aug 2012: First batch of cleaning and documentation Not finished yet and probably is currently broken some changes not backward compatible (testing needed)

25 Jul 2012: Commit of current code (needs cleaning and documentation) Prior work done outside of github.