search

astrofrog / sedfitter

Python version of the SED fitter from Robitaille et al., 2007, ApJS 169 328
http://sedfitter.readthedocs.org
BSD 2-Clause "Simplified" License
20 stars 22 forks source link

Difference in fitting results compared to fortran sedfitter #62

Open mattpovich opened 6 years ago

mattpovich commented 6 years ago

In a test fit of Kurucz model atmospheres to a large (150,000) catalog of combined UKIDSS/2MASS and GLIMPSE point sources in a chunk of the Galactic midplane, I find ~10% more badly fit sources using the python sedfitter than previously with the fortran sedfitter. Best of my knowledge* all fitting parameters are the same (extinction law is your old ex_law_gl.par file, cpd=2, minimum datapoints = 4).

*Am I correct that the format/units of old extinction law files, including lots of extraneous columns, are still appropriate for the python sedfitter?

astrofrog commented 6 years ago

The old extinction files should still work with the new SED fitter. There was a bug I fixed in the Python SED fitter that may have been present in the old fitter that changes the reduced chi^2 values a little. Could you try and compare the chi^2 values for the best fit before and after, and see if the agreement depends on the number of valid fluxes?

mattpovich commented 6 years ago

Yes, I'll see if I can do that when I get some time. Would you be more interested in comparisons using kurucz atmospheres or r06 models? (I have no plans to use the old fortran fitter with the r17 models :-) )

On Thu, Jul 26, 2018 at 6:33 AM Thomas Robitaille notifications@github.com wrote:

The old extinction files should still work with the new SED fitter. There was a bug I fixed in the Python SED fitter that may have been present in the old fitter that changes the reduced chi^2 values a little. Could you try and compare the chi^2 values for the best fit before and after, and see if the agreement depends on the number of valid fluxes?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/astrofrog/sedfitter/issues/62#issuecomment-408099008, or mute the thread https://github.com/notifications/unsubscribe-auth/AnrfIC1J4-JSwuLoeSOmhakgzjmxH7r4ks5uKcVAgaJpZM4VdFG- .

-- --

Matthew S. Povich Associate Professor Department of Physics and Astronomy College of Science Building 8, Rm 216 Cal Poly Pomona http://www.cpp.edu/~mspovich/ mspovich@cpp.edu Office: (909) 869-3608 Mobile: (617) 306-0411