Likelihood profiles from gta.fit() and free sources using gta.free_sources

[danyanim2006]

Data Analysis Question

How can I visualise the 2D likelihood profile as a function of the prefactor and index values after a powerlaw fit. In the .npy files generated after the fit, I could find the info such as dloglike_scan, norm_scan, and flux_scan, but did not find index_scan.

And when I free sources in the roi using gta.free_sources, is there a way to get the list of sources/parameters that are freed just due to this setting (without having to compare the xml files).

[MiltosMichailidis]

Thanks for reaching out!

For a complete understanding of likelihood analysis results and the data saved when you run gta.write_roi('output.npy'), please refer to this resource: https://fermipy.readthedocs.io/en/latest/output.html. Note that index_scan is not included in the saved parameters.

Concerning the sources, parameters, and errors that are allowed to vary when you use a command like gta.free_sources(distance=5.0,pars='norm'), you can track these during the analysis.
Use gta.print_model() to see which sources have varying parameters (marked with an asterisk in the last column).
For the specific parameters being varied, their fitted values, and errors for each source, use gta.print_params().

This information is also conveniently stored in the fermipy.log file located in your outdir.