How to save a single source model

[HealthyPear]

Assume I call the sed method on my gta to get the catalog-like SED

now I want to test a different spectrum

I am doing (feel free to criticize if this is not the usual way to do it)

spectrum_pars = {
    "Prefactor": {
        ... # value, scale, min, max
    },
    "Index1": {
       ...
    },
    "Index2": {
        ...
    },
    "BreakValue": {
        ...
    },
}

gta.set_source_spectrum(
    target_source,
    spectrum_type="BrokenPowerLaw",
    spectrum_pars=spectrum_pars,
    update_source=True
)

and then the usual pre-fit procedures, optimize / fit then ok it works

I want to make this reproducible, so I would like to save just this single source model to a file, so that I can call spectrum_pars from a file

Source. write_xml() seems to be a function for this, but honestly I feel like

with open(output_path / "custom_spectral_model.yaml", mode="w") as f:
    yaml.safe_dump(spectrum_pars, f)

it's better

Of course, I know about gta.write_roi(), but what I want to load is something that I might want to define and save before updating the entire model.

What do you think?

[MiltosMichailidis]

Your YAML approach sounds reasonable to me for this kind of use case. Fermipy already provides methods like write_xml() for exporting the full ROI model and individual source XMLs, but those are more geared toward full model persistence and Fermitools interoperability. Of course, they work perfectly fine as well for what you want to achieve.

Having said that, what you are trying to save here is really just a reusable spectral-model setup, and for that a small YAML file honestly feels easy to work with. I do not think there is a single "right" way to do this -- it mostly depends on what fits best with your workflow and what information you actually want to preserve.