Add back support for fitparam_values arg in public data analysis flux conversion

[tomaskontrimas]

It seems to work as the new Using fitparam_values with ana should match the reference analysis without fitparam_values test passes:

skyllh/tests/publicdata_ps/test_time_integrated_ps.py

Line 380 in b56df81

def test_ana_with_fitparam_values_matches_ana_ref(self):

[Copilot]

Pull request overview

This PR restores support for evaluating public-data flux conversions at explicit fit-parameter values, especially for gamma-dependent scaling factors.

Changes:

• Adds fitparam_values plumbing through analysis-level flux conversion helpers.
• Allows MultiDatasetSignalGenerator to evaluate scaling factors with an explicit source-parameter recarray.
• Adds regression/integration tests for fit-parameter-based flux conversion and unsupported combinations.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.

File	Description
skyllh/core/analysis.py	Adds fitparam_values support to scaling, mu2flux, and flux2mu.
skyllh/core/signal_generator.py	Adds optional source-parameter recarray support and MC-generator rejection path.
tests/core/test_signal_generator.py	Adds unit coverage for the MC-generator src_params_recarray guard.
tests/publicdata_ps/test_time_integrated_ps.py	Adds public-data integration tests for fit-parameter flux conversion behavior.

Comments suppressed due to low confidence (5)

skyllh/core/analysis.py:1771

• This signature reorders the existing second positional argument. Existing code using mu2flux(mu, True) for per-source conversion will now treat True as fitparam_values, so the call no longer returns per-source fluxes and may raise during parameter mapping. Preserve the old positional order or make the new parameter keyword-only.

    def mu2flux(self, mu, fitparam_values=None, per_source=False):

skyllh/core/analysis.py:1792

• This has the same positional-argument compatibility break as mu2flux: existing flux2mu(flux_norm, True) calls now pass a boolean as fitparam_values instead of setting per_source=True. Keep per_source in the original positional slot or require fitparam_values to be keyword-only.

    def flux2mu(self, flux_norm, fitparam_values=None, per_source=False):

skyllh/core/analysis.py:1769

• This unconditionally requires every signal generator implementation to accept the new src_params_recarray keyword, even when fitparam_values is not provided. Existing custom sig_generator_cls implementations that support the previous fluxmodel_scaling_factor(per_source=False) interface will now fail on default flux conversions; only pass the new keyword when needed or keep a compatibility path.

        return self._sig_generator.fluxmodel_scaling_factor(
            per_source=per_source,
            src_params_recarray=src_params_recarray,
        )

skyllh/core/signal_generator.py:335

• When a custom src_params_recarray is provided, this recalculates detector yields at the requested parameters but the later energy-range correction factors remain cached from the reference flux model. For configured energy ranges those factors depend on the spectral parameters, so direct calls to this method can return stale scaling factors even though the analysis-level wrapper rejects that combination.

        src_detsigyield_weights_service.calculate(src_params_recarray=src_params_recarray)

skyllh/core/signal_generator.py:756

• This also reorders the existing per_source positional argument for the MC signal generator. Existing fluxmodel_scaling_factor(True) calls now pass True as src_params_recarray and hit the new NotImplementedError instead of returning per-source factors; keep per_source in the original positional slot or make the new argument keyword-only.

    def fluxmodel_scaling_factor(self, src_params_recarray=None, per_source=False):

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

[tomaskontrimas]

All existing fluxmodel_scaling_factormethods were updated by this PR

[tomaskontrimas]

I'd say it is fine as it is for now

[chiarabellenghi]

This is good to go for me