Inconsistency in Factory parameter setting

[arthurmloureiro]

Describe the bug It seems like setting some parameters in the factory like alphag will actually fix those parameters inside firecrown instead of simply instantiating the factory objects with those values. This can will an issue for CosmoAPI as it will create a discrepancy between factory parameters for synthetic analysis.

For example, at the moment we can set the parameters for a synthetic analysis in CosmoAPI with

    systematics:
      type: WeakLensingFactory
      per_bin_systematics:
      - type: MultiplicativeShearBiasFactory
        mult_bias: [1e-3, 1e-3, 1e-3, 1e-3, 1e-3]
      - type: PhotoZShiftFactory
        delta_z: &delta_z [0.2, 0.1, 0.1, 0.2, 0.1]
      global_systematics:
      - type: LinearAlignmentSystematicFactory
        alphag: 1.0

This allows the user to set the parameters to generate a fake SACC data file. However, if the user recycles this part of the code to then run a likelihood analysis, alphag will be fixed while the other parameters are not.

This is not really a bug just something that can cause some issues as different systematic factories seem to have different behaviour when constructing them.

[vitenti]

In #193, a request was made to allow configuring Firecrown so that specific parameters would be entirely hidden from the sampler—meaning they would not appear as sampled parameters or in configuration files like values.ini in CosmoSIS. This was implemented in #194.

Please let us know if this causes any issues so we can address them. Otherwise, please close the issue. @arthurmloureiro @tilmantroester

[vitenti]

Closing for lack of feedback. If the issue can be described more clearly please open a new issue.