Reorganise form factor parametrizations

[dvandyk]

Proposal to reorganise the form factor parametrizations.

• General parametrizations:

  • SE for Series Expansion; would cover BFW2010, BMRvD2022, ABR2022; possibly BGL1997 (which has no tensor FFs)
  • SSE for Simplified Series Expansion; would cover BSZ2015; SSE should be the default set of form factors wherever possible.
  • default parameters for both should be set to the current best-fit values and their uncertainties

• Specialized parametrizations

  • to be kept, if they are valueable and provide additional functionality (i.e., if they support processes not covered by the above general parametrizations)
  • default parameters should be set to zero (see e.g. Set default value of HQET parameters (Isgur-Wise functions) to zero #787)
    • BGJvD2019
    • BFvD2014 (remove entirely?)
    • FvDV2018
    • KMPW2010
    • KKvDZ2022 (possibly keep this one since there is no SE/SSE parametrization available for B->gamma^* form factors)
  • these parametrizations should issue a warning when they are used for the first time, indicating that their default parameters are set to zero (Add utility class for one-time warnings #791)

[gubernari]

I am not sure that the following suggestion is the optimal:
default parameters for both should be set to the current best-fit values and their uncertainties
This might be good for a basic user, but when performing analysis can lead to mistakes, that is using the defaults instead of the fitted values. Maybe add some warning or something that helps preventing this?

[dvandyk]

I am not sure that the following suggestion is the optimal: default parameters for both should be set to the current best-fit values and their uncertainties This might be good for a basic user, but when performing analysis can lead to mistakes, that is using the defaults instead of the fitted values. Maybe add some warning or something that helps preventing this?

@gubernari That's a trade-off between being user friendly and correct. Default values for the default parametrization(s) should be present IMHO. The heavy-quark expansion should be left to people who know what they're doing, hence we should zero out its parameters.

[gubernari]

Understood.
Probably for the future one could think about setting all the parameters to zero, then add method like eos.load(), which can give you the possibility to load your favorite set of parameters (like eos.load('B>K^*::BSZ2015-paper_ref').
This would also allow to compare different set of results and avoid to constantly update the parameters files, keeping also old results.

[mreboud]

I am have some doubts about item 1.3: The uncertainties are meaningless in the absence of correlations, I would only provide the central values for the parameters