Multiplicity (and control parameter) handling

[butlerpd]

Describe the bug
sasmodels implements control parameters which are hidden. They are documented (lightly) in the developer docs, in particular at:
https://www.sasview.org/docs/dev/sasmodels-api/sasmodels.html#submodules

However, the GUI does not interpret/use them correctly. These are used in the so-called multiplicity models which includes the RPA, the core_multishell, the unified_power_Rg, and the onion models. The RPA is different from the others in that it is really a series of cases rather than a number of shells. While there are issues with all, the RPA is so broken in the GUI that it has been removed form version 5.

Someone needs to relook at the GUI interface to sasmodels with an eye to properly working with control parameters.

This is essentially the issue reported in sasmodels: SasView/sasmodels#396. That really should be a sasview issue as it is a GUI issue so linking that discussion here but will close that issue.

Expected behavior
It should work (give the correct answers)

SasView version (please complete the following information):

• certainly all of 5.x though there are suggestions that RPA may not have worked even in 4.x

Operating system (please complete the following information):

• OS: all

Additional context
Some other related sasview issues are: #607, #965, #1089 (which seems to be the same as #607), #1106, #1816. Note that some of these may have been fixed without closing the issue?

[RichardHeenan]

Additionally, adding S(Q) to say core_multi_shell does not give the correct results, nor does it give any error messages to the user. The attached project file will show this. The fits on the two plots below should, if I got the models right, be the same!

The core_multi_shell is not adding separate P(Q) and s(Q) to the data explorer.

More worrying still, if you remove the S(Q) and then fit just "scale" the core_multi_shell still does not give quite the same result as core_shell_sphere (no plots here). How different they are depends on the value of the flat background. I suspect that for core_multi_shell the flat background is perhaps not being included in the chi squared (though is there in the I(Q) on the plot).

Basically the gui seems to have a lot of issues with multiplicity models, which will not be apparent to the average user.

test_multishellSQ_19Sept2023_506release.json.txt

[smk78]

As far as the rpa model is concerned, the consensus at the 2024 Contributor Camp roadmap discussion was that this model should not be re-implemented in the form it once was (ie, using a control variable to select the calculation case). Instead, each case should be implemented as a separate model.

[butlerpd]

In closing issue #1089 with the merging of PR #2647, @pkienzle's comment about replacing the old multiplicity concept being maintained in sasview with sasmodels' more sophisticated control and hidden parameters is copied over here.

Could redo the sasview model table view so that parameter list is updated whenever any control parameter is updated (sasmodels parameters each have an "is_control" attribute), and drop the notion of a "multiplicity" model.