sum and product models won't work with multiplicity (Trac #1022)

[pkienzle]

See discussion in ticket #1075 and pull request [https://github.com/SasView/sasmodels/pull/53 #281].

This will also be a problem in mixture.py, if you are combining two models, one of which is a multiplicity model (combining two multiplicity models should raise an error). In that case, the control parameter will need to be tagged with the appropriate 'A', 'B', ... prefix before assigning, and the hidden() function, if present, will need to be wrapped so that the parameter names in the returned set are also prefixed.

Other parameters in model_info.ModelInfo should not be a problem. profile, profile_axes (for spherical shell models) are probably going to be drawn from the form_factor, and so don't need to be included. single and opencl are used internally to select a calculator, and can be undefined for this model (this model should present as a python model).

Migrated from http://trac.sasview.org/ticket/1022

{
    "status": "new",
    "changetime": "2019-03-02T00:15:16",
    "_ts": "2019-03-02 00:15:16.703945+00:00",
    "description": "See discussion in ticket #1006 and pull request [https://github.com/SasView/sasmodels/pull/53 #53].\n\n\nThis will also be a problem in mixture.py, if you are combining two models, one of which is a multiplicity model (combining two multiplicity models should raise an error). In that case, the control parameter will need to be tagged with the appropriate 'A_', 'B_', ... prefix before assigning, and the hidden() function, if present, will need to be wrapped so that the parameter names in the returned set are also prefixed.\n\nOther parameters in model_info.ModelInfo should not be a problem. profile, profile_axes (for spherical shell models) are probably going to be drawn from the form_factor, and so don't need to be included. single and opencl are used internally to select a calculator, and can be undefined for this model (this model should present as a python model).\n",
    "reporter": "pkienzle",
    "cc": "",
    "resolution": "",
    "workpackage": "SasView Bug Fixing",
    "time": "2017-10-27T16:33:57",
    "component": "SasView",
    "summary": "sum and product models won't work with multiplicity",
    "priority": "major",
    "keywords": "",
    "milestone": "SasView 4.3.0",
    "owner": "",
    "type": "defect"
}

[butlerpd]

Trac update at 2018/02/13 02:40:27: butler changed milestone from "SasView 4.2.0" to "SasView 4.3.0"

[pkienzle]

Trac update at 2019/01/29 16:23:09: pkienzle commented:

A user wanted a fuzzy core-shell sphere plus a lorentzian. We did this by creating a sphere+lorentz model then edited the model, changing sphere to spherical_sld. The model works, but the user has to know to ignore the parameters for the extra shells.

Adapted model:

from sasmodels.core import load_model_info
from sasmodels.sasview_model import make_model_from_info

model_info = load_model_info('spherical_sld+lorentz')
model_info.name = 'snl'
Model = make_model_from_info(model_info)

[sasview-bot]

Trac update at 2019/01/29 16:26:35: Paul Kienzle pkienzle@nist.gov commented:

In changeset 15c80af4414c35749bdc01df46d248dfddd485d3:

#!CommitTicketReference repository="sasmodels" revision="15c80af4414c35749bdc01df46d248dfddd485d3"
propagate first multiplicity parameter to sum model. Refs #1089.

[sasview-bot]

Trac update at 2019/01/29 18:54:41: Paul Kienzle pkienzle@nist.gov commented:

In changeset b171acdab344b2827f9bdca29d66411c2484991f:

#!CommitTicketReference repository="sasmodels" revision="b171acdab344b2827f9bdca29d66411c2484991f"
move 'multiplicity' handling into sasview model. Refs #1089.

[sasview-bot]

Trac update at 2019/01/29 19:16:41: Paul Kienzle pkienzle@nist.gov commented:

In changeset 4f4d3e3574a18563fda8a9f5e84a9106aa7a1f1e:

#!CommitTicketReference repository="sasmodels" revision="4f4d3e3574a18563fda8a9f5e84a9106aa7a1f1e"
re-enable control parameter for rpa model. Refs #1089

[pkienzle]

Trac update at 2019/01/29 19:17:50: pkienzle commented:

Sasmodels has been fixed.

Sasview needs to be updated so that the list of available models presented on the sum/multi box in sasview includes models with multiplicity.

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.

[butlerpd]

Trac update at 2019/03/02 00:15:16: butler commented:

This is linked to #607 which is more the GUI side. When fully implemented this ticket should be closed.

[butlerpd]

exposing the multiplicity models to the sum|multiply interface should be fairly simple. However refactoring the fitting GUI to handle a mixture model with two multiplicities is not obvious. Recommend we punt on doing this in a final 4.x release and move to 5.1

[wpotrzebowski]

It still doesn't seem to work (from https://github.com/SasView/sasview/pull/2647):

#14:50:17 - ERROR: Fitting failed: Traceback (most recent call last): File "sas/qtgui/Perspectives/Fitting/FitThread.py", line 79, in compute File "sas/qtgui/Perspectives/Fitting/FitThread.py", line 19, in map_apply File "sas/qtgui/Perspectives/Fitting/FitThread.py", line 16, in map_getattr File "sas/sascalc/fit/BumpsFitting.py", line 265, in fit File "bumps/fitproblem.py", line 203, in FitProblem File "bumps/fitproblem.py", line 634, in __init__ File "bumps/fitproblem.py", line 258, in model_reset ValueError: Need more data points than fitting parameters
14:50:17 - ERROR: Need more data points than fitting parameters None
14:50:17 - ERROR: Traceback (most recent call last): File "sas/qtgui/Perspectives/Fitting/FittingWidget.py", line 2053, in fitComplete AttributeError: 'str' object has no attribute 'fitness'
14:50:35 - ERROR: Fitting failed: Traceback (most recent call last): File "sas/qtgui/Perspectives/Fitting/FitThread.py", line 79, in compute File "sas/qtgui/Perspectives/Fitting/FitThread.py", line 19, in map_apply File "sas/qtgui/Perspectives/Fitting/FitThread.py", line 16, in map_getattr File "sas/sascalc/fit/BumpsFitting.py", line 265, in fit File "bumps/fitproblem.py", line 203, in FitProblem File "bumps/fitproblem.py", line 634, in __init__ File "bumps/fitproblem.py", line 258, in model_reset ValueError: Need more data points than fitting parameters
14:50:35 - ERROR: Need more data points than fitting parameters None
14:50:35 - ERROR: Traceback (most recent call last): File "sas/qtgui/Perspectives/Fitting/FittingWidget.py", line 2053, in fitComplete AttributeError: 'str' object has no attribute 'fitness'

[krzywon]

@wpotrzebowski, I just tested this locally using the branch in #2647. I added the core_multi_shell and the cylinder models together using Add/Multiply Models. I'm able to run fits and calculate theories using the resulting plugin model.

Can you walk me through the steps you took to get to the error you are seeing?

[wpotrzebowski]

I think I tested wrong installer as I am not able to reproduce it - sorry.

[butlerpd]

Does this mean this issue has been fixed @krzywon and @wpotrzebowski ? If so we can we close it? It is a rather old issue so nice to finally have it done 😄

[butlerpd]

This should have been closed with the merging of #2647. Closing it manually.