"Fitting did not converge!!!" error with a Sum|Multi plugin model

[smk78]

Reported by User alessandra.delgiudice@…:

Dear SASview developer,

I found problems making fits with Sum|Multi plugin models in the 4.1.2 version. In brief:

In the 4.1.2 version I got a "Fitting did not converge!!!" error, whereas trying with the 4.1 version (installed on the same pc in a different folder) the fit worked.

I did a test with simulated data:

I created a Sum|Multi model with the dedicated tool: sphere_plus_hollowcylinder.py

I saved a calculated curve with parameters:
p1_scale 0.5
p1_background 0
p1_sld 9.45
p1_sld_solvent 9.4
p1_radius 100
p2_scale 0.5
p2_background 0
p2_radius 50
p2_thickness 20
p2_length 1000
p2_sld 9.45
p2_sld_solvent 9.4
scale_factor 1

and 1400 points as sphere_and_hollow_cylinder_out.txt

I loaded this file as a data file to be fitted. I changed the sphere radius value from 100 to 50, and then clicked to "Fit". Nothing happened and I got a "Fitting did not converge!!!" error in the console.

I had an older installation file for SASview 4.1 and I installed it on my laptop in the folder SasView_4_1 instead of SasView (where the 4.1.2 version is)

The same test worked in this case.

When trying a similar test with a simpler model (sphere with default parameters), the fit starting from a different value of the sphere radius (100 instead of 50) worked with the 4.1.2 version.

I attach the sasview log as sasview_20180112.log. What I described starts with the line "2018-01-12 11:50:05,181 INFO —- SasView session started —-"

[smk78]

Alessandras log file

[smk78]

Alessandras PC was Windows 10 Home on a Dell XPS 9360 with processor Intel® Core™ i7-7500U CPU @ 2.70GHz, 2901 Mhz, 2 cores, 4 processors.

I am also using a W10 PC and managed to reproduce this behaviour.

[smk78]

Wojtek has now reproduced this behaviour on a Mac using 4.1.2

[butler]

What is the status of this ticket? Does the problem still exist in 4.2 (i.e. master build?) or can we close this as fixed in 4.2?

[smk78]

Screenshot of SasView? Master pulled 13/02/2018 10:30GMT

[smk78]

Have just pulled from master, built locally (32 bit, tinycc=no64) on Win 7, and re-tried Alessandra's instructions. SasView fitted the model data it had just created. Screen shot attached. The problem seems to be fixed.