Opened 7 months ago

Last modified 6 months ago

#1242 new defect

Resolution smearing is only applied to positive Qx and Qy in 2D

Reported by: smk78 Owned by:
Priority: blocker Milestone: SasView 4.2.2
Component: SasView Keywords:
Cc: Work Package: SasView Bug Fixing

Description

User Bei Tian notes that in 2D fitting/calculation, instrument smearing is only applied to positive values of Qx and Qy.

Attachments (1)

bei.png.jpg (195.3 KB) - added by smk78 7 months ago.
Screen shot from Bei

Download all attachments as: .zip

Change History (6)

Changed 7 months ago by smk78

Screen shot from Bei

comment:1 Changed 7 months ago by smk78

@pkienzle says:

Here's a quick fix.  A full solution involves thinking a lot more about what happens to 
pixels near the origin wrt phi in the polar representation.

diff --git a/src/sas/sasgui/perspectives/fitting/fitpage.py 
b/src/sas/sasgui/perspectives/fitting/fitpage.py
index f3baf06b9..ad82c3081 100644
--- a/src/sas/sasgui/perspectives/fitting/fitpage.py
+++ b/src/sas/sasgui/perspectives/fitting/fitpage.py
@@ -2364,8 +2364,8 @@ class FitPage(BasicPage):
            elif self.dx_percent is not None:
                percent = self.dx_percent/100
                if self._is_2D():
-                    data.dqx_data[data.dqx_data == 0] = percent * data.qx_data
-                    data.dqy_data[data.dqy_data == 0] = percent * data.qy_data
+                    q = np.sqrt(data.qx_data**2 + data.qy_data**2)
+                    data.dqx_data = data.dqy_data = percent*q                 else:
                    data.dx = percent * data.x
            self.current_smearer = smear_selection(data, self.model)

comment:2 Changed 7 months ago by smk78

@pkienzle also says:

The code for generating resolution should not be in the GUI.  Put it in 
sascalc/fit/qsmearing.py or sasmodels/resolution2d.py so that the qt gui 
can use the same code.

comment:3 Changed 6 months ago by pkienzle

A proper estimate of resolution requires too many details about the experiment configuration for the fitting panel. Could we perhaps attach it to the Q resolution estimator window that we already have?

Note: playing with resolution estimator, it seems to always give the same value for sigma x and sigma y even when the resolution is clearly an ellipse. That should be fixed before using the calculator elsewhere. Hammouda's online SANS course (ch 15, pg 9) has formulas for sigma x and sigma y.

comment:4 Changed 6 months ago by pkienzle

Resolution calculator note turned into Ticket #1244.

comment:5 Changed 6 months ago by butler

  • Milestone changed from SasView 4.3.0 to SasView 4.2.2
Note: See TracTickets for help on using tickets.