lamellarPS in sasview, bugs in polydispersity loops (and Caille S(Q) ) (Trac #376)

[RichardHeenan]

Whilst debugging equivalent lamellarCaille routine for sasmodels,

discovered the old one had two bugs in the polydispersity.

Later Paul K also found a separate issue with a summation in the

Caille S(Q) itself. I need to check the status of the fix for this

also for this and lamellarPS_HG.

The polydispersity bugs are illustrated here:

// Perform the computation, with all weight points
double sum = 0.0;
double norm = 0.0;

// Loop over short_edgeA weight points
for(int i=0; i< (int)weights_spacing.size(); i++) {
dp[1] = weights_spacing[i].value;
//for(int j=0; j< (int)weights_spacing.size(); j++) { BUG
for(int j=0; j< (int)weights_delta.size(); j++) {
//dp[2] = weights_delta[i].value; BUG
dp[2] = weights_delta[j].value;

  sum += weights_spacing[i].weight * weights_delta[j].weight * LamellarPS_kernel(dp, q);
  norm += weights_spacing[i].weight * weights_delta[j].weight;
}

}
return sum/norm + background();

Migrated from http://trac.sasview.org/ticket/376

{
    "status": "closed",
    "changetime": "2015-03-24T16:12:32",
    "_ts": "2015-03-24 16:12:32.735829+00:00",
    "description": "Whilst debugging equivalent lamellarCaille routine for sasmodels, [[BR]]\ndiscovered the old one had two bugs in the polydispersity. [[BR]]\nLater Paul K also found a separate issue with a summation in the [[BR]]\nCaille S(Q) itself. I need to check the status of the fix for this [[BR]]\nalso for this and lamellarPS_HG.\n\nThe polydispersity bugs are illustrated here:\n\n  // Perform the computation, with all weight points\n  double sum = 0.0;\n  double norm = 0.0;\n\n  // Loop over short_edgeA weight points\n  for(int i=0; i< (int)weights_spacing.size(); i++) {\n    dp[1] = weights_spacing[i].value;\n    //for(int j=0; j< (int)weights_spacing.size(); j++) {    BUG\n    for(int j=0; j< (int)weights_delta.size(); j++) {\n      //dp[2] = weights_delta[i].value;        BUG\n      dp[2] = weights_delta[j].value;\n\n      sum += weights_spacing[i].weight * weights_delta[j].weight * LamellarPS_kernel(dp, q);\n      norm += weights_spacing[i].weight * weights_delta[j].weight;\n    }\n  }\n  return sum/norm + background();\n",
    "reporter": "richardh",
    "cc": "",
    "resolution": "fixed",
    "workpackage": "SasView Bug Fixing",
    "time": "2015-03-09T16:34:17",
    "component": "SasView",
    "summary": "lamellarPS in sasview, bugs in polydispersity loops (and Caille S(Q) )",
    "priority": "minor",
    "keywords": "lamellarPS",
    "milestone": "SasView 3.1.0",
    "owner": "richardh",
    "type": "defect"
}

[RichardHeenan]

Trac update at 2015/03/09 16:47:36: richardh commented:

Have fixed the Caille s(Q) in sasmodels for lamellarCaille, as Paul K had only done lamellarCailleHG. RichardH

[RichardHeenan]

Trac update at 2015/03/09 18:04:47:

• richardh changed _comment0 from "Have now pushed the fixes for lamellarPS and lamellarPS_HG in sasview, can close the ticket." to "1425924465577707"
• richardh commented:

Have now pushed the fixes for lamellarPS and lamellarPS_HG in sasview. Alas they will now fail the current unit tests, as the results of the routines have changed!

[RichardHeenan]

Trac update at 2015/03/09 18:33:01: richardh commented:

Have made a quick fix to utest_other_models.py , have to go home now, will check more tomorrow.

[pkienzle]

Trac update at 2015/03/24 16:12:32:

• pkienzle changed resolution from "" to "fixed"
• pkienzle changed status from "new" to "closed"