Verify that BUMPS is computing reduced chisq correctly (Trac #270)

[smk78]

The Park optimiser is not computing reduced chisq correctly resulting
in it returning uncertainties that are larger (by a factor 5 - 10)
than they should be.

See attached files.

Migrated from http://trac.sasview.org/ticket/270

{
    "status": "closed",
    "changetime": "2016-03-20T12:07:58",
    "_ts": "2016-03-20 12:07:58.883340+00:00",
    "description": "The Park optimiser is not computing reduced chisq correctly resulting\nin it returning uncertainties that are larger (by a factor 5 - 10) \nthan they should be.\n\nSee attached files.",
    "reporter": "smk78",
    "cc": "",
    "resolution": "wontfix",
    "workpackage": "SasView QA and testing",
    "time": "2014-07-15T09:35:15",
    "component": "SasView",
    "summary": "Verify that BUMPS is computing reduced chisq correctly",
    "priority": "critical",
    "keywords": "Park, uncertainties, chisq",
    "milestone": "SasView 4.0.0",
    "owner": "",
    "type": "task"
}

[smk78]

Trac update at 2014/07/15 09:36:03:

• smk78 changed attachment from "" to "RE Error estimates from fits.pdf"
• smk78 commented:

Initial error report

[smk78]

Trac update at 2014/07/15 09:36:37:

• smk78 changed attachment from "" to "RE Error estimates from fits - updated.pdf"
• smk78 commented:

Evaluation of the promblem by Paul K

[smk78]

Trac update at 2014/07/15 09:37:53: smk78 commented:

Since the Park optimiser is to be replaced by Bumps, does this error need to be fixed?

[ajj]

Trac update at 2015/01/14 12:37:22: ajj changed workpackage from "" to "SasView Bug Fixing"

[butlerpd]

Trac update at 2015/02/15 15:19:14:

• butler changed owner from "" to "smk78"
• butler changed status from "new" to "assigned"
• butler changed summary from "Park optimiser is not computing reduced chisq correctly" to "Verify that BUMPS is computing reduced chisq correctly"

[butlerpd]

Trac update at 2015/02/15 15:21:22: butler commented:

OOps .. should have added the comment when changing title from "PARK does not compute ..." to the current title. Since we are removing PARK the ticket made no sense. However Paul suggests we double check that BUMPS is doing it correctly before closing this ticket. So have renamed.

[butlerpd]

Trac update at 2015/03/10 14:32:55: butler commented:

check that LM and the Nelder Mead simplex give similar results using previous set from the PARK ticket. Steve to submit

[smk78]

Trac update at 2015/03/10 16:07:54: smk78 commented:

Added email thread with Users SasView 3.0.0 fit uncertainties, and users data set.