Fitting result changes depending on log level

[Michael Wedel]

While I was doing some experimental work on POLDI 2D-fitting, I discovered that the result depends on the log level. For log-level information or debug, the fit converges differently and gives a different result.

Since this was experimental work, I will just push the small changes I made to reproduce this problem.

[Michael Wedel]

Refs #10515. Adding latest parameters for comparison.

Current software works with those parameters, so for comparison they are required.

Changeset: 4d928438877bfe718c60382c932b814e7c738455

[Michael Wedel]

Refs #10515. Small changes to PoldiCalculateSpectrum2D

Changeset: 06dafd9224fea8bd0a395941d9f419c52be85947

[Michael Wedel]

Refs #10515. Adding test data

Changeset: 3f4e3053ec31c3ef6f03c37793713e360fc0942d

[Michael Wedel]

Script (UsageData-directory must be configured as data directory):

fileRange = range(73,76)

workspaces = []
for i in fileRange:
	wsName = "243%s" % (i)
	workspaces.append(wsName)
	LoadSINQFile("poldi2014n0243%s.hdf" % (i), "POLDI", OutputWorkspace=wsName)
	LoadInstrument(wsName, InstrumentName="POLDI")
	
SiWs = GroupWorkspaces(InputWorkspaces=','.join(workspaces))
Merged=PoldiMerge(SiWs)

PoldiAutoCorrelation(InputWorkspace='Merged', OutputWorkspace='corr')
PoldiPeakSearch(InputWorkspace='corr', MaximumPeakNumber=30, OutputWorkspace='corrPeaks')
PoldiFitPeaks1D(InputWorkspace='corr', FwhmMultiples=4, PoldiPeakTable='corrPeaks')
DeleteTableRows(TableWorkspace='RefinedPeakTable', Rows='10-15')
PoldiCalculateSpectrum2D(InputWorkspace='Merged', PoldiPeakWorkspace='RefinedPeakTable', PeakProfileFunction='Gaussian', OutputWorkspace='d')

[Stuart Campbell]

This ticket has been transferred to github issue ​11357