3.12

DirectILLDiagnostics v1

../_images/DirectILLDiagnostics-v1_dlg.png

DirectILLDiagnostics dialog.

Summary

Perform detector diagnostics and masking for the direct geometry TOF spectrometers at ILL.

Properties

Name Direction Type Default Description
InputWorkspace Input MatrixWorkspace Mandatory A ‘raw’ workspace from DirectILLCollectData to calculate the diagnostics from.
OutputWorkspace Output Workspace Mandatory A diagnostics mask workspace.
Cleanup Input string Cleanup ON What to do with intermediate workspaces. Allowed values: [‘Cleanup ON’, ‘Cleanup OFF’]
SubalgorithmLogging Input string Logging OFF Enable or disable subalgorithms to print in the logs. Allowed values: [‘Logging OFF’, ‘Logging ON’]
EPPWorkspace Input TableWorkspace   Table workspace containing results from the FindEPP algorithm.
ElasticPeakDiagnostics Input string Peak Diagnostics AUTO Enable or disable elastic peak diagnostics. Allowed values: [‘Peak Diagnostics AUTO’, ‘Peak Diagnostics ON’, ‘Peak Diagnostics OFF’]
ElasticPeakWidthInSigmas Input number 3 Integration width of the elastic peak in multiples of ‘Sigma’ in the EPP table.
ElasticPeakLowThreshold Input number 0.1 Multiplier for lower acceptance limit used in elastic peak diagnostics.
ElasticPeakHighThreshold Input number 3 Multiplier for higher acceptance limit used in elastic peak diagnostics.
ElasticPeakErrorThreshold Input number 3.3 To fail the elastic peak diagnostics, the intensity must also exceed this number of error bars with respect to the median intensity.
BkgDiagnostics Input string Bkg Diagnostics AUTO Control the background diagnostics. Allowed values: [‘Bkg Diagnostics AUTO’, ‘Bkg Diagnostics ON’, ‘Bkg Diagnostics OFF’]
NonBkgRegionInSigmas Input number 10 Width of the range excluded from background integration around the elastic peaks in multiplies of ‘Sigma’ in the EPP table
NoisyBkgLowThreshold Input number 0.1 Multiplier for lower acceptance limit used in noisy background diagnostics.
NoisyBkgHighThreshold Input number 3.3 Multiplier for higher acceptance limit used in noisy background diagnostics.
NoisyBkgErrorThreshold Input number 3.3 To fail the background diagnostics, the background level must also exceed this number of error bars with respect to the median level.
BeamStopDiagnostics Input string Beam Stop Diagnostics AUTO Control the beam stop diagnostics. Allowed values: [‘Beam Stop Diagnostics AUTO’, ‘Beam Stop Diagnostics ON’, ‘Beam Stop Diagnostics OFF’]
BeamStopThreshold Input number 0.67 Multiplier for the lower acceptance limit for beam stop diagnostics.
DefaultMask Input string Default Mask ON Enable or disable instrument specific default mask. Allowed values: [‘Default Mask ON’, ‘Default Mask OFF’]
MaskedDetectors Input int list   List of spectra to mask.
MaskedComponents Input str list   List of instrument components to mask.
OutputReportWorkspace Output TableWorkspace   Output table workspace for detector diagnostics reporting.
OutputReport Output string   Diagnostics report as a string.

Description

This algorithm performs detector diagnostics and masking. It is part of ILL’s direct geometry data reduction suite. The diagnostics are calculated using the counts from InputWorkspace which is preferably the raw workspace provided by the OutputRawWorkspace property of DirectILLCollectData. The output is a special mask workspace which can be further fed to DirectILLReduction to mask the detectors diagnosed as bad. Optionally, an instrument specific default mask, a beam stop mask and/or a user specified hard mask given by MaskedDetectors or MaskedComponents can be added to the diagnostics mask.

A workflow diagram for the diagnostics is shown below:

../_images/DirectILLDiagnostics-v1_wkflw.png

Diagnostics performed

The algorithm performs two tests for each spectrum in InputWorkspace: elastic peak diagnostics and flat background diagnostics. Basically both tests calculate the median of the test values over all spectra, then compare the individual values to the median. For more detailed information, see MedianDetectorTest.

Elastic peak diagnostics

The EPP table given in EPPWorkspace and the value of ElasticPeakWidthInSigmas are used to integrate the spectra around the elastic peaks, giving the elastic intensities. The intensities are further normalised by the opening solid angles of the detectors, given by SolidAngle before the actual diagnostics.

Flat background diagnostics

Similarly to elastic peak diagnostics, EPPWorkspace and NonBgkRegionInSigmas are used to integrate the time-independent background regions of InputWorkspace. NonBkgRegionInSigmas is a factor applied to the ‘Sigma’ column in EPPWorkspace and this interval around the elastic peak positions is excluded from the integration. No opening angle corrections are applied to the background diagnostics.

Beam stop

The shadow cast on the detectors by a beam stop can be masked by the diagnostics, as well. This functionality is automatically enabled when ‘beam_stop_diagnostics_spectra’ instrument parameter is defined and can be disabled by BeamStopDiagnostics. The algorithm tries to mask a continuous region within the spectra listed in ‘beam_stop_diagnostics_spectra’. The BeamStopThreshold property can be used to fine-tune the operation.

The ‘beam_stop_diagnostics_spectra’ instrument parameter lists ranges of spectrum numbers. Each range should cover a region of a physical detector tube, part of which is behind the beam stop.

The masking procedure proceeds as follows:

  1. Pick a range from ‘beam_stop_diagnostics_spectra’.
  2. Integrate the spectra within the range.
  3. Divide the range into two halves from the middle.
  4. Pick one of the halves, take the maximum integrated value.
  5. Starting from the spectrum containing the maximum value, and stepping towards the center of the range, find the first spectrum where the integrated intensity is less than the maximum intensity multiplied by BeamStopThreshold. Lets call this the threshold spectrum.
  6. Mark all spectra from the middle of the range to the threshold spectrum as masked.
  7. Repeat for the other half.

Defaul mask

The default mask file is defined by the ‘Workflow.MaskFile’ instrument parameter.

Currently, there is a default mask available for ILL’s IN5 instrument which masks 8 pixels at both ends of every detector tube.

Diagnostics reporting

The optional OutputReportWorkspace property returns a table workspace summarizing the diagnostics. The table has six columns:

  1. ‘WorkspaceIndex’
  2. ‘UserMask’: Holds non-zero values for spectra masked by the default mask, MaskedDetectors and MaskedComponents.
  3. ‘ElasticIntensity’: Holds the value of integrated elastic peaks used for the diagnostics.
  4. ‘IntensityDiagnosed’: Holds non-zero values for spectra diagnosed as ‘bad’ in elastic peak diagnostics.
  5. ‘FlagBkg’: Holds the value of the flat backgrounds used for the diagnostics.
  6. ‘FlatBkgDiagnosed’: Non-zero values in this column indicate that the spectrum did not pass the background diagnostics.

The columns can be plotted to get an overview of the diagnostics.

Additionally, a string listing the masked and diagnosed detectors can be accessed via the OutputReport property.

ILL’s instrument specific defaults

The following settings are used when the AUTO keyword is encountered:

Property IN4 IN5 IN6 Ohters
ElasticPeakDiagnostics Peak Diagnostics ON Peak Diagnostics OFF Peak Diagnostics ON Peak Diagnostics ON
BkgDiagnostics Bkg Diagnostics ON Bkg Diagnostics OFF Bkg Diagnostics ON Bkg Diagnostics ON
BeamStopDiagnostics Beam Stop Diagnostics OFF Beam Stop Diagnostics ON Beam Stop Diagnostics OFF Beam Stop Diagnostics OFF

Usage

Example - Diagnostics on fake IN4 workspace

import numpy
import scipy.stats

# Create a fake IN4 workspace.
# We need an instrument and a template first.
empty_IN4 = LoadEmptyInstrument(InstrumentName='IN4')
nHist = empty_IN4.getNumberHistograms()
# Make TOF bin edges.
xs = numpy.arange(530.0, 2420.0, 4.0)
# Make some Gaussian spectra.
ys = 1000.0 * scipy.stats.norm.pdf(xs[:-1], loc=970, scale=60)
# Repeat data for each histogram.
xs = numpy.tile(xs, nHist)
ys = numpy.tile(ys, nHist)
ws = CreateWorkspace(
    DataX=xs,
    DataY=ys,
    NSpec=nHist,
    UnitX='TOF',
    ParentWorkspace=empty_IN4
)
# Set some histograms to zero to see if the diagnostics can catch them.
ys = ws.dataY(13)
ys *= 0.0
ys = ws.dataY(101)
ys *= 0.0

# Manually correct monitor spectrum number as LoadEmptyInstrument does
# not know about such details.
SetInstrumentParameter(
    Workspace=ws,
    ParameterName='default-incident-monitor-spectrum',
    ParameterType='Number',
    Value=str(1)
)
# Add incident energy information to sample logs.
AddSampleLog(
    Workspace=ws,
    LogName='Ei',
    LogText=str(57),
    LogType='Number',
    LogUnit='meV',
    NumberType='Double'
)
# Elastic channel information is missing in the sample logs.
# It can be given as single valued workspace, as well.
elasticChannelWS = CreateSingleValuedWorkspace(107)

DirectILLCollectData(
    InputWorkspace=ws,
    OutputWorkspace='preprocessed',
    ElasticChannelWorkspace=elasticChannelWS,
    IncidentEnergyCalibration='Energy Calibration OFF', # Normally we would do this for IN4.
    OutputEPPWorkspace='epps' # Needed for the diagnostics.
)

diagnostics = DirectILLDiagnostics(
    InputWorkspace='preprocessed',
    OutputWorkspace='diagnosed',
    EPPWorkspace='epps',
    NoisyBkgLowThreshold=0.01,
    OutputReportWorkspace='diagnostics_report'
)

print(diagnostics.OutputReport)
print('Some small-angle detectors got diagnosed as bad due to detector solid angle corrections.')
report = mtd['diagnostics_report']
I0 = report.cell('ElasticIntensity', 0)
I304 = report.cell('ElasticIntensity', 303)
print('Solid-angle corrected elastic intensity of spectrum 1: {:.8}'.format(I0))
print('vs. corrected intensity of spectrum 304: {:.8}'.format(I304))

Output:

Spectra masked by default mask file:
None
Spectra masked by user:
None
Spectra masked by beam stop diagnostics:
None
Spectra marked as bad by elastic peak diagnostics:
14, 102, 302-305, 314-317, 326-329, 338-341, 350-353, 362-365, 374-377, 386-389
Spectra marked as bad by flat background diagnostics:
14, 102
Some small-angle detectors got diagnosed as bad due to detector solid angle corrections.
Solid-angle corrected elastic intensity of spectrum 1: 555524.7
vs. corrected intensity of spectrum 304: 1795774.9

Categories: Algorithms | ILL\Direct | Inelastic\Reduction | Workflow\Inelastic

Source

Python: DirectILLDiagnostics.py (last modified: 2018-02-26)