Single Crystal Profile fit in Mantid

[Vickie Lynch]

Attached are the steps for doing the ID profile fitting. We discussed implementing the first 2 steps in Mantid to produce an "expname.profiles" file in the Mantid Workspace. Attached is an example of a profiles file (sapphire.profiles). It is the same as a peaks or integrate file, but with the counts for 100 steps through each profile.

All the Python scripts and config file examples can be found in /SNS/TOPAZ/shared/PythonPrograms.

To summarize, step 1 finds and indexes peaks, outputs UB matrices and outputs binary files with the events as a Qx,Qy,Qz vector.

Step 2 uses the UB matrices and the Qx,Qy,Qz events to create the unfitted profiles for each hkl peak.

[Nick Draper]

Moved to r2.6 at the end of r2.5

[Ruth Mikkelson]

Refs #6940 Fixed logic error,Calculated more lattice Errors

SelectCellOfType and SelectCellWithForm will now calculate(incorrectly?) lattice parameter errors assuming least squares optimization(which may not be correct)

Changeset: 4f8d0f5aed74535b9d393d9bfa8c4dec0982e173

[Ruth Mikkelson]

Refs #6940 Fixed logic error,Calculated more lattice Errors

SelectCellOfType and SelectCellWithForm will now calculate(incorrectly?) lattice parameter errors assuming least squares optimization(which may not be correct)

Changeset: 4f8d0f5aed74535b9d393d9bfa8c4dec0982e173

[Ruth Mikkelson]

Refs #6940 Fixed logic error,Calculated more lattice Errors

SelectCellOfType and SelectCellWithForm will now calculate(incorrectly?) lattice parameter errors assuming least squares optimization(which may not be correct)

Changeset: 4f8d0f5aed74535b9d393d9bfa8c4dec0982e173

[Vickie Lynch]

Refs #6940 output vectors instead of a file

Changeset: 8a64104a78f25cfc084d72d4996278cd807793c8

[Vickie Lynch]

Merge branch 'feature/6940_profile_fit' into develop

[Vickie Lynch]

Refs #6940 expose PeakInfo to python

Changeset: 7d6acc0a8e6b85f5bb4996b8ea129ca7a15fccb7

[Vickie Lynch]

Refs #6940 changed too many

Changeset: 523917c9c6c559ac5b07abe73cec4aaee5a6f110

[Vickie Lynch]

Refs #6940 merge conflict

Changeset: 58ff25f5cecfa4af54ab4350d1e08c7bcb863a1a

[Vickie Lynch]

Refs #6940 cylinder integration added to IntegratePeaksMD

Changeset: 363907280ecd9ddfc58f6214d06f463ef5dec850

[Vickie Lynch]

Refs #6940 fix background for cylinder integration

Changeset: ad0be0d72df96d0a10614376ca951a40bd2a09bd

[Vickie Lynch]

Refs #6940 separate cylinder and sphere to help performance tests

Changeset: 3abd2f19797ee3d7b9f09b90494c5996be4fce97

[Vickie Lynch]

Refs #6940 fix windows build

Changeset: 954f762219971a1127218cf5dbd81bf141237f92

[Vickie Lynch]

Refs #6940 fix windows build

Changeset: a27e92768acdf4217913fabdf42c3c195fdaf816

[Vickie Lynch]

Refs #6940 write profiles along cylinder

Changeset: bbd814e54f6d5ae20af2db273669cebf326e3f21

[Vickie Lynch]

Refs #6940 sum events for profile

Changeset: e94ec2ac1188cebd4dffcac7cde975899bbb9afa

[Vickie Lynch]

Refs #6940 try to remove osx warning

Changeset: f3e7d6c07980700f4d496b2c6e5d6a87be02a464

[Vickie Lynch]

Refs #6940 better number of point for profile

Changeset: dd9d46e2c3e7f2acfdb49ce53d36319b77d21701

[Vickie Lynch]

Refs #6940 add fitting options

Changeset: ee9732df8d66e6cebd29cbf082ce4e8b365c29cd

[Vickie Lynch]

Refs #6940 only output Qvectors in memory if no file requested

Changeset: 13d9353f367f00edeacbefe3c613b7523d09d9f5

[Vickie Lynch]

Refs #6940 write chisq; remove warning

Changeset: 637673b2b875ab8382be7d08bd3c69646f693580

[Vickie Lynch]

Refs #6940 add FixResolution from ticket 7369

Changeset: bdbc797853af41c1d6e539dcfc7c31f230e5caf2

[Vickie Lynch]

Refs #6940 add test for cylindrical integration

Changeset: 22c8d37a37e2bd581711a847626248b460615640

[Vickie Lynch]

I will create another ticket to improve the fits with convolutions and save the fits for plotting, but I would like to get this into the nightly build for TOPAZ instrument scientists to test on July 8 at a meeting setup by Art Schultz. To test:

Load(Filename='TOPAZ_6674_event.nxs',OutputWorkspace='TOPAZ_6674_event',FilterByTofMin='2000',FilterByTofMax='16500',LoadMonitors='1',MonitorsAsEvents='1')
ConvertToDiffractionMDWorkspace(InputWorkspace='TOPAZ_6674_event',OutputWorkspace='TOPAZ_6674_MD',OutputDimensions='Q (sample frame)',SplitThreshold='50',MaxRecursionDepth='11')
FindPeaksMD(InputWorkspace='TOPAZ_6674_MD',PeakDistanceThreshold='0.47099999999999997',MaxPeaks='100',OutputWorkspace='peaks_ws')
CentroidPeaksMD(InputWorkspace='TOPAZ_6674_MD',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws')
FindUBUsingFFT(PeaksWorkspace='peaks_ws',MinD='4',MaxD='12',Tolerance='0.12')
IndexPeaks(PeaksWorkspace='peaks_ws',Tolerance='0.12')
SelectCellOfType(PeaksWorkspace='peaks_ws',CellType='Rhombohedral',Centering='R')
IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.10000000000000001',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder',
Cylinder='1',CylinderLength='0.40000000000000002',PercentBackground='20',ProfileFunction='NoFit')

IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.10000000000000001',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder_gauss',
Cylinder='1',CylinderLength='0.40000000000000002',PercentBackground='20',ProfileFunction='Gaussian')

IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.10000000000000001',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder_b2b',
Cylinder='1',CylinderLength='0.40000000000000002',PercentBackground='20',ProfileFunction='ConvolutionBackToBackGaussian')

IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.10000000000000001',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder_exp',
Cylinder='1',CylinderLength='0.40000000000000002',PercentBackground='20',ProfileFunction='ConvolutionExpGaussian')

[Russell Taylor]

Newer gcc's are producing warnings in MD(Grid)Box relating to the fact that there is a read past the end of a vector if instantiated with nd=1. Less critically, there's also one about testing that an unsigned integer isn't negative. See here: ​http://download.mantidproject.org/jenkins/view/Develop%20Clean%20Builds%20and%20Tests/job/is_clean_ubuntu-12.04_develop/warnings19

These should be addressed before the ticket is merged. Hopefully you can use a development build for the meeting.

[Vickie Lynch]

Refs #6940 fix warnings; add workspaces of each fit

Changeset: b497f31abf0e70382eb09ff425f2c8df7fdc36b0

[Vickie Lynch]

Refs #6940 fix warnings

Changeset: 5b1a8c6734699dce3154a0785b7ec4ac64155990

[Vickie Lynch]

Refs #6940 improve initial guess

Changeset: f1fe567edc6e38f6f72cfcaf2b5d900c4192fa25

[Vickie Lynch]

Refs #6940 workspaces of fit and difference; labeled by hkl

Changeset: b5d8818fb16839e3f41494dad02f9d601545e536

[Vickie Lynch]

Refs #6940 improve initial values for fits

Changeset: 5f3f52645465dc280b051772c1051c55ae9a30a6

[Vickie Lynch]

Refs #6940 ConvolutionExpGaussian working

Changeset: 8bdc591dfe9e9164853000dae71fe378ff240c95

[Vickie Lynch]

Refs #6940 comment failing test

Changeset: d0c190bab82504a556e882186fd210e479c24716

[Vickie Lynch]

Refs #6940 comment more failing tests

Changeset: a6bd821fec2bb994da17f42dde3b077ccb2a795f

[Vickie Lynch]

Would like to get ticket tested while Art Shultz is at SNS this week. To test:

Load(Filename='TOPAZ_6674_event.nxs',OutputWorkspace='TOPAZ_6674_event',FilterByTofMin='2000',FilterByTofMax='16500',LoadMonitors='1',MonitorsAsEvents='1')
ConvertToDiffractionMDWorkspace(InputWorkspace='TOPAZ_6674_event',OutputWorkspace='TOPAZ_6674_MD',OutputDimensions='Q (sample frame)',SplitThreshold='50',MaxRecursionDepth='11')
FindPeaksMD(InputWorkspace='TOPAZ_6674_MD',PeakDistanceThreshold='0.471',MaxPeaks='100',OutputWorkspace='peaks_ws')
CentroidPeaksMD(InputWorkspace='TOPAZ_6674_MD',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws')
FindUBUsingFFT(PeaksWorkspace='peaks_ws',MinD='4',MaxD='12',Tolerance='0.12')
IndexPeaks(PeaksWorkspace='peaks_ws',Tolerance='0.12')
SelectCellOfType(PeaksWorkspace='peaks_ws',CellType='Rhombohedral',Centering='R')
IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.1',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_sphere')
IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.1',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder',
Cylinder='1',CylinderLength='0.4',PercentBackground='20',ProfileFunction='NoFit')

IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.1',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder_gauss',
Cylinder='1',CylinderLength='0.4',PercentBackground='20',ProfileFunction='Gaussian')

for wi in xrange(mtd["ProfilesData"].getNumberHistograms()):
	graph_spec = plotSpectrum (("ProfilesData","ProfilesFit","ProfilesFitDiff"), wi)
	graph_spec.export("ProfileGauss_"+mtd["ProfilesData"].getAxis(1).label(wi)+".png")
	#Remove next 2 lines to keep plots in window
	graph_spec.confirmClose(False)
	graph_spec.close()

IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.1',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder_b2b',
Cylinder='1',CylinderLength='0.4',PercentBackground='20',ProfileFunction='ConvolutionBackToBackGaussian')

for wi in xrange(mtd["ProfilesData"].getNumberHistograms()):
	graph_spec = plotSpectrum (("ProfilesData","ProfilesFit","ProfilesFitDiff"), wi)
	graph_spec.export("ProfileB2B_"+mtd["ProfilesData"].getAxis(1).label(wi)+".png")
	#Remove next 2 lines to keep plots in window
	graph_spec.confirmClose(False)
	graph_spec.close()

IntegratePeaksMD(InputWorkspace='TOPAZ_6674_MD',CoordinatesToUse='Q (sample frame)',PeakRadius='0.1',PeaksWorkspace='peaks_ws',OutputWorkspace='peaks_ws_cylinder_exp',
Cylinder='1',CylinderLength='0.4',PercentBackground='20',ProfileFunction='ConvolutionExpGaussian')

for wi in xrange(mtd["ProfilesData"].getNumberHistograms()):
	graph_spec = plotSpectrum (("ProfilesData","ProfilesFit","ProfilesFitDiff"), wi)
	graph_spec.export("ProfileExp_"+mtd["ProfilesData"].getAxis(1).label(wi)+".png")
	#Remove next 2 lines to keep plots in window
	graph_spec.confirmClose(False)
	graph_spec.close()

[Andrei Savici]

For some reason it puts the png files into the bin directory of Mantid. Not sure this will work for users on the beamline

[Stuart Campbell]

This ticket has been transferred to github issue ​7786