NaNs are lost somewhere on the way from MD Workspace to Matix workspace

[Alex Buts]

If one runs script below and looks at the result in the slice viewer the area which are not covered by the image (and dead pixels) correctly marked as NaN-s(source file is not important -- any nxspe would produce similar result).

When one wants to produce nice image from the observed slice, he takes binned image and converts it into matrix workspace for plotting. At this stage matrix workspace contains zeros in place of NaNs which mixes legitimate zero signal areas with areas with NaN-s

This should be fixed and NaN-s should be retained.

LoadNXSPE(Filename=r'Z:\Helen\MER23357_75meV_on2one.nxspe', OutputWorkspace='cur_ws')
AddSampleLog(Workspace='cur_ws', LogName='Psi', LogText='0.', LogType='Number')
SetUB(Workspace='cur_ws', a=5.9470000000000001, b=10.300000000000001, c=6.7220000000000004, beta=107.16, u='0,0,1')
SetGoniometer(Workspace='cur_ws', Axis0='Psi,0,1,0,1')
ConvertToMD(InputWorkspace='cur_ws', QDimensions='Q3D', Q3DFrames='HKL', QConversionScales='HKL', PreprocDetectorsWS='preprDetMER', OutputWorkspace='target_MDws', OverwriteExisting=False, SplitInto='50,50,50,60', MaxRecursionDepth=1)
BinMD(InputWorkspace='target_MDws', AlignedDim0='[H,0,0],-3,3,150', AlignedDim1='DeltaE,0,50,100', AlignedDim2='[0,K,0],0.4,0.6,1', OutputExtents='-3,3,0,50,0.4,0.6', OutputBins='150,100,1', OutputWorkspace='hslice_0p5')

[Nick Draper]

Roman can you try to track this one down, speak to Alex and Owen to get a start

[Alex Buts]

To simplify the ticket verification I've placed the source file used in the example above at ​ftp://ftp.nd.rl.ac.uk/scratch/abuts/Ticket_10336/MER23357_75meV_on2one.nxspe

[Alex Buts]

Fixing this ticket should involve modifying VATES Validation, http://www.mantidproject.org/Vates_Validation#Slices_comparison, Issue 2 (NaN-s on the plot) and http://www.mantidproject.org/Vates_Validation#ConvertToMD_and_BinMD.26SliceMD_for_dataset_with_non-orthogonal_lattice_against_gen_sqw_and_cut_in_Horace -- text information.

[Nick Draper]

Moved to the backlog at the code freeze of R3.3

[Stuart Campbell]

This ticket has been transferred to github issue ​11178