create an algorithm to transform a processed workspace into MDWorkspace for inelastic instruments

[Alex Buts]

In [14445]:

refs #3675 an template for the algorithm, created from transform to diffraction WS algorithm.

[Alex Buts]

In [14467]:

refs #3675 some intermediate step

[Alex Buts]

In [14468]:

refs #3675 unix error fixed

[Alex Buts]

In [14469]:

refs #3675 failing under Unux test commented until clarified

[Russell Taylor]

In [14481]:

Disable test that appears to be hanging on 32 bit windows for some reason. Re #3675.

[Russell Taylor]

In [14483]:

Not to worry - it was something else. Re #3675.

[Alex Buts]

In [14505]:

refs #3675 some intermediate step

[Alex Buts]

In [14506]:

refs #3675 failing test fixed

[Alex Buts]

In [14554]:

refs #3675 transformation algorithm starts to do something useful.

[Alex Buts]

In [14558]:

refs #3675 fixing warnings (temporary)

[Alex Buts]

In [14586]:

refs #3675 working 1thread algorithm but issues with k-range have to be addressed

[Alex Buts]

In [14609]:

refs #3675 k-dE limits are introduced as input parameters for the algorithm.

[Alex Buts]

In [14610]:

refs #3675 2 warnings should go

[Alex Buts]

It looks like MDEvent workspace has its default internal coordinate system. If this is right, the algorithm is complete, as the algorithm transforms the data into this coordinate system.

I can not currently be sure how user requests for an arbitrary coordinate system would be satisfied.

Also additional checks are needed to be sure, what coordinate system is used for MDEvent workspace.

[Andrei Savici]

I created a matrix workspace using the following script, where the test.mat file contains a UB matrix one on the diagonal, 0 off-diagonal

Load(Filename='~/Mantid/Test/AutoTestData/CNCS_7860_event.nxs',OutputWorkspace='CNCS_7860_event')
ConvertUnits(InputWorkspace='CNCS_7860_event',OutputWorkspace='CNCS_7860_event',Target='DeltaE',EMode='Direct',EFixed='3')
AddSampleLog(Workspace='CNCS_7860_event',LogName='phi',LogText='0',LogType='Number Series')
AddSampleLog(Workspace='CNCS_7860_event',LogName='chi',LogText='0',LogType='Number Series')
AddSampleLog(Workspace='CNCS_7860_event',LogName='omega',LogText='0',LogType='Number Series')
LoadIsawUB(InputWorkspace='CNCS_7860_event',Filename='~/test.mat')
Rebin(InputWorkspace='CNCS_7860_event',OutputWorkspace='s1',Params='-2.9,0.05,2.9',PreserveEvents='0')

Memory used for s1 is about 140MB

Then run ConvertToQ3DdE with the following parameters: InputWorkspace: s1 OutputWorkspace: s2 EnergyInput: 3 UsedPreprocessedDetectors: True MinQdE_values: -2,-2,-2,-3 MinQdE_values: 2,2,2,3

The output s2 is an MDEventWorkspace, with 625 MDBoxes. The boxes take 141 kB. There are no events, and the total size is 141 kB.

Something is obviously wrong

[Alex Buts]

In [14794]:

refs #3675 Set of algorithms and methods modified to specify the energy axis properly.

[Alex Buts]

In [14797]:

refs #3675 temporary modification to revert energy limits to workspace rather then Axis
dirty hack to improve algorithm performance (but works very well)

[Russell Taylor]

In [14798]:

Revert majority of changes in [14794]. Bug-fix coming in #3769 which means that this is not needed. Re #3675.

[Alex Buts]

In [14799]:

refs #3675 fixing failing test

[Alex Buts]

In [14808]:

refs #3675 The problem identified by tester fixed.
Temporary reverted to obtaining energy from x-spectra, until rebin is fixed.

[Martyn Gigg]

Can't test this with the data we have, passing back to the pool.

[Stuart Campbell]

I re-ran Andrei's scriptlet above, but just used the TOPAZ_3007.mat file from the AutoTestData directory.

Michael plotted it using the VATES GUI and it shows something as a function of Et. So I'd say this is good enough until we push it further.

[Stuart Campbell]

This ticket has been transferred to github issue ​4522