\(\renewcommand\AA{\unicode{x212B}}\)

ConvertToMD v1¶

../_images/ImageNotFound.png — Enable screenshots using DOCS_SCREENSHOTS in CMake¶

Summary ¶

Create a MDEventWorkspace with selected dimensions, e.g. the reciprocal space of momentums (Qx, Qy, Qz) or momentums modules mod(Q), energy transfer dE if available and any other user specified log values which can be treated as dimensions.

Properties ¶

Name	Direction	Type	Default	Description
InputWorkspace	Input	MatrixWorkspace	Mandatory	An input Matrix Workspace (2DMatrix or Event workspace)
QDimensions	InOut	string	CopyToMD	String, describing available analysis modes, registered with MD Transformation factory.There are 3 modes currently available and described in details on MD Transformation factory page.The modes names are CopyToMD, \|Q\| and Q3D. Allowed values: [‘CopyToMD’, ‘Q3D’, ‘\|Q\|’]
dEAnalysisMode	InOut	string	Direct	You can analyze neutron energy transfer in Direct, Indirect or Elastic mode.The analysis mode has to correspond to experimental set up. Selecting inelastic mode increasesthe number of the target workspace dimensions by one. See MD Transformation factory for further details. Allowed values: [‘Elastic’, ‘Direct’, ‘Indirect’]
Q3DFrames	Input	string	AutoSelect	Selects Q-dimensions of the output workspace in Q3D case. AutoSelect: Choose the target coordinate frame as the function of goniometer and UB matrix values set on the input workspace. Q (lab frame): Wave-vector converted into the lab frame. Q (sample frame): Wave-vector converted into the frame of the sample (taking out the goniometer rotation). HKL: Use the sample’s UB matrix to convert Wave-vector to crystal’s HKL indices.See MD Transformation factory (Q3D) for more details about this. Allowed values: [‘Q_lab’, ‘Q_sample’, ‘HKL’, ‘AutoSelect’]
QConversionScales	Input	string	Q in A^-1	This property to normalize three momentums obtained in Q3D mode. See MD Transformation factory for description and available scaling modes. The value can be modified depending on the target coordinate system, defined by the property OutputDimensions. Allowed values: [‘Q in A^-1’, ‘Q in lattice units’, ‘Orthogonal HKL’, ‘HKL’]
OtherDimensions	Input	str list		List(comma separated) of additional to Q and DeltaE variables which form additional (orthogonal) to Q dimensions in the target workspace (e.g. Temperature or Magnetic field). These variables had to be logged during experiment and the names of these variables have to coincide with the log names for the records of these variables in the source workspace.
PreprocDetectorsWS	Input	string	PreprocessedDetectorsWS	The name of the table workspace where the part of the detectors transformation into reciprocal space, calculated by PreprocessDetectorsToMD v1 algorithm is stored. If the workspace is not found in analysis data service, PreprocessDetectorsToMD v1 used to calculate it. If found, the algorithm uses existing workspace. The field is useful if one expects to analyze number of different experiments obtained on the same instrument… warning:: Dangerous if one uses number of workspaces with modified derived instrument one after another. In this case this property has to be set to “-” sting (minus without quotes) or empty (possible from script only) to force the workspace recalculation each time the algorithm is invoked.
UpdateMasks	Input	boolean	False	if PreprocessDetectorWS is used to build the workspace with preprocessed detectors at first algorithm call,and the input workspaces instruments are different by just different masked detectors, setting this option to true forces PreprocessDetectorsToMD v1 update only the detectors masks for all subsequent calls to this algorithm… warning:: This is temporary solution necessary until Mantid masks spectra by 0 rather then by NaN.
LorentzCorrection	Input	boolean	False	Correct the weights of events or signals and errors transformed into reciprocal space by multiplying them by the Lorentz multiplier: \(sin(\theta)^2/\lambda^4\). Currently works in Q3D Elastic case only and is ignored in any other case.
IgnoreZeroSignals	Input	boolean	False	Enabling this property forces the algorithm to ignore bins with zero signal for an input matrix workspace. Input event workspaces are not affected. This violates the data normalization but may substantially accelerate calculations in situations when the normalization is not important (e.g. peak finding).
Uproj	Input	dbl list		Defines the first projection vector of the target Q coordinate system in Q3D mode - Default (1,0,0)
Vproj	Input	dbl list		Defines the second projection vector of the target Q coordinate system in Q3D mode - Default (0,1,0).
Wproj	Input	dbl list		Defines the third projection vector of the target Q coordinate system in Q3D mode. - Default (0,0,1)
AbsMinQ	Input	number	0	Do not add events to MD workspace that are closer to the origin in QSample radius than this value. Needed for 3Dviews to remove noise.
OutputWorkspace	Output	MDEventWorkspace	Mandatory	Name of the output MDEventWorkspace.
OverwriteExisting	Input	boolean	True	By default (“1”), existing Output Workspace will be replaced. Select false (“0”) if you want to add new events to the workspace, which already exist. Choosing “0” can be very inefficient for file-based workspaces
MinValues	Input	dbl list		It has to be N comma separated values, where N is the number of dimensions of the target workspace. Values smaller then specified here will not be added to workspace. Number N is defined by properties 4,6 and 7 and described on MD Transformation factory page. See also ConvertToMDMinMaxLocal v1
MaxValues	Input	dbl list		A list of the same size and the same units as MinValues list. Values higher or equal to the specified by this list will be ignored
SplitInto	Input	int list	5	A comma separated list of into how many sub-grid elements each dimension should split; or just one to split into the same number for all dimensions. Default 5.
SplitThreshold	Input	number	1000	How many events in a box before it should be split. Default 1000.
MaxRecursionDepth	Input	number	20	How many levels of box splitting recursion are allowed. The smallest box will have each side length \(l = (extents) / (SplitInto^{MaxRecursionDepth}).\) Default 20.
MinRecursionDepth	Input	number	1	Optional. If specified, then all the boxes will be split to this minimum recursion depth. 0 = no splitting, 1 = one level of splitting, etc. Be careful using this since it can quickly create a huge number of \(boxes = SplitInto^{MinRercursionDepth \times NumDimensions}\). But setting this property equal to MaxRecursionDepth is necessary if one wants to generate multiple file based workspaces in order to merge them later.
TopLevelSplitting	Input	boolean	False	This option causes a split of the top level, i.e. level0, of 50 for the first four dimensions.
Filename	Input	string		The name of the Nexus file to write, as a full or relative path. Only used if FileBackEnd is true. Allowed extensions: [‘.nxs’]
FileBackEnd	Input	boolean	False	If true, Filename must also be specified. The algorithm will create the specified file in addition to an output workspace. The workspace will load data from the file on demand in order to reduce memory use.
ConverterType	Input	string	Default	[Default, Indexed], indexed is the experimental type that can speedup the conversion processfor the big files using the indexing. Allowed values: [‘Default’, ‘Indexed’]

Description ¶

The algorithm is used transform existing Event or Matrix workspace into Multidimensional workspace.

If the target workspace does not exist, the algorithm creates MDEventWorkspace with selected dimensions, e.g. the reciprocal space of momentums (Qx, Qy, Qz) or momentums modules |Q|, energy transfer dE if available and any other user specified log values which can be treated as dimensions. If the target workspace do exist, the MD Events are added to this workspace.

Using the FileBackEnd and Filename properties the algorithm can produce a file-backed workspace. Note that this will significantly increase the execution time of the algorithm.

Used Subalgorithms ¶

The algorithm uses Unit Factory and existing unit conversion procedures from the input Workspace units to the units , necessary for transformation into correspondent MD Event workspace. It also uses PreprocessDetectorsToMD v1 algorithm to help with transformation to reciprocal space.

If min, max or both lists of values (properties 12 and 13) for the algorithm are not specified, ConvertToMDMinMaxLocal v1 is used to estimate missing min-max values. This algorithm is also used to calculate min-max values if specified min-max values are deemed incorrect (e.g. less values then dimensions or some min values are bigger then max values)

\(l = (extents) / (SplitInto^{MaxRecursionDepth}).\)

Notes ¶

For elastic analysis (\(dEAnalysisMode=Elastic\)) the target unit is momentum \(k\).
For no analysis (CopyToMD) mode, the units remain the one, previously defined along the workspace’s axes.
When units of input Matrix 2D workspace (Histogram workspace) are not Momentums for Elastic or EnergyTransfer for inelastic mode, the algorithm uses internal unit conversion of input X-values based on central average of a bin ranges. Namely, value \(X_c = 0.5*(X_i+X_{i+1})\) is calculated and converted to Momentum or EnergyTransfer correspondingly. This can give slightly different result from the case, when input workspace has been converted into correspondent units before converting to MDEvents.
Confusing message “Error in execution of algorithm ConvertToMD: emode must be equal to 1 or 2 for energy transfer calculation” is generated when one tries to process the results of inelastic scattering experiment in elastic mode. This message is generated by units conversion routine, which finds out that one of the workspace axis is in unit of DeltaE. These units can not be directly converted into momentum or energy, necessary for elastic mode. Select Direct or Indirect mode and integrate over whole energy transfer range to obtain MD workspace, which would correspond to an Elastic mode.
A good guess on the limits can be obtained from the ConvertToMDMinMaxLocal v1 algorithm.

Indexed mode ¶

Setting the ConverterType parameter to Indexed uses an alternative intermediate data format within ConvertToMD which may give significant performance boosts in certain cases.

The performance benefit of this method is dependant on the number of events in the input dataset. Once you have data files containing more than 100 million events and have at least 8 cores this method becomes worth enabling. For large files (>500 million events) performance scales well with the number of available CPU cores (i.e. using 32 cores will be notably faster than 8 cores).

Use of this method comes with the following restrictions:

SplitInto should be the power of two (i.e. 2, 4, 8, 16, etc.)
FileBackEnd and TopLevelSplitting are not applicable and should be disabled
Indexing adds a small numerical error to the event coordinates, the magnitude of this error is listed in the log (Error with using Morton indexes is)

How to write custom ConvertToMD plugin ¶

This information intended for developers who have at least basic knowledge of C++ and needs to write its own plugin using Writing a Custom ConvertToMD Transformation.

Usage examples ¶

Note

To run these usage examples please first download the usage data, and add these to your path. In Mantid this is done using Manage User Directories.

Example - Convert re-binned MARI 2D workspace to 3D MD workspace for further analysis/merging with data at different temperatures :

# Load Operation (disabled in test code)
# Load(Filename='MAR11001.nxspe',OutputWorkspace='MAR11001')
# Simulates Load of the workspace above #################
redWS = CreateSimulationWorkspace(Instrument='MAR',BinParams=[-10,1,10],UnitX='DeltaE',OutputWorkspace='MAR11001')
AddSampleLog(redWS,LogName='Ei',LogText='12.',LogType='Number');
# Do fine rebinning, which accounts for poligon intersections
SofQW3(InputWorkspace='MAR11001',OutputWorkspace='MAR11001Qe2',QAxisBinning='0,0.1,7',EMode='Direct')
AddSampleLog(Workspace='MAR11001Qe2',LogName='T',LogText='100.0',LogType='Number Series')
# copy to new MD workspace
ws=ConvertToMD(InputWorkspace='MAR11001Qe2',OutputWorkspace='MD3',QDimensions='CopyToMD',OtherDimensions='T',\
MinValues='-10,0,0',MaxValues='10,6,500',SplitInto='50,50,5')


#Output **MD3** workspace can be viewed in slice-viewer as 3D workspace with T-axis having single value.
#Visualize 3D data using slice viewer:
#plotSlice(ws)

# Look at sample results:
# A way to look at these results as a text:
print("Resulting MD workspace has {0} events and {1} dimensions".format(ws.getNEvents(),ws.getNumDims()))
print("--------------------------------------------")

Output:

Resulting MD workspace has 802 events and 3 dimensions
--------------------------------------------

Example - Convert Set of Event Workspaces (Horace scan) to 4D MD workspace, direct mode:

Meaningful results can be obtained on the basis of CNCS_7860_event.nxs file, available in Mantid test folder. The script below simulates workspace loading but would produce meaningfill result if real experimental data obtained in an experiment and stored in nxspe files are provided to it.

import os
# set up target ws name and remove target workspace with the same name which can occasionally exist.
# list of MD files (workspaces) to combine into target MD workspace
MD_FilesList='';

# define convetr to MD parameters
pars = dict();
pars['InputWorkspace']=''
pars['QDimensions']='Q3D'
pars['dEAnalysisMode']='Direct'
pars['Q3DFrames']='HKL'
pars['QConversionScales']='HKL'
pars['PreprocDetectorsWS']='preprDetMantid'
pars['MinValues']='-3,-3,-3.,-50.0'
pars['MaxValues']='3.,3.,3.,50.0'
pars['SplitInto']=50
pars['MaxRecursionDepth']=1
pars['MinRecursionDepth']=1
pars['OverwriteExisting']=1  # Change this to false, if the files should/can be added in memory
# test script combines all contributed files in memory
pars['OverwriteExisting']=0  # Change this to false, if the files should/can be added in memory
#
#---> Start loop over contributing files
for n in range(0,5,1):
     source_file = 'MER19566_22.0meV_one2one125.nxspe'; # redefine source files list as function of loop number
     target  = 'MDMAP_T1'+str(n)+'.nxs';
     # check if the file already been converted to MD and is there
     if not(os.path.exists(target )):
         print('Converting  {}'.format(source_file))
         #current_ws=LoadNXSPE(Filename=source)
         #### For the sample script, simulate load operation above
         current_ws = CreateSimulationWorkspace(Instrument='MAR',BinParams=[-3,1,3],UnitX='DeltaE',OutputWorkspace=source_file)
         AddSampleLog(Workspace=current_ws,LogName='Ei',LogText='3.0',LogType='Number')

         #### Add iformation which is not stored in the nxspe file
         # Add UB matrix (lattice and the beam direction)
         SetUB(Workspace=current_ws,a='1.4165',b='1.4165',c='1.4165',u='1,0,0',v='0,1,0')
         # Add crystal rotation (assume rotation abgle Psi=5*n where n is file number. Define list of angles if this is not correct
         AddSampleLog(Workspace=current_ws,LogName='Psi',LogText=str(5*n)+'.',LogType='Number')  # --correct Psi value may be already in nxspe file. This operation is then unnecessary
         # set crystal rotation
         SetGoniometer(Workspace=current_ws,Axis0='Psi,0,1,0,1')

         # Convert to MD
         pars['InputWorkspace']=current_ws;
         md_ws=ConvertToMD(**pars)

         # save MD for further usage -- disabled in test script
         #SaveMD(md_ws,Filename=target);
         #DeleteWorkspace(md_ws);  # delete intermediate workspace to save memory
         DeleteWorkspace(current_ws);

     # add the file name of the file to combine
     if (len(MD_FilesList) == 0):
         MD_FilesList = target;
     else:
         MD_FilesList=MD_FilesList+','+target;
#---> End loop

# merge md files into file-based MD workspace
#md_ws = MergeMDFiles(MD_FilesList,OutputFilename='TestSQW_1.nxs',Parallel='0');
# plot results using sliceviewer
#plotSlice(md_ws)
# produce some test output
print("Resulting MD workspace contains {0} events and {1} dimensions".format(md_ws.getNEvents(),md_ws.getNumDims()))
print("--------------------------------------------")

Output:

Converting  MER19566_22.0meV_one2one125.nxspe
Converting  MER19566_22.0meV_one2one125.nxspe
Converting  MER19566_22.0meV_one2one125.nxspe
Converting  MER19566_22.0meV_one2one125.nxspe
Converting  MER19566_22.0meV_one2one125.nxspe
Resulting MD workspace contains 27540 events and 4 dimensions
--------------------------------------------

Example - Convert set of inelastic results obtained in Powder mode (direct) as function of temperature to a 3D workspace:

This example produces 3-dimensional dataset, with a temperature axis.

# set up target ws name and remove target workspace with the same name which can occasionally exist.
RezWS = 'WS_3D'
try:
     DeleteWorkspace(RezWS)
except ValueError:
     print("Target ws  {}  not found in analysis data service\n".format(RezWS))

# define convert to MD parameters
pars = dict();
pars['InputWorkspace']=''
pars['QDimensions']='|Q|'
pars['dEAnalysisMode']='Direct'
pars['OtherDimensions']='T'  # make temperature log to be a dimension
pars['PreprocDetectorsWS']='preprDetMantid'
pars['MinValues']='0,-10,0'
pars['MaxValues']='12,10,10'
pars['SplitInto']='100,100,12'
pars['OverwriteExisting']=0  # contributed MD worskpaces are added in memory

# let's assume this is the temperature range obtained in experiments and
# each data file is obtained for particular temperature.
T = [1.0,2.0,3.0,3.5,4.0,5.0,6.0,7.0,8.0,9.0,9.5,10.0]
for i in range(0,len(T),1):
     # source = sorurce_file_name[i];
     #current_ws=LoadNXSPE(Filename=source)
     # EMULATE LOAD OF DIFFERENT results obtained for different temperatures. ------>
     current_ws = CreateSimulationWorkspace(Instrument='MAR',BinParams=[-3,0.1,3],UnitX='DeltaE')
     AddSampleLog(Workspace=current_ws,LogName='Ei',LogText='3.0',LogType='Number')
     # if the file does not have temperature log, add it here.
     AddSampleLog(Workspace=current_ws,LogName='T',LogText=str(T[i]),LogType='Number Series')
     # simulate changes in scattering with temperature
     current_ws = current_ws*T[i];
     # END EMULATION ---------------------------------------------------------------------

     pars['InputWorkspace']=current_ws;
     md_ws=ConvertToMD(**pars)
     # delete source workspace from memory;
     DeleteWorkspace(current_ws)
 # end loop

# produce some test output
print("Resulting MD workspace contains {0} events and {1} dimensions".format(md_ws.getNEvents(),md_ws.getNumDims()))

Output:

Target ws  WS_3D  not found in analysis data service

Resulting MD workspace contains 605880 events and 3 dimensions

Example - Convert to Q-space, indexed version:

#.. testcode:: ExConvertToMD|Q|

# load test workspace
TOPAZ_3132_event = Load(Filename=r'TOPAZ_3132_event.nxs', ConverterType='Indexed')

# build peak workspace necessary for IntegrateEllipsoids algorithm to work
TOPAZ_3132_md = ConvertToMD(InputWorkspace=TOPAZ_3132_event,QDimensions='Q3D',dEAnalysisMode='Elastic',Q3DFrames='Q_sample',LorentzCorrection='1',\
MinValues='-25,-25,-25',MaxValues='25,25,25',SplitInto='4',SplitThreshold='50',MaxRecursionDepth='13',MinRecursionDepth='7')

# produce some test output
print("Resulting MD workspace contains {0} events and {1} dimensions".format(TOPAZ_3132_md.getNEvents(),TOPAZ_3132_md.getNumDims()))

Output:

Resulting MD workspace contains 15329354 events and 3 dimensions

Categories: AlgorithmIndex | MDAlgorithms\Creation

Source ¶

C++ header: ConvertToMD.h

C++ source: ConvertToMD.cpp