Modify LoadDetectorInfo algorithm to understand detector.nxs files

[Alex Buts]

Detector.dat file describes modified detector's positions which replace Instrument description or raw file's detectors positions with the positions, calculated from a calibration procedure.

Unfortunately, it is an ascii file, which has to be read at each reduction step. This takes substantial time.

In addition to that, there are some cases (e.g. maps) when the detector.dat file is not availible and nxs file with specific format is provided instead.

The purpose of this ticket is to modify LoadDetectorInfo algorithm to understand this kind of nxs files too and modify the detector position/time of flight on the basis of such files.

[Alex Buts]

refs #6543 Reading old (and wrong) libisis format

Changeset: f0be7db69cdb9a6a3a377680a3d3c4113559fcf3

[Alex Buts]

refs #6543 Unit tests and some efficiency modifications

Changeset: a0515ce3ab71f661e56616fce51571a0884543a3

[Alex Buts]

refs #6543 One more unit test and make most protected for testing

methods back private

Changeset: 15d04735a89c581b8f540d95340cd08bc50fa6cd

[Alex Buts]

refs #6543 replaced Mantid Nexus with 3-rd party nexus

Changeset: a81a51d54617ab1b4f0feac52a21f5e42f4bd6e3

[Alex Buts]

refs #6543 Modified Nexus header to match new repo structure.

Changeset: 2970b1933b5c4058b8a7c057de1970bb3ccf7448

[Alex Buts]

refs #6543 One more location of old NeXus file

Changeset: 26f4660c0aedefa26d7259b14ced0b0ad2b2873d

[Alex Buts]

refs #6543 One windows warning.

Changeset: 01118f554c4b2d69fb4503bf23f426b757dd9969

[Alex Buts]

To test this ticket one needs to check that workspace modified by loadDetectorInfo(det.dat) is equivalent to a workspace, modified by loadDetectorInfo(det.nxs).

det.nxs file can be obtained from det.dat file by python script, available within qtiGenie in user scripts (the function convertDetDataToNexus(det.dat) ) To use qtiGenie from user scripts repository, one need to add path to it to python scripts path and restart Mantid.

A tester can invoke loadDetectorIndfor with det.dat and det.nxs and check if detector positions and the time scales for both workspaces are equivalent.

This is ideal test which I would like to try by myself but has not done for its complexity. A unit tests doing similar comparison on test workspaces so this kind of testing is in a some way already done. An alternative check would be to convert to energy using det.dat and det.nxs files obtained from det.dat. The script which uses det.dat file to do conversion should be obtained from an excitation group instrument scientist and I am not providing it below. The parameters used by this script are in ISIS svn repository ​https://svn.isis.rl.ac.uk/InstrumentFiles/trunk and the referred data files -- in the data repository. The illiad function referred in the script is also in qtiGenie repository.

The tester may not bother checking libisis.nxs branch, as this is very specific operation, which no general user would be exposed to.

from qtiGenie import *
from PySlice2 import *

inst='MER'
iliad_setup(inst)
ext='.raw'
mapfile='rings_123'
#det_cal_file must be specified if the reduction sends out put to a workpsace
cal_file='det_corr_123.dat'
#cal_file="12669"
#load vanadium file
whitebeamfile="13961"
LoadRaw(Filename=whitebeamfile,OutputWorkspace="wb_wksp",LoadLogFiles="0")
#LoadRaw(Filename=whitebeamfile,OutputWorkspace="wb_wksp",LoadLogFiles="0")
#---------------------------------------------------------------------------------------------------------------
#---------------------------------------------------------------------------------------------------------------

#ei=50
#rebin_params='-70,0.5,48'
ei=60
rebin_params='-45,0.25,43'


runs=[14185]
############## normal reduction####################

#save .spe file
for runfile in runs:
	save_file=inst+str(runfile)+'.spe'
	LoadRaw(Filename=str(runfile),OutputWorkspace="run_wksp",LoadLogFiles="0")
	w1=iliad("wb_wksp","run_wksp",ei,rebin_params,mapfile,det_cal_file=cal_file,norm_method='current')
	SaveSPE('w1',save_file)
print type(w1)

[Alex Buts]

refs #6543 deal with warnings

Changeset: 2ccd3a73b9ea60fcc43e079cc89e8969f5d154dc

[Alex Buts]

refs #6543 Added OMP around main detector modifications loop

Changeset: d107c378118a33cc6e2428e99d8b51c17e19d9de

[Russell Taylor]

Re #6543. Move continue statement outside of catch block.

Works around a defect with the Intel compiler, which can get stuck when it encounters continue statements inside catch blocks within a parallelized loop.

Changeset: 22688a08aade0bf8638317a6e8e35447c7c1245d

[Russell Taylor]

Re #6543. Remove unnecessary guard that fails on the Mac.

And the test that checks it. If the file doesn't exist, it won't be possible to execute the algorithm as the validator on the Filename property will prevent it.

Changeset: c366a6c6fb83129158d0123f29d8a8b59a4162d6

[Alex Buts]

refs #6543 Trying to fix crash on Mac

Changeset: 32c86e74633e34edb7cbd323de8f53fec6049c06

[Alex Buts]

refs #6543 resolving conflicts

Changeset: c70c63eb0b0d73d4fb9ac771bfb02acb8f83286b

[Alex Buts]

refs #6543 disabling OMP for the time being.

Changeset: f07871399662aad089bad42cbc6095e0463d4a8c

[Russell Taylor]

Re #6543. Remove redundant line.

hFile can never be NULL at that point. If the file cannot be opened the constructor will throw and it will be caught in the algorithm base class.

Changeset: 34fdba33c8057ffebab4f76872944122ad3d12d3

[Alex Buts]

refs #6543 Attempt to fix and enable OMP again

Changeset: 3b3db5d1cf6b7ddb302f8d86b04cff5819109c41

[Alex Buts]

refs #6543 technical changes

Changeset: a2a5f686c6cb903de60df081420b2ef4f0c456ca

[Alex Buts]

refs #6543 -- disabled OMP again

to resolve strange problem with memory leak in DataHandling dll.

Changeset: d589aa3b44bc2d0a903df0524ad4d82ecdf9edba

[Alex Buts]

refs #6543 enabled OMP again

Changeset: 525c8eb56c1eed4650b53b889c0a0d87377154f7

[Alex Buts]

refs #6543 Disable OMP

Changeset: c18b61aabdc52027d303b357f16c8ba8771d4b83

[Russell Taylor]

Despite having been marked as fixed over a week ago, this ticket has fresh commits as recently as yesterday. At a minimum, the 'fixed' assertion needs reaffirming if that really is the case.

[Alex Buts]

It had some changes related to omp but apparently most expensive data modifying operation is not thread safe so OMP does not have any sense. OMP disabled for the time being and the ticket is fixed as above.

The tester should compare workspaces, modified by loadDetectorInfo("det.dat") file and loadDetectorInfo("det.nxs") file and check that they are equivalent. There are apparently operations which compare workspace.

the det.nxs file can be created from det.dat file by onvertDetDataToNexus("det.dat") from qtiGenie user script.

[Nick Draper]

changes are on Master

[Nick Draper]

Checked by loading and cloneing an LET workspace LoadDectorInfo from a matching .nxs and .dat file, with move detectors on The CheckworkspacesMatch including the instrument. Passed

[Stuart Campbell]

This ticket has been transferred to github issue ​7389