Extend "TRANS" Command To Allow "Beam Stop Out" Method

[Peter Parker]

The current functionality of the SANS User File command TRANS is outlined here.

Proposed changes to the syntax of this command:

• Add "MAIN" and "REAR" options to the /TRANSPEC qualifier.
• Add /RADIUS=r qualifier, where r is a decimal value measured in mm.
• Add /MASK=filepath qualifier. This file would be a Mantid Mask File that specifies a particular region of interest.

This should work either with a single line (e.g. TRANS/TRANSPEC=MAIN/RADIUS=3/MASK=mask_file.xml) or via multiple calls to TRANS.

[Nick Draper]

As part of the beam stop out work is this really for the backlog?

[Peter Parker]

Probably not. Moved to 3.2, at least for now. We'll be meeting with the SANS guys again next week to discuss more short term-goals -- this ticket may be on the list.

[Peter Parker]

This ticket was brought up in the SANS update meeting last week and there was general agreement that it should be in for the release. Could you set this to assigned, please?

On a related note, the other Bilby ticket about background corrections (#9350) was seen as less important/useful right now.

[Peter Parker]

As per emails and conversations with SANS scientists, moving this to next release. It is clear that this is number 1 priority for next cycle.

[Peter Parker]

Essentially this should be just a more generic version of /TRANSPEC (which calls CalculateTransmission), i.e. instead of dividing one monitor by another we should allow the division of a summed region of interest by a monitor.

Richard has pointed out that the spectra making up the ROI have to be converted to wavelength before summing. Also, he would like a copy of the summed ws (before it gets fed to CalculateTransmission) kept around in the workspace list.

[Peter Parker]

We need to check that pixel calibration has been applied when we try to work out which detectors are within a certain radius, etc. I'm not sure where that step is in the reduction.

In any event, Richard has said this is not a priority, so we can split it out into a separate ticket if necessary.

[Peter Parker]

Refs #9349 - Move monitor det ID check to validator.

Changeset: 5ecd48a66ca1932574cd3c78e487ff887108a04a

[Peter Parker]

Refs #9349 - Make changes to CalculateTransmission algorithm ...

... to allow for a region of interest to be used for the transmission rather than a monitor.

Changeset: 5bb5bf201c53de0067ce838d56dcdb58f0476d6d

[Peter Parker]

Refs #9349 - Have trans ROI var as well as monitor.

Get rid of the mess to do with the property, getter and setter. It was uncessary.

Changeset: 24b9bd287432c2246e68a8ebdfd117f836f33442

[Peter Parker]

Refs #9349 - Add code to extract spectra from a ws.

Includes addition to the SANSUtility unit tests, which apparently were never being run as the file was not listed in CMakeList.txt.

Changeset: 2dc999095d9c1bfd501c7a29b5e9144c348742d0

[Peter Parker]

Refs #9349 - CalcTrans reduction step now accepts ROI.

Changeset: 399e2d85171af9d1f0a39890cefe05de6191956d

[Peter Parker]

Refs #9349 - Utility to extract masked detector IDs.

Changeset: 910f8a49a39df89f82492efcfd0b799faff60475

[Peter Parker]

Refs #9349 - Utility to get all detector IDs in a component.

Changeset: 0817d4806a31ec1881a3c8ba1b728e6baa11148e

[Peter Parker]

Refs #9349 - Fully support "beam stop out" commands.

Everything should be in place now.

Changeset: 159b1d98099a5a7c3a6cb445f2e32f3796ce2ff3

[Peter Parker]

Refs #9349 - Fix RHEL6 unit tests...

It's now clear why these tests were never ran. I am removing them but will open up a new ticket to reinstate them at a later date. We're not losing anything here, since they were never run to begin with.

Changeset: bf160b8cebdb4d3abcf2c631f7fce550f8c15b85

[Peter Parker]

Refs #9349 - Also fix doxygen warning.

Sorry for the double checkbuild.

Changeset: 64af4e682ec07597677c28c424629debaba9afe2

[Peter Parker]

Refs #9349 - Fix last compiler warning.

The variable was never used unitialised, but initialising it all the same.

Changeset: 304d626b5919e56f6aad6913e5e8d82fe38c6c6c

[Stuart Campbell]

Hi Peter, Just a heads up that (I think) some of your changes (in particular to CalculateTransmission.cpp) break the build for RHEL7.

Just thought I would let you know.

Cheers Stu

[Martyn Gigg]

So it looks like the breakage on RHEL7 is actually a regression introduced in boost 1.53 and fixed in 1.54: ​https://svn.boost.org/trac/boost/ticket/7421. Unfortunately, RHEL7 is already behind the times as both Fedora 20 & Ubuntu 14.04 have 1.54 so neither of those build wouldn't of actually picked this up either.

[Martyn Gigg]

Fix ambiguous overload error with boost 1.53

Refs #9349

Changeset: 4f0b78a85244ae32e5a0a308a4d8c2d812089842

[Peter Parker]

Update:

The code that has been done so far was tested by Richard on a nightly build and seemed to work ok. Unfortunately the code is now very stale so it will either have to be recreated from a fresh copy of master, or painfully brought up to date with a merge.

The problem which held us back from completing this 8 months ago was the dramatic increase in time it took to do reductions with any of the new User File commands. This is still an issue, but Rob's suggestion of grouping ROI spectra rather than masking everything else may bear fruit.

Anton, we should set aside some time to talk a bit about these changes before you start to pick up this ticket. Hopefully it is relatively straightforward and wont take too long.

[Stuart Campbell]

This ticket has been transferred to github issue ​10192