LoadMuonNexus fails to load specified spectra

[Raquel Alvarez Banos]

When SpectraMin, SpectraMax or SpectraList are set in the algorithm's dialog, the set of spectra loaded is the set {sp=1,2,...n}, where n is the total number of spectra specified.

For example:

• Load MUSR00015189
• Set SpectraMin=5, SpectraMax=10 and SpectraList=15,20
• The output workspace contains spectra 1,2,...,8

[Raquel Alvarez Banos]

Re #11165 Removing unnecessary argument

Changeset: 3294017e74ad57dd0cb5d7ce745f95cb8d8c7050

[Raquel Alvarez Banos]

Re #11165 Using correct spectrum indices

Changeset: e0410e74284709c3dab4a4712301326be4f12e04

[Raquel Alvarez Banos]

Re #11165 Further corrections

Changeset: fae3732eadaa8a03c345cd389e45a643cd0166bc

[Raquel Alvarez Banos]

Re #11165 updating unit test SpectrumList index

Changeset: 3beb409f52c8e25fb3e1622320c3bae270e9d842

[Raquel Alvarez Banos]

Re #11165 Add unit test to check selected spectra

Changeset: b7699949cb6d010f4c5a1e15f5a9ce192af0d606

[Raquel Alvarez Banos]

Re #11165 Set bounded validator lower value to 1 for spec

Changeset: 2a4d26df524a4e0217a590a861db9824b5d70fb1

[Raquel Alvarez Banos]

Re #11165 Further corrections to ensure systemtests pass

Changeset: cf38b4cd35d5eef0da0b33273cd997201e0c575f

[Raquel Alvarez Banos]

Re #11165 Bounded validator should be 1 for spec and 0 for entry

Changeset: 10c566626d1d031eb721aa83ed66e834e7429b33

[Raquel Alvarez Banos]

Re #11165 change unit test name

Changeset: 7de06be3352a1e2f4e65e6451cb2c976a3669026

[Raquel]

This is being verified as ​pull request #302.

[Raquel Alvarez Banos]

Re #11165 temporary fix for unit test

Changeset: 81133ad8a2d541b32d3b2fb105477b5ee4ed3c35

[Karl Palmen]

I don't think SpectraMin & SpectraMax are supposed to be used with SpectraList. One should use one or the other. Nevertheless the effect of using both that you describe is not good.

I'd expect either an error message or something sensible such as (5,6,7,8,9,10,15,20) or no spectra and the documentation explains this.

[Karl Palmen]

I tried it out and found a similar situation occurs when spectraList is left blank. Select spectraMin=11 and SpectraMax=20. The spectra displayed are labelled 1 to 10, when the data is shown and when the spectra a plotted. Nethertheless I think the correct spectra are shown, because I get a different data for SpectraMin=1 and SpectaMax=10.

The labels sp-1, sp-2, etc seem to be incorrect.

[Raquel Alvarez Banos]

Yes, you are right, and I chose a bad example in the description: The effect is not caused by the fact that the user provides both SpecMin/Max and SpecList, but to the fact that spectrum numbers are not set properly in the algorithm. For instance, as you mentioned in your last comment, if you try only a range via SpecMin+Max, the result is the same: spectra labeled from 1, 2, ... to n. The same happens if you only supply SpectraList.

But you raised an interesting question, as I thought that Min/Max and List could be used together. Can't they? Which one should take preference? The way I was working on this would allow user to end up with (5,6,7,8,9,10,15,20). I guess this is then incorrect.

[Karl Palmen]

( 5,6,7,8,9,10,15,20) may be what you get and it's just the spectra that are labelled incorrectly. It may be worth checking the behaviour of other load algorithms with SpectraMin etc.

[Raquel Alvarez Banos]

Yes, the point is exactly that: you get the right set, but they are labelled incorrectly. I have searched on mantidproject.org for "SpectrumMin" and from the algorithms returned by the search, the following

• LoadEventNexusv1
• LoadISISNexusv2
• LoadRawv3
• LoadRawBin0

allow using SpectraMin/Max and List simultaneously. After speaking to Anders, I am going to use that convention, that is, if user specifies:

SpectrumMin=5 + SpectrumMax=10 and SpectrumList = 15, 20

He will end up with the set (5,6,7,8,9,10,15,20).

[Karl Palmen]

Yes, this seems to be a good idea. There may be scientists who rely on it even though the spectra are not labelled correctly. Perhaps, something about it could be added to the documentation.

[Raquel Alvarez Banos]

Re #11165 Using specNo instead of spec list

Changeset: e69ba4f22a554f64e1fa3c0535acf66a4da1b9fe

[Raquel Alvarez Banos]

Re #11165 Previous temporary fix will be permanent, comment added

Changeset: 8392233b6f11fa95432b5511dba806912c12c52c

[Raquel Alvarez Banos]

Re #11165 Check and handle duplicate spec numbers

Changeset: f7b83734e248650a6b2b6f16bc6080566eadddc8

[Raquel Alvarez Banos]

Re #11165 Updating property description

Changeset: 5ec71a0400cb6a7898ce37650b45611d38b29652

[Raquel Alvarez Banos]

Re #11165 Update doc and add user examples

Changeset: 5c7d12ed7c7bfb0f08a5e555e1720e13b658c59f

[Karl Palmen]

I see you have removed the reference to single period and multi period data from the comments. Does it now work with multi-period data?

[Raquel Alvarez Banos]

Yes, it does (or it should). By the way, I do not why it was said in the old documentation that these properties only work with single period data, it seemed to me that it was "working" (except for the wrong labelling) with multi period data as well...

[Raquel Alvarez Banos]

Re #11165 Fixing phase quad doc test

Changeset: a45a05f21a18d098caeb642cf7f1990e6c46184a

[Raquel]

Jenkins, retest this please

[Raquel]

Jenkins, retest this please

[KarlPalmen]

Merge pull request #302 from mantidproject/11165_LoadMuonNexus_fails_to_load_specified_spectra

LoadMuonNexus1 fails to load specified spectrum number

Full changeset: ​9315e6c84e8d51a72dfeb7f327711b655b6e224e

[Stuart Campbell]

This ticket has been transferred to github issue ​12004