Improvements to PlotAsymmetryByLogValue

[Anders Markvardsen]

See below message from James:

Error bars from PlotAsymmetryByLogValue

Further odd behaviour. Plotting the individual red and green data sets, they swap places between runs (sometimes Red>Green and sometimes Green>Red). So why is the Red-Green value consistently negative?

Also there are some repeated points in the graph (scripting/instrument control problem so those runs were actually all taken at the same X value, not Mantid's problem). Why is only one of the error bars plotted when all the 4 points have valid error values in the table?

James Lord. From: Lord, James (STFC,RAL,ISIS) Sent: 11 July 2011 17:29 To: Mantid Help Subject: Error bars from PlotAsymmetryByLogValue

I'm using PlotAsymmetryByLogValue to analyse data with 2 periods, where the 2 data sets will have very similar total counts and therefore statistics. As expected I get a table with 4 rows: Red-Green, Red, Green, and Red+Green. I'd expect that the errors on Red and Green are uncorrelated so if Red and Green have error e, the sum and difference would be about e*sqrt(2). But I get an error on the sum of about e*2 and on the difference it's about 6*e. I think the errors on the individual periods are about right based on the scatter between runs.

James Lord. Mantid version 1.29.12931 on Windows XP.

[Anders Markvardsen]

For this ticket fix:

1) when you start up PlotAsymmetryByLogValue with valid First run and Last Run the 'Log value' is not populated - ensure it is populated

2) Ensure that the errorbars are added as according to the data are uncorrelated, that is sum of square roots

The create plot James refer to:

First run
hifi\data\23084 Last run
hifi\data\23099 Log value Field_main Red 1 Green 2 Forward spectra 1-32 Backward spectra 33-64

Alternative use for runs

First run
hifi\data\22927 Last run
hifi\data\22940

[Anders Markvardsen]

Add to list above

3) Understand why the Red-Green value consistently is negative....

[Anders Markvardsen]

In [14016]:

Fix: when you start up PlotAsymmetryByLogValue with valid First run and Last Run the 'Log value' is not populated - ensure it is populated

re #3354

[Anders Markvardsen]

In [14088]:

Fix: for PlotAsymmetryByLogValue

2) Ensure that the errorbars are added as according to the data are uncorrelated, that is sum of square roots
3) fixed Red-Green issue

To test first run

First run:
hifi\data\23084 Last run
hifi\data\23099 Log value Field_main Red 1 Green 2 Forward spectra 1-32 Backward spectra 33-64

then the same as above but with Forward spectra and Backward spectra fields empty

re #3354

[Roman Tolchenov]

I think the error for Red+Green should be sqrt(E2 + E12) not sqrt(E) + sqrt(E1)

[Anders Markvardsen]

In [14813]:

Somehow missed how do add gaussian errors - should be ok now. re #3354

[Stuart Campbell]

This ticket has been transferred to github issue ​4201