Speed up asymmetry calculation

[Raquel Alvarez Banos]

PlotAsymmetryByLogValue loads a set of muon runs and computes the asymmetry as a function of some log value. Currently, the data loading and the asymmetry calculation are performed within the same loop. These tasks should be carried out separately so that users can play with asymmetry parameters (for instance integration limits) without having to load all the datasets every time.

[Raquel Alvarez Banos]

The main loop currently reads:

  for (size_t i = is; i <= ie; i++) {

    // Load run, apply dead time corrections and detector grouping
    Workspace_sptr loadedWs = doLoad(i);

    // Analyse loadedWs
    doAnalysis (loadedWs, i-is);

    progress.report();
  }

We could add an if statment to check whether the loadedWs were already loaded or not. It seems that the best option is to add the temporary workspaces to the ADS, and end up with something like this:

  for (size_t i = is; i <= ie; i++) {

    if ( loadedWs is not in ADS ) {

      Workspace_sptr loadedWs = doLoad(i);
      Add_loadedWs_to_ADS;
    
    } else {

      loadedWs = ADS.retrieve( loadedWs_name );
    }

    // Analyse loadedWs
    doAnalysis (loadedWs, i-is);

    progress.report();
  }

Two boolean flags should be added to:

• Hide/show the temporary workspaces
• Keep/remove loaded workspaces

[Raquel Alvarez Banos]

(In #6931) I have attached a script I have been using to test the changes, which the tester may find useful. It includes descriptions (comments) of what I would expect in each successive execution of the algorithm.

[Stuart Campbell]

This ticket has been transferred to github issue ​12158