Beta from single files far away from beta for all the files

[piotrskowronski]

Attached example showing the issue.
AnIssue.tar.gz
Extract the tarball in /tmp and run by calling run.sh.
It will produce pictures with beta beating (bbx.png, bby.png)

There are 6 input files. Beta calculated for single files are far from beta for all the files.
Of course I use the same options for all the cases. The output is also the same, so I assume that N-BPM method is used in all the cases. Is it?

Which of the beta beating is (more) correct? All the error bars are very small, apparently for one of the cases it is wrong.

Below phase beating

[awegsche]

The issue of running the N-BPM method with single files:

For a single file there is no statistical error. As statistical errors are usually dominating this means for single files that the force of the N-BPM algorithm -- the calculation of the correlation between the different results and their dependence on errors -- will barely be used.

The result can be even less reliable than the 3BPM method since many imperfections that lie in between the BPMs deteriorate the obtained beta values.