Skip to content

Use standard error of the mean instead of variance

Henrik Stooss requested to merge introduce-sem-Dict into main

@schlaicha figured out that the error bars of EpsilonPlanar were too high because I used the variance instead of the standard error of the mean in the error propagation. Trivial mistake... But it leads to the question if we even need the variance, or should scrap it in favor of the SEM.

In this PR I introduced the SEM as a Results Dict just like the means and variances of the observables and applied it to EpsilonPlanar.

Notice that at one point in the calculation of the results, the variance of M_perp is needed. I just replaced that by calculating the mean of the square of M_perp (which also shows the strength of the introduced Dicts!), showing that variances are never needed explicitly.

Merge request reports