Histograms incorrectly normalized

Bug #1204088 reported by Markus Ebert
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
MadAnalysis 5
Fix Released
High
Benjamin Fuks

Bug Description

I observed a wrong normalization in the histograms I obtained when running MadAnalysis on a MadGraph5-output.
There are two aspects:
1) The normalization differs with the output type, i.e. the plots in the HTML output are differently normalized than in the PDF ouput (and both differ from the DVI output).
This concerns:
- histograms plotted with "stackingmethod = normalize2one", independent of "main.normalize"
- all histograms if "main.normalize = lumi"
In the latter case, the histograms from the 3 output formats differ exactly by a factor corresponding to the luminosity (see attached plots)
2) I did not figure out in which cases any of the histograms actually matched the expectation. (I compared results with an analytic calculation)
Sometimes the HTML-output showed the correct normalization, but most often none of the plots matched.

I made an example from the process p p > t t~.
In the attached file, I included the PDF- and DVI-output for both cases "main.normalize = lumi" and "main.normalize = none".

Revision history for this message
Markus Ebert (markus-ebert) wrote :
Revision history for this message
Benjamin Fuks (fuks) wrote :

Hi Markus,

This is fixed for a couple of weeks in version 1.1.8. Could you please test it and tell us if the problem persists?

Cheers,

Benjamin

Changed in madanalysis5:
assignee: nobody → Benjamin Fuks (fuks)
importance: Undecided → Critical
milestone: none → v1.1.8
status: New → Confirmed
status: Confirmed → Fix Committed
importance: Critical → High
Revision history for this message
Markus Ebert (markus-ebert) wrote :

I just tested version 1.1.8.
It seems to solve problem 2, i.e. the histograms in the HTML version match my expectation.
However, the first problem still occurs, namely that the scales in the plots contained in the different output formats still differ.

Revision history for this message
Markus Ebert (markus-ebert) wrote :

Update:
Only the histograms normalized to the luminosity work.
Histograms with the option normalize2one are obviously wrong, see e.g. the plot of rapidity p p > t t~ attached.
Problem 1 also occurs for them.

Cheers,
Markus

Revision history for this message
Markus Ebert (markus-ebert) wrote :
Revision history for this message
Benjamin Fuks (fuks) wrote :

Hi Markus,

Could you please retry with v1.1.8 now. In principle, I have fixed all the problems. Please let me know

Cheers,

Benjamin

Revision history for this message
Markus Ebert (markus-ebert) wrote :

Hi Benjamin,

it now works perfectly fine.
Thanks for the quick fix!

Cheers,
Markus

Benjamin Fuks (fuks)
Changed in madanalysis5:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.