MadGraph5_aMC@NLO

Asymmetric pseudorapidity distribution in p p > mu+ vm after showering

Asked by Jean-Loup Tastet

Dear MadGraph authors,

I obtain some puzzling results when generating `p p > mu+ vm` and showering the resulting events with Pythia 8.

I used the default settings and generated three event sets:
* LO: `p p > mu+ vm`
* NLO: `p p > mu+ vm [QCD]`
* NLO (MadSpin): `p p > w+ [QCD]` then `decay w+ > mu+ vm` in MadSpin.

The pseudorapidity distribution of the w+ mediating the event (selected as the first w+ in the event record) presents a strong forward-backward asymmetry, in opposite directions for LO and NLO (see this plot: https://cernbox.cern.ch/index.php/s/yVd9YpzkVUUAmUe). I did not expect that since the initial pp state is symmetric.

However, the pseudorapidity distribution looks fine at parton level, as can be seen on this plot: https://cernbox.cern.ch/index.php/s/NbtRdhUuiTupgsk.

Do you have any idea what could be causing this effect? (or what I could be doing wrong...)

Many thanks in advance,
Jean-Loup

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Jean-Loup Tastet
Solved:
Last query:
Last reply:
Revision history for this message
Jean-Loup Tastet (jltastet) said : 		#1

I have fixed my analysis to use the last w+ in the event record (which is the one decaying to mu+ vm) instead of the first one. However, this does not qualitatively change the results.

Revision history for this message
Jean-Loup Tastet (jltastet) said : 		#2

I see the exact same effect for `p p > mu+ mu-` and `p p > mu+ mu- [QCD]`, with the same pseudorapidity distributions for the Z, both at LO and NLO: https://cernbox.cern.ch/index.php/s/0Xe7R5SegNtEFfi

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#3

Hi,

Do not have that much time to look at this in details before march 9 (I'm not a full time employee so ...)
But a very quick look seems to indicate that this is a pure parton-shower issue. Don't you agree?
In that case, you should rather ask help to py8 author (but my guess is that this is an issue of your analysis)

Cheers,

Olivier

Revision history for this message
Jean-Loup Tastet (jltastet) said : 		#4

Hi Oliver,

Thanks for your answer. I will try to use a different generator for the parton shower, and see if the issue persists.

Cheers,
Jean-Loup

Revision history for this message
Jean-Loup Tastet (jltastet) said : 		#5

This was indeed a bug in my analysis. MadAnalysis turned out to be very helpful for debugging this issue, by providing an independent cross-check of the analysis.