MadGraph5_aMC@NLO

aMC@NLO p p > z j scale variations

Asked by Luke Lambourne

Dear aMC@NLO authors,

I'm generating theory predictions (8 TeV) for high pT Z->bb. We've generated the Z plus 1 jet process with aMC@NLO and Powheg, where the particle level cross sections passing our fiducial cuts are:

Powheg: 2.02 - 0.19 + 0.25
aMC@NLO: 1.98 - 0.08 + 0.16

where the errors are the scale uncertainties. I was surprised to see that the aMC@NLO errors came out much smaller than the Powheg ones and also that the aMC@NLO errors are quite asymmetric. I'd naively expect the scale errors to come out similar in size, do you have any ideas why this is not the case?

For both generators the scale uncertainties were evaluated by simultaneously scaling muR and muF by 2.0/0.5.

Thanks,
Luke

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Stefano Frixione
Solved:
Last query:
Last reply:
Revision history for this message
Best Stefano Frixione (stefano-frixione) said : 		#1

Hi,
did you get the errors by rerunning the code and by changing the inputs, or did you
obtain them by using the reweighting information stored in the event file, or was
the above quoted from the summary printed out at the end of the run on the screen?

It's hard to tell why the errors are asymmetric, but usually this is not the signal that
something went wrong. Regardless of the asymmetry, please note that MC@NLO and
Powheg are equivalent only up to NLO -- beyond that order, they differ. So, there's
no reason in principle why the scale uncertainties should come out the same

Best, Stefano.

Revision history for this message
Luke Lambourne (luke-lambourne) said : 		#2

Hi,

I got the errors by running the code and changing the inputs. The error is on the particle level cross section that passes a set of fiducial cuts in rivet.

That's useful to know that these scale variations shouldn't necessarily come out roughly the same for the two generators .

Thanks,
Luke

Revision history for this message
Luke Lambourne (luke-lambourne) said : 		#3

Thanks Stefano Frixione, that solved my question.