MadGraph5_aMC@NLO

Different number of real emission diagrams between 2.9.x and 3.3.x

Asked by Sanghyun Ko

Dear Madgraph5 experts,

I found that the xsec of diphoton process is significantly different between version 2.9.x and 3.3.x (with Marco's patch to make it run).

https://bugs.launchpad.net/mg5amcnlo/+bug/1960832

While digging into the issue, I've observed that there is a difference in the number of real emission diagrams in the first place between the two versions. For example, if I do

import model loop_sm-no_b_mass
generate p p > a a [QCD] @0
add process p p > a a j [QCD] @1

then for the process 'p p > a a j [QCD]' (without jet it gives the same diagrams), I get 30 subprocesses with 5050 real emission diagrams, 180 born and 2340 virtual in 2.9.x, but 30 subprocesses with 7400 real emission diagrams, 180 born and 2340 virtual in 3.3.x.

I was assuming they should be the same when we don't have any QED correction, so I was wondering whether there was any change in the logic to estimate the diagrams.

Best,
Sanghyun.

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
davide.pagani.85 Edit question
Last query:
Last reply:
Revision history for this message
Sanghyun Ko (sang.hyun.ko) said : 		#1

PS: we found that the same happens for the V+jets as well. For example,

import loop_sm-no_b_mass
define ell+ = e+ mu+ ta+
define ell- = e- mu- ta-
generate p p > ell+ ell- [QCD]
add process p p > ell+ ell- j [QCD]
add process p p > ell+ ell- j j [QCD]

would generate real emission diagrams with the following numbers in 33x (29x):
0j - 360 (360)
1j - 8560 (6480)
2j - 128820 (82500)

so this also happens when we have no tagged photon in the process.

Revision history for this message
marco zaro (marco-zaro) said : 		#2

Hi,
yes, there is a difference, however the results are identical. In v3, because of the mixed-coupling expansion, some contributions which end up giving zero are generated for the real emission. It should only give a mild efficiency effect.

Cheers,

Marco

Can you help with this problem?

Provide an answer of your own, or ask Sanghyun Ko for more information if necessary.

To post a message you must log in.