Event record shows Z decay to e+mu-, leptons attached to wrong boson?

[Johannes Ebke]

I was using the following proc card with a vanilla MadGraph 1.4.6

import model heft
define l = e- mu- ta-
define l~ = e+ mu+ ta+
define v = ve vm vt
define v~ = ve~ vm~ vt~
generate p p > l v~ l v~ l~ v QED=99 QCD=99
add process p p > l v~ l~ v l~ v QED=99 QCD=99
output PP_VVV

The run card specified proton(4000)-proton(4000) collisions, with no other changes. I used the multicore run mode.

I visually inspected the attached unweighted_events.lhe.gz using mcviz and noticed a Z boson decaying into a positron and a negative muon; a rather unusual combination. In the same event, the W- decays into an electron and a muon antineutrino, so in total it adds up. This suggests that there is a bug in the association of the decay products with their parents.

Cheers,
Johannes

[Johan Alwall (johan-alwall)]

Hello Johannes,

Thanks for your report. Was this a gridpack run, or a regular run?

Thanks,
Johan

[Johannes Ebke (ebke)]

That was a regular, multiprocessing run on one machine.

[Johan Alwall (johan-alwall)]

Dear Johannes,

Thanks a lot for this bug report. It turned out that the problem was related to how matrix elements were combined, so it happened (sometimes, in very complicated processes like this) that two matrix element that had the same diagrams but different relative particle ordering could be combined into a single matrix element. This lead to an error in the particle code assignment at event generation time, giving the Z > e+ mu- events.

I have fixed this in the development version 1.4.7. Can you please branch this version and make sure that the problem doesn't occur there? You simply install bazaar from http://wiki.bazaar.canonical.com/Download and then download the branch using

bzr branch lp:~maddevelopers/madgraph5/1.4.7

Many thanks,
Johan

[Johannes Ebke (ebke)]

Dear Johan,

I got the branch and re-ran the generation, and now i do not see such events anymore!

Thanks for the quick fix,

Johannes