Decaying using MadSpin gives incorrect xsec or max weight

Asked by Jay Sandesara on 2020-05-22


I am trying to generate g g > z z g g [noborn=QCD] with z > l+ l- in MadSpin. I was successful in generating the process with z resonances. In an earlier answer, you mentioned how Loop Induced processes can only be decayed upon setting spin mode none in the MadSpin card. Since the two jet contribution ought to be a small correction to the ggF 0+1 jet processes, I decided to go forward with this reduced accuracy. However the total weight I get after decaying the z resonances is almost similar to the xsec of g g > z z g g process. It is even greater than 1 jet process.

For further checks, I ran g g > z z first and then decayed the z resonance to two leptons and compared the results with g g > l+ l- l+ l- xsec. Again the results are widely different between the two. Doing MadSpin way gives me the max weight of about 5.4102e-03 while g g > l+ l- l+ l- gives 2.861e-03.

Can setting spin mode to none cause such a large discrepancy? Or am I doing something wrong? Heres how the spin card looks like:

 set Nevents_for_max_weigth 2000
set spinmode none
set max_weight_ps_point 400

decay z > l+ l-; z > l+ l-

Question information

English Edit question
MadGraph5_aMC@NLO Edit question
No assignee Edit question
Solved by:
Jay Sandesara
Last query:
Last reply:

This question was reopened

Your Madspin_card is weird.
I would expect a crash with such syntax.

Could you try with this syntax?

set Nevents_for_max_weigth 2000
set spinmode none
set max_weight_ps_point 400

decay z > l+ l-

I do not have time to look today. But for me this sounds a symmetry factor which is wrong set (maybe due to you weird syntax)




I have checked with your syntax and
1) your syntax is indeed not supported as you can see from the warning:
Command "z" not recognized, please try again

2) I have check madspin like this:
generate p p > z z
decay z > l+ -
and it returns 0.044 pb

I have compare it with the decay chain syntax:
generate p p > z z, z > l+ l-
and I also got 0.044 pb
This shows that madspin compute correctly the branching ratio and the symmetry factor.

3) I'm a bit worried about the generation that you try to perform.
g g > z z g g[noborn=QCD] is included in the cross-section of g g > z z [noborn=QCD] (and in g g > z z g [noborn=QCD])
However you have singularity that you need to handle correctly. So if you plan to combine those process you need to use a matching/merging scheme. However for such scheme you can not restrict your self to only gluon flavor.
We do not have an automated way to select only loop-induced diagram + QCD radiation but it is technically possible to create a loop diagram filter to keep only those type of loop in the computation.



PS: If you have twice the line "decay z > l+ l-" in your madspin_card then I expect that madspin put the wrong symmetry factor since it will assume that you have different decay for each branch.

Jay Sandesara (jaysandesara) said : #3

Hello Olivier,

Thanks for replying!

1) Indeed, I mis-typed the madspin syntax. It was supposed to be decay z > l+ l- ; decay z > l+ l-. But you're right this is not good. I retried with just z > l+ l-. The xsec still doesn't match.

2) That example works indeed.

3) About the generation I am trying to perform, yes I do plan to do CKKW merging on these individually generated samples using a Pythia macro later. The reason I am just focusing on g flavors is really only simplicity and speed. I will try the full flavor simulation at a later stage if this succeeds. I have indeed successfully filtered out loop diagrams using Valentin's answer: . In here he also mentions that doing just g g > z z g g shouldn't require any filters.

I got
original cross-section: 1.14015798681 for g g > z z [noborn=QCD]
after MS I got: 0.005396
which is
1.14015798681 * BR**2
1.14015798681 * 0.06877462 **2

For the case p p > z z
you got for the production 9.4348
and after decay 0.044 pb which is also 9.4348 * 0.06877462 **2

Concerning g g > z > l+ l- l+l- /a [noborn=QCD]
I believe that you forget to remove the cuts on the lepton. I do not reproduce your number for that process.



Jay Sandesara (jaysandesara) said : #5

I forgot to mention that in both the samples I put a cut-off on the invariant masses (>180 GeV). For g g > z z sample on the z boson invariant mass and in g g > 4 lep on the 4 lep invariant mass in the run card. That could be why your numbers are different than mine. I removed all the other cuts in the run card of g g > 4 lep so it could be similar to g g > z z . Values are slightly different from before now but still incompatible.

Olivier Mattelaer suggests this article as an answer to your question:
FAQ #2442: “why production and decay cross-section didn't agree.”.

Jay Sandesara (jaysandesara) said : #7

Doing False = cut_decays did the job! Thank you!

Jay Sandesara (jaysandesara) said : #8

One other thing that helped more I think was doing

decay z > mu+ mu- ; decay z > e+ e-

in place of

decay z > l+ l-

Jay Sandesara (jaysandesara) said : #9

One last question I have is, would doing:

decay z > mu+ mu- ; decay z > e+ e-
decay z > mu+ mu- ; decay z > mu+ mu-
decay z > e+ e- ; decay z > e+ e-

in the same MadSpin card work? Or do I need to generate separate samples for each of them?

Jay Sandesara (jaysandesara) said : #10

Nevermind, I get an error doing the above. I apologize for the question. Just to be clear doing

decay z > mu+ mu- ; decay z > e+ e-

will produce events with one z decaying to muon pairs and the other z decaying to electron pairs? Or it will produce all different combinations?

Jay Sandesara (jaysandesara) said : #11

Again, sorry I took a look at the .lhe file and I found the answer. It seems to be equivalent to doing g g > mu+ mu- e+ e-. I think the incompatibility problem here lies with the invariant mass cuts I put. Thank you for your help!


Yes your invariant mass cut can break the Narrow-width approximation since you cut a huge part of the phase-space for the 4l case which is not cut for the not decayed counter-part