MadGraph5_aMC@NLO

zz>4l madspin: factor 2 from declaration of same decay for same two particles

Asked by Oleksii Kurdysh

Hi experts,

we're generating zz>4l EFT aQGC samples and using madspin- start from
  generate p p > j j z z QCD=0 T0==1
then how to define z decays among there two options?:
case1) decay z > l+ l-; decay z > l+ l-
case2) decay z > l+ l-

1) and 2) give the same lepton kinematics and everything the same except in case 1) sum of weights and cross-section is exactly twice larger
cross-section from log file (for FT0 QUAD) before madspin is 0.07496 +- 0.0001524 pb
after madspin
in case1) cross-section (nb)= 6.789e-07
and in case 2) cross-section (nb)= 3.395e-07

estimation starting from production gives
sigma*(BR(eeee)+BR(mumu)+BR(eemumu))/2 = 0.07496*((0.033*2)**2+(0.033*2)**2+2*(0.033*2)**2)/2 = 0.00065
(factor 2 from the indistinguisahble boson)
which is much closer to case1) xsec after decay

-----
another thing with case2) which is kind strange, when generating 13200 events:
decay z > l+ l-; decay z > l+ l- >>>>> generate 13200 decay event for particle z
decay z > l+ l- >>>>> generate 26400 decay event for particle z
why factor 2 for decay events for case 2) ?
------

how to define lepton decays of two Z?
is this factor two coming indistinguishable particles?

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
Last query:
Last reply:
Revision history for this message
Best Olivier Mattelaer (olivier-mattelaer) said : 		#1

Asking for twice the same decay, while the default syntax is to apply the same decay to all particles is not something that we have think of.

So our method to support for double decay is indeed designed for different decay when the decay broke the symmetry between the production and where we need to multiply the branching ration by that combinatorics to take into account that any of the identical particle can decay into any of the requested decay.

So yes, it does not surprise me that if you ask madspin to do
decay z > l+ l-
decay z > l+ l-
it fails to realise that this is in fact the same decay and therefore returns a wrong cross-section (which is in the case of madspin just a mulitplication of branching ratio plus symmetry factor.

Since you seem to suggest that 1) would be the correct syntax (and as stated above, this would suprise me).
I have done the following test (in the SM):
1) in madgraph: generate p p > z z, z> l+ l-: this returns: 0.04219 +- 8.413e-05 pb
2) in madgraph generate p p > z z, z> l+ l-, z > l+ l-: this returns 0.04219 +- 8.413e-05 pb (same seed used here)
3) in madgraph generate p p > z z + single decay in madspin: 0.04419 ± 7.9e-05
4) in madgraph generate p p > z z + double decay in madspin: 0.0442 ± 6.7e-05

So here I actually fail to see a difference between the two syntax.
Now the above computation was done in the "default" (complete) mode of madspin.
That mode does not have any line like "generate 13200 decay event for particle z" so
I guess that you are using some non default mode...

So I switch to "spinmode=none" (which I would not recommend at all in this case).
5) in madgraph generate p p > z z + single decay in madspin (spinmode=none): 0.04432 ± 7.9e-05
6) in madgraph generate p p > z z + double decay in madspin (spinmode=none): 0.08809 ± 0.00014
This mode (and spinmode=onshell) have indeed a different method to compute the cross-section due to the possibility to take cuts into account in this mode (which is not possible in the default mode).

----
another thing with case2) which is kind strange, when generating 13200 events:
decay z > l+ l-; decay z > l+ l- >>>>> generate 13200 decay event for particle z
decay z > l+ l- >>>>> generate 26400 decay event for particle z
why factor 2 for decay events for case 2) ?
------

Well the factor of 2 is coming from the fact that you have two Z to decays. As said above, we have never thought about syntax 1). I would have naively expected to need two lines "generating 13200 events" but since they have the exact same matrix-element, they are likely over-writting each other on disk and therefore the second line is kind of drop (not sure just guessing). In any case, the code is built in a way that such number is kind of irrelevant since madspin can ask for more event if that number is not large enough.

Cheers,

Olivier

Revision history for this message
Oleksii Kurdysh (okurdysh) said : 		#2

Hi Olivier,
many thanks for clarifying and check!

spinmode is indeed none as there are cuts - which spinmode would you recommend in that case? apart from the problem above (can change syntax) why not none?

then if every decay syntax means "hey madspin please decay all Z in this event(two) to (same)leptons" - all Z decay, what happens when decay syntax is called again but Z already decayed? why it's not a crash?

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

Hi,

> spinmode is indeed none as there are cuts

Remember that the cuts applied for spinmode=none are in the restframe of the Z. So for Z > l+ l-
I'm not sure which cut would be relevant in that frame for such decay...

> - which spinmode would you
> recommend in that case? apart from the problem above (can change syntax)
> why not none?

spinmode=None,
does not have spin-correlation neither any offshsell effect.
That mode was designed for the Higgs decay and as a baseline for comparison with other mode since this is basically the most simple/naive (worst?) prediction that you can do.

For Z-decay, I would trust more the parton-shower way to do the decay than spinmode=none.
At bare minimum, I would say to use spinmode=onshell which has the correct spin-correlation but miss off-shell effect.
(a mode which also allows cuts). But I would recommend madspin in normal mode, and apply a post-processing of the lhef file to drop events that do not pass the cuts that you want to use.

> then if every decay syntax means "hey madspin please decay all Z in this
> event(two) to (same)leptons" - all Z decay, what happens when decay
> syntax is called again but Z already decayed? why it's not a crash?

No, if you have a single
"decay X > ..."
this means that such decay applied to all X.

if you have the exact same number of "decay" as the number of decaying particle,
then it means that each particle decays according to a different rule.
If you have a different number of "decay" as the number of decaying particle (typically case is three decay line in madspin for two in production), then we go back to a symmetric case, where all particle can decay into all production mode.

So given the above, implementing a crash is not trivial and at least as complicated as checking fully that two decay syntax are actually 100% equivalent and therefore bypass that symmetry factor in that part of the code (like it is done in the other part of the code).

Given that I want to introduce a new mode which is in spirit close to onshell but as precise as full, I will look at what i can do (not really for you case since you do not really need a fix but for case with triple decay this might be meaningfull).

Cheers,

Olivier

> On 22 Apr 2024, at 22:15, Oleksii Kurdysh <email address hidden> wrote:
>
> Question #809272 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/809272
>
> Status: Answered => Open
>
> Oleksii Kurdysh is still having a problem:
> Hi Olivier,
> many thanks for clarifying and check!
>
> spinmode is indeed none as there are cuts - which spinmode would you
> recommend in that case? apart from the problem above (can change syntax)
> why not none?
>
> then if every decay syntax means "hey madspin please decay all Z in this
> event(two) to (same)leptons" - all Z decay, what happens when decay
> syntax is called again but Z already decayed? why it's not a crash?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
Oleksii Kurdysh (okurdysh) said : 		#4

Hi Olivier,
clear now, thank you very much for clarifying,
problem is solved

just as comment as you mention triple decay. when having
 generate p p > j j w+ w+ w- QCD=0
and both w+ decaying in the same way, correct syntax for decay is
 decay w+ > l+ vl; decay w- > l- vl~
right? instead of "decay w+ > l+ vl; decay w+ > l+ vl; decay w- > l- vl~"

Revision history for this message
Oleksii Kurdysh (okurdysh) said : 		#5

Thanks Olivier Mattelaer, that solved my question.

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#6

Yes you are correct.

I consider this as a "double" decay plus a single decay, triple decay would mean something like triple Z or triple Higgs production.

Cheers,

Olivier

Revision history for this message
Oleksii Kurdysh (okurdysh) said : 		#7

Hi Olivier,
okay, again thanks for your replies

Best regards
Oleksii