MadGraph5_aMC@NLO

Large width will contribute to on-shell propagator

Asked by ParticleCat

As I generate the events with an off-shell proceess as t-channel.
The fermion propagator N have a decay mode N -> W^+ l^- and the width will even greater than the mass of N around M_N>10M_W.
Compare the simulation result with the analytic formula it seems the MadGraph put the width in the propagator.
For off-shell propagator with t-channel, the momentum transfer is space-like.
The width should not contribute to the process.
However, if I set the width myself but not use the "auto" syntax, will I got a wrong answer in the s-channel ?

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
Olivier Mattelaer (olivier-mattelaer) said : 		#1

Hi,

As stated in the MadWidth paper: arXiv:1402.1178 (and as written in the log of the run)
The auto syntax is only valid in Narrow Width Approximation (and LO).
This is clearly not the case when you have the width larger than the mass of the particle.

Now indeed the width is associated to all propagator.
If you want to change that, you can edit the matrix.f to set the width to zero for the particle that you consider as T-channel.
or define a new propagator in the UFO model.

Cheers,

Olivier

On Nov 14, 2014, at 4:11 AM, ParticleCat <email address hidden> wrote:

> New question #257527 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/257527
>
> As I generate the events with an off-shell proceess as t-channel.
> The fermion propagator N have a decay mode N -> W^+ l^- and the width will even greater than the mass of N around M_N>10M_W.
> Compare the simulation result with the analytic formula it seems the MadGraph put the width in the propagator.
> For off-shell propagator with t-channel, the momentum transfer is space-like.
> The width should not contribute to the process.
> However, if I set the width myself but not use the "auto" syntax, will I got a wrong answer in the s-channel ?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
ParticleCat (two-joker) said : 		#2

Thanks, Olivier

I have another 2 questions, one is relative question to the width.
If I want to generate a process with S and T channel and the S channel could be on-shell.
The zero width or small width would affect the cross section of S channel.
Does that means I cannot generate event correctly if I have a large width ?

The 2nd question is : Can we separate the T channel only from total process ?
For example, the process e+ e- > e+ e- through Z.
We can separate the S channel by using e+ e- > z > e+ e-.
Can we separate the T channel by using some syntax or we can only have the non-resonance result by using $z and got a mixture of off-shell S-channel and T-channel data ?

Regards,
Bob

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

Dear ParticleCat,

> I have another 2 questions, one is relative question to the width.
> If I want to generate a process with S and T channel and the S channel could be on-shell.
> The zero width or small width would affect the cross section of S channel.
> Does that means I cannot generate event correctly if I have a large width ?

The presence of a width is related to higher order effect (the width is not a parameter of the Lagrangian).
The idea that a particle has a width is in itself an approximation and the presence of the fix width
introduces gauge and lorentz violation. At leading Order, those effect are typically extremely small and
most of the people ignores those effect. On NLO those effect are typically more important and people
then use other framework (in general the complex mass-scheme). If a particle has a very large width, this
kind of effect need to be study and understand.

> The 2nd question is : Can we separate the T channel only from total process ?
> For example, the process e+ e- > e+ e- through Z.
> We can separate the S channel by using e+ e- > z > e+ e-.
> Can we separate the T channel by using some syntax or we can only have the non-resonance result by using $z and got a mixture of off-shell S-channel and T-channel data ?

You can use $$ for that
e+ e- > e+ e- $$ Z
Note that in general such kind of separation are not physical since you will loose interference (which can affect gauge and lorentz invariance)

Cheers,

Olivier

On Nov 17, 2014, at 7:56 AM, ParticleCat <email address hidden> wrote:

> Question #257527 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/257527
>
> Status: Answered => Open
>
> ParticleCat is still having a problem:
> Thanks, Olivier
>
> I have another 2 questions, one is relative question to the width.
> If I want to generate a process with S and T channel and the S channel could be on-shell.
> The zero width or small width would affect the cross section of S channel.
> Does that means I cannot generate event correctly if I have a large width ?
>
> The 2nd question is : Can we separate the T channel only from total process ?
> For example, the process e+ e- > e+ e- through Z.
> We can separate the S channel by using e+ e- > z > e+ e-.
> Can we separate the T channel by using some syntax or we can only have the non-resonance result by using $z and got a mixture of off-shell S-channel and T-channel data ?
>
> Regards,
> Bob
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
ParticleCat (two-joker) said : 		#4

Thanks Olivier Mattelaer, that solved my question.