Changing BW cutoff to include off shell W-bosons

Asked by Saptaparna Bhattacharya

Dear Olivier,

I am trying to produce a sample of events containing tribosons, specifically WWW. I would like to include the Higgs mediated mode of WWW production (WH, H -> WW). For this, I need to change the BW cutoff from the standard value of 15 to 40 or so. A plot of the variation of the BW cutoff showing the gradual inclusion of the off-shell W’s from the Higgs mediated mode can be seen here: http://nuhep.northwestern.edu/~sapta/ForOlivier/mass_elnu_BW_Madgraphonly.pdf

As the BW cutoff is increased the cross section also increases and I obtain a value of the cross section consistent with the LO prediction (the k-factor is large ~1.9, so within some tolerance the cross section also makes sense). However, I encounter a problem when I the set the BW cutoff to 40. For these values, the rate of event generation is slow and Madgraph produces 73 events when the requested number of events is 10k. For BW cutoff < 40, Madgraph produces 10k events as requested. I have also tried producing a gridpack and after several tries, I get higher numbers of events than regular event production but the rate of event generation remains very slow. I used MG5 2.6.7. Is there a way to solve this problem? I think for this purpose, the ability to set BW cutoff to 40 or a slightly higher value (not more than 50 based on the W width) should be sufficient.

Looking forward to your feedback.
Thanks so much!
Sapta

Question information

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

Dear Sapta,

I do not think that increasing Bwcut makes sense here.
With your syntax, MG5aMC assumes that NWA is valid within tthe Bwcut range.
That assumption is actually heavily used to optimized the phase-space integrator in order to be able to integrate correctly the phase-space.

Looking at your plot, your generation syntax should not be used with a Bw-cut off larger than 10 to 15.
Larger value will decrease the speed efficiency as you have realised but can also lead to bias of the integral.

My advise would be to use the following syntax:
generate p p > w+ w+ w- $h , w- > e- ve~, w+ > mu+ vm
add process p p > w+ h, h > e- ve~ mu+ vm, w+ > mu+ vm

and check with both bwcutoff to 10 and 15 (15 is likely too big for that syntax)

Cheers,

Olivier

> On 22 Mar 2021, at 15:30, Saptaparna Bhattacharya <email address hidden> wrote:
>
> New question #696189 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/696189
>
> Dear Olivier,
>
> I am trying to produce a sample of events containing tribosons, specifically WWW. I would like to include the Higgs mediated mode of WWW production (WH, H -> WW). For this, I need to change the BW cutoff from the standard value of 15 to 40 or so. A plot of the variation of the BW cutoff showing the gradual inclusion of the off-shell W’s from the Higgs mediated mode can be seen here: http://nuhep.northwestern.edu/~sapta/ForOlivier/mass_elnu_BW_Madgraphonly.pdf
>
> As the BW cutoff is increased the cross section also increases and I obtain a value of the cross section consistent with the LO prediction (the k-factor is large ~1.9, so within some tolerance the cross section also makes sense). However, I encounter a problem when I the set the BW cutoff to 40. For these values, the rate of event generation is slow and Madgraph produces 73 events when the requested number of events is 10k. For BW cutoff < 40, Madgraph produces 10k events as requested. I have also tried producing a gridpack and after several tries, I get higher numbers of events than regular event production but the rate of event generation remains very slow. I used MG5 2.6.7. Is there a way to solve this problem? I think for this purpose, the ability to set BW cutoff to 40 or a slightly higher value (not more than 50 based on the W width) should be sufficient.
>
> Looking forward to your feedback.
> Thanks so much!
> Sapta
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
Saptaparna Bhattacharya (saptaparna) said :
#2

Dear Olivier,

Thank you very much for your response! Quick follow up question from our analysis team: would we be taking the interference between these two processes into account if we split them?

Thanks a lot!
Sapta

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

Yes and no,

In the offshell region, you do have the interference term.
In the onshell region, you do not have the interference term.

So you can see the impact of the interference when you pass from one regime to the other (and this is why you have to check the value of BWcut)

Cheers,

Olivier

> On 22 Mar 2021, at 18:45, Saptaparna Bhattacharya <email address hidden> wrote:
>
> Question #696189 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/696189
>
> Status: Answered => Open
>
> Saptaparna Bhattacharya is still having a problem:
> Dear Olivier,
>
> Thank you very much for your response! Quick follow up question from our
> analysis team: would we be taking the interference between these two
> processes into account if we split them?
>
> Thanks a lot!
> Sapta
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
Saptaparna Bhattacharya (saptaparna) said :
#4

Hi Olivier,

One more follow-up question since we are trying to understand the interference between the purely Higgs mediated and non-Higgs mediated modes of WWW production: does the syntax you quote above treat the two processes as uncorrelated? In other words, when one adds processes using the "add process" syntax, are the processes treated by Madgraph as independent processes?

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#5

Hi,

Remember that the "$ h " syntax is a special phase-space cut forbidding the Higgs to be onshell, while " h , h > "
has the exact opposite cut forcing the Higgs to be onshell.

So indeed they are NO interference term between the various generate/add process.
This is why you do not have such intereference term in the on-shell region.

In the offshell region, the second line "h , h >" does not contribute dude to the phase-space cut.
and there you have only the first line:
generate p p > w+ w+ w- $h , w- > e- ve~, w+ > mu+ vm
which also contains H* > W W
including the interference.

In the onshell region, a complex cut is in place to remove the "Higgs" onshell contribution from the first syntax while keeping other contribution. But here the second process kind of compensate by the adding the pure higgs onshell contribution, but the intereference is not included.

If you want to include such interference term, then you likely need to use the following syntax
generate p p > 2e- 2ve~ mu+ vm
and using complex mass-scheme might be appropriate here
import model sm
set complex_mass_scheme True
generate p p > 2e- 2ve~ mu+ vm

Obviously this syntax will be much slower but more accurate.

Can you help with this problem?

Provide an answer of your own, or ask Saptaparna Bhattacharya for more information if necessary.

To post a message you must log in.