# How to generate H+ -> W* A (W off-shell)

Asked by William John Murray on 2021-05-27

I am using MG5_aMC_v3_1_0 & Pythia8 to simulate tt-> WbWH+, H+->WA. When the H+ decay is on shell all is well; MG5 simulates up to the H+ and Pythia8 takes over. But for mH+ < mA + mW I cannot persuade Pythia to run. So I try:

generate p p > t t~, (t~ > b~ w-), (t > b h+, (h+ > h1 f f ))

with f any fermion. I need to remove the (H+ taubar nu) coupling, but OK, that gives the Feynman diagram I want.

However, the kinematics is changed. If we compare the two approaches (Pythia v MG5 for the H+ decay) when using masses allowing on-shell decay, the pT of the A produced by Madgraph is considerably more than that from Pythia. This presumably means the ff mass is reduced. But which is correct?

I try to debug by asking MG5 to simulate a process I know the distribution:
generate p p > Z H , ( Z > p p ) , ( H > Z Z , ( Z > l+ l- ) )
but with or without madspin that typically produces masses for both Z bosons of 60 GeV, not 90/30 you expect. [I have trouble running madspin as it sees the "( Z" in the above as a particle name, and I need to delete all the brackets from the process spec in proc_card_mg5 AFTER mg5_aMC has run. But I think thats not relevant]

Am I doing something wrong in these setups? Any hints much appreciated
Bill

## Question information

Language:
English Edit question
Status:
Solved
For:
Assignee:
No assignee Edit question
Solved by:
William John Murray
Solved:
Last query:
 Revision history for this message Olivier Mattelaer (olivier-mattelaer) said on 2021-05-28: #1

Hi,

>
> However, the kinematics is changed. If we compare the two approaches (Pythia v MG5 for the H+ decay) when using masses allowing on-shell decay, the pT of the A produced by Madgraph is considerably more than that from Pythia. This presumably means the ff mass is reduced. But which is correct?

The role of the parton-shower is to add radiation both from the initial state and the final state.
By doing so the kinematic of the events has to be modified in order to keep conservation of energy-momenta.
So it is normal that some observable are modified by that procedure.
When a particle is onshell, the parton-shower handle such particle in a special way in order to keep his invariant mass on the pole. This is not the case if the particle is offshell. For more details on that (and if you have a way to impact that) you should check parton-shower documentation.

> generate p p > Z H , ( Z > p p ) , ( H > Z Z , ( Z > l+ l- ) )

This syntax does not make sense at madgraph level (wihtin the standard model) since you ask
both the "H" and both "Z" to be onshell. This is why you have weird mass distribution.

Cheers,

Olivier

> On 27 May 2021, at 18:10, William John Murray <email address hidden> wrote:
>
> New question #697280 on MadGraph5_aMC@NLO:
>
> I am using MG5_aMC_v3_1_0 & Pythia8 to simulate tt-> WbWH+, H+->WA. When the H+ decay is on shell all is well; MG5 simulates up to the H+ and Pythia8 takes over. But for mH+ < mA + mW I cannot persuade Pythia to run. So I try:
>
> generate p p > t t~, (t~ > b~ w-), (t > b h+, (h+ > h1 f f ))
>
> with f any fermion. I need to remove the (H+ taubar nu) coupling, but OK, that gives the Feynman diagram I want.
>
> However, the kinematics is changed. If we compare the two approaches (Pythia v MG5 for the H+ decay) when using masses allowing on-shell decay, the pT of the A produced by Madgraph is considerably more than that from Pythia. This presumably means the ff mass is reduced. But which is correct?
>
> I try to debug by asking MG5 to simulate a process I know the distribution:
> generate p p > Z H , ( Z > p p ) , ( H > Z Z , ( Z > l+ l- ) )
> but with or without madspin that typically produces masses for both Z bosons of 60 GeV, not 90/30 you expect. [I have trouble running madspin as it sees the "( Z" in the above as a particle name, and I need to delete all the brackets from the process spec in proc_card_mg5 AFTER mg5_aMC has run. But I think thats not relevant]
>
> Am I doing something wrong in these setups? Any hints much appreciated
> Bill
>
>
>
> --

 Revision history for this message William John Murray (william-john-murray) said on 2021-05-28: #2

Thanks Olivier,
That cleared up what was wrong with the H->ZZ: bad test. H -> llll indeed gives sensible results, and shows me how to use MadGraph to decay bosons!

But I don't think this has answered my real problem: H+ -> WA. You discussed using the parton shower for radiation, but maybe I was not clear I was also using pythia for the H+ decay. Thanks to your hints I have tried a third option:

1) generate p p > t t~, (t~ > b~ w-), (t > b h+, (h+ > h1 f f )) - MG doing all decays bar the h1(==A)
2) generate p p > t t~, (t~ > b~ w-), (t > b h+, (h+ > h1 W+ )) - Pythia decaying the W and h1
3) generate p p > t t~, (t~ > b~ w-), (t > b h+) Pythia decaying H+, W and h1

I am doing this for a test case, mH+=100, mA(==h1) 15, so an onshell W is just possible.

I am new to madanalysis but can plot the pT of the A in all 3 cases as it decays to a mumu resonance. I find pT mean 20 for cases 2 and 3 but 29 for case 1.

Ah...now I begin to suspect the answer. An on-shell W is JUST possible, but typically a full generation will give rise to lower masses? So perhaps the only correct solution is to leave out the intermediate W...as solved the H->ZZ!

Thanks for this....I will test again using higher mH+, so W is more on shell..if results converge we have a solution!
Bill

 Revision history for this message Olivier Mattelaer (olivier-mattelaer) said on 2021-05-28: #3

Your "just" possible might means that Narrow-width approximation is "just" not yet valid.
And that the issue that you have might be some resonance competition and off-shell effects.

Olivier

> On 28 May 2021, at 11:05, William John Murray <email address hidden> wrote:
>
> Question #697280 on MadGraph5_aMC@NLO changed:
>
> William John Murray posted a new comment:
> Thanks Olivier,
> That cleared up what was wrong with the H->ZZ: bad test. H -> llll indeed gives sensible results, and shows me how to use MadGraph to decay bosons!
>
> But I don't think this has answered my real problem: H+ -> WA. You
> discussed using the parton shower for radiation, but maybe I was not
> clear I was also using pythia for the H+ decay. Thanks to your hints I
> have tried a third option:
>
> 1) generate p p > t t~, (t~ > b~ w-), (t > b h+, (h+ > h1 f f )) - MG doing all decays bar the h1(==A)
> 2) generate p p > t t~, (t~ > b~ w-), (t > b h+, (h+ > h1 W+ )) - Pythia decaying the W and h1
> 3) generate p p > t t~, (t~ > b~ w-), (t > b h+) Pythia decaying H+, W and h1
>
> I am doing this for a test case, mH+=100, mA(==h1) 15, so an onshell W
> is just possible.
>
> I am new to madanalysis but can plot the pT of the A in all 3 cases as
> it decays to a mumu resonance. I find pT mean 20 for cases 2 and 3 but
> 29 for case 1.
>
> Ah...now I begin to suspect the answer. An on-shell W is JUST possible, but typically a full generation will give rise to lower masses? So perhaps the only correct solution is to leave out the intermediate W...as solved the H->ZZ!
>
> Thanks for this....I will test again using higher mH+, so W is more on shell..if results converge we have a solution!
> Bill
>
> --