# Singling out diagrams/ large differences in result for small mass differences/ small width in t-channel

Asked by Tania Robens on 2020-09-30

Hi

I am looking at a 2HDM where couplings are determined by electroweak SM parameters, so in principle all results depend only on kinematics. In this model I look at a process at a muon collider, via

mu+ mu- > vm vm~ h+ h-

for charged scalar pair-production and missing ET. Then, I have identified the following major contributions (all from ww fusion)

a) t channel H
b) t channel A
c) wwZ, with Z-> h+h-
d) ww gamma, with gamma -> h+ h-

I look at two model parameter points which have quite similar mass spectra, with the exception of mH. if I generate

a+b only
c+d only
a+b+c
a+b+d

I get similar results, however, if I generate the full process then suddenly there is a factor 6 or so difference... which could be physical but I want to understand it

In total there are 32 diagrams contributing on the same order in couplings. I also checked matrix elements (for one point so far) and found similar values, however, if I turn on the integration something strange happens. The second parameter point has small widths for A , can this be a problem in the t-channel ? Is there a way to debug this in a smart way ? For example, in the Gxx directories there are files called "result.dat", but I am not so sure how to interpret this...

Please get back to me re this ! Thanks Best Tania

## Question information

Language:
English Edit question
Status:
For:
Assignee:
No assignee Edit question
Last query:
2020-09-30
2020-09-30
 Olivier Mattelaer (olivier-mattelaer) said on 2020-09-30: #1

Hi,

In principle for such type of process, you should set all the width to zero to avoid gauge issue related to the width.

Since 2.8.0 all the width of T-channel propagators will be set to zero.
Keeping non zero width in T-channel propagator can often lead to unphysical growing of the amplitude due to a breaking of gauge invariance.
Setting the width to zero only for T-channel is however not a perfect solution and can still lead to gauge issue (and therefore to unphysical growing of the amplitude).
A more correct solution for that problem is of course using the complex-mass scheme which is also supported in MG5aMC.

Now this can also be simply related to an interference term (depending of the direction of your factor 6)
-- which can also be related to gauge cancelation --

Cheers,

Olivier

> On 30 Sep 2020, at 17:11, Tania Robens <email address hidden> wrote:
>
> New question #693172 on MadGraph5_aMC@NLO:
>
> Hi
>
> I am looking at a 2HDM where couplings are determined by electroweak SM parameters, so in principle all results depend only on kinematics. In this model I look at a process at a muon collider, via
>
> mu+ mu- > vm vm~ h+ h-
>
> for charged scalar pair-production and missing ET. Then, I have identified the following major contributions (all from ww fusion)
>
> a) t channel H
> b) t channel A
> c) wwZ, with Z-> h+h-
> d) ww gamma, with gamma -> h+ h-
>
> I look at two model parameter points which have quite similar mass spectra, with the exception of mH. if I generate
>
> a+b only
> c+d only
> a+b+c
> a+b+d
>
> I get similar results, however, if I generate the full process then suddenly there is a factor 6 or so difference... which could be physical but I want to understand it
>
> In total there are 32 diagrams contributing on the same order in couplings. I also checked matrix elements (for one point so far) and found similar values, however, if I turn on the integration something strange happens. The second parameter point has small widths for A , can this be a problem in the t-channel ? Is there a way to debug this in a smart way ? For example, in the Gxx directories there are files called "result.dat", but I am not so sure how to interpret this...
>
> Please get back to me re this ! Thanks Best Tania
>
> --