Loop induced gg->llqq

Hi Lailin,

If you want 1+2+3, you can do it in two steps:

a) generate p p > e+ e- j j [QCD]

Will give you 3) and 2) *as well as NLO corrections* to 2), which is why you also get the real emission diagrams with initial state quarks.
To convince you that this indeed contains the Loop-induced tree interference, you can type 'display diagrams loop' and you will see that the desired loop-induced diagrams are there.

b) Then add the contribution from g g > h > z z > l l q q~. This will give you the pure loop-induced contribution. 1).

If you don't want any NLO correction, and avoid combining an NLO run (step a) with an LO one (step b), then you will have to compute 1) and 2) separately as you have done already, but also the interference term 3) independently.

There is no mainstream way of computing *3) only* in MG5aMC. I have ways of doing so with inelegant but functional hacks, which I refrain from sharing here as I they can be error prone for non-developers.

Maybe Eleni can tell you more.

Hi Valentin,

Thanks for your reply.
For a) you mentioned, by doing 'display diagrams loop' I only see loop induced diagrams with QCD=4, QED=2, while
I don't see diagrams with QCD=2, QED=4. Do you know why they (QCD=2, QED=4) are not included?

Thanks,
Lailin

Hi Lailin,

to get "3) only" you can do:

import model loop_sm
generate g g > e+ e- u u~ /h QED^2==8 QCD^2==4 [noborn=QCD]
output loop_background

you can check if that gives you the right diagrams with display diagrams. You should get the gg>VV>llqq boxes and also the triangles in figure 4b of http://arxiv.org/pdf/1506.01694.pdf which should have a zero contribution. (there are ways to remove those diagrams with some inelegant hacks as Valentin says)

Let me know if it works.

Cheers,

Eleni

Hi Eleni,

Thanks for your help and indeed that works to get "3) only"!
Also could you let me know how to get 2)+3) (background only) and 1)+2)+3 (signal +background) ?

Thanks,
Lailin

Hi Eleni,

Do you think the following would work to generate signal+background?

generate g g > h > e+ e- u u~ [QCD]
add process g g > e+ e- u u~
add process g g > e+ e- u u~ /h QED^2==8 QCD^2==4 [noborn=QCD]

I tried this but it crashed after the command "output":

"output JOBDIR" with error:
AttributeError : 'HelasMatrixElement' object has no attribute 'compute_all_analytic_information'

Do you have any suggestions?

Thanks,
Lailin

Hi,

this will not give you any interference between any of the three contributions, just the sum of each of them squared. And I think it crashes because you are trying to generate tree level and loop-induced processes together.

if you do g g > e+ e- u u~ QED^2==8 QCD^2==4 [noborn=QCD] you will get the signal and the loop background together (including their interference).

The thing which is non-trivial to do is the interference between the tree background and the signal. The way we did it in http://arxiv.org/pdf/1506.01694.pdf was to implement the ggh form factor. I can send you the model and command by email if you want.

Cheers,

Eleni

Hi Eleni,

Thanks for your information.
Yes, I'm interested in the interference between the tree background and the signal.
It would be great if you can send me the model and command to do the study.

Thanks,
Lailin

Hi Lailin,

Unless I misunderstood, what you want, I *don't* think that Eleni's
suggested syntax, namely:

generate g g > e+ e- u u~ /h QED^2==8 QCD^2==4 [noborn=QCD]

will give you what you want, As it would still be part of the loop-induced
signal 1).
If what you need, 3), is indeed the interference of the tree background
with the loop contribution, then that's not it.
Like I said, the only way to get it would then to take it as part of the
NLO contribution of g g > e+ e- u u~ or with some minimal but not
completely trivial hack of MG5_aMC.

Cheers,

On Fri, Apr 1, 2016 at 4:27 AM, Lailin Xu <
<email address hidden>> wrote:

> Question #289697 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/289697
>
> Status: Answered => Solved
>
> Lailin Xu confirmed that the question is solved:
> Hi Eleni,
>
> Thanks for your information.
> Yes, I'm interested in the interference between the tree background and
> the signal.
> It would be great if you can send me the model and command to do the study.
>
> Thanks,
> Lailin
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.
>

--
Valentin

Hi Valentine,

Thanks for your comment.

On Sat, Apr 2, 2016 at 1:07 AM, Valentin Hirschi <
<email address hidden>> wrote:

> Your question #289697 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/289697
>
> Valentin Hirschi posted a new comment:
> Hi Lailin,
>
> Unless I misunderstood, what you want, I *don't* think that Eleni's
> suggested syntax, namely:
>
> generate g g > e+ e- u u~ /h QED^2==8 QCD^2==4 [noborn=QCD]
>
> will give you what you want, As it would still be part of the loop-induced
> signal 1).
>

But this syntax removes the Higgs diagrams, so they should not be part of
the loop-induced signal 1), right?
I checked the diagrams from this syntax, and they are indeed the
loop-induced background:
http://lxu.web.cern.ch/lxu/public/MG/diagrams_0_gg_epemuux_no_h.eps

> If what you need, 3), is indeed the interference of the tree background
> with the loop contribution, then that's not it.
> Like I said, the only way to get it would then to take it as part of the
> NLO contribution of g g > e+ e- u u~ or with some minimal but not
> completely trivial hack of MG5_aMC.
>

Ideally I'd like to have signal only 1) and background only 2)+3) and
signal+background 1)+2)+3).
From what I understand from Eleni, it's easy to get the interference of the
signal and loop-induced background,
but it's not trivial to get the interference of the signal and the tree
background.
As I mentioned, by using g g > e+ e- u u~ [QCD], I only see loop induced
diagrams with QCD=4, QED=2, while
I don't see diagrams with QCD=2, QED=4, which I don't understand yet.

Thank you!

>
> Cheers,
>
>
Regards,
Lailin

>
> On Fri, Apr 1, 2016 at 4:27 AM, Lailin Xu <
> <email address hidden>> wrote:
>
> > Question #289697 on MadGraph5_aMC@NLO changed:
> > https://answers.launchpad.net/mg5amcnlo/+question/289697
> >
> > Status: Answered => Solved
> >
> > Lailin Xu confirmed that the question is solved:
> > Hi Eleni,
> >
> > Thanks for your information.
> > Yes, I'm interested in the interference between the tree background and
> > the signal.
> > It would be great if you can send me the model and command to do the
> study.
> >
> > Thanks,
> > Lailin
> >
> > --
> > You received this question notification because you are an answer
> > contact for MadGraph5_aMC@NLO.
> >
>
>
> --
> Valentin
>
> You received this question notification because you asked the question.
>

Hi Valentin and Lailin,

I took 3) to be the loop background i.e. without the Higgs, that is why I suggested

generate g g > e+ e- u u~ /h QED^2==8 QCD^2==4 [noborn=QCD]

then generate g g > e+ e- u u~ QED^2==8 QCD^2==4 [noborn=QCD] gives the signal, the loop induced background and their interference. So subtracting the signal and the background (obtained with the /h) one can get the interference between signal and loop background.

 g g > e+ e- u u~ [QCD] gives you the QCD corrections to the tree level background. The signal is actually an "electroweak correction" to the tree background so you can't get it like that.

Does this make sense?

Eleni