Add a new model?

Asked by Andreas Kirchhoff on 2018-12-03

Dear MadGraph-authors,

for my Master's thesis I look at the ttgamma process. Unfortunately there is no possibility to unambiguously separate events where the top quark radiates the photon from ISR or b-quarks radiating the photon. To archive this I did a few manipulations (I have some slides describing the changes but I do not see a possibility here to upload them). These manipulations , although neglecting interference terms, did not change the overall distributions as I checked (could be seen in my slides as well). The distributions were later used to feed a neural network to distinguish between events where the photon comes from the top quark and where it comes form ISR, the b quark or the W boson and its decay products.
The question that I was asked now is, whether it makes sense to implement those models (although neglecting interference terms) to make these differentiations in general possible (for example also for W or Z radiation). A possible usage of such models is probing the coupling to the top quark because therefore only events where the top quark radiates the interesting particle are relevant and it is necessary to filter them wrt to ISR, b quarks and W bosons radiating such particles, for instance the photon.
What do you think about this?

Cheers,
Andreas

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Last query:
2018-12-03
Last reply:
19 hours ago

I should add that the manipulations I did, concern the top-photon vertex. Basically I renamed the top-photon coupling but kept the QED properties.

Hi Andreas,

I'm a bit confused by your question.

If by model, you are talking about UFO model, then you can give to me the path to their tarball (somewhere online were you are responsible for those to keep them up-to-date) and I will include those in our online model database.
I would advise to include the __arxiv__ flag in the __init__.py and to include a README in your model in order to explain what you did inside those. This being said making such incomplete model public might be quite dangerous and I would need to refer to my boss if we want to include such type of model.

If this is not what you have in mind, please explain.

Cheers,

Olivier

Hi Olivier,
thanks for your reply. I put everything I have to this git repository: https://gitlab.cern.ch/ankirchh/madgraph_ttgamma

It is basically the normal MadGraph standalone code one can download, but I added a pdf/-file with slides explaining why and what I did. I hope that helps to understand what I did and what my question is about.

If not, I would be glad to explain it in more detail to you.

Cheers,
Andreas

Hi Andreas,

Note that we can compute interference term as well
something like
generate p p > t t~ a TGA^2==1
will compute the interference between ISR and top radiation
Note that such syntax is not compatible with decay-chain syntax.
but you can do
generate p p > t t~ w+ w- > b b~ l+ l- vl vl~ TGA^2==1

For the question about including your model inside our online model database.
I do not see any real objection. Just give me a html link of your model (as .tar.gz or .zip)
and when user will do import model YOUR_MODEL_NAME
they will download your model.

Cheers,

Olivier

Hi Olivier,

here is the Link to the .tar.gz file: https://www.dropbox.com/s/ddyl4w4898l3fwm/MadGraph_ttgamma.tar.gz?dl=0

With respect to the interference terms I have a follow up question:
If I do
generate p p > t t~ a TGA^2=1 QED^2=1
I get the interference terms I want.
If I than do two further productions
define q = u d c s u~ d~ c~ s~
generate q q > t t~ a TGA==1 QCD==2
and
define q = u d c s u~ d~ c~ s~
generate q q > t t~ a TGA==0 QCD==2
to get the non-interfering contributions and than add up the cross sections I get from MadGraph they do not add up to the cross section I get if i generate
define q = u d c s u~ d~ c~ s~
generate q q > t t~ a QCD==2

Why is that the case?
The cross section values are (in the order from above):
-0.0008833
0.1444
0.53017
0.6683

Cheers,
Andreas

Hi,

This is typically due to (and within) systematical error.

Indeed the default choice for the factorization scale is Feynman diagram dependent and does not make that much sense when you are dominated by interference.
So if you change the dynamical scale choice (let's say to "3", i.e. HT/2), then you should get an agreement up to statistical uncertainty.

Cheers,

Olivier

> On 6 Dec 2018, at 15:07, Andreas Kirchhoff <email address hidden> wrote:
>
> Question #676558 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/676558
>
> Andreas Kirchhoff posted a new comment:
> Hi Olivier,
>
> here is the Link to the .tar.gz file:
> https://www.dropbox.com/s/ddyl4w4898l3fwm/MadGraph_ttgamma.tar.gz?dl=0
>
> With respect to the interference terms I have a follow up question:
> If I do
> generate p p > t t~ a TGA^2=1 QED^2=1
> I get the interference terms I want.
> If I than do two further productions
> define q = u d c s u~ d~ c~ s~
> generate q q > t t~ a TGA==1 QCD==2
> and
> define q = u d c s u~ d~ c~ s~
> generate q q > t t~ a TGA==0 QCD==2
> to get the non-interfering contributions and than add up the cross sections I get from MadGraph they do not add up to the cross section I get if i generate
> define q = u d c s u~ d~ c~ s~
> generate q q > t t~ a QCD==2
>
> Why is that the case?
> The cross section values are (in the order from above):
> -0.0008833
> 0.1444
> 0.53017
> 0.6683
>
> Cheers,
> Andreas
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi,

You provide a link to the full version of MG5aMC (with compiled file...)c. This is not what I was expecting.
What I would be fine to include in our database is your UFO model.

Such that if a user do
import model SM_with TTA_separated

it will automatically download your UFO model and can use the
generate command with the TGA command

Cheers,

Olivier

> On 6 Dec 2018, at 15:12, Andreas Kirchhoff <email address hidden> wrote:
>
> Question #676558 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/676558
>
> Andreas Kirchhoff posted a new comment:
> Sorry, here is the correct link:
> https://www.dropbox.com/s/e44vdn54ufowyuf/MadGraph_ttgamma.tar.gz?dl=0
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hey,

first: Thank you very much for the explanation corresponding the interference calculations. I retried it with dynamic scale 3 and systematics. The results are much better now.

I have to admit that I do not now what it means to include an UFO model. Do you just want the model directory in which I changed the files or just the modified files?
I am sorry that this is unclear to me!

Cheers,
Andreas

Hi,

Here is the description of the UFO model:
https://arxiv.org/abs/1108.2040
In two word, this is how a (B)SM model is coded within the HEP community.

But if you do not know what a UFO model is. Then I'm likely not understanding what you want me to do for you.

Cheers,

Olivier

On 6 Dec 2018, at 15:53, Andreas Kirchhoff <<email address hidden><mailto:<email address hidden>>> wrote:

Question #676558 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/676558

Andreas Kirchhoff posted a new comment:
Hey,

first: Thank you very much for the explanation corresponding the
interference calculations. I retried it with dynamic scale 3 and
systematics. The results are much better now.

I have to admit that I do not now what it means to include an UFO model. Do you just want the model directory in which I changed the files or just the modified files?
I am sorry that this is unclear to me!

Cheers,
Andreas

--
You received this question notification because you are an answer
contact for MadGraph5_aMC@NLO.

Hi,

I applied the changes to MadGraph directly in the sm directory of the downloaded code. Therefore I never had to build a UFO-model and I did not know how this works. I think the paper helped me a lot. Thanks for that.
I have taken now just the python files (and some others) and uploaded them again.
https://www.dropbox.com/s/uix127hrwgxm6bs/ttgamma_separated.tar.gz?dl=0
I was not sure if you need the .dat files , so I included them. I hope that is what you need, right?

Cheers,
Andreas

Yes in term of format this is good.
Now looks like Dropbox does not allow script to download your file.
and therefore the functionality does not work.
Could you put your model in your homepage of your home institute website?

Cheers,

Olivier

I have no access to the institutes web page, but I hope that this git link works:
https://gitlab.cern.ch/ankirchh/ttgamma_separated

Cheers,
Andreas

Hi,

Yes that's perfect. That link was accepted by our validation script.
Our server will be updated in less than an hour with that link.

Cheers,

Olivier

Hi Olivier,

thanks a lot for your support. I apologize for any inconviniece due to my lack of knowledge at some points.

Cheers,
Andreas

Hi Olivier,

excuse me that I have to ask you again, but where do I actually find the model now. When looking into the database of models that I am aware of (https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/Models) I could not find it. Could you please give me the link, so I can share it with other people? Thanks a lot.

Cheers,
Andreas

Hi,

You do not have anything to share with them.
just ask them to type "import model ttgamma_separated"
and they will import your model on their machine and load it.

Cheers,

Olivier

Can you help with this problem?

Provide an answer of your own, or ask Andreas Kirchhoff for more information if necessary.

To post a message you must log in.