Single top t-channel process matching question

Asked by Kelci Mohrman on 2020-06-26

Hello MG experts,

I have a question about using matching for a LO single top t-channel sample. We are using the 5f scheme. In our process card, we have:

    generate p p > t j l+ l- $$ w+ w- / h
    add process p p > t~ j l+ l- $$ w+ w- / h
    add process p p > t j j l+ l- $$ w+ w- / h
    add process p p > t~ j j l+ l- $$ w+ w- / h

In our run card, we have set auto_ptj_mjj to False, and have set mmjj and ptj to be smaller than our xqcut (as suggested in Olivier's slides from the June 16, 2020 CMS TOP group meeting).

For the pythia jet matching parameters, we have nJet at its default value of -1, and nJetMax set to 1.

However, this setup does not seem to produce a proper matched sample. For example, when we look at the DJR plots, there is no 0p line (and only a 1p line). We think this indicates that the "j" in "p p > t j l+ l-" is being treated as an extra parton (and matching is being applied to it). Also, when we change nJetMax to be 2, the DJR plots additionally contain a 2p line. This seems to indicate that that the "p p > t j j l+ l-" process is being treated as a process with two additional partons, and all of the events produced by the third and fourth rows in the process card (the processes with an additional jet) are being vetoed when we set nJetMax=1.

We are wondering how we can set up MG and pythia so that matching will be handled properly in this situation.

Thanks very much! - Kelci

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Last query:
2020-06-26
Last reply:
2020-06-26

hi,

I would suggest to double check that to pythia author.
But I would have set
nJet to 1
and
nJetMax to 2

But please double check with pythia author

Cheers,

Olivier

> On 26 Jun 2020, at 02:01, Kelci Mohrman <email address hidden> wrote:
>
> New question #691541 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/691541
>
> Hello MG experts,
>
> I have a question about using matching for a LO single top t-channel sample. We are using the 5f scheme. In our process card, we have:
>
> generate p p > t j l+ l- $$ w+ w- / h
> add process p p > t~ j l+ l- $$ w+ w- / h
> add process p p > t j j l+ l- $$ w+ w- / h
> add process p p > t~ j j l+ l- $$ w+ w- / h
>
> In our run card, we have set auto_ptj_mjj to False, and have set mmjj and ptj to be smaller than our xqcut (as suggested in Olivier's slides from the June 16, 2020 CMS TOP group meeting).
>
> For the pythia jet matching parameters, we have nJet at its default value of -1, and nJetMax set to 1.
>
> However, this setup does not seem to produce a proper matched sample. For example, when we look at the DJR plots, there is no 0p line (and only a 1p line). We think this indicates that the "j" in "p p > t j l+ l-" is being treated as an extra parton (and matching is being applied to it). Also, when we change nJetMax to be 2, the DJR plots additionally contain a 2p line. This seems to indicate that that the "p p > t j j l+ l-" process is being treated as a process with two additional partons, and all of the events produced by the third and fourth rows in the process card (the processes with an additional jet) are being vetoed when we set nJetMax=1.
>
> We are wondering how we can set up MG and pythia so that matching will be handled properly in this situation.
>
> Thanks very much! - Kelci
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Kelci Mohrman (kmohrman) said : #2

Hi Olivier,

Thank you for the suggestion. We tried setting nJet=1, nJetMax=2 as you suggested, but we still see the same issue. The DJR plots contain a 1p line and a 2p line, but no 0p line; also, we see a cutoff in the DJR plots at the qCut (for both the 1p line and 2p line). We believe these issues indicate that the matching is not being handled properly.

We had previously discussed the issue of single top t-channel matching with Steve Mrenna (from Pythia), and he had said that he thought it is not actually possible for MG/Pythia to handle the matching correctly in this situation. After trying all of the settings you have suggested, it seems that this is indeed the case.

We appreciate your help though, and we would like to say thanks again for helping us to look into this! - Kelci

Hi,

MLM was certainly working with Pythia6 for that specific process, I doubt that Pythia8 author drop that support.

Now I would not worry too much about the name of the curve, the important is to plot the DJR plot for the different multiplicity and see how the DJR behaves this should tell you if it is working as expected or not.

Cheers,

Olivier

> On 26 Jun 2020, at 18:25, Kelci Mohrman <email address hidden> wrote:
>
> Question #691541 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/691541
>
> Kelci Mohrman posted a new comment:
> Hi Olivier,
>
> Thank you for the suggestion. We tried setting nJet=1, nJetMax=2 as you
> suggested, but we still see the same issue. The DJR plots contain a 1p
> line and a 2p line, but no 0p line; also, we see a cutoff in the DJR
> plots at the qCut (for both the 1p line and 2p line). We believe these
> issues indicate that the matching is not being handled properly.
>
> We had previously discussed the issue of single top t-channel matching
> with Steve Mrenna (from Pythia), and he had said that he thought it is
> not actually possible for MG/Pythia to handle the matching correctly in
> this situation. After trying all of the settings you have suggested, it
> seems that this is indeed the case.
>
> We appreciate your help though, and we would like to say thanks again
> for helping us to look into this! - Kelci
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Kelci Mohrman (kmohrman) said : #4

Yes, we agree that the names of the lines are not so important; what mainly concerned us is that both the 1p line and 2p line are getting cut off at the value of the qCut. Also, when we discussed this with Steve Mrenna, he mentioned that he thought the UMEPS matching scheme would be needed in order to handle matching correctly in this situation, so we are not sure that MLM will work in this case.

Thanks! - Kelci

Can you help with this problem?

Provide an answer of your own, or ask Kelci Mohrman for more information if necessary.

To post a message you must log in.