About matching in MG5

Asked by vaborla

Hello

       I caculated the cross section of process pp>zj, z>l+l- in parton level (ickkw=0,xqcut=0). Then I want to perform detector simulation with pythia-pgs/pythis delphes. I find these questions:

     1. I write process pp>zj,z>l+l- in proc_card.dat. Then I set ickkw=1,xqcut=15 in run_card.dat. There are other parameters at default in pythia_card.dat and pgs_card.dat. When the MG5 is performed, the error appeared:

       Command "generate_events " interrupted in sub-command:
       "generate_events" with error:
        ZeroDivisionError : float division by zero

    2. I set ickkw=0 xqcut=15. I found that the cross section is not equal the result with xqcut=0.

     I have read matching lectures in MG school, but I am still confused about matching. I just want to perform detector simulation for the process pp>zj,z>l+l-. What should I do about matching?
     Thank you very much!

Cheers
Violin

Question information

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

This question was reopened

Revision history for this message
Johan Alwall (johan-alwall) said :
#1

Hello Violin,

In principle you are right, that you should always include matching whenever you have a process with radiated jets (as is the case for p p > z j). Note however that your xqcut must be lower than the pt cut you plan to use on the first jet in your analysis.

I am not able to reproduce the error you describe. Can you please send all the cards in your Cards directory (or, equivalently, the banner of the run if it has been created) to my email address jalwall_at_ntu.edu.tw?

You do not expect the cross section to be the same if you have ickkw=0, since in that case you have no alpha_s reweighting. Please see our wiki pages about matching for more explanations:
https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/IntroMatching
https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/Matching

All the best,
Johan

Revision history for this message
vaborla (vaborla-tt) said :
#2

Hello Johan

         I have send my cards to your email.
         Thank you very much!

Cheer
Violin

Revision history for this message
karo (karo-t) said :
#3

Hello Johan

        I have changed etaj to 1d5, the Pythia run is successful. Now I have another question. I want to simulate pp>zj, z>l+l-, then before detector simulation I must run pythia. In order to match well, I should set the process in proc_card_mg5.dat like that:

     generate p p > Z j, Z > l+ l- @0
     add process p p > Z j j, Z > l+ l- @1
     add process p p > Z j j j, Z > l+ l- @2

or just like

      generate p p > Z j, Z > l+ l- @0

Cheers
Violin

Revision history for this message
Johan Alwall (johan-alwall) said :
#4

Hello Violin,

How many matrix elements you need to include depends partly on the process, and partly on the analysis you want to do. For p p > W/Z production, you should always include at least the 2-jet multiplicity, since there are enhancements at large momenta when the W/Z is emitted collinear to an outgoing jet (which only show up in the 2-jet multiplicity and higher). If your analysis is only sensitive to the kinematics of the first (and perhaps second) jet, I would suggest to include only the 1- and 2-jet multiplicities, since this will save you quite a lot of simulation time. There is nothing wrong with including a third jet as well, just takes longer time and is unnecessary unless your analysis is sensitive to the momenta of the third jet (e.g., if you are sensitive to very large H_T).

Also note that if you are sensitive to the lepton kinematics away from the Z peak, you should include interference with the photon, i.e.,

p p > l+ l- j @1
p p > l+ l- j j @2
etc.

All the best,
Johan

Revision history for this message
vaborla (vaborla-tt) said :
#6

Hello Johan

          Why the delphes_events.lhco doesn't give matched cross section, that events file only shows the cross section before matched?

Cheers
Violin

Revision history for this message
Johan Alwall (johan-alwall) said :
#7

Hello Violin,

This is a very good point, we will add that number.

In the meantime, you can use the unmatched cross section corresponds to the cross section, multiply by the number of events in the Delphes file, and divide by the number of events before matching. This can be deduced directly from the Delphes file, since each event is numbered, so you can see how many events that failed the matching by just looking at the event numbering.

Olivier, I think you added the line
## Integrated weight (pb) : 73213.1
to the Delphes/PGS LHCO header. Could you also add a corresponding line for the matched cross section (in v. 1.5.9)?

Thanks a lot,
Johan

Revision history for this message
vaborla (vaborla-tt) said :
#8

Hello Johan

         I must use the kinematics distribution of events after detector simulation. So I put the etmiss cut > 0.0001 when I deal with the Delphes_events.lhco. Then I get the correct cross section after simulation. But I don't know whether I am right to do that.

        I refer the question website
        https://answers.launchpad.net/madgraph5/+question/200336

Cheers
Violin

Revision history for this message
Johan Alwall (johan-alwall) said :
#9

Dear Violin,

That is an excellent way to do it, since it removes any events that are empty (i.e., that have been input as placeholders only).

All the best,
Johan

Revision history for this message
vaborla (vaborla-tt) said :
#10

Thanks Johan Alwall, that solved my question.

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#11

Hi Johan,

> Olivier, I think you added the line
> ## Integrated weight (pb) : 73213.1
> to the Delphes/PGS LHCO header. Could you also add a corresponding line for the matched cross section (in v. 1.5.9)?

I don't remember that I did. (or at least not on purpose when passing on 1.4.x)
So I have added the line:
## Matched Integrated weight (pb) : 29490.0
both in the pythia banner and in the pgs/Delphes banner.

Cheers,

Olivier

On Mar 25, 2013, at 5:55 AM, Johan Alwall <email address hidden> wrote:

> Question #224066 on MadGraph5 changed:
> https://answers.launchpad.net/madgraph5/+question/224066
>
> Status: Open => Answered
>
> Johan Alwall proposed the following answer:
> Hello Violin,
>
> This is a very good point, we will add that number.
>
> In the meantime, you can use the unmatched cross section corresponds to
> the cross section, multiply by the number of events in the Delphes file,
> and divide by the number of events before matching. This can be deduced
> directly from the Delphes file, since each event is numbered, so you can
> see how many events that failed the matching by just looking at the
> event numbering.
>
> Olivier, I think you added the line
> ## Integrated weight (pb) : 73213.1
> to the Delphes/PGS LHCO header. Could you also add a corresponding line for the matched cross section (in v. 1.5.9)?
>
> Thanks a lot,
> Johan
>
> --
> You received this question notification because you are a direct
> subscriber of the question.

Revision history for this message
vaborla (vaborla-tt) said :
#12

Hi Johan,

     I have another question about matching. If I calcuate Z+ j exclusively ,not inclusively, I also should add
     pp > z j
     pp > z j j
     p p > z j j j
     in run_card.dat ?

Thanks.
Violin

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#13

Hi Violin,

Sorry for the delay on this answer. Neither me or Johan were available last week for answering questions.

No you don't have to include those in the proc_card. But you still have to use the matching and
force Pythia to not radiate any jets above the xqcut threshold.
You will find the information on the wiki page associated to the matching.

Sorry to not include the exact options to include in the pythia card, neither to put a direct link to the wiki.
But I'm writing this email whiteout internet connection. Don't hesitate to ask me if you don't find the options in the wiki pages.

Cheers,

Olivier

On Apr 16, 2013, at 8:31 AM, violin <email address hidden> wrote:

> Question #224066 on MadGraph5 changed:
> https://answers.launchpad.net/madgraph5/+question/224066
>
> Status: Solved => Open
>
> violin is still having a problem:
> Hi Johan,
>
> I have another question about matching. If I calcuate Z+ j exclusively ,not inclusively, I also should add
> pp > z j
> pp > z j j
> p p > z j j j
> in run_card.dat ?
>
>
> Thanks.
> Violin
>
> --
> You received this question notification because you are a direct
> subscriber of the question.

Can you help with this problem?

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

To post a message you must log in.