# Matching and Merging in madgraph

Asked by prashant sundaram on 2020-08-11

Hi ,

I have generated the following events in madgraph by importing the UFO model :-https://gitlab.com/hepcedar/contur/-/tree/master/Models/DM/DM_vector_mediator_UFO

y1 is the vector Dark matter mediator in the model and xm is the Majorana DM candidate

The following are my commands

generate p p > y1 , y1 > u u~
add process p p > y1 , y1 > d d~
add process p p > y1 , y1 > xm xm
add process p p > y1 j , y1 > u u~
add process p p > y1 j , y1 > d d~
add process p p > y1 j , y1 > xm xm
add process p p > y1 w+ , y1 > u u~
add process p p > y1 w+ , y1 > d d~
add process p p > y1 w+ , y1 > xm xm
add process p p > y1 z , y1 > u u~
add process p p > y1 z , y1 > d d~
add process p p > y1 z , y1 > xm xm
add process p p > y1 a , y1 > u u~
add process p p > y1 a , y1 > d d~
add process p p > y1 a , y1 > xm xm

I launch the process and use Pythia 8 to shower.As i have processes with a hard jet ,I perform MLM matching by setting ickkw=1 in the run_card.dat file . The cross-section that i get is 13.02pb , If i set ickkw=0 (i.e. no Matching) I expect that my cross-section should be larger than the case when ickkw=1 , because my jets should be double counted, but my cross-section turns out to be 12.27pb in this case which is confusing.So my question is why is the cross-section smaller for the unmatched case as compared to the matched case?

Note: I used the default xqcut = 30.0 .I use madgraph version MG5_aMC_v2_6_7

## Question information

Language:
English Edit question
Status:
For:
Assignee:
No assignee Edit question
Last query:
2020-08-11
2020-08-14
 Olivier Mattelaer (olivier-mattelaer) said on 2020-08-11: #1

Hi,

I would not include the process with:
add process p p > y1 w+ , y1 > u u~
add process p p > y1 w+ , y1 > d d~
add process p p > y1 w+ , y1 > xm xm
add process p p > y1 z , y1 > u u~
add process p p > y1 z , y1 > d d~
add process p p > y1 z , y1 > xm xm
add process p p > y1 a , y1 > u u~
add process p p > y1 a , y1 > d d~
add process p p > y1 a , y1 > xm xm

in this computation since they are not QCD related.
Now for the rest, I would expect that this can be scale related what is the theoretical uncertainty on those numbers?
Did you set ptj and mmjj to 30 when setting ickkw=0? you need to set those value to have a fair comparison in this case.

Cheers,

Olivier

 prashant sundaram (psundaram) said on 2020-08-12: #2

Hi Olivier,

Yes I did set the ptj and mmjj cut to 30 while setting ickkw=0, still my cross-section is less than ickkw=1 case.

>Now for the rest, I would expect that this can be scale related what is the theoretical uncertainty on those numbers?
By the numbers do you mean the scale? I don't quite understand this.

Cheers,
Prashant

 prashant sundaram (psundaram) said on 2020-08-13: #3

Hi Olivier,

I was going through the run of the events, at the end of Pythia shower i get an output like this

Cross-section : 17.2 +- 0.05501 pb
Nb of events : 10000
Pythia8 merged cross-sections are:
> Merging scale = 30 : 8.315 +/- 0.03 [pb]
> Merging scale = 45 : 8.3546 +/- 0.03 [pb]
> Merging scale = 60 : 8.3408 +/- 0.03 [pb]
Notice that because Systematics computation is turned on, the merging did not veto events but modified their weights instead.
The resulting hepmc/stdhep file should therefore be use with those weights

What does the cross-section 17pb signify ?Is it the total inclusive cross-section for the process or the matched one? I am a bit confused about this.

Thanks,
Prashant

 prashant sundaram (psundaram) said on 2020-08-13: #4

Hi Olivier,

For the above comment I forgot to specify that i have set ickkw=1 and set xqcut to 15 and qcut to 30.

Cheers,
Prashant

 Olivier Mattelaer (olivier-mattelaer) said on 2020-08-14: #5

Hi,

Yes 17.2 is the sum of the individual cross-section of each multiplicity independently.
Those numbers lack the inclusion of the Sudakov form factor which will make each multiplicity exclusive and remove the double counting. Such Sudakov is performed by Pythia8 via a veto method. And this is why the cross-section after pythia is smaller.

The physical cross-section is the one after pythia8.

Cheers,

Olivier

 prashant sundaram (psundaram) said on 2020-08-14: #6

Dear Olivier,

Thanks a lot for the answer.I still have one small question ,the hepmc file that is generated at the end of the run, does it contain the merged cross-sections? I am trying to run these hepmc files in Rivet-3.1.0 analyses and i see that the generated cross-section is 17.2pb which is not the merged one , so how can i make sure that the physical cross-section after pythia shower is included in my hepmc file.

Thanks,
Prashant

 Olivier Mattelaer (olivier-mattelaer) said on 2020-08-14: #7

You need to use the weight associated to each event within the analysis otherwise the cross-section (and observables) will be wrong.

Cheers,

Olivier

 prashant sundaram (psundaram) said on 2020-08-14: #8

So can i do this at the level of madgraph or do i need to implement it in Rivet?And where do i get the weight associated to each event?

Cheers,
Prashant

 Olivier Mattelaer (olivier-mattelaer) said on 2020-08-14: #9

They are inside the hepmc file, Rivet need to take that into account.

Cheers,

Olivier

> On 14 Aug 2020, at 16:10, prashant sundaram <email address hidden> wrote:
>
> Question #692330 on MadGraph5_aMC@NLO changed:
>
> prashant sundaram posted a new comment:
> So can i do this at the level of madgraph or do i need to implement it
> in Rivet?And where do i get the weight associated to each event?
>
> Cheers,
> Prashant
>
> --