Applying parton cuts

Asked by Blaž Leban on 2019-05-22

Dear Olivier,

I am generating some SM background processes for my analysis. Firstly, I tried to have all inclusive runs (with no parton cuts in MG5, I only applied cuts during analysis) and in most cases I was able to generate enough events with my laptop, but in some cases (because of the large cross section) I would have to generate > 700 mio events for sufficient statistics, so my adviser suggested me to apply cuts already in MG5 to suppress cross section but now I have some problems.

My signal should have 2 leptons (I am restricting to electrons) and 4 jets in final state and after the analysis of the signal and background which I was able to generate, I decided to make two cuts on lepton pT-s for the rest background processes. From the analysis, it was clear that the signal (from BSM) could easily be extracted with lepton pT cuts - pT(l1) > 150 GeV and pT(l2) > 25 GeV and I saw that in MG5, I am able to include those two cuts via ptl1min and ptl2min command in run_card. The problem is that when I do this, it says that the cross section is equal to 0 and one of the reasons could be too strong cuts, but I lower them and the error stayed the same. My backup plan was the ordinary ptl and then also the ptj cut, but it turned out that although the process 'went through' the cross section was still high (actually it stayed the same) even when applying abnormal cuts like ptl = 10000 and ptj = 100000. I have the problems with following processes:

p p > t t~ @0, p p > t t~ j @1, p p > t t~ j j @2 , cs = 1152 pb,
p p > V @0, p p > V j @1, p p > V j j @2, cs = 2.5e+05 pb ,
p p > V V @0, p p > V V j @1, p p > V V j j @2 , cs = 674.93 pb,

where V = w+ w- z.
Here is an example of my script:

generate p p > t t~ @0
add process p p > t t~ j @1
add process p p > t t~ j j @2

output simulations/Background/ttj
launch

# Param card:
shower = Pythia8
detector = Delphes
analysis = OFF

# Run card:
set nevents 1000
set pdlabel lhapdf
set lhaid 21250

# Cuts:
set ptl1min 100.0
set ptl2min 10.0

or

set ptl 100.0
set ptj 20.0

Am I missing something?
Thank you in advance, looking forward to hearing from you,
Blaž

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2019-05-24
Last query:
2019-05-24
Last reply:
2019-05-23

Hi,

MG5aMC can only apply cuts on particle present at the MG5aMC level.
Since you do not decay any of your particle (you ask for t t~ in your examples, the decay of those is not asked within MG5aMC -- will be done by Pythia8), we do not have any lepton for applying the cut... So depending of the cut we either cut 100% of the events or 0%.

So you should either apply a pt cut on the top, or make the top to decay within MG5aMC and then you can use a cut on the leptons.

Note that you should apply much smaller cut compare to your analysis cut since the parton-shower and detector effect can increase the pt of some lepton.

Cheers,

Olivier

Blaž Leban (gasar8) said : #2

Thank you for your quick reply, it clarifies a lot, I somehow thought the cuts speak to Pythia also. :)

> ...or make the top to decay within MG5aMC and then you can use a cut on the leptons.

Can this be done with madspin=ON, because I think that I have already tried this? How else can I achieve this, do I have to specify the exact final states like (taken from https://answers.launchpad.net/mg5amcnlo/+question/236617):

define l+ = l+ ta+
define l- = l- ta-
generate p p > t t~, (t > b w+, w+ > j j), (t~ > b~ w-, w- > l- vl~) @1
add process p p > t t~, (t > b w+, w+ > l+ vl), (t~ > b~ w-, w- > j j) @2
add process p p > t t~, (t > b w+, w+ > l+ vl), (t~ > b~ w-, w- > l- vl~) @3
add process p p > t t~, (t > b w+, w+ > j j), (t~ > b~ w-, w- > j j) @4

or is there any command which would decay t?

Best regards,
Blaž

Yes this is the way to go:

> define l+ = l+ ta+
> define l- = l- ta-
> generate p p > t t~, (t > b w+, w+ > j j), (t~ > b~ w-, w- > l- vl~) @1
> add process p p > t t~, (t > b w+, w+ > l+ vl), (t~ > b~ w-, w- > j j) @2
> add process p p > t t~, (t > b w+, w+ > l+ vl), (t~ > b~ w-, w- > l- vl~) @3
> add process p p > t t~, (t > b w+, w+ > j j), (t~ > b~ w-, w- > j j) @4

be carefull to the cut_decays parameter of the run_card.

Cheers,

Olivier

> On 22 May 2019, at 22:37, Blaž Leban <email address hidden> wrote:
>
> Question #681010 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/681010
>
> Status: Answered => Open
>
> Blaž Leban is still having a problem:
> Thank you for your quick reply, it clarifies a lot, I somehow thought
> the cuts speak to Pythia also. :)
>
>> ...or make the top to decay within MG5aMC and then you can use a cut
> on the leptons.
>
> Can this be done with madspin=ON, because I think that I have already
> tried this? How else can I achieve this, do I have to specify the exact
> final states like (taken from
> https://answers.launchpad.net/mg5amcnlo/+question/236617):
>
> define l+ = l+ ta+
> define l- = l- ta-
> generate p p > t t~, (t > b w+, w+ > j j), (t~ > b~ w-, w- > l- vl~) @1
> add process p p > t t~, (t > b w+, w+ > l+ vl), (t~ > b~ w-, w- > j j) @2
> add process p p > t t~, (t > b w+, w+ > l+ vl), (t~ > b~ w-, w- > l- vl~) @3
> add process p p > t t~, (t > b w+, w+ > j j), (t~ > b~ w-, w- > j j) @4
>
> or is there any command which would decay t?
>
> Best regards,
> Blaž
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Blaž Leban (gasar8) said : #4

Thank you. Unfortunately, I don't quite get what MadSpin is then used for? As far as I understand is that it automatically decays products, so that I don't have to write all those messy lines?

I am asking this also because I am not sure how to further decay the other two processes described in first post (so that I would again be able to apply cuts on final states - leptons), where I have defined a multiparticle v = w+ w- z? Do I have to generate them separately then?

The interest of MadSpin is first for NLO process.
This code does not bring anything new compare to the decay chain syntax (the one that you call messy).
They are however cases at LO where MadSpin is faster than the decay chain syntax and this is why we allow the user to choose between the decay chain syntax and MadSpin.

MadSpin is actually much more constrained on what it can do compare to the decay chain syntax (no cut in MadSpin, no three body decay,...) and less secure (rely more deeply on narrow width approximation compare to the decay-chain syntax). I would refer you to the MadSpin paper for more details.

> I am asking this also because I am not sure how to further decay the
> other two processes described in first post (so that I would again be
> able to apply cuts on final states - leptons), where I have defined a
> multiparticle v = w+ w- z? Do I have to generate them separately then?

If you do not know how to decay them, I do not know either.
I guess something like
z > l+ l-, w+ > l+ vl, w- > l- vl~
should be what you are looking for. But you should try and look at the Feynman diagram to see if this is what you want.

Cheers,

Olivier

> On 22 May 2019, at 23:52, Blaž Leban <email address hidden> wrote:
>
> Question #681010 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/681010
>
> Status: Answered => Open
>
> Blaž Leban is still having a problem:
> Thank you. Unfortunately, I don't quite get what MadSpin is then used
> for? As far as I understand is that it automatically decays products, so
> that I don't have to write all those messy lines?
>
> I am asking this also because I am not sure how to further decay the
> other two processes described in first post (so that I would again be
> able to apply cuts on final states - leptons), where I have defined a
> multiparticle v = w+ w- z? Do I have to generate them separately then?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Blaž Leban (gasar8) said : #6

Hi Olivier, thank you for your reply.

>I guess something like z > l+ l-, w+ > l+ vl, w- > l- vl~ should be what you are looking for.

Thank you, hehe, I know how each individual particle decays, but I don't know the syntax to implement this for a multiparticle v, since I have defined it as v = w+ w- z. Would it be ok to define another multiparticle k = l+ l- vl vl~ and generate p p > V, V > k k?

Should be ok (but please check the Feynman Diagram to be sure)

> On 23 May 2019, at 09:22, Blaž Leban <email address hidden> wrote:
>
> Question #681010 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/681010
>
> Status: Answered => Open
>
> Blaž Leban is still having a problem:
> Hi Olivier, thank you for your reply.
>
>> I guess something like z > l+ l-, w+ > l+ vl, w- > l- vl~ should be
> what you are looking for.
>
> Thank you, hehe, I know how each individual particle decays, but I don't
> know the syntax to implement this for a multiparticle v, since I have
> defined it as v = w+ w- z. Would it be ok to define another
> multiparticle k = l+ l- vl vl~ and generate p p > V, V > k k?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Blaž Leban (gasar8) said : #8

Thank you Olivier for your quick response, that seems to work just fine. :)
That also solves my question.