Pt cuts in run_card.dat

Asked by Shubhani Jain on 2021-02-09

Hi

I want to produce events for processes g g > h2 > h1 h1 > b b~ b b~ [QCD] using 2hdmII_NLO with madspin, p p > b b~ b b~, p p > Z b b~ ,Z > b b~ and p p > t t~ , (t > W+ b , W+ > j j), (t~ > W- b~ , W- > all all) with j and all containing 'b' as well. What would be the sensible way to apply generation level cuts on these processes precisely pt_b and pt_j (where applicable)?

So far what we have done for g g > h2 > h1 h1 > b b~ b b~ [QCD] is to generate g g > h2 > h1 h1 , decay h1 to b b~ in madspin card and applying pt cut on h1 > 300 GeV. This seems to bw giving some weird results after madanalysis runs.

So it would be good to know a sensible way to generate events with pt cuts.

Regards
Shubhani

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2021-02-09
Last query:
2021-02-09
Last reply:
2021-02-09

Hi,

If you use MadSpin you basically have no way to have a cut on the decayed particle.
So this settle it for
> g g > h2 > h1 h1 [QCD] + madspin

For the other processes that you quote
> p p > b b~ b b~,
> p p > Z b b~ ,Z > b b~
> p p > t t~ , (t > W+ b , W+ > j j), (t~ > W- b~ , W- > all all)

You can use the cut from the run_card.
If your b as massless, ptb is not used and instead ptj is used.
For massive b, the ptb is used and not ptj

Cheers,

Olivier

> On 9 Feb 2021, at 12:45, Shubhani Jain <email address hidden> wrote:
>
> New question #695442 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/695442
>
> Hi
>
> I want to produce events for processes g g > h2 > h1 h1 > b b~ b b~ [QCD] using 2hdmII_NLO with madspin, p p > b b~ b b~, p p > Z b b~ ,Z > b b~ and p p > t t~ , (t > W+ b , W+ > j j), (t~ > W- b~ , W- > all all) with j and all containing 'b' as well. What would be the sensible way to apply generation level cuts on these processes precisely pt_b and pt_j (where applicable)?
>
> So far what we have done for g g > h2 > h1 h1 > b b~ b b~ [QCD] is to generate g g > h2 > h1 h1 , decay h1 to b b~ in madspin card and applying pt cut on h1 > 300 GeV. This seems to bw giving some weird results after madanalysis runs.
>
> So it would be good to know a sensible way to generate events with pt cuts.
>
> Regards
> Shubhani
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Shubhani Jain (s2697661) said : #2

Hi Olivier

Thanks for replying back.

So it is okay to just generate g g > h2 > h1 h1 [QCD], put generation level cut for pt(h1) > 400 GeV in run_card and then decaying h1 > b b~ in madspin card.

For ttbar, we have redefined j = g u d s c b and and all definition does have b's so would it be sensible to just put cut on pt_j or pt_b?

Regards
Shubhani

>So it is okay to just generate g g > h2 > h1 h1 [QCD], put generation level cut for pt(h1) > 400 GeV in run_card and then decaying h1 > b b~ in madspin card.

Technically yes, now I can not comment if this is a sensible way to put cut for your analysis.
In general, production cut needs to be mild compare to analysis cuts. But since I do not know your analysis cuts, I do not know if such cut is mild enough to not bias your analysis. Obviously, the point of cut at generation level is to reduce the number of events rejected at analysis level but if you set the cut too strong you will bias yourself.

> For ttbar, we have redefined j = g u d s c b and and all definition does have b's so would it be sensible to just put cut on pt_j or pt_b?

Well you will not have "b" quark out for the decay of the "W" and since the b is massive, it does make sense to have separate cuts.
This being said you can obviously set both cuts to the same value and this also make sense.

Cheers,

Olivier

Shubhani Jain (s2697661) said : #4

Hi Olivier

Thanks for you reply.

Our analysis cuts are 400 < pt(jets) < 500 GeV . So would putting a production cut of 400 GeV bias our analysis?

Regards
Shubhani

Hi,

> Our analysis cuts are 400 < pt(jets) < 500 GeV . So would putting a
> production cut of 400 GeV bias our analysis?

I would naively say yes, but the only way to be sure is to check.
Naively if you want a analysis cut 400 < pt(jets), I would start by considering a generation cut around 300-350GeV.
(I would actually run both to see if they are a difference the two, and check if the efficiency is acceptable or not).

Now since you do not cut at generation level on the jet but on the higgs, you likely need to be even more conservative.
But I have never study that effect.

Cheers,

Olivier

> On 9 Feb 2021, at 14:30, Shubhani Jain <email address hidden> wrote:
>
> Question #695442 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/695442
>
> Shubhani Jain posted a new comment:
> Hi Olivier
>
> Thanks for you reply.
>
> Our analysis cuts are 400 < pt(jets) < 500 GeV . So would putting a
> production cut of 400 GeV bias our analysis?
>
> Regards
> Shubhani
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Shubhani Jain (s2697661) said : #6

Hi

Thanks for the explanation we will try more conservative cuts.

Regards
Shubhani

Shubhani Jain (s2697661) said : #7

Thanks Olivier Mattelaer, that solved my question.

Billy Ford (billyford) said : #8

Hey both,

I am a collaborator of Shubhani's, I hope you don't mind me adding a comment/query.

We are happy with the method of cutting the Higgs pT to get an effective cut on the bjets to increase selection efficiency, and will optimise the exact values later. However the issue we are having at the moment is in the behaviour of the cuts. We find that, despite successfully cutting the Higgs pT at 400GeV (for example), there is a large tail in the bquark (and hence bjet) pT distributions around zero!

I have removed ISR/FSR, and sanity checked that the low pT b's are in fact ones that came from the Higgs decay, but this peculiar behaviour is still there, any ideas what this could be? I wondered if it was something to do with Madspin, as this is where we perform the h1 > b b~ decays.

Thanks very much,
Billy

Hi,

What I do observe is a quite flat spectrum for the decay product.
up to 400 and then it starts to decrease.
The relation between the pt of the decay product and the sum of two pt is not that simple:
pt_h**2 = ptb**2 + ptbbar**2 + 2* Pxb * Pxbar + 2* Pyb*Pybar

I do not see any reason to doubt that such output is correct.
(the pt of the sum is cut correctly and the invariant mass of the sum is correct as well)

Cheers,

Olivier

> On 10 Feb 2021, at 11:55, Billy Ford <email address hidden> wrote:
>
> Question #695442 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/695442
>
> Billy Ford posted a new comment:
> Hey both,
>
> I am a collaborator of Shubhani's, I hope you don't mind me adding a
> comment/query.
>
> We are happy with the method of cutting the Higgs pT to get an effective
> cut on the bjets to increase selection efficiency, and will optimise the
> exact values later. However the issue we are having at the moment is in
> the behaviour of the cuts. We find that, despite successfully cutting
> the Higgs pT at 400GeV (for example), there is a large tail in the
> bquark (and hence bjet) pT distributions around zero!
>
> I have removed ISR/FSR, and sanity checked that the low pT b's are in
> fact ones that came from the Higgs decay, but this peculiar behaviour is
> still there, any ideas what this could be? I wondered if it was
> something to do with Madspin, as this is where we perform the h1 > b b~
> decays.
>
> Thanks very much,
> Billy
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.