How to decay HNL particle of mass just below the W mass into off-shell W

Asked by Ogul Oencel on 2020-11-19

Dear Experts,

I am having a problem with making sense of unexpectedly high raw (number of generated events) MC efficiencies when I generate the process with the mass of BSM particle I am using close to W boson mass which it decays into.

I am working with HeavyN model [1] and attempting at generating NLO tt̄ events and then decay them with Madspin, where we probe the regime with mHNL < mW. We allow W of SM top to decay hadronically or leptonically whereas we strictly ask W from HNL from BSM top to decay only hadronically.

SM Top: t -> Wb -> jj or lv
BSM top: t -> Wb -> Nl where N -> Wl and W -> jj

To translate this into MG5 syntax:

1) Generate tt̄ events with Proc_card, I do:

=========================
generate p p > t t~ [QCD]
=========================

2) In a second step I want to decay these top quarks using MadSpin:

=========================
set spinmode none # 3body decay
decay t > b w+, (w+ > n ssl+ , n > ssl+ j j) # since mN < mW 2-body decay syntax not possible
decay t~ > b~ w-, w- > all_sm all_sm NP=0 #all_sm is defined as l, j and v to avoid Ns here
=========================

Now, we probe the range 15-80GeV for HNL and expecting to have a peak sensitivity of about 50GeV and lower efficiencies towards the end points. We see this shape between 15-75GeV, however at 80.0GeV the raw MC number of events passing selections start to increase again. Making a finer check we also studied 77 and 79GeV and noticed the unexpected rise of efficiency happens at 79GeV as well, but not at 77GeV. To give some feeling it goes like that for some cut level (100K events):

65Gev: 27.8%
70Gev: 21.7%
75Gev: 14.3%
77Gev: 12.6%
79Gev: 14.3%
80Gev: 15.3%

We also note that cross-section results are reasonable at all mass points. Therefore, we suspect this is to do with the BW_cut settings. Checking posts on this forum my understanding is that there are two different BWcuts:

Run_card "bwcutoff" whose default is 15 and regulates |m-m0| < bwcutoff x Gamma
MadSpin "BW_cut" whose default is 1.5 and regulates |m-m0|*|m-m0| < BW_cut x Gamma

We have made some checks with various bwcutoff values(5,15,40) and have not seen any differences.
When one plugs in the Gamma_W=2.085 and BW_cut=1.5 and assuming m0=mW=80.4 one comes to the value of m=78.7GeV for cut-off. This seems to fit inbetween the 77GeV (no efficiency problem) and 79GeV (efficiency problem). However, setting BW_cut to 0 does not change the outcome either.

We would like to then consult you on the following question:

1) Is it possible to generate HNL=80.0GeV mass point in a valid way?
2) Is there a hierarchy between two BW cuts. Do they affect the same particles? Does one overwrite the other if stricter?
3) is the default value for BW_cut still 1.5. Because when I want to enter non-integer values (0.5 for example) program crashes and says BW_cut must be an integer? I am using MG.2.7.3

Cheers,

Ogul

Question information

Language:
English Edit question
Status:
For:
Assignee:
No assignee Edit question
Last query:
2020-11-26
2020-11-26

This question was reopened

 Olivier Mattelaer (olivier-mattelaer) said on 2020-11-21: #1

Hi,

I do not understand what your percentage means.

1) Is it possible to generate HNL=80.0GeV mass point in a valid way?

Well since the mass of the W is at 80.4 and the mass of the hNL is at 80.0 and that the mass of the W is at 2.085
It seems that you are not in a Narrow-width approximation regime. So you should not use a narrow-width approximation issue.

2) Is there a hierarchy between two BW cuts. Do they affect the same particles? Does one overwrite the other if stricter?

In madspin none, they are no offshell effect (that's the definition of the mode none). So the parameter are of madspin should not have any effect. Now with spinmode=none mode you are allow to specify a run_card where you can define a new value of bw_cutoff which will be applied for the "W/n" particles which is independent of all the other ones (the default should be 15 here).

Cheers,

Olivier

 Ogul Oencel (oguloncel) said on 2020-11-26: #2

Hi,

Thanks for the replies.

The percentages refer to raw number of MC events left after the cutflow. As an example, if one generates 10K events, and after the cutflow 1K raw events are left, that percent would be 10%. We use this information to estimate required statistic tics we want to have in the end.

1) I see. Is there an alternative approach you can recommend?

2) Here I am confused I guess. I thought two-body decay syntax enforces on-shellness such as t > W b, whereas many-body decays (enabled through set spin mode = none) are always carried out off-shell such as t > b j j or in our case n > l j j?

I understand that run_card version of BW_cut has a default of 15, but I am not sure about the default and allowed values of the MadSpin_card definition of second BW_cut. Documentation says 1.5 but program accepts only integers?

Cheers,

Ogul

 Olivier Mattelaer (olivier-mattelaer) said on 2020-11-26: #3

1) I guess you need to use a five body decay for the top without any request of having particle onshell.

2) Maspin in "none" is the same as the following code:
- generate the production event with madevent/aMC@NLO
- generate the 1 to N body decay with madevent
for each production event, take one of the decay event and attach it to the production (i.e. by boosting the decay part).

So by construction the particle that you decay with madspin will be exactly onshell.
So in that mode that parameter of madspin is irrelevant.

Cheers,

Olivier