MadGraph5_aMC@NLO

neutrino production in fixed target experiment

Asked by Yulei Zhang

Dear experts,

I'm trying to generate the neutrino pair production process in Madgraph5 3.5.3.
The process comes from an electron beam, hitting some targets (Tungsten), and interacting with nuclei, resulting in a pair of neutrinos.

Following some discussion about fixed target mode and heavy ion mode here, I'm using the following command to generate:

-------------------------------

# Generate
import model sm
define np = u d u~ d~
define nj = u d u~ d~
generate np e- > e- vl vl~ nj
output pair_production

# Launch Jobs
set lpp1 1
set lpp2 0
set ebeam1 0
set ebeam2 10
set nevents 1000
set use_syst False
set no_parton_cut

set nb_neutron1 109
set nb_proton1 74
set mass_ion1 171.25

-------------------------------

My questions are:

1. Is this the correct way to simulate the fixed-target experiment for an electron beam, using Tungsten as a target?
2. Should I use the fixed scale in this case, and how should I choose the fixed scale value?
3. In the heavy ion mode, the form factors are already defined, right?

Note: For question 2, I've tried to produce other processes (like e + N -> e + Ap + N, Ap is a dark photon), and I found that only if I set the fixed scale to ~10, the cross-sections are correct. But for this neutrino production, the reference value is around 0.03 fb in this paper(https://arxiv.org/pdf/1411.1404.pdf, Page. 8), and only when I do not set the fixed scale, the madgraph gives a cross-section around ~ 0.9 fb, and with the warning: failed to generate enough events.

I'm wondering how should I config this process in Madgraph for the fixed target mode. I appreciate your reply.

Best

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Yulei Zhang
Solved:
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#1

Hi,

I'm not an heavy ion expert. So do no trust blindly my comments.

I do not understand here why you do not include other quark (c, s and maybe b and t) into your initial/final state.
I do understand that for your particular process, including the gluon is irrelevant but this is also not a problem to include it.
So I would have done (where here p and j does not include b quark)
generate p e- > e- vl vl~ j

Second, I would have expect the command "launch"
between the output command and the first set command
(wihtout it, it crash on my version of MG5aMC at least)

Third, I would have expect that you would need some cut on the final state to avoid singular behavior of your amplitude.
Why do you think that this is not needed? The computation is at least quite slow on my system and I do see that the slow channel is the one with a t-channel photon. This is exactly the channel where I would have say that a cut is needed to remove the photon singularity. Note that the default dynamical scale choice is automatically associate to a cut preventing scale to be too small this is likely why your cross-section is not infinite here.

For setting the fix scale, this can be done via the line
set fixed_scale 10
I have check that just setting fixed_scale and no cut make the cross-section blowing up (38989 +- 61158). Meaning that you indeed need a cut on the jet in one way or another.

And for your question on form-factor,
the mode with
set nb_neutron1 109
set nb_proton1 74
is using rescaling of the pdf set that you provide (if you do not change it, it is NNPDF2.3LO1 which is used and rescale)
in that case they are no global effect taken into account.

You can also use the mode without specifying the nb_neutron/nb_proton and use a PDF dedicated to the ION that you consider. Those are then more precise.

Cheers,

Olivier

Revision history for this message
Yulei Zhang (avencast) said : 		#2

Dear Olivier,

Many thanks for your reply.

----
I do not understand here why you do not include other quark (c, s and maybe b and t) into your initial/final state.
I do understand that for your particular process, including the gluon is irrelevant but this is also not a problem to include it.
So I would have done (where here p and j does not include b quark)
generate p e- > e- vl vl~ j

--> Yes, I should include them.

---
Second, I would have expect the command "launch"
between the output command and the first set command
(wihtout it, it crash on my version of MG5aMC at least)

--> Yes, I just want to show the main part, so I ignore it here.

---
Third, I would have expect that you would need some cut on the final state to avoid singular behavior of your amplitude.
Why do you think that this is not needed? The computation is at least quite slow on my system and I do see that the slow channel is the one with a t-channel photon. This is exactly the channel where I would have say that a cut is needed to remove the photon singularity. Note that the default dynamical scale choice is automatically associate to a cut preventing scale to be too small this is likely why your cross-section is not infinite here.

--> I'm trying to consider some kinematics cuts on the final states. But when I ran mg5 with version 3.5.3, with setting no parton cut, it could still give some meaningful cross-sections. But I'm new to this, do you have any idea what kind of cut I need here, for an 8GeV electron-hitting-on-target experiment?

---
For setting the fix scale, this can be done via the line
set fixed_scale 10
I have check that just setting fixed_scale and no cut make the cross-section blowing up (38989 +- 61158). Meaning that you indeed need a cut on the jet in one way or another.

--> I have a silly question here, how should I determine the number (10 here)? Should I set this number according to the center-of-mass energy?

---
And for your question on form-factor,
the mode with
set nb_neutron1 109
set nb_proton1 74
is using rescaling of the pdf set that you provide (if you do not change it, it is NNPDF2.3LO1 which is used and rescale)
in that case they are no global effect taken into account.

You can also use the mode without specifying the nb_neutron/nb_proton and use a PDF dedicated to the ION that you consider. Those are then more precise.

--> Thanks for your clarification. So if I simply use the heavy ion mode, it doesn't consider the form factors.
Do you have any idea how should I implement a dedicated PDF? I find this PDF (nCTEQ15_184_74) for Tungsten target. So if I want to use this one in lhapdf, I need to remove all the heavy ion parameters and set the beam energy to be 172.84 (the rest mass of Tungsten)?

I appreciate your reply.

Best,
Yulei

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#3

For what are "relevant" cut, I would not be the correct person to answer that, I have no experience with such type of experiment and therefore do not know what is typically use experimentally. My advise is to either check with your supervisor and/or check experimental paper to check which kind of cut they use for such setup.
(and use even softer version within MG5aMC)

--> I have a silly question here, how should I determine the number (10 here)? Should I set this number according to the center-of-mass energy?

They are no good answer here, changing that value, will lead to different prediction and you typically check for multiple value and take the envellope to get a theoretical error associated to your prediction.

--> Thanks for your clarification. So if I simply use the heavy ion mode, it doesn't consider the form factors.
Do you have any idea how should I implement a dedicated PDF? I find this PDF (nCTEQ15_184_74) for Tungsten target. So if I want to use this one in lhapdf, I need to remove all the heavy ion parameters and set the beam energy to be 172.84 (the rest mass of Tungsten)?

You need to check the manual for such type of PDF, (each PDF might be implemented differently). In general such type of PDF expect the energy per nuclei, in such case it will be around 1.0048
set ebeam1 1.0048
set pdlabel lhapdf
set lhaid XXXXXX
(where XXXX is the lhaid for the pdf set that you want)

Cheers,

Olivier

Revision history for this message
Yulei Zhang (avencast) said : 		#4

Dear Olivier,

Thanks for your patient reply. I'll search more about the theory part.

Best,
Yulei