Validity of madgraph for fixed target experiments

Hi,

Indeed MadGraph is also used for fixed target experiment.
In this case, you need to set the energy of the proton beam to the proton mass (we use 0.938d0).
You also have to set the pdlable of the lepton beam to 0 (no pdf).

Cheers,

Olivier

On 16 Feb 2015, at 15:36, Pilar <email address hidden> wrote:

> New question #262196 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/262196
>
> Hello,
>
> I would like to use MadGraph for a sensitivity study using a fixed target experiment. I know that MadGraph is normally used for collider experiments, but I see no reason why it should not work in my case as well. However, I want to make sure that this is so.
>
> For instance, one of the processes I am interested in is a neutral-current deep inelastic scattering event, where a lepton arrives to the target and interacts via NC, producing a jet in the final state:
>
> l p > l j
>
> In order to make sure I am in this regime (so that the parton model is valid), I am applying a cut of 1 GeV to the energy and Pt of the jets in the final state. Is MadGraph reliable for this type of calculation? Is there something else I should be taking care of, besides the cuts on the jet in the final state?
>
> Many thanks in advance!
> Pilar
>
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thanks for replying so quickly!

Pilar

Thanks Olivier Mattelaer, that solved my question.

Hi again,

I have generated the events for a fixed target experiment as you suggested, setting the proton energy to ~1 GeV and the lepton PDF to 0. This works and I can generate events just fine, but the cross section I am getting is too low (by two orders of magnitude roughly).

So, after looking carefully at the LHE file, I realized that even though the proton energy is set to 1 GeV, the parton energy is always above 0.6 GeV or so (ie, x > 0.6). The partons are always up and down quarks, so this is fine. But the fact that the events are generated only with x > 0.6 means that I am probably getting a strong suppression on the cross section because the PDF in that range is already very small, right? and I am missing the peak around 0.3 or so...

I tried to remove all cuts in the run card, and also a couple of different choices for the PDF function, but still got the same result.

Am I missing something here? Is there anything I should be doing to avoid this issue?

Thanks in advance.
Best,
Pilar

Hi Pilar,

So you are interested in inelastic collision. Did you check the validity of the pdf in that range of energy?
I'm not sure that the factorization theorem can be use for such low energy.

Cheers,

Olivier

Hi Olivier,

thanks for your prompt answer! Yes, I am interested in generating a deep inelastic event. In principle, factorization should be OK as long as the momentum transfer is large enough, right? If this is so, it should not depend on the energy of the event. In order to guarantee this, I was imposing some cuts on the transverse momentum of the jet in the final state. Just for you to know, I am interested in interactions where the initial lepton has a few GeV of energy, and the proton is at rest.

Curiously enough, when doing some checks on the events I found out that, even if I removed all cuts in the run card, the momentum transfer is always larger than 2 GeV or so. Does madgraph apply some internal cut on Q^2 in order to guarantee that factorization is OK?

Thanks,
Pilar

Hi Pilar,

> Curiously enough, when doing some checks on the events I found out that,
> even if I removed all cuts in the run card, the momentum transfer is
> always larger than 2 GeV or so. Does madgraph apply some internal cut on
> Q^2 in order to guarantee that factorisation is OK?

Yes, this is possible, not really for the factorisation but more to avoid problem with hadronisation.
(even if both are actually related). But i have no idea where this cut is done in the code.
That cut was introduce by Johan Alwall who left physics couple of years ago.

One idea is to check if you perform a fix scale computation if you still have that effect or not.

Cheers,

Olivier

On 10 Apr 2015, at 03:16, Pilar <email address hidden> wrote:

> Question #262196 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/262196
>
> Status: Answered => Open
>
> Pilar is still having a problem:
> Hi Olivier,
>
> thanks for your prompt answer! Yes, I am interested in generating a deep
> inelastic event. In principle, factorization should be OK as long as the
> momentum transfer is large enough, right? If this is so, it should not
> depend on the energy of the event. In order to guarantee this, I was
> imposing some cuts on the transverse momentum of the jet in the final
> state. Just for you to know, I am interested in interactions where the
> initial lepton has a few GeV of energy, and the proton is at rest.
>
> Curiously enough, when doing some checks on the events I found out that,
> even if I removed all cuts in the run card, the momentum transfer is
> always larger than 2 GeV or so. Does madgraph apply some internal cut on
> Q^2 in order to guarantee that factorization is OK?
>
> Thanks,
> Pilar
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Ok I found it:

c Check that factorization scale is >= 2 GeV
      if(lpp(1).ne.0.and.q2fact(1).lt.4d0.or.
     $ lpp(2).ne.0.and.q2fact(2).lt.4d0)then

This is in SubProcesses/reweight.f around line 988.

Cheers,

Olivier

On 10 Apr 2015, at 15:41, Olivier Mattelaer <email address hidden> wrote:

> Question #262196 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/262196
>
> Status: Open => Answered
>
> Olivier Mattelaer proposed the following answer:
> Hi Pilar,
>
>> Curiously enough, when doing some checks on the events I found out that,
>> even if I removed all cuts in the run card, the momentum transfer is
>> always larger than 2 GeV or so. Does madgraph apply some internal cut on
>> Q^2 in order to guarantee that factorisation is OK?
>
> Yes, this is possible, not really for the factorisation but more to avoid problem with hadronisation.
> (even if both are actually related). But i have no idea where this cut is done in the code.
> That cut was introduce by Johan Alwall who left physics couple of years ago.
>
> One idea is to check if you perform a fix scale computation if you still
> have that effect or not.
>
> Cheers,
>
> Olivier
>
>
>
> On 10 Apr 2015, at 03:16, Pilar <email address hidden> wrote:
>
>> Question #262196 on MadGraph5_aMC@NLO changed:
>> https://answers.launchpad.net/mg5amcnlo/+question/262196
>>
>> Status: Answered => Open
>>
>> Pilar is still having a problem:
>> Hi Olivier,
>>
>> thanks for your prompt answer! Yes, I am interested in generating a deep
>> inelastic event. In principle, factorization should be OK as long as the
>> momentum transfer is large enough, right? If this is so, it should not
>> depend on the energy of the event. In order to guarantee this, I was
>> imposing some cuts on the transverse momentum of the jet in the final
>> state. Just for you to know, I am interested in interactions where the
>> initial lepton has a few GeV of energy, and the proton is at rest.
>>
>> Curiously enough, when doing some checks on the events I found out that,
>> even if I removed all cuts in the run card, the momentum transfer is
>> always larger than 2 GeV or so. Does madgraph apply some internal cut on
>> Q^2 in order to guarantee that factorization is OK?
>>
>> Thanks,
>> Pilar
>>
>> --
>> You received this question notification because you are an answer
>> contact for MadGraph5_aMC@NLO.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thanks for looking again into this! I have been doing some tests and it seems that fixing the factorization scale solves the problem, as you suggested in your previous response. I still need to do additional checks and so on, the results seem consistent for now. If I find any more issues I will reopen the question.

Thanks a lot again!
Cheers,
Pilar

Hi -

By digging in the code and googling around, I realize that I have also stumbled on this buried 2 GeV cut, which I was hoping to remove for my process (QED only, a a > mu+ mu-, using photons from proton beams).

As there is absolutely no issue with harmonization, Is there a reliable way to lower this cut, in principle with no (or a very low) pT for the outgoing muons? I am experimenting with just switching the flag to use a fixed factorization scale, but the jobs now run very slowly.

(For the record, I need MG for "a a > mu+ mu- a", but i need the simpler process for renormalizing to other calculations).

- Peter

Hi Peter,

Another way to avoid this cut is to set the dynamical scale via HT/2 (or sqrts) instead of the complicated function that we used as default.
You can switch that easily in the run_card.dat via the flag dynamical_scale_choice (look at the associated FAQ on the run_card for more details).

Now the fact that mad graph slows down is a sign that MGaMC as some struggle to converge, more likely due to some bad behaviour of the PDF very close to x=0.
Please consider the PDF error when setting your pt cut that low, I would not be surprised if the contribution at low pt have huge PDF uncertainties.
(You might want to use elastic photon in that regime actually)

Cheers,

Olivier

> On 4 May 2017, at 23:07, Peter Steinberg <email address hidden> wrote:
>
> Question #262196 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/262196
>
> Peter Steinberg posted a new comment:
> Hi -
>
> By digging in the code and googling around, I realize that I have also
> stumbled on this buried 2 GeV cut, which I was hoping to remove for my
> process (QED only, a a > mu+ mu-, using photons from proton beams).
>
> As there is absolutely no issue with harmonization, Is there a reliable
> way to lower this cut, in principle with no (or a very low) pT for the
> outgoing muons? I am experimenting with just switching the flag to use
> a fixed factorization scale, but the jobs now run very slowly.
>
> (For the record, I need MG for "a a > mu+ mu- a", but i need the simpler
> process for renormalizing to other calculations).
>
> - Peter
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi -

Sorry I wasn't clearer, by using lpp=2 for both beams I should already be using elastic photons only (as I am trying to reweight this calculation to compare with similar QED calculations for Pb+Pb, using EPA approximation).

Thus, I should have no PDF issues as you describe, since i'm just using epa_proton from PhotonFlux.f -- correct?

Perhaps this is why I am confused - why do I even see effects from Q2min in this context?

- Peter