Cross section of h production in MadGraph/Pythia

Hi,

The paper does not do shower-kt MLM ( arXiv:0810.5350)
but the more usual KT MLM.
So to reproduce our result, you should have
> JetMatching:doShowerKt = off

In showerkt MLM, you are allowed (and encouraged to set xqcut=qcut) which is the advantages of that method.

When reading the paper, you will also see that that matching scale is set to 50 GeV and not 45.
The xqcut is not mentionned in the paper (since this is a pure technical cut) but I remember to be forced to set it quite low to have smooth DJR plot (even if they are not perfect as explained in the paper).

> 1) there were some errors when running pythia:

This seems to be some issue with your lhapdf setup. This means we are not able to compute
systematics error (scale and PDF variation) so you "lack" the additional weights related to those theoretical uncertainties (which are optional). So all your result are actually fully correct.

> 2) Pythia8 merged cross-sections are:
>> Merging scale = 50 : 15.904 +/- 0.06 [pb]
>> Merging scale = 75 : 15.835 +/- 0.06 [pb]
>> Merging scale = 100 : 15.769 +/- 0.06 [pb],
>
> while MadGraph cross-section is 27.18 pb.

The madgraph cross-section has the doulble counting so it is not relevant, you should typically compare the cross-section after merging with the one with 0j. However in this case, you have a lot of diagram opening due to the box diagram and the pentagon ones. Those can not be handle correctly by the parton-shower and therefore it makes sense that the cross-section after merging differs from the 0j one.

Cheers,

Olivier

> On 23 Apr 2019, at 08:37, Maksym Ovchynnikov <email address hidden> wrote:
>
> New question #680399 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/680399
>
> I tried to reproduce the pT spectrum obtained in https://arxiv.org/abs/1507.00020 with the following:
> generate p p > h [noborn=QCD]
> add process p p > h j [noborn = QCD]
> 1 = ickkw
> 30 = xqcut
> All kinematical cuts for jets (pT, eta, Delta R) are set to zero.
> QCUT = 45.
> JetMatching:doShowerKt = on
> JetMatching:nJetMax =1
> Merging:nJetMax = 2
> However, the obtained spectrum (from the hepmc file, where I have chosen the last row with the Higgs data inside the each event) turns out to be completely different. Namely, it has a maximum in the region pT = 130 GeV.
> Could you please tell me what can be a reason for such large discrepancy? Where can I make a mistake?
>
> Here is some info regarding run:
>
> 1) there were some errors when running pythia:
>
> INFO: Trying to download NNPDF23_lo_as_0130_qed
> Traceback (most recent call last):
> File "/usr/local/bin/lhapdf", line 12, in <module>
> __version__ = lhapdf.__version__
> AttributeError: 'module' object has no attribute '__version__'
> /usr/local/share/LHAPDF/NNPDF23_lo_as_0130_qed.tar.gz: Permission denied
> tar (child): NNPDF23_lo_as_0130_qed.tar.gz: Cannot open: No such file or directory
> tar (child): Error is not recoverable: exiting now
> tar: Child returned status 2
> tar: Error is not recoverable: exiting now
> WARNING: impossible to download all the pdfsets. Bypass systematics
>
> 2) Pythia8 merged cross-sections are:
>> Merging scale = 50 : 15.904 +/- 0.06 [pb]
>> Merging scale = 75 : 15.835 +/- 0.06 [pb]
>> Merging scale = 100 : 15.769 +/- 0.06 [pb],
>
> while MadGraph cross-section is 27.18 pb.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thank you for the answer!

So, to conclude, the setup I should run is

generate p p > h [noborn=QCD]
add process p p > h j [noborn = QCD]
1 = ickkw
30 = xqcut (?)
All kinematical cuts for jets (pT, eta, Delta R) are set to zero.

QCUT = 50
JetMatching:doShowerKt = off
JetMatching:nJetMax =1
Merging:nJetMax = 2,

correct?

This part sounds weird:

JetMatching:nJetMax =1
Merging:nJetMax = 2,

I actually not sure what is the difference here.
Please read the py8 manual to be sure.
The maximum multiplicity for the merging should be in that case "1" and not "2"

Cheers,

Olivier

Dear Olivier,

I have launched the simulation with

generate p p > h [noborn=QCD]
add process p p > h j [noborn = QCD]
1 = ickkw
30 = xqcut (?)
All kinematical cuts for jets (pT, eta, Delta R) are set to zero.
QCUT = 50
JetMatching:doShowerKt = off
JetMatching:nJetMax =1

Now the pT spectrum slightly differs from the one obtained after the first run (it has a maximum neat pT = 50 GeV), but it is still far from the spectrum shown in your paper. Could you please tell me what can be a reason for this?

Sorry, it has a maximum in the vicinity of 30 GeV. Maybe, xqcut matters?

Hi,

This is also what I observe in the plot of Figure 4.
xqcut should have zero impact on your plot (if it does, you have to reduce it as long as your curve is impacted by that value).
The smaller that value is, the more events will be veto (or given weight 0) by the parton-shower.

Cheers,

Olivier

Dear Olivier,

sorry, I did not look careful on the spectrum from the paper. So, the spectrum has a maximum at the vicinity of 30 GeV. Comparing with other programs like powheg/MC@NLO, I notice that the maximum predicted by MadGraph is shifted on ~ 15 GeV (see please Fig. 4 in https://arxiv.org/pdf/1510.08850.pdf).

Could you please tell me what is the main reason for this discrepancy? Different perturbation theory orders considered in these MC simulators?

Another my suggestion is that in the paper I cited there is a spectrum based on the process p p > h only, while the prediction for MG includes also the process p p > h j.

Thanks Olivier Mattelaer, that solved my question.