Higgs gluon fusion production at NLO

The syntax g g > h [all=QCD] which is the exact syntax for NLO in QCD is not valid in this case, since they are not Tree-Level diagram.

You can however compute the LO contribution (at one loop) via the loop-induced syntax
g g > h [noborn=QCD]
Obviously in this case, the PT of the Higgs at parton level is 0.
So you are likely needed to do
generate p p > h j [noborn=QCD]

The [QCD] syntax is by default a synonym of [all=QCD] but in the case where you do not have any tree-level diagram, in such case it is a synonym of [noborn=QCD]

Cheers,

Olivier

PS: a good reference here is: arXiv:1507.00020<http://arXiv.org/abs/arXiv:1507.00020>

On 19 Apr 2019, at 22:03, Maksym Ovchynnikov <<email address hidden><mailto:<email address hidden>>> wrote:

New question #680335 on MadGraph5_aMC@NLO:
https://answers.launchpad.net/mg5amcnlo/+question/680335

Hi.

As far as I understand the notation of NLO, MG5_aMC@NLO allows to calculate the pT spectrum of Higgs bosons produced in gluon fusion, as it is (at lease) the NLO effect. Am I right?

--
You received this question notification because you are an answer
contact for MadGraph5_aMC@NLO.

Dear Olivier,

sorry for stupid question, but is this pT spectrum relevant for the main production channel of the Higgs boson, which is the gluon fusion?

Hi,

I would say yes. But this depend what you are looking for obviously.

Cheers,

Olivier

Dear Olivier,

I am looking for double differential distribution for Higgs bosons produced in gluon fusion, pT and, say, pL. I can obtain the pL distribution from the process g g > h. However, I am not able to get the simultaneously the pT distribution. Am I right that in this case MG5_aMC@NLO can not help?

Hi,

Please read the paper that I mention earlier since this include a full section on Higgs gluon fusion.
We can certainly be used for such plot. I would go for the more accurate one which is a MLM matched/merged computation as done in the paper. Note that we realise that the default gap between xqcut and qcut was too small for this process and had to increase it to have meaningful result.

If you do not want to handle parton-shower, you can also simply do
generate p p > h j [QCD]
Which should give you the real fixed order LO prediction for the PT of the Higgs.

Cheers,

Olivier

> On 20 Apr 2019, at 00:27, Maksym Ovchynnikov <email address hidden> wrote:
>
> Question #680335 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/680335
>
> Status: Answered => Open
>
> Maksym Ovchynnikov is still having a problem:
> Dear Olivier,
>
> I am looking for double differential distribution for Higgs bosons
> produced in gluon fusion, pT and, say, pL. I can obtain the pL
> distribution from the process g g > h. However, I am not able to get the
> simultaneously the pT distribution. Am I right that in this case
> MG5_aMC@NLO can not help?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier,

I do the following:

_______________________________
import model heft

generate p p > h

add process p p > h j

add process p p > h j j

_______________________________

In the run_card I set Delta Rjj > 0.4, pT_min,j > 20, |eta_j| < 5. I got the total cross-section 39.55 pb. However, immediately after running there were error messages:
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv

What do they mean?

And also I've set ickkw = 0 and kqmin = 0.

Hi,

So how do you handle the double counting?

For the error you mention it is an error of your web browser so nothing to worry about

Now one more question. Why are you using heft?
If you look at the paper you will see how bad it is for the pt spectrum

Cheers

Olivier

Dear Olivier,

heft is needed in order to simulate the gluon fusion process and get the pL spectrum. I naively think that I can avoid the double counting by setting pT > 20 GeV and Rjj > 0.4.

Hi,

> heft is needed in order to simulate the gluon fusion process and get the pL spectrum.

Well I expect PL in heft to be as bad at PT (and therefore being completely wrong as soon as PL > 120 GeV.
Now you can/should use the same generation for both PT and PL.

> I naively think that I can avoid the double counting by setting pT > 20 GeV and Rjj > 0.4.

I believe that this is too naive. If you do not want to do a proper MLM (or CKKW-L) merging, then just use
generate p p > h j [QCD]

Cheers,

Olivier

> On 20 Apr 2019, at 20:08, Maksym Ovchynnikov <email address hidden> wrote:
>
> Question #680335 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/680335
>
> Status: Answered => Open
>
> Maksym Ovchynnikov is still having a problem:
> Dear Olivier,
>
>
> heft is needed in order to simulate the gluon fusion process and get the pL spectrum. I naively think that I can avoid the double counting by setting pT > 20 GeV and Rjj > 0.4.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier,

>Well I expect PL in heft to be as bad at PT (and therefore being completely wrong as soon as PL > 120 GeV.
Now you can/should use the same generation for both PT and PL.

Could you please tell me what is a reason for the heft model to break down in such domain?

>I believe that this is too naive. If you do not want to do a proper MLM (or CKKW-L) merging, then just use
generate p p > h j [QCD]

I do not have an experience in Pythia and other MC simulators except MadGraph. Therefore, my choice should be to use your advice. However, I still do not understand why the Higgs pL and pT spectrum can be (with some even rough accuracy) approximated by the process p p > h j. As far as I understand, the main contribution to the Higgs cross-section comes from the gluon fusion, which seem to have different kinematics (indeed at LO). May I kindly ask you to clarify this point?

> On 20 Apr 2019, at 22:17, Maksym Ovchynnikov <email address hidden> wrote:
>
> Question #680335 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/680335
>
> Status: Answered => Open
>
> Maksym Ovchynnikov is still having a problem:
> Dear Olivier,
>
>> Well I expect PL in heft to be as bad at PT (and therefore being completely wrong as soon as PL > 120 GeV.
> Now you can/should use the same generation for both PT and PL.
>
> Could you please tell me what is a reason for the heft model to break
> down in such domain?

The heft is based on a Taylor Expansion of the loop-computation.
This is actually fully valid in the infinite top mass limit...

So as soon as you have observables higher than the top mass, you expect the taylor expansion to break down.

> I do not have an experience in Pythia and other MC simulators except
> MadGraph. Therefore, my choice should be to use your advice. However, I
> still do not understand why the Higgs pL and pT spectrum can be (with
> some even rough accuracy) approximated by the process p p > h j. As far
> as I understand, the main contribution to the Higgs cross-section comes
> from the gluon fusion, which seem to have different kinematics (indeed
> at LO). May I kindly ask you to clarify this point?

If you look at the generated diagram, you will see that this is actually equivalent to the Real contribution of the gluon fusion at NLO.

>
>> I believe that this is too naive. If you do not want to do a proper MLM (or CKKW-L) merging, then just use
> generate p p > h j [QCD]
>
> I do not have an experience in Pythia and other MC simulators except
> MadGraph. Therefore, my choice should be to use your advice. However, I
> still do not understand why the Higgs pL and pT spectrum can be (with
> some even rough accuracy) approximated by the process p p > h j. As far
> as I understand, the main contribution to the Higgs cross-section comes
> from the gluon fusion, which seem to have different kinematics (indeed
> at LO). May I kindly ask you to clarify this point?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier, thank you, this makes sense.

What model should I use when generating p p > h j [QCD]? If I just run this command after launching MG5_aMC@NLO, it writes "No amplitude generated for this process". If I import heft, then for the syntax p p > h j there are 13 processes with 16 diagrams. However, if I use the syntax p p > h j [QCD], it writes "The model heft cannot handle loop processes".

this is for the "sm" model

Cheers,

Olivier

> On 20 Apr 2019, at 22:57, Maksym Ovchynnikov <email address hidden> wrote:
>
> Question #680335 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/680335
>
> Status: Answered => Open
>
> Maksym Ovchynnikov is still having a problem:
> Dear Olivier, thank you, this makes sense.
>
> What model should I use when generating p p > h j [QCD]? If I just run
> this command after launching MG5_aMC@NLO, it writes "No amplitude
> generated for this process". If I import heft, then for the syntax p p >
> h j there are 13 processes with 16 diagrams. However, if I use the
> syntax p p > h j [QCD], it writes "The model heft cannot handle loop
> processes".
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

After importing this model and trying to generate p p > h j [QCD] it writes:

"Error detected in "generate p p > h j [QCD]"
write debug file MG5_debug
If you need help with this issue please contact us on https://answers.launchpad.net/mg5amcnlo
MadGraph5Error :
 Automatic process-order determination lead to negative constraints:
 QED: 6, QCD: -4
 Please specify the coupling orders from the command line.
"

I do not reproduce that, which version of MG5aMC are you using?

Cheers,

Olivier

Version 3.0.1.

OK this version is still beta.

I would advise to use 2.6.5 which is the stable version

Cheers,

Olivier

> On 20 Apr 2019, at 23:57, Maksym Ovchynnikov <email address hidden> wrote:
>
> Question #680335 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/680335
>
> Status: Answered => Open
>
> Maksym Ovchynnikov is still having a problem:
> Version 3.0.1.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thank you.

Sorry for a lot of questions. I would like to clarify the following.

The process p p > h j, as far as I understand, goes through the same effective hgg vertex as the process p p > h. The only difference is that in p p > h both the gluon lines are incoming, while in p p > h j there are both incoming and outcoming lines. However, as far as I understand, the position of the gluon line does not affect the validity conditions of the hgg vertex, and again in the case of large Higgs energy the predictions based on the p p > h j process become untrustable. Where I am wrong?

You are correct heft also break down in this case. This is why I suggested to do the full loop computation

Cheers

Olivier

Get Outlook for iOS<https://aka.ms/o0ukef>

________________________________
From: <email address hidden> on behalf of Maksym Ovchynnikov <email address hidden>
Sent: Sunday, April 21, 2019 12:22 AM
To: Olivier Mattelaer
Subject: Re: [Question #680335]: Higgs gluon fusion production at NLO

Question #680335 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/680335

Status: Answered => Open

Maksym Ovchynnikov is still having a problem:
Thank you.

Sorry for a lot of questions. I would like to clarify the following.

The process p p > h j, as far as I understand, goes through the same
effective hgg vertex as the process p p > h. The only difference is that
in p p > h both the gluon lines are incoming, while in p p > h j there
are both incoming and outcoming lines. However, as far as I understand,
the position of the gluon line does not affect the validity conditions
of the hgg vertex, and again in the case of large Higgs energy the
predictions based on the p p > h j process become untrustable. Where I
am wrong?

--
You received this question notification because you are an answer
contact for MadGraph5_aMC@NLO.

Dear Olivier,

thank you very much for the answer. As far as I understand, the sm model provides more reliable calculations of these ggh induced processes.

At the moment I am not familiar with jet/parton merging/matching in MG5. Am I right that it does not work without pythia? Also, am I right that it helps to avoid double counting?

Correct

Get Outlook for iOS<https://aka.ms/o0ukef>
________________________________
From: <email address hidden> <email address hidden> on behalf of Maksym Ovchynnikov <email address hidden>
Sent: Sunday, April 21, 2019 10:03:32 AM
To: Olivier Mattelaer
Subject: Re: [Question #680335]: Higgs gluon fusion production at NLO

Question #680335 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/680335

    Status: Answered => Open

Maksym Ovchynnikov is still having a problem:
Dear Olivier,

thank you very much for the answer. As far as I understand, the sm model
provides more reliable calculations of these ggh induced processes.

At the moment I am not familiar with jet/parton merging/matching in MG5. Am I right that it does not work without pythia? Also, am I right that it helps to avoid double counting?

--
You received this question notification because you are an answer
contact for MadGraph5_aMC@NLO.

Thanks Olivier Mattelaer, that solved my question.

Dear Olivier, sorry for another questions.

1) Could you please tell me whether the following will give reliable prediction for the Higgs pT spectrum?

import model sm
generate p p > h [noborn=QCD]
add process p p > h j [noborn = QCD]
add process p p > h j j [noborn = QCD]

In run_card.dat the parameters are set to the following:

1 = ickkw
25 = xqcut

All the kinematical cuts are set to zero.

In the pythia card the QCUT parameter is
QCUT = 40

2) After launching process, what output should I use? pythia_events_lhe or pythia_evens_hep? Do I understand correctly that the former is automatically obtained from the latter using some built-in script?

Dear Olivier,

thank you for your help again.

I tried to reproduce the pT spectrum obtained in your paper 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 row 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?

I should mention here that there is the following error during using 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

Also.

 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]

And the total cross section obtained before pythia merging/matching is 27 pb. Therefore some problem is present at this stage...

Thanks Olivier Mattelaer, that solved my question.