QCUT value for ttll process

Asked by Nicolas Tonon on 2020-05-22

Dear experts,

I want to generate a ttll/vv sample using a LO EFT model, with an extra parton (to better reproduce NLO predictions). The process is generated in the 5FS, with a cut inv_mll > 10.

Excerpt from proc card:
=============
define p = p b b~
define j = p
define l+ = e+ mu+ ta+
define l- = e- mu- ta-
define vl = ve vm vt
define vl~ = ve~ vm~ vt~

import model dim6top_LO_UFO
generate p p > t t~ l+ l- / h FCNC=0 DIM6=1 @0
add process p p > t t~ vl vl~ / h FCNC=0 DIM6=1 @1
add process p p > t t~ l+ l- j / h FCNC=0 DIM6=1 @2
add process p p > t t~ vl vl~ j / h FCNC=0 DIM6=1 @3
=============

Excerpt from run card:
=============
#*********************************************************************
# Matching parameter (MLM only)
#*********************************************************************
 1 = ickkw ! 0 no matching, 1 MLM
 1.0 = alpsfact ! scale factor for QCD emission vx
 False = chcluster ! cluster only according to channel diag
 5 = asrwgtflavor ! highest quark flavor for a_s reweight
 False = auto_ptj_mjj ! Automatic setting of ptj and mjj if xqcut >0
                                   ! (turn off for VBF and single top processes)
 30.0 = xqcut ! minimum kt jet measure between partons
...
 10.0 = mmll ! min invariant mass of l+l- (same flavour) lepton pair
...
 5 = maxjetflavor ! Maximum jet pdg code
=============

The addition of the extra parton (@2 and @3) requires matching, and I am unsure what would be the most relevant choice for the QCUT scale in this scenario.

From the matching doc [0], I read that a value QCUT=30 is suggested for tt~, which suggests that I should use QCUT >= 30 in my case.

I already produced DJR plots for 3 QCUT values (using 30K LHE events each):
a) QCUT=20 (xqcut=10) [1]
b) QCUT=40 (xqcut=30) [2]
c) QCUT=50 (xqcut=30) [3]

Scenario b) does not exhibit a smooth transition, and c) does not look too satisfying neither. I may retry with QCUT=60 to see if it works better.
However I am surprised that the scenario a) I was using initially, which corresponds to a small QCUT value, seems to provide the smoothest transition, whereas I was expecting that a higher value is required for this kind of processes.

Could you please let me know what are your thoughts on this ? In particular, what do you think would be the most sensible QCUT choice for such a process, and are there figures of merits other than the DJR plot to assess the correctness of the matching procedure ?

Thank you, cheers,
Nicolas

[0] https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/IntroMatching
[1] https://cernbox.cern.ch/index.php/s/rcpXpB2Y0YIC4Op
[2] https://cernbox.cern.ch/index.php/s/5TdqCUszdrzCQdQ
[3] https://cernbox.cern.ch/index.php/s/ZnvgPyji5imRjTf

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2020-05-24
Last query:
2020-05-24
Last reply:
2020-05-23

Hi,

Do you have chromo-dynamic operator in this model?
If yes MLM will certainly not work (and if you use the latest version of MG5aMC we have prevent the user to even try to run MLM for such type of model.
In general, I'm not fully sure that such type of model support MLM features out of the box.

Otherwise, I would indeed expect a higer value of QCut like 100GeV

Cheers,

Olivier

> On 22 May 2020, at 12:20, Nicolas Tonon <email address hidden> wrote:
>
> New question #690892 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/690892
>
> Dear experts,
>
> I want to generate a ttll/vv sample using a LO EFT model, with an extra parton (to better reproduce NLO predictions). The process is generated in the 5FS, with a cut inv_mll > 10.
>
> Excerpt from proc card:
> =============
> define p = p b b~
> define j = p
> define l+ = e+ mu+ ta+
> define l- = e- mu- ta-
> define vl = ve vm vt
> define vl~ = ve~ vm~ vt~
>
> import model dim6top_LO_UFO
> generate p p > t t~ l+ l- / h FCNC=0 DIM6=1 @0
> add process p p > t t~ vl vl~ / h FCNC=0 DIM6=1 @1
> add process p p > t t~ l+ l- j / h FCNC=0 DIM6=1 @2
> add process p p > t t~ vl vl~ j / h FCNC=0 DIM6=1 @3
> =============
>
>
> Excerpt from run card:
> =============
> #*********************************************************************
> # Matching parameter (MLM only)
> #*********************************************************************
> 1 = ickkw ! 0 no matching, 1 MLM
> 1.0 = alpsfact ! scale factor for QCD emission vx
> False = chcluster ! cluster only according to channel diag
> 5 = asrwgtflavor ! highest quark flavor for a_s reweight
> False = auto_ptj_mjj ! Automatic setting of ptj and mjj if xqcut >0
> ! (turn off for VBF and single top processes)
> 30.0 = xqcut ! minimum kt jet measure between partons
> ...
> 10.0 = mmll ! min invariant mass of l+l- (same flavour) lepton pair
> ...
> 5 = maxjetflavor ! Maximum jet pdg code
> =============
>
> The addition of the extra parton (@2 and @3) requires matching, and I am unsure what would be the most relevant choice for the QCUT scale in this scenario.
>
>> From the matching doc [0], I read that a value QCUT=30 is suggested for tt~, which suggests that I should use QCUT >= 30 in my case.
>
> I already produced DJR plots for 3 QCUT values (using 30K LHE events each):
> a) QCUT=20 (xqcut=10) [1]
> b) QCUT=40 (xqcut=30) [2]
> c) QCUT=50 (xqcut=30) [3]
>
> Scenario b) does not exhibit a smooth transition, and c) does not look too satisfying neither. I may retry with QCUT=60 to see if it works better.
> However I am surprised that the scenario a) I was using initially, which corresponds to a small QCUT value, seems to provide the smoothest transition, whereas I was expecting that a higher value is required for this kind of processes.
>
> Could you please let me know what are your thoughts on this ? In particular, what do you think would be the most sensible QCUT choice for such a process, and are there figures of merits other than the DJR plot to assess the correctness of the matching procedure ?
>
> Thank you, cheers,
> Nicolas
>
> [0] https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/IntroMatching
> [1] https://cernbox.cern.ch/index.php/s/rcpXpB2Y0YIC4Op
> [2] https://cernbox.cern.ch/index.php/s/5TdqCUszdrzCQdQ
> [3] https://cernbox.cern.ch/index.php/s/ZnvgPyji5imRjTf
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Nicolas Tonon (ntonon) said : #2

Hi Olivier,

thank you for your feedback. Although I do not consider it, the chromo-dynamic operator is included in the model; the authors have adapted the model so that matching should work.

I also produced the DJR plot [0] using QCUT=100, and the distribution indeed looks pretty smooth.
In such cases where the DJR plot looks smooth both for QCUT values of 20 and 100 GeV (but not necessarily for intermediate values), is there a figure of merit to decide the choice for QCUT ?

Cheers,
Nicolas

[0] https://cernbox.cern.ch/index.php/s/yVBSLZh74JvnscE

Hi,

> thank you for your feedback. Although I do not consider it, the chromo-
> dynamic operator is included in the model; the authors have adapted the
> model so that matching should work.

I would be interested to read the paper where they explain what they did to make this consistent.

> I also produced the DJR plot [0] using QCUT=100, and the distribution indeed looks pretty smooth.
> In such cases where the DJR plot looks smooth both for QCUT values of 20 and 100 GeV (but not necessarily for intermediate values), is there a figure of merit to decide the choice for QCUT ?

Matching uncertainty should be your figure of merit.
In this case I would not really trust a QCUT at 20GeV

Cheers,

Olivier

> On 23 May 2020, at 16:30, Nicolas Tonon <email address hidden> wrote:
>
> Question #690892 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/690892
>
> Status: Answered => Open
>
> Nicolas Tonon is still having a problem:
> Hi Olivier,
>
> thank you for your feedback. Although I do not consider it, the chromo-
> dynamic operator is included in the model; the authors have adapted the
> model so that matching should work.
>
> I also produced the DJR plot [0] using QCUT=100, and the distribution indeed looks pretty smooth.
> In such cases where the DJR plot looks smooth both for QCUT values of 20 and 100 GeV (but not necessarily for intermediate values), is there a figure of merit to decide the choice for QCUT ?
>
> Cheers,
> Nicolas
>
> [0] https://cernbox.cern.ch/index.php/s/yVBSLZh74JvnscE
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Nicolas Tonon (ntonon) said : #4

Thanks Olivier Mattelaer, that solved my question.

Nicolas Tonon (ntonon) said : #5

Hi Olivier, thanks.

> I would be interested to read the paper where they explain what they did to make this consistent.

I was rather referring to a recent fix added to the UFO model (see [0] > 'Changelog') :
"added a configuration option to normalize all chromo-dipole operator coefficients (ctG and FCNC cousins) with a factor of the strong coupling G. This is currently required for jet merging not to crash. Note however that the jet merging code was not written with non-SM interactions in mind. So results it produces with chromo-dipole operators should be validated. Note also that G runs and that its value at the scale MZ depends on the parton distribution function employed. See configuration.py for the hack, to be used with great care."

In practice, the order of couplings involving the chromodynamic operator are modified in this manner :
"order = {'FCNC':1,'QED':1, 'QCD':1 if configuration.norm_chromo_gs else 0})"

I have no expertise to judge whether this addresses the issue you mentioned, although that was my undertanding.

Cheers,
Nicolas

[0] https://feynrules.irmp.ucl.ac.be/wiki/dim6top

So in fact they have no clue if this produces correct physics.
They just implement a way to avoid the restrictions of MadGraph to be trigger (and to make such computation to crash)

Cheers,

Olivier

> On 24 May 2020, at 11:50, Nicolas Tonon <email address hidden> wrote:
>
> Question #690892 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/690892
>
> Nicolas Tonon posted a new comment:
> Hi Olivier, thanks.
>
>> I would be interested to read the paper where they explain what they
> did to make this consistent.
>
> I was rather referring to a recent fix added to the UFO model (see [0] > 'Changelog') :
> "added a configuration option to normalize all chromo-dipole operator coefficients (ctG and FCNC cousins) with a factor of the strong coupling G. This is currently required for jet merging not to crash. Note however that the jet merging code was not written with non-SM interactions in mind. So results it produces with chromo-dipole operators should be validated. Note also that G runs and that its value at the scale MZ depends on the parton distribution function employed. See configuration.py for the hack, to be used with great care."
>
> In practice, the order of couplings involving the chromodynamic operator are modified in this manner :
> "order = {'FCNC':1,'QED':1, 'QCD':1 if configuration.norm_chromo_gs else 0})"
>
> I have no expertise to judge whether this addresses the issue you
> mentioned, although that was my undertanding.
>
> Cheers,
> Nicolas
>
> [0] https://feynrules.irmp.ucl.ac.be/wiki/dim6top
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.