AQGC SM_LT8_LT9 EWK cross section unrealistic NEEDS FIX URGENT *

Asked by Saif Mustafa

Hi,

I'm using the SM_LT8_LT9 model from: http://feynrules.irmp.ucl.ac.be/wiki/AnomalousGaugeCoupling

where I manually set all EFT parameters = 1.00000e-99 (off) first.

Then I set FT8 = 1.0 and the cross section value I got = .28581E+21
Then I set FT8 = 5.0 and the cross section value I got = .71453E+22

Then I tried FT9 = 1.0 and the cross section = N/A as it did not even generate properly (too small number of events I think so could not be fed into Rivet to produce kinematics histograms)

Then I tried FT9 = 5.0 and the cross section = .15209E+22

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

As you can see from the above EWK cross-sections, these are very unrealistic and much too large for an EWK process. The QCD processes look to be of the correct magnitude though ... How do I fix these EWK cross-sections???? URGENT *****

thank you,

S. Mustafa.

Question information

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

A lot of cross-section are indeed infinite at LO accuracy (even in the standard model)
So such behavior can happen in many situationn if you do not apply the correct cut to kill the divergence of the matrix-element (to kill the part of the phase-space where the theory is not perturbative).

This is the best comment that I can give you with that few information (if you need help, please help me to help you by giving me much more details).

Cheers,

Olivier

PS: If you think this is a problem of the model, then you need to use another model and/or contact the author of the model

> On 7 Aug 2020, at 21:15, Saif Mustafa <email address hidden> wrote:
>
> New question #692269 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/692269
>
> Hi,
>
> I'm using the SM_LT8_LT9 model from: http://feynrules.irmp.ucl.ac.be/wiki/AnomalousGaugeCoupling
>
> where I manually set all EFT parameters = 1.00000e-99 (off) first.
>
>
> Then I set FT8 = 1.0 and the cross section value I got = .28581E+21
> Then I set FT8 = 5.0 and the cross section value I got = .71453E+22
>
> Then I tried FT9 = 1.0 and the cross section = N/A as it did not even generate properly (too small number of events I think so could not be fed into Rivet to produce kinematics histograms)
>
> Then I tried FT9 = 5.0 and the cross section = .15209E+22
>
>
> -----------------------------
>
>
> As you can see from the above EWK cross-sections, these are very unrealistic and much too large for an EWK process. The QCD processes look to be of the correct magnitude though ... How do I fix these EWK cross-sections???? URGENT *****
>
>
> thank you,
>
>
> S. Mustafa.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
Saif Mustafa (ucapmus) said :
#2

Hi Olivier,

I have heard that to fix this issue I have to enable the so-called 'Feynman Gauge' - where can this be done in MG5?

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

If your model does not support unitary gauge,
Then the easiest is to edit the model to let madgraph know that your model only support that gauge choice.

For that you can edit the __init__.py file of your UFO model
and modify the line
gauge=[0,1]
and set it to
gauge=[1]

Cheers,

Olivier

PS: I would advise you to use the "check process X y > A B C" in order to check gauge invariance and lorentz invariance of your process. If your model only support Feynman gauge but does not specify it, you should see some issue identified by that syntax.

> On 8 Aug 2020, at 12:55, Saif Mustafa <email address hidden> wrote:
>
> Question #692269 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/692269
>
> Status: Answered => Open
>
> Saif Mustafa is still having a problem:
> Hi Olivier,
>
>
> I have heard that to fix this issue I have to enable the so-called 'Feynman Gauge' - where can this be done in MG5?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
Saif Mustafa (ucapmus) said :
#4

Hi Olivier,

I am testing EWK variations from the SM limit of this AQGC model, and the author recommended to start from parameter values of 1.0e-12.

This gave me much better cross-sections.

I am now performing chi-squared minimisation tests to find the optimum kinematic discriminating variables for certain parameters where I require small chi-squared values for a delta-chi sqaured of 4 for 2sigma uncertainty.

However, when I change the parameter values in the AQGC model i.e. FT8 and FT9 EFT parameters to values = 1e-15 there is no variation at all in the chi-squared from the EWK SM limit for this corresponding model? Is there something I am missing that I haven't enabled in this model to show variations for lower parameter order parameter values in MG5?

To clarify, I am producing yoda files from these mg5 runs using Rivet, and then plotting the chi-squared of some amount of kinematic variables for the given parameter values against the Standard Model limit i.e. all parameters = 0.0 but FT8 / FT9 varied for parameter values. At 1e-12 values for the FT8 / FT9 parameters I see variations for the chi-squared EWK, but for smaller values e.g. 1e-15 I see no variation in EWK. Why is this?

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#5

Hi,

Did you look at the conntribution of the SM only?
One possible issue is that at some point you will be dominated by the SM contribution and therefore you will loose sensitivity to the EFT operator.
This is particularly a problem for dimension 8 operator. Those are typically either fully dominated by the SM (and therefore you are not sensitive to the operator) or you are in the regime where the EFT is not valid (i.e. you break unitarity).

Obviously with number like 1e-15 one can simply worry about numerical accuracy of the code since any formula like (1+FT8) will simply be equal to 1 for value at 1e-15. (but in general the "1" is the SM value so this is equivalent to the above case).

Cheers,

Olivier

> On 18 Aug 2020, at 19:10, Saif Mustafa <email address hidden> wrote:
>
> Question #692269 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/692269
>
> Status: Answered => Open
>
> Saif Mustafa is still having a problem:
> Hi Olivier,
>
> I am testing EWK variations from the SM limit of this AQGC model, and
> the author recommended to start from parameter values of 1.0e-12.
>
> This gave me much better cross-sections.
>
> I am now performing chi-squared minimisation tests to find the optimum
> kinematic discriminating variables for certain parameters where I
> require small chi-squared values for a delta-chi sqaured of 4 for 2sigma
> uncertainty.
>
>
> However, when I change the parameter values in the AQGC model i.e. FT8 and FT9 EFT parameters to values = 1e-15 there is no variation at all in the chi-squared from the EWK SM limit for this corresponding model? Is there something I am missing that I haven't enabled in this model to show variations for lower parameter order parameter values in MG5?
>
> To clarify, I am producing yoda files from these mg5 runs using Rivet,
> and then plotting the chi-squared of some amount of kinematic variables
> for the given parameter values against the Standard Model limit i.e. all
> parameters = 0.0 but FT8 / FT9 varied for parameter values. At 1e-12
> values for the FT8 / FT9 parameters I see variations for the chi-squared
> EWK, but for smaller values e.g. 1e-15 I see no variation in EWK. Why is
> this?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Can you help with this problem?

Provide an answer of your own, or ask Saif Mustafa for more information if necessary.

To post a message you must log in.