aQGC: Discrepancy in SM and reweighted SM

My first guess is that this is due to the dynamical scale that you use.
You are using the default of MG5 which is Feynman diagram based.
Consequently, the scale choice depend of the process definition.

Can you check with another scale choice?

Cheers,

Olivier

> On 25 Apr 2017, at 23:37, Ramkrishna <email address hidden> wrote:
>
> New question #628556 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/628556
>
> Dear experts,
>
> I need your suggestion/help in understanding one of my comparison plot at generator level, where I tried to compare "SM" with the "reweighted SM".
>
> => "SM" : SM is the pure EWK sample generated for p p > W+ W- JJ. And this is decayed using madspin. W+ -> l+ nu and W- -> JJ. You can see the cards here [1].
>
> => "Reweighted SM": This is also pure EWK sample with the same channel and decayed using madspin, as described for "SM". But, this is generated with option NP = 1 for model "SM_LS_LM_LT_UFO" along with reweighting card. Before generation, in the customised card, I set one of the aQGC parameters to non-zero. And generated SM using reweight card along with
> another aQGC parameter.
>
> At [6], you can see that one of parameter number 11 (which corresponds to FT0) is fixed to 12.5e-12. And as the sample generation depends a lot on this initialization. So, I tried to vary it. I varied it by few values they are : 14.0e-12, 12.5e-12, 10e-12, and 8e-12.
>
> Ideally, all of them should match. But, if you look at the plot [3] (same plot with extended range [4], [5]), then it started deviating from 750 GeV onwards.
>
> Do you think that I am doing something wrong? If not, then why it is behaving like this?
>
>
> Please let me know if you need more information.
>
> with regards,
> Ram
>
> [1] SM gridpack:
> https://github.com/ram1123/genproductions/tree/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_SM_RunCardChanged
>
> [2] aQGC gridpack:
> https://github.com/ram1123/genproductions/tree/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RunCardChanged
>
> [3] https://drive.google.com/open?id=0B2Yvp8xMAYECeTZvVmI2T3ZIQ0U
>
> [4] https://drive.google.com/open?id=0B2Yvp8xMAYECVWdBeE1nSW9OV0E
>
> [5] https://drive.google.com/open?id=0B2Yvp8xMAYECTC1lWkFrWVdJVTA
>
> [6] https://github.com/ram1123/genproductions/blob/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RunCardChanged/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RunCardChanged_customizecards.dat
>
>
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier,

Thanks for quick reply.

Are you saying instead of setting FT0 try to set any other variable?

Also, is there any cleaver way to guess approx. the value of these parameter for initialization?

with regards,
Ram

No.

I’m just speaking of the QCD scale.

Cheers,

Olivier
> On 25 Apr 2017, at 23:53, Ramkrishna <email address hidden> wrote:
>
> Question #628556 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/628556
>
> Status: Answered => Open
>
> Ramkrishna is still having a problem:
> Dear Olivier,
>
> Thanks for quick reply.
>
> Are you saying instead of setting FT0 try to set any other variable?
>
> Also, is there any cleaver way to guess approx. the value of these
> parameter for initialization?
>
> with regards,
> Ram
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier,

Are you saying about "as" in the block below:

For SM:

                   41 Block sminputs
                   42 1 1.325070e+02 # aEWM1
                   43 2 1.166390e-05 # Gf
                   44 3 1.180000e-01 # aS

For aQGC:

                  72 Block sminputs
                  73 1 1.279000e+02 # aEWM1
                  74 2 1.166370e-05 # Gf
                  75 3 1.184000e-01 # aS

Hi,

that value of aS is never used since
1) we use the value of aS (at MZ) given by the PDF and not that one (we only use it when the user run at fixed energy )
2) the problem here the evolution of aS at different scale. i.e. for each event at which scale should we evaluate aS?

This is control by the run_card by couple of parameter the most important one is
dynamical_scale_choice
the default is at -1 which is a CKKW back clustering based on a given Feynman diagram (choose thanks to the diagram enhancement method).
This method is likely to create your bias in your case. So you should use a simpler (purely kinematic) running scheme (like Ht/2 which is setting the above parameter on 3.

Cheers,

Olivier

Dear Olivier,

I did not have much knowledge about generators so asking these small details. Sorry for that.

So, as I understand you said to change the line:

 -1 = dynamical_scale_choice

instead of -1, I should use just Ht/2? Did the madgraph recognise Ht? Or, else how I need to define it.

with regards,
Ram

Hi Olivier,

As I understand it your suggestion is to run with dynamical_scale_choice = 3, right?

In your opinion does this affect mean that only the SM point is unreliable? Would you still expect the comparison of reweighted aQGC points to a separate SM sample to be meaningful?

Thanks,

Kenneth

Hi Kenneth,

yes 3 is HT/2, if you want HT, then you can set it on 2

I copy here the help present in the code:

## Information on parameter dynamical_scale_choice from class RunCardLO
## current value: -1 (parameter should be of type <type 'int'>)
## '-1' is based on CKKW back clustering (following feynman diagram).
## '1' is the sum of transverse energy.
## '2' is HT (sum of the transverse mass)
## '3' is HT/2
## '4' is the center of mass energy
## This value is considered as been set by the user

>In your opinion does this affect mean that only the SM point is unreliable? Would you still expect the comparison of reweighted aQGC points to a separate SM sample to be meaningful?

Everything is realiable, even if you use -1. The only point is that -1 for the SM will not pick the same scale as -1 for EFT (at least not always) and therefore both predictions will only match within scale uncertainty (which can be huge). Nothing is wrong but that property is of course not nice for method validation.

Cheers,

Olivier

Dear Olivier,

I tried to fix the dynamical scale to 3 (which is HT/2). You can see the comparison plot at

https://drive.google.com/file/d/0B2Yvp8xMAYECRlpkdmN5bHptcVU/view?usp=sharing

It seems that now the discrepancies between them increased somehow.

You can see my run cards at [3] and [4] (all cards here [1] and [2]).

What do you suggest now?

Let me know if you need more information.

Thanks & regards,
Ram

[1] https://github.com/ram1123/genproductions/tree/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RwgtValidation

[2] https://github.com/ram1123/genproductions/tree/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_SM_RwgtValidation

[3] https://github.com/ram1123/genproductions/blob/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RwgtValidation/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RwgtValidation_run_card.dat

[4] https://github.com/ram1123/genproductions/blob/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_SM_RwgtValidation/aQGC_WPlepWMhadJJ_EWK_LO_SM_RwgtValidation_run_card.dat

Hi,

Can you put your model somewhere online, such that I can take a look myself?
I’m not sure to understand the plot that you are doing, so I would like to do such plot myself.

Cheers,

Olivier

> On 1 May 2017, at 01:37, Ramkrishna <email address hidden> wrote:
>
> Question #628556 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/628556
>
> Status: Answered => Open
>
> Ramkrishna is still having a problem:
> Dear Olivier,
>
> I tried to fix the dynamical scale to 3 (which is HT/2). You can see the
> comparison plot at
>
> https://drive.google.com/file/d/0B2Yvp8xMAYECRlpkdmN5bHptcVU/view?usp=sharing
>
> It seems that now the discrepancies between them increased somehow.
>
> You can see my run cards at [3] and [4] (all cards here [1] and [2]).
>
> What do you suggest now?
>
> Let me know if you need more information.
>
> Thanks & regards,
> Ram
>
>
>
> [1] https://github.com/ram1123/genproductions/tree/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV/AnomalousCouplings-SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RwgtValidation
>
> [2]
> https://github.com/ram1123/genproductions/tree/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV
> /AnomalousCouplings-
> SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_SM_RwgtValidation
>
> [3]
> https://github.com/ram1123/genproductions/blob/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV
> /AnomalousCouplings-
> SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RwgtValidation/aQGC_WPlepWMhadJJ_EWK_LO_NPle1_RwgtValidation_run_card.dat
>
> [4]
> https://github.com/ram1123/genproductions/blob/mg240/bin/MadGraph5_aMCatNLO/cards/production/13TeV
> /AnomalousCouplings-
> SMP/aQGC_WpWm_SemiLeptonic/aQGC_WPlepWMhadJJ_EWK_LO_SM_RwgtValidation/aQGC_WPlepWMhadJJ_EWK_LO_SM_RwgtValidation_run_card.dat
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier,

You can find the model at link:

https://drive.google.com/file/d/0B2Yvp8xMAYECLU1ySVpVS0QtVWs/view?usp=sharing

with regards,
Ram

Dear Olivier,

If you looked into this, please let us know if I am doing something wrong, or what's happening here?

Thanks & regards,
Ram

Hi Ram,

I did take a look but the i forget about it.
Indeed something was not running fine in my tests. Therefore, I do not think that you did something wrong.
This being said, I'm not 100% sure that this is related to a bug.
It can also be that the two model are too different and therefore needs very large statistic of events to provide reasonable results.

Cheers,

Olivier

Dear Olivier,

The statistics that I used is

Entries SM = 1670400
Entries SM fixed QCD = 697800
Entries aQGC fixed QCD = 403200

Do you think that this statistics is low? If so, I can try to generate more and let you know.

Hi,

My tests were with 10k events, and in that case, the results were wrong by order of magnitude.
Your result seems better than mine since you have a nice agreement up to 1.2 TeV and after that it starts to have huge discrepancy.
So I would say that your stat is already quite high. What I would do actually is to split your sample in 5 different sample, and look at the variance between the 5 reweighed sample. This will give you a nice estimation of the statistical error associated with the re-weighted sample and a clear indication of how much you are sensitive to that.

Note that in your plot the interesting ratio is actually RWGT(fixed QCD)/ SM fixed QCD.
Plotting the (normalise) distribution for aQGC can be interesting.

Cheers,

Olivier

> On 15 May 2017, at 16:14, Ramkrishna <email address hidden> wrote:
>
> Question #628556 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/628556
>
> Status: Answered => Open
>
> Ramkrishna is still having a problem:
> Dear Olivier,
>
> The statistics that I used is
>
> Entries SM = 1670400
> Entries SM fixed QCD = 697800
> Entries aQGC fixed QCD = 403200
>
> Do you think that this statistics is low? If so, I can try to generate
> more and let you know.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier,

I will try your suggestion and let you know my finding.

Also, I would like to share few distributions for dynamical QCD scale. Here I generated sample 4 different times by choosing different initialization parameter. And compared them. You can have a look at it here:

https://docs.google.com/presentation/d/1gelz95nywkhivuQ_k3JncSDZlgoYLSqB0JXsef7Zh-k/edit?usp=sharing

Let me know what do you think? And what do you suggest about the choosing the initialization of parameter before generation sample? As the distribution varies with different parameters so it might be contributing along with some other things in the deviation. What do you think?

with regards,
Ram

Hi,

I’m not sure to understand your legend,
What does “Reweighed SM (FT0=8.0)” means

I’m confused about the SM here. From the upper left, It looks like that you are not plotting the SM.
And from the information in paranthesis, it seems that the initial sample is not SM either.
Should legend be “Reweighed FROM FT0=8.0” ?

From what I see from your plot, it can indeed be a statistical problem.
Can you show the plot fro the initial sample?

Cheers,

Olivier

> On 15 May 2017, at 17:03, Ramkrishna <email address hidden> wrote:
>
> Question #628556 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/628556
>
> Status: Answered => Open
>
> Ramkrishna is still having a problem:
> Dear Olivier,
>
> I will try your suggestion and let you know my finding.
>
> Also, I would like to share few distributions for dynamical QCD scale.
> Here I generated sample 4 different times by choosing different
> initialization parameter. And compared them. You can have a look at it
> here:
>
> https://docs.google.com/presentation/d/1gelz95nywkhivuQ_k3JncSDZlgoYLSqB0JXsef7Zh-k/edit?usp=sharing
>
> Let me know what do you think? And what do you suggest about the
> choosing the initialization of parameter before generation sample? As
> the distribution varies with different parameters so it might be
> contributing along with some other things in the deviation. What do you
> think?
>
> with regards,
> Ram
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Olivier,

Sorry for the confusion.

"Reweighed SM (FT0=8.0)" means "Reweighted from FT0 = 8.0x10^{-12}".

"plot from the initial sample" => Do you mean that just plot distribution without any weight?

with regards,
Ram

> "plot from the initial sample" => Do you mean that just plot
> distribution without any weight?

Yes indeed.

Cheers,

Olivier

> On 16 May 2017, at 11:27, Ramkrishna <email address hidden> wrote:
>
> Question #628556 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/628556
>
> Status: Answered => Open
>
> Ramkrishna is still having a problem:
> Dear Olivier,
>
> Sorry for the confusion.
>
> "Reweighed SM (FT0=8.0)" means "Reweighted from FT0 = 8.0x10^{-12}".
>
> "plot from the initial sample" => Do you mean that just plot
> distribution without any weight?
>
> with regards,
> Ram
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Dear Oliver,

Sorry for late response.

Please have a look at the plot that you asked.

Plot [1]: Here I plotted all the distribution without weight. Each distribution I generated with different initialization parameters.

[1] https://drive.google.com/file/d/0B2Yvp8xMAYECU3FSUE9pMUl5TlE/view?usp=sharing

with regards,
Ram

This question was expired because it remained in the 'Open' state without activity for the last 15 days.