Propagating EFT model into decays via MadSpin

> 1) Is what I am observing expected? If so, how? Or there is something I am missing here?

The total width of a particles does not depend of a particular decay that you select. So the computation of the total-width is fully indpendent of any syntax that you use.

Now in the auto-mode (which is called automatically by MadSpin if the full informattion --including BR-- are not available) we do not play with non definite-positive matrix-element. If you want to play with that, you have to compute the width with a more manual method (like it was done before the "auto" option).

Concerning the branching ratio, stuff are here more complicated.
indeed the BR of h > b b~ and the BR of h > b b~ NP=0 (or other coupling order restriction) might not be the same. The separation of the coupling order per coupling contribution is not compatible with the LHE banner (SLHA formaty) and therefore you do not have way to provide that information within the banner.

Since MadSpin use the information from the banner (and use the auto-width to fill such information if the banner is not complete, i.e. missing BR), this is also something that can not be handle automatically by madspin.

For both issue, one solution is to not use MadSpin.

Indeed if you use the following syntax
generate p p > t t~ h FCNC=0 DIM6<=1, (t > w+b dim6<=1, w+>all all), (t~ > w- b~ DIM6<=1), h > b b~ DIM6<=1
Then you have the same final state compare to MadSpin, the EXACT same set of Feynman Diagram that control the full-distribution (so the differential cross-section is the same up to statistical and theoretical uncertainty). The difference is that this syntax does not use any hard-coded branching ratio, but depend of the value of the coupling and of the total width.
so if you do
generate p p > t t~ h FCNC=0 DIM6<=1, (t > w+b dim6<=1, w+>all all), (t~ > w- b~ DIM6<=1), h > b b~ DIM6<=1
or
generate p p > t t~ h FCNC=0 DIM6<=1, (t > w+b dim6<=1, w+>all all), (t~ > w- b~ DIM6<=1), h > b b~ DIM6=0
You will get a different cross-section

Second solution is to use either spinmode=none or spinmode=onshell.
In neither of those case, we are using the BR, we do actually compute the partial width for whatever cuts/coupling orders constraint/.. that you give and then divide by the total-width.
The drawback of such method is that you will loose some precision (the offshell effect for the "onshell" and in top of that the spin-correlation between production and decay for "none")

> 2) In this case, I am doing one EFT insertion in the production and another one in the decay. How do I make sure that the EFT insertion is either in the production OR the decay but not both, i.e. avoid double insertions?

MadSpin can not prevent double insertion, this will break the narrow-width approximation (NWA) and will create a lot of convergence issue of the madspin procedure since the un-weighting efficiency of MadSpin will drop considerably. It is technically possible to do (we tried it). but the physical hyppothesis (NWA) was not met to make the code to correctly behave.

Consequently, here the only solution is to drop madspin and to use the decay chain syntax --the same as above--. In top, you can add restriction to avoid double insertion by adding at the end of the line "@0 DIM6<=1"

In place where you have "@1" or "@2" , ... you can use those obviously. If not this is a flag with only technical implication (i.e. no physics impact) which allows to flag events of the lhe with a number (you need to use more than one process for that). Here we use this as a pure separator.

Any coupling order restriction applied after the "@" applies to the full process and therefore avoid such double insertion.

This is technically a very old syntax (introduces in 1106.0522 <https://arxiv.org/abs/1106.0522> to keep full retro-compatibility with MG4).
but no-one used it for years up to ~6 month ago. A LOT of very old bug have been detected in the last 6 month related to this syntax. So I would strongly recommend to use the latest version of the code (at least 2.9.2)

Cheers,

Olivier

> On 22 Feb 2021, at 16:55, Hesham El Faham <email address hidden> wrote:
>
> New question #695693 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/695693
>
> Hi,
> I am trying to 'generate p p > t t~ h FCNC=0 DIM6=1' using the dim6top model. I want to decay both tops and the Higgs, I also want to do an EFT insertion into the decays. In MadSpin I do for example 'decay t > w+ b DIM6=1, w+ > all all', I also do 'decay h > all all DIM6=1' (I am not sure if that latter makes sense). I don't set the decay width in the param_card.dat to 'auto'. The following didn't change when I used 'auto', it just took a huge amount of running time to produce the same thing.
>
> I expected MadSpin to recalculate the widths of the tops and the Higgs according to the EFT model, and thus will get different BR and total widths than what I get when I don't use DIM6=1. However, that doesn't seem to be the case, i.e. whether I put DIM6=1 or not, the width and the BR are the same. I have a couple of questions please:
>
> 1) Is what I am observing expected? If so, how? Or there is something I am missing here?
> 2) In this case, I am doing one EFT insertion in the production and another one in the decay. How do I make sure that the EFT insertion is either in the production OR the decay but not both, i.e. avoid double insertions?
>
> Your help is much appreciated.
> Best,
> Hesham
>
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi Olivier,

Thanks for taking the time to write such a helpful response.

I will probably use the full decay chain syntax since I am anticipating I will have to avoid the double insertions at some point. You mentioned that such a syntax will use the coupling and the total width. That shall be of course the couplings according to dim6top, but for the total width, this will be the ones we know from the SM, right? Do I need to change those widths (even if I have to compute them 'manually') for I am to studying the process in the SMEFT?

If so, and if I managed to compute the total width 'manually' in SMEFT (for the Higgs for example), can I smoothly feed it in during the run without hidden issues that can spoil my results?

Best,
Hesham

Hi,

> That shall
> be of course the couplings according to dim6top, but for the total
> width, this will be the ones we know from the SM, right?

The width of EFT effect is a complex subject. (it is actually already in the SM).
To my point of view if you consider insertion in the production and decay then it does not make sense to use the SM total width. However using a non SM width is equivalent to have multiple insertion (since the width appears in the denominator).

So in https://arxiv.org/pdf/1803.10379.pdf <https://arxiv.org/pdf/1803.10379.pdf> (contribution 12 page 88), I actually discussed how to correctly handle such width effects. As far as I know this has been implemented in only one (set) of pubic UFO model (SMEFTSIM).

> If so, and if I managed to compute the total width 'manually' in SMEFT
> (for the Higgs for example), can I smoothly feed it in during the run
> without hidden issues that can spoil my results?

This technically depends of the UFO model that you consider.
In general the width is a free parameter of the model and you can set the value that you want in the param_card.
(you decide which type of width you include and at which precision).

Now the UFO author can decide to make some width an internal parameter of the model where the user can not modify such value. However it is mandatory to write such information in the param_card and in theory this is the responsability of the user to ensure that the value is consistent. We try to overwrite that value automatically in that case but they are few cases of workflow where this is diffficult to do and therefore we do not do it. In both cases (we overwrite you or fail to check if we need to overwrite you), you will get a warning.

Cheers,

Olivier

> On 24 Feb 2021, at 13:46, Hesham El Faham <email address hidden> wrote:
>
> Question #695693 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/695693
>
> Status: Answered => Open
>
> Hesham El Faham is still having a problem:
> Hi Olivier,
>
> Thanks for taking the time to write such a helpful response.
>
> I will probably use the full decay chain syntax since I am anticipating
> I will have to avoid the double insertions at some point. You mentioned
> that such a syntax will use the coupling and the total width. That shall
> be of course the couplings according to dim6top, but for the total
> width, this will be the ones we know from the SM, right? Do I need to
> change those widths (even if I have to compute them 'manually') for I am
> to studying the process in the SMEFT?
>
> If so, and if I managed to compute the total width 'manually' in SMEFT
> (for the Higgs for example), can I smoothly feed it in during the run
> without hidden issues that can spoil my results?
>
> Best,
> Hesham
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

https://arxiv.org/pdf/2012.11343.pdf

contains a lot of information on this topic as well.

Thanks Olivier, this was a very useful discussion.