Off-shell intermediate bosons for Drell-Yan processes

For the impact on the parton-shower side, the best is likely to ask the parton-shower author how their code is impacted in presence/absence of intermediate particle written into the file.

I know that for BSM particles, we do have a security that forbid to write the particle if the width is larger than 10% of the mass.
And experimentalist are able to observe that threshold in their observables due to the change that it implies.

The basic impact of writing a particle in the lhef is that you put a constraint on the way the parton-shower does emit radiation and does the recoil.

Now be carefull that setting bwcutoff too large can also make MG5aMC to fail to integrate correctly (and typically miss some contribution of the cross-section) --typically under-estimating in the same time the error so it can be left as un-detected.

My advice would be to NEVER use bwcutoff too large (i.e. never more than 30). And if you need to have the particle written in the lhe file, then add a post-processing script where you add such particles (like CMS was doing in the past).

Cheers,

Olivier

Hi Olivier,

Thank you for the swift response, and your information/advice.

Do you have any advice on how to add the post-processing script to add the off-shell intermediate bosons? Would I simply produce a script to edit the LHE file and make sure there is an intermediate particle for each event, before I perform the parton showering? In which case, how is this different from simply increasing the BW cutoff? Both would seem to result in an intermediate particle for each event.

Thanks,
James

> Would I simply produce a script to
> edit the LHE file and make sure there is an intermediate particle for
> each event, before I perform the parton showering?

Yes.

> In which case, how is
> this different from simply increasing the BW cutoff? Both would seem to
> result in an intermediate particle for each event.

The BW cutoff has also impacts on the method used to integrate the phase-space.
Not only on how to write the file. So setting bwcutoff to very large value can lead to a bad
parametrization of the phase-space and consequently to wrong cross-section.
(someone reported the problem last week)

The second difference is that you are then in full control on how to do it and you only write the particle that you do want (since setting bwcutoff too large might lead to spurious writing of unwanted particles).

Cheers,

Olivier

> On 28 Aug 2023, at 12:40, James <email address hidden> wrote:
>
> Question #707760 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/707760
>
> James posted a new comment:
> Hi Olivier,
>
> Thank you for the swift response, and your information/advice.
>
> Do you have any advice on how to add the post-processing script to add
> the off-shell intermediate bosons? Would I simply produce a script to
> edit the LHE file and make sure there is an intermediate particle for
> each event, before I perform the parton showering? In which case, how is
> this different from simply increasing the BW cutoff? Both would seem to
> result in an intermediate particle for each event.
>
> Thanks,
> James
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thanks Olivier Mattelaer, that solved my question.

Hi Olivier,

Thank you for your help so far. I had a go at writing a post-processing script to modify the LHE file to include the intermediate boson with the correct properties in each event. The resulting LHE file is as expected. I am generating the charged-current Drell-Yan process in five different cases: one is the SM case, and the other four correspond to having a one of four Wilson coefficients (SMEFT operators) present. I proceed to plot kinematic distributions of different variables for each generation process, applying the mc_weight as a weight for each process.

Without modifying the LHE file, I observe a series of kinematic distributions which are similar to those of the neutral-current Drell-Yan processes, which is to be expected. In particular, the invariant mass distribution of the process with the Wilson coefficient clq3 present has an increasing cross section with invariant mass. This results in the ratio of clq3 to SM increasing as invariant mass increases.

However, after modifying the LHE file with my post processing script, this increase in the ratio is no long observed. Instead the ratio of clq3 to SM is constant, at roughly 4, and doesn’t increase with invariant mass. Although I now observe the correct delta phi distribution after the post-processing, I am no longer observing the expected effect in other distributions.

Do you have any idea why I am observing this strange effect? I can’t see how modifying the LHE file to include an intermediate particle in each event would alter the ratio of clq3 to SM. Do you have any ideas how to overcome this effect?

Many thanks,
James

Hi,

In which type of plot?
For plot of the lhef file, this would indicate an error in your script.
If this is a plot after parton-shower that is possible since you can change the parton-shower algorithm by adding particle.

Cheers,

Olivier

> On 25 Sep 2023, at 11:40, James <email address hidden> wrote:
>
> Question #707760 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/707760
>
> Status: Solved => Open
>
> James is still having a problem:
> Hi Olivier,
>
> Thank you for your help so far. I had a go at writing a post-processing
> script to modify the LHE file to include the intermediate boson with the
> correct properties in each event. The resulting LHE file is as expected.
> I am generating the charged-current Drell-Yan process in five different
> cases: one is the SM case, and the other four correspond to having a one
> of four Wilson coefficients (SMEFT operators) present. I proceed to plot
> kinematic distributions of different variables for each generation
> process, applying the mc_weight as a weight for each process.
>
> Without modifying the LHE file, I observe a series of kinematic
> distributions which are similar to those of the neutral-current Drell-
> Yan processes, which is to be expected. In particular, the invariant
> mass distribution of the process with the Wilson coefficient clq3
> present has an increasing cross section with invariant mass. This
> results in the ratio of clq3 to SM increasing as invariant mass
> increases.
>
> However, after modifying the LHE file with my post processing script,
> this increase in the ratio is no long observed. Instead the ratio of
> clq3 to SM is constant, at roughly 4, and doesn’t increase with
> invariant mass. Although I now observe the correct delta phi
> distribution after the post-processing, I am no longer observing the
> expected effect in other distributions.
>
> Do you have any idea why I am observing this strange effect? I can’t see
> how modifying the LHE file to include an intermediate particle in each
> event would alter the ratio of clq3 to SM. Do you have any ideas how to
> overcome this effect?
>
> Many thanks,
> James
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi,

These are plots produced after showering.

Thanks,
James

Hi,

Then I guess this means that your observable is quite (too) sensitive to the detail of the parton-shower.
And that the associate theory prediction is impacted by theory/modelling uncertainty.

Maybe they are internal option of the parton-shower to reduce such dependence/... but you would need a parton-shower expert on such topic.

Cheers,

Olivier