NLO QCD for elastic photon processes

Asked by James Howarth on 2021-01-11

Hi All,

I would like to ask about a warning that occurs when trying to do NLO QCD effects in processes with elastic photons. I can run the following:

a a > t t~ [QCD]

without any issues (and I get a k-factor relative to LO of ~1.2) but if I try to run:

a p > t t~ [QCD]

A warning pops up to say that DIS-type interactions are not supported at NLO. Now, I can easily disable this warning and it runs fine, and I again get a k-factor of ~1.2, but obviously I'd like to understand why this warning is there. The process I'm generating isn't really DIS at all, and the check seems to be simply looking to see if someone has turned on elastic photon from protons and also asked for NLO.

Could you explain why this process isn't supported at NLO? When it seems to work fine and give sensible results?

Cheers,

Jay

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
James Howarth
Solved:
2021-01-15
Last query:
2021-01-15
Last reply:
2021-01-15

I guess that you are not running the latest version of the code. Are you?

Olivier

> On 11 Jan 2021, at 13:45, James Howarth <email address hidden> wrote:
>
> New question #694892 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/694892
>
> Hi All,
>
> I would like to ask about a warning that occurs when trying to do NLO QCD effects in processes with elastic photons. I can run the following:
>
> a a > t t~ [QCD]
>
> without any issues (and I get a k-factor relative to LO of ~1.2) but if I try to run:
>
> a p > t t~ [QCD]
>
> A warning pops up to say that DIS-type interactions are not supported at NLO. Now, I can easily disable this warning and it runs fine, and I again get a k-factor of ~1.2, but obviously I'd like to understand why this warning is there. The process I'm generating isn't really DIS at all, and the check seems to be simply looking to see if someone has turned on elastic photon from protons and also asked for NLO.
>
> Could you explain why this process isn't supported at NLO? When it seems to work fine and give sensible results?
>
> Cheers,
>
> Jay
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

James Howarth (jhowarth) said : #2

Hi Oliver,

Sorry, I should have said, I'm running v2.8.2. This is the most recent of the v2 tag right?

Cheers,

Jay

James Howarth (jhowarth) said : #3

Sorry, Olivier*

Hi,

You are right even with the latest version we still have that behaviour. I thought that we were allowing it but looks like not.

So the fact that it was not allowed was not introduced by designed but at the same time such feature was never validated. Couple of person told us they were going to use/validate such type of generation but looks like we never heard back from them (since otherwise we would have allowed this to run).

So i guess that bypassing such check is reasonable.

Cheers,

Olivier

PS: You can call me Oliver or Olivier both are fine.

> On 11 Jan 2021, at 15:50, James Howarth <email address hidden> wrote:
>
> Question #694892 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/694892
>
> James Howarth posted a new comment:
> Sorry, Olivier*
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

James Howarth (jhowarth) said : #5

Thanks Olivier!

Would it be possible to remove or change the warning for future versions? I want to patch this in an ATLAS release but it's of course neater if it's fixed in a future version.

Cheers,

Jay

Richard Ruiz (rruiz) said : #6

Hi Jay,

Just a minor addition. Modern PDF sets with the LUXqed or a LUX-like formalism, eg., NNPDF+LUXqed, MMHT+LUXqed, and CT14qed, contain both elastic and inelastic contributions to the photon PDF. So if you desire the double elastic, single elastic-single inelastic, and double inelastic regions of phase space at NLO in QCD, one option is to run the following:

generate a a > t t~ [QCD]
with
lpp = 1
lhaid = something with LUXqed

It is super important to activate QED evolution in the parton shower. For PY8 (see bottom of http://home.thep.lu.se/~torbjorn/pythia82html/SpacelikeShowers.html) , these switches are off by default. More specifically in file
<output dir>/MCatNLO/Scripts/MCatNLO_MadFKS_PYTHIA8.Script
you want to set the following to `on' :

TimeShower:QEDshowerByQ = off
TimeShower:QEDshowerByL = off
TimeShower:QEDshowerByGamma = off
SpaceShower:QEDshowerByQ = off
SpaceShower:QEDshowerByL = off

These will match/connect your initial-state photon to a quark line, which might have low |eta| (inelastic region) or arbitrarily large (elastic region).

Well the main reason why I do not fix this in an official version is that i have never seen a validation plot for such case.
I guess it should be fine but before having something available without hacking the code it would be nice to be sure that such implementation is working and that we are not missing a counter-term or something like that.
I know that the formula is correct (since this is the same as for LO so that part is validated to my point of view but the interaction with NLO is less clear. I will bring this point to our next MG5dev meeting (if I do not forget and ask the full team what they think)

Cheers,

Olivier

James Howarth (jhowarth) said : #8

Okay, thanks a lot Olivier! And thanks Richard for the helpful Pythia advice.

I'll mark the problem as solved.

Cheers,

Jay