Dynamical scale settings in the same-sign WW process

Thanks a lot for this.

I will have to take a look in details in this scale issue (likely that -1 takes ridiculous scale choice here).

Cheers,

Olivier

On 19 Sep 2019, at 23:13, Stefanie <<email address hidden><mailto:<email address hidden>>> wrote:

Question #684056 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/684056

Description changed to:
Dear experts,

I'm using MG version 5.2.6.5 for GP generation and 5.2.6.2 for event
generation. I'm running Madgraph with the following simplified process
for the same-sign WW fully leptonic final state:

generate u u > e+ ve mu+ vm d d QCD=0 @1

Though I'm generating the SM, I'm using an EFT model, however the total
cross-section are identical to when I run the default sm model.

I was testing the different dynamical scale settings -1 through 4. I
compare the total cross-section after integration in this spreadsheet:

https://docs.google.com/spreadsheets/d/1nM19Mv9DZn7Q8gNV2AkyKv2OxYRrFmpz2kVtSN_8PF0/edit?usp=sharing

There are differences, these are in the range of the expected scale
uncertainty for this process. However, if I look into differential
distributions, I find this in the invariant diboson mass at parton
level, mWW:

https://cernbox.cern.ch/index.php/s/8gwOonQuPQbYcuN

(Please ignore the label with the F parameter in this plot. It is all
generated as described above, SM only)

Now what makes us wonder is the very different shape of the default
scale in the high mWW tail and we are wondering if the clustering
algorithm is somehow behaving unexpected in this regime? All other
distributions I look at seem to be fine (apart from the slight offset
already seen at cross section level).

Unfortunately, since we found this in the course of the validation of
our EFT model (and as a consequence of this launchpad:
https://answers.launchpad.net/mg5amcnlo/+question/683400) we are not
able to invest much more time into this (e.g. doing many more tests). It
would be great if someone from the experts could have a look what is
going on with the scales here. If you need more information, I'm happy
to post logfiles, cards, settings.

Thanks,
Stefanie

--
You received this question notification because you are an answer
contact for MadGraph5_aMC@NLO.

Hi Olivier,

Thanks a lot for having a look!

Here maybe some more information for you to pin down the problem. We have tested this for some other VBS final states.

For lvlljj (WZ fully leptonic), the mWZ looks like:

https://cernbox.cern.ch/index.php/s/d7XhnnUbzE5jMuB

For semi-leptonic VBS some plots of the mVV are here:

https://cernbox.cern.ch/index.php/s/7QUu4u6KyumfwBx

For this we have tested different final states: same-sign WW, opposite sign WW, and ZZ.

In these final states, we do not observe such a feature in the mVV invariant mass distribution.

Cheers,
Stefanie

PS: I can remember from very early times that we have seen problem with the simulation of same-sign WW showers as well, where the two same charge leptons and a dipole radiation pattern led to incorrect results. I know that here we are at ME level and probably it's not comparable at all. I just wanted to give you one more hint where we have already seen problems in the same-sign WW final state only (while other final states were totally fine).