correct flavour scheme for ttbar production

Hi Wojciech,
I think the 5 flavour scheme is fine, as you are anyway inclusive on the radiation generated by the real emission, and furthermore only a fraction of that will include b-quarks (in a way which is proportional to the b quark pdf, so quite small).

About the default choices, there is no particular reason for that. I agree that it is formally inconsistent to use a 5 flavour set when doing a computation with massive b's, but again, there is no particular reason to do that (also built-in PDFs are rather old, so we do not have 4 flavour built-in sets).

I hope this answer your questions, if not, don't hesitate to ask again.

All the best!

Marco

Hi Marco,

Thanks a lot for your answer and sorry for replying with such a delay. I wanted to ask about the PYTHIA8 setup in that case. Should one set b mass in shower card to 0 (i.e. set the corresponding flag to -1 and allow PYTHIA8 to take the mass used in generation of ME) and also use 5F PDF or should one use massive b's in PYTHIA?

Wojciech

Dear Wojciech,
the setup we suggest to use with b quarks is to leave PY8 to use the
default mass, i.e. 4.8 gev.
This works if the b-quarks in the LHE file are massless.
If you use massive b-quarks in the LHE file, pythia 8 keeps their mass,
regardless (even if i'm not 100% sure) of what you have written in the
shower_card. This is a peculiar behaviour of py8, not of all MCs.
If you want to change the b-quark mass, you should anyway never set it to
zero, but rather to a small value (0.3gev or so). Please note that we do
not guarantee for the results to be meaningful, and note also that the tune
of py8 may assume mb=4.8 gev, so by changing the b mass to an unphysical
value you can screw up amny things, in primis b-hadron realted observables.

Please let me know if this answer is satisfactory.
I have added Paolo Torrielli to the discussion, who is the MC expert (and
who suggested me the answer I wrote you)...
Cheers,

Marco

Marco Zaro

On Fri, Feb 28, 2014 at 7:31 AM, Wojciech Kotlarski <
<email address hidden>> wrote:

> Question #242904 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/242904
>
> Status: Answered => Open
>
> Wojciech Kotlarski is still having a problem:
> Hi Marco,
>
> Thanks a lot for your answer and sorry for replying with such a delay. I
> wanted to ask about the PYTHIA8 setup in that case. Should one set b
> mass in shower card to 0 (i.e. set the corresponding flag to -1 and
> allow PYTHIA8 to take the mass used in generation of ME) and also use 5F
> PDF or should one use massive b's in PYTHIA?
>
> Wojciech
>
> --
> You received this question notification because you are a direct
> subscriber of the question.
>

Dear Marco

Thanks a lot. I seem to undestand it now. The only problem is that the PYTHIA8 shower interface seems to break after a handfull of events. This is true for both 4F and 5F so if you'd like I can ask this question in a separate thread as this is not directly connected with this topic. PYTHIA8 stops after rougly 20 events. HERWIG6 interface works fine. For a given decayed sample, the number of events after which it fails depends on the seed in the shower card. And it seams to be a process dependent thing. It does fail for zw+,zz and tt~tt~ but it doesn't for ttbar.

P.s. You wrote, that "If you use massive b-quarks in the LHE file, pythia 8 keeps their mass". And if you have massless in real emission? Will it rescale its momenta?

cheers,
Wojciech

Thanks marco zaro, that solved my question.

Dear Wojcech,
You can get the pythia log file in MCatNLO/RUN_PYTHIA8_.../mcatnlo_run.log,
where you should have the error message.

If quarks are massless, momenta will be reshuffled internally in the
parton-shower in order to put them on the mass-shell.

Cheers,

Marco

Marco Zaro

On Wed, Mar 5, 2014 at 10:51 AM, Wojciech Kotlarski <
<email address hidden>> wrote:

> Question #242904 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/242904
>
> Status: Answered => Open
>
> Wojciech Kotlarski is still having a problem:
> Dear Marco
>
> Thanks a lot. I seem to undestand it now. The only problem is that the
> PYTHIA8 shower interface seems to break after a handfull of events. This
> is true for both 4F and 5F so if you'd like I can ask this question in a
> separate thread as this is not directly connected with this topic.
> PYTHIA8 stops after rougly 20 events. HERWIG6 interface works fine. For
> a given decayed sample, the number of events after which it fails
> depends on the seed in the shower card. And it seams to be a process
> dependent thing. It does fail for zw+,zz and tt~tt~ but it doesn't for
> ttbar.
>
> P.s. You wrote, that "If you use massive b-quarks in the LHE file,
> pythia 8 keeps their mass". And if you have massless in real emission?
> Will it rescale its momenta?
>
> cheers,
> Wojciech
>
> --
> You received this question notification because you are a direct
> subscriber of the question.
>

Dear Marco,

So I've checked that PYTHIA8 runs fail because of momentum non-conservation. By default, the Check:epTolErr = 0.001. I have an LHE file from MadGraph where you would need to set it to 0.5 to get 100% events from the sample. Otherwise PYTHIA throws abort an skips events returning less than 1% of events.

cheers,
Wojciech

Dear Wojciech,
thanks for your feedback.
Please, send us some failing events in order for us to understand whether
there is a problem on the MG5_aMC@NLO side (for example, to see up to
what accuracy momentum is conserved in the events).
Anyway, since things work fine with HW6, the source of the problem should
not be a true momentum non-conservation, but just the fact that probably
PY8 does not like certain masses you are using (for example, it does not like
massless leptons, so there is a new version of our interface that sets leptons
massive).
If the problem is due to leptons, please try setting them massive in the PY8
script. If this does not work (or if it is not the source of the problem) I think
it is best if you ask the authors of PY8 what is the reason for the failure.
Anyway, the setting of Check:epTolErr = 0.001 was explicitly recommended
by one of them, so I'm not sure if setting it to 0.5 is fine.
Best regards.
Paolo

Dear Pablo,

I also don't think that this is a "real issue". I looked at the LHE file and it seems fine as far as momentum conservationj is concerned. Also this Check:epTolErr shouldn't be larger than this 0.001 as this is "Maximum allowed summed deviation of E, p_x, p_y and p_z between the incoming beams and the final state, as a fraction of the initial energy, above which the event is counted as aborted." according to PY8 online manual. So 0.5 signals that there is a problem somewhere. I put event file containing 10k decayed events pp>zw+ under the link
https://dl.dropboxusercontent.com/u/30673620/events.lhe.gz

I use PY8 8183.

cheers,
Wojciech

Dear Paolo,

Any news on the problem?

cheers,
Wojciech

Dear Wojciech,
sorry for my late reply, somehow your previous email had not been notified
to my email account. I'll take a look at the event sample and get back to you
as soon as possible.
Sorry again,
cheers.
Paolo

Dear Wojciech,
the problem is due to the tau mass in the event file generated by MadSpin.
The momentum squared is 0 while the mass is not, hence Pythia's complaint.
This should be fixed in MadGraph5_aMC@NLO version 2.1.1, that will be released
soon (matter of days): I've tried with this version and the events featuring a tau
respect the on-shell condition, so it should be fine.
Cheers.
Paolo

Dear Paolo,

Thanks a lot.

cheers,
Wojciech