Very different cross sections between gridpack and event generations

Asked by Sihyun Jeon on 2020-02-03

Hi Olivier,

I've been working on LRSM model [1] kfactors and saw something that wasn't expected.
The process I made was "p p > n1 e+" and "p p > n1 e+ [QCD]" for LO and NLO respectively.
And the tested mass point was "m(WR) = 5000 GeV & m(N) = 2500 GeV".

gridpack NLO LO [pb]
0.0002704 0.000189 -> kfactor ~ 1.5
event generation NLO LO [pb] (100K events)
0.0002694 0.000132 -> kfactor ~ 2.0

If you look at the results, it seems that NLO cross sections for event generation and gridpack are similar.
However there is somewhat difference between LO event generation and gridpack which of course leads to difference in kfactors.

I also checked m(N)=100 and 4900 GeVs and they all show similar behavior (similar NLO but event generation LO smaller than gridpack results).

I wonder if there's any deep reason for this.

Regards,
Sihyun.

[1] http://feynrules.irmp.ucl.ac.be/wiki/EffLRSM

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2020-02-14
Last query:
2020-02-14
Last reply:
2020-02-13

Can you give more details?
In particular we do not have any official gridpack for NLO. So please specify what you call gridpack. In top of that could you quote statistical and systematical uncertainties?

Thanks,

Olivier

> On 3 Feb 2020, at 05:43, Sihyun Jeon <email address hidden> wrote:
>
> New question #688459 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/688459
>
> Hi Olivier,
>
> I've been working on LRSM model [1] kfactors and saw something that wasn't expected.
> The process I made was "p p > n1 e+" and "p p > n1 e+ [QCD]" for LO and NLO respectively.
> And the tested mass point was "m(WR) = 5000 GeV & m(N) = 2500 GeV".
>
> gridpack NLO LO [pb]
> 0.0002704 0.000189 -> kfactor ~ 1.5
> event generation NLO LO [pb] (100K events)
> 0.0002694 0.000132 -> kfactor ~ 2.0
>
> If you look at the results, it seems that NLO cross sections for event generation and gridpack are similar.
> However there is somewhat difference between LO event generation and gridpack which of course leads to difference in kfactors.
>
> I also checked m(N)=100 and 4900 GeVs and they all show similar behavior (similar NLO but event generation LO smaller than gridpack results).
>
> I wonder if there's any deep reason for this.
>
> Regards,
> Sihyun.
>
> [1] http://feynrules.irmp.ucl.ac.be/wiki/EffLRSM
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Sihyun Jeon (shjeon) said : #2

Sorry Olivier,

never mind the NLO gridpacks since it's internal CMS stuff.

But regarding the LO gridpacks (which is actually also a CMS internal),
I would at least want to check and compare the following :

1. Cross section from actual event generating with MG gridpacks
2. Cross section from initial MadGraph pilot run while making gridpacks
3. Cross section from CMSSW gridpacks

For 2., when I get the run_01_gridpack.tar.gz and do ./run.sh, the output doesn't seem to give cross sections from the event generation.
Is it only possible by reading the lhe file and adding up the weights?

Thanks.

HI,

For 2., when I get the run_01_gridpack.tar.gz and do ./run.sh, the output doesn't seem to give cross sections from the event generation.
Is it only possible by reading the lhe file and adding up the weights?

Cross-section is not computed when you do ./run.sh.
So the one reported is (and should be) the one obtained when creating the pilot run.
The reason is that if you do
./run.sh 1 1
if we do not compute the cross-section, we can actually return the single events requested very fast (while re-computing the cross-section would need much more time)

Cheers,

Olivier

> On 13 Feb 2020, at 08:34, Sihyun Jeon <email address hidden> wrote:
>
> Question #688459 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/688459
>
> Status: Answered => Open
>
> Sihyun Jeon is still having a problem:
> Sorry Olivier,
>
> never mind the NLO gridpacks since it's internal CMS stuff.
>
> But regarding the LO gridpacks (which is actually also a CMS internal),
> I would at least want to check and compare the following :
>
> 1. Cross section from actual event generating with MG gridpacks
> 2. Cross section from initial MadGraph pilot run while making gridpacks
> 3. Cross section from CMSSW gridpacks
>
> For 2., when I get the run_01_gridpack.tar.gz and do ./run.sh, the output doesn't seem to give cross sections from the event generation.
> Is it only possible by reading the lhe file and adding up the weights?
>
>
> Thanks.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Sihyun Jeon (shjeon) said : #4

Ah I see. I was wondering why the event generation was so fast.
Hmm okay. Thanks anyways.

Regards,
Sihyun.

Sihyun Jeon (shjeon) said : #5

Thanks Olivier Mattelaer, that solved my question.