MadGraph5_aMC@NLO

Cross section and luminosity includes missing events?

Asked by Gabriel Magill

Hi,

I'm generating a neutrino event:
generate p vm > p vm mu+ mu-
and removing most of the cuts from the run_card as a check I'm doing for something else. I set
10000 = nevents
1 = lpp1 ! beam 1 type
0 = lpp2 ! beam 2 type
0.940 = ebeam1 ! beam 1 total energy in GeV
20.0 = ebeam2 ! beam 2 total energy in GeV
and put ptj = 0, etamin=0, etc.

When I generate events, it quotes a cross section and says that the number of events generated was ~100 out of 10000 I specified. Reading similar questions, I understand that it skips over events in which the divergences are too problematic because of my lack of cuts. However, when I multiply the cross section by the effective luminosity, I get something close to the initial number of events I specified. So my question is, does the cross section it quotes assumes the initial number of events specified, or has it already been multiplied by 1% to account for those that have been lost in the generation? I'm using Madgraph 2.4.0.

Best,
Gabriel Magill

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Gabriel Magill
Solved:
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#1

Dear Gabriel,

If you do not put cut, the cross-section is infinite.
We do not apply any cut in order to make the events, all your 100 events will be very close of the divergence.

The problem of the generation of events is that the function is not bounded.
Our algorithm tries the find the maximum bound by looking at the phase-space generated and then generate events based on this maximum. If it does not have enough events, it generates more phase-space.

Since your function is not bounded, what happens is that at the second step the maximum increase (typically by order of magnitude) and consequently only a couple of events are generated at that step. We then increase the number of PS -> the maximum increase by order of magnitude -> only few event generated. And it continues like that for a while.
So at some (long running time) we decide to give up and generate only a few events.

>So my question is, does the cross section it quotes assumes the initial number of events specified, or has it already been multiplied by 1% to account for those that have been lost in the generation?

The number quote is not related at all with the efficiency of generation. But I would certainly not trust this cross-section, since it is unlikely to be stable (and likely to have a very large statistical error)

Cheers,

Olivier

We do not skip any events

Revision history for this message
Gabriel Magill (magillgw) said : 		#2

Thanks Olivier,

I wasn't sure what algorithm it was doing, and I was skeptical of the effects of removing the cuts, so this clarifies a lot what is happening and why it was taking long.

Cheers,
Gabriel