Event numbers of screen output and run_01_tag_1_banner.txt are different

Asked by Gang Li

Dear MG5 team,

I want generate p p > w+ w- h at the LHC in the llljj final state in the SM at the parton-level. Here is how I generate events:

import model sm
generate p p > w+ w- h, w+ > j j, w- > l- vl~, h > l+ vl l- vl~ / z
add process p p > w+ w- h, w+ > l+ vl, w- > j j, h > l+ vl l- vl~ / z
add process p p > w+ w- h, w+ > l+ vl, w- > l- vl~, h > j j l- vl~
add process p p > w+ w- h, w+ > l+ vl, w- > l- vl~, h > l+ vl j j

When I set nevents 100 or 1000, event numbers of screen output and run_01_tag_1_banner.txt are the same. But if I set nevents 10000, then event numbers of screen output and run_01_tag_1_banner.txt are different. Events numbers of screen output is about 1.5K.

Did I miss something? Thanks a lot in advance.

Best,
Gang Li

Question information

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

Hi,

The run_XXX is your input (i.e what you request). In some case, what is asked by the user is too complicated to reach the number of events. In that case, indeed the two number would be different.

Cheers,

Olivier

Revision history for this message
Gang Li (ligangpku) said :
#2

Hi Olivier,

Thanks a lot for your information.

However, when I analyze the lhe file with ExRootAnalysis (here I am doing parton-level analysis). Using Root command treeReader->GetEntries() I obtain that the unweighted number of events if 1566, which is output to screen.

So it seems that there are actually 1566 events generated, not 10000 that I required.

Do you have any hint? Thanks.

Best,
Gang Li

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#3

Hi,

Yes that's correct if the two number differ, the actual number of event generated is the one written in the html page (or inside the lhe file).

The why is difficult to know like that, you can look at the html diagnostic page (by cliking on the cross-section) there you will see
all the channel of integration and the associated luminosity, then we have found the channel with the smallest luminosity, you can take a look at the associate log.
You will typically see that the code has trouble to converge correctly, this can be due to the presence of some cuts deforming the typical method of integrations.

Cheers,

Olivier

Note that for decay-chain, you have various options that you can try and that can have more efficient results

Revision history for this message
Gang Li (ligangpku) said :
#4

Hi Olivier,

That solved my problem.

Thanks a lot.

Best,
Gang Li