LO event generation with NLO command with specific light parton

Asked by Sung Hak Lim on 2017-12-13

Dear MG5 team,

I'm trying to understand some initial state dependent feature in my NLO analysis MG5 analysis.

I know this statement is only valid in LO but I checked my distribution scales constant from NLO to LO, so I'm trying to generate LO event with specific initial state.

But I want to generate LO event with seamless parameterization from NLO, i.e. using same run_card and keeping same run output structure.

It could be crazy command but I tried

 generate g g > t t~ [QCD]

Of course, this command is invalid as NLO, but I just wanted to know that using this process to generate LO events after launch by specifying


is okay or not. It seems MG5 gives some output to me at least.

Best wishes,
Sung Hak

Question information

English Edit question
MadGraph5_aMC@NLO Edit question
marco zaro Edit question
Solved by:
marco zaro
Last query:
Last reply:

This question was reopened

marco zaro (marco-zaro) said : #1

Dear Sung,
yes, if you do only LO runs, the results will be correct. Note that you can generate just the LO matrix elements, and still be able to use the same setup as for your NLO runs, with

generate g g > t t~ [LOonly=QCD]

Finally, note that in the html page after the run you will get the cross section per partonic channel, so you can also use that to get the gg-initiated cross section if you define a process with the multiparticle labels.

Let me know if you need more
Best wishes,


Sung Hak Lim (sunghak-lim) said : #2

Dear Marco,

Thanks for quick answers and suggestions!

LOonly=QCD option seems suitable for my analysis. I'll also check html page after this run then also.

Have a nice day!

Best wishes,
Sung Hak

Sung Hak Lim (sunghak-lim) said : #3

Thanks marco zaro, that solved my question.

Sung Hak Lim (sunghak-lim) said : #4

Dear Marco,

I just noticed that the command got crashes when I add

   set low_mem_multicore_nlo_generation true

at the beginning of MG5. May I know what's going on slightly?

I just put this here for those who might see this question again.

Best wishes,
Sung Hak

Best marco zaro (marco-zaro) said : #5

Dear Sung Hak ,
the low_mem_multicore_generation mode is a way to generate very complex processes (e.g. with many partonic subprocesses) in a more cpu and memory efficient way. In particular the process generation is spread across all cores and intermediate results are written to disk (as .pkl files) instead of being kept in the RAM. After output, what you get is identical to the 'standard' generation mode.
Now, concerning your problem, I got an error in version 2.6.0, but the error is not there any longer in 2.6.2, so it should have been fixed.
Anyway, thanks for reporting this problem.


Sung Hak Lim (sunghak-lim) said : #6

Thanks marco zaro, that solved my question.

Sung Hak Lim (sunghak-lim) said : #7

Dear Marco,

I see. Thanks for quick answsers!

Best wishes,
Sung Hak