MadGraph5_aMC@NLO

MadGraph gets stuck generating p p > b b~ b b~ [QCD]

Asked by Nihad Hidić

Hello,

I installed MadGraph a few days ago and I am trying to generate 4b events at NLO.
These are the commands that I used for the generation

generate p p > b b~ b b~ [QCD]
output fourB
launch fourB
shower off
set nevents = 1000

Currently I don't need showering.
I have also tried to set nevents lower 10 and 100, I got the same problem.

The problem is that it gets stuck at the last job
INFO: Idle: 0, Running: 1, Completed: 97 [ 10h 58m ]

I left it to run for 3 days and it was stuck at this.

The only warnings in the output in the terminal are these two

Gtk-Message: 20:08:39.218: Not loading module "atk-bridge": The functionality is provided by GTK natively. Please try to not load it.

WARNING: Could not compile StdHEP because its source directory could not be found in the SOURCE folder.
 Check the MG5_aMC option 'output_dependencies'.
 This will prevent the use of HERWIG6/Pythia6 shower.

From the text it doesn't seem like they should be the problem.

MadGraph seems to work fine and I also tried to generate the same process at LO and it worked and I also tried a NLO process to check if that was the problem (generate p p > e+ ve [QCD]), it worked fine.

Except for the changes in the run card stated above I didn't make any other changes.

These are the programs that were run
/=========================================================\
| 1. Type of perturbative computation order = NLO |
| 2. No MC@[N]LO matching / event generation fixed_order = OFF |
| 3. Shower the generated events shower = OFF |
| 4. Decay onshell particles madspin = OFF |
| 5. Add weights to events for new hypp. reweight = OFF |
| 6. Run MadAnalysis5 on the events generated madanalysis = Not Avai. |
\=========================================================/

I also have another question for the LO process p p > b b~ b b~.
I changed the ht4min in the run card several times (values 0, 10, 20, 40, 20000) but I got the same cross section.
I used "set ht4min = value that I want" to change the run card. From my understanding this should put a minimum for the Ht of the four leading jets. Shouldn't it affect the cross section?

I am still a beginner with MadGraph I don't know what the possible issues could be or if I didn't install something that is necessary for the process that I want to generate.

Best,
Nihad

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
Rikkert Frederix Edit question
Solved by:
Rikkert Frederix
Solved:
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#1

Hi,

For the LO computation,
I guess that you are doing a four flavor computation, in that case
maxjetflavor is set to 4 and b-quark are not consider as part of the jet for any of the cut (but they have their own separate cut.
In that case, it does make sense that it does not impact the cross-section at all.

The following line in the run_card is more appropriate for your case:
  0.0 = ihtmin !inclusive Ht for all partons (including b)
  -1.0 = ihtmax !inclusive Ht for all partons (including b)

For NLO, I'm a bit confused about your question. AT NLO, they are two type of computations:
- fixed order NLO, where running parton-shower is NOT possible (so no event generation).
- MC@NLO where running parton-shower is mandatory for getting physical result. In that mode you need to specify which parton-shower will be used later on since the result is specific for a given parton-shower.
Note that by construction both mode returns the same cross-section (up to statistical error).
So if you do not care about parton-shower, I would advice to run the fixed_order mode on "ON" and not on "OFF".

For the rest of the NLO question, I will assign this to Rik.

Revision history for this message
Rikkert Frederix (frederix) said : 		#2

Hello Nihad,

Which version of MadGraph5_aMC@NLO are you using?

Best,
Rikkert

Revision history for this message
Nihad Hidić (fenrir98) said : 		#3

Hello Rikket,

I am using MG5_aMC_v2.9.19

Best,
Nihad

Revision history for this message
Rikkert Frederix (frederix) said : 		#4

Dear Nihad,

Can you try with reducing the value of the OSThres parameter from the default 1d-13 to 1d-8 in the <YourProc>/Cards/MadLoopParams.dat? I.e., set

#OSThres
1.0d-8

(Note that you have to remove the exclamation mark before the value for it to take effect). Does this help?

Best,
Rikkert

Revision history for this message
Nihad Hidić (fenrir98) said : 		#5

Hello Rikket,

I am sorry for the late reply.
I tried running it with your suggestion. For the last 30 hours it is stuck at refining results step 2.
INFO: Refining results, step 2
INFO: Idle: 9, Running: 16, Completed: 0 [ current time: 03h03 ]

Best,
Nihad

Revision history for this message
Rikkert Frederix (frederix) said : 		#6

Hello Nihad,

This is a complicated process to compute at NLO. What is the accuracy you've requested for the run?

best,
Rikkert

Revision history for this message
Nihad Hidić (fenrir98) said : 		#7

Hello Rikket,

I haven't changed the accuracy. I began using MadGraph recently and I didn't make any changes to the run card except for the ones I stated at the beginning and the ones suggested here (i.e. I only made these changes shower off, fixed order on, set nevents = 5, OSThres 1.0d-8).

Best,
Nihad

Revision history for this message
Best Rikkert Frederix (frederix) said : 		#8

Hello Nihad,

I think this is just the way it is. This is a complicated process, with small masses that must regulate singularities. Hence, a rather difficult integral to do, which indeed might take ~weeks (if not more) to complete.

Best,
Rikkert

Revision history for this message
Nihad Hidić (fenrir98) said : 		#9

Thanks Rikkert Frederix, that solved my question.