Critical warning in spinmode=none

Asked by Baptiste Ravina on 2020-11-06

I generated 100k events in MG v2.8.1 with "generate p p > t t~ z" and I would now like to decay them with and without spin correlations. The former runs fine, but in the latter case, using "set spinmode none" in the MadSpin card, I see

************************************************************
* *
* W E L C O M E to M A D S P I N *
* *
************************************************************
INFO: Extracting the banner ...
INFO: process: p p > t t~ z
INFO: options:
INFO: detected model: sm. Loading...
Set group_subprocesses to Auto
Note that you need to regenerate all processes
set ninja to /Users/bravina/Documents/LocalMadGraph/MG5_aMC_v2_8_1/HEPTools/lib
set f2py compiler to /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/bin/f2py
set lhapdf to /Users/bravina/Documents/LocalMadGraph/MG5_aMC_v2_8_1/HEPTools/lhapdf6/bin/lhapdf-config
set f2py compiler to /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/bin/f2py
set lhapdf to /Users/bravina/Documents/LocalMadGraph/MG5_aMC_v2_8_1/HEPTools/lhapdf6/bin/lhapdf-config
set fortran compiler to /usr/local/bin/gfortran
set seed 1
set Nevents_for_max_weight 75 # number of events for the estimate of the max. weight
set BW_cut 15 # cut on how far the particle can be off-shell
set spinmode none # Use one of the madspin special mode
set max_weight_ps_point 400 # number of PS to estimate the maximum for each event
define l+ = e+ mu+ ta+
define l+ = e+ mu+ ta+
Defined multiparticle l+ = e+ mu+ ta+
define l- = e- mu- ta-
define l- = e- mu- ta-
Defined multiparticle l- = e- mu- ta-
decay t > w+ b, w+ > all all
decay t~ > w- b~, w- > all all
decay w+ > l+ vl
decay w- > l- vl~
decay z > l+ l-
launch
INFO: generate 100000 decay event for particle t~
INFO: Will use seed 121581711
INFO: generate 100000 decay event for particle t
INFO: generate 100000 decay event for particle z
CRITICAL: The number of event generated is only 17957/120000. This typically indicates that you need specify cut on the decay process.
CRITICAL: We strongly suggest that you cancel/discard this run.
decaying event number 100 [0.446672916412 s]
decaying event number 200 [0.95703792572 s]
decaying event number 300 [1.51839399338 s]
decaying event number 400 [2.05968809128 s]
decaying event number 500 [2.58510303497 s]
decaying event number 600 [3.17977190018 s]
decaying event number 700 [3.69768500328 s]
decaying event number 800 [4.19961190224 s]
decaying event number 900 [4.67924499512 s]
decaying event number 1000 [5.19240808487 s]
decaying event number 2000 [9.9806330204 s]
decaying event number 3000 [14.8901329041 s]
decaying event number 4000 [19.5546720028 s]
decaying event number 5000 [24.5294570923 s]
decaying event number 6000 [29.2021939754 s]
decaying event number 7000 [34.2057130337 s]
decaying event number 8000 [38.9195849895 s]
decaying event number 9000 [44.06011796 s]
decaying event number 10000 [48.8661599159 s]
INFO: generate 87042 decay event for particle z
decaying event number 20000 [264.263995886 s]
decaying event number 30000 [319.74861598 s]
decaying event number 40000 [370.156426907 s]
decaying event number 50000 [419.519126892 s]
decaying event number 60000 [468.584677935 s]
decaying event number 70000 [519.457434893 s]
decaying event number 80000 [569.591913939 s]
decaying event number 90000 [619.03350091 s]

There does seem to be 100k decayed events in the output LHE file, but what does this CRITICAL warning mean? Can I trust the output?
Note: this does not happen if I "generate p p > t t~" and decay with "set spinmode none". Only in the ttZ version.

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Last query:
2020-11-08
Last reply:
2020-11-08

This question was reopened

Hi,

Looks like some issue occur when trying to generate the decay of the Z.
The phase-space integrator only generates 18k events instead of the 120k expected.

This explains why you have the lines:
> decaying event number 10000 [48.8661599159 s]
> INFO: generate 87042 decay event for particle z
> decaying event number 20000 [264.263995886 s]

Since the code fails short of the number of events it needs for the decay.
From the log, You can see that we did not had any issue to generate those.

So to be on the safe side, I would at least discard the first 18k events.
This being said it is unlikely that they are anything wrong to those 18k events (especially since 1>2 matrix element are constant). But at the same time this is very surprising (and clearly seed related) so I would encourage you to discard them.

Note that you do not only "loose" the spin-correlation but also the off-shell effects.

Cheers,

Olivier

> On 6 Nov 2020, at 12:31, Baptiste Ravina <email address hidden> wrote:
>
> New question #693857 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/693857
>
> I generated 100k events in MG v2.8.1 with "generate p p > t t~ z" and I would now like to decay them with and without spin correlations. The former runs fine, but in the latter case, using "set spinmode none" in the MadSpin card, I see
>
> ************************************************************
> * *
> * W E L C O M E to M A D S P I N *
> * *
> ************************************************************
> INFO: Extracting the banner ...
> INFO: process: p p > t t~ z
> INFO: options:
> INFO: detected model: sm. Loading...
> Set group_subprocesses to Auto
> Note that you need to regenerate all processes
> set ninja to /Users/bravina/Documents/LocalMadGraph/MG5_aMC_v2_8_1/HEPTools/lib
> set f2py compiler to /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/bin/f2py
> set lhapdf to /Users/bravina/Documents/LocalMadGraph/MG5_aMC_v2_8_1/HEPTools/lhapdf6/bin/lhapdf-config
> set f2py compiler to /System/Library/Frameworks/Python.framework/Versions/2.7/Extras/bin/f2py
> set lhapdf to /Users/bravina/Documents/LocalMadGraph/MG5_aMC_v2_8_1/HEPTools/lhapdf6/bin/lhapdf-config
> set fortran compiler to /usr/local/bin/gfortran
> set seed 1
> set Nevents_for_max_weight 75 # number of events for the estimate of the max. weight
> set BW_cut 15 # cut on how far the particle can be off-shell
> set spinmode none # Use one of the madspin special mode
> set max_weight_ps_point 400 # number of PS to estimate the maximum for each event
> define l+ = e+ mu+ ta+
> define l+ = e+ mu+ ta+
> Defined multiparticle l+ = e+ mu+ ta+
> define l- = e- mu- ta-
> define l- = e- mu- ta-
> Defined multiparticle l- = e- mu- ta-
> decay t > w+ b, w+ > all all
> decay t~ > w- b~, w- > all all
> decay w+ > l+ vl
> decay w- > l- vl~
> decay z > l+ l-
> launch
> INFO: generate 100000 decay event for particle t~
> INFO: Will use seed 121581711
> INFO: generate 100000 decay event for particle t
> INFO: generate 100000 decay event for particle z
> CRITICAL: The number of event generated is only 17957/120000. This typically indicates that you need specify cut on the decay process.
> CRITICAL: We strongly suggest that you cancel/discard this run.
> decaying event number 100 [0.446672916412 s]
> decaying event number 200 [0.95703792572 s]
> decaying event number 300 [1.51839399338 s]
> decaying event number 400 [2.05968809128 s]
> decaying event number 500 [2.58510303497 s]
> decaying event number 600 [3.17977190018 s]
> decaying event number 700 [3.69768500328 s]
> decaying event number 800 [4.19961190224 s]
> decaying event number 900 [4.67924499512 s]
> decaying event number 1000 [5.19240808487 s]
> decaying event number 2000 [9.9806330204 s]
> decaying event number 3000 [14.8901329041 s]
> decaying event number 4000 [19.5546720028 s]
> decaying event number 5000 [24.5294570923 s]
> decaying event number 6000 [29.2021939754 s]
> decaying event number 7000 [34.2057130337 s]
> decaying event number 8000 [38.9195849895 s]
> decaying event number 9000 [44.06011796 s]
> decaying event number 10000 [48.8661599159 s]
> INFO: generate 87042 decay event for particle z
> decaying event number 20000 [264.263995886 s]
> decaying event number 30000 [319.74861598 s]
> decaying event number 40000 [370.156426907 s]
> decaying event number 50000 [419.519126892 s]
> decaying event number 60000 [468.584677935 s]
> decaying event number 70000 [519.457434893 s]
> decaying event number 80000 [569.591913939 s]
> decaying event number 90000 [619.03350091 s]
>
>
> There does seem to be 100k decayed events in the output LHE file, but what does this CRITICAL warning mean? Can I trust the output?
> Note: this does not happen if I "generate p p > t t~" and decay with "set spinmode none". Only in the ttZ version.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Baptiste Ravina (bravina) said : #2

Hi Olivier,

Thanks for your reply. Where does the number 120000 come from, in "CRITICAL: The number of event generated is only 17957/120000" ?
Would changing the seed or reducing the number of events per job to maybe 10k help? I'm looking to generate maybe up to 1M events, so what you're saying is I need to collect and inspect the logs of each run, and discard N events in each? that's not practical, but ok...

Cheers,
Baptiste

Hi,

> Where does the number 120000 come from, in "CRITICAL: The number of event generated is only 17957/120000" ?

This is 100k (the number of events of the sample) * 1.2 (for security reason).

> Would changing the seed or reducing the number of events per job to maybe 10k help?

This is clearly a seed specific issue so so changing the combination seed/number of events will likely fix the issue. (also increasing to 250k events can be a good strategy).
This is the first time ever that I heard that we fail to reach a given number of events for Z-decay. So you should have been very unlucky for your seed (or face an hardware issue like a full /tmp).

> I'm looking to generate maybe up to 1M events, so what you're saying is I need to collect and inspect the logs of each run, and discard N events in each? that's not practical, but ok...

In theory, you should always inspect log. However I do not expect this issue to occur very often.

Cheers,

Olivier

> On 8 Nov 2020, at 13:50, Baptiste Ravina <email address hidden> wrote:
>
> Question #693857 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/693857
>
> Status: Answered => Open
>
> Baptiste Ravina is still having a problem:
> Hi Olivier,
>
> Thanks for your reply. Where does the number 120000 come from, in "CRITICAL: The number of event generated is only 17957/120000" ?
> Would changing the seed or reducing the number of events per job to maybe 10k help? I'm looking to generate maybe up to 1M events, so what you're saying is I need to collect and inspect the logs of each run, and discard N events in each? that's not practical, but ok...
>
> Cheers,
> Baptiste
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Can you help with this problem?

Provide an answer of your own, or ask Baptiste Ravina for more information if necessary.

To post a message you must log in.