IEEE_DIVIDE_BY_ZERO error for very-high-mass H+ signals

Asked by Shigeki Hirose on 2021-06-03

Dear MadGraph authors,

I am one of the ATLAS collaborators, and am working in the H+ -> tb analysis team with targeting very-high-mass region such as > 3 TeV.
When we were trying to test-produce signal MC samples with m(H+) = 3-5 TeV with MadGraph, we faced a zero-division error.

...
tau_min 3 1 : 0.46767E+04 0.46767E+04 0.46767E+04
tau_min 4 1 : 0.46767E+04 0.46767E+04 0.46767E+04
tau_min 5 1 : 0.46767E+04 -- 0.46767E+04
tau_min 6 1 : 0.46767E+04 -- 0.46767E+04
 ERROR #39 in genps_fks.f 21871522.890000008 21871522.889999997
 Error #13 in genps_fks.f
   21827584.000000000 22.089999999998291 -1
Note: The following floating-point exceptions are signalling: IEEE_INVALID_FLAG IEEE_DIVIDE_BY_ZERO
Thanks for using LHAPDF 6.2.3. Please make sure to cite the paper:
  Eur.Phys.J. C75 (2015) 3, 132 (http://arxiv.org/abs/1412.7420)
Time in seconds: 0

Please find the following log file for full details, which is from the setting with m(H+) = 4.5 TeV:

https://cernbox.cern.ch/index.php/s/WDzuHAH4nyPERzu

At first, we thought the 4.5 TeV mass point coincided with a singularity by chance, but the same error arose when the mass was set at 4.0 < m(H+) < 5.0 TeV (i.e. 4.0 and 5.0 TeV were fine, but any mass points greater than 4.0 TeV and less than 5.0 TeV seemed to cause a problem).

Do you have the cause of this error occurs, and if it is possible for to elude this?
We are a little afraid that our mass points are too heavy and problematic, like the ME calculation does not work properly.

Please let us know if you need more inputs from our side.

Many thanks for your kind help,
Shigeki

Question information

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

How did you implement such mass window?
For such type of implementation, you have not only to implement the cut but also to modify the phase-space integrator accordingly.

> We are a little afraid that our mass points are too heavy and problematic, like the ME calculation does not work properly.

I do not expect any issue in that direction.

Cheers,

Olivier

> On 3 Jun 2021, at 07:10, Shigeki Hirose <email address hidden> wrote:
>
> New question #697365 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/697365
>
> Dear MadGraph authors,
>
> I am one of the ATLAS collaborators, and am working in the H+ -> tb analysis team with targeting very-high-mass region such as > 3 TeV.
> When we were trying to test-produce signal MC samples with m(H+) = 3-5 TeV with MadGraph, we faced a zero-division error.
>
> ...
> tau_min 3 1 : 0.46767E+04 0.46767E+04 0.46767E+04
> tau_min 4 1 : 0.46767E+04 0.46767E+04 0.46767E+04
> tau_min 5 1 : 0.46767E+04 -- 0.46767E+04
> tau_min 6 1 : 0.46767E+04 -- 0.46767E+04
> ERROR #39 in genps_fks.f 21871522.890000008 21871522.889999997
> Error #13 in genps_fks.f
> 21827584.000000000 22.089999999998291 -1
> Note: The following floating-point exceptions are signalling: IEEE_INVALID_FLAG IEEE_DIVIDE_BY_ZERO
> Thanks for using LHAPDF 6.2.3. Please make sure to cite the paper:
> Eur.Phys.J. C75 (2015) 3, 132 (http://arxiv.org/abs/1412.7420)
> Time in seconds: 0
>
> Please find the following log file for full details, which is from the setting with m(H+) = 4.5 TeV:
>
> https://cernbox.cern.ch/index.php/s/WDzuHAH4nyPERzu
>
> At first, we thought the 4.5 TeV mass point coincided with a singularity by chance, but the same error arose when the mass was set at 4.0 < m(H+) < 5.0 TeV (i.e. 4.0 and 5.0 TeV were fine, but any mass points greater than 4.0 TeV and less than 5.0 TeV seemed to cause a problem).
>
> Do you have the cause of this error occurs, and if it is possible for to elude this?
> We are a little afraid that our mass points are too heavy and problematic, like the ME calculation does not work properly.
>
> Please let us know if you need more inputs from our side.
>
> Many thanks for your kind help,
> Shigeki
>
>
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
Shigeki Hirose (shhirose) said :
#2

Dear Olivier,

Thank you very much for your answer, and I aplogise for my slow response.
I was taking to our software experts to understand how MadGraph is implemented to our software.
It seems that MadGraph is implemented "as it is", meaning that there is no special treatment and therefore default way of phase-space integral is expected.
They seemed not sure how MadGraph behaves by default if we do not apply any specific treatment on this point.
Do you think we should accommodate the phase space integrator by ourselves? (i.e. the phase space integrator is not automatically modified by MadGraph, depending on the input parameters such as mass?)

Best regards,
Shigeki

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

Then I'm clearly missing some point on how you run the code.
Could you give me more information
(all command that you run / parameter that you modify/...)

Thanks,

Olivier

> On 17 Jun 2021, at 07:00, Shigeki Hirose <email address hidden> wrote:
>
> Question #697365 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/697365
>
> Status: Answered => Open
>
> Shigeki Hirose is still having a problem:
> Dear Olivier,
>
> Thank you very much for your answer, and I aplogise for my slow response.
> I was taking to our software experts to understand how MadGraph is implemented to our software.
> It seems that MadGraph is implemented "as it is", meaning that there is no special treatment and therefore default way of phase-space integral is expected.
> They seemed not sure how MadGraph behaves by default if we do not apply any specific treatment on this point.
> Do you think we should accommodate the phase space integrator by ourselves? (i.e. the phase space integrator is not automatically modified by MadGraph, depending on the input parameters such as mass?)
>
> Best regards,
> Shigeki
>
> --
> 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 Shigeki Hirose for more information if necessary.

To post a message you must log in.