warning: small width

Asked by Sarah

I got this warning( Particle 36 with small width detected)
I saw the answer https://answers.launchpad.net/mg5amcnlo/+faq/3053, but I don't understand it.

do I have to write this line 1e-12 = small_width_treatment in run card after the running of the process is done?

Question information

English Edit question
MadGraph5_aMC@NLO Edit question
No assignee Edit question
Solved by:
Olivier Mattelaer
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :

Depend on how you want to handle that width.
In general, not doing anything is the best thing to do.



Revision history for this message
Sarah (sarahwarad) said :

Thank you for your answer

Is it something related to narrow width approximation?

I am trying to reproduce data of a published paper and I got slightly different results.
Maybe the story be behind this

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

Yes and No,

If the width is too small, then the numerical evaluation of the matrix-element and the associated phase-space integration can be numerically unstable.
This mode triggers special mode of computation to avoid such numerical issue.
To avoid such issue we indeed assume narrow-width-approximation (which should be a very good approximation when this mode triggers) to keep the cross-section correct, however offshell effect will not be handle in an exact fashion (which given the small width should be reasonable once again). Indeed, the full matrix-element will use a fake width (artificially large to avoid numerical issue) and this will impact offshell effect.

the parameter 'small_width_treatment' can change the threshold where this mode activates and how much artificially large the width is going to be in this mode.



Revision history for this message
Sarah (sarahwarad) said :

Thank you Olivier for the clarification.
I appreciate you

Revision history for this message
Sarah (sarahwarad) said :

Thanks Olivier Mattelaer, that solved my question.