Problem with decay via dipole operator

[Brian Shuve]

I have a model with two Majorana fermions (call them chi2 and chi1) coupled to the photon and Z via a dimension-5 dipole operator. I have implemented this in a UFO model, andI am also applying a form factor to resolve the vertex at high momenta. Everything seems to function reasonably well in the model for any terms involving the production of chi2 and chi1 through Z/gamma.

If, however, I want to implement the decay chi2 > chi1 a, I run into issues. First of all, the width is correctly generated if I do something like

generate chi2 > chi1 a

If instead I look at a decay chain like

generate p p > chi1 chi2, chi2 > chi1 a,

then it returns zero events.

I get even funnier behaviour if I use MadSpin to implement the decay. If I turn off spin correlations via "set spinmode none", then the decay is implemented fine. If, however, I keep spin correlations, then I get the error

Exception : Partial width for chi2 > a chi1 negative: -0.00120140334966

This suggests that there is something about how the spins are being treated. I do not know if it is a problem with the UFO file or a genuine bug, but I thought I would ask to see if you had any thoughts for things I could try. I am attaching a tarball of the UFO file and can also send the log with traceback if it would be helpful.

Thanks!

[Brian Shuve (bshuve-d)]

Here is the debug report.

[Olivier Mattelaer (olivier-mattelaer)]

Hi,

 If instead I look at a decay chain like

 generate p p > chi1 chi2, chi2 > chi1 a,

 then it returns zero events.

This is related to a bug in the ALOHA package (arXiv:1108.2041<http://arXiv.org/abs/arXiv:1108.2041>).
In presence of Majorana, the use of form-factor is not supported (for conjugate helicity amplitude).
Since MG5aMC do not always use conjugate helicity amplitude (it does not need to for chi2 > chi1 a) this explains why the decay work but not production+decay.

 I get even funnier behaviour if I use MadSpin to implement the decay.
 If I turn off spin correlations via "set spinmode none", then the
 decay is implemented fine.

This one works because it never use the full matrix-element. It only generate events for chi2 > chi1 a.
It also works because that mode does not need to know in advance the partial width (it is actually computed on the flight when the event are generated by madevent)

If, however, I keep spin correlations, then
 I get the error

 Exception : Partial width for chi2 > a chi1 negative:
 -0.00120140334966

It seems to be related to a second problem within the file decay.py of your UFO model.
However, since you indicate that you are using a FormFactor, I’m surprised that you have not removed the decay.py file.
Independently of the fact that the result is correct or not, I guess that the problem is that you have numerical inaccuracy related to the analytical formula.
(Since the value computed by direct integration is of the order of 1e-13 this is clearly possible).

Now due to your breaking of the UFO convention for your model. A lot of functionality of MG5aMC will not work out of the box.
In particular the automatic computation of the width (arXiv:1402.1178<http://arXiv.org/abs/arXiv:1402.1178>).
Since MadSpin (arXiv:1212.3460<http://arXiv.org/abs/arXiv:1212.3460>) call the automatic computation of the width if the partial width are not present in your file, then you have such problem.
The only solution to use MadSpin is to have the partial width defined for all the particle which have a decay defined in the madspin_card.dat
(So remove the decay line for particle that you do not care in that file)

Cheers,

Olivier

On 23 Feb 2017, at 06:09, Brian Shuve <<email address hidden><mailto:<email address hidden>>> wrote:

Here is the debug report.

** Attachment added: "run_03_tag_1_debug.log"
  https://bugs.launchpad.net/mg5amcnlo/+bug/1667211/+attachment/4824977/+files/run_03_tag_1_debug.log

--
You received this bug notification because you are subscribed to
MadGraph5_aMC@NLO.
https://bugs.launchpad.net/bugs/1667211

Title:
 Problem with decay via dipole operator

Status in MadGraph5_aMC@NLO:
 New

Bug description:
 I have a model with two Majorana fermions (call them chi2 and chi1)
 coupled to the photon and Z via a dimension-5 dipole operator. I have
 implemented this in a UFO model, andI am also applying a form factor
 to resolve the vertex at high momenta. Everything seems to function
 reasonably well in the model for any terms involving the production of
 chi2 and chi1 through Z/gamma.

 If, however, I want to implement the dec...

[Olivier Mattelaer (olivier-mattelaer)]

Here is the ALOHA patch. Be carefull that I can not guarantee that such patch is going to work with your out-dated version of MG5aMC.

Cheers,

Olivier

[Olivier Mattelaer (olivier-mattelaer)]

And by the way thanks a lot for your very nice and very helpfull bug report.

Cheers,

Olivier

[Brian Shuve (bshuve-d)]

Hi Olivier,

Thank you very much for the very quick and detailed response. This is very helpful for using the model moving forward!

Cheers,

Brian