MadGraph5_aMC@NLO

Color Octet Scalar in MG 3 vs 4

Asked by Matthew Low

Hi,

Some time ago I tried adding a color octet scalar (coupling to quarks and gluons) to the SM (via the usr v4 method) and then I tried to generate the process p p > octet > g g in MadGraph 3.24. The parton-level distribution of the dijet invariant mass was not as I expected. There was a peak at the octet mass however it was on top of a background which increased at lower dijet mass.

Now, I've completed the same process in MadGraph 4.8 and the dijet mass is sharply peaked at the octet mass without the falling background (how I would have thought it would be). I'm just wondering what was the change from 3->4 that caused this correction to the plot and since the gluon-gluon-octet operator is a dimension-five operator, what is the vertex that MadGraph is using?

Thanks for your help,
- Matthew

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
Last query:
Last reply:
Revision history for this message
Matthew Low (mattlow) said : 		#1

I had another question actually about the resulting LHE file of pp > octet > gg. I've noticed that if I try to use the LHE file as input to Pythia8 then most events fail with the error:

 PYTHIA Error in ProcessLevel::checkColours: unphysical colour flow
 PYTHIA Abort from Pythia::next: processLevel failed; giving up

But a few (~5/500) will be showered and hadronized. The ones that fail have the form:

<event>
 5 0 0.1064220E+03 0.2060213E+03 0.7818608E-02 0.1151168E+00
       21 -1 0 0 503 501 0.00000000000E+00 0.00000000000E+00 0.64898362944E+02 ...
       21 -1 0 0 501 502 0.00000000000E+00 0.00000000000E+00 -0.96122250908E+03 ...
  9100000 2 1 2 503 502 0.00000000000E+00 0.00000000000E+00 -0.89632414614E+03 ...
       21 1 3 3 503 504 0.36924051067E+01 -0.20598825584E+03 -0.15811446599E+03 ...
       21 1 3 3 504 502 -0.36924051067E+01 0.20598825584E+03 -0.73820968015E+03 ...
</event>

While those that are okay have the form:
<event>
 5 0 0.1064220E+03 0.2060213E+03 0.7818608E-02 0.1151168E+00
       21 -1 0 0 503 501 0.00000000000E+00 0.00000000000E+00 0.64898362944E+02 ...
       21 -1 0 0 501 502 0.00000000000E+00 0.00000000000E+00 -0.96122250908E+03 ...
       21 1 3 3 503 504 0.36924051067E+01 -0.20598825584E+03 -0.15811446599E+03 ...
       21 1 3 3 504 502 -0.36924051067E+01 0.20598825584E+03 -0.73820968015E+03 ...
</event>

I'm wondering if you have any recommendations on how to deal with this?

Thanks again for your assistance,
- Matthew

Revision history for this message
Best Olivier Mattelaer (olivier-mattelaer) said : 		#2

Hi Matthew,

> I'm just wondering what was the change from 3->4 that caused this correction to the plot and since the gluon-gluon-octet operator is a dimension-five operator, what is the vertex that MadGraph is using?

Like this, I don't see any changes in the updateNote files which should creates such differences. Are you sure that you were using the same syntax of generation (and having the same diagrams?). Were you using the same cuts?

Concerning the operator, this should be indicated in your model and in order to have a dimension-five operator probably creates your own HELAS routine (for v4 model). It's probably using in your case a \eta_{mu,nu}. I will need to have the model to be sure.

In this case, it's probably easier to implement a UFO model rather than a v4 model. In UFO, you can specify directly the FeynmanRules, and Madgraph
will call a program ALOHA in order to create the HELAS routine that MG needs in order to do the computation.
In order to create the UFO model, you can also think to use Feynrules, in this case, you need to write your Lagrangian and FR will create the UFO files for you.

Concerning the event:

> <event>
> 5 0 0.1064220E+03 0.2060213E+03 0.7818608E-02 0.1151168E+00
> 21 -1 0 0 503 501 0.00000000000E+00 0.00000000000E+00 0.64898362944E+02 ...
> 21 -1 0 0 501 502 0.00000000000E+00 0.00000000000E+00 -0.96122250908E+03 ...
> 9100000 2 1 2 503 502 0.00000000000E+00 0.00000000000E+00 -0.89632414614E+03 ...
> 21 1 3 3 503 504 0.36924051067E+01 -0.20598825584E+03 -0.15811446599E+03 ...
> 21 1 3 3 504 502 -0.36924051067E+01 0.20598825584E+03 -0.73820968015E+03 ...
> </event>

The color of those events are correct for an octet. Did you use the last version of Pythia8?
The difference between the two type of events is that in some case we write the scalar octet (as above) and
sometimes not (since the octet is too off-shell and therefore we can't neglect interference effect).

One solution, in order to have pythia pass would be to remove the line specifying the octet (line starting with 9100000).
But This will prevent any direct radiation from the octet.

Cheers,

Olivier

On Aug 13, 2012, at 3:25 PM, Matthew Low <email address hidden> wrote:

> Question #205733 on MadGraph5 changed:
> https://answers.launchpad.net/madgraph5/+question/205733
>
> Matthew Low gave more information on the question:
> I had another question actually about the resulting LHE file of pp >
> octet > gg. I've noticed that if I try to use the LHE file as input to
> Pythia8 then most events fail with the error:
>
> PYTHIA Error in ProcessLevel::checkColours: unphysical colour flow
> PYTHIA Abort from Pythia::next: processLevel failed; giving up
>
> But a few (~5/500) will be showered and hadronized. The ones that fail
> have the form:
>
> <event>
> 5 0 0.1064220E+03 0.2060213E+03 0.7818608E-02 0.1151168E+00
> 21 -1 0 0 503 501 0.00000000000E+00 0.00000000000E+00 0.64898362944E+02 ...
> 21 -1 0 0 501 502 0.00000000000E+00 0.00000000000E+00 -0.96122250908E+03 ...
> 9100000 2 1 2 503 502 0.00000000000E+00 0.00000000000E+00 -0.89632414614E+03 ...
> 21 1 3 3 503 504 0.36924051067E+01 -0.20598825584E+03 -0.15811446599E+03 ...
> 21 1 3 3 504 502 -0.36924051067E+01 0.20598825584E+03 -0.73820968015E+03 ...
> </event>
>
> While those that are okay have the form:
> <event>
> 5 0 0.1064220E+03 0.2060213E+03 0.7818608E-02 0.1151168E+00
> 21 -1 0 0 503 501 0.00000000000E+00 0.00000000000E+00 0.64898362944E+02 ...
> 21 -1 0 0 501 502 0.00000000000E+00 0.00000000000E+00 -0.96122250908E+03 ...
> 21 1 3 3 503 504 0.36924051067E+01 -0.20598825584E+03 -0.15811446599E+03 ...
> 21 1 3 3 504 502 -0.36924051067E+01 0.20598825584E+03 -0.73820968015E+03 ...
> </event>
>
> I'm wondering if you have any recommendations on how to deal with this?
>
> Thanks again for your assistance,
> - Matthew
>
> --
> You received this question notification because you are a member of
> MadTeam, which is an answer contact for MadGraph5.

Revision history for this message
Matthew Low (mattlow) said : 		#3

Hi Olivier,

Thanks for your reply. I reimplemented color octet to check and the distributions do match in between MG versions. I must have messed on something somewhere else in the code.

In regards to Pythia, I've been using Pythia8162, so not quite the latest (Pythia8165). Your suggestion of manually removing 9100000 line works fine and will be fine for my uses.

One last thing, though perhaps this is more of a Pythia question. If I wanted to leave 9100000 in the LHE file how would Pythia know how to properly compute radiation from it since Pythia doesn't know what the value of the 9100000 coupling to gluons is (or to quarks for that matter)?

Thanks again,
- Matthew

Revision history for this message
Matthew Low (mattlow) said : 		#4

Thanks Olivier Mattelaer, that solved my question.

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#5

Hi Matthew,

> One last thing, though perhaps this is more of a Pythia question. If I
> wanted to leave 9100000 in the LHE file how would Pythia know how to
> properly compute radiation from it since Pythia doesn't know what the
> value of the 9100000 coupling to gluons is (or to quarks for that
> matter)?

I don't know Pythia enough in order to give an accurate answer,
I can be wrong if I give an attempt of solutions, so I prefer not to.

Cheers,

Olivier

On Aug 14, 2012, at 7:20 PM, Matthew Low <email address hidden> wrote:

> Question #205733 on MadGraph5 changed:
> https://answers.launchpad.net/madgraph5/+question/205733
>
> Matthew Low posted a new comment:
> Hi Olivier,
>
> Thanks for your reply. I reimplemented color octet to check and the
> distributions do match in between MG versions. I must have messed on
> something somewhere else in the code.
>
> In regards to Pythia, I've been using Pythia8162, so not quite the
> latest (Pythia8165). Your suggestion of manually removing 9100000 line
> works fine and will be fine for my uses.
>
> One last thing, though perhaps this is more of a Pythia question. If I
> wanted to leave 9100000 in the LHE file how would Pythia know how to
> properly compute radiation from it since Pythia doesn't know what the
> value of the 9100000 coupling to gluons is (or to quarks for that
> matter)?
>
> Thanks again,
> - Matthew
>
> --
> You received this question notification because you are a member of
> MadTeam, which is an answer contact for MadGraph5.

Revision history for this message
Johan Alwall (johan-alwall) said : 		#6

Hello Matthew,

In regards to your second question, Pythia is reading the QNUMBERS block from the param_card, which defines the quantum numbers of the new particles. Perhaps the problem in your case is that this block is missing. It is automatically included in the param_card by the ConversionScript in the v4 usrmod, as well as from the UFO model if you use that to define your model.

All the best,
Johan