compute_widths vs. calculate_decay_widths

Dear Matthew,

In short, the command
“compute_widths” is an external routine which is able to compute the total width for any particle of the model.
“calculate_decay_widths” is a routine which runs the numerical code created by the “generate” command to compute
the partial width (total width if you include all the channel).
The first one is an automated version of the second which provides more control to the user.

More in details:
“compute_widths” is a recent function which was introduced in the following paper: arXiv:1402.1178
In short that command compute the width of any particle of the model and determine everything automatically
this means:
 - creates the Feynman Diagram that are needed for the computation (removing radiation diagram and double counting with lower body decay)
 - determines if it is relevant to include 3/4 body decay automatically
 - determines apriori the contribution of all the body decay to integrate numerically only the relevant one.
Since this command creates it’s own Feynman Diagram.
 - it needs to have access to the MG5_aMC main code
 - it can be run without having run any “generate” command. The following code is for example working:
    import model XXX
    compute_widths z w+ t
 - Even if you have run the “generate”, you can compute width for different particle.
 - Whatever your “generate” command is, the result of that function will be the same.
 - This is the command used when you put “Auto” in the param_card for a width. (so that command can also be use if your “generate” command is a 2>N process)

“calculate_decay_widths” is actually much older and does not require specific citation.
 - That command is only accessible in the shell interface created by ./bin/madevent (or via the command “launch -i”)
 - This command is actually very close of the command “generate_events" expect that the number of requested events is ignored
   and the only purpose of the command is to compute all the partial width with a 1% precision
 - Since you generate the Feynman Diagram which are considered for the computation, you have to handle the double counting by hand, …

One technical difference is that “compute_widths” can only be used with UFO model while, the “calculate_decay_widths” can be used with any model

> I have also noticed that they can give different results (I have only observed this for a 3-body decay so far).

If you do not have the same results for a partial width, this means that you didn’t agree with our code on the Feynman Diagram to include for the computation of the three-body decay.
I would advise that you compare the Feynman Diagram used in both case, and if you consider that we are wrong, then I can take a look.
Note that the inclusion/exclusion of 4 point interaction are always tricky and deeply depends if the UFO model. So if the model is not designed carefully this can leads to some trouble.

Cheers,

Olivier

On Nov 21, 2014, at 6:51 AM, Matthew Low <email address hidden> wrote:

> New question #258001 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/258001
>
> Hi MadGraph team,
>
> What is the difference between calculate_decay_widths and compute_widths. I have read the help entry for both and it is not obvious to me what the difference is. I notice that setting auto appears to use compute_widths. I have also noticed that they can give different results (I have only observed this for a 3-body decay so far). What are they actually doing and which one should be used?
>
> Thanks,
> - Matthew
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi Olivier,

Thanks for the detailed response, I have not accounted for double counting in using calculate_decay_widths so I will check that this makes the two computations agree. When calling compute_widths, are the processes/diagrams stored somewhere such that I can make direct comparisons?

In the particular model I am using I expected to see a 4-pt interaction in the diagrams for the 3-body decay, however, I only see diagrams will off-shell intermediate particles and 3-pt interactions. How does the inclusion of a 4-pt interaction depend on the UFO? Does it come down to defining the interaction order for the 4-pt vertex?

Thanks,
- Matthew

Hi Matthew,

> When calling compute_widths, are the
> processes/diagrams stored somewhere such that I can make direct
> comparisons?

The information is present but in a temporary file (i.e. removed when the computation is done).
One way to have access to hit is to press ctrl-z when you see the line:

Pass to numerical integration for computing the widths:
INFO: More info in temporary files:
    /Users/omatt/Documents/eclipse/NewLoopInduced_genximprove/tmp0uWKY0/temp_decay/index.html

Then open that file and you will have all the information.

> In the particular model I am using I expected to see a 4-pt interaction
> in the diagrams for the 3-body decay, however, I only see diagrams will
> off-shell intermediate particles and 3-pt interactions. How does the
> inclusion of a 4-pt interaction depend on the UFO? Does it come down to
> defining the interaction order for the 4-pt vertex?

Yes it does. The rules are not easy to explained, so I would simply refer to the paper (arXiv:1402.1178) for those details, rather than retry to explain those.

Cheers,

Olivier

On Nov 24, 2014, at 6:56 AM, Matthew Low <email address hidden> wrote:

> Question #258001 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/258001
>
> Status: Answered => Open
>
> Matthew Low is still having a problem:
> Hi Olivier,
>
> Thanks for the detailed response, I have not accounted for double
> counting in using calculate_decay_widths so I will check that this makes
> the two computations agree. When calling compute_widths, are the
> processes/diagrams stored somewhere such that I can make direct
> comparisons?
>
> In the particular model I am using I expected to see a 4-pt interaction
> in the diagrams for the 3-body decay, however, I only see diagrams will
> off-shell intermediate particles and 3-pt interactions. How does the
> inclusion of a 4-pt interaction depend on the UFO? Does it come down to
> defining the interaction order for the 4-pt vertex?
>
> Thanks,
> - Matthew
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.