Higgs decay branchings

Hi,

I can only comment for the standard model since I do not know your model.

Already for the standard model, the computation of the Higgs width does not work.
When running you should see, the following warning
> Please note that the automatic computation of the width is
> only valid in narrow-width approximation and at tree-level.

In the case of the Higgs, you have non negligible loop-induced decay which are not included. and therefore in other mode will be over-estimated (and the total width under-estimated).

One work-around is to use the heft model, which contract the top loop to a tree-level vertex (in the infinite top mass limit)
Which will then include the Higgs decaying to gluon and to photon (which is very small)

Here is the result for heft model:

# PDG Width
DECAY 25 5.621961e-03
# BR NDA ID1 ID2 ...
   7.657115e-01 2 -5 5 # 0.0043048001902515
   4.602487e-02 2 -15 15 # 0.00025875002417007
   3.470497e-02 2 21 21 # 0.00019510998784617
   2.305423e-02 3 -2 1 24 # 0.00012960998194503
   2.305423e-02 3 -4 3 24 # 0.00012960998194503
   2.283278e-02 3 -24 -1 2 # 0.00012836499868158
   2.283278e-02 3 -24 -3 4 # 0.00012836499868158
   7.623497e-03 3 -12 11 24 # 4.2859002817617e-05
   7.623497e-03 3 -14 13 24 # 4.2859002817617e-05
   7.598950e-03 3 -24 -11 12 # 4.272100054095e-05
   7.598950e-03 3 -24 -13 14 # 4.272100054095e-05
   7.549324e-03 3 -16 15 24 # 4.2442005104364e-05
   7.536872e-03 3 -24 -15 16 # 4.2372000445992e-05
   2.270560e-03 3 -1 1 23 # 1.276499976816e-05
   2.270560e-03 3 -3 3 23 # 1.276499976816e-05
   1.821429e-03 3 -5 5 23 # 1.0240002802269e-05
   1.793502e-03 3 -2 2 23 # 1.0082998297422e-05
   1.793502e-03 3 -4 4 23 # 1.0082998297422e-05
   1.726124e-03 2 22 22 # 9.704201809164e-06
   1.020700e-03 3 -12 12 23 # 5.7383355927e-06
   1.020700e-03 3 -14 14 23 # 5.7383355927e-06
   1.020700e-03 3 -16 16 23 # 5.7383355927e-06
   5.083903e-04 3 -11 11 23 # 2.8581504393783e-06
   5.083903e-04 3 -13 13 23 # 2.8581504393783e-06
   4.989540e-04 3 -15 15 23 # 2.805099928794e-06

Those number can be compare to Table 7 of the reference paper: 1402.1178
So everything seems fine here to my point of view. Even if this does not give 21% for WW.
Where that number is coming from? Is this LO computation or experimental value or higher order computation?

If I do the computation in the SM, I do reproduce the number that you quote.
I observe two differences here with the heft model
1) the default yukawa coupling is different in both model explaining (relatively small) difference between the computed width.
2) some of the H > ZZ* have not been included in the automatic computation since the algorithm consider them has too small and therefore irrelevant for the computation of the total width.

If I do force the computation of all three body decay via the command:
compute_widths h --body_decay=3

I do have the following result:

# PDG Width
DECAY 25 6.500357e-03
# BR NDA ID1 ID2 ...
   8.293051e-01 2 5 -5 # 0.005390778958043033
   3.980520e-02 2 15 -15 # 0.000258747981072848
   1.976891e-02 3 -24 -1 2 # 0.00012850499376002
   1.976891e-02 3 -24 -3 4 # 0.00012850499376002
   1.955585e-02 3 -2 1 24 # 0.00012712001301158
   1.955585e-02 3 -4 3 24 # 0.00012712001301158
   6.703556e-03 3 -12 11 24 # 4.357550345594e-05
   6.703556e-03 3 -14 13 24 # 4.357550345594e-05
   6.611484e-03 3 -24 -11 12 # 4.2977004249726004e-05
   6.611484e-03 3 -24 -13 14 # 4.2977004249726004e-05
   6.529180e-03 3 -16 15 24 # 4.2442002471486006e-05
   6.518411e-03 3 -24 -15 16 # 4.2371996195542004e-05
   1.985045e-03 3 -1 1 23 # 1.2903497564244e-05
   1.985045e-03 3 -3 3 23 # 1.2903497564244e-05
   1.570837e-03 3 -5 5 23 # 1.0211004120654e-05
   1.534916e-03 3 -2 2 23 # 9.977502390342e-06
   1.534916e-03 3 -4 4 23 # 9.977502390342e-06
   8.755930e-04 3 -12 12 23 # 5.6916667869304e-06
   8.755930e-04 3 -14 14 23 # 5.6916667869304e-06
   8.755930e-04 3 -16 16 23 # 5.6916667869304e-06
   4.489292e-04 3 -11 11 23 # 2.9181998712744005e-06
   4.489292e-04 3 -13 13 23 # 2.9181998712744005e-06
   4.271150e-04 3 -15 15 23 # 2.7763997259766e-06

Which indeed include additional three body decay.

So far I do not see anything suspicious about the code.
And do not know where to look from here. Can you comment more?

Cheers,

Olivier

> On 28 Jan 2022, at 08:15, Sarif Khan <email address hidden> wrote:
>
> New question #700425 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/700425
>
> Hi,
>
> I was trying to use the default SM file in Madgraph and also using model files generated from Feynrules. I am not getting the correct branching for Higgses: for example, b bbar is coming around 83% instead of 60% and WW* is coming around 5 % (not 21%). Can you please let me know how I can fix it?
>
> # PDG Width
> DECAY 25 6.475074e-03
> # BR NDA ID1 ID2 ...
> 8.325432e-01 2 5 -5 # 0.005390778958043033
> 3.996062e-02 2 15 -15 # 0.000258747981072848
> 1.986464e-02 3 -24 -1 2 # 0.0001286250107726
> 1.986464e-02 3 -24 -3 4 # 0.0001286250107726
> 1.963221e-02 3 -2 1 24 # 0.0001271200385916
> 1.963221e-02 3 -4 3 24 # 0.0001271200385916
> 6.729730e-03 3 -12 11 24 # 4.357549948313e-05
> 6.729730e-03 3 -14 13 24 # 4.357549948313e-05
> 6.643398e-03 3 -24 -11 12 # 4.301649684302e-05
> 6.643398e-03 3 -24 -13 14 # 4.301649684302e-05
> 6.554674e-03 3 -16 15 24 # 4.244199868572e-05
> 6.550041e-03 3 -24 -15 16 # 4.241199830774e-05
> 1.994417e-03 3 -1 1 23 # 1.2913998699120002e-05
> 1.994417e-03 3 -3 3 23 # 1.2913998699120002e-05
> 1.578206e-03 3 -5 5 23 # 1.0219000517620001e-05
> 1.542221e-03 3 -2 2 23 # 9.985998132340001e-06
> 1.542221e-03 3 -4 4 23 # 9.985998132340001e-06
>
> Best regards,
> Sarif Khan
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi Olivier,

Thanks a lot for the reply.

>Those number can be compare to Table 7 of the reference paper: 1402.1178
So everything seems fine here to my point of view. Even if this does not give 21% for WW.
Where that number is coming from? Is this LO computation or experimental value or higher order computation?

I was trying to match with the Higgs branching (https://twiki.cern.ch/twiki/bin/view/LHCPhysics/CERNYellowReportPageBR):

H -> b b : 58%

H -> WW : 21%

H -> g g : 8% (I need to provide one loop expressions)

In the default file of micromega calchep files they do generate approximately around the experimental ballpark value. So I was wondering where I am missing something. I could not reproduce them using the Feynrules SM_old.fr file as well (my model is based on this base file).

I will be fine if I can at least match with the bb and ww branching.

Can you please let me know if I need to provide additional things?

Best regards,
Sarif Khan

Hi,

That report/numbers are not LO computation but higher order computation.
So this explains the difference and nothing that can really be fixed.

The only place in the code where the BR information is going to be used is within MadSpin.
Cross-section computation and event generation only depends of the value of the coupling and of the total width.

If the total-width is not LO accurate, you can/will face effective BR which are larger than one due to that.
The typical solution is then to rescale the cross-section to get the BR that you want. This would allow you to set the width that you want and still have the correct effect on cuts/distribution (which forbids to use BR formula).

Cheers,

Olivier

> On 28 Jan 2022, at 09:35, Sarif Khan <email address hidden> wrote:
>
> Question #700425 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/700425
>
> Sarif Khan posted a new comment:
> Hi Olivier,
>
> Thanks a lot for the reply.
>
>> Those number can be compare to Table 7 of the reference paper: 1402.1178
> So everything seems fine here to my point of view. Even if this does not give 21% for WW.
> Where that number is coming from? Is this LO computation or experimental value or higher order computation?
>
> I was trying to match with the Higgs branching
> (https://twiki.cern.ch/twiki/bin/view/LHCPhysics/CERNYellowReportPageBR):
>
> H -> b b : 58%
>
> H -> WW : 21%
>
> H -> g g : 8% (I need to provide one loop expressions)
>
> In the default file of micromega calchep files they do generate
> approximately around the experimental ballpark value. So I was wondering
> where I am missing something. I could not reproduce them using the
> Feynrules SM_old.fr file as well (my model is based on this base file).
>
> I will be fine if I can at least match with the bb and ww branching.
>
>
> Can you please let me know if I need to provide additional things?
>
> Best regards,
> Sarif Khan
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.